RU2811503C2 - Methods of detecting and monitoring cancer by personalized detection of circulating tumor dna - Google Patents

Abstract

FIELD: biotechnology.

SUBSTANCE: method of monitoring and detecting minimal residual disease in breast cancer, as well as a method of monitoring and detecting early recurrence or metastasis of colorectal cancer, is described. The claimed methods involve the use of high-throughput sequencing with a read depth of at least 100,000 per amplicon and the detection of two or more patient-specific single nucleotide variants above a confidence threshold of 0.97.

EFFECT: invention expands the range of diagnostic tools for breast cancer treatment.

46 cl, 100 dwg, 18 tbl, 12 ex

Description

CROSS REFERENCE TO RELATED APPLICATIONS

[1] This application claims priority to U.S. Provisional Patent Application No. 62/657,727, filed April 14, 2018; US Provisional Patent Application No. 62/669330, filed May 9, 2018; US Provisional Patent Application No. 62/693843, filed July 3, 2018; US Provisional Patent Application No. 62/715143, filed August 6, 2018; US Provisional Patent Application No. 62/746210, filed October 16, 2018; US Provisional Patent Application No. 62/777973, filed December 11, 2018; US Provisional Patent Application No. 62/804566, filed February 12, 2019. Each of the above applications is incorporated herein by reference in its entirety.

BACKGROUND OF THE ART

[2] Detection of early cancer recurrence or metastasis has traditionally relied on imaging and tissue biopsy. Tumor tissue biopsy is an invasive procedure and carries the risk of potential metastasis or surgical complications, while imaging-based detection is not sensitive enough to detect recurrence or metastasis at an early stage. More effective and less invasive ways to detect cancers are needed to detect recurrence or metastasis.

BRIEF DESCRIPTION OF THE INVENTION

[3] One aspect of the invention presented herein relates to a method for monitoring and detecting early recurrence or metastasis of cancer (for example, breast cancer, bladder cancer or colorectal cancer), comprising obtaining a set of amplicons by performing a multiplex nucleic acid amplification reaction isolated from a sample of blood or urine, or a fraction thereof, from a patient being treated for cancer (e.g., breast cancer, bladder cancer, or colorectal cancer), wherein each amplicon of the set of amplicons spans at least one single nucleotide variant locus of the set of single nucleotide loci patient-specific variants associated with cancer (eg, breast cancer, bladder cancer or colorectal cancer); and determining the sequence of at least a segment of each amplicon from the set of amplicons that contains the patient-specific single nucleotide variant locus, wherein the detection of one or more (or two or more, or three or more, or four or more, or five or more, or six or more, or seven or more, or eight or more, or nine or more, or ten or more) patient-specific single nucleotide variants indicate early recurrence or metastasis of cancer (eg, breast cancer, bladder cancer, or colorectal cancer).

[4] In addition to breast cancer, bladder cancer and colorectal cancer, the methods presented herein can also be used to monitor and detect early relapse or metastasis of other types of cancer, such as: acute lymphoblastic leukemia; acute myeloid leukemia; adrenocortical carcinoma; AIDS-related cancer; AIDS-related lymphoma; anal cancer; appendix cancer; astrocytomas; atypical teratoid/rhabdoid tumor; basal cell carcinoma; brainstem glioma; brain tumor (including brainstem glioma, atypical teratoid/rhabdoid tumor of the central nervous system, embryonal tumors of the central nervous system, astrocytomas, craniopharyngioma, ependymoblastoma, ependymoma, medulloblastoma, medulloepithelioma, pineal parenchymal tumors of intermediate differentiation, supratentorial undifferentiated neuroectodermal tumors and pineoblastoma); bronchial tumors; Burkitt's lymphoma; cancer of unknown primary location; carcinoid tumor; carcinoma of unknown primary location; atypical teratoid/rhabdoid tumor of the central nervous system; fetal tumors of the central nervous system; cervical cancer; cancer in children; chordoma; chronic lymphocytic leukemia; chronic myelogenous leukemia; chronic myeloproliferative disorders; colon cancer; craniopharyngioma; cutaneous T-cell lymphoma; tumors of islet cells of the endocrine system of the pancreas; endometrial cancer; ependymoblastoma; ependymoma; esophageal carcinoma; esthesioneuroblastoma; Ewing's sarcoma; extracranial germ cell tumor; extragonadal germ cell tumor; extrahepatic bile duct cancer; gallbladder cancer; gastrointestinal (stomach) cancer; gastrointestinal carcinoid tumors; gastrointestinal stromal cell tumors; gastrointestinal stromal tumors (GIST); gestational trophoblastic tumor; glioma; hairy cell leukemia; head and neck cancer; heart cancer; Hodgkin's lymphoma; hypopharyngeal cancer; intraocular melanoma; islet cell tumors; Kaposi's sarcoma; kidney cancer; Langerhans cell histiocytosis; laryngeal cancer; lip cancer; liver cancer; malignant fibrous histiocytoma, bone cancer; medulloblastoma; medulloepithelioma; melanoma; Merkel cell carcinoma; Merkel cell skin cancer; mesothelioma; metastatic squamous cell carcinoma of the neck with a primary cancer of unknown origin; oral cancer; multiple endocrine neoplasia syndromes; multiple myeloma; multiple myeloma/plasmocyte neoplasm; mycosis fungoides; myelodysplastic syndromes; myeloproliferative neoplasms; nasal cavity cancer; nasopharyngeal cancer; neuroblastoma; non-Hodgkin's lymphoma; non-melanoma skin cancer; non-small cell lung cancer; oral cancer; oral cancer; oropharyngeal cancer; osteosarcoma; other tumors of the brain and spinal cord; ovarian cancer; epithelial ovarian cancer; germ cell tumor of the ovaries; borderline ovarian tumor; pancreas cancer; papillomatosis; sinus cancer; parathyroid cancer; pelvic cancer; penile cancer; pharynx cancer; pineal parenchymal tumors of intermediate differentiation; pineoblastoma; pituitary tumor; plasma cell neoplasm/multiple myeloma; pleuropulmonary blastoma; primary lymphoma of the central nervous system (CNS); primary hepatocellular liver cancer; prostate cancer; rectal cancer; kidney cancer; renal cell (kidney) cancer; renal cell carcinoma; respiratory tract cancer; retinoblastoma; rhabdomyosarcoma; salivary gland cancer; Sézary syndrome; small cell lung cancer; small bowel cancer; soft tissue sarcoma; squamous cell carcinoma; squamous cell carcinoma of the neck; gastrointestinal cancer (stomach cancer); supratentorial undifferentiated neuroectodermal tumors; T-cell lymphoma; testicular cancer; throat cancer; thymus carcinoma; thymoma; thyroid cancer; transitional cell carcinoma; transitional cell carcinoma of the renal pelvis and ureter; trophoblastic tumor; ureteral cancer; urethral cancer; uterine cancer; uterine sarcoma; vaginal cancer; vulvar cancer; Waldenström's macroglobulinemia; or Wilma's tumor.

[5] In some embodiments, nucleic acids are isolated from a patient's tumor, and somatic mutations in the tumor are identified for a set of patient-specific single nucleotide variant loci before sequencing at least a segment of each amplicon of a set of amplicons for a blood or urine sample, or a fraction thereof, is determined, and where there are single nucleotide variants.

[6] In some embodiments, the method includes collecting and sequencing blood or urine samples from a patient over time.

[7] In some embodiments, at least 2 or at least 5 NNVs are detected, and the presence of at least 2 or at least 5 NNVs is indicative of early recurrence or metastasis of breast cancer, bladder cancer, or colorectal cancer.

[8] In some embodiments, the breast cancer, bladder cancer, or colorectal cancer is stage 1 or stage 2 breast cancer, bladder cancer, or colorectal cancer. In some embodiments, the breast cancer, bladder cancer, or colorectal cancer is is stage 3 or stage 4 breast cancer, bladder cancer, or colorectal cancer.

[9] In some embodiments, the subject has undergone surgical treatment before issuing a blood or urine sample.

[10] In some embodiments, the individual was treated with chemotherapy before issuing a blood or urine sample.

[11] In some embodiments, the individual has been treated with adjuvant or neoadjuvant therapy before obtaining a blood or urine sample.

[12] In some embodiments, the individual was treated with radiation therapy before issuing a blood or urine sample.

[13] In some embodiments, the method further comprises administering a compound to an individual where the compound is known to be particularly effective in treating breast cancer, bladder cancer, or colorectal cancer having one or more of certain single nucleotide variants.

[14] In some embodiments, the method further includes determining the variant allele frequency for each of the single nucleotide variants based on the sequence determination.

[15] In some embodiments, the treatment plan for breast cancer, bladder cancer, or colorectal cancer is determined based on variant allele frequency determinations.

[16] In some embodiments, the method further comprises administering a compound to an individual where the compound is known to be particularly effective for treating breast cancer, bladder cancer, or colorectal cancer having one of the variable allele frequency single nucleotide variants of greater than at least at least half of the other single nucleotide variants identified.

[17] In some embodiments, the sequence is determined using high-throughput DNA sequencing of multiple single nucleotide variation loci.

[18] In some embodiments, the method further includes detecting a clonal single nucleotide variant in breast cancer, bladder cancer, or colorectal cancer by determining the frequency of the variant allele for each of the SNV loci based on the multiple copy sequence of the amplicon series, where the greater relative frequency of the alleles compared to with other single nucleotide variants at multiple single nucleotide variant loci indicates a clonal single nucleotide variant in breast cancer, bladder cancer or colorectal cancer.

[19] In some embodiments, the method further includes administering to the individual a compound that targets one or more clonal single nucleotide variants, but not other single nucleotide variants.

[20] In some embodiments, a variant allele frequency greater than 1.0% is indicative of a clonal single nucleotide variant.

[21] In some embodiments, the method further includes forming an amplification reaction mixture by combining a polymerase, nucleotide triphosphates, nucleic acid fragments from a library of nucleic acids obtained from the sample, and a set of primers, each of which binds within 150 base pairs of single nucleotide variant loci, or a set of primer pairs, each covering a region of 160 base pairs or less containing single nucleotide variant loci, and subjecting the amplification reaction mixture to amplification conditions to produce a set of amplicons.

[22] In some embodiments, determining whether a single nucleotide variant is present in a sample includes identifying a confidence value for each allele definition at each of the sets of single nucleotide variant loci based, at least in part, on the read depth of the loci.

[23] In some embodiments, single nucleotide variant recognition is performed if the confidence value for the presence of the single nucleotide variant is greater than 90%.

[24] In some embodiments, single nucleotide variant recognition is performed if the confidence value for the presence of the single nucleotide variant is greater than 95%.

[25] In some embodiments, the set of single nucleotide variation loci comprises all single nucleotide variation loci defined in the TCGA and COSMIC data sets for breast cancer, bladder cancer, or colorectal cancer.

[26] In some embodiments, the set of single nucleotide variation sites comprises all single nucleotide variation sites identified in the TCGA and COSMIC datasets for breast cancer, bladder cancer, or colorectal cancer.

[27] In some embodiments, the method is performed with a read depth for a set of single nucleotide variation loci of at least 1000.

[28] In some embodiments, the set of single nucleotide variation loci comprises from 25 to 1000 single nucleotide variation loci known to be associated with breast cancer, bladder cancer, or colorectal cancer.

[29] In some embodiments, the efficiency and error rate per cycle are determined for each amplification reaction of the multiplex single nucleotide variation loci amplification reaction, and the efficiency and error rate are used to determine the presence of a single nucleotide variant in a set of single nucleotide variant loci in a sample.

[30] In some embodiments, the amplification reaction is a PCR reaction, and the annealing temperature is 1-15°C above the melting temperature of at least 50% of the primers in the primer set.

[31] In some embodiments, the amplification reaction is a PCR reaction, and the duration of the annealing step in the PCR reaction is from 15 to 120 minutes.

[32] In some embodiments, the amplification reaction is a PCR reaction, and the duration of the annealing step in the PCR reaction is from 15 to 120 minutes.

[33] In some embodiments, the concentration of primer in the amplification reaction is from 1 to 10 nM.

[34] In some embodiments, the primers in the primer set are designed to minimize the formation of a primer dimer.

[35] In some embodiments, the amplification reaction is a PCR reaction, the annealing temperature is 1-15 °C higher than the melting temperature of at least 50% of the primers of the primer set, the duration of the annealing step in the PCR reaction is from 15 to 120 minutes, the concentration The number of primers in the amplification reaction ranges from 1 to 10 nM, and the primers in the primer set are designed to minimize the formation of primer dimers.

[36] In some embodiments, the multiplex amplification reaction is conducted under limiting primer conditions.

[37] Another aspect of the invention presented herein relates to a composition containing tumor circulating nucleic acid fragments containing a universal adapter, wherein the tumor circulating nucleic acids are derived from breast cancer, bladder cancer or colorectal cancer.

[38] In some embodiments, the circulating tumor nucleic acids are obtained from a blood or urine sample or a fraction thereof from an individual with breast cancer, bladder cancer, or colorectal cancer.

[39] Another aspect of the invention presented herein relates to a composition comprising a solid support containing a plurality of clonal populations of nucleic acids, wherein the clonal populations comprise amplicons derived from a sample of circulating free nucleic acids, wherein the circulating tumor nucleic acids are derived from breast cancer gland, bladder or colorectal cancer.

[40] In some embodiments, the circulating free nucleic acids are obtained from a blood or urine sample, or a fraction thereof, from an individual with breast cancer, bladder cancer, or colorectal cancer.

[41] In some embodiments, nucleic acid fragments in different clonal populations contain the same universal adapter.

[42] In some embodiments, clonal populations of nucleic acids are derived from nucleic acid fragments from a set of samples from two or more individuals.

[43] In some embodiments, the nucleic acid fragments comprise one of a number of molecular barcodes corresponding to a sample in the sample set.

[44] Another aspect of the invention presented herein relates to a method for monitoring and detecting early recurrence or metastasis of breast cancer, bladder cancer or colorectal cancer, comprising selecting a set of at least 8 or 16 patient-specific single nucleotide variant loci based on somatic mutations identified in a tumor sample from a patient who has been diagnosed with breast cancer, bladder cancer, or colorectal cancer; collecting one or more blood or urine samples over time from a patient after the patient has undergone surgery, first-line chemotherapy, and/or adjuvant therapy; obtaining a set of amplicons by performing a multiplex amplification reaction on nucleic acids isolated from each blood or urine sample or fraction thereof, where each amplicon from the set of amplicons covers at least one single nucleotide variant locus from a set of patient-specific single nucleotide variant loci associated with breast cancer glands, bladder or colorectal cancer; and determining the sequence of at least a segment of each amplicon from the set of amplicons that contains the patient-specific single nucleotide variant locus at which one or more (or two or more, or three or more, or four or more, or five or more, or six or more, or seven or more, or eight or more, or nine or more, or ten or more) patient-specific single nucleotide variants from a blood or urine sample indicating early recurrence or metastasis of breast, bladder, or colorectal cancer cancer.

[45] Another aspect of the invention presented herein relates to a method of treating breast cancer, bladder cancer or colorectal cancer, comprising treating a patient who has been diagnosed with breast cancer, bladder cancer or colorectal cancer by surgery , first-line chemotherapy and/or adjuvant therapy; dynamically collecting one or more blood or urine samples from a patient; obtaining a set of amplicons by performing a multiplex amplification reaction on nucleic acids isolated from each blood or urine sample or fraction thereof, wherein each amplicon from the set of amplicons covers at least one single nucleotide variant locus from a set of at least 8 or 16 patient-specific loci single nucleotide variants associated with breast cancer, bladder cancer or colorectal cancer that were selected on the basis of somatic mutations identified in the patient's tumor sample; determining the sequence of at least a segment of each amplicon from a set of amplicons that contains a patient-specific single nucleotide variant locus, wherein the detection of one or more (or two or more, or three or more, or four or more, or five or more, or six or more, or seven or more, or eight or more, or nine or more, or ten or more) patient-specific single nucleotide variants from a blood or urine sample indicates early recurrence or metastasis of breast cancer, bladder cancer, or colorectal cancer; and administering the compound to an individual where the compound is known to be effective in treating breast cancer, bladder cancer or colorectal cancer having one or more single nucleotide variants identified from a blood or urine sample.

[46] Another aspect of the invention presented herein relates to a method for monitoring or predicting response to treatment for breast cancer, bladder cancer or colorectal cancer, comprising over time collecting one or more blood or urine samples from a patient who is undergoing treatment. breast cancer, bladder cancer or colorectal cancer; obtaining a set of amplicons by performing a multiplex amplification reaction on nucleic acids isolated from each blood or urine sample or fraction thereof, where each amplicon from the set of amplicons covers at least one single nucleotide variant locus from a set of at least 8 or 16 patient-specific single nucleotide variant loci variants associated with breast cancer, bladder cancer or colorectal cancer that were selected on the basis of somatic mutations identified in the patient's tumor sample; and determining the sequence of at least a segment of each amplicon from the set of amplicons that contains the patient-specific single nucleotide variant locus at which one or more (or two or more, or three or more, or four or more, or five or more, or six or more, or seven or more, or eight or more, or nine or more, or ten or more) patient-specific single nucleotide variants from a blood or urine sample, indicating poor response to treatment for breast cancer, bladder cancer, or colorectal cancer.

[47] In some embodiments, the methods provided herein include detecting ctDNA in the plasma of breast cancer patients before therapy and/or during neoadjuvant therapy (e.g., after cycle 1, cycle 2, cycle 3, cycle 4, etc. .d.). In some embodiments, a treatment plan is provided based on determining ctDNA concentration (eg, presence/absence) and rate of decline during neoadjuvant therapy.

[48] In some embodiments, the methods provided herein include assessing the presence and levels of ctDNA for each cancer patient (ie, targeting mutations that are actually present in the tumor). In some embodiments, the methods provided herein include detecting 2 or more, 4 or more, 10 or more, 16 or more, 32 or more, 50 or more, 64 or more, or 100 or more mutations that are actually present in tumor(s) of patients.

[49] In accordance with some embodiments of the present invention, at least 50%, or at least 60%, or at least 70%, or at least 80%, or at least 90%, or approximately 100% of patients, who will experience metastatic recurrence (eg, after neoadjuvant therapy and surgery) have ctDNA detected at baseline.

[50] In accordance with some embodiments of the present invention, at least 50%, or at least 60%, or at least 70%, or at least 80%, or at least 90%, or approximately 100% of patients, who will have metastatic recurrence (eg, after neoadjuvant therapy and surgery) have ctDNA detected after cycle 1 of neoadjuvant therapy.

[51] In accordance with some embodiments of the present invention, at least 50%, or at least 60%, or at least 70%, or at least 80%, or at least 90%, or approximately 100% of patients, who will have metastatic recurrence (eg, after neoadjuvant therapy and surgery) have ctDNA detected after cycle 2 of neoadjuvant therapy.

[52] In accordance with some embodiments of the present invention, at least 50%, or at least 60%, or at least 70%, or at least 80%, or at least 90%, or approximately 100% of patients, who will have metastatic recurrence (eg, after neoadjuvant therapy and surgery) have ctDNA detected after neoadjuvant therapy and before surgery.

[53] In accordance with some embodiments of the present invention, at least 50%, or at least 60%, or at least 70%, or at least 80%, or at least 90%, or approximately 100% of patients, who will have metastatic recurrence (eg, after neoadjuvant therapy and surgery) have ctDNA detected after surgery.

[54] In accordance with some embodiments of the present invention, at least 50%, or at least 60%, or at least 70%, or at least 80%, or at least 90%, or approximately 100% of patients, in which ctDNA is detected (eg, after surgery) will have metastasis without further treatment for relapse (eg, after neoadjuvant therapy and surgery).

[55] In accordance with some embodiments of the present invention, at least 50%, or at least 60%, or at least 70%, or at least 80%, or at least 90%, or approximately 100% of patients, who have elevated ctDNA levels between baseline and cycle 1, or cycle 2, etc. will have metastatic recurrence after surgery unless additional treatment is given.

[56] In some embodiments, the methods provided herein include detecting the occurrence, recurrence, or metastasis of certain subtypes of cancer, including certain subtypes of breast cancer. In some embodiments, the methods provided herein include detecting the occurrence, recurrence, or metastasis of an HR+/HER2- tumor, including HR+/HER2- breast cancer (eg, hormone receptor-positive-ERα+ and/or PR+). HR+ tumors are usually less aggressive and have a favorable prognosis with a 5-year survival rate of over 90%.

[57] In some embodiments, the methods provided herein include detecting the occurrence, recurrence or metastasis of a HER2+ tumor, including HER2+ (human epidermal growth factor receptor 2 positive) breast cancer. HER2+ tumors tend to be more invasive, have a worse prognosis, and are more susceptible to recurrence and metastasis than HR+/HER2− breast cancers.

[58] In some embodiments, the methods provided herein include detecting the occurrence, recurrence or metastasis of an HR-/HER2- tumor, including HR-/HER2 breast cancer (TNBC or triple negative breast cancer). Triple-negative breast cancer (TNBC) does not express ERα, PR, or HER2. These tumors tend to be the most aggressive and have the worst prognosis of all breast cancer subtypes.

[59] In some embodiments, the method provided herein is capable of detecting patient-specific single nucleotide variants in at least 75%, at least 80%, at least 85%, at least 90%, or at least in 95% of patients with early recurrence or metastasis of cancer.

[60] In some embodiments, the method provided herein is capable of detecting patient-specific single nucleotide variants in at least 80%, at least 85%, at least 90%, at least 95%, or at least in 98% of patients with early relapse or metastasis of HER2+ breast cancer.

[61] In some embodiments, the method provided herein is capable of detecting patient-specific single nucleotide variants in at least 80%, at least 85%, at least 90%, at least 95%, or at least in 98% of patients with early relapse or metastasis of triple-negative breast cancer.

[62] In some embodiments, the method provided herein is capable of detecting patient-specific single nucleotide variants in at least 75%, at least 80%, at least 85%, at least 90%, or at least in 95% of patients with early relapse or metastasis of HR+/HER2- breast cancer.

[63] In some embodiments, the method provided herein is capable of detecting patient-specific single nucleotide variants in patients who have early cancer recurrence or metastasis at least 100 days, at least 150 days, at least 200 days, at least 250 days or at least 300 days before clinical recurrence or metastasis of cancer detected by imaging, and/or at least 100 days, at least 150 days, at least 200 days, at least 250 days or at least 300 days before CA15-3 levels rise.

[64] In some embodiments, the method provided herein is capable of detecting patient-specific single nucleotide variants in patients who have early recurrence or metastasis of HER2+ breast cancer in at least 100 days, at least 150 days, at least 200 days, at least 250 days or at least 300 days before clinical recurrence or metastasis of HER2+ breast cancer detected by imaging, and/or at least 100 days, at least 150 days, at least 200 days, at least 250 days or at least 300 days before CA15-3 levels rise.

[65] In some embodiments, the method provided herein is capable of detecting patient-specific single nucleotide variants in patients having early relapse or metastasis of triple negative breast cancer in at least 100 days, at least 150 days, at least in 200 days, at least 250 days or at least 300 days before clinical recurrence or metastasis of triple-negative breast cancer detected by imaging, and/or at least 100 days at least, 150 days at least at least 200 days, at least 250 days, or at least 300 days before CA15-3 levels rise.

[66] In some embodiments, the method provided herein is capable of detecting patient-specific single nucleotide variants in patients having early recurrence or metastasis of HR+/HER2- breast cancer in at least 100 days, in at least 150 days, in at least measures, 200 days, at least 250 days, or at least 300 days before clinical recurrence or metastasis of HR+/HER2- breast cancer detected by imaging and/or at least 100 days, at least 150 days, at least 200 days, at least 250 days, or at least 300 days before the increase in CA15-3 levels.

[67] In some embodiments, the method presented herein does not detect patient-specific single nucleotide variants in at least 95%, at least 98%, at least 99%, at least 99.5%, by at least 99.8% or at least 99.9% of patients who do not have early recurrence or metastasis of cancer.

[68] In some embodiments, the method presented herein does not detect patient-specific single nucleotide variants in at least 95%, at least 98%, at least 99%, at least 99.5%, by at least 99.8% or at least 99.9% of patients who do not have early recurrence or metastasis of HER2+ breast cancer.

[69] In some embodiments, the method presented herein does not detect patient-specific single nucleotide variants in at least 95%, at least 98%, at least 99%, at least 99.5%, by at least 99.8% or at least 99.9% of patients who do not have early recurrence or metastasis of triple-negative breast cancer.

[70] In some embodiments, the method presented herein does not detect patient-specific single nucleotide variants in at least 95%, at least 98%, at least 99%, at least 99.5%, by at least 99.8% or at least 99.9% of patients who do not have early recurrence or metastasis of HR+/HER2- breast cancer.

[71] In some embodiments, a method provided herein has a specificity of at least 95%, at least 98%, at least 99%, at least 99.5%, at least 99.8%, or at least at least 99.9% in detecting early cancer recurrence or metastasis when two or more patient-specific single nucleotide variants are detected above a predefined confidence threshold (eg, 0.95, 0.96, 0.97, 0.98, or 0.99) .

[72] In some embodiments, a method provided herein has a specificity of at least 95%, at least 98%, at least 99%, at least 99.5%, at least 99.8%, or at least at least 99.9% in detecting early recurrence or metastasis of HER2+ breast cancer when two or more patient-specific single nucleotide variants are detected above a predefined confidence threshold (eg, 0.95, 0.96, 0.97, 0.98, or 0 ,99).

[73] In some embodiments, a method provided herein has a specificity of at least 95%, at least 98%, at least 99%, at least 99.5%, at least 99.8%, or at least at least 99.9% in detecting early recurrence or metastasis of triple-negative breast cancer when two or more patient-specific single nucleotide variants are detected above a predefined confidence threshold (eg, 0.95, 0.96, 0.97, 0.98, or 0.99).

[74] In some embodiments, a method provided herein has a specificity of at least 95%, at least 98%, at least 99%, at least 99.5%, at least 99.8%, or at least at least 99.9% in detecting early recurrence or metastasis of HR+/HER2- breast cancer when two or more patient-specific single nucleotide variants are detected above a predefined confidence threshold (eg, 0.95, 0.96, 0.97, 0. 98 or 0.99).

[75] In some embodiments, the method provided herein detects patient-specific single nucleotide variants in at least 75%, at least 80%, at least 85%, at least 90%, or at least 95% of patients with early recurrence or metastasis of muscle-invasive bladder cancer (MIBC).

[76] In some embodiments, the method provided herein detects patient-specific single nucleotide variants in patients who have had early cancer recurrence or metastasis at least 100 days, at least 150 days, at least 200 days, or at least at least 250 days before clinical recurrence or metastasis of MIBC detected by imaging.

[77] In some embodiments, the method presented herein does not detect patient-specific single nucleotide variants in at least 95%, at least 98%, at least 99%, at least 99.5%, by at least 99.8% or at least 99.9% of patients who do not have early recurrence or metastasis of MIBC.

[78] In some embodiments, a method provided herein has a specificity of at least 95%, at least 98%, at least 99%, at least 99.5%, at least 99.8%, or at least 99.9% in detecting early relapse or metastasis of MIBC when two or more patient-specific single nucleotide variants are detected above a predefined confidence threshold (eg, 0.95, 0.96, 0.97, 0.98, or 0.99).

[79] In addition to or as an alternative to single nucleotide variants, the methods presented herein can also be based on the detection of other genomic variants, such as insertions/deletions, multiple nucleotide variants and/or gene fusions.

[80] Accordingly, an additional aspect of the invention presented herein relates to a method for monitoring and detecting early recurrence or metastasis of breast cancer, bladder cancer or colorectal cancer, comprising selecting multiple genomic variant loci (e.g., HB, insertion/deletion, multiple nucleotide variant and gene fusion) based on somatic mutations identified in a tumor sample from a patient diagnosed with breast cancer, bladder cancer, or colorectal cancer; collecting one or more blood or urine samples over time from a patient after the patient has undergone surgery, first-line chemotherapy, and/or adjuvant therapy; obtaining a set of amplicons by performing a multiplex amplification reaction on nucleic acids isolated from each blood or urine sample or fraction thereof, where each amplicon from the set of amplicons covers at least one genomic variant locus from a set of patient-specific genomic variant loci associated with breast cancer glands, bladder or colorectal cancer; and determining the sequence of at least a segment of each amplicon from the set of amplicons that contains the patient-specific locus of genomic variants in which one or more (or two or more, or three or more, or four or more, or five or more, or six or more, or seven or more, or eight or more, or nine or more, or ten or more) patient-specific genomic variants from a blood or urine sample, indicating early recurrence or metastasis of breast, bladder, or colorectal cancer cancer.

[81] A further aspect of the invention provided herein relates to a method of treating breast cancer, bladder cancer or colorectal cancer, comprising treating a patient who has been diagnosed with breast cancer, bladder cancer or colorectal cancer with surgery, first-line chemotherapy and/or adjuvant therapy; dynamically collecting one or more blood or urine samples from the patient; obtaining a set of amplicons by performing a multiplex amplification reaction on nucleic acids isolated from each blood or urine sample or fraction thereof, where each amplicon from the set of amplicons covers at least one locus of genomic variants (for example, NCB, insertion/deletion, multiple nucleotide variant and gene fusion) from a set of at least 8 or 16 patient-specific genomic variant loci associated with breast cancer, bladder cancer or colorectal cancer that were selected based on somatic mutations identified in the patient's tumor sample; determining the sequence of at least a segment of each amplicon from a set of amplicons that contains a patient-specific locus of genomic variants, wherein detecting one or more (or two or more, or three or more, or four or more, or five or more, or six or more, or seven or more, or eight or more, or nine or more, or ten or more) patient-specific genomic variants from a blood or urine sample indicate early recurrence or metastasis of breast cancer, bladder cancer, or colorectal cancer; and administering the compound to an individual where the compound is known to be effective for treating breast cancer, bladder cancer or colorectal cancer having one or more genomic variants detected in a blood or urine sample.

[82] A further aspect of the invention provided herein relates to a method for monitoring or predicting response to treatment for breast cancer, bladder cancer or colorectal cancer, comprising over time collecting one or more blood or urine samples from a patient who is undergoing cancer treatment breast, bladder or colorectal cancer; obtaining a set of amplicons by performing a multiplex amplification reaction on nucleic acids isolated from each blood or urine sample or fraction thereof, where each amplicon from the set of amplicons covers at least one locus of genomic variants (for example, NCB, insertion/deletion, multiple nucleotide variant and gene fusion) from a set of at least 8 or 16 patient-specific genomic variant loci associated with breast cancer, bladder cancer or colorectal cancer that were selected based on somatic mutations identified in the patient's tumor sample; and determining the sequence of at least a segment of each amplicon from the set of amplicons that contains the patient-specific locus of genomic variants in which one or more (or two or more, or three or more, or four or more, or five or more, or six or more, or seven or more, or eight or more, or nine or more, or ten or more) patient-specific genomic variants from a blood or urine sample indicating poor response to treatment for breast, bladder, or colorectal cancer cancer.

[83] In addition to or as an alternative to patient-specific genomic variants, the methods presented herein can also be based on the detection of recurrent mutations associated with cancer (eg, cancer hot spot mutations, markers associated with drug resistance, cancer mutations). panels) that are repeated in many patients with cancer.

[84] Accordingly, a further aspect of the invention presented herein relates to a method for monitoring and detecting early recurrence or metastasis of breast cancer, bladder cancer or colorectal cancer, comprising selecting a plurality of recurrent cancer-associated mutations; collecting one or more blood or urine samples from a patient over time after the patient has undergone surgery, first-line chemotherapy, and/or adjuvant therapy; obtaining a set of amplicons by performing a multiplex amplification reaction on nucleic acids isolated from each blood or urine sample or fraction thereof, where each amplicon from the set of amplicons covers at least one of a set of recurrent mutations associated with breast cancer, bladder cancer or colorectal cancer cancer; and determining the sequence of at least a segment of each amplicon from the set of amplicons that includes a recurrent cancer-associated mutation, wherein the detection of one or more (or two or more, or three or more, or four or more, or five or more, or six or more, or seven or more, or eight or more, or nine or more, or ten or more) recurring, cancer-associated mutations from a blood or urine sample, indicating early recurrence or metastasis of breast, bladder, or colorectal cancer cancer.

[85] A further aspect of the invention provided herein relates to a method of treating breast cancer, bladder cancer or colorectal cancer, comprising treating a patient who has been diagnosed with breast cancer, bladder cancer or colorectal cancer with an operation, first-line chemotherapy and/or adjuvant therapy; dynamically collecting one or more blood or urine samples from the patient; producing a set of amplicons by performing a multiplex amplification reaction on nucleic acids isolated from each blood or urine sample or fraction thereof, wherein each amplicon of the set of amplicons covers at least one recurrent cancer-associated mutation (e.g., cancer hot spot mutation, marker associated drug-resistant, cancer panel mutation) from a set of at least 8 or 16 recurrent mutations associated with breast cancer, bladder cancer or colorectal cancer; determining the sequence of at least a segment of each amplicon from a set of amplicons that contains a recurrent cancer-associated mutation, wherein detecting one or more (or two or more, or three or more, or four or more, or five or more, or six or more, or seven or more, or eight or more, or nine or more, or ten or more) recurring cancer-associated mutations from a blood or urine sample, indicating early recurrence or metastasis of breast cancer, bladder cancer, or colorectal cancer ; and administering the compound to an individual where the compound is known to be effective in treating breast cancer, bladder cancer, or colorectal cancer having one or more recurrent cancer-associated mutations found in the blood or urine sample.

[86] A further aspect of the invention provided herein relates to a method for monitoring or predicting response to treatment for breast cancer, bladder cancer or colorectal cancer, comprising over time collecting one or more blood or urine samples from a patient who is undergoing cancer treatment breast, bladder or colorectal cancer; producing a set of amplicons by performing a multiplex amplification reaction on nucleic acids isolated from each blood or urine sample or fraction thereof, wherein each amplicon of the set of amplicons covers at least one recurrent cancer-associated mutation (e.g., cancer hot spot mutation, marker, drug resistance associated, cancer panel mutation) from a set of at least 8 or 16 recurrent mutations associated with breast cancer, bladder cancer or colorectal cancer; and determining the sequence of at least a segment of each amplicon from the set of amplicons that includes a recurrent cancer-associated mutation, wherein the detection of one or more (or two or more, or three or more, or four or more, or five or more, or six or more, or seven or more, or eight or more, or nine or more, or ten or more) recurring cancer-related mutations from a blood or urine sample indicate a poor response to treatment for breast cancer, bladder cancer, or colorectal cancer.

[87] In addition to or as an alternative to initially identifying somatic mutations from a tumor sample of a patient diagnosed with breast cancer, bladder cancer, or colorectal cancer, the methods presented herein can also rely on the detection of somatic mutations from other biological samples patients, such as blood, serum, plasma, urine, hair, tears, saliva, skin, nails, feces, bile, lymph, cervical mucus or semen.

[88] Accordingly, an additional aspect of the invention presented herein relates to a method for monitoring and detecting early recurrence or metastasis of breast cancer, bladder cancer or colorectal cancer, comprising selecting multiple genomic variant loci (e.g., HB, insertion/deletion, multinucleotide variant and gene fusion) based on somatic mutations identified in a biological sample including cancer-associated mutations (eg, blood, serum, plasma, urine, hair, tears, saliva, skin, nails, feces, bile, lymph, cervical mucus or semen) from a patient who has been diagnosed with breast cancer, bladder cancer, or colorectal cancer; collecting one or more blood or urine samples over time from a patient after the patient has undergone surgery, first-line chemotherapy, and/or adjuvant therapy; obtaining a set of amplicons by performing a multiplex amplification reaction on nucleic acids isolated from each blood or urine sample or fraction thereof, wherein each amplicon from the set of amplicons covers at least one genomic variant locus from a set of patient-specific genomic variant loci associated with breast cancer glands, bladder or colorectal cancer; and determining the sequence of at least a segment of each amplicon from the set of amplicons that contains a patient-specific locus of genomic variants in which one or more (or two or more, or three or more, or four or more, or five or more, or six or more, or seven or more, or eight or more, or nine or more, or ten or more) patient-specific genomic variants from a blood or urine sample, indicating early recurrence or metastasis of breast cancer, bladder cancer, or colorectal cancer.

[89] A further aspect of the invention provided herein relates to a method of treating breast cancer, bladder cancer or colorectal cancer, comprising treating a patient who has been diagnosed with breast cancer, bladder cancer or colorectal cancer with surgery, first-line chemotherapy and/or adjuvant therapy; dynamically collecting one or more blood or urine samples from the patient; obtaining a set of amplicons by performing a multiplex amplification reaction on nucleic acids isolated from each blood or urine sample or fraction thereof, wherein each amplicon from the set of amplicons covers at least one locus of genomic variants (e.g., HB, insertion/deletion, multiple nucleotide variant and gene fusion) a set of at least 8 or 16 patient-specific genomic variant loci associated with breast cancer, bladder cancer or colorectal cancer that have been selected on the basis of somatic mutations identified in a patient biological sample containing mutations associated with cancer (eg, blood, serum, plasma, urine, hair, tears, saliva, skin, nails, feces, bile, lymph, cervical mucus or semen); determining the sequence of at least a segment of each amplicon from a set of amplicons that contains a patient-specific locus of genomic variants, wherein detecting one or more (or two or more, or three or more, or four or more, or five or more, or six or more, or seven or more, or eight or more, or nine or more, or ten or more) patient-specific genomic variants from a blood or urine sample indicate early recurrence or metastasis of breast cancer, bladder cancer, or colorectal cancer; and administering the compound to an individual where the compound is known to be effective for treating breast cancer, bladder cancer or colorectal cancer having one or more genomic variants detected from a blood or urine sample.

[90] A further aspect of the invention provided herein relates to a method for monitoring or predicting response to treatment for breast cancer, bladder cancer or colorectal cancer, comprising over time collecting one or more blood or urine samples from a patient who is undergoing cancer treatment breast, bladder or colorectal cancer; obtaining a set of amplicons by performing a multiplex amplification reaction on nucleic acids isolated from each blood or urine sample or fraction thereof, wherein each amplicon from the set of amplicons covers at least one locus of genomic variants (for example, HB, insertion/deletion, multiple nucleotide variant and gene fusion) a set of at least 8 or 16 patient-specific genomic variant loci associated with breast cancer, bladder cancer or colorectal cancer that have been selected on the basis of somatic mutations identified in a patient biological sample containing mutations associated with cancer (eg, blood, serum, plasma, urine, hair, tears, saliva, skin, nails, feces, bile, lymph, cervical mucus or semen); and determining the sequence of at least a segment of each amplicon from the set of amplicons that contains the patient-specific locus of genomic variants in which one or more (or two or more, or three or more, or four or more, or five or more, or six or more, or seven or more, or eight or more, or nine or more, or ten or more) patient-specific genomic variants from a blood or urine sample, indicating poor response to treatment for breast cancer, bladder cancer, or colorectal cancer.

[91] Other embodiments, features and advantages of the disclosed inventions will be apparent from the following detailed description and claims.

BRIEF DESCRIPTION OF THE DRAWINGS

[92] The patent or application file contains at least one drawing in color. Copies of this patent or patent application publication with color drawing(s) will be made available by the Office upon request and upon payment of the required fee.

[93] The embodiments disclosed herein will be further explained with reference to the accompanying drawings, in which like structures are designated by like numerals in multiple views. The drawings shown are not necessarily to scale, but instead tend to focus on illustrating the principles of the embodiments currently disclosed.

[94] FIG. 1 is a workflow diagram.

[95] FIG. 2 Top panel: number of NDCs per sample; bottom panel: performance tests sorted by driver category.

[96] FIG. 3. Measured concentration of scDNA. Each data point refers to a plasma sample.

[97] FIG. 4. Samples showing good correlation between tissue FNA measurements determined previously (x-axis) and in this case using mPCR-SNP (y-axis). Each sample is shown in a separate frame, and the FNA data points are colored according to the tissue region.

[98] FIG. 5. Samples showing weak correlation between tissue FNA measurements determined previously (x-axis) and in this case using mPCR-SNP (y-axis). Each sample is shown in a separate frame, and the FNA data points are colored according to the tissue region.

[99] FIG. 6A-B. Histogram of reading depth depending on the received recognition. Top: The analysis did not detect the expected plasma NDC. Bottom: The analysis detected the expected plasma NDC.

[100] FIG. 7. Number of ONVs detected in plasma according to histological type.

[101] FIG. 8. ONV detection (left) and sample detection (right) in plasma by tumor stage.

[102] FIG. 9. Plasma NVA as a function of tumor stage and CNV clonality.

[103] FIG. 10. Number of ONVs detected in plasma from each sample as a function of the amount of cDNA input.

[104] FIG. 11. Plasma NVA as a function of the average tumor NVA. The average tumor NNA was calculated across all tumor sites analyzed for each tumor.

[105] FIG. Figure 12 shows the clone ratio (red to blue) and the mutant variant allele frequency (MyTVA) of each detected NCB. The total number of NNVs detected in each sample is placed in one column, and samples are classified by tumor stage (pTNM stage). Samples without detected NDCs were included. The clonal ratio is defined as the ratio between the number of tumor sites in which a CNV was observed and the total number of sites analyzed from that tumor.

[106] FIG. Figure 13 shows the clonal status (blue for clonal and red for subclonal) and mutant variant allele frequency (MutFVA) of each detected SNV. The total number of NNVs detected in each sample is placed in one column, and samples are classified by tumor stage (pTNM stage). Samples without detected NDCs were included. Clonal status was determined using PyCloneCluster using whole exome sequencing data from tumor tissue.

[107] FIG. 14 shows the clonal status (blue for clonal and red for subclonal) and mutant variant allele frequency (MutVA) of each detected NNV, where the top panel shows only clonal NNV and the lower panel shows only subclonal NNV. The total number of NNVs detected in each sample is placed in one column, and samples are classified by tumor stage (pTNM stage). Samples without detected NDCs were included. Clonal status was determined using PyCloneCluster using whole exome sequencing data from tumor tissue.

[108] FIG. 15 shows the number of ONVs detected in plasma as a function of histological type and tumor size. The histological type and tumor size were determined based on the pathological report. Each data point is colored according to size, with red representing the largest tumor size and blue representing the smallest tumor size.

[109] FIG. 16 is a cDNA analysis table showing DNA concentration, genome copy equivalents in the library preparation, degree of hemolysis in plasma, and cDNA profile in all samples.

[110] FIG. 17 is a table of the NDCs detected in plasma for each sample.

[111] FIG. 18 is a table of additional NCBs identified in plasma.

[112] FIG. 19 is an example of the identified assays and their initial allelic fractions for a plasma sample at the time of relapse (LTX103).

[113] FIG. 20A-B: Clinical and molecular protocol diagram.

[114] FIG. 21: Research Overview.

[115] FIG. 22: Summary of patients for 36 months of follow-up and plasma collection.

[116] FIG. 23A-B: Risk of relapse after therapy stratified by postoperative ctDNA status.

[117 FIG. 24A-B: Risk of relapse after therapy stratified by postoperative ctDNA status.

[118] FIG. 25: Efficacy of adjuvant therapy in preventing relapses.

[119] FIG. 26A-B: Time to relapse based on radiological studies and ctDNA.

[120] FIG. 27A-D: Early detection of relapse and prediction of response to treatment.

[121] FIG. 28: Clinical specimen collection flowchart.

[122] FIG. 29: QC plasma sequencing.

[123] FIG. 30A-F: Early detection of relapse.

[124] FIG. 31: Disease-free survival and ctDNA status at diagnosis and after cystectomy.

[125] FIG. 32A-B: Response to neoadjuvant treatment.

[126] FIG. 33: Signatera Process (RUO).

[127] FIG. 34: QC plasma sequencing.

[128] FIG. 35: Sensitivity of detecting a single NVG.

[129] FIG. 36: Expected Input vs. Observed NEA with Signatera (RUO).

[130] FIG. 37: Summary of patients for the breast cancer study in Example 6.

[131] FIG. 38A-H: is a table of information about the samples analyzed in the study of Example 6. FIG. 38A is part 1 of the table. FIG. 38B is a continuation of the table. FIG. 38C is a continuation of the table. FIG. 38D is a continuation of the table. FIG. 38E is a continuation of the table. 38F is a continuation of the table. FIG. 38G is a continuation of the table. FIG. 38H is a continuation of the table.

[132] FIG. 39: Patient demographics in the breast cancer study from Example 6. Raw PES data obtained from 50 patients (with driver variations for 35 patients). 218 plasma samples were obtained at various time points (from 1 to 8). 108 additional isolated DNA samples were obtained. Relapse status was also collected. Blood samples were collected after adjuvant therapy at 6-month intervals.

[133] FIG. 40: Summary of PES analysis and pool design for the breast cancer study in Example 6. Pool A is based on the Signatera method. Pool B contains 25 patients and is indicated by an asterisk in the bar and box plots. 19 patients in pool B had low tumor purity. 6 patients had very early stage HER2- tumors. Pool B contains driver options.

[134] FIG. 41: Plasma samples for breast cancer study in Example 6. Median plasma volume was 4 ml. The median DNA input was 26 ng. The median DNA input is lower than that of CRC and MIBC samples (45 ng and 66 ng, respectively).

[135] FIG. 42: Sequencing quality control showing the median process error rate for each type and the median depth of analysis for the breast cancer study in Example 6. A total of 326 plasma sequencing samples were processed. The false mutation detection rate is estimated at 0.28%.

[136] FIG. 43: Replicate studies for plasma samples for breast cancer research in Example 6. 319 sequenced samples with 214 unique plasma samples are presented for 49 patients.

[137] FIG. 44: Results from Pool A for the breast cancer study in Example 6. Of the 49 patients, 11 were initially positive. 3 have only one time point. The remaining 8 patients remain positive all the time. Pool B and driver produced similar results. Driver information: 16 relapsed samples have driver mutations. 11 relapse samples had at least one driver assay.

[138] FIG. 45: Summary of 16 patients with detected ctDNA.

[139] FIG. 46: Graphical representation of data corresponding to patient CD047 (TNBC) in FIG. 38.

[140] FIG. 47: Graphical representation of data corresponding to patient CD033 (TNBC) in FIG. 38.

[141] FIG. 48: Graphical representation of data corresponding to patient CD037 (HER2+) in FIG. 38.

[142] FIG. 49: Graphical representation of data corresponding to patient CD040 (HER2+) in FIG. 38.

[143] FIG. 50: Graphical representation of data corresponding to patient CD048 (HER2-) in FIG. 38.

[144] FIG. 51: Graphical representation of data corresponding to patient CD005 (HER2-) in FIG. 38.

[145] FIG. 52: Graphical representation of data corresponding to patient CD036 (HER2-) in FIG. 38.

[146] FIG. 53: Graphical representation of data corresponding to patient CD044 (HER2-) in FIG. 38.

[147] FIG. 54: Graphical representation of data corresponding to patient CD049 in FIG. 38.

[148] FIG. 55: Graphical representation of data corresponding to patient CD029 in FIG. 38.

[149] FIG. 56: Graphical representation of data corresponding to patient CD026 in FIG. 38.

[150] FIG. 57: Graphical representation of data corresponding to patient CD017 in FIG. 38.

[151] FIG. 58: Graphical representation of data corresponding to patient CD031 in FIG. 38. HW: SHC2, PKD1, COLEC12.

[152] FIG. 59: Graphical representation of data corresponding to patient CD025 in FIG. 38. In this patient, ctDNA was examined for the presence of a mutation in FGF9 at 2 consecutive time points. This patient may have a relapse in the near future.

[153] FIG. 60: Patient enrollment and clinical specimen collection. For the 49 women with breast cancer followed in this study, collected tumor tissue samples and serial plasma samples were analyzed in a blinded manner using the Signatera™ RUO workflow. Exome alterations were determined by paired-end sequencing of formalin-fixed paraffin-embedded (FFPE) tumor tissue samples and matched normal DNA. Patient-specific panels were developed that included 16 somatic mutations identified by PES. Plasma samples were processed using appropriate individual panels. A total of 208 samples were analyzed to detect ctDNA.

[154] FIG. 61A-C: Summary and results of ctDNA analysis. (A) Summary of each patient's treatment regimen (n=49) and results of analysis of serial plasma samples (n=208). (B) Summary table showing the total number of patients in each breast cancer subtype, number of recurrences, percentage detected by ctDNA analysis, and median lead time in days. (C) Comparison of molecular and clinical recurrence colored according to breast cancer subtype - HR+, HER2+, TNBC using paired Wilcoxon signed rank test (p value less than 0.001).

[155] FIG. 62A-B: Detection of ctDNA in serial plasma samples predicts disease-free survival (A) Disease-free survival according to detection of ctDNA in any subsequent plasma samples after surgery [HR: 35.84 (7.9626 - 161.32] p value less than 0.001. (B Disease-free survival based on detection of ctDNA in the first postoperative plasma sample [HR: 11.784 (4.2784 - 32.457). Data are from n = 49 patients with a p value less than 0.001.

[156] FIG. 63: (A-E) Plasma levels of ctDNA at multiple plasma time points for five breast cancer patients (one per panel). Whole exome sequencing of the primary tumor and matched controls identified patient-specific somatic mutations. Using the analytically validated Signatera™ RUO workflow, each patient-specific assay was designed to identify 16 somatic ONVs and insertion/deletion variants using massively parallel sequencing (median depth greater than 100,000X per target). Median NVA values are indicated by a blue circle, and the solid line represents the average NVA profile over time. Lead time is calculated by the difference for clinical relapse and molecular relapse. CA15-3 levels are plotted over time, with baseline levels indicated in a light blue hue. (F) Summary of NFA and number of targets detected at molecular and clinical relapse for all ctDNA-positive samples excluding single time point patients.

[157] FIG. 64A-C: Signatera option selection strategy for 49 patient-specific panels. (Top) Distribution of tumor tissue NVA on an individual patient panel. Different colors represent different subtypes: HER2- (dark blue), triple negative (orange), and HER2+ (green). (Center) Number of putative clonal and subclonal variants in an individual patient panel. The median number of clonal variants across the 49 individual panels is 13 out of 16. (Bottom) Number of putative clonal and subclonal variants in patient PES data.

[158] FIG. 65: (A-L) Plasma ctDNA levels at multiple plasma time points for 12 (11 relapsed and 1 nonrelapsed) patients with breast cancer. Whole exome sequencing of the primary tumor and matched controls identified patient-specific somatic mutations. Using the analytically validated Signatera™ workflow, each patient-specific assay was designed to identify 16 somatic ONVs and insertion/deletion variants using massively parallel sequencing (average depth greater than 100,000X per target). The mean NVA values are indicated by the blue circle, and the solid line represents the average NVA profile over time. Lead time is calculated by the difference for clinical relapse and molecular relapse. CA15-3 levels are plotted over time, with baseline levels indicated in a light blue hue.

[159] FIG. 66: Distribution of NVA and number of mutants. A total of 251 targets were detected in ctDNA-positive plasma samples. The NVA of detected targets ranged from 0.01% to 64% with a median value of 0.82%. We used the observed mutation-specific TNA and the total number of DNA molecules in each sample to calculate the number of tumor molecules present in the patient's plasma sample. The number of mutant molecules detected in the 251 positive targets ranged from 1 to 6500 mutant molecules, with a median of 39 molecules.

[160] FIG. 67A-D: Signatera Quality Control Process: Quality control was carried out at every stage of the work process. Of the 215 plasma samples, 208 passed the sample quality control process, and of the 784 unique assays developed, 767 assays passed the assay quality control process (representing a total of 3237 tests, of which 3328 were passed across all samples). A) Extracted cDNA per ml. The scDNA extracted from each plasma sample was quantified using the Quant-iT High Sensitivity dsDNA Assay Kit. Samples where the amount of cDNA was less than 5 ng were marked as ATTENTION. Extracted cDNA per ml ranged from 1 to 21.4 ng with a median of 4.7 ng. B) Preparing the DNA quantity input for the library. In the library preparation protocol, up to 66 ng of scDNA from each plasma sample was used as input. The amount of DNA input for the library ranged from 1 to 66 ng with a median of 25.02. Purified libraries were subjected to quality control before moving to the next step. C) Sequencing coverage. Tests with coverage less than 5000x were excluded from the analysis. Subsequently, samples with fewer than 8 tests failed sequence coverage quality control. The median read depth for QC tested assays was 110,000x. D) Pattern consistency. To track sample integrity, SNP tags were used to measure consistency between patient samples. A genotyping concordance score was calculated for each plasma sample compared to their corresponding paired normal genotyping data. Samples were considered to originate from the same patients when at least 85% of their SNVs had identical genotypes. Six plasma samples identified as replacements were excluded from ctDNA analyses.

[161] FIG. 68A-B: Results of analytical validation. (A) Single target detection sensitivity. Using Signatera, an analytical sensitivity of approximately 60% for mutation detection was achieved with approximately 0.03% added tumor DNA. (b) Sample-level sensitivity assessment for Signatera when at least two mutations are detected from a set of 16 target variants.

[162] FIG. 69: After screening and recruitment of patients, 6 monthly blood draws were performed. HER2 status was determined using immunohistochemical analysis and fluorescence in situ hybridization analysis. If either test came back positive, the patient was considered to have HER2-positive cancer. NACT: neoadjuvant chemotherapy ACT: adjuvant chemotherapy.

[163] FIG. 70: Workflow diagram for the muscle-invasive bladder cancer study in Example 9.

[164] FIG. 71A-G: Patient summary for the muscle-invasive bladder cancer study in Example 9. FIG. 71A shows the frequency of synonymous and nonsynonymous mutations recognized from the RPE. One patient's tumor was hypermutated with a mutational load of 126 mutations/kb. and she demonstrated a POLD1 mutation, which has previously been associated with hypermutators. (Campbell, B.B. et al. Comprehensive Analysis of Hypermutation in Human Cancer. Cell 171, 1042-1056.e10 (2017). FIG. 71B shows the relative contribution of mutational signatures associated with bladder cancer. FIG. 71C shows mutations in commonly mutated genes in bladder cancer (TCGA) (Robertson, A. G. et al. Comprehensive Molecular Characterization of Muscle-Invasive Bladder Cancer. Cell 171, 540-556.e25 (2017)) FIG. 71D shows deleterious mutations in genes associated with the response for DNA damage response (DDR), mutated in more than 5% of 68 samples.FIG. 71E shows the total number of deleterious DDR mutations.FIG.71F shows clinical and histopathological characteristics.FIG.71G shows the total ctDNA status.

[165] FIG. 72: Diagram outlining the clinical protocol and sampling schedule for the muscle-invasive bladder cancer study in Example 9.

[166] FIG. 73: Diagram outlining the Signatera™ workflow.

[167] FIG. 74: Consistent presentation of ctDNA results for all samples analyzed corresponding to the muscle-invasive bladder cancer study in Example 9. Patients are divided into three groups based on ctDNA status: the top panel shows patients with positive ctDNA before and after cystectomy (CE); middle panel shows patients positive for ctDNA before CE only; bottom panel shows ctDNA negative patients. Horizontal lines indicate the disease course of each patient and circles indicate ctDNA status, red circles indicate samples with at least 2 positive tests. Treatment and imaging information is provided for each patient.

[168] FIG. 75A-E: Graphical representation of the predictive value of ctDNA detection for the study of muscle-invasive bladder cancer in Example 9. Kaplan-Meier survival analysis showing the probability of disease-free survival (RFS) and overall survival (OS) stratified by pre-chemotherapy ctDNA status ( FIG. 75A), before cystectomy (CE) (FIG. 75B) and after cystectomy (CE) (FIG. 75C). FIG. 75D shows the relationship between disease recurrence and pre-chemotherapy, pre-cystectomy and post-cystectomy ctDNA status, and with disease recurrence and pre-cystectomy lymph node status. FIG. 75E shows the relationship between pre-cystectomy ctDNA status (CE) and pathological status at cystectomy (CE). Statistical significance was assessed using the Wilcoxon signed rank test for continuous variables and the Fisher exact test for categorical variables.

[169] FIG. 76: Graphs showing ctDNA changes during selected disease courses for the muscle-invasive bladder cancer study in Example 9. FIG. 76 shows a presentation of detailed disease courses, treatments used, and associated sequential ctDNA analyzes from selected patients. ctDNA status, treatment used, and imaging results are presented as per caption. A positive lead time for ctDNA-based relapse detection is indicated.

[170] FIG. 77: Graphs showing the difference in time between molecular relapse (ctDNA positivity) and clinical relapse (radiography positive) for the muscle-invasive bladder cancer study in Example 9. P value was calculated using the paired Wilcoxon signed rank test.

[171] FIG. 78A-H: Graphs showing predictive markers of response to chemotherapy for the muscle-invasive bladder cancer study in Example 9. FIG. 78A shows the relationship between disease relapse and response to chemotherapy. FIG. 78B shows, respectively, the relative contribution of signature 5 for all patients stratified by chemotherapy response and ERCC2 mutation status. FIG. 78C shows the proportion of patients responding to therapy relative to ERCC2 mutation status. FIG. 78D RNA subtype figure figures_NEW. FIG. 78E shows the relationship between ctDNA and response to chemotherapy for patients who are ctDNA negative throughout the disease course, patients whose ctDNA levels drop to zero, and patients whose ctDNA levels remain positive. FIG. 78F shows the ctDNA level for all patients with detectable ctDNA before, during and after chemotherapy. Patients are grouped by response to chemotherapy and relapse status is indicated.

[172] FIG. 79A-D: Graphs showing the total number of identified mutations per patient as a function of ERCC2 status or number of DNA damage response (DDR) mutations for the muscle-invasive bladder cancer study in Example 9.

[173] FIG. 80: Graphs depicting genomic heterogeneity in primary tumor and metastatic relapse for the muscle-invasive bladder cancer study in Example 9. Whole exome sequencing (WES) data from primary tumors were compared with ctDNA. PES data from plasma samples with high variant allele frequency (VAF) in ctDNA detected in metastatic relapse. Genomic positions with mutations identified in either plasma or tumor exome data were examined for base counts. The resulting allele frequencies identified in plasma and tumor exome data are shown. Individual mutations are color coded according to the statistical probability (power) of recognizing the mutation. Venn diagrams represent the number of mutations identified exclusively in tumor, plasma, or both.

[174] FIG. 81: Graphs showing variant allele frequency (%NA) on different days relative to cystectomy (CE) in 8 patients from the muscle-invasive bladder cancer study in Example 9.

[175] FIG. 82: Graphs showing plasma ctDNA levels in 10 patients previously analyzed by digital droplet PCR compared with ultra-deep sequencing for the study of muscle-invasive bladder cancer in Example 9.

[176] FIG. 83A-E: Graphs showing clinical, histopathological and molecular parameters for all 125 patients. FIG. 83A shows the relative contribution of the five most common mutational signatures associated with colorectal cancer. FIG. 83B shows the frequency of synonymous and nonsynonymous mutations recognized from the RPE. FIG. 83C depicts a graph showing mutations in commonly mutated genes in colorectal cancer (TCGA) {Cancer Genome Atlas, 2012 #52}. FIG. 83D shows clinical and histopathological characteristics. FIG. 83E shows a graph summarizing the ctDNA status before and after surgery.

[177] FIG. 84: Diagram showing patient inclusion, sample collection, and definitions of patient subgroups used to address specific clinical questions. Abbreviations: ctDNA, circulating tumor DNA; CT, computed tomography; post-operative, postoperative; TTR, time to relapse.

[178] FIG. 85A-C: Graphs showing workflow quality control (QC) tests for the entire exome sequence of patient samples. 793 (99%) of 795 plasma samples passed the sample quality control process. 194 samples (from 70 patients) were run with the SNP marker to verify the match between the plasma sample and the corresponding tissue biopsy. All 194 plasma samples underwent QC. FIG. 85A shows the amount of DNA input into the library preparation. Up to 66 ng of cell-free DNA (cfDNA) from each plasma sample was used as input to the library preparation protocol. The amount of library DNA input varied from 1 to 66 ng with a median of 45.66. Purified libraries were subjected to quality control before moving to the next step. One sample did not pass the library preparation QC. FIG. 85B shows sequencing coverage. Analyzes with coverage less than 5000x were excluded from the analysis. Subsequently, samples with fewer than 8 assay runs failed quality control of sequence coverage. One sample does not meet the sequence coverage requirement. The median read depth for assays that passed QC coverage was 105,000x. FIG. 85C shows the sequencing error rate measured in all plasma samples. The average error rate for transitions is 5e-5, and the average error rate for transversions is 8e-6.

[179] FIG. 86 shows the results and dynamics of circulating tumor DNA (ctDNA) for each individual patient.

[180] FIG. 87A-F shows the status of ctDNA before surgery, on day 30 after surgery and during adjuvant chemotherapy (ACT). FIG. 87A shows preoperative detection of ctDNA. FIG. 87B shows the relapse rate. FIG. 87C shows Kaplan-Meier TTR estimates for 94 stage I-III patients stratified by postoperative day 30 status and ctDNA status. FIG. 87D shows the effect of ACT on ctDNA-positive patients, assessed by relapse rates and ctDNA status over time. FIG. 87E shows the relapse rate stratified by ctDNA status at the first visit after chemotherapy. FIG. 87F shows Kaplan-Meier TTR estimates for 58 patients treated with ACT, stratified by ctDNA status at the first visit after ACT.

[181] FIG. 88 demonstrates preoperative detection of carcinoembryonic antigen (CEA) in 125 patients with stage I-III CRC.

[182] FIG. 89 shows a schematic overview of ctDNA profiling results from plasma samples included in the day 30 ctDNA analysis, ordered by relapse status and disease stage. Patients marked with a sign(s) have synchronous CRC. Plasma marked ** is positive only in the second pool (n=1).

[183] FIG. 90A-B shows a schematic overview of ctDNA profiling results for a subset of plasma samples included in the Day 30 ctDNA analysis and treated with AChT, ordered by relapse status and disease stage. Patients marked with a sign(s) have synchronous CRC.

[184] FIG. 91 shows a schematic overview of ctDNA profiling results from plasma samples included in sequential post-ACT ctDNA analysis, ordered by relapse status, postoperative ctDNA status, and length of follow-up. Patients marked with a sign(s) have synchronous CRC (n=2). Plasma samples marked with ** are positive only in the second pool (n=1).

[185] FIG. 92 shows a schematic overview of CEA profiling results of plasma samples included in sequential ctDNA analysis after ACT, ordered by relapse status, postoperative ctDNA status, and duration of follow-up. Patients marked with letter(s) a have synchronous CRC (n=2). Plasma marked ** is positive only in the second pool (n=1).

[186] FIG. 93A-D: Graphs showing the relationship between ctDNA status and relapse after curative treatment. FIG. 93A shows the relapse rate stratified by ctDNA status over time. FIG. 93B shows Kaplan-Meier TTR estimates for 75 patients with time-course samples, stratified by time-course ctDNA status. FIG. 93C shows a graph comparing time to radiological recurrence and ctDNA recurrence. FIG. 93D shows that the variant allele frequency (VA) of plasma ctDNA increases in the direction of radiological relapse. Early periods before and during AHT are omitted.

[187] FIG. 94: Schematic overview of the results of ctDNA profiling of plasma samples over time in patients with and without relapses. Patients with one positive plasma sample during follow-up are considered positive.

[188] FIG. 95: Schematic overview of CEA profiling results for serum samples over time in patients with and without relapse. Patients with one positive plasma sample during follow-up are considered positive.

[189] FIG. 96: Graph comparing time to radiological recurrence and CEA recurrence.

[190] FIG. 97A-C: Identification of a significant mutation in relapsed patients. Figure 97A shows the percentage of ctDNA+ relapses in patients with significant mutations detected during follow-up. First ctDNA+ sample (left column) and all ctDNA+ plasma samples (right column). FIG. 97B shows significant variants recognized in blood. The correlation between mean blood NFA values calculated using Signatera ctDNA+ assays and variant allele frequencies (VAFs) of significant mutations is plotted on a logarithmic scale along the horizontal and vertical axes. FIG. 97C shows serial ctDNA profiling of two representative relapsed patients with significant mutations.

[191] FIG. 98: Schematic comparison of current standard of medical practice and potential postoperative patient management determined by ctDNA.

[192] FIG. 99: Graph showing the decrease in ctDNA during adjuvant chemotherapy (ACT).

[193] FIG. 100A-B: Graphs showing the relationship between ctDNA status and relapse after radical treatment. FIG. 100A shows relapse rates stratified by ctDNA status over time and Kaplan-Meier TTR estimates for 58 patients with over time samples stratified by ctDNA analysis over time. FIG. 100B shows relapse rates stratified by CEA analysis and Kaplan-Meier TTR estimates for 58 patients with over-time samples stratified by CEA analysis.

[194] The above figures are presented by way of representation and not limitation.

DETAILED DESCRIPTION OF THE INVENTION

[195] The methods and compositions provided herein improve the detection, diagnosis, staging, screening, treatment, and management of cancer (eg, breast cancer, bladder cancer, or colorectal cancer). The methods provided herein, in illustrative embodiments, analyze single nucleotide variant (SNV) mutations in circulating fluids, especially circulating tumor DNA. The methods provide the advantage of identifying more tumor-detectable mutations and clonal as well as subclonal mutations in a single test rather than multiple tests that would be required, if at all, to be effective when using tumor samples. The methods and compositions may be useful on their own, or they may be useful when used in conjunction with other methods for the detection, diagnosis, staging, screening, treatment and management of cancer (for example, breast cancer, bladder cancer or colorectal cancer), for example, to help support the results of these other methods in order to provide greater confidence and/or a certain result.

[196] Accordingly, herein, in one embodiment, a method is provided for determining single nucleotide variants present in cancer (e.g., breast cancer, bladder cancer, or colorectal cancer) by determining single nucleotide variants present in a ctDNA sample from an individual, e.g. a person who has or is suspected of having cancer (eg, breast cancer, bladder cancer, or colorectal cancer) using the ctDNA CNV amplification/sequencing workflow presented in this document.

[197] The terms "cancer" and "cancerous" refer to or describe a physiological condition in animals that is typically characterized by unregulated cell growth. "Tumor" includes one or more cancer cells. There are several main types of cancer. Carcinoma is a cancer that begins on the skin or in the tissues that line or cover internal organs. Sarcoma is a cancer that begins in bone, cartilage, fat, muscle, blood vessels, or other connective or supporting tissues. Leukemia is a cancer that begins in blood-forming tissue, such as bone marrow, and causes large numbers of abnormal blood cells to form and enter the bloodstream. Lymphoma and multiple myeloma are cancers that begin in the cells of the immune system. Central nervous system cancer is cancer that begins in the tissues of the brain and spinal cord.

[198] In some embodiments, the cancer includes acute lymphoblastic leukemia; acute myeloid leukemia; adrenocortical carcinoma; AIDS-related cancer; AIDS-related lymphoma; anal cancer; appendix cancer; astrocytomas; atypical teratoid/rhabdoid tumor; basal cell carcinoma; bladder cancer; brainstem glioma; brain tumor (including brainstem glioma, atypical teratoid/rhabdoid tumor of the central nervous system, embryonal tumors of the central nervous system, astrocytomas, craniopharyngioma, ependymoblastoma, ependymoma, medulloblastoma, medulloepithelioma, pineal parenchymal tumors of intermediate differentiation, supratentorial undifferentiated neuroectodermal tumors and pineoblastoma); breast cancer; bronchial tumors; Burkitt's lymphoma; cancer of unknown primary location; carcinoid tumor; carcinoma of unknown primary location; atypical teratoid/rhabdoid tumor of the central nervous system; fetal tumors of the central nervous system; cervical cancer; cancer in children; chordoma; chronic lymphocytic leukemia; chronic myelogenous leukemia; chronic myeloproliferative disorders; colon cancer; colorectal cancer; craniopharyngioma; cutaneous T-cell lymphoma; tumors of islet cells of the endocrine system of the pancreas; endometrial cancer; ependymoblastoma; ependymoma; esophageal carcinoma; esthesioneuroblastoma; Ewing's sarcoma; extracranial germ cell tumor; extragonadal germ cell tumor; extrahepatic bile duct cancer; gallbladder cancer; gastrointestinal (stomach) cancer; gastrointestinal carcinoid tumor; gastrointestinal stromal cell tumor; gastrointestinal stromal tumor (GIST); gestational trophoblastic tumor; glioma; hairy cell leukemia; head and neck cancer; heart cancer; Hodgkin's lymphoma; hypopharyngeal cancer; intraocular melanoma; islet cell tumors; Kaposi's sarcoma; kidney cancer; Langerhans cell histiocytosis; laryngeal cancer; lip cancer; liver cancer; malignant fibrous histiocytoma, bone cancer; medulloblastoma; medulloepithelioma; melanoma; Merkel cell carcinoma; Merkel cell skin cancer; mesothelioma; metastatic squamous cell carcinoma of the neck with a primary cancer of unknown origin; oral cancer; multiple endocrine neoplasia syndromes; multiple myeloma; multiple myeloma/plasmocyte neoplasm; mycosis fungoides; myelodysplastic syndromes; myeloproliferative neoplasms; nasal cavity cancer; nasopharyngeal cancer; neuroblastoma; non-Hodgkin's lymphoma; non-melanoma skin cancer; non-small cell lung cancer; oral cancer; oral cancer; oropharyngeal cancer; osteosarcoma; other tumors of the brain and spinal cord; ovarian cancer; epithelial ovarian cancer; germ cell tumor of the ovaries; borderline ovarian tumor; pancreas cancer; papillomatosis; sinus cancer; parathyroid cancer; pelvic cancer; penile cancer; pharynx cancer; pineal parenchymal tumors of intermediate differentiation; pineoblastoma; pituitary tumor; plasma cell neoplasm/multiple myeloma; pleuropulmonary blastoma; primary lymphoma of the central nervous system (CNS); primary hepatocellular liver cancer; prostate cancer; rectal cancer; kidney cancer; renal cell (kidney) cancer; renal cell carcinoma; respiratory tract cancer; retinoblastoma; rhabdomyosarcoma; salivary gland cancer; Sézary syndrome; small cell lung cancer; small bowel cancer; soft tissue sarcoma; squamous cell carcinoma; squamous cell carcinoma of the neck; gastrointestinal cancer (stomach cancer); supratentorial undifferentiated neuroectodermal tumors; T-cell lymphoma; testicular cancer; throat cancer; thymus carcinoma; thymoma; thyroid cancer; transitional cell carcinoma; transitional cell carcinoma of the renal pelvis and ureter; trophoblastic tumor; ureteral cancer; urethral cancer; uterine cancer; sarcoma of the uterus; vaginal cancer; vulvar cancer; Waldenström's macroglobulinemia; or Wilma's tumor.

[199] In another embodiment, provided herein is a method for detecting cancer (eg, breast cancer, bladder cancer, or colorectal cancer) in a sample of blood or a fraction thereof from an individual, such as an individual suspected of having cancer. This involves identifying single nucleotide variants present in a sample by identifying single nucleotide variants present in a ctDNA sample using the ctDNA CNV amplification/sequencing workflow presented herein. Presence of 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14 or 15 NDC at the lower end of the range, as well as 2, 3, 4, 5 6, 7, 8 , 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 30, 40, or 50 SNPs at the upper end of the range in the sample on multiple single nucleotides loci indicates the presence of cancer (for example, breast cancer, bladder cancer or colorectal cancer).

[200] In another embodiment, provided herein is a method for detecting a clonal single nucleotide variant in an individual's tumor (eg, breast cancer, bladder cancer, or colorectal cancer). The method includes performing a CNV ctDNA amplification/sequencing workflow as provided herein and determining the variant allele frequency for each of the CNV loci based on multiple copy sequencing of the amplicon series. A higher relative allele frequency compared to other single nucleotide variants at multiple single nucleotide variant loci is indicative of a clonal single nucleotide variant in the tumor. Variant allele frequencies are well known in the sequencing field. Support for this embodiment is provided, for example, in FIG. 12-14.

[201] In some embodiments, the method further includes determining a plan for treatment, therapy, and/or administration to an individual of a compound that targets one or more clonal single nucleotide variants. In some examples, subclonal and/or other clonal ONVs are not a target for therapy. Specific treatments and associated mutations are presented elsewhere herein and are known in the art. Accordingly, in certain examples, the method further comprises administering a compound to an individual where the compound is known to be particularly effective for treating cancer (eg, breast cancer, bladder cancer, or colorectal cancer) having one or more of certain single nucleotide variants.

[202] In certain aspects of this embodiment, a variant allele frequency of greater than 0.25, 0.5, 0.75, 1.0, 5, or 10% is indicative of a clonal single nucleotide variant. These cut points are supported by the data in tabular form in FIG. 20A-B.

[203] In certain examples of this embodiment, the cancer is stage 1a, 1b, or 2a breast cancer, bladder cancer, or colorectal cancer. In some examples of this embodiment, the cancer is stage 1a or 1b breast cancer, bladder cancer, or colorectal cancer. In certain examples of this embodiment, the individual does not undergo surgery. In certain examples of this embodiment, the individual is not biopsied.

[204] In some examples of this embodiment, clonal ONV is identified, or is further identified if other tests, such as direct tumor testing, suggest that the ONV in the test is clonal ONV, for any ONV in the test that has a variable allele frequency greater than at least one quarter, one third, one half or three quarters of other specified single nucleotide variants.

[205] In some embodiments, the methods described herein for detecting CNVs in ctDNA can be used instead of directly analyzing DNA from a tumor. The results presented herein show that NNVs that are more likely to be clonal NNVs have higher NVAs (see, for example, FIGS. 12-14).

[206] In certain examples of any of the embodiments of the method presented herein, before targeted amplification is performed on ctDNA from an individual, data are provided on the CNVs that were detected in the tumor from the individual. Accordingly, in these embodiments, the CNV amplification/sequencing reaction is performed on one or more tumor samples from an individual. In these methods, the CNV ctDNA amplification/sequencing reaction presented herein is still advantageous as it provides a liquid biopsy of clonal and subclonal mutations. In addition, as provided herein, clonal mutations can be more clearly identified in an individual who has cancer (eg, breast cancer, bladder cancer, or colorectal cancer) if a high percentage of NVA is detected for NNV, for example, more than 1 , 2, 3, 4, 5, 6, 7, 8, 9, 10% NVA in a ctDNA sample from an individual.

[207] In a certain embodiment, the method provided herein can be used to determine whether ctDNA from circulating free nucleic acids from an individual with cancer (eg, breast cancer, bladder cancer, or colorectal cancer) should be isolated and analyzed. First, it is determined whether the cancer is breast cancer, bladder cancer, or colorectal cancer. If the cancer is breast cancer, bladder cancer or colorectal cancer, circulating free nucleic acids are isolated from the individual. The method, in some examples, further includes determining the stage of the cancer.

[208] Some methods of the present invention provide compositions of the invention and/or solid carriers. A composition comprising circulating tumor nucleic acid fragments containing a universal adapter, wherein the circulating tumor nucleic acid fragments are derived from breast cancer, bladder cancer or colorectal cancer.

[209] In some embodiments, provided herein is a composition of the invention that includes tumor circulating nucleic acid fragments comprising a universal adapter, wherein the tumor circulating nucleic acid is derived from a blood sample or fraction thereof from an individual with cancer (e.g., breast cancer). glands, bladder cancer or colorectal cancer). These methods typically involve the formation of a ctDNA fragment that includes a universal adapter. In addition, such methods typically involve the formation of a solid support, especially a solid support for high throughput sequencing, that includes multiple clonal populations of nucleic acids, where the clonal populations contain amplicons derived from a sample of circulating free nucleic acids where the ctDNA is present. In illustrative embodiments of the invention, based on the unexpected results presented herein, the ctDNA is derived from cancer (eg, breast cancer, bladder cancer, or colorectal cancer).

[210] Similarly, an embodiment of the invention herein provides a solid support containing a plurality of clonal populations of nucleic acids, wherein the clonal populations comprise nucleic acid fragments derived from a sample of circulating free nucleic acids from a blood sample, or a fraction thereof, from an individual with cancer (such as breast cancer, bladder cancer or colorectal cancer).

[211] In some embodiments, nucleic acid fragments in different clonal populations contain the same universal adapter. Such a composition is typically generated during a high throughput sequencing reaction in the methods of the present invention.

[212] Clonal populations of nucleic acids can be obtained from nucleic acid fragments from a set of samples from two or more individuals. In these embodiments, the nucleic acid fragments comprise one of a number of molecular barcodes corresponding to a sample in the sample set.

[213] Detailed analytical methods are presented in this document in the analytical section as Method 1 RNV and Method 2 RNV. Any of the methods presented herein may further include the analytical steps presented herein. Accordingly, in some examples, methods for determining whether a single nucleotide variant is present in a sample include identifying a confidence value for each allele definition at each of the sets of single nucleotide variation loci, which may be based at least in part on the read depth for the loci. The confidence limit may be set to at least 75%, 80%, 85%, 90%, 95%, 96%, 97%, 98%, or 99%. The confidence limit for different types of mutations can be set at different levels.

[214] The method can be performed with a read depth for multiple single nucleotide variation loci of at least 5, 10, 15, 20, 25, 50, 100, 150, 200, 250, 500, 1000, 10000, 25000, 50000, 100000, 250,000, 500,000 or 1 million.

[215] In some embodiments, a method of any of the embodiments presented herein includes determining the efficiency, and/or determining the error rate for a cycle for each amplification reaction of a multiplex reaction for amplifying single nucleotide variant loci. The efficiency and error rate can then be used to determine whether a single nucleotide variant from a set of single nucleotide variant loci is present in a sample. Also, some embodiments may include more detailed analytical steps, as presented in Method 2 of the HB, presented in the analytical method.

[216] In illustrative embodiments of any of the methods presented herein, the set of single nucleotide variation loci includes all single nucleotide variation loci identified in the TCGA and COSMIC data sets for cancer (eg, breast cancer, bladder cancer, or colorectal cancer).

[217] In some embodiments of any of the methods described herein, the set of single nucleotide variant loci includes 2, 3, 4, 5, 6, 7, 8, 9, 10, 15, 20, 25, 30, 40, 50, 75 ,100, 250, 500, 1000, 2500, 5000, or 10000 single nucleotide variation loci known to be associated with cancer (eg, breast cancer, bladder cancer, or colorectal cancer) at the low end of the range, and,5. 6, 7, 8, 9, 10, 15, 20, 25, 30, 40, 50, 75, 100, 250, 500, 1000, 2500, 5000, 10000, 20000 and 25000 at the high end of the range.

[218] Any of the CNV detection methods presented herein that include a CNV ctDNA amplification/sequencing workflow can utilize improved amplification parameters for multiplex PCR. For example, where the amplification reaction is a PCR reaction and the annealing temperature is 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 °C greater than the melting temperature at the lower end of the range and 2, 3, 4 5, 6, 7, 8, 9, 10, 11, 12, 13, 14 or 15° at the high end of the range, for at least 10, 20, 25, 30, 40, 50, 60, 70, 75, 80, 90, 95 or 100% primers from a primer set.

[219] In some embodiments in which the amplification reaction is a PCR reaction, the duration of the annealing step in the PCR reaction is from 10, 15, 20, 30, 45, and up to 60 minutes at the lower end of the range, and 15 to 20, 30, 45 60, 120, 180 or 240 minutes at the top end of the range. In some embodiments, the concentration of the primer in an amplification such as a PCR reaction is from 1 nM to 10 nM. Additionally, in exemplary embodiments, the primers in the primer set are designed to minimize primer dimer formation.

[220] Accordingly, in any of the methods presented herein that include an amplification step, the amplification reaction is a PCR reaction whose annealing temperature is 1-10 °C higher than the melting temperature of at least 90% of the primers in the primer set, The duration of the annealing step in the PCR reaction is from 15 to 60 minutes, the concentration of primers in the amplification reaction is from 1 to 10 nM, and the primers in the primer set are designed to minimize the formation of primer dimer. In a further aspect of this example, the multiplex amplification reaction is carried out under limiting primer conditions.

[221] In another embodiment, provided herein is a method of supporting a diagnosis of cancer (e.g., breast cancer, bladder cancer, or colorectal cancer) for an individual, such as an individual who is suspected of having cancer (e.g., breast cancer, bladder cancer). bladder or colorectal cancer), from a blood sample or fraction thereof from an individual, which includes performing a ctDNA CNV amplification/sequencing workflow as provided herein to determine the presence of one or more single nucleotide variants at multiple single nucleotide variant loci. In this embodiment, the following elements, statements, guidelines, or rules apply: the absence of a single nucleotide variant supports a diagnosis of stage 1a, 1b, or 2a adenocarcinoma, the presence of a single nucleotide variant supports a diagnosis of squamous cell carcinoma or stage 2b or 3a adenocarcinoma, and/or the presence of ten or more single nucleotide variants confirms diagnosis of squamous cell carcinoma or adenocarcinoma stage 2b or 3.

[222] These results identify the analysis using the CNV ctDNA amplification/sequencing workflow of adenocarcinoma and small cell lung cancer samples from an individual as a valuable method for identifying CNVs found in an adenocarcinoma tumor, especially for stage 2b and 3a adenocarcinoma tumors, and especially small cell tumors cancer at any stage (see, for example, FIG. 15 and FIG. 20A-B).

[223] In some embodiments, the methods presented herein for detecting CNVs can be used to guide a therapy regimen. There are therapies available and in development that target specific mutations associated with adenocarcinoma and small cell carcinoma (Nature Review Cancer. 14:535–551 (2014). For example, detection of an EGFR mutation at L858R or T790M may be informative for choice of therapy. Erlotinib, gefitinib, afatinib, AZK9291, CO-1686 and HM61713 are current treatments approved in the US or in clinical trials that target specific EGFR mutations. In another example, the G12D, G12C or G12V mutation in KRAS may be used to Referring a subject to therapy with a combination of selumetinib plus docetaxel As another example, the V600E mutation in BRAF can be used to refer a subject to treatment with vemurafenib, dabrafenib and trametinib.

[224] The sample analyzed by the methods of the present invention, in some illustrative embodiments, is a blood sample or a fraction thereof. The methods presented herein are, in some embodiments, specifically adapted to amplify DNA fragments, especially tumor DNA fragments that are found in circulating tumor DNA (ctDNA). Such fragments are typically approximately 160 nucleotides in length.

[225] It is known in the art that cell-free nucleic acid (scNA), such as DNA, can be released into the bloodstream through various forms of cell death, such as apoptosis, necrosis, autophagy and necroptosis. The scDNA is fragmented, and the fragment size distribution varies from 150-350 bp. up to more than 10,000 bp (See Kalnina et al. World J Gastroenterol. 2015 Nov 7; 21(41): 11636-11653). For example, the size distribution of plasma DNA fragments in patients with hepatocellular carcinoma (HCC) spanned the range of 100–220 bp. with a peak counting frequency of approximately 166 bp. and the highest concentration of tumor DNA in fragments 150-180 bp long. (See: Jiang et al. Proc Natl Acad Sci USA 112:E1317-E1325).

[226] In an exemplary embodiment, circulating tumor DNA (ctDNA) is isolated from blood using an EDTA-2Na tube after removal of cellular debris and platelets by centrifugation. Plasma samples can be stored at −80°C until DNA is extracted using, for example, the QIAamp DNA Mini Kit (Qiagen, Hilden, Germany), (e.g. Hamakawa et al., Br J Cancer. 2015; 112:352-356). Hamakawa et al. reported a median extracted cell-free DNA concentration across all samples of 43.1 ng per mL of plasma (range 9.5–1338 ng/mL) and a mutant fraction range of 0.001–77.8% with a median of 0.90%.

[227] In some illustrative embodiments, the sample is a tumor. Methods are known in the art for isolating nucleic acid from a tumor and creating a nucleic acid library from such a DNA sample following the teachings herein. Additionally, taking into account the teachings herein, one skilled in the art will understand how, in addition to ctDNA samples, to create a nucleic acid library suitable for the methods presented herein from other samples, such as other liquid samples , where DNA floats freely,

[228] The methods of the present invention, in some embodiments, typically include the step of obtaining and amplifying a library of nucleic acids from a sample (ie, library preparation). Nucleic acids from a sample during the library preparation step may have ligation adapters, often referred to as library tags or ligation tags (LT), which are added where the ligation adapters contain a universal start sequence followed by universal amplification. In one embodiment, this can be done using a standard protocol designed to generate sequencing libraries after fragmentation. In one embodiment, the DNA sample may be blunt ended and then an A may be added at the 3' end. A Y adapter may be added and ligated to a T overhang. In some cases, sticky ends other than the A or T sticky end may be used. In some cases, other adapters, such as loop ligation adapters, can be added. In some cases, claimed adapters may be labeled for PCR amplification.

[229] A number of embodiments presented herein involve detecting CNVs in a ctDNA sample. Such methods, in illustrative embodiments, include an amplification step and a sequencing step (sometimes referred to herein as a “ctDNA CNF amplification/sequencing workflow”). In an illustrative example, a ctDNA amplification/sequencing workflow may include generating a set of amplicons by performing a multiplex amplification reaction on nucleic acids isolated from a blood sample or fraction thereof from an individual, such as an individual suspected of having cancer, such as breast cancer, bladder cancer or colorectal cancer, wherein each amplicon of the set of amplicons spans at least one single nucleotide variant locus of the set of single nucleotide variant loci, such as HB loci known to be associated with cancer (e.g., breast cancer, bladder cancer bladder or colorectal cancer); and determining the sequence of at least a segment in each amplicon of the set of amplicons, where the segment contains single nucleotide variant loci. Thus, this exemplary method identifies single nucleotide variants present in a sample.

[230] Exemplary workflows for ctDNA NCB amplification/sequencing may involve in more detail the formation of an amplification reaction mixture by combining polymerase, nucleotide triphosphates, nucleic acid fragments from a nucleic acid library generated from the sample, and a set of primers, each of which binds an effective distance from the loci single nucleotide variant or a set of primer pairs, each of which covers an effective region including the locus of the single nucleotide variant. Single nucleotide variant loci in exemplary embodiments are known to be associated with cancer, such as breast cancer, bladder cancer, or colorectal cancer. Subjecting the amplification reaction mixture to amplification conditions then generates a set of amplicons comprising at least one single nucleotide variant locus from a set of single nucleotide variant loci known to be preferentially associated with cancer (e.g., breast cancer, bladder cancer, or colorectal cancer ); and determining the sequence of at least a segment of each amplicon from the set of amplicons, where the segment contains single nucleotide variant loci.

[231] The effective binding distance of primers can be in the range of 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 20, 25, 30, 35, 40 , 45, 50, 75, 100, 125 or 150 base pairs of the ONV locus. The effective range that the primer pair covers usually includes the HB and is typically 160 base pairs or less, and may be 150, 140, 130, 125, 100, 75, 50 or 25 base pairs or less. In other embodiments, the effective range that the primer pair covers is 20, 25, 30, 40, 50, 60, 70, 75, 100, 110, 120, 125, 130, 140, or 150 nucleotides from the HB loci at the lower end of the range and 25, 30, 40, 50, 60, 70, 75, 100, 110, 120, 125, 130, 140 or 150, 160, 170, 175 or 200 at the higher end of the range.

[232] Additional details regarding amplification methods that can be used in the process of amplification/sequencing of CNV ctDNA to detect CNV for use in the methods of the invention are presented elsewhere in this specification.

[233] NVG recognition analytics

[234] The methods presented herein generate nucleic acid sequencing data for amplicons generated by multiplex tiling PCR. There are algorithm development tools that can be used and/or adapted to analyze this data to determine, within certain limits of confidence, whether a mutation such as ONV is present in a target gene.

[235] Sequencing reads can be demultiplexed using a proprietary instrument and displayed using Burrows-Wheeler software, Bwa mem function (BWA, Burrows-Wheeler Alignment Software (see Li H. and Durbin R. (2010) Fast and accurate long -read alignment with Burrows-Wheeler Transform. Bioinformatics, Epub. [PMID: 20080505]) in single-user mode using hg19 genome reads using the PEAR program. QA of amplification statistics can be performed by analyzing the total number of reads, the number of mapped reads, the number of mapped reads on the target and the number of reads counted.

[236] In some embodiments, any analytical method for detecting CNVs from nucleic acid sequencing data can be used with the methods of the present invention that include the step of detecting CNVs or determining the presence of CNVs. Some exemplary embodiments of the invention employ methods of the invention that utilize the CNV METHOD 1 below. In other, even more illustrative embodiments, the methods of the invention that include the step of detecting HNV or determining whether HNV is present at HNV loci use HNV METHOD 2 below.

[237] METHOD 1 ONV: For this embodiment, a background error model is built using normal plasma samples that were sequenced in the same sequencing run to account for run-specific artifacts. In some embodiments, 5, 10, 15, 20, 25, 30, 40, 50, 100, 150, 200, 250, or more than 250 normal plasma samples are analyzed in the same sequencing run. In some illustrative embodiments, 20, 25, 40, or 50 normal plasma samples are analyzed in the same sequencing run. Noise-saturated positions with a normal median variant allele frequency greater than the cutoff value are removed. For example, this cutoff value in some embodiments is greater than 0.1, 0.2, 0.25, 0.5, 1, 2, 5, or 10%. In some illustrative embodiments, interference-saturated positions with a normal medial variant allele frequency greater than 0.5% are removed. Outliers were iteratively removed from the model to account for noise and contamination. In some embodiments, samples with a Z score greater than 5, 6, 7, 8, 9, or 10 are removed from the data analysis. For each base substitution of each genomic locus, a weighted depth mean and error standard deviation are calculated. Positions of tumor or cell-free plasma samples with at least 5 read variants and a Z-score of 10, for example, against a background error, can be called a candidate mutation.

[238] METHOD 2 SNV: For this embodiment, single nucleotide variants (SNVs) are determined using plasma ctDNA data. The PCR process is modeled as a stochastic process, estimating the parameters using the training set and performing the final recognition of the CNV requires a separate testing set. The propagation of the error over multiple PCR cycles is determined, the mean and variance of the background error are calculated, and in illustrative embodiments, the background error is different from the actual mutations.

[239] For each basis, the following parameters are assessed:

[240] p = efficiency (probability that each read is replicated in each cycle)

[241] p _e = error rate per cycle for mutations of type e (probability that an error of type e will occur)

[242] X ₀ = initial number of molecules

[243] As reads are replicated, more errors occur during the PCR process. Therefore, the error profile of the reads is determined by the degrees of separation from the original read. We call a read generation k ^th when it has gone through k replications until it is generated.

[244] Let's define the following variables for each base:

[245] X _ij = number of generations i of reads generated in PCR cycle j

[246] Y _ij = total number of generations of i reads at the end of cycle j

[247] X _ij ^e = number of generations of i reads with mutation e generated in a PCR cycle j

_[ 248] Moreover, in addition to normal X ₀ molecules, if there are additional fe _X 0 molecules with mutation e at the beginning of the PCR process (hence fe/(1+fe) will be the fraction of mutant molecules in the original mixture).

[249] Given the total number of i-1 read generations in cycle j-1, the number of i-read generations generated in cycle j has a binomial distribution with sample size Y _i-1,j-1 and probability parameter p . Therefore, E( X _ij, |Y _i-1,j-1 , p ) = p Y _i-1,j-1 and Var( X _ij, |Y _i-1,j-1 , p ) = p( 1-p) Y _i-1,j-1 .

[250] We also have . Therefore, using recursion, simulation or similar methods, we can determine E( X _ij, ). Similarly, using the distribution of p , we can define Var( X _ij ) = E(Var( X _ij, | p )) + Var(E( X _ij, | p )).

[251] finally, E( X _ij ^e |Y _i-1,j-1 , p _e ) = p _e Y _i-1,j-1 and Var( X _ij ^e |Y _i-1,j-1 , p ) = p _e (1- p _e ) Y _i-1,j-1 , and we can use this to calculate E( X _ij ^e ) and Var( X _ij ^e ).

[252] In some embodiments, Method 2 NVG is performed as follows:

[253] a) Estimate PCR efficiency and error rate per cycle using the training data set;

[254] b) Estimate the number of seed molecules for the training data set at each base using the efficiency distribution estimated in step (a);

[255] c) If necessary, update the efficiency estimate for the training data set using the initial number of molecules estimated in step (b);

[256] d) Estimate the mean and variance for the total number of molecules, background error molecules, and real mutation molecules (for a search space consisting of the initial percentage of real mutation molecules) using the training set data and parameters estimated in steps (a). (b) and (c);

[257] e) Select a distribution based on the number of molecules with a common error (background error and real mutation) in all molecules and calculate the probability for each percentage of real mutations in the search space; And

[258] f) Determine the most likely percentage of real mutations and calculate the confidence using the data from step (e).

[259] A confidence limit can be used to identify ONVs at ONV loci. For example, a 90 percent, 95 percent, 96 percent, 97 percent, 98 percent, or 99 percent confidence limit could be used to recognize an NDC.

[260] Approximate algorithm for METHOD 2 ONV

[261] The algorithm begins by estimating the efficiency and error rate per cycle using the training set. Let n denote the total number of PCR cycles.

[262] The number of reads R _b at each base b can be approximated by (1+p _b ) ⁿ X ₀ , where p _b is the efficiency per base b. Then (R _b /X ₀ ) ^1/n can be used to approximate 1+p _b . We can then determine the mean and standard deviation p _b across all training samples to estimate the parameters of a probability distribution (such as a normal, beta, or similar distribution) for each base.

[263] Similarly, the number of reads in error e R _b ^e at each base b can be estimated as . After determining the mean and standard deviation of the error rate across all training samples, we approximate its probability distribution (e.g., normal, beta, or similar distribution) whose parameters are estimated using this mean and standard deviation value.

[264] Then, for the training data, we estimate the original seed copy at each base as , where f(.) is the estimated distribution from the training set.

[265] , where f(.) is the estimated distribution from the training set.

[266] Consequently, we estimated the parameters to be used in the stochastic process. Using these estimates, we can then estimate the mean and variance of the molecules created in each cycle (note that we do this separately for normal molecules, error molecules, and mutant molecules).

[267] Finally, using a probabilistic method (such as maximum likelihood or similar methods), we can determine the best value of f _e that best fits the distribution of faulty, mutant, and normal molecules. More specifically, we estimate the expected ratio of error molecules to total molecules for different values of f _e in the final reads and determine the probability of our data for each of these values, and then select the value with the highest probability.

[268] Primer tails can improve the detection of fragmented DNA from universally labeled libraries. If the library tag and primer tails contain homologous sequence, hybridization may be improved (e.g., melting temperature (Tm) is reduced) and primers may be extended if only part of the primer target sequence is in a DNA fragment of the sample. In some embodiments, 13 may be used. or more target base pairs. In some embodiments, 10 to 12 target base pairs may be used. In some embodiments, 8 to 9 target base pairs may be used. In some embodiments, 6 to 7 target base pairs may be used.

[269] In one embodiment, libraries are generated from the above samples by ligating adapters to the ends of DNA fragments in the samples or to the ends of DNA fragments obtained from DNA isolated from the samples. The fragments can then be amplified using PCR, for example according to the following example protocol:

[270] 95°C, 2 min; 15 x [95°C, 20 sec, 55°C, 20 sec, 68°C, 20 sec], 68°C 2 min, hold at 4°C.

[271] There are many kits and methods known in the art for generating nucleic acid libraries that include universal primer binding sites for subsequent amplification, such as clonal amplification, and for subsequence sequencing. To facilitate adapter ligation, library preparation and amplification may involve end repair and adenylation (ie, A-tailing). The practical methods presented herein may benefit from kits specifically adapted for preparing libraries from small nucleic acid fragments, especially circulating free DNA. For example, NEXTflex Cell Free kits available from Bioo Scientific () or Natera Library Prep Kit (available from Natera, Inc. San Carlos, CA). However, such kits are typically modified to include adapters that are customized for the amplification and sequencing steps of the methods provided herein. Ligation of the adapter can be accomplished using commercially available kits, such as the ligation kit provided in the AGILENT SURESELECT kit (Agilent, CA).

[272] Target regions of a nucleic acid library generated from DNA isolated from a sample, in particular a circulating free DNA sample for the methods of the present invention, are then amplified. For this amplification, a series of primers or primer pairs that may include 5, 10, 15, 20, 25, 50, 100, 125, 150, 250, 500, 1000, 2500, 5000, 10000, 20000, 25000 or 50000 The lower end of the range and 15, 20, 25, 50, 100, 125, 150, 250, 500, 1000, 2500, 5000, 10,000, 20,000, 25000, 50000, 60000, 75000 or 100,000 primmers at the upper end of the range, each of which binds to one of a number of primer binding sites.

[273] Primer designs can be generated with Primer3 (Untergrasser A, Cutcutache I, Koressaar T, Ye J, Faircloth BC, Remm M, Rozen SG (2012) “Primer3 - new capabilities and interfaces.” Nucleic Acids Research 40(15) :e115 and Koressaar T, Remm M (2007) “Enhancements and modifications of primer design program Primer3.” Bioinformatics 23(10):1289-91 (source code available at primer3.sourceforge.net). Primer specificity can be assessed using BLAST and added to existing primer design criteria:

[274] Primer specificity can be determined using the BLASTn program from the ncbi-blast-2.2.29+ package. The “blastn-short” job option can be used to map primers to the human hg19 genome. Primer designs can be defined as “specific” if the primer has fewer than 100 matches to the genome and the top match is the target binding region of the complementary primer of the genome and is at least two points higher than the other matches (score determined by BLASTn). This may be done in order to have a unique hit in the genome and not have many other hits throughout the genome.

[275] The final selected primers can be visualized in IGV browsers (James T. Robinson, Helga Thorvaldsdóttir, Wendy Winckler, Mitchell Guttman, Eric S. Lander, Gad Getz, Jill P. Mesirov. Integrative Genomics Viewer. Nature Biotechnology 29, 24-26 (2011)) and UCSC (Kent WJ, Sugnet CW, Furey TS, Roskin KM, Pringle TH, Zahler AM, Haussler D. The human genome browser at UCSC. Genome Res. 2002 Jun;12(6):996-1006) , using reference files and coverage maps for validation.

[276] The methods of the present invention, in some embodiments, include forming an amplification reaction mixture. The reaction mixture is usually formed by combining the polymerase, nucleotide triphosphates, nucleic acid fragments from a library of nucleic acids generated from the sample, and a set of forward and reverse primers specific for the target regions containing NCBs. The reaction mixtures presented herein themselves form a separate aspect of the invention in the illustrative embodiments.

[277] The amplification reaction mixture used for the present invention includes components known in the art for nucleic acid amplification, especially PCR amplification. For example, the reaction mixture typically includes nucleotide triphosphates, polymerase and magnesium. Polymerases that are useful for the present invention may include any polymerase that can be used in an amplification reaction, especially those useful in PCR reactions. In some embodiments, hot-start Taq polymerases are particularly useful. Amplification reaction mixtures useful for performing the methods presented herein, such as AmpliTaq Gold master mix (Life Technologies, Carlsbad, CA), are commercially available.

[278] Amplification conditions (eg, temperature cycling) for PCR are well known in the art. The methods provided herein may include any PCR cycling conditions that result in the amplification of target nucleic acids, such as target nucleic acids, from a library. Non-limiting exemplary cycling conditions are provided in the Examples section of this document.

[279] There are many workflows that are possible when performing PCR; Some workflows typical of the methods disclosed herein are presented in this document. The steps presented herein are not intended to exclude other possible steps and do not imply that any of the steps presented herein are required for proper operation of the method. There are a large number of parameter variations or other modifications known in the literature that can be made without affecting the spirit of the invention.

[280] In some embodiments of the method presented herein, at least a portion thereof, and in illustrative examples, the entire sequence of an amplicon, such as a target amplicon for an external primer, is determined. Methods for determining the sequence of an amplicon are known in the art. To determine such a sequence, any of the sequencing methods known in the art, for example, Sanger sequencing, can be used. In illustrative embodiments, high-throughput next-generation sequencing techniques (also referred to herein as massively parallel sequencing techniques), such as those used in MYSEQ (ILLUMINA), can be used to sequence amplicons generated by the methods presented herein. HISEQ (ILLUMINA), ION TORRENT (LIFE TECHNOLOGIES), GENOME ANALYZER ILX (ILLUMINA), GS FLEX+ (ROCHE 454), but not limited to them.

[281] High-throughput genetic sequencers are suitable for using barcodes (that is, tagging samples with distinctive nucleic acid sequences) to identify specific samples from individuals, allowing multiple samples to be analyzed simultaneously in a single DNA sequencer run. The number of times the sequence of a given genomic region in a library preparation (or other nucleic acid preparation of interest) is sequenced (number of reads) will be proportional to the number of copies of that sequence in the genome of interest (or expression level in the case of preparations containing cDNA). In such quantification, variations in amplification efficiency can be taken into account.

[282] Target genes

[283] The target genes of the present invention are, in exemplary embodiments, cancer-associated genes, and in many exemplary embodiments, are cancer-associated genes. A cancer-associated gene (eg, cancer-associated gene, or bladder cancer-associated gene, or colorectal cancer-associated gene) refers to a gene associated with an altered risk of developing cancer (eg, breast cancer, bladder cancer, bladder or colorectal cancer) or altered cancer prognosis. Typical cancer-associated genes that contribute to cancer development include oncogenes; genes that enhance cell proliferation, invasion or metastasis; genes that inhibit apoptosis; and genes that stimulate angiogenesis. Cancer-associated genes that inhibit cancer include, but are not limited to, tumor suppressor genes; genes that inhibit cell proliferation, invasion or metastasis; genes promoting apoptosis; and antiangiogenesis genes.

[284] An embodiment of a method for detecting a mutation begins with selecting a region of a gene that becomes a target. The region with known mutations is used to design mPCR-SNP primers for amplification and mutation detection.

[285] The methods presented herein can be used to detect virtually any type of mutation, especially mutations known to be associated with cancer, and most specifically, the methods presented herein are directed at mutations, especially CNVs. associated with cancer, in particular breast cancer, bladder cancer or colorectal cancer. Exemplary ONVs may be in one or more of the following genes: EGFR, FGFR1, FGFR2, ALK, MET, ROS1, NTRK1, RET, HER2, DDR2, PDGFRA, KRAS, NF1, BRAF, PIK3CA, MEK1, NOTCH1, MLL2, EZH2, TET2, DNMT3A, SOX2, MYC, KEAP1, CDKN2A, NRG1, TP53, LKB1 and PTEN, which were identified in various lung cancer samples as mutated, having increased copy numbers, or fused with other genes and their combinations (Non-small-cell lung cancers: a heterogeneous set of diseases. Chen et al. Nat. Rev. Cancer. 2014 Aug 14(8):535-551). In another example, the gene list is the above genes in which NNVs have been reported, such as those listed in Chen et al.

[286] Reaction mixtures for amplification (for example, PCR ) :

[287] The methods of the present invention, in some embodiments, include forming an amplification reaction mixture. The reaction mixture is typically formed by combining polymerase, nucleotide triphosphates, nucleic acid fragments from a nucleic acid library generated from the sample, a series of target-specific outer forward primers, and an outer first-strand universal reverse primer. Another exemplary embodiment is a reaction mixture that includes forward target-specific internal primers in place of forward target-specific external primers and amplicons from a first PCR reaction using external primers in place of nucleic acid fragments from a nucleic acid library. The reaction mixtures presented herein themselves form a separate aspect of the invention in the illustrative embodiments. In illustrative embodiments, the reaction mixtures are PCR reaction mixtures. PCR reaction mixtures typically include magnesium.

[288] In some embodiments, the reaction mixture includes ethylenediaminetetraacetic acid (EDTA), magnesium, tetramethylammonium chloride (TMAC), or any combination thereof. In some embodiments, the concentration of TMAC is from 20 to 70 mM, inclusive. Although not intended to be tied to any particular theory, TMAC is thought to bind DNA, stabilize duplexes, increase primer specificity, and/or equalize the melting temperatures of different primers. In some embodiments, TMAC improves the consistency of the amount of amplified products for different purposes. In some embodiments, the concentration of magnesium (eg, magnesium from magnesium chloride) is from 1 to 8 mM.

[289] A large number of primers used for multiplex PCR with a large number of targets can chelate a lot of magnesium (2 phosphates in the primers chelate 1 magnesium). For example, if enough primers are used such that the phosphate concentration in the primers is approximately 9 mM, then the primers can reduce the effective magnesium concentration by approximately 4.5 mM. In some embodiments, EDTA is used to reduce the amount of magnesium available as a cofactor for the polymerase, since high concentrations of magnesium can lead to PCR errors such as amplification of off-target loci. In some embodiments, the EDTA concentration reduces the amount of available magnesium to 1-5 mM (eg, 3-5 mM).

[290] In some embodiments, the pH is from 7.5 to 8.5, such as from 7.5 to 8, from 8 to 8.3, or from 8.3 to 8.5, inclusive. In some embodiments, Tris is used, for example, at a concentration of 10 to 100 mM, such as 10 to 25 mM, 25 to 50 mM, 50 to 75 mM, or 25 to 75 mM, inclusive. In some embodiments, any of these concentrations of Tris is used at a pH of 7.5 to 8.5. In some embodiments, a combination of KCl and (NH ₄ ) ₂ SO ₄ is used, for example, 50 to 150 mM KCl and 10 to 90 mM (NH ₄ ) ₂ SO ₄ , inclusive. In some embodiments, the KCl concentration is from 0 to 30 mM, from 50 to 100 mM, or from 100 to 150 mM, inclusive. In some embodiments, the concentration of (NH ₄ ) ₂ SO ₄ is 10 to 50 mM, 50 to 90 mM, 10 to 20 mM, 20 to 40 mM, 40 to 60 mM, or 60 to 80 mM (NH ₄ ) ₂ SO ₄ inclusive. In some embodiments, the ammonium concentration [NH ₄ ⁺ ] is from 0 to 160 mM, such as from 0 to 50, from 50 to 100, or from 100 to 160 mM, inclusive. In some embodiments, the sum of the potassium and ammonium concentration ([K ⁺ ] + [NH ₄ ⁺ ]) is 0 to 160 mM, such as 0 to 25, 25 to 50, 50 to 150, 50 to 75, 75 to 100, 100 to 125, or 125 to 160 mM inclusive. An exemplary buffer with [K ⁺ ] + [NH ₄ ⁺ ] = 120 mM is 20 mM KCl and 50 mM (NH ₄ ) ₂ SO _4. In some embodiments, the buffer contains from 25 to 75 mM Tris, pH from 7.2 to 8, from 0 to 50 mM KCl, from 10 to 80 mM ammonium sulfate and from 3 to 6 mM magnesium, inclusive. In some embodiments, the buffer comprises 25 to 75 mM Tris, pH 7 to 8.5, 3 to 6 mM MgCl ₂ , 10 to 50 mM KCl, and 20 to 80 mM (NH ₄ ) ₂ SO ₄ , inclusive. In some embodiments, 100 to 200 units/ml of polymerase is used. Some embodiments use 100 mM KCl, 50 mM (NH ₄ ) ₂ SO ₄ , 3 mM MgCl ₂ , 7.5 nM of each primer in the library, 50 mM TMAC, and 7 μL template DNA in a 20 μL final volume at pH 8 ,1.

[291] In some embodiments, a crowding agent such as polyethylene glycol (PEG, such as PEG 8000) or glycerol is used. In some embodiments, the amount of PEG (such as PEG 8000) is 0.1 to 20%, such as 0.5 to 15%, 1 to 10%, 2 to 8%, or 4 to 8%, inclusive. In some embodiments, the amount of glycerin is from 0.1 to 20%, such as from 0.5 to 15%, from 1 to 10%, from 2 to 8%, or from 4 to 8%, inclusive. In some embodiments, the crowding agent allows the use of a lower polymerase concentration and/or shorter annealing time. In some embodiments, the crowding agent improves depth of read (DOR) uniformity and/or reduces attrition (undetected alleles). Polymerases In some embodiments, a polymerase with proofreading activity, a polymerase with no or little proofreading activity, or a mixture of a polymerase with proofreading activity and a polymerase with no or little proofreading activity is used. In some embodiments, a hot start polymerase, a non-hot start polymerase, or a mixture of a hot start polymerase and a non-hot start polymerase is used. In some embodiments, HotStarTaq DNA polymerase is used (see, for example, QIAGEN Catalog No. 203203). In some embodiments, AmpliTaq Gold® DNA polymerase is used. In some embodiments, PrimeSTAR GXL DNA polymerase is used, a high fidelity polymerase that provides efficient PCR amplification when excess template is present in the reaction mixture, and is also used in the amplification of long products (Takara Clontech, Mountain View, Calif.). In some embodiments, KAPA Taq DNA polymerase or KAPA TaqHotStart DNA polymerase is used; they are based on the wild-type Taq single-subunit DNA polymerase of the thermophilic bacterium Thermus aquaticus . DNA polymerases KAPA Taq and KAPA TaqHotStart have 5'-3' polymerase activity and 5'-3' exonuclease activity, but not 3'-5' exonuclease (proofreading) activity (see, for example, KAPA BIOSYSTEMS catalog No. BK1000). In some embodiments, Pfu DNA polymerase is used; it is a highly thermostable DNA polymerase from the hyperthermophilic archaeon Pyrococcus furiosus . This enzyme catalyzes the template-dependent polymerization of nucleotides into duplex DNA in the 5'→3' direction. Pfu DNA polymerases also exhibit 3'→5' exonuclease (proofreading) activity, which allows the polymerase to correct nucleotide incorporation errors. It does not have 5'→3' exonuclease activity (see, for example, Thermo Scientific catalog no. EP0501). In some embodiments, Klentaq1 is used; it is an analogue of Taq DNA polymerase in the Klenow fragment; it does not have exonuclease or endonuclease activity (see, for example, the catalog of DNA POLYMERASE TECHNOLOGY, Inc, St. Louis, Missouri, No. 100). In some embodiments, the polymerase is a PHUSION DNA polymerase, such as PHUSION High Fidelity DNA polymerase (M0530S, New England BioLabs, Inc.) or PHUSION Hot Start Flex DNA polymerase (M0535S, New England BioLabs, Inc.). In some embodiments, the polymerase is a Q5® DNA polymerase, such as Q5® High-Fidelity DNA Polymerase (M0491S, New England BioLabs, Inc.) or Q5® Hot Start High-Fidelity DNA Polymerase (M0493S, New England BioLabs, Inc. ). In some embodiments, the polymerase is T4 DNA polymerase (M0203S, New England BioLabs, Inc.).

[292] In some embodiments, 5 to 600 units/ml (units per 1 ml of reaction volume) of polymerase is used, e.g., 5 to 100, 100 to 200, 200 to 300, 300 to 400, 400 to 500 or from 500 to 600 units/ml inclusive.

PCR methods

[293] In some embodiments, hot start PCR is used to reduce or prevent polymerization before thermal cycling of the PCR. Typical hot-start PCR methods involve initially inhibiting the DNA polymerase reaction or physically separating the reaction components until the reaction mixture reaches higher temperatures. In some embodiments, slow release of magnesium is used. DNA polymerase requires the activity of magnesium ions, so magnesium is chemically separated from the reaction by binding to a chemical compound and released into solution only at high temperature. In some embodiments, non-covalent binding of the inhibitor is used. In this method, a peptide, antibody or aptamer is non-covalently bound to an enzyme at a low temperature and inhibits its activity. After incubation at elevated temperature, the inhibitor is released and the reaction begins. Some studies use cold-sensitive Taq polymerase, such as a modified DNA polymerase that shows little or no activity at low temperatures. In some embodiments, chemical modification is used. In this method, the molecule is covalently linked to an amino acid side chain in the active site of DNA polymerase. The molecule is released from the enzyme by incubating the reaction mixture at elevated temperature. Once the molecule is released, the enzyme is activated.

[294] In some embodiments, the amount of template nucleic acids (such as an RNA or DNA sample) is 20 to 5000 ng, such as 20 to 200, 200 to 400, 400 to 600, 600 to 1000; From 1000 to 1500; or from 2000 to 3000 ng inclusive.

[295] In some embodiments, the QIAGEN Multiplex PCR Kit (QIAGEN catalog no. 206143) is used. For 100 x 50 µl multiplex PCR reactions, the kit includes 2x QIAGEN Multiplex PCR Master Mix (giving a final concentration of 3 mM MgCl ₂ , 3 x 0.85 ml), 5x Q-Solution (1 x 2.0 ml), and free water from RNase (RNase-Free Water) (2 x 1.7 ml). QIAGEN Multiplex PCR Master Mix (MM) contains a combination of KCl and (NH ₄ ) ₂ SO ₄ , as well as the PCR auxiliary Factor MP, which increases the local concentration of primers on the matrix. MP factor stabilizes specifically bound primers, allowing efficient DNA primer extension by HotStarTaq DNA Polymerase. HotStarTaq DNA polymerase is a modified form of Taq DNA polymerase and does not have polymerase activity at room temperatures. In some embodiments, HotStarTaq DNA polymerase is activated by a 15-minute incubation at 95°C, which can be included in any existing thermal cycling program.

[296] In some embodiments, a final concentration of 1x QIAGEN MM (recommended concentration), 7.5 nM of each primer in the library, 50 mM TMAC, and 7 μl of template DNA in a 20 μl final volume is used. In some embodiments, PCR thermal cycling conditions include 95°C for 10 minutes (hot start); 20 cycles at 96 °C for 30 seconds; 65 °C for 15 minutes; and 72 °C for 30 seconds; then 72°C for 2 minutes (final extension); and then maintain at 4 °C.

[297] Some embodiments use a final concentration of 2x QIAGEN MM (twice the recommended concentration), 2 nM of each primer in the library, 70 mM TMAC, and 7 μl of DNA template in a 20 μl final volume. In some embodiments, up to 4 mM EDTA is also included. In some embodiments, PCR thermal cycling conditions include 95°C for 10 minutes (hot start); 25 cycles at 96 °C for 30 seconds; 65 °C for 20, 25, 30, 45, 60, 120 or 180 minutes; and optionally 72°C for 30 seconds; then 72°C for 2 minutes (final extension); and then maintain at 4 °C.

[298] Another exemplary set of conditions includes a semi-nested PCR approach. The first PCR reaction uses a 20 µl reaction volume with a final concentration of 2x QIAGEN MM, 1.875 nM of each primer in the library (external forward and reverse primers) and DNA template. Thermal cycling parameters include 95°C for 10 minutes; 25 cycles: 96°C for 30 seconds, 65°C for 1 minute, 58°C for 6 minutes, 60°C for 8 minutes, 65°C for 4 minutes and 72°C for 30 seconds ; and then 72°C for 2 minutes, followed by holding at 4°C. Then 2 μl of the resulting product, diluted 1:200, is used as starting material for the second PCR reaction. This reaction uses a 10 µl reaction volume with a final concentration of 1x QIAGEN MM, 20 nM each internal forward primer and 1 µM reverse primer tag. Thermal cycling parameters include 95°C for 10 minutes; 15 cycles: 95°C for 30 seconds, 65°C for 1 minute, 60°C for 5 minutes, 65°C for 5 minutes and 72°C for 30 seconds; and then 72°C for 2 minutes, followed by holding at 4°C. The annealing temperature may not necessarily be higher than the melting temperature of some or all of the primers, as discussed herein (see US Patent Application No. 14/918,544, filed October 20, 2015, which is incorporated herein by reference in its entirety).

[299] Melting temperature (T _m ) is the temperature at which half (50%) of the DNA duplex of an oligonucleotide (such as a primer) and its ideal complement dissociates and becomes single-stranded DNA. Annealing temperature (T _A ) is the temperature at which the PCR protocol is performed. In previous methods, it is usually 5 °C below the lowest T _m of the primers used, so that almost all possible duplexes are formed nearby (such that essentially all primer molecules bind the template nucleic acid). Although this is very effective, more nonspecific reactions occur at lower temperatures. One consequence of too low a T _A is that primers may anneal to sequences other than the true target because internal single-base mismatches or partial annealing may occur. In some embodiments of the present invention, T _A is higher than T _m where, at a given time, only a small portion of the targets have an annealed primer (eg, only about 1-5%). If they elongate, they are removed from the equilibrium of the annealing and dissociating primers and the target (since elongation rapidly increases T _m to temperatures above 70 °C), and approximately 1-5% of the targets have primers new. Thus, by allowing the annealing reaction to take a long time, it is possible to obtain approximately 100% of the targets copied per cycle.

[300] In various embodiments, the annealing temperature is 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13°C and 2, 3, 4, 5, 6, 7. 8, 9, 10 11, 12, 13 or 15°C at the upper end of the range above the melting point (e.g. empirically measured or calculated _Tm ) by at least 25, 50, 60, 70, 75, 80, 90 95 or 100% non-identical primers. In various embodiments, the annealing temperature ranges from 1 to 15 °C (e.g., 1 to 10, 1 to 5, 1 to 3, 3 to 5, 5 to 10, 5 to 8, 8 to 10 , from 10 to 12 or from 12 to 15 °C inclusive) greater than the melting point (for example, empirically measured or calculated _Tm ) of at least 25; 50; 75; 100; 300; 500; 750; 1000; 2000; 5000; 7500; 10000; 15000; 19000; 20000; 25000; 27000; 28000; 30000; 40000; 50000; 75000; 100000; or all non-identical primers. In various embodiments, the annealing temperature is from 1 to 15° C. (e.g., 1 to 10, 1 to 5, 1 to 3, 3 to 5, 3 to 8, 5 to 10, 5 to 8 , 8 to 10, 10 to 12, or 12 to 15 °C inclusive) greater than the melting point (e.g., empirically measured or calculated _Tm ) of at least 25, 50, 60, 70, 75, 80, 90, 95% or all non-identical primers, and the duration of the annealing step (per PCR cycle) is from 5 to 180 minutes, for example from 15 to 120 minutes, from 15 to 60 minutes, from 15 to 45 minutes or from 20 to 60 minutes inclusive.

[301] Exemplary Multiplex PCR Methods

[302] Various embodiments use long annealing times (as discussed herein and exemplified in Example 10) and/or low primer concentrations. In fact, in some embodiments, limiting primer concentrations and/or limiting conditions are used. In various embodiments, the duration of the annealing step is 15, 20, 25, 30, 35, 40, 45 or 60 minutes at the low end of the range and 20, 25, 30, 35, 40, 45, 60, 120 or 180 minutes at the high end range. In various embodiments, the duration of the annealing step (per PCR cycle) ranges from 30 to 180 minutes. For example, the annealing step can be from 30 to 60 minutes, and the concentration of each primer can be less than 20, 15, 10 or 5 nM. In other embodiments, the primer concentration is 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 15, 20, or 25 nM at the low end of the range and 2, 3, 4, 5, 6, 7, 8 , 9, 10, 15, 20, 25 and 50 at the higher end of the range.

[303] At high levels of multiplexing, the solution may become viscous due to the large number of primers in the solution. If the solution is too viscous, the primer concentration can be reduced to an amount that is still sufficient for the primers to bind template DNA. In various embodiments, from 1000 to 100,000 different primers are used, and the concentration of each primer is less than 20 nM, such as less than 10 nM, or from 1 to 10 nM inclusive.

[304] Detection of copy number variation (CNV)

[305] In addition to ONV and insertions/deletions, the methods presented herein for monitoring and detecting early relapses and metastasis may also benefit from detection of ICH.

[306] In one aspect, the present invention generally relates, at least in part, to improved methods for determining the presence or absence of copy number variations, such as deletions or duplications of chromosome segments or entire chromosomes. The methods are particularly useful for detecting small deletions or duplications, which may be difficult to detect with high specificity and sensitivity using previous methods due to the small amount of data available from the corresponding chromosome segment. These methods include improved analytical methods, improved bioassay methods, and combinations of improved analytical methods and bioassay methods. The methods of the invention can also be used to detect deletions or duplications that are present in only a small percentage of the cells or nucleic acid molecules tested. This makes it possible to detect deletions or duplications before the onset of disease (for example, at a precancerous stage) or in the early stages of disease, for example, before a large number of affected cells (such as cancer cells) with deletion or duplication accumulate. More accurate detection of deletions or duplications associated with a disease or disorder allows for improved methods of diagnosing, predicting, preventing, delaying, stabilizing or treating the disease or disorder. Some deletions or duplications are known to be associated with cancer or severe mental or physical disabilities.

[307] In another aspect, the present invention generally relates, at least in part, to improved methods for detecting single nucleotide variants (SNVs). These improved methods include improved analytical methods, improved bioassay methods, and improved methods that use a combination of improved analytical methods and bioassay methods. Methods in some illustrative embodiments are used to detect, diagnose, monitor, or stage cancer, for example, in samples where OCB is present in very low concentrations, for example, less than 10, 5, 4, 3, 2.5, 2, 1, 0 .5, 0.25, or 0.1% relative to the total number of normal copies of the HNV locus, such as circulating free DNA samples. That is, these methods, in certain illustrative embodiments, are particularly well suited for samples where there is a relatively low percentage of mutation or variant relative to the normal polymorphic alleles present for that genetic locus. Finally, this document presents methods that combine improved methods for detecting copy number variations with improved methods for detecting single nucleotide variations.

[308] Successful treatment of a disease such as cancer often depends on early diagnosis, correct staging of the disease, selection of an effective therapeutic regimen, and careful monitoring to prevent or detect relapse. For diagnosing cancer, the most reliable method is often considered to be histological evaluation of tumor material obtained from tissue biopsy. However, the invasive nature of biopsy-based sampling has made it impractical for mass screening and routine surveillance. Therefore, the present methods have the advantage that they can be performed non-invasively, if desired, at relatively low cost and fast turnaround time. Targeted sequencing, which can be used by the methods of the invention, requires fewer reads than fractional sequencing, for example, several million reads instead of 40 million reads, which reduces cost. Multiplex PCR and next generation sequencing that can be used increase throughput and reduce costs.

[309] In some exemplary embodiments, analysis of AAI patterns in ctDNA provides a more detailed understanding of the clonal architecture of tumors to help predict their therapeutic responses and optimize treatment strategies. Therefore, in some embodiments, mmPCR-SNP panels are selected that are designed for clinically significant ICH and NVC. Such panels, in some illustrative embodiments, are particularly useful for patients with cancers where ICHs represent a significant proportion of the mutational load, as is often the case in breast, ovarian, and lung cancers.

[310] In some embodiments, methods are used to detect a deletion, duplication, or single nucleotide variant in a person. A sample from an individual may be analyzed that contains cells or nucleic acids suspected of having a deletion, duplication, or single nucleotide variant. In some embodiments, the sample is from a tissue or organ suspected of having a deletion, duplication, or single nucleotide variant, such as a cell or lesion suspected of being cancerous. The methods of the invention can be used to detect a deletion, duplication or single nucleotide variant that is present in only one cell or a small number of cells in a mixture containing cells with the deletion, duplication or single nucleotide variant and cells without the deletion, duplication or single nucleotide variant. In some embodiments, scDNA or scRNA from a blood sample of an individual is analyzed. In some embodiments, the scDNA or scRNA is secreted from cells, such as cancer cells. In some embodiments, the scDNA or scRNA is released by cells undergoing necrosis or apoptosis, such as cancer cells. The methods of the invention can be used to detect deletions, duplications, or single nucleotide variants that are present in only a small percentage of scDNA or scRNA. In some embodiments, one or more embryonic cells are tested.

[311] In addition to determining the presence or absence of a copy number change, one or more other factors can be analyzed if necessary. These factors can be used to improve diagnostic accuracy (eg, determining the presence or absence of cancer or increased risk of cancer, cancer classification, or cancer staging) or prognosis. These factors can also be used to select a particular therapy or treatment regimen that is likely to be effective in a subject. Examples of factors include the presence or absence of polymorphisms or mutations; altered (increased or decreased) levels of total or specific scDNA, scRNA, microRNA (miRNA); changed (increased or decreased) tumor fraction; altered (increased or decreased) methylation levels, altered (increased or decreased) DNA integrity, altered (increased or decreased) or alternative splicing of mRNA.

[312] The following sections describe methods for detecting deletions or duplications using phased data (such as inferred or measured phased data) or unphased data; samples that can be tested; methods of sample preparation, amplification and quantification; methods for phasing genetic data; polymorphisms, mutations, nucleic acid changes, changes in mRNA splicing and changes in nucleic acid levels that may be detected; databases with results of methods, other risk factors and screening methods; cancer that can be diagnosed or treated; cancer treatment; cancer models for testing treatments; and ways to formulate and prescribe treatments.

[313] Example methods for determining ploidy using phased data

[314] Some of the methods of the invention are based in part on the discovery that using phased data to detect ICH reduces the rate of false negatives and false positives compared to using non-phased data. This improvement is greatest for samples with ICH present at low levels. Thus, phased data improves the accuracy of ICH detection compared to the use of non-phased data (such as methods that calculate allele ratios at one or more loci or cumulative allele ratios to obtain an aggregated value (such as an average) across a chromosome or chromosomal segment without considering whether the ratios of alleles at different loci indicate that the same or different haplotypes are present in abnormal numbers). Using phased data allows us to more accurately determine whether differences between measured and expected allele ratios are due to noise or the presence of ICH. For example, if differences between measured and expected allele ratios at most or all loci in an area indicate that the same haplotype is overrepresented, then ICH is more likely to be present. Using the relationship between alleles in a haplotype allows one to determine whether the measured genetic data corresponds to the same haplotype that is represented by an overrepresentation (rather than random noise). In contrast, if the differences between measured and expected allele ratios are due only to noise (such as experimental error), then in some embodiments, about half the time the first haplotype is represented as being present in excessive amounts, and about the second half of the time the second haplotype is represented as being present in excessive amounts. large quantities.

[315] In some embodiments, phased genetic data is used to determine whether there is an excess number of copies of a first homologous chromosome segment relative to a second homologous chromosome segment in an individual's genome (eg, in the genome of one or more cells or in scDNA or scRNA). Typical over-presentations include duplication of the first homologous chromosome segment or deletion of the second homologous chromosome segment. In some embodiments, there is no overrepresentation because the first and homologous chromosome segments are present in equal proportions (eg, one copy of each segment in a diploid sample). In some embodiments, the calculated allele ratios in the nucleic acid sample are compared with the expected allele ratios to determine whether overrepresentation exists, as described further below. As used herein, the phrase “a first homologous chromosome segment versus a second homologous chromosome segment” means a first homologous chromosome segment and a second homologous chromosome segment.

[316] In some embodiments, the method includes obtaining phased genetic data for a first homologous chromosome segment, including the identity of an allele present at that locus on the first homologous chromosome segment for each locus in a set of polymorphic loci on the first homologous chromosome segment, obtaining phased genetic data for a second homologous chromosomal segment containing the identity of the allele present at that locus on the second homologous chromosomal segment for each locus in the set of polymorphic loci on the second homologous chromosomal segment, and obtaining measured genetic data of the alleles containing for each of the alleles at each of the loci in the set of polymorphic loci the number of each allele present in a sample of DNA or RNA from one or more target cells and one or more non-target cells from an individual. In some embodiments, the method includes listing a set of one or more hypotheses indicating the degree of overrepresentation of a first homologous chromosome segment; calculating, for each of the hypotheses, expected genetic data for a plurality of loci in the sample from the obtained phased genetic data for one or more possible ratios of DNA or RNA from one or more target cells to total DNA or RNA in the sample; calculating (eg, computer calculation) for each possible DNA or RNA ratio and for each hypothesis whether the data from the sample's obtained genetic data for that possible DNA or RNA ratio and for that hypothesis correspond to the expected genetic data for the sample; ranking one or more hypotheses according to the fit of the data; and selecting the hypothesis that has the highest ranking, thereby determining the degree to which the copy number of the first homologous chromosome segment is overrepresented in the genome of one or more cells from an individual.

[317] In some embodiments, the method includes obtaining phased genetic data using any of the methods presented herein or any known method. In some embodiments, the method includes, simultaneously or sequentially in any order, (i) obtaining phased genetic data for a first homologous chromosome segment, including the identity of an allele present at that locus on the first homologous chromosome segment, for each locus in a set of polymorphic loci on the first homologous segment chromosome, (ii) obtaining phased genetic data for a second homologous chromosomal segment containing the identity of the allele present at that locus on the second homologous chromosomal segment for each locus in the set of polymorphic loci on the second homologous chromosomal segment, and (iii) obtaining measured genetic data of the allele , including the amount of each allele at each of the loci in a set of polymorphic loci in a DNA sample from one or more cells of an individual.

[318] In some embodiments, the method includes calculating allele ratios for one or more loci in a set of polymorphic loci that are heterozygous in at least one cell from which the sample was obtained. In some embodiments, the calculated allele ratio for a particular locus is the measured amount of one of the alleles divided by the total measured amount of all alleles for the locus. In some embodiments, the calculated allele ratio for a particular locus is the measured amount of one of the alleles (e.g., an allele on a first segment of a homologous chromosome) divided by the measured amount of one or more other alleles (e.g., an allele on a second segment of a homologous chromosome) for the locus. The calculated allele ratios may be calculated using any of the methods presented herein or any standard method (such as any mathematical transformation of the calculated allele ratios presented herein).

[319] In some embodiments, the method includes determining whether an excessive number of copies of a first homologous chromosome segment is present by comparing one or more calculated allele ratios for a locus with the allele ratio that would be expected for that locus if the first and second homologous chromosome segments were present in equal proportions. In some embodiments, the expected allele ratio assumes that the possible alleles for a locus have an equal probability of being present. In some embodiments, in which the calculated allele ratio for a particular locus is the measured amount of one of the alleles divided by the total measured amount of all alleles for the locus, the corresponding expected allele ratio is 0.5 for a biallelic locus or 1/3 for a triallelic locus. In some embodiments, the expected allele ratio is the same for all loci, such as 0.5 for all loci. In some embodiments, the expected allele ratio assumes that the possible alleles for a locus may have a different probability of presence, such as a probability based on the frequency of each of the alleles in a particular population to which the subject belongs, such as a population based on the subject's ancestry. Such allele frequencies are publicly available (see, for example , HapMap Project; Perlegen Human Haplotype Project; web at ncbi.nlm.nih.gov/projects/SNP/; Sherry ST, Ward MH, Kholodov M, et al. dbSDNP: the NCBI database of genetic variation. Nucleic Acids Res. 2001 Jan 1;29(1):308-11, each of which is incorporated by reference in its entirety). In some embodiments, the expected allele ratio is the allele ratio that is expected for a particular individual tested for a particular hypothesis determining the degree of overrepresentation of the first homologous chromosome segment. For example, the expected allele ratio for a particular individual can be determined based on phased or unphased genetic data from the individual (for example, from a sample from an individual that is unlikely to have a deletion or duplication, such as a non-cancerous sample) or data from one or more relatives of the individual .

[320] In some embodiments, a calculated allele ratio indicates copy number overrepresentation of the first segment of a homologous chromosome if either (i) the allele ratio for the measured number of alleles present at that locus on the first homologous chromosome divided by the total measured number of all alleles for locus is greater than the expected allele ratio for that locus, or (ii) the allele ratio for the measured number of alleles present at that locus on the second homologous chromosome divided by the total measured number of all alleles for the locus is less than the expected allele ratio for that locus . In some embodiments, the calculated allele ratio is considered an indicator of overrepresentation only if it is significantly higher or lower than the expected ratio for that locus. In some embodiments, the calculated allele ratio indicates that there is no copy number overrepresentation of the first homologous chromosome segment if either (i) the allele ratio for the measured number of alleles present at that locus on the first homologous chromosome divided by the total measured number of all alleles for the locus, is less than or equal to the expected allele ratio for that locus, or (ii) the allele ratio for the measured number of alleles present at that locus on the second homologous chromosome, divided by the total measured number of all alleles for the locus, is greater than or equal to the expected allele ratio for that locus. In some embodiments, calculated coefficients equal to the corresponding expected coefficient are ignored (as they indicate no overrepresentation).

[321] In various embodiments, one or more of the following methods are used to compare one or more calculated allele ratios with corresponding expected allele ratios. In some embodiments, it is determined whether the calculated allele ratio is higher or lower than the expected allele ratio for a particular locus, regardless of the magnitude of the difference. In some embodiments, the magnitude of the difference between the calculated allele ratio and the expected allele ratio for a particular locus is determined, regardless of whether the calculated allele ratio is above or below the expected allele ratio. In some embodiments, it is determined whether the calculated allele ratio is higher or lower than the expected allele ratio and the magnitude of the difference for a particular locus. In some embodiments, it is determined whether the average or weighted average of the calculated allele ratios is higher or lower than the average or weighted average of the expected allele ratios regardless of the magnitude of the difference. In some embodiments, the magnitude of the difference between the average or weighted average of the calculated allele ratios and the average or weighted average of the expected allele ratios is determined, regardless of whether the average or weighted average of the calculated allele ratio is higher or lower than the average or weighted average of the expected allele ratio. In some embodiments, it is determined whether the average or weighted average of the calculated allele ratios is higher or lower than the average or weighted average of the expected allele ratios, and the magnitude of the difference. In some embodiments, an average or weighted average of the magnitude of the difference between the calculated allele ratios and the expected allele ratios is determined.

[322] In some embodiments, the magnitude of the difference between the calculated allele ratio and the expected allele ratio for one or more loci is used to determine whether the overrepresentation of the copy number of the first homologous chromosome segment is due to a duplication of the first homologous chromosome segment or a deletion of the second homologous chromosome segment in the genome of one or more cells.

[323] In some embodiments, overrepresentation of the copy number of the first segment of a homologous chromosome is determined if one or more of the following conditions are met. In some embodiments, the number of calculated allele ratios that indicate copy number overrepresentation of the first homologous chromosome segment is greater than a threshold value. In some embodiments, the number of calculated allele ratios that indicate no copy number overrepresentation of the first homologous chromosome segment is below a threshold. In some embodiments, the magnitude of the difference between the calculated allele ratios that indicate copy number overrepresentation of the first homologous chromosome segment and the corresponding expected allele ratios is greater than a threshold. In some embodiments, for all calculated allele ratios that indicate overrepresentation, the sum of the magnitude of the differences between the calculated allele ratio and the corresponding expected allele ratio is greater than a threshold. In some embodiments, the magnitude of the difference between calculated allele ratios that indicate no copy number overrepresentation of the first homologous chromosome segment and the corresponding expected allele ratios is below a threshold. In some embodiments, the average or weighted average of the calculated allele ratios for the measured amount of an allele present on the first homologous chromosome divided by the total measured amount of all alleles for the locus is greater than the average or weighted average of the expected allele ratios by at least a threshold value. In some embodiments, the average or weighted average of the calculated allele ratios for the measured amount of an allele present on the second homologous chromosome divided by the total measured amount of all alleles for the locus is less than the average or weighted average of the expected allele ratio at least within a threshold value. In some embodiments, the data fit between the calculated allele ratios and the allele ratios that are predicted for the copy number overrepresentation of the first homologous chromosome segment is below a threshold (indicating a good fit to the data). In some embodiments, the data fit between the calculated allele ratios and the allele ratios that are predicted to not overrepresent the copy number of the first homologous chromosome segment exceeds a threshold (indicating a poor fit to the data).

[324] In some embodiments, copy number overrepresentation of the first homologous chromosome segment is determined to be absent if one or more of the following conditions are met. In some embodiments, the number of calculated allele ratios that indicate copy number overrepresentation of the first homologous chromosome segment is below a threshold. In some embodiments, the number of calculated allele ratios that indicate no copy number overrepresentation of the first homologous chromosome segment is greater than a threshold value. In some embodiments, the magnitude of the difference between the calculated allele ratios that indicate copy number overrepresentation of the first homologous chromosome segment and the corresponding expected allele ratios is below a threshold. In some embodiments, the magnitude of the difference between the calculated allele ratios that indicate non-overrepresentation of the copy number of the first homologous chromosome segment and the corresponding expected allele ratios is greater than a threshold. In some embodiments, the average or weighted average of the calculated allele ratios for the measured amount of an allele present on the first homologous chromosome divided by the total measured amount of all alleles for the locus minus the average or weighted average of the expected allele ratios is less than a threshold. In some embodiments, the average or weighted average of the expected allele ratios minus the average or weighted average of the calculated allele ratios for the measured amount of an allele present on the second homologous chromosome, divided by the total measured amount of all alleles for the locus, is less than a threshold value. In some embodiments, the agreement of the data between the calculated allele ratios and the allele ratios that are predicted for the copy number overrepresentation of the first homologous chromosome segment is greater than a threshold. In some embodiments, the agreement of the data between the calculated allele ratios and the allele ratios that are predicted to not overrepresent the copy number of the first homologous chromosome segment is below a threshold. In some embodiments, the threshold value is determined based on empirical testing of samples known to have the ICH of interest and/or samples known to be free of ICH.

[325] In some embodiments, determining whether a first homologous chromosome segment copy number is overrepresented includes listing a set of one or more hypotheses indicating the degree of overrepresentation of the first homologous chromosome segment. A rough hypothesis is that there is no overrepresentation because the first and homologous chromosome segments are present in equal proportions (e.g., one copy of each segment in a diploid sample). Other exemplary hypotheses include a first homologous chromosomal segment being duplicated one or more times (eg, 1, 2, 3, 4, 5 or more additional copies of the first homologous chromosome compared to the number of copies of the second homologous chromosomal segment). Another exemplary hypothesis involves deletion of a second homologous chromosome segment. Another tentative hypothesis is the deletion of both the first and second homologous segments of the chromosome. In some embodiments, estimated allele ratios for loci that are heterozygous in at least one cell are estimated for each hypothesis taking into account the degree of overrepresentation determined by that hypothesis. In some embodiments, the probability that a hypothesis is true is calculated by comparing the calculated allele ratios with the predicted allele ratios, and the hypothesis with the highest likelihood is selected.

[326] In some embodiments, the expected distribution of the test statistic is calculated using the predicted allele ratios for each hypothesis. In some embodiments, the probability that a hypothesis is true is calculated by comparing a test statistic that is calculated using the predicted allele ratios with the expected distribution of the test statistic that is calculated using the predicted allele ratios, and the hypothesis with the highest likelihood is selected.

[327] In some embodiments, predicted allele ratios for loci that are heterozygous in at least one cell are estimated taking into account phased genetic data for the first homologous chromosome segment, phased genetic data for the second homologous chromosome segment, and the degree of overrepresentation determined by this hypothesis. In some embodiments, the probability that a hypothesis is true is calculated by comparing the calculated allele ratios with the predicted allele ratios; and the hypothesis with the greatest likelihood is selected.

[328] Using mixed samples

[329] It is understood that for many embodiments the sample is a mixed sample with DNA or RNA from one or more target cells and one or more non-target cells. In some embodiments, target cells are cells that have an ICH, such as a deletion or duplication of interest, and non-target cells are cells that do not have a copy number change of interest (for example, a mixture of cells with a cell deletion or duplication and cells of interest without any deletions or duplications). In some embodiments, target cells are cells that are associated with a disease or disorder or an increased risk of a disease or disorder (for example, cancer cells), and non-target cells are cells that are not associated with a disease or disorder or increased risk for a disease or disorder (such as non-cancerous cells) In some embodiments, all target cells have the same ICH. In some embodiments, two or more target cells have different ICHs. In some embodiments, one or more target cells have an ICH, polymorphism, or mutation associated with a disease or disorder or an increased risk of a disease or disorder that is not found in at least one other target cell. In some such embodiments, the fraction of cells that are associated with the disease or disorder or increased risk of the disease or disorder, of the total number of cells in the sample, is assumed to be greater than or equal to the fraction of the most common of these ICHs, polymorphisms, or mutations in the sample. For example, if 6% of cells have a K-ras mutation and 8% of cells have a BRAF mutation, at least 8% of the cells are assumed to be cancerous.

[330] In some embodiments, the ratio of DNA (or RNA) from one or more target cells to total DNA (or RNA) in the sample is calculated. In some embodiments, a series of one or more hypotheses are listed that determine the degree of overrepresentation of the first homologous chromosome segment. In some embodiments, predicted allele ratios for loci that are heterozygous in at least one cell are estimated taking into account the calculated DNA or RNA ratio and the degree of overrepresentation determined by that hypothesis is estimated for each hypothesis. In some embodiments, the probability that a hypothesis is true is calculated by comparing the calculated allele ratios with the predicted allele ratios, and the hypothesis with the highest likelihood is selected.

[331] In some embodiments, the expected distribution of the test statistic calculated using the predicted allele ratios and the calculated DNA or RNA ratio is estimated for each hypothesis. In some embodiments, the probability that a hypothesis is true is determined by comparing a test statistic calculated using the predicted allele ratios and the calculated DNA or RNA ratio with the expected distribution of the test statistic calculated using the predicted allele ratios and the calculated DNA or RNA ratio, and the hypothesis with the greatest likelihood is selected.

[332] In some embodiments, the method includes enumerating a set of one or more hypotheses determining the degree of overrepresentation of the first homologous chromosome segment. In some embodiments, the method includes estimating, for each hypothesis, either (i) predicted allele ratios for loci that are heterozygous in at least one cell, taking into account the degree of overrepresentation determined by that hypothesis or (ii) for one or more possible DNA or RNA ratios - the expected distribution of a test statistic calculated using the predicted allele ratios and the possible ratio of DNA or RNA from one or more target cells to the total DNA or RNA in the sample. In some embodiments, data fit is calculated by comparing either (i) the calculated allele ratios with the predicted allele ratios or (ii) a test statistic calculated using the calculated allele ratios and the possible DNA or RNA ratio to the expected distribution of the test statistic calculated using the predicted ratios alleles and possible DNA or RNA ratio. In some embodiments, one or more hypotheses are ranked according to the data match and the hypothesis that has the highest ranking is selected. In some embodiments, a technique or algorithm, such as a search algorithm, is used to perform one or more of the following steps: calculating a match between the data, ranking the hypotheses, or selecting the hypothesis that has the highest ranking. In some embodiments, the data fit is a fit to a beta binomial distribution or a fit to a binomial distribution. In some embodiments, the technique or algorithm is selected from the group consisting of maximum likelihood estimator, maximum posterior estimator, Bayesian estimator, dynamic estimator (such as dynamic Bayesian estimator), and expectation maximization estimator. In some embodiments, the method includes applying a technique or algorithm to the obtained genetic data and the expected genetic data.

[333] In some embodiments, the method includes creating a split of possible ratios that range from a lower limit to an upper limit for the ratio of DNA or RNA from one or more target cells to the total DNA or RNA in the sample. In some embodiments, one or more hypotheses are listed that determine the degree of overrepresentation of the first homologous chromosome segment. In some embodiments, the method includes estimating, for each of the possible DNA or RNA ratios in a split and for each hypothesis, either (i) predicted allele ratios for loci that are heterozygous in at least one cell given the possible DNA or RNA ratio, and the degree the overrepresentation determined by this hypothesis, or (ii) the expected distribution of the test statistic calculated using the predicted allele ratios and the possible DNA or RNA ratio. In some embodiments, the method includes calculating, for each of the possible DNA or RNA ratios in a split and for each hypothesis, the probability that the hypothesis is true by comparing either (i) the calculated allele ratios with the predicted allele ratios, or (ii) a test statistic, calculated using the predicted allele ratios and the possible DNA or RNA ratio to the expected distribution of the test statistic, calculated using the predicted allele ratios and the possible DNA or RNA ratio. In some embodiments, the combined probability for each hypothesis is determined by combining the probabilities of that hypothesis for each of the possible relationships in the partition; and the hypothesis with the highest cumulative probability is selected. In some embodiments, the combined probability for each hypothesis is determined by weighting the probability of the hypothesis for a particular candidate relationship based on the probability that the candidate relationship is the correct relationship.

[334] In some embodiments, a technique that is selected from the group consisting of maximum likelihood estimator, maximum posterior estimator, Bayesian estimator, dynamic estimator (such as dynamic Bayes estimator), and expectation maximization estimator is used to estimate the ratio of DNA or RNA from one or more target cells to the total DNA or RNA in the sample. In some embodiments, the ratio of DNA or RNA from one or more target cells to total DNA or RNA in a sample is assumed to be the same for two or more (or all) ICHs of interest. In some embodiments, the ratio of DNA or RNA from one or more target cells to total DNA or RNA in the sample is calculated for each ICH of interest.

[335] Example uses of under-phased data

[336] It should be understood that many embodiments use imperfectly phased data. For example, for one or more loci on the first and/or second segment of a homologous chromosome, it may not be known with 100% certainty which allele is present. In some embodiments, the calculation of the probability of each hypothesis uses input data for the individual's possible haplotypes (such as haplotypes based on population-based haplotype frequencies). In some embodiments, the input data for candidate haplotypes is adjusted either by another method to phase the generation of genetic data, or by phased data from other subjects (e.g., previous subjects) to refine the population data used to phase the individual based on bioinformatics.

[337] In some embodiments, the phased genetic data comprises probabilistic data for two or more possible sets of phased genetic data, wherein each possible set of phased genetic data contains a possible identity of an allele present at each locus in a set of polymorphic loci on a first homologous chromosome segment, and a possible the identity of the allele present at each locus in a set of polymorphic loci on the second homologous segment of the chromosome. In some embodiments, a probability for at least one of the hypotheses is determined for each of the possible sets of phased genetic data. In some embodiments, a joint probability for a hypothesis is determined by combining the hypothesis probabilities for each of the possible sets of phased genetic data; and the hypothesis with the highest cumulative probability is selected.

[338] Any of the methods disclosed herein, or any known method, can be used to obtain incompletely phased data (such as using population haplotype frequencies to determine the most likely phase) for use in the claimed methods. In some embodiments, phased data is obtained by probabilistically combining haplotypes of smaller segments. For example, possible haplotypes can be determined based on possible combinations of one haplotype from a first region with another haplotype from another region from the same chromosome. The likelihood that specific haplotypes from different regions are part of the same larger block of haplotypes on the same chromosome can be determined using, for example, population haplotype frequencies and/or known recombination rates between different regions.

[339] In some embodiments, a single hypothesis rejection test is used for the null hypothesis of disomy. In some embodiments, the probability of a disomy hypothesis is calculated, and the disomy hypothesis is rejected if the probability is below a predetermined threshold (eg, less than 1 in 1000). If the null hypothesis is rejected, this may be due to errors in incompletely phased data or the presence of ICH. In some embodiments, more accurate phased data is obtained (such as phased data from any of the molecular phasing methods disclosed herein to obtain actual phased data rather than bioinformatics-derived phased data). In some embodiments, the probability of the disomy hypothesis is recalculated using more accurate phase data to determine whether the disomy hypothesis should be rejected. Rejection of this hypothesis indicates the presence of a duplication or deletion of a chromosome segment. If necessary, the false positive rate can be changed by adjusting the threshold value.

[340] Further exemplary embodiments for determining ploidy using phased data

[341] In illustrative embodiments, provided herein is a method for determining the ploidy of a chromosomal segment in a sample of an individual. The method includes the following steps: obtaining allele frequency data, including the amount of each allele present in the sample at each locus in a set of polymorphic loci on a chromosomal segment; obtaining phased allelic information for a set of polymorphic loci by estimating the phase of allele frequency data; obtaining individual allele frequency probabilities for polymorphic loci for different ploidy states using allele frequency data; obtaining joint probabilities for a set of polymorphic loci using individual probabilities and phased allelic information; and selecting, based on the joint probabilities, the best fit model indicating the ploidy of the chromosomes, thereby determining the ploidy of the chromosomal segment.

[342] As disclosed herein, allele frequency data (also referred to herein as measured genetic allele data) can be obtained by methods known in the art. For example, data can be obtained using qPCR or microarrays. In one exemplary embodiment, the data is generated using nucleic acid sequencing data, especially high throughput nucleic acid sequencing data.

[343] In certain illustrative examples, allele frequency data are corrected for errors before they are used to generate individual probabilities. In specific illustrative embodiments, the errors that are corrected include bias in allele amplification efficiency. In other embodiments, errors that are corrected include environmental contamination and genotype contamination. In some embodiments, the errors corrected include allelic amplification bias, sequencing errors, environmental contamination, and genotypic contamination.

[344] In some embodiments, individual probabilities are generated using a set of models of both different ploidy states and allelic imbalance fractions for a set of polymorphic loci. In these embodiments and other embodiments, pooled probabilities are generated taking into account the relationship between polymorphic loci on a chromosome segment.

[345] Accordingly, in one illustrative embodiment that combines some of these embodiments, provided herein is a method for detecting chromosomal ploidy in a sample of an individual, which includes the following steps: obtaining nucleic acid sequence data for alleles at a set of polymorphic loci on chromosome segment in an individual; discovery of allele frequencies in a set of loci using nucleic acid sequencing data; correcting allele amplification efficiency bias in detected allele frequencies to generate corrected allele frequencies for a set of polymorphic loci; obtaining individual allele frequency probabilities for polymorphic loci for different ploidy states by comparing phased allelic information with a set of models of different ploidy states and allelic imbalance fractions of a set of polymorphic loci; obtaining phased allelic information for a set of polymorphic loci by assessing nucleic acid sequencing data; obtaining joint probabilities for a set of polymorphic loci by combining individual probabilities taking into account the relationship between polymorphic loci on a chromosome segment; and, based on the pooled probabilities, selecting the best fit model indicating chromosomal aneuploidy.

[346] As disclosed herein, individual probabilities can be generated using a set of models or hypotheses of both different ploidy states and average fractions of allelic imbalance for a set of polymorphic loci. For example, in a particularly illustrative example, individual probabilities are generated by modeling the ploidy states of a first chromosome segment homologue and a second chromosome segment homolog. Modeled ploidy states include the following: (1) all cells have no deletion or amplification of the first homolog or second homolog of a chromosome segment; (2) at least some cells have a deletion of the first homolog or amplification of the second homologue of the chromosome segment; and (3) at least some cells have a deletion of the second homolog or amplification of the first homologue of the chromosome segment.

[347] It should be understood that the above models may also be referred to as a hypothesis, which is used to constrain the model. Thus, 3 hypotheses have been demonstrated that can be used.

[348] The simulated average allelic imbalance fractions may include any range of average allelic imbalance that includes the actual average allelic imbalance of a chromosomal segment. For example, in some illustrative embodiments, the range of average allelic imbalance that is modeled may be between 0, 0.1, 0.2, 0.25, 0.3, 0.4, 0.5, 0.6, 0. 75, 1, 2, 2.5, 3, 4 and 5% on the low end and 1, 2, 2.5, 3, 4, 5, 10, 15, 20, 25, 30, 40, 50, 60, 70, 80, 90, 95 and 99% at the high end. The intervals for range modeling can be any interval depending on the processing power used and the time allocated for analysis. For example, you can model intervals of 0.01, 0.05, 0.02, or 0.1.

[349] In certain illustrative embodiments, the sample has an average allelic imbalance for a chromosomal segment of 0.4% to 5%. In some embodiments, the average allelic imbalance is low. In these embodiments, the average allelic imbalance is typically less than 10%. In some illustrative embodiments, the allelic imbalance is between 0.25, 0.3, 0.4, 0.5, 0.6, 0.75, 1, 2, 2.5, 3, 4, and 5% at the lower end and up to 1, 2, 2.5, 3, 4 and 5% at the high end. In other exemplary embodiments, the average allelic imbalance is from 0.4, 0.45, 0.5, 0.6, 0.7, 0.8, 0.9, or 1.0% on the low end to 0.5. 0.6, 0.7, 0.8, 0.9, 1.0, 1.5, 2.0, 3.0 4.0 or 5.0% at the high end. For example, the average allelic imbalance of the sample in the illustrative example ranges from 0.45 to 2.5%. In another example, average allelic imbalance is detected with sensitivities of 0.45, 0.5, 0.6, 0.7, 0.8. 0.9 or 1.0%. That is, the test method is capable of detecting chromosomal aneuploidy up to AAI (average allelic imbalance) of 0.45, 0.5, 0.6, 0.7, 0.8. 0.9 or 1.0%. An exemplary sample with low allelic imbalance in the methods of the present invention includes plasma samples from individuals with cancer having circulating tumor DNA, or plasma samples from pregnant women having circulating fetal DNA.

[350] It should be understood that for NCB, the proportion of abnormal DNA is usually measured using mutant allele frequency (number of mutant alleles at a locus/total number of alleles at that locus). Because the difference between the abundance of two homologues in tumors is similar, we measure the proportion of abnormal DNA for ICH by the average allelic imbalance (AAI), defined as |(H1 - H2)|/(H1 + H2), where Hi is the average copy number of homolog i in sample, and Hi/(H1 + H2) is the relative abundance or ratio of homologs of homologue i. The maximum homologue ratio is the homolog ratio of the more common homolog.

[351] The assay attrition rate is the percentage of SNPs without readout, calculated using all SNPs. The single allele drop-out rate (ADO) is the percentage of SNPs in which only one allele is present, as estimated using only heterozygous SNPs. Genotype confidence can be determined by fitting a binomial distribution to the number of reads in each SNP that were B-allele reads and using the ploidy status of the focal region of the SNP to estimate the probability of each genotype.

[352] For tumor tissue samples, chromosomal aneuploidy (of which ICH is an example in this paragraph) can be delineated by transitions between allele frequency distributions. In plasma samples from cancer patients, individuals suspected of having cancer, individuals previously diagnosed with cancer, or as a cancer screen for individuals at risk or the general population, ICHs can be identified using a maximum likelihood algorithm that looks for plasma ICH in areas known to exhibit aneuploidy in cancer and/or where a tumor sample from the same individual also has ICH. In illustrative embodiments, the algorithm uses haplotype phase information of an individual whose sample is being analyzed for circulating tumor DNA to match the allele count of the measured and adjusted test sample to the expected allele count, for example, using a co-allocation mode. Such haplotype phase information can be obtained from any sample from an individual that comprises substantially or at least 60, 70, 80, 90, 95, 96, 97, 98, 99% or all of normal cellular DNA, such as restrictions, a buffy coat sample, a saliva sample, or a skin sample, from parental genotype information or de novo haplotype phasing, which can be achieved in a variety of ways (see, for example, Snyder, M., et al., Haplotype-resolved genome sequencing: experimental methods and applications. Nat Rev Genet 16 , 344-358 (2015)), such as haplotyping by dilution (Kaper, F., et al., Whole-genome haplotyping by dilution, amplification, and sequencing. Proc Natl Acad Sci USA 110 , 5552-5557 (2013)) or sequencing with long reads (Kuleshov, V. et al. Whole-genome haplotyping using long reads and statistical methods. Nat Biotech 32 , 261-266 (2014)). This algorithm can model expected allele frequencies across all allelic imbalance ratios at 0.025% intervals for three sets of hypotheses: (1) all cells are normal (no allelic imbalance), (2) some/all cells have homolog 1 deletion or homolog 2 amplification or ( 3) some/all cells have a homolog 2 deletion or a homolog 1 amplification. The likelihood of each hypothesis can be determined in each SNP using a Bayesian classifier based on a beta binomial model of expected and observed allele frequencies for all heterozygous SNPs and can then be calculated a combined probability across multiple SNPs, in some illustrative embodiments, using the association of exemplary SNP loci. In fact, in illustrative embodiments, the normal cellular haplotype phase information obtained as described above is used by the algorithm to fit the calculated and typically adjusted allele counts in the sample under study to the expected allele counts using a joint distribution model. The maximum likelihood hypothesis can then be selected.

[353] Consider a chromosomal region with an average number of copies N in a tumor, and let c denote the proportion of plasma DNA derived from a mixture of normal and tumor cells in the disomic region. AAI is calculated as:

[354] $$AAI=\frac{c|N-2|}{2+c\left(N-2\right)}$$

[355] In certain illustrative examples, allele frequency data are corrected for errors before they are used to generate individual probabilities. Various types of error correction and/or offset are disclosed herein. In specific illustrative embodiments, the errors that are corrected are biases in the efficiency of amplification of alleles. In other embodiments, errors that are corrected include sequence errors, environmental contamination, and genotypic contamination. In some embodiments, the errors corrected include allelic amplification bias, sequencing errors, environmental contamination, and genotypic contamination.

[356] It should be understood that allele amplification efficiency bias may be determined for an allele as part of an experiment or laboratory determination that includes determination on a test sample, or it may be determined at another time using a set of samples that include the allele whose efficiency is being calculated . Environmental contamination and genotype contamination are typically determined in the same run as the test sample is analyzed.

[357] In some embodiments, environmental contamination and genotype contamination are determined for homozygous alleles in the sample. It should be understood that for any given sample from an individual, some loci in the sample will be heterozygous and others will be homozygous, even if the locus is selected for analysis because it has relatively high heterozygosity in the population. In some embodiments, it is useful to determine the ploidy of a chromosomal segment using heterozygous loci for an individual, while environmental and genotype contamination can be calculated using homozygous loci.

[358] In certain illustrative examples, selection is performed by analyzing the magnitude of the difference between the phased allelic information and the estimated allelic frequencies generated for the models.

[359] In illustrative examples, individual allele frequency probabilities are generated based on a beta binomial model of expected and observed allele frequencies at a set of polymorphic loci. In illustrative examples, individual probabilities are generated using a Bayesian classifier.

[360] In certain illustrative embodiments, nucleic acid sequence data is generated by performing high-throughput DNA sequencing of multiple copies of a series of amplicons generated using a multiplex amplification reaction, wherein each amplicon of the amplicon series spans at least one polymorphic locus of a set of polymorphic loci, and where each from the polymer loci of the set is amplified. In some embodiments, the multiplex amplification reaction is conducted under limiting primer conditions for at least 1/2 of the reaction. In some embodiments, limiting primer concentrations are used in 1/10, 1/5, 1/4, 1/3, 1/2, or all of the multiplex reactions. This document presents factors that should be considered to achieve limiting primer conditions in an amplification reaction such as PCR.

[361] In some embodiments, the methods presented herein detect ploidy for multiple chromosomal segments on multiple chromosomes. Accordingly, chromosomal ploidy in these embodiments is determined for a set of chromosome segments in a sample. These embodiments require reactions with higher multiplex amplification. Accordingly, for these embodiments, the multiplex amplification reaction may comprise, for example, 2,500 to 50,000 multiplex reactions. In some embodiments, the following ranges of multiplex reactions are performed: between 100, 200, 250, 500, 1000, 2500, 5000, 10000, 20000, 25000, 50000 at the lower end of the range and between 200, 250, 500, 1000, 2500, 5000, 10,000, 20,000, 25,000, 50,000 and 100,000 at the higher end of the range.

[362] In illustrative embodiments, the set of polymorphic loci is a set of loci that are known to exhibit high heterozygosity. However, for any given individual, some of these loci are expected to be homozygous. In some illustrative embodiments, the methods of the invention use nucleic acid sequence information for both homozygous and heterozygous loci for an individual. An individual's homozygous loci are used, for example, to correct errors, while heterozygous loci are used to determine the allelic imbalance of a sample. In some embodiments, at least 10% of the polymorphic loci are heterozygous loci for the individual.

[363] As disclosed herein, preference is given to analyzing target SNP loci that are known to be heterozygous in the population. Accordingly, in some embodiments, polymorphic loci are selected where at least 10, 20, 25, 50, 75, 80, 90, 95, 99, or 100% of the polymorphic loci in the population are known to be heterozygous.

[364] As disclosed herein, in some embodiments, the sample is a plasma sample from a pregnant woman.

[365] In some examples, the method further includes performing the method on a control sample with a known average allelic imbalance ratio. The control may have an average allelic imbalance ratio for a particular allelic condition indicating aneuploidy of a chromosomal segment of 0.4 to 10% to simulate the average allelic imbalance of an allele in a sample that is present in low concentrations, such as would be expected from circulating free DNA from tumors.

[366] In some embodiments, PlasmArt controls are used as controls. Accordingly, in certain aspects, a sample is obtained by a method including fragmenting a sample of nucleic acid that is known to exhibit chromosomal aneuploidy into fragments that mimic the size of DNA fragments circulating in the plasma of an individual. In certain aspects, a control is used that does not have aneuploidy for the chromosome segment.

[367] In illustrative embodiments, data from one or more controls may be analyzed in the method along with a test sample. For example, controls may include a sample other than an individual that is not suspected of having chromosomal aneuploidy, or a sample that is suspected of containing ICH or chromosomal aneuploidy. For example, if the test sample is a plasma sample suspected of containing circulating free tumor DNA, the method may also be performed on a control sample from the subject's tumor along with the plasma sample. As disclosed herein, a control sample can be obtained by fragmenting a DNA sample known to have chromosomal aneuploidy. Such fragmentation can result in a DNA sample that mimics the DNA composition of an apoptotic cell, especially when the sample is from an individual affected by cancer. Data from the control sample will increase the confidence in detecting chromosomal aneuploidy.

[368] In some embodiments of methods for determining ploidy, the sample is a plasma sample from an individual suspected of having cancer. In these embodiments, the method further comprises determining, based on the selection, whether copy number variation is present in the individual's tumor cells. For these embodiments, the sample may be a plasma sample from an individual. For these embodiments, the method may further include determining, based on the selection, whether cancer is present in the individual.

[369] These embodiments for determining the ploidy of a chromosomal segment may further include detecting a single nucleotide variant at a single nucleotide variation site in a set of single nucleotide variation locations, where the detection of either chromosomal aneuploidy or the single nucleotide variant, or both, indicates the presence of circulating tumor nucleic acids in the sample.

[370] These embodiments may further include obtaining haplotype information about a chromosome segment for an individual's tumor and using the haplotype information to generate a set of models of various ploidy states and allelic imbalance fractions of a set of polymorphic loci.

[371] As disclosed herein, some embodiments of ploidy determination methods may further include removing outliers from the original or adjusted allele frequency data before comparing the original or adjusted allele frequency to a set of models. For example, in some embodiments, allele frequencies of loci that are at least 2 or 3 standard deviations above or below the mean for other loci on a chromosome segment are removed from the data before being used for modeling.

[372] As mentioned herein, it should be understood that many of the embodiments presented herein, including for determining the ploidy of a chromosomal segment, preferably use under or perfectly phased data. It should also be understood that this document presents a number of features that provide improvements over prior ploidy detection methods, and that many different combinations of these features can be used.

[373] In some embodiments, provided herein are computer systems and computer readable media for performing any of the methods of the present invention. These include systems and computer-readable media for performing ploidy determination methods. Accordingly, and as non-limiting examples of system embodiments, to demonstrate that any of the methods presented herein can be performed using a system and a computer readable medium using the disclosure herein, in another aspect, a system for detecting chromosomal ploidy in a sample of an individual, a system comprising: an input processor configured to receive allele frequency data including the amount of each allele present in the sample at each locus in a set of polymorphic loci on a chromosomal segment; a modeling unit configured to: generate phased allelic information for a set of polymorphic loci by estimating the phase of allele frequency data; and obtaining individual allele frequency probabilities for polymorphic loci for different ploidy states using allele frequency data; and obtaining joint probabilities for a set of polymorphic loci using individual probabilities and phased allelic information; and a hypothesis manager configured to select, based on the pooled probabilities, a best fit model indicative of chromosomal ploidy, thereby determining the ploidy of the chromosomal segment.

[374] In some embodiments of this system, the allele frequency data is data generated by a nucleic acid sequencing system. In some embodiments, the system further comprises an error correction unit configured to correct errors in the allele frequency data, wherein the corrected allele frequency data is used by the modeling unit to obtain individual probabilities. In some embodiments, the error correction unit corrects for bias in allele amplification efficiency. In some embodiments, the modeling unit generates individual probabilities using a set of models of both different ploidy states and allelic imbalance fractions for a set of polymorphic loci. The modeling block in some example embodiments generates joint probabilities by considering the association between polymorphic loci on a chromosome segment.

[375] In one illustrative embodiment, provided herein is a system for detecting chromosomal ploidy in a sample of an individual, which includes the following: an input processor configured to obtain nucleic acid sequence data for alleles at a set of polymorphic loci on a chromosome segment in the individual and determining allele frequencies at a set of loci using nucleic acid sequence data; an error correction unit configured to correct errors in the detected allele frequencies and obtain corrected allele frequencies for a set of polymorphic loci; a modeling unit configured to: obtain phased allelic information for a set of polymorphic loci by estimating the phase of the nucleic acid sequence data; obtaining individual allele frequency probabilities for polymorphic loci for different ploidy states by comparing phased allelic information with a set of models of different ploidy states and allelic imbalance fractions of a set of polymorphic loci; and obtaining joint probabilities for a set of polymorphic loci by combining the individual probabilities taking into account the relative distance between the polymorphic loci on a chromosome segment; and a hypothesis manager configured to select, based on the pooled probabilities, a best fit model indicative of chromosomal aneuploidy.

[376] In some exemplary embodiments of the system presented herein, the set of polymorphic loci includes from 1000 to 50,000 polymorphic loci. In some exemplary embodiments of the system provided herein, the set of polymorphic loci includes 100 known heterozygosity hot spot loci. In some exemplary embodiments of the system presented herein, the set of polymorphic loci includes 100 loci that are located at or within 0.5 kb. from the recombination hot spot.

[377] In some exemplary embodiments of the system presented herein, the best-fit model analyzes the following ploidy states of the first chromosome segment homolog and the second chromosome segment homolog: (1) all cells do not have a deletion or amplification of the first homolog or the second chromosome segment homolog; (2) some or all cells have a deletion of the first homologue or amplification of the second homolog of a chromosome segment; and (3) some or all cells have a deletion of the second homolog or amplification of the first homologue of the chromosome segment.

[378] In some exemplary system embodiments presented herein, errors that are corrected include allelic amplification efficiency bias, contamination, and/or sequencing errors. In some exemplary system embodiments provided herein, the contamination includes environmental contamination and genetic contamination. In some exemplary system embodiments presented herein, environmental contamination and genotype contamination are determined for homozygous alleles.

[379] In some exemplary system implementations presented herein, the hypothesis manager is configured to analyze the magnitude of the difference between the phased allelic information and the estimated allelic frequencies generated for the models. In some exemplary system implementations presented herein, the modeling engine generates individual allele frequency probabilities based on a beta binomial model of expected and observed allele frequencies at a set of polymorphic loci. In some exemplary system implementations presented herein, the modeling unit generates individual probabilities using a Bayesian classifier.

[380] In some exemplary embodiments of the system provided herein, nucleic acid sequence data is generated by performing high-throughput DNA sequencing of multiple copies of a series of amplicons generated using a multiplex amplification reaction, wherein each amplicon of the series of amplicons spans at least one polymorphic locus from a set of polymorphic loci, and where each of the polymorphic loci of the set is amplified. In some exemplary embodiments of the system presented herein, the multiplex amplification reaction is conducted under conditions of at least 1/2 of the reaction limiting primer. In some exemplary embodiments of the system provided herein, the sample has an average allelic imbalance of 0.4 to 5%.

[381] In some exemplary embodiments of the system provided herein, the sample is a plasma sample from an individual suspected of having cancer, and a hypothesis manager based on the best fit model is further configured to determine the presence of a copy number alteration in the individual's tumor cells.

[382] In some exemplary embodiments of the system provided herein, the sample is a plasma sample from an individual, and a hypothesis manager based on the best fit model is further configured to determine that cancer is present in the individual. In these embodiments, the hypothesis manager may be further configured to detect a single nucleotide variant at a single nucleotide variation site in a set of single nucleotide variation locations where the detection of either a chromosomal aneuploidy or a single nucleotide variant, or both, indicates the presence of circulating tumor nucleic acids in the sample.

[383] In some exemplary system implementations provided herein, the input processor is further configured to obtain haplotype information about a chromosome segment for an individual's tumor, and the modeling unit is configured to use the haplotype information to generate a set of models of various ploidy states and allelic imbalance fractions set of polymorphic loci.

[384] In some exemplary system embodiments presented herein, the modeling engine generates models for allelic imbalance fractions ranging from 0 to 25%.

[385] It should be understood that any of the methods provided herein may be performed using computer-readable code that is stored on a non-transitory computer-readable medium. Accordingly, provided herein in one embodiment is a non-transitory computer-readable medium for detecting chromosomal ploidy in a sample of an individual, comprising machine-readable code that, when executed by a processing device, causes the processing device to: receive allele frequency data including the amount of each allele present in the sample, in each locus in a set of polymorphic loci on a chromosomal segment; generate phased allelic information for a set of polymorphic loci by estimating the phase of allele frequency data; generate individual allele frequency probabilities for polymorphic loci for different ploidy states using allele frequency data; generate pooled probabilities for a set of polymorphic loci using individual probabilities and phased allelic information; and select, based on the pooled probabilities, the best fit model indicative of chromosomal ploidy, thereby determining the ploidy of the chromosomal segment.

[386] In some embodiments of the computer-readable medium, allele frequency data is generated from nucleic acid sequence data. Certain embodiments of the computer-readable medium further comprise correcting errors in the allele frequency data and using the corrected allele frequency data for the step of generating individual probabilities. In some embodiments of the computer-readable medium, the errors that are corrected are biases in the efficiency of amplification of the allele. In some embodiments of the computer-readable medium, individual probabilities are generated using a set of models of both different ploidy states and allelic imbalance fractions for a set of polymorphic loci. In some embodiments of the computer-readable medium, pooled probabilities are generated by taking into account the relationship between polymorphic loci on a chromosome segment.

[387] In one specific embodiment, provided herein is a non-transitory computer-readable medium for detecting chromosomal ploidy in a sample of an individual, comprising machine-readable code that, when executed by a processing device, causes the processing device to: receive nucleic acid sequence data for alleles at a set of polymorphic loci on a chromosome segment individual; determine allele frequencies at a set of loci using nucleic acid sequence data; perform allele amplification efficiency bias correction at detected allele frequencies to generate corrected allele frequencies for a set of polymorphic loci; generate phased allelic information for a set of polymorphic loci by estimating the phase of the nucleic acid sequence data; generate individual allele frequency probabilities for polymorphic loci for different ploidy states by comparing adjusted allele frequencies with a set of models of different ploidy states and allelic imbalance fractions of multiple polymorphic loci; generate joint probabilities for a set of polymorphic loci by combining individual probabilities taking into account the relationship between polymorphic loci on a chromosome segment; and select, based on the pooled probabilities, the best fit model indicating chromosomal aneuploidy.

[388] In certain illustrative embodiments for computer readable media, the selection is made by analyzing the magnitude of the difference between the phased allelic information and the estimated allelic frequencies generated for the models.

[389] In certain illustrative computer-readable medium embodiments, individual allele frequency probabilities are generated based on a beta binomial model of expected and observed allele frequencies at a set of polymorphic loci.

[390] It should be understood that any of the method embodiments provided herein may be performed by executing code stored on a non-transitory computer readable medium.

[391] Exemplary Embodiments for Cancer Detection

[392] In certain aspects, the present invention provides a method for detecting cancer. It should be understood that the sample may be a tumor sample or a liquid sample, such as plasma, from an individual suspected of having cancer. These methods are particularly effective at detecting genetic mutations, such as single nucleotide changes such as SNV, or copy number changes such as ICH, in samples with low levels of these genetic changes as a fraction of the total DNA in the sample. Thus, the sensitivity for detecting cancer DNA or RNA in samples is exceptional. Methods may combine any or all of the improvements provided herein to detect ICH and NVC to achieve this exceptional sensitivity.

[393] Accordingly, in some embodiments provided herein, there is provided a method for determining whether circulating tumor nucleic acids are present in a sample from an individual, and a non-transitory computer-readable medium comprising computer-readable code that, when executed by a processing device, causes the processing device to to implement the method. The method includes the following steps: analysis of a sample to determine ploidy in a set of polymorphic loci on a chromosome segment in an individual; and determining the level of average allelic imbalance present at the polymorphic loci based on the ploidy determination, where the average allelic imbalance is equal to or greater than 0.4, 0.45, 0.5, 0.6, 0.7, 0.75, 0, 8, 0.9, or 1% indicates the presence of circulating tumor nucleic acids, such as ctDNA, in the sample.

[394] In certain illustrative examples, an average allelic imbalance of greater than 0.4, 0.45, or 0.5% is indicative of the presence of ctDNA. In some embodiments, the method of determining the presence of circulating tumor nucleic acids further comprises detecting a single nucleotide variant at a single nucleotide variation site in a set of single nucleotide variation locations in which either an allelic imbalance equal to or greater than 0.5% is detected, a single nucleotide variant is detected, or both, which indicates the presence of circulating tumor nucleic acids in the sample. It should be understood that any of the methods provided for detecting chromosomal ploidy or ICH can be used to determine the level of allelic imbalance, usually expressed as average allelic imbalance. It should be understood that any of the methods proposed herein for detecting NNVs can be used to detect a single nucleotide for this aspect of the present invention.

[395] In some embodiments, a method for determining the presence of circulating tumor nucleic acids further comprises performing the method on a control sample with a known average allelic imbalance ratio. The control, for example, may be a sample from an individual's tumor. In some embodiments, the control has the average allelic imbalance expected for the sample being analyzed. For example, AAI from 0.5 to 5% or an average allelic imbalance coefficient of 0.5%.

[396] In some embodiments, the analysis step of a method for determining the presence of circulating tumor nucleic acids includes analysis of a number of chromosomal segments that are known to exhibit aneuploidy in cancer. In some embodiments, the analysis step of a method for determining the presence of circulating tumor nucleic acids includes analyzing 1000 to 50,000 or 100 to 1000 polymorphic loci for ploidy. In some embodiments, the analysis step of a method for determining the presence of circulating tumor nucleic acids includes analysis of 100 to 1000 single nucleotide variant sites. For example, in these embodiments, the analysis step may include performing multiplex PCR to amplify amplicons at 1000-50000 polymer loci and 100-1000 single nucleotide variant sites. This multiplex reaction can be configured as a single reaction or as a collection of different subset multiplex reactions. The multiplex reaction methods proposed herein, such as the mass multiplex PCR disclosed herein, provide an exemplary process for conducting an amplification reaction to help achieve improved multiplexing and therefore sensitivity levels.

[397] In some embodiments, the multiplex PCR reaction is run under limiting primer conditions for at least 10, 20, 25, 50, 75, 90, 95, 98, 99, or 100% of the reactions. Improved bulk multiplex reaction conditions presented herein can be used.

[398] In certain aspects, the above method of determining whether circulating tumor nucleic acids are present in a sample from an individual, and all embodiments thereof, can be accomplished by a system. The disclosure provides guidance regarding specific functional and structural features for implementing the methods. By way of non-limiting example, the system includes the following:

[399] An input processor configured to analyze data from a sample to determine ploidy at a set of polymorphic loci on a chromosome segment in an individual; And

[400] A modeling unit configured to determine the level of allelic imbalance present at polymorphic loci based on a determination of ploidy, where allelic imbalance equal to or greater than 0.5% indicates the presence of circulation.

[401] Exemplary Embodiments for Detecting Single Nucleotide Variants

[402] In certain aspects, methods are provided herein for detecting single nucleotide variants in a sample. The improved methods presented herein can achieve detection limits of 0.015, 0.017, 0.02, 0.05, 0.1, 0.2, 0.3, 0.4, or 0.5 percent ODS in a sample. All embodiments for detecting CNVs can be performed using the system. The disclosure provides guidance regarding specific functional and structural features for implementing the methods. Also provided herein are embodiments comprising a non-transitory computer-readable medium comprising computer-readable code that, when executed by a processing device, causes the processing device to perform the CNV detection methods presented herein.

[403] Accordingly, provided herein in one embodiment is a method for determining whether a single nucleotide variant is present at a set of genomic positions in a sample from an individual, the method comprising: for each genomic position, obtaining an estimate of the efficiency and error rate per cycle for the amplicon , covering this genomic position, using the training data set; obtaining information about the observed nucleotide identity for each genomic position in the sample; determining a set of probabilities for the percentage of single nucleotide variants resulting from one or more actual mutations at each genomic position by comparing the observed nucleotide identity information at each genomic position with a model of the various percentage variants using the estimated amplification efficiency and round-robin error rates for each genomic position independently ; and determining the most likely actual percentage variant and confidence from the set of probabilities for each genomic position.

[404] In exemplary embodiments of a method for determining the presence of a single nucleotide variant, efficiency and per-cycle error rate estimates are generated for a set of amplicons that span a genomic position. For example, 2, 3, 4, 5, 10, 15, 20, 25, 50, 100 or more amplicons that span a genomic position may be included.

[405] In exemplary embodiments of a method for determining the presence of a single nucleotide variant, the observed nucleotide identity information comprises an observed number of complete reads for each genomic position and an observed number of variant allele reads for each genomic position.

[406] In exemplary embodiments of the method for determining the presence of a single nucleotide variant, the sample is a plasma sample and the single nucleotide variant is present in circulating tumor DNA of the sample.

[407] In another embodiment provided herein, a method is provided for estimating the percentage of single nucleotide variants that are present in a sample from an individual. The method includes the following steps: obtaining, for a set of genomic positions, a run-by-cycle efficiency and error rate estimate for one or more amplicons spanning those genomic positions using the training data set; obtaining observable nucleotide identity information for each genomic position in the sample; obtaining a calculated mean and variance for the total number of molecules, background error molecules, and real mutant molecules for a search space including the initial percentage of real mutant molecules using amplification efficiency and amplicon error rates per cycle; and determining the percentage of single nucleotide variants present in the sample as a result of the actual mutations by determining the most likely actual percentage of the single nucleotide variant by fitting a distribution using the estimated means and variances for the observed nucleotide identity information in the sample.

[408] In illustrative examples of this method for estimating the percentage of single nucleotide variants that are present in a sample, the sample is a plasma sample and the single nucleotide variant is present in circulating tumor DNA of the sample.

[409] The training data set for this embodiment typically includes samples from a single healthy individual or, preferably, a group of healthy individuals. In some illustrative embodiments, the training data set is analyzed on the same day, or even in the same cycle, as one or more study samples. To obtain a set of training data, for example, samples from a group of 2, 3, 4, 5, 10, 15, 20, 25, 30, 36, 48, 96, 100, 192, 200, 250, 500, 1000 can be used or healthier individuals. Where data are available from a larger number of healthy individuals, for example 96 or more, confidence in estimates of amplification efficiency increases, even if runs are performed before the method is performed on the test samples. The PCR error rate can make use of the nucleic acid sequence information generated not only for the localization of the base with the ONV, but also for the entire amplified region around the ONV, since the error rate is calculated per amplicon. For example, using samples from 50 individuals and sequencing an amplicon around a 20-bp ONV, error rate data from 1000 base reads can be used to determine error rates.

[410] Typically, amplification efficiency is estimated by estimating the mean and standard deviation of the amplification efficiency for the amplified segment and then fitting it to a distribution model such as a binomial distribution or a beta-binomial distribution. The error rate is determined for a PCR reaction with a known number of cycles, and the error rate per cycle is then estimated.

[411] In certain illustrative embodiments, estimating the initial molecules of the test data set further includes updating the performance estimate for the test data set using the initial number of molecules estimated in step (b) if the observed number of reads differs significantly from the estimated number of reads. The score can then be updated for new potencies and/or parent molecules.

[412] The search space used to estimate the total number of molecules, background error molecules, and real mutation molecules may include a search space of 0.1, 0.2, 0.25, 0.5, 1, 2.5, 5, 10, 15, 20 or 25% at the lower end and 1, 2, 2.5, 5, 10, 12.5, 15, 20, 25, 50, 75, 90 or 95% at the upper end of the base copies in the provisions of the NDC, which is the basis for the NDC. Lower ranges, 0.1, 0.2, 0.25, 0.5 or 1% at the low end and 1, 2, 2.5, 5, 10, 12.5 or 15% at the high end may be used for illustrative purposes. examples for plasma samples, where the method detects circulating tumor DNA. Higher ranges are used for tumor samples.

[413] The distribution fits the total number of molecules with error (background error and real mutation) in the total molecules in order to calculate the likelihood or probability for each possible real mutation in the search space. This distribution can be binomial or beta binomial.

[414] The most likely real mutation is determined by determining the most likely percentage of real mutations and calculating the confidence using data from the fitting distribution. As an illustrative example, and not intended to limit the clinical interpretation of the methods presented herein, if the average mutation rate is high, then the percentage of confidence required to make a positive determination of ONV is lower. For example, if the average mutation rate for an ONV in a sample using the most likely hypothesis is 5% and the percentage confidence is 99%, then a positive ONV recognition will be made. On the other hand, for this illustrative example, if the average mutation rate for a CNV in a sample using the most likely hypothesis is 1% and the percentage confidence is 50%, then in certain situations a positive CNV recognition will not be made. It should be understood that clinical interpretation of the data will depend on the sensitivity, specificity, prevalence, and availability of the alternative product.

[415] In one illustrative embodiment, the sample is a sample of circulating DNA, such as a sample of circulating tumor DNA.

[416] In another embodiment, provided herein is a method for detecting one or more single nucleotide variants in a test sample from an individual. The method according to this embodiment includes the following steps:

[417] Determining the median variant allele frequency for a plurality of control samples from each of a plurality of normal individuals, for each single nucleotide variant position in a set of single nucleotide variation positions, based on the results obtained in a sequencing run to identify selected single nucleotide variant positions having median frequencies variant alleles in normal samples below the threshold, and determine the background error for each of the positions of the single-nucleotide variant after removing samples with outliers for each of the positions of the single-nucleotide variant; determining the observed depth of the read weighted mean and variance for selected single nucleotide variant positions for the test sample based on the data generated during sequencing for the test sample; and computer-assisted detection of one or more single nucleotide variant positions with a statistically significant weighted average read depth compared to the background error for that position, thereby detecting the one or more single nucleotide variants.

[418] In some embodiments of this method of detecting one or more CNVs, the sample is a plasma sample, the control samples are plasma samples, and the detected one or more single nucleotide variants are present in circulating tumor DNA of the sample. In some embodiments of this method for detecting one or more CNVs, the set of control samples includes at least 25 samples. In certain illustrative embodiments, the plurality of control samples is at least 5, 10, 15, 20, 25, 50, 75, 100, 200, or 250 samples at the low end and 10, 15, 20, 25, 50, 75, 100, 200, 250, 500 and 1000 samples at the high end.

[419] In some embodiments of this method for detecting one or more CNVs, outliers are removed from the data generated in the high-throughput sequencing run to calculate the observed depth of the weighted average read and determine the observed variance. In some embodiments of this method for detecting one or more SNVs, the read depth for each single nucleotide variant position for the test sample is at least 100 reads.

[420] In some embodiments of this method, to detect one or more CNVs, the sequencing cycle includes a multiplex amplification reaction conducted under limited primer reaction conditions. To perform these embodiments, the illustrative examples utilize the improved methods for performing multiplex amplification reactions presented herein.

[421] Without being limited by theory, the methods of the present embodiment utilize a background error model using normal plasma samples that are sequenced in the same sequencing run as the test sample to account for run-specific artifacts. Noise-rich positions with normal median variant allele frequencies above a threshold, such as greater than 0.1, 0.2, 0.25, 0.5, 0.75, and 1.0%, are removed.

[422] Outlier samples are iteratively extracted from the model to account for noise and contamination. For each base substitution of each genomic locus, the depth-weighted mean read depth and error standard deviation are calculated. In certain illustrative embodiments, samples, such as tumor or cell-free plasma samples, with single nucleotide variant positions with at least a threshold number of reads, such as at least 2, 3, 4, 5, 6, 7, 8, 9, 10, 15, 20, 25, 50, 100, 250, 500, or 1000 variant reads and an a1 Z-score greater than 2.5, 5, 7.5, or 10 compared to the background error model in some embodiments are considered a candidate mutation.

[423] In some embodiments, read depths greater than 100, 250, 500, 1000, 2000, 2500, 5000, 10000, 20000, 250000, 50000, or 100000 at the low end of the range and 2000, 2500, 5000, 7500, 10 000, 25000, 50,000, 100,000, 250,000, or 500,000 top end reads are achieved per sequencing run for each single nucleotide variant position in a set of single nucleotide variant positions. Generally, sequencing is high-throughput sequencing. The average or median values generated for the test samples are, in exemplary embodiments, weighted by read depth. Therefore, the probability that a variant allele determination is valid in a sample with 1 variant allele detected in 1000 reads is weighted higher than in a sample with 1 variant allele detected in 10,000 reads. Because variant allele (i.e., mutation) determinations are not made with 100% confidence, the identified single nucleotide variant can be considered a candidate variant or candidate mutations.

[424] Example test statistics for phased data analysis

[425] The following describes exemplary test statistics for the analysis of phased data from a sample known or suspected to be a mixed sample containing DNA or RNA that is derived from two or more cells that are not genetically identical. Let f denote the DNA or RNA fraction of interest, for example, the DNA or RNA fraction from ICH of interest, or the DNA or RNA fraction from cells of interest, such as cancer cells. In some embodiments, for cancer testing, f is the fraction of DNA or RNA from cancer cells in a mixture of cancer and normal cells, or f is the fraction of cancer cells in a mixture of cancer and normal cells. Note that this refers to a fraction of DNA from the cells of interest, provided that each cell of interest produces two copies of DNA. This is distinct from the fraction of DNA from the cells of interest in the segment that undergoes deletion or duplication.

[426] The possible allele values for each SNP are designated A and B. AA, AB, BA, and BB are used to designate all possible ordered pairs of alleles. In some embodiments, SNPs with ordered AB or BA alleles are provided. Let N _i denote the number of sequence reads of the i-th SNP and let A _i and B _i denote the number of reads of the i-th SNP, which denote, respectively, allele A and B. It is assumed that:

[427] .

[428] The ratio of alleles R _i is defined as:

[429]

[430] Let T denote the number of target SNPs.

[431] Without loss of universality, some embodiments focus on a single chromosomal segment. For clarity, in this specification, the expression “a first homologous chromosome segment versus a second homologous chromosome segment” means a first homologous chromosome segment and a second homologous chromosome segment. In some such embodiments, all of the target SNPs are contained in the chromosome segment of interest. In other embodiments, multiple segments of a chromosome are analyzed for possible copy number variations.

[432] Estimation of maximum a posteriori (MAP)

And
[433] This method uses knowledge of phasing across ordered alleles to detect deletion or duplication of a target segment. For each SNP i, determine

And

And

And
[434]

[435] Then determine

All SNPs
[436]

[437] The distributions of Xi _and S under various copy number hypotheses (such as disomy, first or second homolog deletions, or first or second homolog duplications) are described below.

[438] Disomy hypothesis

[439] According to this hypothesis, the target segment does not undergo deletion or duplication.

[440]

[441]

[442] where

[443]

[444] If we assume a constant read depth N , this gives us a binomial distribution S with parameters

[445] and T.

[446] Deletion hypotheses

[447] According to this hypothesis, the first homologue undergoes a deletion (that is, the AB SNP becomes B and the BA SNP becomes A), then R _i has a binomial distribution with the parameters and T for AB SNP, and and T for BA SNP. Hence,

[448]

[449] Assuming a constant read depth N , this gives a binomial distribution S with parameters

[450] and T.

[451] According to this hypothesis, the second homologue undergoes a deletion (that is, the AB SNP becomes A, and the BA SNP becomes B), then R _i has a binomial distribution with the parameters and T for AB SNP, and and T for BA SNP. Hence,

[452]

[453] Assuming a constant read depth N , this gives a binomial distribution S with parameters

[454] and T.

[455] Duplication hypotheses

[456] According to this hypothesis, the first homolog undergoes duplication (that is, the AB SNP becomes AAB, and the BA SNP becomes BBA), then R _i has a binomial distribution with the parameters and T for AB SNP, and and T for BA SNP. Hence,

[457]

[458] Assuming a constant read depth N , this gives a binomial distribution S with parameters

[459] and T.

[460] According to this hypothesis, the second homolog undergoes duplication (that is, the AB SNP becomes ABB, and the BA SNP becomes BAA), then R _i has a binomial distribution with the parameters and T for AB SNP, and and T for BA SNP. Hence,

[462] Assuming a constant read depth N , this gives a binomial distribution S with parameters

[463] and T.

[464] Classification

[465] As shown in the sections above, Xi , is _a binary random variable with

[466]

this disomy homolog 1 deletion homolog 2 deletion duplication of homologue 1 homolog 2 duplication

[467] This allows the probability of the test statistic S to be calculated for each hypothesis. The probability of each hypothesis given the measured data can be calculated. In some embodiments, the hypothesis with the highest probability is stated. If necessary, the distribution on S can be simplified either by approximating each N _i , with a constant coverage depth N , or by truncating the read depth to a constant N . This simplification gives

[468]

this disomy homolog 1 deletion homolog 2 deletion duplication of homolog 1 homolog 2 duplication

[469] The value for f can be estimated by selecting the most likely value of f given the measured data, such as the value of f that generates the best fit to the data using an algorithm (e.g., a search algorithm) such as maximum likelihood estimation, maximum posterior estimation, or Bayesian estimation. In some embodiments, multiple chromosome segments are analyzed and the value of f is estimated based on the data for each segment. If all target cells have these duplications or deletions, the estimated values for f based on the data for these different segments are similar. In some embodiments, f is measured experimentally, for example, by determining the fraction of DNA or RNA from cancer cells based on differences in methylation (hypomethylation or hypermethylation) between cancer and noncancerous DNA or RNA.

[470] Failure to accept a single hypothesis

[471] The distribution of S for the disomy hypothesis is independent of f . Thus, the probability of the measured data can be calculated for the disomy hypothesis without calculating f . For the null hypothesis of disomy, a single hypothesis rejection test can be used. In some embodiments, the probability S of the disomy hypothesis is calculated, and the disomy hypothesis is rejected if the probability is below a predetermined threshold (eg, less than 1 in 1000). This indicates the presence of a duplication or deletion of a chromosomal segment. If necessary, the false positive rate can be changed by adjusting the threshold value.

[472] Example Methods for Analyzing Phased Data

[473] Exemplary methods are described below for analyzing data from a sample that is known or suspected to be a mixed sample containing DNA or RNA that is derived from two or more cells that are not genetically identical. In some embodiments, phased data is used. In some embodiments, for each calculated allele ratio, the method includes determining whether the calculated allele ratio is higher or lower than the expected allele ratio and the magnitude of the difference for a particular locus. In some embodiments, a probable distribution is determined for the ratio of alleles at a locus for a particular hypothesis, and the closer the calculated ratio of alleles is to the center of the probable distribution, the greater the likelihood that the hypothesis is true. In some embodiments, the method includes determining the probability that the hypothesis is true for each locus. In some embodiments, the method includes determining the probability that a hypothesis is true for each locus, and combining the probabilities of that hypothesis for each locus, and selecting the hypothesis with the highest combined probability. In some embodiments, the method includes determining the probability that the hypothesis is true for each locus and for each possible ratio of DNA or RNA from one or more target cells to the total DNA or RNA in the sample. In some embodiments, a joint probability for each hypothesis is determined by combining the probabilities of that hypothesis for each locus and each possible relationship, and the hypothesis with the highest combined probability is selected.

[474] In one embodiment, the following hypotheses are considered: H ₁₁ (all cells are normal), H ₁₀ (presence of cells with a single homolog 1, therefore with deletion of homolog 2), H ₀₁ (presence of cells with a single homologue 2, therefore with deletion of homologue 1), H ₂₁ (presence of cells with duplication of homologue 1), H ₁₂ (presence of cells with duplication of homologue 2). For fraction f of target cells, such as cancer cells or mosaic cells (or fractions of DNA or RNA from target cells), the expected allele ratio for heterozygous ( AB or BA ) SNPs can be found as follows:

[475] Equation (1):

[476]

[477] Fix offset, contamination and sequencing errors:

[478] An observation of D _s in an SNP consists of a number of original mapped reads with the presence of each allele, n _A ⁰ and n _B ⁰ . We can then find the corrected reads n _A and n _B using the expected amplification bias of the A and B alleles.

[479] Let c _a denote environmental contamination (eg, DNA contamination in the air or environment) and r(c _a ) denote the allele ratio for an environmental contaminant (which is initially assumed to be 0.5). Moreover, c _g denotes the degree of contamination by genotype (eg, the degree of contamination from another sample), and r(c _g ) is the allele ratio for the contaminant. Let s _e (A,B) and s _e (B,A) denote sequencing errors for the recognition of one allele by another allele (for example, the erroneous detection of allele A when allele B is present).

[480] It is possible to find the observed ratio of alleles q(r, c _a , r(c _a ), c _g , r(c _g ), s _e (A,B), s _e (B,A)) for a given expected ratio r alleles by adjusting for environmental contamination, genotype contamination, and sequencing error.

[481] Since the genotypes of the pollutants are unknown, population frequencies can be found to find P(r(c _g )) . More specifically, let p be the population frequency for one of the alleles (which may be referred to as the reference allele). Then, we have P(r(c _g ) = 0) = (1-p) ² , P(r(c _g ) = 0) = 2p(1-p) , and P(r(cg) = 0) =p ² . The conditionally expected value above r(c _g ) can be used to determine E [q(r, c _a , r(c _a ) , c _g , r(c _g ), s _e (A,B), s _e (B, A)) ] . Note that environmental and genotype contaminants are determined using homozygous SNPs and are therefore not affected by the absence or presence of deletions or duplications. In addition, if necessary, environmental and genotype contamination can be measured using a reference chromosome.

[482] Plausibility for each SNP:

[483] The equation below gives the probability of observing n _A and n _B given the allele ratio r :

[484] Equation (2):

[485]

[486] Let D _s denote the data for the SNP . For each hypothesis h ϵ { H ₁₁ , H ₀₁ , H ₁₀ , H ₂₁ , H ₁₂ }, you can represent r=r(AB,h) or r=r(BA,h) in equation (1) and find the conditionally expected a value greater than r(c _g ) to determine the observed allele ratio E [q(r, c _a , r(c _a ) , c _g , r(c _g )) ]. Then, setting r= E [q(r, c _a , r(c _a ) , c _g , r(c _g ), s _e (A,B), s _e (B,A)) ] in equation (2 ) we can define P(D _s |h,f).

[487] Search algorithm:

[488] In some embodiments, SNPs with allele ratios that appear to be outliers are ignored (eg, by ignoring or excluding SNPs with allele ratios that are at least 2 or 3 standard deviations above or below the mean). Note that the benefit identified for this approach is that when there is a higher percentage of mosaicism, the variability in allele ratios can be high, hence ensuring that SNPs are not truncated due to mosaicism.

[489] Let F = {f ₁ , ...., f _N } denote the search space for the percentage of mosaicism (such as tumor fraction). We can define P(D _s |h,f) in each SNP s and f ϵ F and combine the probability for all SNPs.

[490] The algorithm tries each f for each hypothesis. Using the search method, we can conclude that mosaicism exists if there is a range of F* from f scores where the confidence of the deletion or duplication hypothesis is higher than the confidence of the no-deletion and no-duplication hypotheses. In some embodiments, a maximum likelihood estimate for P(D _s |h,f) in F * is determined. If necessary, you can determine the conditionally expected value f ϵ F *. If necessary, confidence can be determined for each hypothesis.

[491] In some embodiments, a beta binomial distribution is used instead of a binomial distribution. In some embodiments, a reference chromosome or chromosomal segment is used to determine specific beta binomial parameters in a sample.

[492] Theoretical characteristics when using simulations:

[493] If necessary, the theoretical performance of an algorithm can be assessed by randomly assigning the number of SNP reference reads to a given read depth (DOR). For the normal case, use p = 0.5 for the binomial probability parameter, and for deletions or duplications, respectively, p is revised. The approximate input parameters for each simulation are: (1) the number S of SNPs (2) the DOR constant D per SNP, (3) p and (4) the number of experiments.

[494] First simulation experiment:

[495] This experiment focused on S ϵ {500, 1000}, D ϵ {500, 1000} and p ϵ {0%, 1%, 2%, 3%, 4%, 5%}. We ran 1000 simulation experiments in each setting (hence 24,000 experiments with phase and 24,000 without phase). We simulated the number of reads from a binomial distribution (other distributions can be used if necessary). The frequency of false positive (in the case of p=0% ) and false negative (in the case of p>0% ) results was determined both with and without phase information. Please note that phase information is very useful, especially for S=1000, D=1000 . However, for S=500, D=500, this algorithm has the highest false positive rate, both with and without phase, regardless of the conditions tested.

[496] Phase information is especially useful for low percentages of mosaicism (≤ 3% ). Without phase information , a high false-negative rate was observed for p=1% because confidence in a deletion is determined by assigning equal odds to H ₁₀ and H ₀₁ , and a small bias in favor of one hypothesis is not enough to compensate for the low likelihood of the other hypothesis. This also applies to duplications. We also note that the algorithm seems to be more sensitive to read depth compared to the number of SNPs. For results with phase information, we assume that accurate phase information is available for a large number of consecutive heterozygous SNPs. If necessary, haplotype information can be obtained by probabilistically combining haplotypes on smaller segments.

[497] Second simulation experiment:

[498] This experiment focused on S ϵ {100, 200, 300, 400, 500}, D ϵ {1000, 2000, 3000, 4000, 5000} and p ϵ {0, 1,1, 5, 2, 2 ,5, 3%} and 10,000 random experiments at each setting. The rate of false positive results (in the case of p = 0%) and false negative results (in the case of p greater than 0% ) was determined both with and without phase information. The false negative rate was 10% for D ≥ 3000 and N ≥ 200 using haplotype information, while the same performance was achieved for D = 5000 and N ≥ 400 . The difference between false-negative rates was particularly noticeable for small percentages of mosaicism. For example, when p=1%, a false negative rate of less than 20% is never achieved without haplotype data, whereas it is close to 0% for N ≥ 300 and D ≥ 3000 . For p=3%, a 0% false negative rate is observed with haplotype data, while achieving the same performance without haplotype data requires N≥300 and D≥3000 .

[499] Example methods for detecting deletions and duplications without phased data

[500] In some embodiments, unphased genetic data is used to determine whether there is an excessive number of copies of a first homologous chromosome segment relative to a second homologous chromosome segment in an individual's genome (e.g., in the genome of one or more cells or in scDNA or scRNA). In some embodiments, phased genetic data is used, but the phasing is ignored. In some embodiments, the DNA or RNA sample is a mixed sample of scDNA or scRNA from an individual that includes scDNA or scRNA from two or more genetically different cells. In some embodiments, the method uses the magnitude of the difference between the calculated allele ratio and the expected allele ratio for each of the loci.

[501] In some embodiments, the method includes obtaining genetic data at a set of polymorphic loci on a chromosome or chromosome segment in a sample of DNA or RNA from one or more cells from an individual by measuring the amount of each allele at each locus. In some embodiments, allele ratios are calculated for loci that are heterozygous in at least one cell from which the sample was obtained. In some embodiments, the calculated allele ratio for a particular locus is the measured amount of one of the alleles divided by the total measured amount of all alleles for the locus. In some embodiments, the calculated allele ratio for a particular locus is the measured amount of one of the alleles (e.g., an allele on a first homologous segment of a chromosome) divided by the measured amount for a locus of one or more other alleles (e.g., an allele on a second homologous segment of a chromosome). The calculated allele ratios and expected allele ratios can be calculated using any of the methods presented herein or any standard method (eg, any mathematical transformation of the calculated allele ratios or expected allele ratios presented herein).

[502] In some embodiments, a test statistic is calculated based on the magnitude of the difference between the calculated allele ratio and the expected allele ratio for each of the loci. In some embodiments, the test statistic Δ is calculated using the following formula

[503] where _δi represents the magnitude of the difference between the calculated allele ratio and the expected allele ratio for the i-th locus;

[504] where μ _i represents the average value of δ _i ; And

[505] where σ _i ² represents the average deviation _δi .

[506] For example, we can define _δi as follows when the expected allele ratio is 0.5:

[507] .

[508] Values for μ _i and σ _i can be calculated using the fact thatR _i is a binomial random variable. In some embodiments, all loci are assumed to have the same standard deviation. In some embodiments, an average or weighted average of the standard deviation or an estimate of the standard deviation is used for the value of σ _i ² . In some embodiments, the test statistic is assumed to have a normal distribution. For example, from the central limit theorem it follows that the distribution Δ approaches the standard normal as the number of loci becomes larger (for example, the numberT SNP).

[509] In some embodiments, a set of one or more hypotheses is listed that determine the number of copies of a chromosome or chromosome segment in the genome of one or more cells. In some embodiments, a hypothesis is selected that is most likely based on the test statistic, thereby determining the number of copies of a chromosome or chromosome segment in the genome of one or more cells. In some embodiments, a hypothesis is selected if the probability that the test statistic belongs to the distribution of the test statistic for that hypothesis exceeds an upper threshold; one or more hypotheses are rejected if the probability that the test statistic belongs to the distribution of the test statistic for that hypothesis is below a lower threshold; or a hypothesis is not selected or rejected if the probability that the test statistic belongs to the distribution of the test statistic for that hypothesis is between the lower and upper thresholds, or if the probability is not determined with sufficiently high confidence. In some embodiments, the upper and/or lower threshold is determined from an empirical distribution, such as a distribution from training data (such as samples with a known copy number, such as diploid samples or samples that are known to have a particular deletion or duplication). This empirical distribution can be used to select a threshold for testing the rejection of a single hypothesis. Note that the Δ test statistic is independent of S , and so both measures can be used independently if necessary.

[510] Exemplary methods for detecting deletions and duplications using allele distributions or patterns

[511] Included in this section are methods for determining whether there is an excessive number of copies of a first homologous chromosome segment compared to a second homologous chromosome segment. In some embodiments, the method includes listing (i) a plurality of hypotheses determining the number of copies of a chromosome or chromosome segment that is present in the genome of one or more cells (such as cancer cells) of an individual, or (ii) a plurality of hypotheses determining the degree of copy number overrepresentation of the first homologous chromosome segment compared to a second homologous chromosome segment in the genome of one or more cells of an individual. In some embodiments, the method includes obtaining genetic data from an individual at multiple polymorphic loci (such as SNP loci) on a chromosome or chromosome segment. In some embodiments, a probability distribution of the individual's expected genotypes for each of the hypotheses is generated. In some embodiments, a data fit between the individual's obtained genetic data and a probability distribution of the individual's expected genotypes is calculated. In some embodiments, one or more hypotheses are ranked according to the fit of the data, and the highest ranking hypothesis is selected. In some embodiments, a technique or algorithm, such as a search algorithm, is used to perform one or more of the following steps: calculating the fit of the data, ranking the hypotheses, or selecting the top-ranked hypothesis. In some embodiments, the data fit is a beta binomial fit or a binomial fit. In some embodiments, the technique or algorithm is selected from the group consisting of maximum likelihood estimator, maximum posterior estimator, Bayesian estimator, dynamic estimator (eg, dynamic Bayesian estimator), and expectation maximization estimator. In some embodiments, the method includes applying a technique or algorithm to the obtained genetic data and the expected genetic data.

[512] In some embodiments, the method includes listing (i) a plurality of hypotheses determining the number of copies of a chromosome or chromosome segment that are present in the genome of one or more cells (such as cancer cells) of an individual, or (ii) a plurality of hypotheses determining the degree of overrepresentation the number of copies of the first homologous chromosome segment compared to the second homologous chromosome segment in the genome of one or more cells of an individual . In some embodiments, the method includes obtaining genetic data from an individual at multiple polymorphic loci (eg, SNP loci) on a chromosome or chromosome segment. In some embodiments, the genetic data includes the number of alleles for multiple polymorphic loci. In some embodiments, a joint distribution model is generated for the expected number of alleles at a plurality of polymorphic loci on a chromosome or chromosome segment for each hypothesis. In some embodiments, the relative probability for one or more hypotheses is determined using a joint distribution model and the number of alleles measured in the sample, and the hypothesis with the highest probability is selected.

[513] In some embodiments, the distribution or pattern of alleles (eg, the pattern of calculated allele ratios) is used to determine the presence or absence of CC, such as a deletion or duplication. If necessary, this scheme can be used to determine the initial origin of ICH.

[514] Approximate methods of counting/Quantitative methods

[515] In some embodiments, one or more counting methods (also referred to as quantitative methods) are used to detect one or more ICHs, such as deletions or duplications of chromosome segments or entire chromosomes. In some embodiments, one or more scoring methods are used to determine whether the excessive number of copies of the first homologous chromosome segment is caused by a duplication of the first homologous chromosome segment or a deletion of the second homologous chromosome segment. In some embodiments, one or more counting methods are used to determine the number of additional copies of a chromosome segment or chromosome that is duplicated (eg, whether there are 1, 2, 3, 4, or more additional copies). In some embodiments, one or more scoring methods are used to differentiate a sample with many duplications and a smaller tumor fraction from a sample with fewer duplications and a larger tumor fraction. For example, one or more scoring methods may be used to differentiate a sample with four extra chromosome copies and a 10% tumor fraction from a sample with two extra chromosome copies and a 20% tumor fraction. Exemplary methods are disclosed, for example, in US Patent Application Publication Nos. 2007/0184467; 2013/0172211; and 2012/0003637; US Patent No. 8467976; 7888017; 8008018; 8296076; and 8195415; U.S. Provisional Patent Application Serial No. 62/008235, filed June 5, 2014, and U.S. Provisional Patent Application Serial No. 62/032785, filed Aug. 4, 2014, each of which is hereby incorporated by reference in its entirety.

[516] In some embodiments, the counting method includes counting the number of reads based on DNA sequences that map to one or more given chromosomes or chromosome segments. Some such methods involve creating a reference value (threshold value) for the number of DNA sequence reads mapping to a particular chromosome or chromosome segment, with a number of reads exceeding this value indicating a particular genetic abnormality.

[517] In some embodiments, the total measured amount of all alleles for one or more loci (eg, the total amount of a polymorphic or non-polymorphic locus) is compared to a reference amount. In some embodiments, the reference amount is (i) a threshold value or (ii) an expected amount for a particular copy number hypothesis. In some embodiments, the reference count (in the absence of ICH) is the total measured count of all alleles for one or more loci for one or more chromosomes or chromosome segments that are known or expected to be free of deletion or duplication. In some embodiments, the reference (for the presence of ICH) is the total measured number of all alleles for one or more loci for one or more chromosomes or chromosome segments that are known or expected to have a deletion or duplication. In some embodiments, the reference amount is the total measured amount of all alleles for one or more loci for one or more reference chromosomes or chromosome segments. In some embodiments, the reference amount is the average or median of values determined for two or more different chromosomes, chromosomal segments, or different samples. In some embodiments, random (eg, massively parallel fractional sequencing) or targeted sequencing is used to quantify one or more polymorphic or non-polymorphic loci.

[518] In some embodiments, using a reference amount, the method includes (a) measuring the amount of genetic material on a chromosome or chromosome segment of interest; (b) comparing the quantity obtained in step (a) with the reference quantity; and (c) identifying the presence or absence of a deletion or duplication based on comparison.

[519] In some embodiments, using a reference chromosome or chromosome segment, the method includes sequencing DNA or RNA from the sample to obtain a plurality of sequence tags aligned to target loci. In some embodiments, the sequence tags are of sufficient length to be assigned to a specific target locus (eg, 15-100 nucleotides in length); the target loci belong to a plurality of different chromosomes or chromosome segments that include at least one first chromosome or chromosome segment expected to be abnormally distributed in the sample and at least one second chromosome or chromosome segment expected to be normally distributed in the sample. In some embodiments, multiple sequence tags are assigned to corresponding target loci. In some embodiments, the number of sequence tags aligned to target loci of the first chromosome or chromosomal segment and the number of sequence tags aligned to target loci of the second chromosome or chromosomal segment are determined. In some embodiments, these numbers are compared to determine the presence or absence of an abnormal distribution (eg, deletion or duplication) of the first chromosome or chromosome segment.

[520] In some embodiments, the definition of ICH uses an f value (eg, tumor fraction), for example, to compare the observed difference between the number of two chromosomes or chromosomal segments with the difference that would be expected for a particular type of ICH given the f value ( see, for example, US Patent Application Publication No. 2012/0190020; US Patent Application Publication No. 2012/0190021; US Patent Application Publication No. 2012/0190557; US Patent Application Publication No. 2012/0191358, each of which is hereby incorporated by reference in its entirety). For example, the difference in the amount of a chromosome segment that is duplicated in a tumor compared to a disomic reference chromosome segment increases as the tumor fraction increases. In some embodiments, the method includes comparing the relative frequency of a chromosome or chromosome segment of interest to a reference chromosome or chromosome segment (e.g., a chromosome or chromosome segment expected or known to be disomic) with an f value to determine the probability of ICH. For example, the difference in abundance between the first chromosomes or chromosome segment and the reference chromosome or chromosome segment can be compared with what would be expected given the f value for various possible ICHs (e.g., one or two additional copies of the chromosome segment of interest.

[521] The following examples of possible use illustrate the use of a counting/quantitative method to distinguish between a duplication of a first homologous chromosome segment and a deletion of a second homologous chromosome segment. Considering the host's normal disomic genome as the baseline, analysis of a mixture of normal and cancer cells provides the average difference between the baseline and the cancer DNA in the mixture. For example, imagine a case where 10% of the DNA in a sample came from cells with a deletion in the region of the chromosome targeted by the assay. In some embodiments, the quantitative approach indicates that the number of reads corresponding to this region should be 95% of the number of reads expected for a normal sample. This is because one of the two target chromosomal regions in each of the tumor cells with a deletion of the target region is missing, and thus the total amount of DNA mapping in this region is 90% (for normal cells) plus ½ x 10% (for tumor cells) = 95%. Alternatively, in some embodiments, the allelic approach reveals that the ratio of alleles at heterozygous loci averaged 19:20. Now imagine a case where 10% of the DNA in the sample came from cells with a fivefold focal amplification of the region of the chromosome targeted by the analysis. In some embodiments, the quantitative approach indicates that the number of reads corresponding to this region should be 125% of that expected for a normal sample. This is because one of the two target chromosomal regions in each of the tumor cells with fivefold focal amplification is copied an additional five times over the target region, and thus the total DNA mapping in that region is 90% (for normal cells) plus (2 + 5) x 10%/2 (for tumor cells) = 125%. Alternatively, in some embodiments, the allelic approach reveals that the ratio of alleles at heterozygous loci averages 25:20. Note that when using an allelic-only approach, a focal fivefold amplification across a chromosomal region in a sample with 10% cDNA may look the same as a deletion across the same region in a sample with 40% cDNA; in these two cases, the haplotype that is underrepresented in the case of a deletion turns out to be the non-ICH haplotype in the case of a focal duplication, and the non-ICH haplotype in the case of a deletion turns out to be an overrepresented haplotype in the case of a focal duplication. Combining the probabilities obtained from this allelic approach with those obtained from the quantitative approach differentiates the two possibilities.

[522] Approximate counting methods/Quantitative methods using reference samples

[523] An exemplary quantitative method using one or more reference samples is described in U.S. Provisional Patent Application Serial No. 62/008235, filed June 5, 2014, and U.S. Provisional Patent Application Serial No. 62/032785, filed Aug. 4. 2014, documents are incorporated herein by reference in their entirety. In some embodiments, one or more reference samples that are not likely to have any ICH on one or more chromosomes or chromosomes of interest (eg, a normal sample) are identified by selecting samples with the largest fraction of tumor DNA, selecting samples with z-score closest to zero, selecting samples in which the data are consistent with the hypothesis corresponding to the absence of ICH with the greatest confidence or probability, selecting samples known to be normal, selecting samples from individuals with the least likelihood of cancer (for example, being young, male gender in breast cancer screening, lack of family history, etc.), selecting samples with the highest amount of DNA input, selecting samples with the highest signal-to-noise ratio, selecting samples based on other criteria believed to correlate with the likelihood cancer, or sampling using some combination of criteria. Once a reference set is selected, one can make the assumption that these cases are disomic and then estimate SNP bias, that is, experiment-specific amplification and other processing bias for each locus. This experiment-specific estimate of bias can then be used to correct for bias in measurements of the chromosome of interest, such as chromosome 21 loci, and, if necessary, for other chromosomal loci for samples that are not part of the subset where disomy is suspected for the chromosome 21. Once biases have been corrected for these samples of unknown ploidy, the data for these samples can then be analyzed a second time using the same or a different method to determine whether individuals have trisomy 21. For example, a quantitative method can be used to remaining samples of unknown ploidy, and a z-score can be calculated using the corrected measured genetic data on chromosome 21. Alternatively, as part of a preliminary assessment of chromosome 21 ploidy status, the tumor fraction can be calculated for samples from an individual with suspected cancer. For this tumor fraction, the proportion of corrected reads expected in the case of disomy (disomy hypothesis) and the proportion of corrected reads expected in the case of trisomy (trisomy hypothesis) can be calculated. Alternatively, if the tumor fraction has not been previously measured, a set of disomy and trisomy hypotheses can be generated for different tumor fractions. For each case, the expected distribution of the proportion of corrected reads can be calculated, taking into account the expected statistical spread when different DNA loci are selected and measured. The observed adjusted fraction of reads can be compared with the distribution of the expected fraction of corrected reads, and the likelihood ratio for the disomy and trisomy hypotheses can be calculated for each of the samples of unknown ploidy. The ploidy state associated with the hypothesis with the highest calculated likelihood can be selected as the correct ploidy state.

[524] In some embodiments, a subset of samples with a sufficiently low probability of cancer may be selected as a control set of samples. The subset may be a fixed number or may be a variable number based on selecting only those samples that do not reach the threshold. Quantitative data from a subset of samples can be pooled, averaged, or combined using a weighted average, where the weighting is based on the probability that the sample is normal. Quantitative data can be used to determine locus bias for sequencing amplification samples in the current batch of control samples. The locus bias may also include data from other batches of samples. Bias at a locus may indicate the relative over- or under-amplification that is observed for that locus compared to other loci, based on the assumption that a subset of samples does not contain any ICH and that any observed over- or under-amplification is due to amplification and /or sequencing or other offset. Locus bias may account for the GC content of the amplicon. Loci can be grouped into loci groups for the purpose of calculating the bias for each locus. Once the bias for each locus has been calculated for each locus in the set of loci, sequencing data for one or more samples that are not in the sample subset, and optionally one or more samples that are in the sample subset, in order to to remove the effect of bias at that locus can be adjusted by adjusting the quantitative measurements for each locus. For example, if in a subset of patients SNP 1 was observed to have a read depth twice the average, the adjustment might involve replacing the number of reads corresponding to SNP 1 with a number that is half the average. If the locus in question is an SNP, the adjustment may involve halving the number of reads corresponding to each of the alleles at that locus. Once the sequencing data for each locus in one or more samples has been corrected, it can be analyzed using a method to detect the presence of ICH in one or more chromosomal regions.

[525] In one example, sample A is a mixture of amplified DNA originating from a mixture of normal and cancer cells, which is analyzed using a quantitative method. The following illustrates sample possible data. The q-arm region on chromosome 22 was found to have only 90% of the DNA mapped to this region, as expected; the focal region corresponding to the HER2 gene was found to have 150% more DNA displayed in that region, as expected; and the p-arm of chromosome 5 was found to have 105% DNA mapping as expected. The doctor may conclude that the sample contains a deletion of a region on the q arm of chromosome 22 and a duplication of the HER2 gene. The physician may conclude that because 22q deletions are common in breast cancer and that because cells with a deletion of the 22q region on both chromosomes usually do not survive, approximately 20% of the DNA in the sample comes from cells with a 22q deletion on one of the two chromosomes. The physician may also conclude that if the DNA from the mixed tumor cell-derived sample came from a set of genetically tumor cells whose HER2 region and 22q regions were homogeneous, then the cells contained a fivefold duplication of the HER2 region.

[526] In one example, sample A is also analyzed in an allelic manner. The following illustrates sample possible data. Two haplotypes in the same region on the q arm of chromosome 22 are present in a 4:5 ratio; two haplotypes in the focal region corresponding to the HER2 gene are present in a 1:2 ratio; and two haplotypes on the p-arm of chromosome 5 are represented in a ratio of 20:21. All other studied genome regions do not have a statistically significant excess of any haplotype. The clinician can conclude that the sample contains tumor DNA with ICH in the 22q region, the HER2 region, and the 5p arm. Based on the knowledge that 22q deletions are very common in breast cancer, and/or based on quantitative analysis showing an underrepresentation of the amount of DNA mapped to the 22q region of the genome, the clinician can infer the existence of a tumor with a 22q deletion. Based on knowledge that HER2 amplifications are very common in breast cancer and/or based on quantitative analysis showing an overrepresentation of the amount of DNA mapping to the HER2 region of the genome, a clinician can infer the existence of a tumor with HER2 amplification.

[527] Exemplary reference chromosomes or chromosomal segments

[528] In some embodiments, any of the methods presented herein are also performed on one or more reference chromosomes or chromosome segments, and the results are compared with the results for one or more chromosomes or chromosome segments of interest.

[529] In some embodiments, a reference chromosome or chromosome segment is used as a control for what would be expected in the absence of ICH. In some embodiments, the reference sample is the same chromosome or chromosome segment from one or more different samples that are known or expected not to have a deletion or duplication on that chromosome or chromosome segment. In some embodiments, the reference sample is a chromosome or chromosome segment different from the test sample that is presumed to be disomic. In some embodiments, the reference sample is a segment different from one of the chromosomes of interest in the same sample that is being tested. For example, the reference pattern may be one or more segments outside the region of the potential deletion or duplication. Having a reference sample for the same chromosome being tested avoids variability between different chromosomes, such as differences in metabolism, apoptosis, histones, inactivation and/or amplification between chromosomes. Analysis of non-ICH segments on the same chromosome as the test chromosome can also be used to determine differences in metabolism, apoptosis, histones, inactivation and/or amplification between homologs, allowing the level of variability between homologues in the absence of ICH to be determined for comparison with results for potential VCHK. In some embodiments, the magnitude of the difference between the calculated and expected allele ratio for a potential ICH is greater than the corresponding value for the reference sample, thereby confirming the presence of ICH.

[530] In some embodiments, a reference chromosome or chromosome segment is used as a control for what would be expected to result from the presence of ICH, such as a particular deletion or duplication of interest. In some embodiments, the reference sample is the same chromosome or chromosome segment from one or more different samples known or expected to have a deletion or duplication on that chromosome or chromosome segment. In some embodiments, the reference sample is a chromosome or segment of a chromosome different from the test sample that is known or expected to have ICH. In some embodiments, the magnitude of the difference between the calculated and expected allele ratio for a potential ICH is similar to (eg, slightly different) that of a reference sample for ICH, thereby confirming the presence of ICH. In some embodiments, the magnitude of the difference between the calculated and expected allele ratio for the potential ICH is less (eg, significantly less) than the corresponding value of the reference sample for ICH, thereby confirming the absence of ICH. In some embodiments, one or more loci for which the genotype of the cancer cell (or DNA or RNA from a cancer cell, such as scDNA or scRNA) is different from the genotype of a non-cancerous cell (or DNA or RNA from a non-cancerous cell, such as scDNA or scRNA). The tumor fraction can be used to determine whether the excessive number of copies of the first homologous chromosome segment is due to a duplication of the first homologous chromosome segment or a deletion of the second homologous chromosome segment. The tumor fraction can also be used to determine the number of extra copies of a segment of a chromosome or chromosome that is being duplicated (e.g., whether there are 1, 2, 3, 4, or more extra copies), for example, to differentiate a sample with four extra copies of a chromosome and a tumor fraction of 10% from a sample with two additional copies of the chromosome and a tumor fraction of 20%. Tumor fraction can also be used to determine how well observed data match expected data for possible ICHs. In some embodiments, the degree of prevalence of ICH is used to select a specific therapy or therapeutic regimen for an individual. For example, some therapeutic agents are effective only on at least four, six, or more copies of a chromosome segment.

[531] In some embodiments, one or more loci used to determine tumor fraction are on a reference chromosome or chromosome segment, such as a chromosome or chromosome segment that is known or expected to be a disomic chromosome or chromosome segment that is rare undergo duplications or deletions in cancer cells in general or in a particular type of cancer that an individual is known to have or is at increased risk of having, or a chromosome or chromosomal segment that is unlikely to be aneuploid (such as a segment that is expected to result in cell death in the event of deletion or duplication). In some embodiments, any of the methods of the invention is used to confirm that a reference chromosome or chromosome segment is disomic in both cancer and non-cancerous cells. In some embodiments, one or more chromosomes or chromosome segments are used for which there is a high degree of confidence in recognizing disomy.

[532] Exemplary loci that can be used to determine tumor fraction include polymorphisms or mutations (such as SNPs) in a cancer cell (or DNA or RNA, such as scDNA or scRNA from a cancer cell) that are absent in a non-cancerous cell (or DNA or RNA from a non-cancerous cell) in an individual. In some embodiments, the tumor fraction is determined by identifying those polymorphic loci at which the cancer cell (or DNA or RNA from a cancer cell) has an allele that is not present in non-cancerous cells (or DNA or RNA from a non-cancerous cell) in the sample (e.g., sample plasma or tumor biopsy) from the individual; and using the amount of an allele unique to the cancer cell at one or more identified polymorphic loci to determine the tumor fraction in the sample. In some embodiments, the non-cancerous cell is homozygous for the first allele at the polymorphic locus and the cancer cell is (i) heterozygous for the first allele and the second allele or (ii) homozygous for the second allele at the polymorphic locus. In some embodiments, the non-cancerous cell is heterozygous for the first allele and the second allele at the polymorphic locus, and the cancer cell (i) has one or two copies of the third allele at the polymorphic locus. In some embodiments, cancer cells are assumed or known to have only one copy of the allele, which is not present in non-cancerous cells. For example, if the genotype of non-cancerous cells is AA and cancer cells are AB, and 5% of the signal at that locus in a sample comes from the B allele and 95% comes from the A allele, then the proportion of tumor in the sample is 10%. In some embodiments, cancer cells are assumed or known to have two copies of an allele that is absent in non-cancerous cells. For example, if the genotype of non-cancerous cells is AA and cancer cells are BB, and 5% of the signal at that locus in a sample is from the B allele and 95% is from the A allele, the tumor fraction in the sample is 5%. In some embodiments, multiple loci for which cancer cells have an allele other than non-cancerous cells are analyzed to determine which loci in the cancer cells are heterozygous and which are homozygous. For example, for loci at which non-cancerous cells are AA, if the signal from the B allele is approximately 5% at some loci and approximately 10% at some loci, then the cancer cells are assumed to be heterozygous at the loci with approximately 5% B allele and homozygous at loci with approximately 10% B allele (indicating a tumor fraction of approximately 10%).

[533] Exemplary loci that can be used to determine tumor fraction include loci for which a cancer cell and a non-cancerous cell share one allele (for example, loci in which the cancer cell is AB and the non-cancerous cell is BB, or the cancer cell is BB, and the non-cancerous cell is AB). The amount of signal A, the amount of signal B, or the ratio of signal A to signal B in a mixed sample (containing DNA or RNA from a cancer cell and a non-cancerous cell) is compared with the corresponding value for (i) a sample containing DNA or RNA from cancer cells only or (ii) a sample , containing DNA or RNA only from non-cancerous cells. The difference in values is used to determine the proportion of tumor in a mixed sample.

[534] In some embodiments, loci that can be used to determine tumor fraction are selected based on the genotype of (i) a sample containing DNA or RNA only from cancer cells, and/or (ii) a sample containing DNA or RNA only from non-cancerous cells cells. In some embodiments, loci are selected based on analysis of a mixed sample, such as loci for which the absolute or relative abundance of each allele differs from what would be expected if both cancer and non-cancerous cells had the same genotype at a particular locus. For example, if cancer and non-cancerous cells have the same genotype, the loci would be expected to produce 0% B signal if all cells are AA, 50% B signal if all cells are AB, or 100% B signal if all cells are cells are BB. Other values for signal B indicate that the genotype of cancerous and noncancerous cells at this locus is different, and thus this locus can be used to determine the tumor fraction.

[535] In some embodiments, the tumor fraction calculated based on alleles at one or more loci is compared with the tumor fraction calculated using one or more scoring methods presented herein.

[536] Exemplary Methods for Phenotype Detection or Multiple Mutation Analysis

[537] In some embodiments, the method includes analyzing a sample for the presence of a set of mutations associated with a disease or disorder (such as cancer) or an increased risk of a disease or disorder. There are strong correlations between events within classes (eg, cancer classes M or C) that can be used to improve the method's signal-to-noise ratio and classify tumors into distinct clinical subgroups. For example, borderline results for multiple mutations (such as multiple ICHs) on one or more chromosomes or chromosome segments considered together can be a very strong signal. In some embodiments, determining the presence or absence of multiple polymorphisms or mutations of interest (such as 2, 3, 4, 5, 8, 10, 12, 15 or more) increases the sensitivity and/or specificity of determining the presence or absence of a disease or disorder such such as cancer, or an increased risk of a disease or disorder such as cancer. In some embodiments, the correlation between events on multiple chromosomes is used to analyze the signal more efficiently than looking at each one individually. To better classify tumors, the design of the method itself can be optimized. This, compared to relapse, can be incredibly useful for early detection and screening where sensitivity to one specific mutation/ICH may be paramount. In some embodiments, the events are not always correlated, but there is a possibility that they are correlated. In some embodiments, a matrix estimator formulation is used with a noise covariance matrix having off-diagonal parameters.

[538] In some embodiments, the invention provides a method for detecting a phenotype (such as a cancer phenotype) in an individual, wherein the phenotype is determined by the presence of at least one mutation from a set of mutations. In some embodiments, the method includes obtaining DNA or RNA measurements for a sample of DNA or RNA from one or more cells of an individual, wherein one or more cells are presumed to have a phenotype; and analyzing DNA or RNA measurements to determine, for each mutation in a set of mutations, the probability that at least one of the cells has that mutation. In some embodiments, the method includes determining that an individual has a phenotype if either (i) for at least one of the mutations, the probability that at least one of the cells contains the mutation exceeds a threshold, or (ii) for at least one of the mutations, the probability that at least one of the cells has these mutations is less than a threshold value, and for a plurality of mutations, the combined probability that at least one of the cells has at least one of the mutations is greater threshold. In some embodiments, one or more cells have a subset or all of the mutations in the set of mutations. In some embodiments, a subset of mutations is associated with cancer or an increased risk of cancer. In some embodiments, the set of mutations includes a subset or all of the mutations in class M of cancer mutations (Ciriello, Nat Genet. 45(10):1127-1133, 2013, doi: 10.1038/ng.2762, incorporated herein by reference in its entirety). In some embodiments, the set of mutations includes a subset or all of the mutations in class C cancer mutations (Ciriello, supra ) . In some embodiments, the sample includes cell-free DNA or RNA. In some embodiments, the DNA or RNA measurements include measurements (eg, the amount of each allele at each locus) at a set of polymorphic loci on one or more chromosomes or chromosome segments of interest.

[539] Approximate combination of methods

[540] To improve the accuracy of the results, two or more methods (eg, any of the methods of the present invention or any known method) are used to detect the presence or absence of ICH. In some embodiments, one or more methods are performed to analyze a factor (eg, any of the methods presented herein or any known method) indicating the presence or absence of a disease or disorder, or an increased risk of a disease or disorder.

[541] In some embodiments, standard mathematical techniques are used to calculate covariance and/or correlation between two or more methods. Standard mathematical techniques can also be used to determine the cumulative probability of a particular hypothesis based on two or more tests. Examples of methods include meta-analysis, Fisher's combined probability test for independent tests, Brown's method for combining dependent p-values with known covariance, and Costa's method for combining dependent p-values with unknown covariance. In cases where the probabilities are determined by the first method, which is orthogonal to or unrelated to how the probability is determined for the second method, combining the probabilities is simple and can be done by multiplication and normalization, or by using a formula such as:

[542] R _comb = R ₁ R ₂ /[R ₁ R ₂ + (1-R ₁ )(1-R ₂ )]

[543] R _comb is the combined probability, and R1 and R2 are the individual probabilities. For example, if the probability of trisomy by mode 1 is 90% and the probability of trisomy by mode 2 is 95%, then combining the results of the two methods allows the clinician to conclude that the fetus is trisomic with a probability of (0.90)(0.95)/[ (0.90)(0.95) + (1 - 0.90)(1 - 0.95)] = 99.42%. In cases where the first and second methods are not orthogonal, that is, when there is a correlation between the two methods, the probabilities can still be combined.

[544] Exemplary methods for analyzing multiple factors or variables are disclosed in US Patent No. 8,024,128, issued September 20, 2011; US Patent Application Publication No. 2007/0027636, filed July 31, 2006; and U.S. Patent Application Publication No. 2007/0178501, filed December 6, 2006, each of which is incorporated herein by reference in its entirety.

[545] In various embodiments, the cumulative probability of a particular hypothesis or diagnosis is greater than 80%, 85%, 90%, 92%, 94%, 96%, 98%, 99%, or 99.9%, or greater than some other threshold.

[546] Detection limit

[547] As demonstrated by the experiments presented in the Examples section, the methods presented herein are capable of detecting average allelic imbalance in a sample with a detection or sensitivity limit of 0.45% AAI, which is the aneuploidy detection limit of the exemplary method of the present invention. Likewise, in some embodiments, the methods provided herein are capable of detecting an average allelic imbalance in a sample of 0.45, 0.5, 0.6, 0.7, 0.8. 0.9 or 1.0%. That is, the testing method is capable of detecting chromosomal aneuploidy in a sample up to 0.45, 0.5, 0.6, 0.7, 0.8. 0.9 or 1.0% AAI. As demonstrated by the experiments presented in the Examples section, the methods presented herein are capable of detecting the presence of NDCs in a sample for at least some NDCs with a detection limit or sensitivity of 0.2%, which is the detection limit for at least several NDCs in one illustrative embodiment. Likewise, in some embodiments, the method can detect NVGs or AAI NVGs at rates of 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8. 0.9 or 1.0%. That is, this testing method is capable of detecting NDCs in a sample with a detection limit of 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8. 0.9 or 1.0% of the total number of alleles in the chromosomal locus ONV.

[548] In some embodiments, the detection limit of a mutation (such as NNV or ICH) of the method of the present invention is less than or equal to 10, 5, 2, 1, 0.5, 0.1, 0.05, 0.01, or 0.005% . In some embodiments, the detection limit for a mutation (such as NNV or ICH) of the method of the invention is 15 to 0.005%, e.g., 10 to 0.005%, 10 to 0.01%, 10 to 0.1%, 5 up to 0.005%, from 5 to 0.01%, from 5 to 0.1%, from 1 to 0.005%, from 1 to 0.01%, from 1 to 0.1%, from 0.5 to 0.005%, from 0.5 to 0.01%, from 0.5 to 0.1% or from 0.1 to 0.01 inclusive.

[549] In some embodiments, the limit of detection is such that a mutation (e.g., CNV or ICH) is detected (or can be detected) that is present in an amount of less than or equal to 10, 5, 2, 1, 0.5, 0. 1, 0.05, 0.01, or 0.005% of the DNA or RNA molecule with that locus in a sample (e.g., a scDNA or scRNA sample). For example, a mutation may be detected even if less than or equal to 10, 5, 2, 1, 0.5, 0.1, 0.05, 0.01, or 0.005% of DNA or RNA molecules having that locus have that mutation at a locus (instead of, for example, a wild-type version of the locus or a non-mutant version of the locus or another mutation at that locus). In some embodiments, the limit of detection is such that a mutation (such as ONV or ICH) is detected (or capable of being detected) present in an amount less than or equal to 10, 5, 2, 1, 0.5, 0.1, 0, 05, 0.01 or 0.005% of the DNA or RNA molecule in a sample (such as a scDNA or scRNA sample). In some embodiments, where the ICH is a deletion, the deletion may be detected even if it is only present in an amount less than or equal to 10, 5, 2, 1, 0.5, 0.1, 0.05, 0.01 or 0.005% of DNA or RNA molecules having a region of interest in the sample, which may or may not contain a deletion. In some embodiments, where the ICH is a deletion, the deletion may be detected even if it is only present in an amount less than or equal to 10, 5, 2, 1, 0.5, 0.1, 0.05, 0.01 or 0.005% of DNA or RNA molecules in the sample. In some embodiments in which the ICH is a duplication, the duplication can be detected even if the additional duplicated DNA or RNA present is less than or equal to 10, 5, 2, 1, 0.5, 0.1, 0.05, 0.01 , or 0.005% of DNA or RNA molecules having a region of interest that may or may not be duplicated in the sample. In some embodiments, where the ICH is a duplication, the duplication can be detected even if the additional duplicated DNA or RNA present is less than or equal to 10, 5, 2, 1, 0.5, 0.1, 0.05, 0.01, or 0.005% of DNA or RNA molecules in the sample.

[550] Sample samples

[551] In some embodiments of any aspect of the invention, the sample includes cellular and/or extracellular genetic material from cells suspected of having a deletion or duplication, such as cells suspected of being cancerous. In some embodiments, the sample includes any tissue or biological fluid suspected of containing cells, DNA or RNA having a deletion or duplication, such as tumors or other samples that include cancer cells, DNA or RNA. Genetic measurements used as part of these methods can be performed on any sample containing DNA or RNA, such as, but not limited to, tissue, blood, serum, plasma, urine, hair, tears, saliva, skin, nails, feces, bile, lymph, cervical mucus, sperm, tumor or other cells or materials containing nucleic acids. Samples can include any type of cell, or DNA or RNA from any type of cell can be used (for example, cells of any organ or tissue suspected of being cancerous, or neurons). In some embodiments, the sample includes nuclear and/or mitochondrial DNA. In some embodiments, the sample is taken from any of the target individuals provided herein. In some embodiments, the target individual is a patient with cancer.

[552] Exemplary samples include samples containing scDNA or scRNA. In some embodiments, the scDNA is available for analysis without the need for a cell lysis step. Cell-free DNA can be obtained from various tissues, such as tissues that are in liquid form, such as blood, plasma, lymph, ascites fluid or cerebrospinal fluid. In some cases, cDNA consists of DNA obtained from fetal cells. In some cases, scDNA is isolated from plasma isolated from whole blood that has been centrifuged to remove cellular material. The scDNA can be a mixture of DNA obtained from target cells (such as cancer cells) and non-target cells (such as non-cancerous cells).

[553] In some embodiments, the sample contains or is believed to contain a mixture of DNA (or RNA), such as a mixture of DNA (or RNA) derived from cancer cells and DNA (or RNA) derived from non-cancerous (ie, normal) cells. In some embodiments, at least 0.5, 1, 3, 5, 7, 10, 15, 20, 30, 40, 50, 60, 70, 80, 90, 92, 94, 95, 96, 98, 99 or 100% of the cells in the sample are cancer cells. In some embodiments, at least 0.5, 1, 3, 5, 7, 10, 15, 20, 30, 40, 50, 60, 70, 80, 90, 92, 94, 95, 96, 98, 99 or 100% of the DNA (eg, scDNA) or RNA (eg, scRNA) in the sample comes from cancer cells. In various embodiments, the percentage of cells in the sample that are cancerous is from 0.5 to 99%, for example, from 1 to 95%, from 5 to 95%, from 10 to 90%, from 5 to 70%, from 10 to 70%, from 20 to 90%, or from 20 to 70% inclusive. In some embodiments, the sample is enriched with cancer cells or DNA or RNA of cancer cells. In some embodiments, in which the sample is enriched in cancer cells by at least 0.5, 1, 2, 3, 4, 5, 6, 7, 10, 15, 20, 30, 40, 50, 60, 70, 80, 90, 92, 94, 95, 96, 98, 99 or 100% of the cells in the enriched sample are cancer cells. In some embodiments, wherein the sample is enriched in DNA or RNA from cancer cells by at least 0.5, 1, 2, 3, 4, 5, 6, 7, 10, 15, 20, 30, 40, 50, 60, 70, 80, 90, 92, 94, 95, 96, 98, 99, or 100% of the DNA or RNA in the enriched sample is from cancer cells. In some embodiments, cell sorting (such as Fluorescent Activated Cell Sorting (FACS)) is used to enrich for cancer cells (Barteneva et. al ., Biochim Biophys Acta., 1836(1):105-22, Aug 2013. doi: 10.1016/j.bbcan.2013.02.004. Epub 2013 Feb 24, and Ibrahim et al ., Adv Biochem Eng Biotechnol. 106:19-39, 2007, each of these documents is incorporated herein by reference in its entirety).

[554] In some embodiments, the sample is enriched with fetal cells. In some embodiments, in which the sample is enriched in fetal cells, at least 0.5, 1, 2, 3, 4, 5, 6, 7% or more of the cells in the enriched sample are fetal cells. In some embodiments, the percentage of cells in the sample that are fetal cells is from 0.5 to 100%, for example, from 1 to 99%, from 5 to 95%, from 10 to 95%, from 10 to 95%, from 20 to 90% or from 30 to 70% inclusive. In some embodiments, the sample is enriched with fetal DNA. In some embodiments, in which the sample is enriched with fetal DNA, at least 0.5, 1, 2, 3, 4, 5, 6, 7% or more of the DNA in the enriched sample is fetal DNA. In some embodiments, the percentage of DNA in the sample that is fetal DNA is from 0.5 to 100%, for example, from 1 to 99%, from 5 to 95%, from 10 to 95%, from 10 to 95%, from 20 to 90% or from 30 to 70% inclusive.

[555] In some embodiments, the sample includes a single cell or includes DNA and/or RNA from a single cell. In some embodiments, multiple single cells (eg, at least 5, 10, 20, 30, 40, or 50 cells from the same subject or from different subjects) are analyzed in parallel. In some embodiments, cells from multiple samples from a single individual are pooled, which reduces the amount of work compared to analyzing samples individually. Pooling multiple samples can also allow multiple tissues to be tested for cancer simultaneously (which can be used to provide either more thorough cancer screening or to determine whether cancer may have metastasized to other tissues).

[556] In some embodiments, the sample contains a single cell or a small number of cells, such as 2, 3, 5, 6, 7, 8, 9, or 10 cells. In some embodiments, the sample contains from 1 to 100, from 100 to 500, or from 500 to 1000 cells, inclusive. In some embodiments, the sample contains from 1 to 10 picograms, from 10 to 100 picograms, from 100 picograms to 1 nanogram, from 1 to 10 nanograms, from 10 to 100 nanograms, or from 100 nanograms to 1 microgram of RNA and/or DNA, inclusive.

[557] In some embodiments, the sample is embedded in paraffin. In some embodiments, the sample is stored with a preservative, such as formaldehyde, and optionally embedded in paraffin, which can cause cross-linking of the DNA so that less of it is available for PCR. In some embodiments, the sample is a formaldehyde fixed-paraffin embedded (FFPE) sample. In some embodiments, the sample is a fresh sample (eg, a sample obtained from 1 or 2 days of analysis). In some embodiments, the sample is frozen prior to analysis. In some embodiments, the sample is a historical sample.

[558] These samples can be used in any of the methods of the invention.

[559] Example Sample Preparation Methods

[560] In some embodiments, the method includes isolating or purifying DNA and/or RNA. A number of standard procedures are known to achieve this goal. In some embodiments, the sample may be centrifuged to separate the different layers. In some embodiments, DNA or RNA can be isolated using filtration. In some embodiments, preparation of DNA or RNA may involve amplification, separation, chromatographic purification, liquid separation, isolation, preferential enrichment, preferential amplification, targeted amplification, or any of a number of other methods known in the art or presented herein. In some embodiments, RNase is used to degrade the RNA to isolate DNA. In some embodiments, a DNase (such as DNase I from Invitrogen, Carlsbad, CA, USA) is used to isolate RNA to degrade DNA. In some embodiments, the RNeasy mini kit (Qiagen) is used to isolate RNA according to the manufacturer's protocol. In some embodiments, small RNA molecules are isolated using the mirVana PARIS kit (Ambion, Austin, TX, USA) according to the manufacturer's protocol (Gu et al ., J. Neurochem. 122:641-649, 2012, entire document incorporated here via link). RNA concentration and purity can optionally be determined using Nanovue (GE Healthcare, Piscataway, NJ, USA), and RNA integrity can optionally be determined using a 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA) (Gu et al ., J. Neurochem 122:641–649, 2012, incorporated herein by reference in its entirety). In some embodiments, TRIzol or RNAlater (Ambion) is used to stabilize the RNA during storage.

[561] In some embodiments, universal tagged adapters are added to create the library. Before ligation, the DNA sample can be blunt ended and then a single adenosine base is added to the 3' end. Before ligation, the DNA can be digested with a restriction enzyme or some other digestion method. During ligation, the 3' adenosine of the sample fragments and the complementary 3' tyrosine sticky end of the adapter can enhance the ligation efficiency. In some embodiments, ligation of the adapter is performed using a ligation kit found in the AGILENT SURESELECT kit. In some embodiments, the library is amplified using universal primers. In one embodiment, the amplified library is fractionated by size separation or using products such as AGENCOURT AMPURE microbeads or other similar methods. In some embodiments, PCR amplification is used to amplify target loci. In some embodiments, the amplified DNA is sequenced (eg, sequencing using an ILLUMINA IIGAX or HiSeq sequencer). In some embodiments, the amplified DNA is sequenced from each end of the amplified DNA to reduce sequencing errors. If there is a sequence error at a particular base when sequencing one end of the amplified DNA, then a sequence error at the complementary base when sequencing the other side of the amplified DNA is less likely to occur (compared to sequencing the same end of the amplified DNA multiple times).

[562] In some embodiments, whole genome amplification (WGA) is used to amplify a nucleic acid sample. There are a number of methods for WGA: ligation-mediated PCR (LM-PCR), PCR with degenerate nucleotide primers (DOP-PCR) and multiple displacement amplification (MDA). In LM-PCR, short DNA sequences called adapters are ligated to the blunt ends of DNA. These adapters contain universal amplification sequences that are used to amplify DNA using PCR. In DOP-PCR, the first round of annealing and PCR uses random primers that also contain universal amplification sequences. A second round of PCR is then used to further amplify the sequences using universal primer sequences. MDA uses phi-29 polymerase, which is a high-throughput and non-specific enzyme that replicates DNA and is used for single cell analysis. In some embodiments, WGA is not performed.

[563] In some embodiments, selective amplification is used to amplify or enrich target loci. In some embodiments, the amplification and/or selective enrichment method may include PCR, such as ligation-mediated PCR, fragment capture hybridization, invertible molecular probes, or other circularizing probes. Some embodiments use real-time quantitative PCR (RT-qPCR), digital PCR, or emulsion PCR, single-base extension reaction followed by mass spectrometry (Hung et al., J Clin Pathol 62 :308-313, 2009, incorporated herein by reference in its entirety). In some embodiments, capture by hybridization with hybrid capture probes is used to preferentially enrich DNA. In some embodiments, amplification or selective enrichment methods may involve the use of probes in which, when properly hybridized to the target sequence, the 3' end or 5' end of the nucleotide probe is separated from the polymorphic site of the polymorphic allele by a small number of nucleotides. This separation reduces the preferential amplification of one allele, called allele bias. This is an improvement over methods that involve the use of probes in which the 3' end or 5' end of the correctly hybridized probe is immediately adjacent or very close to the polymorphic site of the allele. In one embodiment, probes in which the hybridizing region may or necessarily contains a polymorphic site are excluded. Polymorphic sites at the site of hybridization may cause unequal hybridization or even suppress hybridization of some alleles, resulting in preferential amplification of certain alleles. These embodiments are improvements over other methods that involve targeted amplification and/or selective enrichment in that they better preserve the sample's original allele frequencies at each polymorphic locus, regardless of whether the sample is a pure genomic sample from one individual or from a mixture of individuals.

[564] In some embodiments, PCR (referred to as mini-PCR) is used to generate very short amplicons (U.S. Patent Application No. 13/683604, filed November 21, 2012, U.S. Patent Application Publication No. 2013/0123120, U.S. Patent Application No. 13/300235, filed November 18, 2011, US Patent Application Publication No. 2012/0270212, filed November 18, 2011, and US Provisional Patent Application Serial No. 61/994791, filed May 16, 2014, each in its entirety incorporated herein by reference). scDNA (such as cancer DNA released due to necrosis or apoptosis) is highly fragmented. For fetal scDNA, fragment sizes are approximately Gaussian with a mean of 160 bp, a standard deviation of 15 bp, and a minimum size of approximately 100 bp. and a maximum size of approximately 220 bp. The polymorphic site of one particular target locus can occupy any position among the various fragments originating from that locus, from beginning to end. Because scDNA fragments are short, the probability that both primer sites are present and the probability that a fragment of length L contains both forward and reverse primer sites is the ratio of the amplicon length to the fragment length. Under ideal conditions, assays in which the amplicon is 45, 50, 55, 60, 65, or 70 bp will successfully amplify from, respectively, 72, 69, 66, 63, 59, or 56% of the available template fragment molecules. In some embodiments, most preferably for scDNA from samples of individuals suspected of having cancer, the scDNA is amplified using primers that produce a maximum amplicon length of 85, 80, 75, or 70 bp, and in some preferred embodiments, 75 bp .n., and which have a melting point of from 50 to 65 °C, and in some preferred embodiments from 54 to 60.5 °C. Amplicon length is the distance between the 5' ends of the forward and reverse primed sites. Amplicon lengths that are shorter than those typically used by methods known in the art can result in more efficient measurements of desired polymorphic loci while only requiring short sequence reads. In one embodiment, a significant portion of the amplicons are less than 100 bp, less than 90 bp, less than 80 bp, less than 70 bp, less than 65 bp, less than 60 bp ., less than 55 bp, less than 50 bp. or less than 45 bp

[565] In some embodiments, amplification is performed using direct multiplex PCR , sequential PCR, nested PCR, double nested PCR, one-and-a-half-way nested PCR, fully nested PCR, one-way fully nested PCR, one-way nested PCR, hemi-nested PCR, hemi-nested PCR , triple hemi-nested PCR, semi-nested PCR, one-way semi-nested PCR, reverse semi-nested PCR or one-way PCR, which are described in US Patent Application No. 13/683604, filed November 21, 2012, Application Publication No. US Patent No. 2013/0123120, US Patent Application No. 13/300235, filed November 18, 2011, US Patent Application Publication No. 2012/0270212 and US Provisional Patent Application Serial No. 61/994791, filed May 16, 2014, documents in full incorporated herein by reference. If necessary, any of these methods can be used for mini-PCR.

[566] If necessary, the extension step in PCR amplification can be time-limited to reduce amplification from fragments longer than 200 nucleotides, 300 nucleotides, 400 nucleotides, 500 nucleotides, or 1000 nucleotides. This may result in enrichment of fragmented or shorter DNA (eg, fetal DNA or DNA from cancer cells that have undergone apoptosis or necrosis) and improved test results.

[567] In some embodiments, multiplex PCR is used. In some embodiments, a method for amplifying target loci in a nucleic acid sample comprises (i) contacting the nucleic acid sample with a library of primers that simultaneously hybridize to at least 100; 200; 500; 750; 1000; 2000; 5000; 7500; 10,000; 20,000; 25000; 30,000; 40000; 50000; 75000; or 100,000 different target loci to obtain a reaction mixture; and (ii) subjecting the reaction mixture to primer extension reaction conditions (such as PCR conditions) to produce amplified products that include the target amplicons. In some embodiments, at least 50, 60, 70, 80, 90, 95, 96, 97, 98, 99, or 99.5% of the target loci are amplified. In various embodiments, less than 60, 50, 40, 30, 20, 10, 5, 4, 3, 2, 1, 0.5, 0.25, 0.1, or 0.05% of the amplification products are primer dimers. In some embodiments, the primers are in solution (eg, dissolved in a liquid phase rather than in a solid phase). In some embodiments, the primers are in solution and are not immobilized on a solid support. In some embodiments, the primers are not part of the microarray. In some embodiments, the primers do not include molecular invertible probes (MIPs).

[568] In some embodiments, two or more (eg, 3 or 4) target amplicons (such as amplicons from a mini-PCR method disclosed herein) are ligated together and the ligated products are then sequenced. Combining multiple amplicons into a single ligation product increases the efficiency of the subsequent sequencing step. In some embodiments, the target amplicons are less than 150, 100, 90, 75, or 50 base pairs in length before ligation. Selective enrichment and/or amplification may involve tagging each individual molecule with various tags, molecular barcodes, amplification tags, and/or sequencing tags. In some embodiments, the amplified products are analyzed by sequencing (eg, high-throughput sequencing) or hybridization to an array, such as an SNP array, an ILLUMINA INFINIUM array, or an AFFYMETRIX gene chip. In some embodiments, nanopore sequencing is used, such as nanopore sequencing technology developed by Genia (see, for example, the World Wide Web at geniachip.com/technology, the document is incorporated herein by reference in its entirety). In some embodiments, duplex sequencing is used (Schmitt et al., “Detection of ultra-rare mutations by next-generation sequencing,” Proc Natl Acad Sci U S A. 109(36): 14508-14513, 2012, incorporated herein in its entirety by links). This approach significantly reduces errors by independently labeling and sequencing each of the two strands of the DNA duplex. Because the two strands are complementary, true mutations are found at the same position on both strands. In contrast, PCR or sequencing errors result in mutations in only one strand and can thus be ruled out as a technical error. In some embodiments, the method includes tagging both strands of duplex DNA with a random but complementary double-stranded nucleotide sequence, called a duplex tag. Double-stranded tag sequences are incorporated into standard sequencing adapters by first introducing a single-stranded random nucleotide sequence into one strand of the adapter and then extending the opposite strand by DNA polymerase to produce a complementary double-stranded tag. After ligation of tagged adapters to cut DNA, the individually tagged strands are amplified by PCR from asymmetric primer sites on the tails of the adapter and subjected to paired-end sequencing. In some embodiments, the sample (eg, a DNA or RNA sample) is divided into multiple fractions, such as into different wells (eg, wells from a WaferGenSmartChip). Dividing the sample into different fractions (e.g., at least 5, 10, 20, 50, 75, 100, 150, 200, or 300 fractions) can increase the sensitivity of the assay because some of the wells have a higher percentage of molecules with the mutation than the overall sample. . In some embodiments, each fraction contains less than 500, 400, 200, 100, 50, 20, 10, 5, 2, or 1 molecule of DNA or RNA. In some embodiments, the molecules in each fraction are sequenced separately. In some embodiments, the same barcode (e.g., random or non-human sequence) is added to all molecules in the same fraction (e.g., by amplification with a primer containing the barcode or by ligation barcode), and different barcodes are added to the molecules in different fractions. Barcoded molecules can be combined and sequenced together. In some embodiments, the molecules are amplified before pooling and sequencing, for example, using nested PCR. In some embodiments, one forward and two reverse primers or two forward and one reverse primers are used.

[569] In some embodiments, a mutation is (or may be) detected (e.g., ONV or ICH) that is present in less than 10, 5, 2, 1, 0.5, 0.1, 0.05, 0. 01 or 0.005% of the DNA or RNA molecules in a sample (e.g., a scDNA or scRNA sample). In some embodiments, a mutation is (or may be) detected (e.g., ONV or ICH) that is present in less than 1000, 500, 100, 50, 20, 10, 5, 4, 3, or 2 original DNA or RNA molecules ( before amplification) in a sample (such as a scDNA sample or scRNA from, for example, a blood sample). In some embodiments, a mutation is (or may be) detected (e.g., ONV or ICH) that is present in only 1 original DNA or RNA molecule (prior to amplification) in a sample (e.g., a scDNA or scRNA sample from, e.g., a blood sample ).

[570] For example, if the detection limit of a mutation (such as a single nucleotide variant (SNV)) is 0.1%, the mutation present in 0.01% can be detected by dividing the fraction into several fractions, such as 100 wells. In most wells there are no copies of the mutation. For several wells with a mutation, the mutation is in a much higher percentage of reads. In one example, there are 20,000 initial copies of DNA from a target locus, and two of these copies include the ONV of interest. If the sample is divided into 100 wells, 98 wells contain OBC and 2 wells contain 0.5%. The DNA in each well can be encoded, amplified, combined with DNA from other wells, and sequenced. Wells without OCBs can be used to measure the background amplification/sequencing error rate to determine whether the signal from outlier wells exceeds the background noise level.

[571] In some embodiments, amplified products are detected using an array, such as an array, particularly a microarray with probes for one or more chromosomes of interest (eg, chromosomes 13, 18, 21, X, Y, or any combination thereof). It is clear, for example, that one can use commercially available SNP detection microarrays, such as, for example, the Illumina (San Diego, CA) GoldenGate, DASL, Infinium, or CytoSNP-12 genotyping assay, or a SNP detection product from Affymetrix, such as the OncoScan microarray .

[572] In some embodiments involving sequencing, read depth is the number of sequencing reads that correspond to a given locus. The read depth can be normalized by the total number of reads. In some embodiments, for a sample read depth, the read depth is the average read depth across the target loci. In some embodiments, for the read depth of a locus, the read depth is the number of reads measured by sequencer mapping of that locus. In general, the greater the read depth of a locus, the closer the ratio of alleles at the locus is to the ratio of alleles in the original DNA sample. Reading depth can be expressed in many different ways, including but not limited to percentage or proportion. So, for example, in a highly parallel DNA sequencer such as Illumina HISEQ, which for example produces a sequence of 1 million clones, sequencing a single locus 3000 times results in a read depth of 3000 reads at that locus. The proportion of reads at this locus is 3000 divided by 1 million total reads, or 0.3% of the total reads.

[573] In some embodiments, allelic data is obtained, wherein the allelic data includes quantitative measurements indicating the number of copies of a particular allele of a polymorphic locus. In some embodiments, the allelic data includes quantitative measurements indicating the copy number of each of the alleles observed at the polymorphic locus. Typically, quantitative measurements are obtained for all possible alleles of a polymorphic locus of interest. For example, any of the methods presented in the previous paragraphs for identifying an allele for a SNP or SNV locus, such as, for example, microarrays, qPCR, DNA sequencing, such as high-throughput DNA sequencing, can be used to generate quantitative copy number measurements of a particular allele of a polymorphic locus. This quantitative measurement is referred to herein as allele frequency data or measured genetic allele data. Methods using allelic data are sometimes called quantitative allelic methods; this is in contrast to quantitative methods, which use exclusively quantitative data from non-polymorphic loci or from polymorphic loci but without regard to allelic identity. When allelic data is measured using high-throughput sequencing, the allelic data typically includes the number of reads of each allele mapping to the locus of interest.

[574] In some embodiments, non-allelic data is obtained, wherein the non-allelic data includes quantitative measurements indicating the copy number of a particular locus. A locus may be polymorphic or non-polymorphic. In some embodiments, when a locus is not polymorphic, the non-allelic data does not contain information about the relative or absolute number of individual alleles that may be present at the locus. Methods that use only non-allelic data (that is, quantitative data on non-polymorphic alleles or quantitative data on polymorphic loci, but without taking into account the allelic identity of each fragment) are called quantitative methods. Typically, quantitative measurements are obtained for all possible alleles at a polymorphic locus of interest, with one value associated with the measured value for all alleles at that locus as a whole. Non-allelic data for a polymorphic locus can be obtained by summing the allele score for each allele at that locus. When allelic data is measured using high-throughput sequencing, nonallelic data typically includes the number of mapping reads per locus of interest. Sequencing measurements can indicate the relative and/or absolute amount of each of the alleles present at a locus, and non-allelic data includes the sum of reads mapped to the locus, regardless of allelic identity. In some embodiments, the same set of sequencing measurements may be used to obtain both allelic and non-allelic data. In some embodiments, the allelic data is used as part of a method for determining the copy number on a chromosome of interest, and the resulting non-allelic data can be used as part of another method for determining the copy number on a chromosome of interest. In some embodiments, the two methods are statistically orthogonal and combined to provide a more accurate determination of the copy number on the chromosome of interest.

[575] In some embodiments, obtaining genetic data includes (i) obtaining DNA sequence information using laboratory methods, for example, using an automated high-throughput DNA sequencer, or (ii) obtaining information that has previously been obtained by laboratory methods, wherein the information transmitted electronically, such as by computer over the Internet or electronic transmission from a sequencing device.

[576] Additional exemplary methods for sample preparation, amplification and quantitation are described in US Patent Application No. 13/683604, filed November 21, 2012 (US Patent Application Publication No. 2013/0123120 and US Provisional Patent Application Serial No. 61/ 994791, filed May 16, 2014, which is incorporated herein by reference in its entirety). These methods can be used to analyze any of the samples disclosed herein.

[577] Exemplary Methods for Quantifying Cell-Free DNA

[578] If necessary, this amount or concentration of scDNA or scRNA can be measured by standard methods. In some embodiments, the amount or concentration of free cell mitochondrial DNA (scmDNA) is determined. In some embodiments, the amount or concentration of cell-free DNA derived from nuclear DNA (scDNA) is determined. In some embodiments, the amount or concentration of csmDNA and csDNA is determined simultaneously.

[579] In some embodiments, qPCR is used to measure csDNA and/or ccmDNA (Kohler et al. “ Levels of plasma circulating cell free nuclear and mitochondrial DNA as potential biomarkers for breast tumors.” Mol Cancer 8:105, 2009, 8: doi:10.1186/1476-4598-8-105, document incorporated herein by reference in its entirety). For example, one or more loci from csDNA (such as glyceraldehyde-3-phosphate dehydrogenase, GAPDH) and one or more loci from csDNA (ATPase 8, MTATP 8) can be measured using multiplex qPCR. In some embodiments, fluorescently tagged PCR is used to measure csDNA and/or csmDNA (Schwarzenbach et al., “ Evaluation of cell-free tumor DNA and RNA in patients with breast cancer and benign breast disease.” Mol Biosys 7:2848-2854 , 2011, document incorporated herein by reference in its entirety). If necessary, the normality distribution of the data can be determined using standard methods such as the Shapiro-Wilk test. If necessary, scDNA and mDNA levels can be compared using standard methods such as the Mann-Whitney U test. In some embodiments, cDNA and/or mDNA levels are compared with other established prognostic factors using standard methods such as the Mann-Whitney U test or the Kruskal-Wallis test.

[580] Exemplary methods for amplification, quantitation and analysis of RNA

[581] Any of the following typical methods can be used to amplify and optionally quantify RNA, such as scRNA, cellular RNA, cytoplasmic RNA, coding cytoplasmic RNA, non-coding cytoplasmic RNA, mRNA, siRNA, mitochondrial RNA, rRNA, or tRNA. In some embodiments, the miRNA is any of the miRNA molecules listed in the miRBase database available on the World Wide Web at mirbase.org, the document being incorporated herein by reference in its entirety. Exemplary miRNA molecules include miR-509; miR-21 and miR-146a.

[582] In some embodiments, reverse-transcriptase multiplex ligation-dependent probe amplification (RT-MLPA) is used to amplify RNA. In some embodiments, each set of hybridization probes consists of two short synthetic oligonucleotides spanning the SNP and one long oligonucleotide (Li et al ., Arch Gynecol Obstet. “Development of noninvasive prenatal diagnosis of trisomy 21 by RT-MLPA with a new set of SNP markers,” July 5, 2013, DOI 10.1007/s00404-013-2926-5; Schouten et al . “Relative quantification of 40 nucleic acid sequences by multiplex ligation-dependent probe amplification.” Nucleic Acids Res 30:e57, 2002 ; Deng et al. (2011) “Non-invasive prenatal diagnosis of trisomy 21 by reverse transcriptase multiplex ligation-dependent probe amplification,” Clin, Chem. Lab Med. 49:641-646, 2011, each of these documents is included here in its entirety via link).

[583] In some embodiments, the RNA is amplified using reverse transcriptase PCR. In some embodiments, the RNA is amplified using real-time reverse transcriptase PCR, such as one-step real-time reverse transcriptase PCR with SYBR GREEN I, as previously described. (Li et al ., Arch Gynecol Obstet. “Development of noninvasive prenatal diagnosis of trisomy 21 by RT-MLPA with a new set of SNP markers,” July 5, 2013, DOI 10.1007/s00404-013-2926-5; Lo et al., “Plasma placental RNA allelic ratio permits noninvasive prenatal chromosomal aneuploidy detection,” Nat Med 13:218-223, 2007; Tsui et al., Systematic micro-array based identification of placental mRNA in maternal plasma: towards non-invasive prenatal gene expression profiling, J Med Genet 41:461–467, 2004; Gu et al ., J Neurochem 122:641–649, 2012, each of which is incorporated herein by reference in its entirety).

[584] In some embodiments, a microarray is used to detect RNA. For example, the human miRNA microarray from Agilent Technologies can be used according to the manufacturer's protocol. Briefly, the isolated RNA is dephosphorylated and ligated to pCp-Cy3. Labeled RNA is purified and hybridized to miRNA templates containing mature human miRNA probes based on Sanger release of miRBase 14.0. The arrays are washed and scanned using a microarray scanner (G2565BA, Agilent Technologies). The intensity of each hybridization signal is assessed using Agilent extraction software v9.5.3. Labeling, hybridization, and scanning can be performed according to Agilent microarray protocols for the microarray system (Gu et al ., J. Neurochem. 122:641-649, 2012, incorporated herein by reference in its entirety).

[585] In some embodiments, a TaqMan assay is used to detect RNA. An example test is the TaqMan Array Human MicroRNA v1.0 Panel (Early Access) (Applied Biosystems), which contains 157 TaqMan MicroRNA Assays, including appropriate reverse transcription primers, PCR primers, and TaqMan probe (Chim et al., “Detection and characterization of placental microRNAs in maternal plasma,” Clin Chem. 54(3):482–90, 2008, incorporated herein by reference in its entirety).

[586] If necessary, the mRNA splicing pattern of one or more mRNAs can be determined using standard methods (Fackenthal1 and Godley, Disease Models & Mechanisms 1: 37-42, 2008, doi:10.1242/dmm.000331, incorporated herein in its entirety via link). For example, high-density microarrays and/or high-throughput DNA sequencing can be used to detect mRNA splice variants.

[587] In some embodiments, whole transcriptome fractional or array sequencing is used to measure the transcriptome.

[588] Exemplary amplification methods

[589] Improved PCR amplification methods have also been developed that minimize or prevent interference due to amplification of nearby or adjacent target loci in the same reaction volume (eg, part of a multiplex PCR reaction of a sample that simultaneously amplify all target loci). These methods can be used to simultaneously amplify nearby or adjacent target loci, which is faster and cheaper than having to separate nearby target loci into different reaction volumes so they can be amplified separately to avoid interference.

[590] In some embodiments, amplification of target loci is performed using a polymerase (eg, DNA polymerase, RNA polymerase, or reverse transcriptase) with low 5'→3' exonuclease activity and/or low strand displacement activity. In some embodiments, a low level of 5'→3' exonuclease reduces or prevents degradation of a nearby primer (eg, an unextended primer or a primer to which one or more nucleotides have been added during primer extension). In some embodiments, a low level of strand displacement activity reduces or prevents displacement of a nearby primer (eg, an unextended primer or a primer to which one or more nucleotides have been added during primer extension). In some embodiments, target loci that are adjacent to each other (e.g., there are no bases between the target loci) or nearby (e.g., loci that are within 50, 40, 30, 20, 15, 10, 9, 8, 7, 6, 5, 4, 3, 2 or 1 base). In some embodiments, the 3' end of one locus is within 50, 40, 30, 20, 15, 10, 9, 8, 7, 6, 5, 4, 3, 2, or 1 base from the 5' end of the next locus located further down.

[591] In some embodiments, at least 100, 200, 500, 750, 1000 are amplified; 2000; 5000; 7500; 10000; 20000; 25000; 30000; 40000; 50000; 75000; or 100,000 different target loci, for example, by simultaneous amplification in a single reaction volume. In some embodiments, at least 50, 60, 70, 80, 90, 95, 96, 97, 98, 99, or 99.5% of the amplified products are target amplicons. In various embodiments, the amount of amplified products that are target amplicons is from 50 to 99.5%, for example, from 60 to 99%, from 70 to 98%, from 80 to 98%, from 90 to 99.5%, or from 95 to 99.5% inclusive. In some embodiments, at least 50, 60, 70, 80, 90, 95, 96, 97, 98, 99, or 99.5% of the target loci are amplified (e.g., at least 5, 10, 20, 30, 50 are amplified or 100-fold compared to the amount before amplification), for example, by simultaneous amplification in one reaction volume. In various embodiments, the number of target loci that are amplified (e.g., amplified by at least 5, 10, 20, 30, 50, or 100 times the amount before amplification) is from 50 to 99.5%, such as from 60 up to 99%, from 70 to 98%, from 80 to 99%, from 90 to 99.5%, from 95 to 99.9% or from 98 to 99.99% inclusive. In some embodiments, fewer non-target amplicons are produced, such as fewer amplicons generated from the forward primer of the first primer pair and the reverse primer of the second primer pair. Such unwanted off-target amplicons may be generated using prior amplification methods if, for example, the reverse primer from the first primer pair and/or the forward primer from the second primer pair are disrupted and/or replaced.

[592] In some embodiments, these methods allow the use of longer extension times because the polymerase bound to the primer being extended is less likely to degrade and/or displace an adjacent primer (such as the next downstream primer) at low concentrations of 5'→3' exonuclease and/or low chain displacement polymerase activity. In various embodiments, reaction conditions (such as extension time and temperature) are used such that the rate of extension by the polymerase allows the number of nucleotides added to the extension primer to be equal to or greater than 80, 90, 95, 100, 110, 120, 130, 140 , 150, 175, or 200% of the number of nucleotides between the 3' end of a primer binding site and the 5' end of the next downstream primer binding site on the same strand.

[593] In some embodiments, DNA polymerase is used to produce DNA amplicons using DNA as a template. In some embodiments, RNA polymerase is used to produce RNA amplicons using DNA as a template. In some embodiments, reverse transcriptase is used to generate cDNA amplicons using RNA as a template.

[594] In some embodiments, the low level of 5'→3' exonuclease polymerase is less than 80, 70, 60, 50, 40, 30, 20, 10, 5, 1, or 0.1% of the activity of the same amount of polymerase Thermus aquaticus under the same conditions ("Taq" polymerase, which is a commonly used DNA polymerasefrom thermophilic bacteria,P.D.B. 1BGX, EC 2.7.7.7, Muraliet al., “Crystal structure of Taq DNA polymerase in complex with an inhibitory Fab: the Fab is directed against an intermediate in the helix-coil dynamics of the enzyme,” Proc. Natl. Acad. Sci. USA 95:12562-12567, 1998, incorporated herein by reference in its entirety). In some embodiments, the low level of strand displacement polymerase activity is less than 80, 70, 60, 50, 40, 30, 20, 10, 5, 1, or 0.1% of the same amount of Tag polymerase under the same conditions.

[595] In some embodiments, the polymerase is a PUSHION DNA polymerase, such as PHUSION High Fidelity DNA Polymerase (M0530S, New England BioLabs, Inc.) or PHUSION Hot Start Flex (M0535S, New England BioLabs, Inc.; Frey and Suppman BioChemica 2:34-35, 1995; Chester and Marshak Analytical Biochemistry 209:284-290, 1993, each of which is incorporated herein by reference in its entirety). PHUSION DNA polymerase is a Pyrococcus -like enzyme fused to a processivity-enhancing domain. PHUSION DNA polymerase has 5´→3´ polymerase activity and 3´→5´ exonuclease activity, and generates blunt-ended products. PHUSION DNA polymerase lacks 5´→3´ exonuclease and strand displacement activity.

[596] In some embodiments, the polymerase is a high fidelity DNA polymeraseQ5®, such as Q5® High Fidelity (M0491S, New England BioLabs, Inc.) or Q5® Hot Start High-Fidelity (M0493S, New England BioLabs, Inc.). DNA polymerase Q5® High-Fidelity is a high-fidelity, thermostable DNA polymerase with 3´→5´ exonuclease activity fused to the processivity-enhancing domain of Sso7d. DNA polymerase Q5® High-Fidelity lacks 5´→3´ exonuclease and strand displacement activity.

[597] In some embodiments, the polymerase is a T4 DNA polymerase (M0203S, New England BioLabs, Inc.; Tabor and Struh. (1989). “DNA-Dependent DNA Polymerases,” In Ausebel et al. (Ed.), Current Protocols in Molecular Biology 3.5.10-3.5.12 New York: John Wiley & Sons, Inc., 1989 Sambrook et al. Molecular Cloning: A Laboratory Manual (2nd ed.), 5.44-5.47 Cold Spring Harbor : Cold Spring Harbor Laboratory Press, 1989, each of these documents is incorporated herein by reference in its entirety). T4 DNA polymerase catalyzes DNA synthesis in the 5´→3´ direction and requires a template and a primer. This enzyme has 3´→5´ exonuclease activity, which is much more active than that of DNA polymerase I. T4 DNA polymerase lacks 5´→3´ exonuclease and strand displacement activities.

[598] In some embodiments, the polymerase is Sulfolobus DNA polymerase IV (M0327S, New England BioLabs, Inc.; (Boudsocq, et al. (2001). Nucleic Acids Res., 29:4607-4616, 2001; McDonald, et al (2006). Nucleic Acids Res., 34:1102-1111, 2006, each of which is incorporated herein by reference in its entirety.) Sulfolobus DNA polymerase IV is a thermostable Y-family damage bypass DNA polymerase that efficiently synthesizes DNA through multiple DNA template lesions (McDonald, JP et al. (2006). Nucleic Acids Res., . 34, 1102-1111, incorporated herein by reference in its entirety). Sulfolobus DNA polymerase IV lacks 5´→3´ exonuclease activity and strand displacement activity.

[599] In some embodiments, if a primer binds a region to an SNP, the primer may bind and amplify different alleles with different efficiencies or may bind and amplify only one allele. For subjects who are heterozygous, one of the alleles cannot be amplified by the primer. In some embodiments, a primer is designed for each allele. For example, if there are two alleles (eg, a bi-allelic SNP), then two primers can be used to bind the same location of the target locus (eg, a forward primer to bind the "A" allele and a forward primer to bind the "B" allele). . Standard methods such as the dbSNP database can be used to locate known SNPs, such as SNP hotspots with high heterozygosity.

[600] In some embodiments, the amplicons are uniform in size. In some embodiments, the length range of the target amplicons is less than 100, 75, 50, 25, 15, 10, or 5 nucleotides. In some embodiments (such as amplification of target loci in fragmented DNA or RNA), the length of the target amplicons is from 50 to 100 nucleotides, for example, from 60 to 80 nucleotides, or from 60 to 75 nucleotides, inclusive. In some embodiments (such as amplifying multiple target loci within an exon or gene), the target amplicons are 100 to 500 nucleotides in length, such as 150 to 450 nucleotides, 200 to 400 nucleotides, 200 to 300 nucleotides, or 300 up to 400 nucleotides inclusive.

[601] In some embodiments, multiple target loci are simultaneously amplified using a primer pair that includes a forward and reverse primer for each target locus amplified in a given reaction volume. In some embodiments, one PCR cycle is performed with one primer for each target locus, and then a second PCR cycle is performed with a pair of primers for each target locus. For example, the first round of PCR may be performed with one primer for each target locus, such that all primers bind to the same strand (eg, using a forward primer for each target locus). This allows PCR to amplify in a linear manner and reduces or eliminates amplification bias between amplicons due to differences in sequence or length. In some embodiments, the amplicons are then amplified using a forward and reverse primer for each target locus.

[602] Exemplary methods for designing primers

[603] If necessary, multiplex PCR can be performed using primers with a reduced likelihood of primer dimers. In particular, highly multiplexed PCR can often result in the formation of a very high proportion of DNA product, which results from unproductive side reactions such as primer dimer formation. In one embodiment, specific primers that are most likely to cause non-productive side reactions can be removed from the primer library in order to obtain a primer library that will result in an increase in the proportion of amplified DNA that maps to the genome. The step of removing problematic primers, that is, those primers that are likely to strengthen dimers, unexpectedly allowed for extremely high levels of PCR multiplexing for subsequent analysis by sequencing.

[604] There are several methods for selecting primers for a library that minimize the amount of non-mapping primer dimer or other detrimental primer products. Empirical data show that a small number of “bad” primers are responsible for a large number of non-mapping side reactions of primer dimers. Removing these “bad” primers can increase the percentage of sequence reads that map to target loci. One way to identify “bad” primers is to evaluate DNA sequencing data that has been amplified using targeted amplification; those primer dimers that are observed with the greatest frequency can be removed to obtain a primer library that is significantly less likely to produce a DNA by-product that does not map to the genome. Additionally, there are publicly available programs that can calculate the binding energies of different primer combinations and remove those with the highest binding energies, resulting in a primer library that is significantly less likely to produce a DNA byproduct that does not map to the genome.

[605] In some embodiments, an initial library of candidate primers is created for selecting primers by designing one or more primers or primer pairs for target candidate loci. A set of potential target loci (eg, SNPs) can be selected based on publicly available information about the desired parameters of the target loci, such as the frequency of the SNP in the target population or the level of heterozygosity of the SNP. In one embodiment, PCR primers can be designed using the Primer3 program (world wide web at primer3.sourceforge.net; libprimer3 release 2.2.3, incorporated herein by reference in its entirety). If desired, primers may be designed to anneal over a specific range of annealing temperatures, have a specific range of GC content, have a specific size range, produce target amplicons in a specific size range, and/or have other parameter characteristics. Using multiple primers or primer pairs for each candidate target locus increases the likelihood that the primer or primer pair will remain in the library for most or all target loci. In one embodiment, the selection criteria may require that at least one primer pair per target locus remain in the library. Thus, most or all of the target loci will be amplified using the latest primer library. This is desirable for applications such as screening for the presence of deletions or duplications at a large number of sites in the genome or screening for a large number of sequences (such as polymorphisms or other mutations) associated with disease or increased risk of disease. If a primer pair from a library produces a target amplicon that overlaps with a target amplicon produced by another primer pair, one of the primer pairs may be removed from the library to prevent interference.

[606] In some embodiments, a “undesirability score” (the highest score representing the least desirability) is calculated (eg, computer calculation) for most or all possible combinations of two primers from a library of candidate primers. In various embodiments, the undesirability score is calculated for at least 80, 90, 95, 98, 99, or 99.5% of the possible candidate primer combinations in the library. Each undesirability score is based at least in part on the probability of dimer formation between two candidate primers. If necessary, the undesirability assessment may also be based on one or more other parameters selected from the group consisting of degree of heterozygosity of the target locus, prevalence of disease associated with the sequence (e.g., polymorphism) at the target locus, penetrance of disease associated with the sequence (e.g., polymorphism) at the target locus, the specificity of the candidate primer for the target locus, the size of the candidate primer, the melting temperature of the target amplicon, the GC content of the target amplicon, the amplification efficiency of the target amplicon, the size of the target amplicon, and the distance from the center of the recombination hotspot. In some embodiments, the specificity of a candidate primer for a target locus includes the likelihood that the candidate primer will nonspecifically function by binding to and amplifying a locus other than the target locus it was designed to amplify. In some embodiments, one or more or all candidate primers that function nonspecifically are removed from the library. In some embodiments, to increase the number of possible primers to be selected, candidate primers capable of nonspecifically functioning are not removed from the library. If several factors are considered, an undesirability score can be calculated based on a weighted average of the various parameters. Parameters can be assigned different weights depending on their importance for the specific application for which the primers will be used. In some embodiments, the primer with the highest undesirability score is removed from the library. If the removed primer is a member of a primer pair that hybridizes to one target locus, then the other member of the primer pair may be removed from the library. If desired, the primer removal process can be repeated. In some embodiments, the selection method is performed until all undesirability scores for the candidate primer combinations remaining in the library are equal to or below a minimum threshold. In some embodiments, the selection method is performed until the number of candidate primers remaining in the library is reduced to the desired number.

[607] In various embodiments, after undesirability scores are calculated, the candidate primer that is part of the largest number of combinations of two candidate primers with an undesirability score above the first minimum threshold is removed from the library. At this stage, interactions equal to or below the first minimum threshold are ignored because these interactions are less significant. If the removed primer is a member of a primer pair that hybridizes to one target locus, then the other member of the primer pair may be removed from the library. If desired, the primer removal process can be repeated. In some embodiments, the selection method is performed until all undesirability scores for the candidate primer combinations remaining in the library are equal to or below the first minimum threshold. If the number of candidate primers remaining in the library exceeds the desired number, the number of primers can be reduced by reducing the first minimum threshold to a lower second minimum threshold and repeating the primer removal process. If the number of candidate primers remaining in the library is less than desired, the method can continue by increasing the first minimum threshold to a higher second minimum threshold and repeating the primer removal process using the original candidate primer library, thereby allowing more candidate primers to remain in library. In some embodiments, this selection method is performed until all undesirability scores for candidate primer combinations remaining in the library are equal to or below a second minimum threshold, or until the number of candidate primers remaining in the library is reduced to the desired number .

[608] If desired, primer pairs that produce a target amplicon that overlaps with a target amplicon produced by another primer pair can be separated into separate amplification reactions. Multiple PCR amplification reactions may be desirable for applications in which it is desirable to analyze all candidate target loci (rather than excluding candidate target loci from analysis due to overlap of target amplicons).

[609] These selection methods minimize the number of candidate primers that must be removed from the library to achieve the desired reduction in primer dimers. By removing fewer candidate primers from the library, more (or all) of the target loci can be amplified using the resulting primer library.

[610] Multiplexing a large number of primers places significant restrictions on the tests that can be included. Unintentionally, interacting tests result in false amplification products. Size limitations in miniPCR may lead to additional limitations. In one embodiment, one may start with a very large number of potential SNP targets (from about 500 to more than 1 million) and attempt to design primers to amplify each SNP. Where primers can be designed, an attempt can be made to identify primer pairs that are likely to form spurious products by estimating the probability of formation of a spurious primer duplex between all possible primer pairs using published thermodynamic parameters for DNA duplex formation. Primer interactions can be ranked using a scoring function associated with the interaction, and primers with the worst interaction scores are removed until the required number of primers is reached. In cases where the most useful SNPs are likely to be heterozygous, one can also rank the list of tests and select the tests most compatible with heterozygosity. Experiments have confirmed that primers with high interaction rates are most likely to form primer dimers. With high multiplexing, it is not possible to eliminate all spurious interactions, but it is important to remove primers or primer pairs with the highest in silico interaction scores, as they can dominate the entire reaction, significantly limiting the amplification from reaching its intended targets. We performed this procedure to create multiplex primer sets of up to 10,000 primers, and in some cases, more than 10,000 primers. The improvement associated with this procedure is significant, allowing amplification of greater than 80%, greater than 90%, greater than 95%, greater than 98%, and even greater than 99% of the target products, as determined by sequencing all PCR products, according to compared with 10% for a reaction in which the worst primers were not removed. When combined with a partial semi-nested approach as described previously, more than 90% and even more than 95% of amplicons can map to target sequences.

[611] Please note that there are other ways to determine which PCR probes can form dimers. In one embodiment, analysis of a pool of DNA that has been amplified using a non-optimized primer set may be sufficient to identify problematic primers. For example, the analysis can be performed using sequencing, and those dimers that are present in the greatest abundance are determined to be those most likely to form dimers and can be removed. In one embodiment, the primer design method can be used in combination with the mini-PCR method presented herein.

[612] The use of tags on primers can reduce the amplification and sequencing of dimeric primer products. In some embodiments, the primer contains an internal region that forms a loop structure with the label. In particular embodiments, the primers include a 5' region that is specific for the target locus, an internal region that is not specific for the target locus and forms a loop structure, and a 3' region that is specific for the target locus. In some embodiments, the loop region may be located between two binding regions, wherein the two binding regions are designed to bind adjacent or neighboring regions of template DNA. In various embodiments, the 3' region is at least 7 nucleotides in length. In some embodiments, the length of the 3' region is from 7 to 20 nucleotides, such as from 7 to 15 nucleotides, or from 7 to 10 nucleotides, inclusive. In various embodiments, the primers include a 5' region that is not specific to the target locus (e.g., a tag or universal primer binding site), followed by a region that is specific to the target locus, an internal region that is not specific to the target locus and forms a loop structure, and a 3' region specific for the target locus. Tagged primers can be used to reduce the required target specific sequences to less than 20, less than 15, less than 12, and even less than 10 base pairs. This may be an accident with a standard primer design where the target sequence is fragmented within the primer binding site, or it may be engineered into the primer design. The advantages of this method include the following: it increases the number of tests that can be designed for a given maximum amplicon length, and it reduces “uninformative” sequencing of the primer sequence. It can also be used in combination with internal tagging.

[613] In an embodiment, the relative amount of non-productive products in a multiplex targeted PCR amplification can be reduced by increasing the annealing temperature. In cases where libraries are amplified with the same tag as target-specific primers, the annealing temperature can be increased compared to genomic DNA because the tags will promote primer binding. In some embodiments, reduced primer concentrations are used, optionally coupled with longer annealing times. In some embodiments, the annealing time may be greater than 3 minutes, greater than 5 minutes, greater than 8 minutes, greater than 10 minutes, greater than 15 minutes, greater than 20 minutes, greater than 30 minutes, greater than 60 minutes, greater than 120 minutes, greater than 240 minutes, greater than 480 minutes and even more than 960 minutes. In certain illustrative embodiments, longer annealing times are used at lower primer concentrations. In various embodiments, longer than usual extension times are used, for example, greater than 3, 5, 8, 10 or 15 minutes. In some embodiments, primer concentrations are as low as 50 nM, 20 nM, 10 nM, 5 nM, 1 nM, and less than 1 nM. This surprisingly results in reliable performance for highly multiplexed reactions, such as 1000-plexed reactions, 2000-plexed reactions, 5000-plexed reactions, 10,000-plexed reactions, 20,000-plexed reactions, 50,000-plexed reactions, and even 100,000-plexed reactions. In an embodiment, the amplification uses one, two, three, four or five cycles of long annealing times, followed by PCR cycles of more conventional annealing times with labeled primers.

[614] To select target locations, one can start with a pool of candidate primer pair designs and create a thermodynamic model of the potentially unfavorable interactions between primer pairs, and then use the model to eliminate designs that are inconsistent with other designs in the pool.

[615] In an embodiment, the invention relates to a method of reducing the number of target loci (such as loci that may contain polymorphisms or mutations associated with a disease or disorder, or increased risk of a disease or disorder, such as cancer) and/or increasing the severity of the detected diseases (for example, an increase in the number of detected polymorphisms or mutations). In some embodiments, the method includes ranking (e.g., ranking from highest to lowest) loci by frequency or repetition of a polymorphism or mutation (e.g., a single nucleotide variation, insertion or deletion, or any other variations provided herein) at each locus among subjects with a disease or disorder such as cancer. In some embodiments, the PCR primers are designed for some or all of the loci. During the selection of PCR primers for a primer library, primers for loci that have higher frequency or repetition (higher-rank loci) are preferred over primers with lower frequency or repetition (lower-rank loci). In some embodiments, this parameter is included as one of the parameters in the calculation of undesirability scores presented herein. If necessary, primers (eg primers for high-ranking loci) that are incompatible with other designs in the library can be included in another library/PCR pool. In some embodiments, multiple libraries/pools (eg, 2, 3, 4, 5 or more) are used in individual PCR reactions to ensure amplification of all (or most) of the loci represented by all libraries/pools. In some embodiments, this method continues until a sufficient number of primers are included in one or more libraries/pools such that the primers collectively cover the desired disease load for the disease or disorder (e.g., in a manner such as by detecting at least 80, 85, 90, 95 or 99% disease load).

[616] Exemplary primer libraries

[617] In one aspect, the invention includes libraries of primers, such as primers selected from a library of candidate primers using any of the methods of the invention. In some embodiments, the library includes primers that simultaneously hybridize (or are capable of simultaneously hybridizing) with or that simultaneously amplify (or are capable of simultaneously amplifying) at least 100; 200; 500; 750; 1000; 2000; 5000; 7500; 10000; 20000; 25000; 30000; 40000; 50000; 75000; or 100,000 different target loci in one reaction volume. In various embodiments, the library includes primers that simultaneously amplify (or are capable of simultaneously amplifying) from 100 to 500; from 500 to 1000; from 1000 to 2000; from 2000 to 5000; from 5000 to 7500; from 7,500 to 10,000; from 10000 to 20000; from 20,000 to 25,000; from 25,000 to 30,000; from 30,000 to 40,000; from 40,000 to 50,000; from 50,000 to 75,000; or from 75,000 to 100,000 different target loci in one reaction volume, inclusive. In various embodiments, the library includes primers that simultaneously amplify (or are capable of simultaneously amplifying) 1,000 to 100,000 different target loci in a single reaction volume, such as 1,000 to 50,000; from 1000 to 30000; from 1000 to 20000; from 1000 to 10000; from 2000 to 30000; from 2000 to 20000; from 2000 to 10000; from 5000 to 30000; from 5000 to 20000; or from 5,000 to 10,000 different target loci, inclusive. In some embodiments, the library includes primers that simultaneously amplify (or are capable of simultaneously amplifying) target loci in a single reaction volume such that less than 60, 40, 30, 20, 10, 5, 4, 3, 2, 1, 0.5 , 0.25, 0.1, or 0.5% of the amplified products are primer dimers. In various embodiments, the amount of amplified products that are primer dimers is from 0.5 to 60%, for example from 0.1 to 40%, from 0.1 to 20%, from 0.25 to 20%, from 0. 25 to 10%, from 0.5 to 20%, from 0.5 to 10%, from 1 to 20% or from 1 to 10% inclusive. In some embodiments, the primers simultaneously amplify (or are capable of simultaneously amplifying) target loci in a single reaction volume such that at least 50, 60, 70, 80, 90, 95, 96, 97, 98, 99, or 99.5% of the amplified products are target amplicons. In various embodiments, the amount of amplified products that are target amplicons is from 50 to 99.5%, for example, from 60 to 99%, from 70 to 98%, from 80 to 98%, from 90 to 99.5%, or from 95 to 99.5% inclusive. In some embodiments, the primers simultaneously amplify (or are capable of simultaneously amplifying) target loci in a single reaction volume such that at least 50, 60, 70, 80, 90, 95, 96, 97, 98, 99, or 99.5% of the target loci are amplified (eg, amplified by at least 5, 10, 20, 30, 50, or 100 times the amount before amplification). In various embodiments, the number of target loci that are amplified (e.g., amplified by at least 5, 10, 20, 30, 50, or 100 times the amount before amplification) is from 50 to 99.5%, such as from 60 up to 99%, from 70 to 98%, from 80 to 99%, from 90 to 99.5%, from 95 to 99.9% or from 98 to 99.99% inclusive. In some embodiments, the primer library includes at least 100; 200; 500; 750; 1000; 2000; 5000; 7500; 10000; 20,000; 25000; 30000; 40000; 50000; 75000; or 100,000 primer pairs, where each primer pair includes a forward test primer and a reverse test primer, where each test primer pair hybridizes to the target locus. In some embodiments, the primer library includes at least 100; 200; 500; 750; 1000; 2000; 5000; 7500; 10000; 20000; 25000; 30000; 40000; 50000; 75000; or 100,000 individual primers, each hybridizing to a different target locus, with the individual primers not being part of primer pairs.

[618] In various embodiments, the concentration of each primer is less than 100, 75, 50, 25, 20, 10, 5, 2, or 1 nM, or less than 500, 100, 10, or 1 μM. In various embodiments, the concentration of each primer is from 1 μM to 100 nM, for example, from 1 μM to 1 nM, from 1 to 75 nM, from 2 to 50 nM, or from 5 to 50 nM, inclusive. In various embodiments, the GC content of the primers is from 30 to 80%, such as from 40 to 70%, or from 50 to 60%, inclusive. In some embodiments, the range of GC content of the primers is less than 30, 20, 10, or 5%. In some embodiments, the range of GC content of the primers is from 5 to 30%, such as from 5 to 20%, or from 5 to 10%, inclusive. In some embodiments, the melting temperature (T _m ) of the test primers is 40 to 80°C, such as 50 to 70°C, 55 to 65°C, or 57 to 60.5°C, inclusive. In some embodiments, T _m is calculated using the Primer3 program (libprimer3, release 2.2.3) using the built-in parameters of SantaLucia (web address: primer3.sourceforge.net). In some embodiments, the melting temperature range of the primers is less than 15, 10, 5, 3, or 1 °C. In some embodiments, the melting temperature range of the primers is from 1 to 15°C, such as from 1 to 10°C, from 1 to 5°C, or from 1 to 3°C, inclusive. In some embodiments, the length of the primers is from 15 to 100 nucleotides, such as from 15 to 75 nucleotides, from 15 to 40 nucleotides, from 17 to 35 nucleotides, from 18 to 30 nucleotides, or from 20 to 65 nucleotides, inclusive. In some embodiments, the primer length range is less than 50, 40, 30, 20, 10, or 5 nucleotides. In some embodiments, the length range of the primers is from 5 to 50 nucleotides, such as from 5 to 40 nucleotides, from 5 to 20 nucleotides, or from 5 to 10 nucleotides, inclusive. In some embodiments, the length of the target amplicons is from 50 to 100 nucleotides, for example, from 60 to 80 nucleotides, or from 60 to 75 nucleotides, inclusive. In some embodiments, the length range of the target amplicons is less than 50, 25, 15, 10, or 5 nucleotides. In some embodiments, the length range of the target amplicons is from 5 to 50 nucleotides, such as from 5 to 25 nucleotides, from 5 to 15 nucleotides, or from 5 to 10 nucleotides, inclusive. In some embodiments, the library does not contain a microchip. In some embodiments, the library includes a microchip.

[619] In some embodiments, some (eg, at least 80, 90, or 95%) or all of the adapters or primers include one or more bonds between adjacent nucleotides other than a naturally occurring phosphodiester bond. Examples of such linkages include phosphoramide, phosphorothioate and phosphorodithioate linkages. In some embodiments, some (eg, at least 80, 90, or 95%) or all of the adapters or primers include a thiophosphate (eg, monothiophosphate) between the last 3' nucleotide and the penultimate 3' nucleotide. In some embodiments, some (eg, at least 80, 90, or 95%) or all of the adapters or primers include a thiophosphate (eg, monothiophosphate) between the last 2, 3, 4, or 5 nucleotides at the 3' end. In some embodiments, some (e.g., at least 80, 90, or 95%) or all of the adapters or primers include a thiophosphate (e.g., monothiophosphate) between at least 1, 2, 3, 4, or 5 nucleotides of the last 10 nucleotides at 3' -end. In some embodiments, such primers are less likely to be cleaved or destroyed. In some embodiments, the primers do not contain an enzyme cleavage site (eg, a protease cleavage site).

[620] Additional exemplary multiplex PCR methods and libraries are described in US Patent Application No. 13/683604, filed November 21, 2012 (US Patent Application Publication No. 2013/0123120) and US Provisional Patent Application Serial No. 61/994791, filed May 16, 2014, each of which is incorporated herein by reference in its entirety). These methods and libraries can be used to analyze any of the samples disclosed herein and for use in any of the methods of the present invention.

[621] Exemplary primer libraries for recombination detection

[622] In some embodiments, the primers in the primer library are designed to determine whether recombination has occurred at one or more known recombination hot spots (such as crossovers between homologous human chromosomes). Knowing which crossovers have occurred between chromosomes allows us to determine more precise step-by-step genetic data for an individual. Recombination hotspots are local regions of chromosomes in which recombination events tend to be concentrated. They are often surrounded by “cold spots,” areas with below-average recombination rates. Recombination hotspots tend to have similar morphologies and are approximately 1 to 2 kb in length. Hotspot distribution is positively correlated with GC content and repeat element distribution. The partially degenerate 13-mer motif CCNCCNTNNCCNC plays a role in the activity of some hotspots. A zinc finger protein called PRDM9 has been shown to bind to this motif and initiate recombination at its location. The average distance between recombination hotspot centers is reported to be ~80 kb. In some embodiments, the distance between the centers of recombination hot spots is from about 3 kb. up to approximately 100 kb. Publicly available databases include a large number of known human recombination hot spots, such as the HUMHOT and International HapMap Project databases (see, for example, Nishant et al., “HUMHOT: a database of human meiotic recombination hot spots,” Nucleic Acids Research, 34: D25-D28, 2006, Database issue; Mackiewicz et al., “Distribution of Recombination Hotspots in the Human Genome - A Comparison of Computer Simulations with Real Data” PLoS ONE 8(6): e65272, doi:10.1371/journal. pone.0065272; and the web address hapmap.ncbi.nlm.nih.gov/downloads/index.html.en, each of which is incorporated herein by reference in its entirety).

[623] In some embodiments, the primers in the primer library are clustered at or near recombination hotspots (eg, known human recombination hotspots). In some embodiments, the corresponding amplicons are used to determine the sequence within or adjacent to a recombination hot spot to determine whether recombination has occurred at that particular hot spot (e.g., whether the amplicon sequence is the sequence expected if recombination has occurred, or sequence expected if recombination did not occur). In some embodiments, the primers are designed to amplify part or all of a recombination hot spot (and, optionally, the sequence flanking the recombination hot spot). In some embodiments, long-read sequencing (such as sequencing using Moleculo technology developed by Illumina for sequencing up to approximately 10 kb) or paired-end paired-end sequencing is used to sequence part or all of a recombination hot spot. Knowing whether a recombination event has occurred can be used to determine which blocks of haplotypes flank a hotspot. If necessary, the presence of certain haplotype blocks can be confirmed using primers specific for regions within the haplotype blocks. In some embodiments, it is assumed that there are no crossovers between known recombination hot spots. In some embodiments, the primers in the primer library are grouped at or near the ends of chromosomes. For example, such primers can be used to determine the presence or absence of a particular arm or region at the end of a chromosome. In some embodiments, the primers in the primer library are clustered at or near recombination hotspots and at or near the ends of chromosomes.

[624] In some embodiments, the primer library includes one or more primers (e.g., at least 5; 10; 50; 100; 200; 500; 750; 1000; 2000; 5000; 7,500; 10,000; 20,000; 25,000; 30,000; 40,000; or 50,000 different primers or different primer pairs) that are specific to a recombination hotspot (e.g., a known human recombination hotspot) and/or specific to a region adjacent to a recombination hotspot (e.g., within 10, 8, 5, 3 , 2, 1, or 0.5 kb from the 5' or 3' end of the recombination hotspot). In some embodiments, at least 1, 5, 10, 20, 40, 60, 80, 100, or 150 different primers (or primer pairs) are specific for the same recombination hot spot or are specific for the same recombination hot spot or areas near a recombination hotspot. In some embodiments, at least 1, 5, 10, 20, 40, 60, 80, 100, or 150 different primers (or primer pairs) are specific for a region between recombination hot spots (eg, a region that is unlikely to have undergone recombination); these primers can be used to confirm the presence of blocks of haplotypes (such as those that would be expected depending on whether recombination has occurred). In some embodiments, at least 10, 20, 30, 40, 50, 60, 70, 80, or 90% of the primers in the primer library are specific for a recombination hot spot and/or specific for a region adjacent to the recombination hot spot (e.g. within 10, 8, 5, 3, 2, 1 or 0.5 kb from the 5' or 3' end of the recombination hotspot). In some embodiments, a library of primers is used to determine whether recombination has occurred at a number of different recombination hot spots greater than or equal to 5; 10; 50; 100; 200; 500; 750; 1000; 2000; 5000; 7500; 10000; 20000; 25000; 30000; 40000; or 50,000 (such as known human recombination hotspots). In some embodiments, the regions targeted by primers targeting a recombination hotspot or nearby region are approximately evenly distributed along that portion of the genome. In some embodiments, at least 1, 5, 10, 20, 40, 60, 80, 100, or 150 different primers (or primer pairs) are specific for a region at or approximately the end of a chromosome (e.g., regions within 20, 10, 5, 1, 0.5, 0.1, 0.01 or 0.001 Mb from the end of the chromosome). In some embodiments, at least 10, 20, 30, 40, 50, 60, 70, 80, or 90% of the primers in the primer library are specific for a region at or approximately the end of the chromosome (e.g., regions within 20, 10, 5, 1, 0.5, 0.1, 0.01 or 0.001 Mb from the end of the chromosome). In some embodiments, at least 1, 5, 10, 20, 40, 60, 80, 100, or 150 different primers (or primer pairs) are specific for a region within a potential microdeletion in a chromosome. In some embodiments, at least 10, 20, 30, 40, 50, 60, 70, 80, or 90% of the primers in the primer library are specific for a region within a potential microdeletion in the chromosome. In some embodiments, at least 10, 20, 30, 40, 50, 60, 70, 80, or 90% of the primers in the primer library are specific for a recombination hotspot, a region adjacent to a recombination hotspot, a region at or approximately the end of a chromosome, or regions within a potential microdeletion in a chromosome.

[625] Exemplary Multiplex PCR Methods

[626] In one aspect, the invention relates to methods for amplifying target loci in a nucleic acid sample, which include (i) contacting the nucleic acid sample with a library of primers that simultaneously hybridize with at least 1000; 2000; 5000; 7500; 10000; 15000; 19000; 20000; 25000; 27000; 28000; 30000; 40000; 50000; 75000; or 100,000 different target loci to obtain a reaction mixture; and (ii) exposing the reaction mixture to primer extension reaction conditions (such as PCR conditions) to produce amplified products that include the target amplicons. In some embodiments, the method also includes determining the presence or absence of at least one target amplicon (eg, at least 50, 60, 70, 80, 90, 95, 96, 97, 98, 99, or 99.5% of the target amplicons). In some embodiments, the method also includes determining the sequence of at least one target amplicon (eg, at least 50, 60, 70, 80, 90, 95, 96, 97, 98, 99, or 99.5% of the target amplicons). In some embodiments, at least 50, 60, 70, 80, 90, 95, 96, 97, 98, 99, or 99.5% of the target loci are amplified. In some embodiments, at least 25; 50; 75; 100; 300; 500; 750; 1000; 2000; 5000; 7500; 10,000; 15000; 19000; 20000; 25000; 27000; 28000; 30000; 40000; 50000; 75000; or 100,000 different target loci are amplified at least 5-fold, 10-fold, 20-fold, 40-fold, 50-fold, 60-fold, 80-fold, 100-fold, 120-fold, 150-fold, 200-fold, 300-fold, or 400-fold. In some embodiments, at least 50, 60, 70, 80, 90, 95, 96, 97, 98, 99, 99.5, or 100% of the target loci are amplified in at least 5, 10, 20, 40, 50, 60, 80, 100, 120, 150, 200, 300 or 400 times. In various embodiments, less than 60, 50, 40, 30, 20, 10, 5, 4, 3, 2, 1, 0.5, 0.25, 0.1, or 0.05% of the amplified products are primer dimers. In some embodiments, the method includes multiplex PCR and sequencing (eg, high-throughput sequencing).

[627] Various embodiments utilize long annealing times and/or low primer concentrations. In various embodiments, the duration of the annealing step is greater than 3, 5, 8, 10, 15, 20, 30, 45, 60, 75, 90, 120, 150, or 180 minutes. In various embodiments, the duration of the annealing step (per PCR cycle) is from 5 to 180 minutes, such as from 5 to 60, from 10 to 60, from 5 to 30, or from 10 to 30 minutes, inclusive. In various embodiments, the duration of the annealing step is greater than 5 minutes (eg, greater than 10 or 15 minutes) and the concentration of each primer is less than 20 nM. In various embodiments, the duration of the annealing step is greater than 5 minutes (eg, greater than 10 or 15 minutes), and the concentration of each primer is from 1 to 20 nM, or from 1 to 10 nM, inclusive. In various embodiments, the duration of the annealing step is greater than 20 minutes (eg, greater than 30, 45, 60, or 90 minutes) and the concentration of each primer is less than 1 nM.

[628] At high levels of multiplexing, the solution may become viscous due to the large number of primers in the solution. If the solution is too viscous, the primer concentration can be reduced to an amount sufficient for the primer to bind the template DNA. In various embodiments, fewer than 60,000 different primers are used, and the concentration of each primer is less than 20 nM, such as less than 10 nM, or from 1 to 10 nM, inclusive. In various embodiments, more than 60,000 different primers are used (eg, from 60,000 to 120,000 different primers), and the concentration of each primer is less than 10 nM, such as less than 5 nM, or from 1 to 10 nM, inclusive.

[629] It has been discovered that the annealing temperature may not necessarily be higher than the melting temperatures of some or all of the primers (unlike other methods that use annealing temperatures below the melting temperatures of the primers). Melting temperature ( _Tm ) is the temperature at which half (50%) of the DNA duplex of an oligonucleotide (eg, a primer) and its perfect complement dissociate and become single-stranded DNA. Annealing temperature (T _A ) is the temperature at which the PCR protocol is performed. For previous methods, it is usually 5 °C below the lowest T _m of the primers used, so that almost all possible duplexes are formed (such that essentially all primer molecules bind to the template nucleic acid). Although this is very effective, at lower temperatures more nonspecific reactions are bound to occur. One consequence of too low a T _A is that primers may anneal to sequences other than the true target, since internal single-base mismatches or partial annealing may occur. In some embodiments of the present invention, T _A is higher than (T _m) , where at a given time only a small portion of the targets have an annealed primer (eg, only about 1-5%). If they are extended, then they are removed from the equilibrium of annealing and primer-target dissociation (as extension rapidly increases T _m to over 70°C), and approximately 1-5% of new targets have primers. Thus, by giving the reaction a long annealing time, approximately 100% copies of the targets can be obtained per cycle. Thus, the most stable pairs of molecules (with perfect DNA pairing between primer and template DNA) are preferentially extended to produce the correct target amplicons. For example, the same experiment was performed with 57 °C as the annealing temperature and 63 °C as the annealing temperature with primers that had a melting point below 63 °C. When the annealing temperature was 57°C, the percentage of mapped reads for the amplified PCR products was only 50% (with approximately 50% of the amplified products being primer dimers). When the annealing temperature was 63 °C, the percentage of amplified products that were primer dimers dropped to approximately 2%.

[630] In various embodiments, the annealing temperature is at least 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, or 15 °C higher than the melting temperature (e.g. empirically measured or calculated T_m) at least 25; 50; 75; 100; 300; 500; 750; 1000; 2000; 5000; 7500; 10000; 15000; 19000; 20000; 25000; 27000; 28000; 30000; 40000; 50000; 75000; 100000; or all non-identical primers. In some embodiments, the annealing temperature is at least 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, or 15 °C higher than the melting temperature (e.g., empirically measured or calculated T_m) at least 25; 50; 75; 100; 300; 500; 750; 1000; 2000; 5000; 7500; 10000; 15000; 19000; 20000; 25000; 27000; 28000; 30000; 40000; 50000; 75000; 100000; or all non-identical primers, and the duration of the annealing step (per PCR cycle) is more than 1, 3, 5, 8, 10, 15, 20, 30, 45, 60, 75, 90, 120, 150 or 180 minutes.

[631] In various embodiments, the annealing temperature is 1 to 15 °C (e.g., 1 to 10, 1 to 5, 1 to 3, 3 to 5, 5 to 10, 5 to 8, 8 to 10, 10 to 12, or 12 to 15 °C inclusive) above the melting point (such as empirically measured or calculated _Tm ) of at least 25; 50; 75; 100; 300; 500; 750; 1000; 2000; 5000; 7500; 10000; 15000; 19000; 20000; 25000; 27000; 28000; 30000; 40000; 50000; 75000; 100000; or all non-identical primers. In various embodiments, the annealing temperature is from 1 to 15 °C (for example, from 1 to 10, from 1 to 5, from 1 to 3, from 3 to 5, from 5 to 10, from 5 to 8, from 8 to 10, 10 to 12 or 12 to 15 °C inclusive) above the melting point (such as empirically measured or calculated _Tm ) of at least 25; 50; 75; 100; 300; 500; 750; 1000; 2000; 5000; 7500; 10000; 15000; 19000; 20000; 25000; 27000; 28000; 30000; 40000; 50000; 75000; 100000; or all non-identical primers, and the duration of the annealing step (per PCR cycle) is from 5 to 180 minutes, for example from 5 to 60, from 10 to 60, from 5 to 30 or from 10 to 30 minutes inclusive.

[632] In some embodiments, the annealing temperature is at least 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, or 15 °C higher than the highest melting temperature ( for example, empirically measured or calculated T _m ) of primers. In some embodiments, the annealing temperature is at least 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, or 15 °C higher than the highest melting temperature (e.g., empirically measured or calculated T _m ) of primers, and the length of the annealing step (per PCR cycle) is greater than 1, 3, 5, 8, 10, 15, 20, 30, 45, 60, 75, 90, 120, 150 or 180 minutes.

[633] In some embodiments, the annealing temperature is 1 to 15 °C (e.g., 1 to 10, 1 to 5, 1 to 3, 3 to 5, 5 to 10, 5 to 8, 8 to 10, 10 to 12, or 12 to 15 °C inclusive) higher than the highest melting temperature (e.g., empirically measured or calculated T _m ) of the primers. In some embodiments, the annealing temperature is 1 to 15 °C (e.g., 1 to 10, 1 to 5, 1 to 3, 3 to 5, 5 to 10, 5 to 8, 8 to 10 , from 10 to 12 or from 12 to 15 °C inclusive) higher than the highest melting temperature (e.g. empirically measured or calculated T _m ) of the primers, and the duration of the annealing step (per PCR cycle) is from 5 to 180 minutes, e.g. 5 to 60, from 10 to 60, from 5 to 30 or from 10 to 30 minutes inclusive.

[634] In some embodiments, the annealing temperature is at least 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, or 15 °C higher than the average melting temperature (e.g. , empirically measured or calculated T _m ) at least 25; 50; 75; 100; 300; 500; 750; 1000; 2000; 5000; 7500; 10000; 15000; 19000; 20000; 25000; 27000; 28000; 30000; 40000; 50000; 75000; 100000; or all non-identical primers. In some embodiments, the annealing temperature is at least 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, or 15 °C higher than the average melting temperature (e.g., empirically measured or calculated Tm) of at least 25; 50; 75; 100; 300; 500; 750; 1000; 2000; 5000; 7500; 10000; 15000; 19000; 20000; 25000; 27000; 28000; 30000; 40000; 50000; 75000; 100000; or all non-identical primers, and the length of the annealing step (per PCR cycle) is greater than 1, 3, 5, 8, 10, 15, 20, 30, 45, 60, 75, 90, 120, 150 or 180 minutes.

[635] In some embodiments, the annealing temperature is 1 to 15 °C (e.g., 1 to 10, 1 to 5, 1 to 3, 3 to 5, 5 to 10, 5 to 8, 8 to 10, 10 to 12, or 12 to 15 °C inclusive) above the average melting temperature (such as empirically measured or calculated _Tm ) of at least 25; 50; 75; 100; 300; 500; 750; 1000; 2000; 5000; 7500; 10000; 15000; 19000; 20000; 25000; 27000; 28000; 30000; 40000; 50000; 75000; 100000; or all non-identical primers. In some embodiments, the annealing temperature is 1 to 15 °C (e.g., 1 to 10, 1 to 5, 1 to 3, 3 to 5, 5 to 10, 5 to 8, 8 to 10 , from 10 to 12 or from 12 to 15 °C inclusive) above the average melting temperature (such as empirically measured or calculated _Tm ) of at least 25; 50; 75; 100; 300; 500; 750; 1000; 2000; 5000; 7500; 10000; 15000; 19000; 20000; 25000; 27000; 28000; 30000; 40000; 50000; 75000; 100000; or all non-identical primers, and the duration of the annealing step (per PCR cycle) is from 5 to 180 minutes, for example from 5 to 60, from 10 to 60, from 5 to 30 or from 10 to 30 minutes inclusive.

[636] In some embodiments, the annealing temperature is 50 to 70 °C, such as 55 to 60, 60 to 65, or 65 to 70 °C, inclusive. In some embodiments, the annealing temperature is 50 to 70 °C, such as 55 to 60, 60 to 65, or 65 to 70 °C, inclusive, and either (i) the length of the annealing step (per PCR cycle) is greater than 3 , 5, 8, 10, 15, 20, 30, 45, 60, 75, 90, 120, 150 or 180 minutes or (ii) the duration of the annealing step (per PCR cycle) is from 5 to 180 minutes, for example from 5 to 60, from 10 to 60, from 5 to 30 or from 10 to 30 minutes inclusive.

[637] In some embodiments, one or more of the following conditions are used to empirically measure _Tm or are used to calculate _Tm : temperature: 60.0 °C, primer concentration 100 nM, and/or salt concentration 100 mM. In some embodiments, other conditions are used, such as those that would be used for multiplex PCR with a library. In some embodiments, 100 mM KCl, 50 mM (NH ₄ ) ₂ SO ₄ , 3 mM MgCl ₂ , 7.5 nM each primer, and 50 mM TMAC at pH 8.1 are used. In some embodiments, T _m is calculated using the Primer3 program (libprimer3 release 2.2.3) using built-in SantaLucia parameters (web address primer3.sourceforge.net, document incorporated herein by reference in its entirety). In some embodiments, the calculated melting temperature of the primer is the temperature at which half of the primer molecules are expected to anneal. As discussed above, even at temperatures above the calculated melting temperature, some of the primers will anneal and hence the PCR may be extended. In some embodiments, the empirically measured T _m (actual T _m ) is determined using a thermostated cell in a UV spectrophotometer. In some embodiments, a plot of temperature versus absorbance produces an S-shaped curve with two plateaus. The absorption value halfway between the plateaus corresponds to _Tm .

[638] In some embodiments, absorbance at 260 nm is measured as a function of temperature on an ultrospec 2100 pr UV/visible spectrophotometer. (Amershambiosciences) (see, for example, Takiya et al ., “An empirical approach for thermal stability (Tm) prediction of PNA/DNA duplexes,” Nucleic Acids Symp Ser (Oxf); (48):131-2, 2004, document is incorporated herein by reference in its entirety . In some embodiments, absorbance at 260 nm is measured by decreasing the temperature in 2 °C increments per minute from 95 to 20 °C. In some embodiments, the primer and its ideal complement (e.g., 2 μM of each paired oligomer) are mixed and then annealed by heating the sample to 95 °C, holding it there for 5 minutes, then cooling to room temperature over 30 minutes, and holding samples at 95°C for at least 60 minutes. In some embodiments, the melting point is determined by analyzing data using the SWIFT Tm program. In some embodiments of any of the methods of the invention, the method includes empirically measuring or calculating (e.g., computer calculation) the melting temperature of at least 50, 80, 90, 92, 94, 96, 98, 99, or 100% of the primers in library before or after using primers for PCR amplification of target loci.

[639] In some embodiments, the library includes a microchip. In some embodiments, the library does not contain a microchip.

[640] In some embodiments, most or all of the primers are extended to produce amplification products. Using all primers in a PCR reaction increases the uniformity of amplification across different target loci because the same or similar number of primer molecules are converted into target amplicons for each target locus. In some embodiments, at least 80, 90, 92, 94, 96, 98, 99, or 100% of the primer molecules are extended to form amplified products. In some embodiments, for at least 80, 90, 92, 94, 96, 98, 99, or 100% of the target loci, at least 80, 90, 92, 94, 96, 98, 99, or 100% of the primer molecules for those targets loci are extended to form amplified products. In some embodiments, multiple cycles are performed until this percentage of primers is consumed. In some embodiments, multiple cycles are performed until all or substantially all of the primers are consumed. If necessary, a higher percentage of primers can be used by decreasing the initial primer concentration and/or increasing the number of PCR cycles performed.

[641] In some embodiments, PCR methods can be performed with microliter reaction volumes, which may be more difficult to achieve specific PCR amplification (due to the lower local concentration of template nucleic acids) compared to nanoliter or picoliter reaction volumes used in microfluidic applications . In some embodiments, the reaction volume is 1 to 60 μL, such as 5 to 50 μL, 10 to 50 μL, 10 to 20 μL, 20 to 30 μL, 30 to 40 μL, or 40 to 50 μL, inclusive .

[642] In one embodiment, the method disclosed herein uses high-throughput, highly multiplexed targeted PCR to amplify DNA, followed by high-throughput sequencing to determine allele frequencies at each target locus. The ability to multiplex more than 50 or 100 PCR primers in a single reaction volume such that the majority of the resulting sequence reads map to target loci is new and not obvious. One technique that allows highly multiplexed, highly targeted PCR involves designing primers that are unlikely to hybridize to each other. PCR probes, commonly called primers, are selected by creating a thermodynamic model of potentially unfavorable interactions between at least 300; at least 500; at least 750; at least 1000; at least 2000; at least 5000; at least 7500; at least 10,000; at least 20,000; at least 25,000; at least 30,000; at least 40,000; at least 50,000; at least 75,000; or at least 100,000 potential primer pairs, or unintended interactions between the primers and the DNA sample, and then use the model to eliminate constructs that are inconsistent with other constructs in the pool. Another technique that allows for highly multiplexed, high-throughput targeted PCR is to use a partial or full nesting approach to targeted PCR. Using one or a combination of these approaches allows at least 300, at least 800, at least 1200, at least 4000, or at least 10,000 primers to be multiplexed into a single pool, with the resulting amplified DNA containing the majority of the DNA molecules that, when sequenced, will map to target loci. Using one or a combination of these approaches allows you to multiplex a large number of primers in one pool with the resulting amplified DNA containing more than 50%, more than 60%, more than 67%, more than 80%, more than 90%, more than 95%, more than 96%, more than 97 %, greater than 98%, greater than 99%, or greater than 99.5% of DNA molecules that map to target loci.

[643] In some embodiments, detection of target genetic material can be performed in a multiplex manner. The number of target genetic sequences that can be executed in parallel can vary from one to ten, ten to one hundred, one hundred to one thousand, one thousand to ten thousand, ten thousand to one hundred thousand, one hundred thousand to a million, or from one million to ten million. Previous attempts to multiplex more than 100 primers per pool resulted in significant problems with unwanted side reactions such as primer dimer formation.

[644] Targeted PCR

[645] In some embodiments, PCR can be used to target specific regions of the genome. In plasma samples, the original DNA is highly fragmented (usually less than 500 bp, with an average length of less than 200 bp). In PCR, both forward and reverse primers anneal to the same fragment to make amplification possible. Therefore, if the fragments are short, PCR assays should also amplify relatively short regions. Similar to MIPS, if polymorphic positions are too close to the polymerase binding site, this can lead to errors in amplification from different alleles. Currently, PCR primers targeting polymorphic regions, such as those containing SNPs, are typically designed such that the 3' end of the primer will hybridize to the base immediately adjacent to the polymorphic base or bases. In one embodiment of the present invention, the 3' ends of both the forward and reverse PCR primers are designed to hybridize to bases that are one or more positions further from variant positions (polymorphic sites) of the target allele. The number of bases between the polymorphic site (SNP or other) and the base to which the 3' end of the primer is intended to hybridize could be one base, it could be two bases, it could be three bases, it could be four bases, it could be five bases, it could be six bases, it could be seven to ten bases, it could be eleven to fifteen bases, or it could be sixteen to twenty bases. Forward and reverse primers can be designed to hybridize different numbers of bases away from the polymorphic site.

[646] PCR assays can be performed in large numbers, but interactions between different PCR assays make it difficult to multiplex beyond about a hundred assays. Various complex molecular approaches can be used to increase multiplexing levels, but may still be limited to less than 100, perhaps 200, or perhaps 500 assays per reaction. Samples with large amounts of DNA can be split into multiple subreactions and then recombined before sequencing. For samples in which either the total sample or some subpopulation of DNA molecules is limited, separating the sample will introduce statistical noise. In one embodiment, a small or limited amount of DNA may refer to an amount below 10 pg, from 10 to 100 pg, from 100 pg to 1 ng, from 1 to 10 ng, or from 10 to 100 ng. Note that while this method is particularly useful for small quantities of DNA, where other methods involving partitioning into multiple pools may cause significant problems due to introduced stochastic noise, this method still provides the benefit of minimizing bias when performed on samples with any amount of DNA. In these situations, a versatile preamplification step can be used to increase the total sample quantity. Ideally, this preamplification step should not appreciably alter allelic distributions.

[647] In one embodiment, the method according to the present description can generate PCR products that are specific for a large number of target loci, in particular for 1000 to 5000 loci, 5000 to 10000 loci, or more than 10000 loci, for genotyping by sequencing or some other method of genotyping from a limited number of samples, such as single cells or DNA from body fluids. Currently, performing multiplex PCR reactions for more than 5-10 targets is a major challenge and is often hampered by primer by-products such as primer dimers and other artifacts. When detecting target sequences using microarrays with hybridization probes, primer dimers and other artifacts can be ignored as they are not detected. However, when using sequencing as a detection method, the vast majority of sequencing reads will sequence such artifacts rather than the desired target sequences in the sample. Methods presented in the prior art used to multiplex more than 50 or 100 reactions in a single reaction volume followed by sequencing typically result in greater than 20%, and often greater than 50%, in many cases greater than 80%, and in some cases, more than 90% of the reads are off-target sequence.

[648] In general, to perform targeted sequencing of multiple (n) targets on a sample (greater than 50, greater than 100, greater than 500, or greater than 1000), the sample can be divided into a number of parallel reactions that amplify one individual target. This has been performed in multiwell PCR plates or can be performed on commercial platforms such as FLUIDIGM ACCESS ARRAY (48 reactions per sample in microfluidic chips) or DROPLET PCR from RAIN DANCE TECHNOLOGY (100 to several thousand targets). Unfortunately, these splitting and pooling methods are problematic for samples with limited amounts of DNA because there are often not enough copies of the genome to ensure that there is one copy of each genomic region in each well. This is a particularly serious problem when polymorphic loci are targeted and the relative proportions of alleles at the polymorphic loci are required, since the stochastic noise introduced by splitting and merging will lead to very inaccurate measurements of the proportions of alleles that are present in the original DNA sample. This document describes a method for efficient and effective amplification of many PCR reactions, which is applicable in cases where only a limited amount of DNA is available. In one embodiment, the method can be used to analyze single cells, biological fluids, mixtures of DNA, such as free-floating DNA found in plasma, biopsies, environmental samples, and/or forensic samples.

[649] In one embodiment, targeted sequencing may include one, many, or all of the following steps: a) Generate and amplify a library with adapter sequences at both ends of the DNA fragments. b) Divide into several reactions after library amplification. c) Generate and optionally amplify a library with adapter sequences at both ends of the DNA fragments. d) Perform 1000- to 10,000-plex amplification of selected targets using one target-specific “forward” primer for each target and one tag-specific primer. e) Perform a second amplification of this product using target-specific “reverse” primers and one (or more) primers specific for the universal tag that was introduced as part of the target-specific forward primers in the first round. f) Perform 1000-plex pre-amplification of the selected target for a limited number of cycles. g) Divide the product into several aliquots and amplify target subpools in individual reactions (e.g., 50 to 500-plex amplification, although this can be used up to single-plex amplification. h) Combine the products of parallel subpool reactions. i) During these amplifications, primers can carry sequencing-compatible tags (partial or full length) so that these products can be sequenced.

[650] Highly multiplex PCR

[651] Disclosed herein are methods that enable targeted amplification of more than hundreds to tens of thousands of target sequences (eg, SNP loci) from a nucleic acid sample, such as plasma-derived genomic DNA. The amplified sample may be relatively free of primer dimer products and have low allelic shift at target loci. If sequencing-compatible adapters are added to the products during or after amplification, analysis of these products can be performed by sequencing.

[652] Performing highly multiplexed PCR amplifications using methods known in the art results in primer dimer products that are larger than the desired amplification products and are not suitable for sequencing. They can be reduced empirically by eliminating primers that form these products or by performing in silico primer selection. However, the greater the number of analyses, the more complex this problem becomes.

[653] One solution is to split the 5000-plex reaction into several lower-plex amplifications, such as one hundred 50-plex or fifty 100-plex reactions, or use microfluidics, or even split the sample into separate PCR reactions. However, if the DNA sample is limited, such as in non-invasive prenatal diagnostics from pregnancy plasma, splitting the sample into multiple reactions should be avoided as this will create bottlenecks.

[654] Methods are described herein to first amplify the plasma DNA of a sample as a whole and then split the sample into multiple multiplexed target enrichment reactions with a more moderate number of target sequences per reaction. In one embodiment, the method of the present invention can be used to preferentially enrich a mixture of DNA at multiple loci, the method comprising one or more of the following steps: creating and amplifying a library from a mixture of DNA, wherein the molecules in the library have adapter sequences ligated at both ends of the DNA fragments, dividing the amplified library into several reactions, performing a first round of multiplex amplification of selected targets using one target-specific “forward” primer per target and one or more adapter-specific universal “reverse” primers. In one embodiment, the method of the present invention further includes performing a second amplification using target-specific reverse primers and one or more primers specific for a universal tag that was introduced as part of the target-specific forward primers in the first round. In one embodiment, the method may include a fully nested, hemi-nested, semi-nested, one-way fully nested, one-way hemi-nested, or one-way semi-nested PCR approach. In an embodiment, the method of the present invention is used to preferentially enrich a mixture of DNA at multiple loci, the method comprising performing multiplex pre-amplification of selected targets over a limited number of cycles, dividing the product into multiple aliquots and amplifying subpools of targets in individual reactions, and combining the reaction products of parallel subpools . Note that this approach can be used to perform targeted amplification in a manner that will result in low levels of allelic bias for 50 to 500 loci, for 500 to 5000 loci, for 5000 to 50,000 loci, or even for 50,000 to 500,000 loci. In an embodiment, the primers carry tags compatible with partial or full sequencing.

[655] The workflow may include (1) extraction of DNA, such as plasma DNA, (2) preparation of a fragment library with universal adapters at both ends of the fragments, (3) amplification of the library using universal adapter-specific primers, (4) separation amplified sample "library" into multiple aliquots, (5) performing multiplex amplifications on the aliquots (e.g., approximately 100-plex, 1000-plex, or 10,000-plex with one target-specific primer per target and a tag-specific primer), (6) pooling the aliquots one sample, (7) sample barcoding, (8) sample mixing and concentration adjustment, (9) sample sequencing. The workflow may include several substeps that contain one of the listed steps (for example, step (2) of library preparation may include three enzymatic steps (blunt end, dA tail extension and adapter ligation) and three purification steps). Workflow steps can be combined, separated, or performed in different orders (for example, barcoding and sample merging).

[656] It is important to note that library amplification can be done in such a way that it is biased to more efficiently amplify short fragments. In this way, shorter sequences, such as fragments of mononucleosomal DNA in the form of cell-free fetal DNA (of placental origin) found in the bloodstream of pregnant women, can be preferentially amplified. Please note that PCR assays may have labels, such as sequencing tags (usually a truncated form of 15-25 bases). After multiplexing, the sample PCR multiplexes are combined and then tags are completed (including barcoding) using tag-specific PCR (ligation can also be performed). Additionally, full sequencing tags can be added in the same reaction as multiplexing. In the first cycles, targets can be amplified using target-specific primers, subsequently tag-specific primers come into play to complete the SQ adapter sequence. PCR primers may not contain labels. Sequencing tags can be added to amplification products by ligation.

[657] In an embodiment, highly multiplexed PCR followed by evaluation of the amplified material by clonal sequencing may be used for various applications, such as detection of fetal aneuploidy. While traditional multiplex PCR assays evaluate up to fifty loci simultaneously, the approach presented here can be used to simultaneously evaluate more than 50 loci simultaneously, more than 100 loci simultaneously, more than 500 loci simultaneously, more than 1000 loci simultaneously, more than 5000 loci simultaneously, more than 10,000 loci at the same time, more than 50,000 loci at the same time and more than 100,000 loci at the same time. Experiments have shown that up to 10,000 different loci can be assessed simultaneously, in a single reaction, with efficiency and specificity good enough to make non-invasive prenatal aneuploidy diagnoses and/or replicate various recognitions with high accuracy. Assays can be combined into a single reaction with the entire sample, such as a plasma-derived scDNA sample, a fraction thereof, or a further processed derivative of the scDNA sample. The sample (e.g., scDNA or derivative) can also be divided into several parallel multiplex reactions. Optimal sample separation and multiplexing is determined by compromising various technical characteristics. Due to the limited amount of material, dividing a sample into multiple fractions can result in sampling noise, increased processing time, and increased potential for error. Conversely, higher multiplexing may result in more spurious amplification and greater disparity in amplification, both of which can reduce test performance.

[658] Two critical interrelated considerations when using the methods presented herein are the limited amount of starting sample (eg, plasma) and the number of starting molecules in that material from which allele frequency or other measurements are obtained. If the number of starting molecules falls below a certain level, random sampling noise becomes significant and can affect the accuracy of the test. Typically, data of sufficient quality to make a noninvasive diagnosis of prenatal aneuploidy can be obtained if measurements are made on a sample containing the equivalent of 500–1000 parent molecules per target locus. There are several ways to increase the number of individual measurements, such as increasing the sample volume. Each manipulation applied to a sample also potentially results in material loss. It is important to characterize the losses incurred due to various manipulations and avoid them or, if necessary, improve the performance of certain manipulations to avoid losses that could degrade test performance.

[659] In an embodiment, potential losses in downstream steps can be reduced by amplifying all or part of the original sample (eg, a scDNA sample). Various methods are available to amplify all the genetic material in a sample, increasing the volume available for subsequent procedures. In an embodiment of ligation-mediated PCR (LM-PCR), DNA fragments are amplified by PCR after ligation of one single adapter, two different adapters, or many separate adapters. An embodiment of multiple displacement amplification (MDA) uses phi-29 polymerase to amplify all DNA isothermally. DOP-PCR and its variations use random priming to amplify DNA from the starting material. Each method has certain characteristics, such as uniformity of amplification in all represented regions of the genome, efficiency of capture and amplification of source DNA, and amplification performance depending on fragment length.

[660] In an embodiment, LM-PCR can be used with a single heteroduplex adapter having a 3'-tyrosine. A heteroduplex adapter allows the use of a single adapter molecule, which can be converted into two separate sequences at the 5' and 3' ends of the original DNA fragment during the first round of PCR. In an embodiment, the amplified library can be fractionated by size or product separation, such as AMPURE, TASS, or other similar methods. Before ligation, the DNA sample can be blunt ended and then a single adenosine base is added to the 3' end. Before ligation, the DNA can be digested with a restriction enzyme or some other digestion method. During ligation, the 3'-adenosine of the sample fragments and the additional 3'-tyrosine overhang of the adapter can increase the ligation efficiency. The extension step of the PCR amplification may be time limited to reduce amplification from fragments longer than about 200 bp, about 300 bp, about 400 bp, about 500 bp. or approximately 1000 bp. A series of reactions were carried out using the conditions specified in commercially available kits; as a result, less than 10% of the sample DNA molecules were successfully ligated. To achieve this, a series of optimizations of reaction conditions improved ligation to approximately 70%.

[661] Mini-PCR

[662] The following mini-PCR method is desirable for samples containing short nucleic acids, digested nucleic acids or fragmented nucleic acids such as scDNA. Traditional PCR assay designs result in significant losses of individual fetal molecules, but losses can be significantly reduced by developing very short PCR assays called mini-PCR assays. Fetal cDNA in maternal serum is highly fragmented and fragment sizes are approximately Gaussian with a mean of 160 bp, a standard deviation of 15 bp, and a minimum size of approximately 100 bp. and a maximum size of approximately 220 bp. The distribution of starting and ending positions of fragments relative to target polymorphisms, although not necessarily random, varies widely among individual targets and among all targets collectively, and a polymorphic site of one particular target locus among different fragments originating from that locus can occupy any position from from the beginning to the end. Note that the term mini-PCR can also refer to normal PCR without additional restrictions or limitations.

[663] During PCR, amplification will only occur from template DNA fragments containing both forward and reverse primer sites. Since fetal scDNA fragments are short, the probability of both primer sites being present and the probability that a fetal fragment of length L contains both forward and reverse primer sites is the ratio of the amplicon length to the fragment length. Under ideal conditions, tests in which the amplicon is 45, 50, 55, 60, 65, or 70 bp will successfully amplify from, respectively, 72, 69, 66, 63, 59, or 56% of the available template fragment molecules. Amplicon length is the distance between the 5' ends of the forward and reverse priming sites. Amplicon lengths that are shorter than those typically used by methods known in the art can result in more efficient measurements of desired polymorphic loci while only requiring short sequence reads. In an embodiment, a significant proportion of the amplicons must be less than 100 bp, less than 90 bp, less than 80 bp, less than 70 bp, less than 65 bp, less than 60 bp, less than 55 bp, less than 50 bp or less than 45 bp

[664] Please note that short assays such as those presented herein are generally avoided in methods known in the art because they are unnecessary and they place significant constraints on primer design by limiting primer length, characteristics annealing and distance between forward and reverse primers.

[665] Also note that there is the possibility of biased amplification if the 3' end of any primer is within approximately 1-6 bases of a polymorphic site. This single base difference at the site of initial polymerase binding can result in preferential amplification of one allele, which can alter observed allele frequencies and impair efficiency. All of these limitations make it very difficult to identify primers that will successfully amplify a particular locus and, furthermore, to design large sets of primers that are compatible in the same multiplex reaction. In an embodiment, the 3' end of the internal forward and reverse primers are designed to hybridize to a region of DNA upstream of the polymorphic site and are separated from the polymorphic site by a small number of bases. Ideally, the number of bases would be 6 to 10 bases, but could equally be 4 to 15 bases, three to 20 bases, two to 30 bases, or 1 to 60 bases, and essentially achieve the same end .

[666] A multiplex PCR may involve one PCR cycle in which all targets are amplified, or may involve one PCR cycle followed by one or more cycles of nested PCR or some variant of nested PCR. Nested PCR consists of a subsequent cycle or cycles of PCR amplification using one or more new primers that are internally linked by at least one base pair to the primers used in the previous cycle. Nested PCR reduces the number of false amplification targets by amplifying in subsequent reactions only those amplification products from the previous one that have the correct internal sequence. Reducing the number of off-target amplification targets increases the number of useful measurements that can be obtained, especially with sequencing. Nested PCR typically entails the design of primers completely internal to the previous primer binding sites, which necessarily increases the minimum DNA segment size required for amplification. For samples such as plasma scDNA, in which the DNA is highly fragmented, a larger assay size reduces the number of individual scDNA molecules from which a measurement can be obtained. In an embodiment, to compensate for this effect, a partially nested approach can be used in which one or both second-round primers overlap the first binding sites, extending a number of bases within to achieve additional specificity with minimal increase in overall assay size.

[667] In an embodiment, a multiplex pool of PCR assays is designed to amplify potentially heterozygous SNPs or other polymorphic or non-polymorphic loci on one or more chromosomes, and these assays are used in a single DNA amplification reaction. The number of PCR assays can be from 50 to 200 PCR assays, from 200 to 1000 PCR assays, from 1000 to 5000 PCR assays, or from 5000 to 20,000 PCR assays (respectively, from 50 to 200 plex, from 200 to 1000 plex , from 1000- to 5000-plex, from 5000- to 20000-plex, more than 20000-plex analysis). In an embodiment, a multiplex pool of approximately 10,000 PCR assays (10,000-plex) is designed to amplify potentially heterozygous SNP loci on chromosomes X, Y, 13, 18, 21 and 1 or 2, and these assays are used in a single reaction to amplify scDNA obtained from a plasma sample of material, chorionic villus samples, amniocentesis samples, single cells or small numbers of cells, other body fluids or tissues, cancer or other genetic material. The SNP frequencies of each locus can be determined by clonal or other amplicon sequencing techniques. Statistical analysis of allele frequency distributions or ratios of all assays can be used to determine whether a sample contains trisomy on one or more of the chromosomes included in the test. In another embodiment, the original scDNA samples are split into two samples and parallel 5000-plex assays are performed. In another embodiment, the original scDNA samples are divided into n samples and parallel (~10,000/n)-plex assays are performed, where n is 2 to 12, or 12 to 24, or 24 to 48, or 48 to 96. Data is collected and analyzed in the same way as already described. Please note that this method is equally suitable for detecting translocations, deletions, duplications and other chromosomal abnormalities.

[668] In an embodiment, tails having no homology to the target genome can also be added to the 3' or 5' end of any of the primers. These tails facilitate subsequent manipulations, procedures or measurements. In an embodiment, the tail sequence may be the same for the target-specific forward and reverse primers. In an embodiment, different tails may be used for the target-specific forward and reverse primers. In an embodiment, multiple different tails may be used for different loci or sets of loci. Some tails may be common to all loci or to a subset of loci. For example, the use of forward and reverse tails corresponding to the forward and reverse sequences required by any of the existing sequencing platforms can provide forward sequencing after amplification. In an embodiment, the tails can be used as common priming sites among all amplified targets, which can be used to add other useful sequences. In some embodiments, the internal primers may contain a region that is designed to hybridize either upstream or downstream of a target locus (eg, a polymorphic locus). In some embodiments, the primers may contain a molecular barcode. In some embodiments, the primer may comprise a universal priming sequence designed to enable PCR amplification.

[669] In an embodiment, a 10,000-plex PCR assay is pooled such that the forward and reverse primers have tails corresponding to the desired forward and reverse sequences needed by a high-throughput sequencing tool (often called a massively parallel sequencing tool) such as HISEQ, GAIIX, or MYSEQ, available from ILLUMINA. In addition, an additional 5' sequence is included in the sequencing tails, which can be used as a priming site in subsequent PCR to add nucleotide barcode sequences to amplicons, allowing multiplex sequencing of multiple samples in a single lane of a high-throughput sequencing instrument .

[670] In an embodiment, a 10,000-plex PCR assay is pooled such that the reverse primers have tails corresponding to the desired reverse sequences needed by a high-throughput sequencing instrument. After amplification with the first 10,000-plex assay, subsequent PCR amplification can be performed using another 10,000-plex pool having partially nested forward primers (e.g., nested 6 bases) for all targets, and a reverse primer corresponding to the reverse sequencing tail included in first cycle. This subsequent round of partially nested amplification with only one target-specific primer and a universal primer limits the required assay size, reducing sampling noise, but significantly reduces the number of spurious amplicons. Sequencing tags can be added to the supplied ligation adapters and/or as part of PCR probes so that the tag is part of the final amplicon.

[671] Tumor fraction influences test performance. There are several ways to enrich the tumor DNA fraction found in a patient's plasma. The tumor fraction can be increased using the previously presented LM-PCR method, which has already been discussed, as well as by targeted removal of long fragments. In an embodiment, prior to multiplex PCR amplification of target loci, an additional multiplex PCR reaction may be performed to selectively remove long and primarily maternal fragments corresponding to the target loci in the subsequent multiplex PCR. Additional primers are designed to anneal to a region located at a greater distance from the polymorphism than expected among cell-free fetal DNA fragments. These primers can be used in a single-cycle multiplex PCR reaction before multiplex PCR of target polymorphic loci. These distal primers are labeled with a molecule or fragment that can provide selective recognition of the labeled DNA fragments. In an embodiment, these DNA molecules can be covalently modified with a biotin molecule, which allows the newly formed double-stranded DNA containing these primers to be removed after one round of PCR. The double-stranded DNA produced during this first cycle is likely of maternal origin. Removal of the hybrid material can be accomplished using magnetic streptavidin microbeads. There are other tagging methods that can also work well. In an embodiment, size selection methods may be used to enrich the sample with shorter DNA strands; for example, less than about 800 bp, less than about 500 bp. or less than approximately 300 bp. After this, amplification of short fragments can continue as usual.

[672] The mini-PCR method of this invention provides highly multiplexed amplification and analysis of hundreds to thousands or even millions of loci in a single reaction from a single sample. At the same time, detection of amplified DNA can be multiplexed; tens to hundreds of samples can be multiplexed into a single sequencing lane using barcoding PCR. This multiplex detection has been successfully tested to 49-plex and much higher multiplexing is now possible. In fact, this allows hundreds of samples to be genotyped for thousands of SNPs in a single sequencing run. For these samples, the method allows determination of genotype and degree of heterozygosity and simultaneously determination of copy number, both of which can be used for the purpose of detecting aneuploidy. It can be used as part of a mutation dosing method. This method can be used for any amount of DNA or RNA, and the target regions can be SNPs, other polymorphic regions, non-polymorphic regions, and combinations thereof.

[673] In some embodiments, ligation-mediated universal-PCR amplification of fragmented DNA can be used. Versatile ligation-mediated PCR amplification can be used to amplify plasma DNA, which can then be separated into multiple parallel reactions. It can also be used to preferentially amplify short fragments, thereby enriching the tumor fraction. In some embodiments, adding tags to fragments by ligation may allow detection of shorter fragments, use of shorter portions of target sequence-specific primers, and/or annealing at higher temperatures, thereby reducing nonspecific reactions.

[674] The methods described herein can be used for a number of purposes where there is a target set of DNA that is mixed with some amount of contaminating DNA. In some embodiments, the target DNA and the contaminating DNA may be from people who are genetically related. For example, genetic abnormalities in the fetus (target) can be detected from maternal plasma, which contains fetal (target) DNA as well as maternal (contaminating) DNA; abnormalities include whole chromosomal abnormalities (eg, aneuploidy), partial chromosomal abnormalities (eg, deletions, duplications, inversions, translocations), polynucleotide polymorphisms (eg, STR simple tandem repeat), single nucleotide polymorphisms, and/or other genetic abnormalities or differences. In some embodiments, the target and contaminating DNA may be from the same individual, but the target and contaminating DNA differ in one or more mutations, such as in the case of cancer. (See, e.g., H. Mamon et al. Preferential Amplification of Apoptotic DNA from Plasma: Potential for Enhancing Detection of Minor DNA Alterations in Circulating DNA . Clinical Chemistry 54:9 (2008). In some embodiments, DNA may be detected in cell culture supernatant (apoptosis). In some embodiments, apoptosis can be induced in biological samples (eg, blood) for subsequent library preparation, amplification, and/or sequencing. A number of workflows and protocols for achieving this goal are presented elsewhere in this description.

[675] In some embodiments, the target DNA can come from single cells, from DNA samples consisting of less than one copy of the target genome, from small amounts of DNA, from DNA of mixed origin (for example, plasma from a cancer patient and a tumor: a mixture of healthy and cancerous DNA, transplantation, etc.), from other body fluids, from cell cultures, from culture supernatants, from forensic DNA samples, from ancient DNA samples (e.g. insects encased in amber), from other DNA samples and their combinations.

[676] In some embodiments, small amplicons can be used. Short amplicon sizes are especially suitable for fragmented DNA (see, for example, A. Sikora, et al. Detection of increased amounts of cell-free fetal DNA with short PCR amplicons. Clin Chem. 2010 Jan;56(1):136-8 .)

[677] The use of small amplicons can provide some significant advantages. Small amplicon sizes can lead to optimization of amplification efficiency. Short amplicon sizes tend to produce shorter products, so nonspecific priming is less likely to occur. Shorter products may be clustered more densely on a sequencing flow cell because the clusters will be smaller. Please note that the methods presented here may work equally well for longer PCR amplicons. If necessary, the amplicon length can be increased, for example, when sequencing large sections of the sequence. Targeted 146-plex amplification experiments were performed on single cells and on genomic DNA with positive results, with assays ranging from 100 to 200 bp in length as the first step in the nested PCR protocol.

[678] In some embodiments, the methods provided herein can be used to amplify and/or detect SNPs, copy number, nucleotide methylation, mRNA levels, other types of RNA expression levels, other genetic and/or epigenetic features. The mini-PCR methods presented herein can be used in conjunction with next-generation sequencing; it can be used with other following methods such as microarrays, digital PCR enumeration, real-time PCR, mass spectrometry analysis, etc.

[679] In some embodiments, the mini-PCR amplification methods presented herein can be used as part of a method for accurately quantifying populations present in a minority. It can be used for absolute quantitation using peak calibrators. It can be used to quantify mutations/minor alleles through very deep sequencing and can be performed in a highly multiplexed manner. It can be used to routinely test paternity and the identity of relatives or ancestors in humans, animals, plants or other creatures. Can be used for forensic examination. It can be used for rapid genotyping and copy number (CN) analysis on any type of material, such as amniotic fluid and chorionic villus sampling, semen, product of conception (POC). It can be used for single cell analysis, such as genotyping embryo biopsy samples. It can be used to rapidly analyze embryos (less than one, one or two days after biopsy) by targeted sequencing using mini-PCR.

[680] In some embodiments, mini-PCR amplification methods can be used to analyze tumors: a tumor biopsy is often a mixture of healthy and tumor cells. Targeted PCR allows deep sequencing of SNPs and loci with virtually no background sequences. It can be used to analyze copy number and loss of heterozygosity of tumor DNA. This tumor DNA may be present in many different body fluids or tissues of patients with tumors. It can be used to detect tumor recurrence and/or tumor screening. Can be used to control seed quality. Can be used for raising animals or fishing. Note that any of these methods can be used equally well to target non-polymorphic loci for ploidy discrimination.

[681] Some literature describing some of the fundamental techniques that underlie the methods presented herein include: (1) Wang HY, Luo M, Tereshchenko IV, Frikker DM, Cui X, Li JY, Hu G, Chu Y , Azaro MA, Lin Y, Shen L, Yang Q, Kambouris ME, Gao R, Shih W, Li H. Genome Res. 2005 Feb;15(2):276-83. Department of Molecular Genetics, Microbiology and Immunology/The Cancer Institute of New Jersey, Robert Wood Johnson Medical School, New Brunswick, New Jersey 08903, USA. (2) High-throughput genotyping of single nucleotide polymorphisms with high sensitivity. Li H, Wang HY, Cui X, Luo M, Hu G, Greenawalt DM, Tereshchenko IV, Li JY, Chu Y, Gao R. Methods Mol Biol. 2007;396 - PubMed PMID: 18025699. (3) A method involving multiplexing an average of 9 sequencing assays is described in: Nested Patch PCR enables highly multiplexed mutation discovery in candidate genes. Varley KE, Mitra RD. Genome Res. 2008 Nov;18(11):1844-50. Epub 2008 Oct 10. Please note that the methods disclosed herein allow for multiplexing that is several orders of magnitude greater than in the above references.

[682] Sample sets

[683] In one aspect, the invention is a kit, such as a kit for amplifying target loci in a nucleic acid sample to detect deletions and/or duplications of chromosomal segments or entire chromosomes using any of the methods presented herein. In some embodiments, the kit may include any of the primer libraries of the invention. In an embodiment, the set includes a plurality of internal forward primers and, optionally, a plurality of internal reverse primers, and optionally external forward primers and external reverse primers, wherein each of the primers is designed to hybridize to a region of DNA located immediately upstream and/or downstream of one of the target sites (eg, polymorphic sites) on the target chromosome(s) or chromosome segment(s), and optionally additional chromosomes or chromosome segments. In some embodiments, the kit includes instructions for using a library of primers to amplify target loci, for example, to detect one or more deletions and/or duplications of one or more chromosome segments or entire chromosomes using any of the methods presented herein.

[684] In some embodiments, the kits of the invention provide primer pairs for detecting chromosomal aneuploidy and detecting ICH, such as primer pairs for massive multiplex reactions for detecting chromosomal aneuploidy such as ICH (CoNVERGe) (Copy Number Variant Events Revealed Genotypically) and/ or ONV. In these embodiment options, sets may include at least 100, 200, 250, 300, 500, 1000, 2000, 2500, 3000, 5000, 10,000, 20,000, 25000, 28000, 50,000 or 75000 and not more than 200, 250, 300, 500, 1000, 2000, 2500, 3000, 5000, 10000, 20000, 25000, 28000, 50000, 75000 or 100000 primer pairs supplied together. The primer pairs may be contained in a single container, such as a single tube or box, or multiple tubes or boxes. In some embodiments, the primer pairs are pre-qualified by a commercial supplier and sold together, and in other embodiments, the customer selects customized gene targets and/or primers and the commercial supplier prepares and ships a pool of primers to the customer in either a tube or multiple tubes. In some exemplary embodiments, the kits include primers for detecting both ICH and NNV, especially ICH and NVC, which are known to correlate with at least one type of cancer.

[685] Kits for detecting circulating DNA in accordance with some embodiments of the present invention include standards and/or controls for detecting circulating DNA. For example, in some embodiments, standards and/or controls are sold and optionally shipped and packaged along with primers used to perform the amplification reactions provided herein, such as primers to perform CoNVERGe. In some embodiments, the controls include polynucleotides, such as DNA, including isolated genomic DNA, that exhibits one or more chromosomal aneuploidies, such as ICH, and/or includes one or more ONVs. In some embodiments, standards and/or controls are referred to as PlasmArt standards and include polynucleotides that are identical in sequence to regions of the genome that are known to exhibit ICH, particularly in certain inherited diseases and in certain disease states such as cancer, as well as size distribution , which reflects the distribution of scDNA fragments naturally occurring in plasma. Examples of ways to create PlasmArt standards are provided in the examples provided in this document. Typically, genomic DNA from a source known to include a chromosomal aneuploidy is isolated, fragmented, purified, and size selected.

[686] Accordingly, standards and/or controls for artificial scDNA polynucleotides can be prepared by adding samples of isolated polynucleotides prepared as described above to DNA samples that are not known to exhibit chromosomal aneuploidy and/or ONV, at concentrations similar to observed for scDNA in vivo, for example, from 0.01% to 20%, from 0.1 to 15%, or from 0.4 to 10% DNA in such a liquid. These standards/controls may be used as controls for assay design, characterization, development, and/or validation, and as quality control standards during testing, such as cancer testing performed in a CLIA laboratory, and/or as standards included in research use or diagnostic test kits only.

[687] Approximate methods of normalization/correction

[688] In some embodiments, measurements for different loci, chromosomal segments, or chromosomes are adjusted for bias, such as bias due to differences in GC content or bias due to other differences in amplification efficiency or to account for sequencing errors. In some embodiments, measurements for different alleles of the same locus are adjusted for differences in metabolism, apoptosis, histones, inactivation, and/or amplification between alleles. In some embodiments, measurements for different alleles of the same RNA locus are adjusted to account for differences in transcription rate or stability between different RNA alleles.

[689] Example methods for phasing genetic data

[690] In some embodiments, genetic data is phased using the methods presented herein or any known method of phasing genetic data (see, for example, PCT Publication No. WO2009/105531, filed February 9, 2009, and PCT Publication No. WO2010/017214 , filed August 4, 2009; US Patent Application Publication No. 2013/0123120, November 21, 2012; US Patent Application Publication No. 2011/0033862, filed October 7, 2010; US Patent Application Publication No. 2011/0033862, filed 19 August 2010; US Patent Application Publication No. 2011/0178719, filed February 3, 2011; US Patent No. 8515679, filed March 17, 2008; US Patent Application Publication No. 2007/0184467, filed November 22, 2006; US Patent No. 2008/0243398, filed March 17, 2008, and US Provisional Patent Application Serial No. 61/994791, filed May 16, 2014, each of which is incorporated herein by reference in its entirety). In some embodiments, the phase is determined for one or more regions that are known or suspected to contain the ICH of interest. In some embodiments, phase is also determined for one or more regions flanking the ICH region(s) and/or for one or more reference regions. In one embodiment, an individual's genetic data is phased by inference by measuring the individual's haploid tissue, for example, by measuring one or more sperm or eggs. In one embodiment, an individual's genetic data is phased by inference using measured genotypic data from one or more first-degree relatives, such as the individual's parents (eg, sperm from the individual's father) or siblings.

[691] In one embodiment, an individual's genetic data is phased when DNA or RNA is diluted into one or more wells, for example, using digital PCR. In some embodiments, the DNA or RNA is diluted to the point where no more than approximately one copy of each haplotype is expected in each well, and then the DNA or RNA in one or more wells is measured. In some embodiments, cells are arrested in the mitotic phase where the chromosomes are tightly bundled and microfluidics is used to place individual chromosomes into individual wells. Because the DNA or RNA is dilute, it is unlikely that there will be more than one haplotype in a single fraction (or tube). Thus, in fact, there can be one DNA molecule in a test tube, which makes it possible to determine the haplotype on one DNA or RNA molecule. In some embodiments, the method includes dividing a DNA or RNA sample into a plurality of fractions such that at least one of the fractions includes one chromosome or one chromosome segment of a pair of chromosomes, and genotyping (e.g., determining the presence of two or more polymorphic loci) the DNA sample or RNA in at least one of the fractions, thereby defining the haplotype. In some embodiments, genotyping includes sequencing (such as fractional or single molecule sequencing), SNP array to detect polymorphic loci, or multiplex PCR. In some embodiments, genotyping includes using an SNP array to detect polymorphic loci, such as at least 100; 200; 500; 750; 1000; 2000; 5000; 7500; 10000; 20000; 25000; 30000; 40000; 50000; 75000; or 100,000 different polymorphic loci. In some embodiments, genotyping involves the use of multiplex PCR. In some embodiments, the method includes contacting a sample in a fraction with a library of primers that simultaneously hybridize with at least 100; 200; 500; 750; 1000; 2000; 5000; 7500; 10000; 20000; 25000; 30000; 40000; 50000; 75000; or 100,000 different polymorphic loci (such as SNPs) to obtain a reaction mixture; and exposing the reaction mixture to primer extension reaction conditions to produce amplification products, which are measured by a high-throughput sequencer to obtain sequencing data. In some embodiments, the RNA (eg, mRNA) is sequenced. Because mRNA contains only exons, mRNA sequencing allows the determination of alleles at polymorphic loci (such as SNPs) over large distances in the genome, such as several megabases. In some embodiments, an individual's haplotype is determined by chromosome sorting. A typical way to sort chromosomes involves arresting cells in the mitotic phase, when the chromosomes are in dense bundles, and using microfluidics to place individual chromosomes into individual wells. Another method involves collecting individual chromosomes using FACS-mediated single chromosome sorting. Standard techniques (such as sequencing or array) can be used to identify alleles on a single chromosome to determine an individual's haplotype.

[692] In some embodiments, an individual's haplotype is determined by long-read sequencing, such as using Illumina's Moleculo technology. In some embodiments, the library preparation step includes cutting the DNA into fragments, e.g., ~10 kb fragments, diluting the fragments and placing them in wells (so that there are approximately 3000 fragments in one well), amplifying the fragments in each well by PCR of long fragments, cutting into short fragments and barcoding the fragments, and combining the barcoded fragments from each well to sequence them together. After sequencing, computational steps include separating the reads from each well based on the attached barcodes and grouping them into fragments, assembling the fragments in overlapping heterozygous HNVs into haplotype blocks and statistically phasing the blocks based on a stepwise reference panel, and obtaining long haplotype contigs.

[693] In some embodiments, an individual's haplotype is determined using data from a relative of the individual. In some embodiments, an SNP matrix is used to determine the presence of at least 100; 200; 500; 750; 1000; 2000; 5000; 7500; 10000; 20000; 25000; 30000; 40000; 50000; 75000; or 100,000 different polymorphic loci in a DNA or RNA sample from an individual and his relative. In some embodiments, the method includes contacting a DNA sample from the individual and/or a relative of the individual with a library of primers that simultaneously hybridize with at least 100; 200; 500; 750; 1000; 2000; 5000; 7500; 10000; 20000; 25000; 30000; 40000; 50000; 75000; or 100,000 different polymorphic loci (such as SNPs) to obtain a reaction mixture; and exposing the reaction mixture to primer extension reaction conditions to produce amplification products, which are measured by a high-throughput sequencer to obtain sequencing data.

[694] In one embodiment, an individual's genetic data is phased using a computer program that uses population haplotype frequencies to determine the most likely phase, for example, HapMap-based phasing. For example, haploid data sets can be inferred directly from diploid data using statistical methods that use known blocks of haplotypes in the general population (eg, those created for the public HapMap project and for the Perlegen human haplotype project). A haplotype block is essentially a series of correlated alleles that occur repeatedly in different populations. Because these haplotype blocks are often ancient and widespread, they can be used to predict haplotypes based on diploid genotypes. Publicly available algorithms that solve this problem include imperfect phylogenetic approaches, Bayesian approaches based on conjugate inputs, and inputs from population genetics. Some of these algorithms use a hidden Markov model.

[695] In one embodiment, individual genetic data is phased using an algorithm that estimates haplotypes based on genotype data, such as an algorithm using localized haplotype clustering (see, e.g. , Browning and Browning, “Rapid and Accurate Haplotype Phasing and Missing -Data Inference for Whole-Genome Association Studies By Use of Localized Haplotype Clustering” Am J Hum Genet Nov 2007;81(5):1084-1097, incorporated herein by reference in its entirety). An example program is Beagle version: 3.3.2 or version 4 (available at hfaculty.washington.edu/browning/beagle/beagle.html, the document is incorporated herein by reference in its entirety).

[696] In one embodiment, an individual's genetic data is phased using an algorithm that estimates haplotypes based on genotype data, such as an algorithm that uses linkage disequilibrium decay with distance, the order and spacing of genotyped markers, imputation of missing data, estimation recombination rates or a combination thereof (see, for example, Stephens and Scheet, “Accounting for Decay of Linkage Disequilibrium in Haplotype Inference and Missing-Data Imputation” Am. J. Hum. Genet. 76:449-462, 2005, entire document included here via link). An example program is PHASE v.2.1 or v2.1.1. (available at stephenslab.uchicago.edu/software.html, the document is incorporated herein by reference in its entirety).

[697] In one embodiment, an individual's genetic data is phased using an algorithm that estimates haplotypes from population genotype data, such as an algorithm that allows cluster membership to vary continuously along a chromosome in accordance with a hidden Markov model. This approach is flexible, allowing both “block” models of linkage disequilibrium and a gradual decrease in linkage disequilibrium with distance (see, for example, Scheet and Stephens, “A fast and flexible statistical model for large-scale population genotype data: applications to inferring missing genotypes and haplotypic phase.” Am J Hum Genet 78:629–644, 2006, incorporated herein by reference in its entirety). An example program is fastPHASE (available at stephenslab.uchicago.edu/software.html, the document is incorporated herein by reference in its entirety).

[698] In one embodiment, an individual's genetic data is phased using a conditional genotype imputation method, such as a method that uses one or more of the following reference data sets: a HapMap data set, data sets of controls genotyped on multiple SNP chips, and densely typed samples from the 1,000 Genomes Project. An exemplary approach is a flexible modeling framework that improves accuracy and integrates information from multiple reference panels (see, for example, Howie, Donnelly, and Marchini (2009) “A flexible and accurate genotype imputation method for the next generation of genome-wide association studies " PLoS Genetics 5(6): e1000529, 2009, incorporated herein by reference in its entirety). Example programs are IMPUTE or IMPUTE version 2 (also known as IMPUTE2) (available at mathgen.stats.ox.ac.uk/impute/impute_v2.html, which are incorporated herein by reference in their entirety).

[699] In one embodiment, an individual's genetic data is phased using an algorithm that infers haplotypes, such as an algorithm that infers haplotypes according to a genetic model of fusion with recombination, such as developed by Stevens in PHASE v2.1. Major algorithmic improvements rely on the use of binary trees to represent sets of candidate haplotypes for each individual. These binary tree representations: (1) speed up calculations of posterior haplotype probabilities by avoiding the redundant operations done in PHASE v2.1, and (2) overcome the exponential aspect of the haplotype inference problem by intelligently exploring most likely paths (i.e., haplotypes) in binary trees (see, for example, Delaneau, Coulonges and Zagury, “Shape-IT: a new rapid and accurate algorithm for haplotype inference,” BMC Bioinformatics 9:540, 2008 doi:10.1186/1471-2105-9-540, incorporated herein by reference in its entirety. An example program is SHAPEIT (available at mathgen.stats.ox.ac.uk/genetics_software/shapeit/shapeit.html, the document is incorporated herein by reference in its entirety).

[700] In one embodiment, an individual's genetic data is phased using an algorithm that estimates haplotypes from population genotype data, such as an algorithm that uses haplotype fragment frequencies to derive empirically based probabilities for longer haplotypes. In some embodiments, the algorithm reconstructs haplotypes so that they have maximum local coherence (see, for example, Eronen, Geerts, and Toivonen, “HaploRec: Efficient and accurate large-scale reconstruction of haplotypes,”BMC Bioinformatics 7:542, 2006, incorporated herein by reference in its entirety). An example program is HaploRec, for example HaploRec version 2.3. (available at cs.helsinki.fi/group/genetics/haplotyping.html, the document is incorporated herein by reference in its entirety).

[701] In one embodiment, an individual's genetic data is phased using an algorithm that estimates haplotypes based on population genotype data, such as an algorithm that uses a split-ligation strategy and an algorithm based on expectation maximization (see, for example, Qin, Niu, and Liu, “Partition-Ligation-Expectation-Maximization Algorithm for Haplotype Inference with Single-Nucleotide Polymorphisms,” Am J Hum Genet. 71(5): 1242-1247, 2002, incorporated herein by reference in its entirety). An example program is PL-EM (available at people.fas.harvard.edu/~junliu/plem/click.html, the document is incorporated herein by reference in its entirety).

[702] In one embodiment, an individual's genetic data is phased using an algorithm that estimates haplotypes from population genotype data, such as an algorithm for simultaneously phasing genotypes into haplotypes and block partitioning. In some embodiments, an expectation maximization algorithm is used (see, for example, Kimmel and Shamir, “GERBIL: Genotype Resolution and Block Identification Using Likelihood,” Proceedings of the National Academy of Sciences of the United States of America (PNAS) 102: 158 -162, 2005, incorporated herein by reference in its entirety). An example program is GERBIL, available as part of the GEVALT program version 2 (available at the web address acgt.cs.tau.ac.il/gevalt/, the document is incorporated herein by reference in its entirety).

[703] In one embodiment, an individual's genetic data is phased using an algorithm that estimates haplotypes based on population genotype data, such as an algorithm that uses the EM algorithm to compute ML estimates of haplotype frequencies given genotype measurements that do not indicate phase. The algorithm also allows for the absence of some genotype measurements (for example, due to PCR failure). It also allows for multiple conditional substitutions of individual haplotypes (see, for example, Clayton, D. (2002), "SNPHAP: A Program for Estimating Frequencies of Large Haplotypes of SNPs", incorporated herein by reference in its entirety. An example program is SNPHAP (available at gene.cimr.cam.ac.uk/clayton/software/snphap.txt, the document is incorporated herein by reference in its entirety).

[704] In one embodiment, an individual's genetic data is phased using an algorithm that estimates haplotypes based on population genotype data, such as a haplotype inference algorithm based on genotype statistics collected for SNP pairs. This software can be used to relatively accurately phase large numbers of long genome sequences, such as those obtained from DNA templates. The example program takes a genotype matrix as input and outputs the corresponding haplotype matrix. (See, for example , Brinza and Zelikovsky, “2SNP: scalable phasing based on 2-SNP haplotypes,” Bioinformatics.22(3):371-3, 2006, incorporated herein by reference in its entirety). An example program is 2SNP (available at alla.cs.gsu.edu/~software/2SNP, the document is incorporated herein by reference in its entirety).

[705] In various embodiments, an individual's genetic data is phased using data on the probability of chromosome crossing over at different locations on a chromosome or chromosome segment (e.g., using recombination data that can be found in the HapMap database to create a recombination risk score for any interval) to model the dependence between polymorphic alleles on a chromosome or chromosome segment. In some embodiments, the number of alleles at polymorphic loci is calculated on a computer based on sequencing data or SNP array data. In some embodiments, a plurality of hypotheses are generated (e.g., computer-generated), each of which relates to a different possible state of a chromosome or chromosome segment (e.g., an overrepresentation of the copy number of a first homologous chromosome segment relative to a second homologous chromosome segment in the genome of one or more cells from an individual, duplication of the first homologous chromosome segment, deletion of the second homologous chromosome segment, or equal representation of the first and second homologous chromosome segments); for each hypothesis, a model (for example, a joint distribution model) of the expected number of alleles in polymorphic chromosome loci is built (for example, on a computer); the relative probability of each of the hypotheses is determined (eg, computer determined) using a joint distribution and allele counting model; and the hypothesis with the highest probability is selected. In some embodiments, constructing a joint distribution model for the number of alleles and the step of determining the relative likelihood of each hypothesis is performed using a method that does not require the use of a reference chromosome.

[706] In some embodiments, a sample (e.g., a biopsy such as a tumor biopsy, a blood sample, a plasma sample, a serum sample, or other sample that may contain primarily or only cells, DNA, or RNA of the ICH of interest) from the individual is analyzed to determine the phase for one or more regions known or suspected to contain the ICH of interest (eg, deletion or duplication). In some embodiments, the sample has a high tumor fraction (eg, 30, 40, 50, 60, 70, 80, 90, 95, 98, 99, or 100%).

[707] In some embodiments, the sample has a haplotype imbalance or any aneuploidy. In some embodiments, the sample includes any mixture of two types of DNA, where the two types have different ratios of two haplotypes and have at least one haplotype. For example, in the case of a tumor, normal tissue is 1:1 and tumor tissue is 1:0 or 1:2, 1:3, 1:4, etc. In some embodiments, at least 10 are analyzed; 100; 500; 1000; 2000; 3000; 5000; 8000; or 10,000 polymorphic loci to determine the phase of alleles at some or all loci. In some embodiments, the sample is from a cell or tissue that has been treated to produce aneuploidy, such as aneuploidy induced by long-term cell culture.

[708] In some embodiments, a large percentage or all of the DNA or RNA in the sample has the ICH of interest. In some embodiments, the ratio of DNA or RNA from one or more target cells containing the ICH of interest to total DNA or RNA in the sample is at least 80%, 85%, 90%, 95%, or 100%. For samples with a deletion, there is only one haplotype for the cells (either DNA or RNA) with the deletion. This first haplotype can be determined by standard methods for determining the identity of alleles present in the region of the deletion. In samples that contain only cells (or DNA or RNA) with a deletion, there will only be a signal from the first haplotype present in those cells. In samples that also contain a small number of cells (or DNA or RNA) without deletion (for example, a small number of benign cells), the weak signal from the second haplotype in those cells (or DNA or RNA) can be ignored. A second haplotype that is present in other cells, the DNA or RNA of an individual that lacks the deletion, can be determined by inference. For example, if the genotype of an individual's cells without a deletion is (AB, AB), and the phased data for the individual indicates that the first haplotype is (A, A); then another haplotype can be inferred as (B, B).

[709] For samples in which both cells (or DNA or RNA) with a deletion and cells (or DNA or RNA) without a deletion are present, the phase can still be determined. For example, graphs can be created in which the x-axis represents the linear position of individual loci along the chromosome and the y-axis represents the number of A allele reads as a proportion of the total (A + B) allele reads. In some embodiments, for a deletion, the pattern includes two central bands that represent SNPs for which the individual is heterozygous (the top band represents AB from cells without the deletion and A from cells with the deletion, and the bottom band represents AB from cells without the deletion and B from the cell with deletion). In some embodiments, the separation of the two bands increases as the proportion of cells with DNA or RNA with the deletion increases. Thus, the identity of the A alleles can be used to determine the first haplotype, and the identity of the B alleles can be used to determine the second haplotype.

[710] For samples with a duplication, an additional copy of the haplotype is present for the cells (either DNA or RNA) with the duplication. This haplotype of the duplicated region can be determined using standard methods for determining the identity of alleles present in increased abundance in the region of duplication, or the haplotype of the region that is not duplicated can be determined using standard methods for determining the identity of alleles present in decreased abundance. Once one haplotype is determined, another haplotype can be determined by inference.

[711] For samples in which there are both cells (or DNA or RNA) with duplication and cells (or DNA or RNA) without duplication, the phase can still be determined using a method similar to that presented above for deletions. For example, graphs can be created in which the x-axis represents the linear position of individual loci along the chromosome and the y-axis represents the number of A allele reads as a proportion of the total (A + B) allele reads. In some embodiments, for a deletion, the pattern includes two central bands that represent SNPs for which the individual is heterozygous (the top band represents AB from cells without duplication and AAB from cells with duplication, and the bottom band represents AB from cells without duplication and ABB from cells with duplication). In some embodiments, the separation of these two bands increases as the proportion of cells, DNA or RNA with duplication increases. Thus, the identity of the A alleles can be used to determine the first haplotype, and the identity of the B alleles can be used to determine the second haplotype. In some embodiments, the phase of one or more regions of ICH (e.g., the phase of at least 50, 60, 70, 80, 90, 95, or 100% of the polymorphic loci in the measured region) is determined for a sample (e.g., tumor biopsy or plasma sample) from an individual known to have cancer, and is used to analyze subsequent samples from the same individual in order to monitor the progression of cancer (eg, monitoring remission or recurrence of cancer). In some embodiments, a sample with a high tumor fraction (eg, a tumor biopsy or a plasma sample from an individual with a high tumor burden) is used to obtain phased data that is used to analyze subsequent samples with a lower tumor fraction (for example, a plasma sample from an individual undergoing being treated for cancer or in remission).

[712] In some embodiments, two or more of the methods presented herein are used to phase an individual's genetic data. In some embodiments, both a bioinformatics method (e.g., using population-based haplotype frequencies to determine the most likely phase) and a molecular biology method (e.g., any of the molecular phasing methods disclosed herein rather than prospective phased data based on bioinformatics). In some embodiments, phased data from other subjects (eg, previous subjects) is used to refine the population data. For example, phased data from other subjects can be added to population data to calculate inputs for possible haplotypes for another subject. In some embodiments, phased data from other subjects (eg, previous subjects) is used to calculate inputs for possible haplotypes for another subject.

[713] In some embodiments, probabilistic data may be used. For example, due to the probabilistic nature of the representation of DNA molecules in a sample, as well as various amplification and measurement errors, the relative number of DNA molecules measured from two different loci or from different alleles at a given locus does not always reflect the relative number of molecules in the mixture or at the individual. If one were to attempt to determine the genotype of a normal diploid individual at a given locus on an autosomal chromosome by sequencing DNA from the individual's plasma, one would expect that either only one allele (homozygous) or approximately equal numbers of two alleles (heterozygous) would be observed. If ten molecules of allele A were observed on this allele and two molecules of allele B were observed, it would not be clear whether the individual was homozygous at the locus and the two molecules of allele B were due to noise or contamination, or if the individual was heterozygous and fewer molecules of the allele B was caused by a random statistical change in the number of DNA molecules in the plasma, amplification bias, contamination, or any number of other reasons. In this case, the probability that an individual was homozygous and the corresponding probability that the individual was heterozygous could be calculated, and these probabilistic genotypes could be used in further calculations.

[714] Consider that for a given ratio of alleles, the probability that this ratio accurately represents the ratio of DNA molecules in an individual increases as the number of molecules observed increases. For example, if 100 molecules of A and 100 molecules of B were to be measured, the probability that the actual ratio is 50% is significantly higher than if 10 molecules of A and 10 molecules of B were to be measured. In one embodiment, Bayesian theory is used in combined with a detailed model of the data to determine the likelihood that a particular hypothesis is true given an observation. For example, if one were to consider two hypotheses, one that corresponds to a trisomic individual and one that corresponds to a disomic individual, then the probability that the disomic hypothesis would be correct would be significantly higher for the case where 100 molecules of each of the two alleles were observed , compared to the case where 10 molecules of each of the two alleles were observed. As the data becomes noisier due to bias, contamination, or some other source of noise, or as the number of observations at a given locus decreases, the probability that the maximum likelihood hypothesis will be true, given the observed data drops. In practice, it is possible to aggregate probabilities across multiple loci to increase confidence that the maximum likelihood hypothesis can be determined to be correct. In some embodiments, the probabilities are simply summed without taking into account recombination. In some embodiments, crossovers are taken into account in the calculations.

[715] In one embodiment, probabilistically phased data is used in determining copy number variation. In some embodiments, the probabilistically phased data is haplotype block frequency data based on a population data source, such as a HapMap database. In some embodiments, the probabilistic phasing data is molecularly derived haplotype data, such as dilution phasing where individual chromosome segments are diluted to one molecule per reaction, but due to stochastic noise the identity of the haplotypes may be completely unknown. In some embodiments, the probabilistically phased data is molecularly derived haplotype data where the identity of the haplotypes can be known with a high degree of confidence.

[716] Imagine a hypothetical case where a physician wanted to determine whether an individual's body had some cells with a deletion in a certain segment of a chromosome by measuring the DNA of that individual's plasma. The physician could use the information that if all the cells from which plasma DNA originated were diploid and had the same genotype, then for heterozygous loci the relative number of DNA molecules observed for each of the two alleles would fall into one distribution centered on 50% of the A alleles and 50% of the B alleles. However, if the fraction of cells from which plasma DNA originated had a deletion in a particular chromosome segment, then for heterozygous loci one would expect the relative number of DNA molecules observed for of each of the two alleles, will be divided into two distributions, one centered above 50% of the A allele for loci where a deletion of the chromosome segment containing the B allele has occurred, and one centered below 50% for loci where a deletion of the chromosome segment containing allele A. The greater the proportion of cells from which the plasma DNA originated contained the deletion, the further away from 50% these two distributions will be.

[717] In this hypothetical case, imagine a clinician who wants to determine whether an individual has had a deletion of a chromosomal region in a certain portion of cells in the individual's body. The clinician may draw blood from the individual into a vacutainer or other type of tube, centrifuge the blood, and isolate a layer of plasma. The clinician can isolate DNA from plasma, enrich DNA at target loci, possibly through targeted or other amplification, loci capture techniques, size expansion, or other enrichment techniques. The clinician may analyze, for example, by measuring the number of alleles in a set of SNPs, in other words, generating allele frequency data, enriched and/or amplified DNA, using an analysis such as qPCR, sequencing, microarray or other methods of measuring the amount of DNA in a sample. We will consider data analysis for a case in which a clinician amplified cell-free plasma DNA using a targeted amplification method and then sequenced the amplified DNA to obtain the following estimated possible data for six SNPs found on a segment of the chromosome that indicates cancer where the individual ended up heterozygous for these SNPs:

[718] SNP 1: 460 reads of allele A; 540 reads of allele B (46% A)

[719] SNP 2: 530 reads of allele A; 470 B allele reads (53% A)

[720] SNP 3: 40 reads of allele A; 60 reads of allele B (40% A)

[721] SNP 4: 46 reads of allele A; 54 B allele reads (46% A)

[722] SNP 5: 520 reads of allele A; 480 B allele reads (52% A)

[723] SNP 6: 200 reads of allele A; 200 reads of allele B (50% A)

[724] From this data set, it may be difficult to distinguish between the case where an individual is normal, where all cells are disomic, or where an individual may have cancer, with some portion of the cells whose DNA contributed to free-cell DNA. found in plasma, with a deletion or duplication on the chromosome. For example, the two highest likelihood hypotheses would be that the individual has a deletion in that chromosome segment with a tumor fraction of 6%, and where the deleted chromosome segment has a genotype for six SNPs (A, B, A, A, B, B) or (A, B, A, A, B, A). In this representation of an individual's genotype over a set of SNPs, the first letter in parentheses corresponds to the haplotype genotype for SNP 1, the second to SNP 2, etc.

[725] If a method were used to determine the haplotype of an individual on that chromosome segment and it was found that the haplotype for one of the two chromosomes was (A, B, A, A, B, B), this would be consistent with the maximum likelihood hypothesis. and the calculated probability that an individual has a deletion in this segment and therefore may have cancerous or precancerous cells will be significantly increased. On the other hand, if the haplotype (A, A, A, A, A, A) were found in an individual, then the likelihood that the individual has a deletion in that chromosome segment would be greatly reduced, and perhaps the likelihood of the hypothesis in the absence of a deletion will be higher (actual probability values will depend on other parameters such as, among other things, measured noise in the system).

[726] There are many ways to determine an individual's haplotype, many of which are described elsewhere in this document. This document provides a partial list and is not intended to be exhaustive. One method is a biological method in which individual DNA molecules are diluted until approximately one molecule from each chromosomal region is present in any given reaction volume, and then methods such as sequencing are used to measure the genotype. Another method is based on computer science, where population data on different haplotypes, combined with their frequencies, can be used in a probabilistic manner. Another way is to measure the diploid data of an individual along with one or many related individuals that are expected to share haplotype blocks with the individual, and then infer the haplotype blocks. Another method could be to sample tissue with a high concentration of a segment with a deletion or duplication and determine the haplotype based on allelic imbalance, for example, genotype measurements of a tumor tissue sample with a deletion could be used to determine phased data for that region of deletion, and these data could then be used to determine cancer regrowth after resection.

[727] In practice, it is common to measure more than 20 SNPs, more than 50 SNPs, more than 100 SNPs, more than 500 SNPs, more than 1000 SNPs, or more than 5000 SNPs on a given chromosome segment.

[728] Example Mutations

[729] Examples of mutations associated with a disease or disorder, such as cancer, or increased risk (eg, a higher than normal level of risk) for a disease or disorder, such as cancer, include single nucleotide variants (SNVs), multiple nucleotide mutations, deletions (eg , as a deletion of a region of 2 to 30 million base pairs), duplications or tandem repeats. In some embodiments, the mutation is in DNA, such as cDNA, extracellular mitochondrial DNA (scmDNA), extracellular DNA derived from nuclear DNA (scDNA), cellular DNA, or mitochondrial DNA. In some embodiments, the mutation is in RNA, such as scRNA, cellular RNA, cytoplasmic RNA, coding cytoplasmic RNA, non-coding cytoplasmic RNA, mRNA, miRNA, mitochondrial RNA, rRNA, or tRNA. In some embodiments, the mutation occurs at a higher frequency in subjects with a disease or disorder (such as cancer) than in subjects without the disease or disorder (such as cancer). In some embodiments, the mutation is indicative of cancer, such as a mutation that causes cancer. In some embodiments, the mutation is a driver mutation that plays a causative role in a disease or disorder. In some embodiments, the mutation is not a causal mutation. For example, in some forms of cancer multiple mutations accumulate, but some of them are not causative mutations. Mutations (such as those that are present at a higher frequency in subjects with a disease or disorder than in subjects without the disease or disorder) that are not causal may still be useful in diagnosing the disease or disorder. In some embodiments, the mutation is a loss-of-heterozygosity (LOH) for one or more microsatellites.

[730] In some embodiments, the subject is screened for one of several polymorphisms or mutations that the subject is known to have (e.g., to test for their presence, changes in the number of cells, DNA or RNA with those polymorphisms or mutations, or in remission cancer or re-development). In some embodiments, the subject is screened for the presence of one or more polymorphisms or mutations for which the subject is known to be at risk (eg, a subject who has a relative with the polymorphism or mutation). In some embodiments, the subject is screened for a panel of polymorphisms or mutations associated with a disease or disorder, such as cancer (e.g., at least 5, 10, 50, 100, 200, 300, 500, 750, 1000, 1500, 2000, or 5000 polymorphisms or mutations).

[731] Many coding variants associated with cancer are described in Abaan et al ., “The Exomes of the NCI-60 Panel: A Genomic Resource for Cancer Biology and Systems Pharmacology,” Cancer Research, July 15, 2013, and on the web -address dtp.nci.nih.gov/branches/btb/characterizationNCI60.html, each of these documents is incorporated herein by reference in its entirety). The NCI-60 Human Cancer Cell Panel consists of 60 different cell lines representing lung, colon, brain, ovarian, breast, prostate and kidney cancers, as well as leukemia and melanoma. The genetic variations that were identified in these cell lines consisted of two types: type I variants, which occur in the normal population, and type II variants, which are specific to cancer.

[732] Exemplary polymorphisms or mutations (such as deletions or duplications) are found in one or more of the following genes: TP53, PTEN, PIK3CA, APC, EGFR, NRAS, NF2, FBXW7, ERBBs, ATAD5, KRAS, BRAF, VEGF, EGFR , HER2, ALK, p53, BRCA, BRCA1, BRCA2, SETD2, LRP1B, PBRM, SPTA1, DNMT3A, ARID1A, GRIN2A, TRRAP, STAG2, EPHA3/5/7, POLE, SYNE1, C20orf80, CSMD1, CTNNB1, ERBB2, FBXW7 , KIT, MUC4, ATM, CDH1, DDX11, DDX12, DSPP, EPPK1, FAM186A, GNAS, HRNR, KRTAP4-11, MAP2K4, MLL3, NRAS, RB1, SMAD4, TTN, ABCC9, ACVR1B, ADAM29, ADAMTS19, AGAP10, AKT1 , AMBN, AMPD2, ANKRD30A, ANKRD40, APOBR, AR, BIRC6, BMP2, BRAT1, BTNL8, C12orf4, C1QTNF7, C20orf186, CAPRIN2, CBWD1, CCDC30, CCDC93, CD5L, CDC27, CDC42BPA, CDH9, CDKN2A, CHD8, CHEK2, CHRNA 9 , CIZ1, CLSPN, CNTN6, COL14A1, CREBBP, CROCC, CTSF, CYP1A2, DCLK1, DHDDS, DHX32, DKK2, DLEC1, DNAH14, DNAH5, DNAH9, DNASE1L3, DUSP16, DYNC2H1, ECT2, EFHB, RRN3P2, TRIM49B, TUBB8P5, EPHA7 , ERBB3, ERCC6, FAM21A, FAM21C, FCGBP, FGFR2, FLG2, FLT1, FOLR2, FRYL, FSCB, GAB1, GABRA4, GABRP, GH2, GOLGA6L1, GPHB5, GPR32, GPX5, GTF3C3, HECW1, HIST1H3B, HLA-A, HRAS , HS3ST1, HS6ST1, HSPD1, IDH1, JAK2, KDM5B, KIAA0528, KRT15, KRT38, KRTAP21-1, KRTAP4-5, KRTAP4-7, KRTAP5-4, KRTAP5-5, LAMA4, LATS1, LMF1, LPAR4, LPPR4, LRRFIP1 Lum, Lyst, MAP2K1, March1, Marco, MB21D2, MEGF10, MMP16, Morc1, MRE11AA, MTMR3, MUC12, MUC17, MUC2, MUC20, NBPFF10, NBPF20, NEK1, NFE2L2, NLRP4, NOTCH2, NR K, nup93, OBSCN, OR11H1 , OR2B11, OR2M4, OR4Q3, OR5D13, OR8I2, OXSM, PIK3R1, PPP2R5C, PRAME, PRF1, PRG4, PRPF19, PTH2, PTPRC, PTPRJ, RAC1, RAD50, RBM12, RGPD3, RGS22, ROR1, RP11-671M22.1, RP13 -996F3.4, RP1L1, RSBN1L, RYR3, SAMD3, SCN3A, SEC31A, SF1, SF3B1, SLC25A2, SLC44A1, SLC4A11, SMAD2, SPTA1, ST6GAL2, STK11, SZT2, TAF1L, TAX1BP1, TBP, TGFBI, TIF1, TMEM14B TMEM74 TPTE, TRAPPC8, TRPS1, TXNDC6, USP32, UTP20, VASN, VPS72, Wash3P, WWTR1, XPO1, ZFHX4, ZMIZ1, ZNF167, ZNF436, ZNF492, ZNF598, ZRSR2, AB1, AB1, AK T2, AKT3, ARAF, ArFRP1, Arid2, ASXL1 , ATR, ATRX, AURKA, AURTHUS. P.N., AXL, BAP1, BARD1, BCL2, BCL2L2, BCL6, BCOR, BCORL1, BLM, BRIP1, BTK, CARD11, CBFB, CBL, CCND1, CCND2, CCND3, CCNE1, CD79A, CD79B, CDC73, CDK12, CDK4 , CDK6, CDK8, CDKN1B, CDKN2B, CDKN2C, CEBPA, CHEK1, CIC, CRKL, CRLF2, CSF1R, CTCF, CTNNA1, DAXX, DDR2, DOT1L, EMSY (C11orf30), EP300, EPHA3, EPHA5, EPHB1, ERBB4, ERG, ESR1, EZH2, FAM123B (WTX), FAM46C, FANCA, FANCC, FANCD2, FANCE, FANCF, FANCG, FANCL, FGF10, FGF14, FGF19, FGF23, FGF3, FGF4, FGF6, FGFR1, FGFR2, FGFR3, FGFR4, FLT3, FLT4 , FOXL2, GATA1, GATA2, GATA3, GID4 (C17orf39), GNA11, GNA13, GNAQ, GNAS, GPR124, GSK3B, HGF, IDH1, IDH2, IGF1R, IKBKE, IKZF1, IL7R, INHBA, IRF4, IRS2, JAK1, JAK3, JUN, KAT6A (MYST3), KDM5A, KDM5C, KDM6A, KDR, KEAP1, KLHL6, MAP2K2, MAP2K4, MAP3K1, MCL1, MDM2, MDM4, MED12, MEF2B, MEN1, MET, MITF, MLH1, MLL, MLL2, MPL, MSH2 , MSH6, MTOR, MUTYH, MYC, MYCL1, MYCN, MYD88, NF1, NFKBIA, NKX2-1, NOTCH1, NPM1, NRAS, NTRK1, NTRK2, NTRK3, PAK3, PALB2, PAX5, PBRM1, PDGFRA, PDGFRB, PDK1, PIK3CG , PIK3R2, PPP2R1A, PRDM1, PRKAR1A, PRKDC, PTCH1, PTPN11, RAD51, RAF1, RARA, RET, RICTOR, RNF43, RPTOR, RUNX1, SMARCA4, SMARCB1, SMO, SOCS1, SOX10, SOX2, SPEN, SPOP, SRC, STAT4 , SUFU, TET2, TGFBR2, TNFAIP3, TNFRSF14, TOP1, TP53, TSC1, TSC2, TSHR, VHL, WISP3, WT1, ZNF217, ZNF703 and combinations thereof (Su et al ., J Mol Diagn 2011, 13:74-84; DOI:10.1016/j.jmoldx.2010.11.010; and Abaan et al ., “The Exomes of the NCI-60 Panel: A Genomic Resource for Cancer Biology and Systems Pharmacology,” Cancer Research, July 15, 2013, each of which is incorporated herein by reference in its entirety). In some embodiments, the duplication is a duplication of chromosome 1p (“Chr1p”), which is associated with breast cancer. In some embodiments, one or more polymorphisms or mutations are found in BRAF, for example, the V600E mutation. In some embodiments, one or more polymorphisms or mutations are in K-ras. In some embodiments, there is a combination of one or more polymorphisms or mutations in K-ras and APC. In some embodiments, there is a combination of one or more polymorphisms or mutations in K-ras and p53. In some embodiments, there is a combination of one or more polymorphisms or mutations in APC and p53. In some embodiments, there is a combination of one or more polymorphisms or mutations in K-ras, APC, and p53. In some embodiments, there is a combination of one or more polymorphisms or mutations in K-ras and EGFR. Exemplary polymorphisms or mutations are found in one or more of the following microRNAs: miR-15a, miR-16-1, miR-23a, miR-23b, miR-24-1, miR-24-2, miR-27a, miR-27b , miR-29b-2, miR-29c, miR-146, miR-155, miR-221, miR-222, and miR-223 (Calin et al. “A miRNA signature associated with prognosis and progression in chronic lymphocytic leukemia. N Engl J Med 353:1793-801, 2005, incorporated herein by reference in its entirety).

[733] In some embodiments, the deletion is a deletion of at least 0.01 kb, 0.1 kb, 1 kb, 10 kb, 100 k. bp, 1 MB, 2 MB, 3 MB, 5 MB, 10 MB, 15 MB, 20 MB, 30 MB or 40 MB. In some embodiments, the deletion is a deletion of 1 kb. up to 40 MB, such as from 1 kb. up to 100 kb, from 100 kb up to 1 MB, from 1 to 5 MB, from 5 to 10 MB, from 10 to 15 MB, from 15 to 20 MB, from 20 to 25 MB, from 25 to 30 MB or from 30 to 40 MB inclusive.

[734] In some embodiments, the duplication is a duplication of at least 0.01 kb, 0.1 kb, 1 kb, 10 kb, 100 tbp, 1 MB, 2 MB, 3 MB, 5 MB, 10 MB, 15 MB, 20 MB, 30 MB or 40 MB. In some embodiments, the duplication is a duplication of 1 kb or greater. up to 40 MB, such as from 1 kb. up to 100 kb, from 100 kb up to 1 MB, from 1 to 5 MB, from 5 to 10 MB, from 10 to 15 MB, from 15 to 20 MB, from 20 to 25 MB, from 25 to 30 MB or from 30 to 40 MB inclusive.

[735] In some embodiments, the tandem repeat is a repeat of 2 to 60 nucleotides, such as 2 to 6, 7 to 10, 10 to 20, 20 to 30, 30 to 40, 40 to 50 or from 50 to 60 nucleotides inclusive. In some embodiments, the tandem repeat is a 2-nucleotide repeat (dinucleotide repeat). In some embodiments, the tandem repeat is a repeat of 3 nucleotides (trinucleotide repeat).

[736] In some embodiments, the polymorphism or mutation is prognostic. Typical prognostic mutations include K-ras mutations, such as K-ras mutations, which are indicators of postoperative disease recurrence in colorectal cancer (Ryan et al." A prospective study of circulating mutant KRAS2 in the serum of patients with colorectal neoplasia: strong prognostic indicator in postoperative follow up," Gut 52:101-108, 2003; and Lecomte T et al. Detection of free-circulating tumor-associated DNA in plasma of colorectal cancer patients and its association with prognosis," Int J Cancer 100:542- 548, 2002, each of which is incorporated herein by reference in its entirety).

[737] In some embodiments, the polymorphism or mutation is associated with an altered response to a particular treatment (eg, increased or decreased effectiveness or side effects). Examples include K-ras mutations associated with decreased response to EGFR-based treatment in non-small cell lung cancer (Wang et al. “ Potential clinical significance of a plasma-based KRAS mutation analysis in patients with advanced non-small cell lung cancer,” Clin Canc Res16:1324-1330, 2010, incorporated herein by reference in its entirety).

[738] K-ras is an oncogene that is activated in many cancers. Typical K-ras mutations are mutations at codons 12, 13, and 61. K-ras scDNA mutations have been identified in pancreatic, lung, colorectal, bladder, and gastric cancers (Fleischhacker & Schmidt, “Circulating nucleic acids (CNAs) and caner - a survey,” Biochim Biophys Acta 1775:181-232, 2007, incorporated herein by reference in its entirety).

[739] p53 is a tumor suppressor that is mutated in many forms of cancer and promotes tumor progression (Levine & Oren, “The first 30 years of p53: growing ever more complex. Nature Rev Cancer,” 9:749–758, 2009, paper incorporated herein by reference in its entirety). Many different codons can mutate, such as Ser249. p53 cDNA mutations have been identified in breast, lung, ovarian, bladder, gastric, pancreatic, colorectal, intestinal, and hepatocellular cancers (Fleischhacker & Schmidt, “Circulating nucleic acids (CNAs) and caner - a survey,” Biochim Biophys Acta 1775:181-232, 2007, incorporated herein by reference in its entirety).

[740] BRAF is an oncogene located downstream of Ras. BRAF mutations have been identified in glial neoplasms, melanoma, thyroid and lung cancers (Dias-Santagata et al. BRAF V600E mutations are common in pleomorphic xanthoastrocytoma: diagnostic and therapeutic implications. PLOS ONE 2011;6:e17948, 2011; Shinozaki et al. Utility of circulating B-RAF DNA mutation in serum for monitoring melanoma patients receiving biochemotherapy. Clin Canc Res 13:2068-2074, 2007 and Board et al. Detection of BRAF mutations in the tumor and serum of patients enrolled in the AZD6244 (ARRY- 142886) advanced melanoma phase II study. Brit J Canc2009;101:1724-1730, each of these documents is incorporated herein by reference in its entirety). The BRAF V600E mutation occurs, for example, in melanoma tumors and is more common in advanced stages. The V600E mutation was found in cDNA.

[741] EGFR promotes cell proliferation and is misregulated in many forms of cancer (Downward J. Targeting RAS signaling pathways in cancer therapy. Nature Rev Cancer 3:11–22, 2003 and Levine & Oren “The first 30 years of p53: growing ever more complex. Nature Rev Cancer,” 9:749–758, 2009, incorporated herein by reference in its entirety). Exemplary EGFR mutations include mutations in exons 18–21, which have been identified in patients with lung cancer. EGFR mutations in cDNA have been identified in patients with lung cancer (Jia et al. “ Prediction of epidermal growth factor receptor mutations in the plasma/pleural effusion to the efficacy of gefitinib treatment in advanced non-small cell lung cancer,” J Canc Res Clin Oncol 2010; 136:1341–1347, 2010, incorporated herein by reference in its entirety).

[742] Examples of polymorphisms or mutations associated with breast cancer include LOH on microsatellites (Kohler et al. “ Levels of plasma circulating cell free nuclear and mitochondrial DNA as potential biomarkers for breast tumors,” Mol Cancer 8:doi:10.1186/ 1476-4598-8-105, 2009, incorporated herein by reference in its entirety), p53 mutations (such as mutations in exons 5-8) (Garcia et al.” Extracellular tumor DNA in plasma and overall survival in breast cancer patients, “Genes, Chromosomes & Cancer 45:692-701, 2006, incorporated herein by reference in its entirety), HER2 (Sorensen et al. “ Circulating HER2 DNA after trastuzumab treatment predicts survival and response in breast cancer,” Anticancer Res30:2463-2468 , 2010, document incorporated herein by reference in its entirety), PIK3CA, MED1, and GAS6 polymorphisms or mutations (Murtaza et al. “ Non-invasive analysis of acquired resistance to cancer therapy by sequencing of plasma DNA,” Nature 2013;doi:10.1038/nature12065 , 2013, document incorporated herein by reference in its entirety).

[743] Increased levels of cDNA and LOH are associated with decreased overall survival and disease-free survival. Mutations in p53 (exons 5–8) are associated with decreased overall survival. Reduced levels of circulating HER2 cDNA are associated with better response to HER2-targeted treatments in subjects with HER2-positive breast tumors. Activating mutation in PIK3CA, truncation of MED1, and splicing mutation in GAS6 lead to treatment resistance.

[744] Exemplary polymorphisms or mutations associated with colorectal cancer include mutations in p53, APC, K-ras, and thymidylate synthase, as well as methylation of the p16 gene. (Wang et al. “ Molecular detection of APC, K-ras, and p53 mutations in the serum of colorectal cancer patients as circulating biomarkers,” World J Surg 28:721-726, 2004; Ryan et al. “ A prospective study of circulating mutant KRAS2 in the serum of patients with colorectal neoplasia: strong prognostic indicator in postoperative follow up,” Gut 52:101-108, 2003; Lecomte et al. “ Detection of free-circulating tumor-associated DNA in plasma of colorectal cancer patients and its association with prognosis,” Int J Cancer 100:542-548, 2002; Schwarzenbach et al. “ Molecular analysis of the polymorphisms of thymidylate synthase on cell-free circulating DNA in the blood of patients with advanced colorectal carcinoma,” Int J Cancer 127:881-888, 2009, each of these documents is incorporated herein by reference in its entirety). Postoperative detection of K-ras mutations in serum is a reliable predictor of disease relapse. Detection of K-ras mutations and p16 gene methylation is associated with decreased survival and increased disease relapse. Detection of K-ras, APC and/or p53 mutations is associated with relapse and/or metastasis. Polymorphisms (including LOH, SNP, variable number tandem repeats, and deletion) in the thymidylate synthase gene (a target of fluoropyrimidine-based chemotherapy) using scDNA may be associated with treatment response.

[745] Exemplary polymorphisms or mutations associated with lung cancer (such as non-small cell lung cancer) include K-ras (such as mutations in codon 12) and EGFR mutations. Examples of prognostic mutations include EGFR mutations (exon 19 deletion or exon 21 mutation) are associated with increased overall survival and progression-free survival, and K-ras mutations (at codons 12 and 13) are associated with decreased progression-free survival (Jian et al. “ Prediction of epidermal growth factor receptor mutations in the plasma/pleural effusion to efficacy of gefitinib treatment in advanced non-small cell lung cancer,” J Canc Res Clin Oncol 136:1341-1347, 2010; Wang et al. “ Potential clinical significance of a plasma-based KRAS mutation analysis in patients with advanced non-small cell lung cancer,” Clin Canc Res 16:1324-1330, 2010, each of which is incorporated herein by reference in its entirety.) Exemplary polymorphisms or mutations indicative of treatment response include EGFR mutations (exon 19 deletion or exon 21 mutation), which improve response to treatment, and K-ras mutations (codons 12 and 13), which reduce response to treatment. A mutation conferring resistance to EFGR has been identified. (Murtaza et al. “ Non-invasive analysis of acquired resistance to cancer therapy by sequencing of plasma DNA,” Nature doi:10.1038/nature12065, 2013, the document is incorporated herein by reference in its entirety).

[746] Exemplary polymorphisms or mutations associated with melanoma (such as ulcerative melanoma) include those in GNAQ, GNA11, BRAF and p53. Exemplary GNAQ and GNA11 mutations include the R183 and Q209 mutations. Q209 mutations in GNAQ or GNA11 are associated with bone metastases. BRAF V600E mutations may be found in patients with metastatic/advanced melanoma. BRAF V600E is an indicator of invasive melanoma. The presence of the BRAF V600E mutation after chemotherapy is associated with lack of response to treatment.

[747] Exemplary polymorphisms or mutations associated with pancreatic carcinomas include those in K-ras and p53 (such as p53 Ser249). p53 Ser249 is also associated with hepatitis B infection and hepatocellular carcinoma, as well as ovarian cancer and non-Hodgkin's lymphoma.

[748] Even polymorphisms or mutations that are present at low frequency in a sample can be detected using the methods of the present invention. For example, a polymorphism or mutation that is present at a frequency of 1 in a million can be observed 10 times by performing 10 million sequencing reads. If necessary, the number of sequencing reads can be changed depending on the desired level of sensitivity. In some embodiments, the sample is reanalyzed or a different sample from the subject is analyzed using more sequencing reads to increase sensitivity. For example, if no or only a small number (eg, 1, 2, 3, 4, or 5) polymorphisms or mutations associated with cancer or increased risk of cancer are found, the sample is reanalyzed or a different sample is tested.

[749] In some embodiments, multiple polymorphisms or mutations are required for cancer or metastatic cancer. In such cases, screening for multiple polymorphisms or mutations improves the ability to accurately diagnose cancer or metastatic cancer. In some embodiments, when a subject has a subset of multiple polymorphisms or mutations required for cancer or metastatic cancer, the subject may be rescreened at a later date to see if the subject acquires additional mutations.

[750] In some embodiments in which multiple polymorphisms or mutations are required for a cancer or metastatic cancer, the frequency of each polymorphism or mutation can be compared to see if they occur at the same frequency. For example, if a cancer requires two mutations (labeled "A" and "B"), some cells will have neither of them, some cells will have A, some will have B, and some will have A and B. If A and B are observed with the same frequency, a subject is more likely to have multiple cells with A and B. If A and B are observed at different frequencies, the subject is more likely to have different populations of cells.

[751] In some embodiments in which multiple polymorphisms or mutations are required for cancer or metastatic cancer, the number or identity of such polymorphisms or mutations that are present in a subject can be used to predict the likelihood or rate of occurrence of the disease or disorder in the subject. In some embodiments, in which polymorphisms or mutations tend to occur in a particular order, the subject may be tested periodically to see if the subject has acquired other polymorphisms or mutations.

[752] In some embodiments, determining the presence or absence of multiple polymorphisms or mutations (such as 2, 3, 4, 5, 8, 10, 12, 15 or more) increases the sensitivity and/or specificity of determining the presence or absence of a disease or disorder, such as cancer, or an increased risk of a disease or disorder such as cancer.

[753] In some embodiments, the polymorphism(s) or mutation(s) are detected in a direct manner. In some embodiments, the polymorphism(s) or mutation(s) are detected indirectly by detecting one or more sequences (eg, a polymorphic locus, such as an SNP) associated with the polymorphism or mutation.

[754] Approximate nucleic acid changes

[755] In some embodiments, there is a change in RNA or DNA integrity (e.g., a change in the size of fragmented scRNA or scDNA, or a change in nucleosomal composition) that is associated with a disease or disorder, such as cancer, or an increased risk of a disease or disorder, such as cancer. In some embodiments, there is a change in RNA or DNA methylation pattern associated with a disease or disorder, such as cancer, or an increased risk of a disease or disorder, such as cancer (eg, hypermethylation of tumor suppressor genes). For example, methylation of CpG islands in the promoter region of tumor suppressor genes has been proposed to trigger local gene silencing. Aberrant methylation of the p16 tumor suppressor gene occurs in subjects with liver, lung, and breast cancer. Other frequently methylated tumor suppressor genes, including APC, Ras association domain family protein 1A (RASSF1A), glutathione S-transferase P1 (GSTP1), and DAPK, have been found in various types of cancer, such as nasopharyngeal carcinoma, colorectal cancer, lung, esophageal cancer, prostate cancer, bladder cancer, melanoma and acute leukemia. Methylation of certain tumor suppressor genes such as p16 has been described as an early event in cancer formation and is thus useful for early cancer screening.

[756] In some embodiments, bisulfite conversion or a non-bisulfite-based strategy using methylation-sensitive restriction enzyme digestion is used to determine the methylation pattern (Hung et al., J Clin Pathol 62:308-313, 2009, entire document incorporated here via link). In bisulfite conversion, methylated cytosines remain cytosines, while unmethylated cytosines are converted to uracils. Methylation-sensitive restriction enzymes (e.g., BstUI) cleave unmethylated DNA sequences at specific recognition sites (e.g., 5´-CG vs. CG-3´ for BstUI), while methylated sequences remain intact. In some embodiments, intact methylated sequences are detected. In some embodiments, stem-loop primers are used to selectively amplify unmethylated restriction enzyme-digested fragments without co-amplifying methylated, non-enzymatically digested DNA.

[757] Approximate measurements in mRNA splicing

[758] In some embodiments, a change in mRNA splicing is associated with a disease or disorder, such as cancer, or an increased risk of a disease or disorder, such as cancer. In some embodiments, the change in mRNA splicing occurs in one or more of the following nucleic acids associated with cancer or increased risk of cancer: DNMT3B, BRCA1, KLF6, Ron, or Gemin5. In some embodiments, the detected mRNA splice variant is associated with a disease or disorder, such as cancer. In some embodiments, healthy cells (such as non-cancerous cells) produce multiple mRNA splice variants, but a change in the relative amounts of mRNA splice variants is associated with a disease or disorder, such as cancer. In some embodiments, the change in mRNA splicing occurs due to a change in the mRNA sequence (e.g., a mutation at a splice site), a change in the levels of splicing factors, a change in the amount of splicing factor available (e.g., a decrease in the amount of splicing factor available due to binding of the splicing factor to a repeat ), changes in splicing regulation or tumor microenvironment.

[759] The splicing reaction is carried out by a complex of several proteins and RNA called the spliceosome (Fackenthal1 and Godley, Disease Models & Mechanisms 1: 37–42, 2008, doi:10.1242/dmm.000331, incorporated herein by reference in its entirety). The spliceosome recognizes intron-exon boundaries and removes intervening introns through two transesterification reactions that result in the ligation of two adjacent exons. The precision of this reaction must be perfect because if ligation occurs incorrectly, the normal protein coding potential may be disrupted. For example, in cases where exon skipping preserves the triplet codon reading frame that determines the identity and order of amino acids during translation, the alternatively spliced mRNA may indicate a protein that is missing important amino acid residues. Most often, exon skipping disrupts the translational reading frame, resulting in premature stop codons. These mRNAs are typically degraded by at least 90% through a process known as nonsense-mediated mRNA degradation, making it less likely that such defective messages will accumulate to form truncated protein products. If misspliced mRNAs escape this pathway, truncated, mutated, or unstable proteins are produced.

[760] Alternative splicing is a method of expressing several or many different transcripts from the same genomic DNA and results from the inclusion of a subset of available exons for a particular protein. By deleting one or more exons, certain protein domains may be lost from the encoded protein, which may result in loss or gain of protein function. Several types of alternative splicing have been described: exon skipping; alternative 5' or 3' splice sites; mutually exclusive exons; and, much less commonly, intron retention. Others have compared the amount of alternative splicing in cancer cells compared to normal cells using a bioinformatics approach and determined that cancers have lower levels of alternative splicing than normal cells. In addition, the distribution of types of alternative splicing events differed in cancer and normal cells. Cancer cells showed less exon skipping, but greater selection of alternative 5' and 3' splice sites and intron retention than normal cells. When the phenomenon of exonization (the use of sequences as exons that are preferentially used by other tissues as introns) was investigated, genes associated with exonization in cancer cells were predominantly associated with mRNA processing, indicating a direct link between cancer cells and the generation of aberrant forms mRNA splicing.

[761] Approximate changes in DNA or RNA levels

[762] In some embodiments, there is a change in the total amount or concentration of one or more types of DNA (such as scDNA, scmDNA, scDNA, cellular DNA, or mitochondrial DNA) or RNA (scRNA, cellular RNA, cytoplasmic RNA, coding cytoplasmic RNA, non-coding cytoplasmic RNA, mRNA, miRNA, mitochondrial RNA, rRNA or tRNA). In some embodiments, there is a change in the amount or concentration of one or more specific molecules of DNA (such as scDNA, scmDNA, scDNA, cellular DNA, or mitochondrial DNA) or RNA (scRNA, cellular RNA, cytoplasmic RNA, coding cytoplasmic RNA, non-coding cytoplasmic RNA, mRNA, miRNA, mitochondrial RNA, rRNA or tRNA). In some embodiments, one allele is expressed more strongly than another allele of the locus of interest. Typical miRNAs are short RNA molecules of 20–22 nucleotides that regulate gene expression. In some embodiments, there is a change in the transcriptome, such as a change in the identity or quantity of one or more RNA molecules.

[763] In some embodiments, an increase in the total amount or concentration of scDNA or scRNA is associated with a disease or disorder, such as cancer, or an increased risk of a disease or disorder, such as cancer. In some embodiments, the total concentration of a type of DNA (such as scDNA, scmDNA, scDNA, cellular DNA, or mitochondrial DNA) or RNA (scRNA, cellular RNA, cytoplasmic RNA, coding cytoplasmic RNA, non-coding cytoplasmic RNA, mRNA, miRNA, mitochondrial RNA, rRNA or tRNA) increases by at least 2, 3, 4, 5, 6, 7, 8, 9, 10 times or more compared to the total concentration of that type of DNA or RNA in healthy (eg, non-cancerous) subjects. In some embodiments, the total scDNA concentration is from 75 to 100 ng/mL, from 100 to 150 ng/mL, from 150 to 200 ng/mL, from 200 to 300 ng/mL, from 300 to 400 ng/mg, from 400 to 600 ng/ml, 600 to 800 ng/ml, 800 to 1000 ng/ml inclusive, or total cDNA concentration greater than 100 ng/ml, e.g. greater than 200, 300, 400, 500, 600, 700, 800, 900 or 1000 ng/ml indicates cancer, increased risk of cancer, increased risk of the tumor being cancerous rather than benign, decreased likelihood of the cancer going into remission, or a worse prognosis for the cancer. In some embodiments, the amount of a type of DNA (such as scDNA, scmDNA, scDNA, cellular DNA, or mitochondrial DNA) or RNA (scRNA, cellular RNA, cytoplasmic RNA, coding cytoplasmic RNA, non-coding cytoplasmic RNA, mRNA, miRNA, mitochondrial RNA, rRNA or tRNA) having one or more polymorphisms/mutations (such as deletions or duplications) associated with a disease or disorder, such as cancer, or an increased risk of a disease or disorder, such as cancer, is at least 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 14, 16, 18, 20 or 25% of the total amount of that type of DNA or RNA. In some embodiments, at least 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 14, 16, 18, 20, or 25% of the total amount of that type of DNA (such as cDNA , scmDNA, scDNA, cellular DNA or mitochondrial DNA) or RNA (scRNA, cellular RNA, cytoplasmic RNA, coding cytoplasmic RNA, non-coding cytoplasmic RNA, mRNA, miRNA, mitochondrial RNA, rRNA or tRNA) has a specific polymorphism or mutation (eg, deletion or duplication) associated with a disease or disorder such as cancer, or an increased risk of a disease or disorder such as cancer.

[764] In some embodiments, the scDNA is encapsulated. In some embodiments, the scDNA is not encapsulated.

[765] In some embodiments, the proportion of tumor DNA out of total DNA is determined (eg, the proportion of tumor cDNA out of total cDNA, or the proportion of tumor cDNA with a particular mutation out of total cDNA). In some embodiments, the tumor DNA fraction can be determined for a plurality of mutations, where the mutations can be single nucleotide variants, copy number variants, differential methylation, or combinations thereof. In some embodiments, the average tumor fraction calculated for the single mutation or set of mutations with the highest calculated tumor fraction is taken to be the actual tumor fraction in the sample. In some embodiments, the average tumor fraction calculated for all mutations is taken to be the actual tumor fraction in the sample. In some embodiments, this tumor fraction is used to determine the stage of the cancer (since higher tumor fractions may be associated with more advanced stages of cancer). In some embodiments, the tumor fraction is used to determine the size of the cancer, since larger tumors may correlate with the tumor DNA fraction in the plasma. In some embodiments, the tumor fraction is used to determine the size of the tumor fraction affected by one or multiple mutations, since there may be a correlation between the measured tumor fraction in the plasma sample and the size of the tissue with the given mutation(s). For example, the size of tissue with a given mutation(s) can be correlated with the tumor DNA fraction, which can be calculated to target those specific mutation(s).

[766] Sample Databases

[767] The invention also provides databases containing one or more results of the method of the invention. For example, the database may include records with any of the following information for one or more subjects: any identified polymorphisms/mutations (eg, ICH), any known association of polymorphisms/mutations with a disease or disorder or increased risk of a disease or disorder, impact of polymorphisms/mutations on the expression or activity level of the encoded mRNA or protein, the fraction of DNA, RNA or cells associated with a disease or disorder (for example, DNA, RNA or cells having a polymorphism/mutation associated with a disease or disorder) from the total DNA, RNA or cells in sample, the source of the sample used to identify polymorphisms/mutations (for example, a blood sample or a sample from a specific tissue), the number of cells affected, the results of a subsequent repeat of the test (for example, a repeat of the test to monitor the progression or remission of a disease or disorder), results of other tests on disease or disorder, type of disease or disorder with which the subject has been diagnosed, treatment(s) prescribed, response to such treatment(s), side effects of such treatment(s), symptoms (such as symptoms associated with the disease or disorder ), duration and number of remissions, duration of survival (eg, length of time from initial test to death or length of time from diagnosis to death), cause of death, and combinations thereof.

[768] In some embodiments, the database includes records with any of the following information for one or more subjects: any identified polymorphisms/mutations, any known association of polymorphisms/mutations with cancer or increased risk of cancer, effect of polymorphisms/mutations on expression or activity levels encoded mRNA or protein, fraction of cancer DNA, RNA or cells from the total DNA, RNA or cells in the sample, source of the sample used to identify polymorphisms/mutations (for example, a blood sample or a sample from a specific tissue), number of cancer cells, tumor size ( tumors), results of subsequent repeat test (eg, repeat test to monitor cancer progression or remission), results of other cancer tests, type of cancer that was diagnosed in the subject, treatment(s) prescribed, response to such treatment(s), side effects of such treatment(s), symptoms (eg, cancer-related symptoms), duration and number of remissions, length of survival (eg, length of time from initial test to death or length of time from cancer diagnosis before death), cause of death, and combinations thereof. In some embodiments, the response to treatment includes any of the following: reducing or stabilizing the size of a tumor (e.g., benign or cancerous tumor), slowing or preventing the increase in tumor size, reducing or stabilizing the number of tumor cells, increasing the time of disease-free survival between the disappearance of a tumor and its recurrence occurrence, prevention of initial or subsequent occurrence of a tumor, reduction or stabilization of an adverse symptom associated with a tumor, or a combination thereof. In some embodiments, results of one or more other tests for a disease or disorder, such as cancer, are included, such as results of screening tests, medical imaging, or microscopic examination of a tissue sample.

[769] In one such aspect, the invention provides an electronic database comprising at least 5, 10, 10 ² , 10 ³ , 10 ⁴ , 10 ⁵ , 10 ⁶ , 10 ⁷ , 10 ⁸ or more entries. In some embodiments, the database contains records for at least 5, 10, 10 ² , 10 ³ , 10 ⁴ , 10 ⁵ , 10 ⁶ , 10 ⁷ , 10 ⁸ or more different entities.

[770] In another aspect, the invention relates to a computer including a database of the invention and a user interface. In some embodiments, the user interface may display some or all of the information contained in one or more records. In some embodiments, the user interface is capable of displaying (i) one or more cancer types that have been identified as containing a polymorphism or mutation that is stored in the computer, (ii) one or more polymorphisms or mutations that have been identified in a particular type of cancer, a record of which is stored in a computer, (iii) information about the prognosis for a particular type of cancer or a particular polymorphism or mutation, a record of which is stored in a computer, (iv) one or more compounds or other treatments beneficial for cancer with a polymorphism or mutation of which a computer-recorded record is stored, (v) one or more compounds that modulate the expression or activity of an mRNA or protein of which a computer-recorded record is stored, and (vi) one or more mRNA molecules or proteins of which the expression or activity is stored modulated by a connection, a record of which is stored in the computer. The internal components of a computer typically include a processor connected to memory. External components typically include a storage device such as a hard drive; user input devices such as keyboard and mouse; display, such as a monitor; and, optionally, a network link capable of connecting the computer system to other computers to enable sharing of data and processing tasks. During operation, programs can be loaded into the memory of this system.

[771] In another aspect, the invention is a computer-implemented process that includes one or more steps of any of the methods of the invention.

[772] Sample risk factors

[773] In some embodiments, the subject is also assessed for one or more risk factors for a disease or disorder, such as cancer. Examples of risk factors include family history of disease or disorder, lifestyle (eg, smoking and exposure to carcinogens), and levels of one or more hormones or serum proteins (eg, alpha-fetoprotein (AFP) for liver cancer, carcinoembryonic antigen (CEA) for colorectal cancer or prostate-specific antigen (PSA) for prostate cancer). In some embodiments, the size and/or number of tumors are measured and used to determine a subject's prognosis or select treatment for the subject.

[774] Example Screening Methods

[775] If necessary, the presence or absence of a disease or disorder, such as cancer, can be confirmed, or the disease or disorder, such as cancer, can be classified using any standard method. For example, a disease or disorder such as cancer can be detected in several ways, including the presence of certain signs and symptoms, a tumor biopsy, screening tests, or medical imaging (such as mammography or ultrasound). Once a possible cancer is found, it can be diagnosed by microscopic examination of a tissue sample. In some embodiments, a subject who has been diagnosed is retested using a method of the invention or known testing for a disease or disorder at multiple time points to monitor the progression of the disease or disorder, or remission or recurrence of the disease or disorder.

[776] Example types of cancer

[777] Exemplary cancers that can be diagnosed, predicted, stabilized, treated, prevented, and for which response to treatment can be predicted or monitored using any of the methods of the invention include solid tumors, carcinomas, sarcomas, lymphomas, leukemias, germ cell tumors or blastoma. In various embodiments, the cancer is acute lymphoblastic leukemia, acute myeloid leukemia, adrenocortical carcinoma, AIDS-related cancer, AIDS-related lymphoma, anal cancer, appendiceal cancer, astrocytoma (eg, cerebellar or cerebral astrocytoma in children), basal cell carcinoma, bile duct cancer (eg, extrahepatic bile duct cancer), bladder cancer, bone tumor (eg, osteosarcoma or malignant fibrous histiocytoma), brainstem glioma, brain cancer (eg, cerebellar astrocytoma, cerebral astrocytoma/malignant glioma , ependymoma, medulloblastoma, supratentorial primitive neuroectodermal tumors, or optic pathway/hypothalamic glioma), glioblastoma, breast cancer, bronchial adenoma or carcinoid, Burkitt's lymphoma, carcinoid tumor (eg, childhood carcinoid tumor or gastrointestinal carcinoid tumor), carcinoma, central nervous system lymphoma, cerebellar astrocytoma or malignant glioma (eg, pediatric cerebellar astrocytoma or malignant glioma), cervical cancer, pediatric cancer, chronic lymphocytic leukemia, chronic myelogenous leukemia, chronic myeloproliferative disorders, colon cancer, cutaneous T- cell lymphoma, desmoplastic small round cell tumor, endometrial cancer, ependymoma, esophageal cancer, Ewing's sarcoma, tumor in the Ewing tumor family, extracranial germ cell tumor (eg, extracranial germ cell tumor of childhood), extragonadal germ cell tumor, eye cancer (eg, intraocular melanoma or retinoblastoma ocular cancer), gallbladder cancer, gastric cancer, gastrointestinal carcinoid tumor, gastrointestinal stromal tumor, germ cell tumor (eg, extracranial, extragonadal, or ovarian germ cell tumor), gestational trophoblastic tumor, glioma (eg, truncal glioma brain, childhood cerebral astrocytoma or childhood optic pathway glioma/hypothalamic glioma), gastric carcinoid, hairy cell leukemia, head and neck cancer, heart cancer, hepatocellular (liver) cancer, Hodgkin's lymphoma, hypopharyngeal cancer, hypothalamic and optic pathway glioma (eg, pediatric optic pathway glioma), islet cell carcinoma (eg, endocrine carcinoma or pancreatic islet cell carcinoma), Kaposi's sarcoma, kidney cancer, laryngeal cancer, leukemia (eg, acute lymphoblastic, acute myeloid, chronic lymphocytic, chronic myelogenous or hairy cell leukemia), cancer of the lip or mouth, liposarcoma, liver cancer (eg, non-small cell or small cell cancer), lung cancer, lymphoma (eg, AIDS-related, Burkitt's, cutaneous T-cell lymphoma, Hodgkin's lymphoma, non-Hodgkin's lymphoma, or central nervous system lymphoma), macroglobulinemia (eg, Waldenström's macroglobulinemia, malignant fibrous histiocytoma of bone or osteosarcoma, medulloblastoma (eg, childhood medulloblastoma), melanoma, Merkel cell carcinoma, mesothelioma (eg, adult or pediatric mesothelioma), metastatic squamous cell carcinoma of the neck of unknown origin, oral cancer , multiple endocrine neoplasia syndrome (eg, multiple endocrine neoplasia syndrome of childhood), multiple myeloma or plasma cell neoplasm, mycosis fungoides, myelodysplastic syndrome, myelodysplastic or myeloproliferative disease, myelogenous leukemia (such as chronic myelogenous leukemia), myeloid leukemia (eg, acute myeloid leukemia in adults or acute myeloid leukemia in children), a myeloproliferative disorder (such as chronic myeloproliferative disorder), cancer of the nasal cavity or paranasal sinuses, cancer of the nasopharynx, neuroblastoma, cancer of the oral cavity, cancer of the oropharynx, osteosarcoma or malignant fibrous bone histiocytoma, ovarian cancer, epithelial ovarian cancer, ovarian germ cell tumor, low-grade ovarian tumor, pancreatic cancer (such as pancreatic islet cell cancer), cancer of the paranasal sinuses or nasal cavity, parathyroid cancer, penile cancer, cancer pharynx, pheochromocytoma, pineal astrocytoma, pineal germinoma, pineoblastoma or supratentorial primitive neuroectodermal tumor (eg, infantile pineoblastoma or supratentorial primitive neuroectodermal tumor), pituitary adenoma, plasma cell neoplasia, pleuropulmonary blastoma, primary central nervous system lymphoma, cancer, rectal cancer colon, renal cell carcinoma or cancer of the renal pelvis (such as renal pelvis cancer or transitional cell carcinoma of the ureter), retinoblastoma, rhabdomyosarcoma (eg, childhood rhabdomyosarcoma), salivary gland cancer, sarcoma (eg, sarcoma of the Ewing tumor family, Kaposi, soft tissue, or uterine sarcoma), Sézary syndrome, skin cancer (eg, nonmelanoma, melanoma, or Mergkel cell skin cancer), small bowel cancer, squamous cell carcinoma, supratentorial primitive neuroectodermal tumor (eg, childhood supratentorial primitive neuroectodermal tumor), T-cell lymphoma (eg, cutaneous T-cell lymphoma), testicular cancer, throat cancer, thymoma (eg, childhood thymoma), thymoma or carcinoma of the thymus, thyroid cancer (eg, childhood thyroid cancer), trophoblastic tumor (eg, gestational trophoblastic tumor), carcinoma of unknown primary site (eg, carcinoma of unknown primary site in adults or children), urethral cancer (eg, endometrial cancer of the uterus), uterine sarcoma, vaginal cancer, optic pathway glioma, or hypothalamic glioma (eg, glioma optic pathway/hypothalamus in children), vulvar cancer, Waldenström's macroglobulinemia, or Wilms' tumor (eg, Wilms' tumor in children). In various embodiments, the cancer has metastasized or not metastasized.

[778] The cancer may or may not be hormone-related or hormone-dependent cancer (eg, estrogen or androgen cancer). Benign tumors or malignant tumors can be diagnosed, prognosis, stabilized, treated or prevented using the methods and/or compositions of the present invention.

[779] In some embodiments, the subject has a cancer syndrome. Cancer syndrome is a genetic disease in which genetic mutations in one or more genes predispose affected individuals to developing cancer and can also cause early onset of these cancers. Cancer syndromes often show not only a high lifetime risk of developing cancer, but also the development of multiple independent primary tumors. Many of these syndromes are caused by mutations in tumor suppressor genes, genes that are involved in protecting cells from becoming cancerous. Other genes that may be affected are DNA repair genes, oncogenes, and genes involved in the formation of blood vessels (angiogenesis). Common examples of hereditary cancer syndromes are hereditary breast and ovarian cancer syndrome and hereditary nonpolyposis colon cancer (Lynch syndrome).

[780] In some embodiments, a subject with one or more polymorphisms or mutations in K-ras, p53, BRA, EGFR, or HER2 is administered a treatment targeting, respectively, K-ras, p53, BRA, EGFR, or HER2.

[781] The methods of the present invention can generally be used to treat malignant or benign tumors of any type of cell, tissue or organ.

[782] Sample treatments

[783] If necessary, a subject may be prescribed any treatment to stabilize, treat, or prevent a disease or disorder, such as cancer, or an increased risk of a disease or disorder, such as cancer (for example, a subject who is diagnosed with cancer or an increased risk of cancer using any of the methods of the invention). In various embodiments, the treatment is a known treatment or combination of treatments for a disease or disorder, such as cancer, including, but not limited to, cytotoxic agents, targeted therapy, immunotherapy, hormonal therapy, radiation therapy, surgical removal of cancer cells or cells that may become malignant, stem cell transplantation, bone marrow transplantation, photodynamic therapy, palliative treatment, or a combination thereof. In some embodiments, a treatment (such as a prophylactic medication) is used to prevent, delay, or reduce the severity of a disease or disorder, such as cancer, in a subject at increased risk of the disease or disorder, such as cancer. In some embodiments, treatment is surgery, first-line chemotherapy, adjuvant therapy, or neoadjuvant therapy.

[784] In some embodiments, targeted therapy is a treatment that targets specific cancer genes, proteins, or tissue environments that promote cancer growth and survival. This type of treatment blocks the growth and spread of cancer cells while limiting damage to normal cells, which usually results in fewer side effects than other cancer drugs.

[785] One of the most successful approaches is to target angiogenesis, the growth of new blood vessels around the tumor. Targeted therapies such as bevacizumab (Avastin), lenalidomide (Revlimid), sorafenib (Nexavar), sunitinib (Sutent), and thalidomide (Thalomid) interfere with angiogenesis. Another example is the use of HER2-targeted treatments, such as trastuzumab or lapatinib, for cancers that overexpress HER2 (such as some breast cancers). In some embodiments, a monoclonal antibody is used to block a specific target on the outside of cancer cells. Examples include alemtuzumab (Campath-1H), bevacizumab, cetuximab (Erbitux), panitumumab (Vectibix), pertuzumab (Omnitarg), rituximab (Rituxan), and trastuzumab. In some embodiments, the monoclonal antibody tositumomab (Bexxar) is used to deliver radiation to the tumor. In some embodiments, the oral small molecule inhibits the cancer process within a cancer cell. Examples include dasatinib (Sprycel), erlotinib (Tarceva), gefitinib (Iressa), imatinib (Gleevec), lapatinib (Tykerb), nilotinib (Tasigna), sorafenib, sunitinib, and temsirolimus (Torisel). In some embodiments, a proteasome inhibitor (eg, the multiple myeloma drug bortezomib (Velcade)) interferes with specialized proteins called enzymes that break down other proteins in the cell.

[786] In some embodiments, immunotherapy is provided to stimulate the body's natural defenses to fight cancer. Exemplary types of immunotherapy use materials produced in the body or in the laboratory to support, target, or restore immune system function.

[787] In some embodiments, hormone therapy treats cancer by reducing the amount of hormones in the body. Some cancers, including some breast and prostate cancers, grow and spread only in the presence of natural chemicals in the body called hormones. In various embodiments, hormone therapy is used to treat prostate, breast, thyroid, and reproductive cancers.

[788] In some embodiments, the treatment includes a stem cell transplant in which the diseased bone marrow is replaced with highly specialized cells called hematopoietic stem cells. Hematopoietic stem cells are found both in the bloodstream and in the bone marrow.

[789] In some embodiments, treatment includes photodynamic therapy, which uses special drugs called photosensitizing agents as well as light to kill cancer cells. Medicines work after they have been activated by a certain light.

[790] In some embodiments, treatment includes surgical removal of cancer cells or cells that may become cancerous (such as a lumpectomy or mastectomy). For example, a woman with a breast cancer susceptibility gene mutation ( BRCA1 or BRCA2 gene mutation) can reduce her risk of breast and ovarian cancer by having a risk-reducing salpingo-oophorectomy (removal of the fallopian tubes and ovaries) and/or a risk-reducing bilateral mastectomy (removal of both breasts). ). Lasers, which are very powerful and precise beams of light, can be used instead of blades (scalpels) for very careful surgical work, including the treatment of some types of cancer.

[791] In addition to treatment to slow, stop, or eliminate cancer (also called disease-directed treatment), an important part of cancer treatment is relieving the subject's symptoms and side effects, such as pain and nausea. It involves supporting the subject with physical, emotional and social needs, an approach called palliative or supportive care. People often receive therapy to target the disease and treatment to relieve symptoms at the same time.

[792] Examples of treatments include Actinomycin D, Adcetris, Adriamycin, Aldesleukin, Alemtuzumab, Alimta, Amsidin, Amsacrine, Anastrozole, Aredia, Arimidex, Aromasin, Asparaginase, Avastin, Bevacizumab, Bicalutamide, Bleomycin, Bondronate, Bonefos, Bortezomib, Busilvex, Busulf en , campto, capecitabine, carboplatin, carmustine, casodex, cetuximab, chymax, chlorambucil, cimetidine, cisplatin, cladribine, clodronate, clofarabine, crisantaspase, cyclophosphamide, cyproterone acetate, cyprostat, cytarabine, cytoxan, dacarbozine, dactinomycin, dasatinib, yes unorubicin, dexamethasone, diethylstilbestrol, docetaxel, doxorubicin, drogenil, emcyt, epirubicin, epozine, erbitux, erlotinib, estracit, estramustine, etopofos, etoposide, evoltra, exemestane, fareston, femaru, filgrastim, fludaru, fludarabine, fluorouracil, flutamide, gefinitib, gemcitabine, gem zar, Gleevec, Gleevec, gonapeptil depot, goserelin, Halaven, Herceptin, hycamptin, hydroxycarbamide, ibandronic acid, ibritumomab, idarubicin, ifsfomide, interferon, imatinib mesylate, iressu, irinotecan, jevtan, Lanvis, lapatinib, letrozole, leukeran, leuprorelin, lestat , lomustine , Mabcampate, Mabthera, Megas, Megestrol, Methotrexate, Mitosanthrone, Mitomycin, Mutulan, Mileran, Navelbine, Neulastu, Neupogen, Nexavar, Nipent, Nolvadex D, Novantrone, Oncovin, Paclitaxel, Pamidronate, PCV, Pemetrexed, Pentostatin, Perjeta, Procarbazine, provenge, prednisolone, prostrup, raltitrexed, rituximab, sprycel, sorafenib, saltamox, streptozocin, stilbestrol, stimuvax, sunitinib, sutent, tabloid, tagamet, tamofen, tamoxifen, tarceva, taxol, taxotere, tegafur with uracil, temodal, temozolomide, thalidomide, thioplex, thiotepa, thioguanine, tomudex, topotecan, toremifene, trastuzumab, tretinoin, treosulfan, triethylenethioforamide, triptorelin, tyverb, uftoral, velcad, vepezid, vesanoid, vincristine, vinorelbine, xalkori, xeloda, ervoy, zaktimu, zanosar, zavedos, zevelin , zoladex, zoledronate, zoledronic acid, zometa and zitiga.

[793] In some embodiments, the cancer is breast cancer, and the treatment or compound administered to the individual is one or more of the following: Abemaciclib, Abraxane (a nanoparticle formulation stabilized by Paclitaxel Albumin), Ado-Trastuzumab Emtansine, Afinitor (Everolimus ), Anastrozole, Aredia (Pamidronate Disodium), Arimidex (Anastrozole), Aromasin (Exemestane), Capecitabine, Cyclophosphamide, Docetaxel, Doxorubicin Hydrochloride, Ellets (Epirubicin Hydrochloride), Epirubicin Hydrochloride, Eribulin Mesylate, Everolimus, Exemestane, 5-FU (f torouracil for injection), Fareston (Toremifene), Faslodex (Fulvestrant), Femara (Letrozole), Fluorouracil for injection, Fulvestrant, Gemcitabine Hydrochloride, Gemzar (Gemcitabine Hydrochloride), Goserelin Acetate, Halaven (Eribulina Mesylate), Herceptin (Trastuzumab), Ibrance ( Palbociclib), Ixabepilone, Ixempra (Ixabepilone), Kadcyla (Ado-Trastuzumab Emtansine), Kisqali (Ribociclib), Lapatinib Ditosylate, Letrozole, Lynparza (Olaparib), Megestrol acetate, Methotrexate, Neratinib Maleate, Nerlinx (Neratinib Maleate), Olaparib, Paclitaxel , Composition of nanoparticles stabilized by Paclitaxel Albumin, Palbociclib, Pamidronate disodium, Perjeta (Pertuzumab), Pertuzumab, Ribociclib, Tamoxifen Citrate, Taxol (Paclitaxel), Taxotere (Docetaxel), Thiotepa, Toremifene, Trastuzumab, Trexall (Methotrexilate), Tickerb (Lapatinib Ditosylate ), Verzenio (Abemaciclib), Vinblastine sulfate, Xeloda (Capecitabine), Zoladex (Goserelin acetate), Evista (Raloxifene hydrochloride), Raloxifene hydrochloride, Tamoxifene citrate. In some embodiments, the cancer is breast cancer, and the treatment or compound administered to the individual is a combination selected from the following: Doxorubicin hydrochloride (Adriamycin) and Cyclophosphamide; Doxorubicin hydrochloride (Adriamycin), Cyclophosphamide and Paclitaxel (Taxol); Doxorubicin hydrochloride (Adriamycin), Cyclophosphamide and Fluorouracil; Methotrexate, Cyclophosphamide and Fluorouracil; Epirubicin hydrochloride, Cyclophosphamide and Fluorouracil; and Doxorubicin hydrochloride (Adriamycin), Cyclophosphamide and Docetaxel (Taxotere).

[794] For subjects that express both a mutant form (eg, a form associated with cancer) and a wild-type form (eg, a form not associated with cancer) of an mRNA or protein, therapy preferably inhibits the expression or activity of the mutant form by at least at least 2, 5, 10, or 20 times more potent than it inhibits the expression or activity of the wild-type form. The simultaneous or sequential use of multiple therapeutic agents can significantly reduce the incidence of cancer and reduce the number of treated cancers that become resistant to therapy. In addition, therapeutic agents that are used as part of a combination therapy may require a lower dose to treat cancer than the corresponding dose required when the therapeutic agents are administered individually. The low dose of each compound in combination therapy reduces the severity of the potential adverse side effects of these compounds.

[795] In some embodiments, a subject identified as having an increased risk of cancer (by the methods of the invention or any standard method) must avoid specific risk factors or make lifestyle changes to reduce any additional risk of cancer.

[796] In some embodiments, polymorphisms, mutations, risk factors, or any combination thereof are used to select a treatment regimen for a subject. In some embodiments, a higher dose or more treatments are selected for a subject with an increased risk of cancer or a worse prognosis.

[797] Other compounds for inclusion in single or combination therapies

[798] If necessary, additional compounds for stabilizing, treating, or preventing a disease or disorder, such as cancer, or an increased risk of a disease or disorder, such as cancer, can be identified from large libraries of both natural products and synthetic (or semi-synthetic) extracts or chemical libraries, in accordance with methods known in the art. Those skilled in the art of drug discovery and development will appreciate that the exact source of the extracts or compounds tested is not critical to the methods of the present invention. Accordingly, virtually any number of chemical extracts or compounds can be screened for their effect on cells of their particular cancer type or from a particular subject, or can be screened for their effect on the activity or expression of cancer-related molecules (such as cancer-associated molecules known to have altered activity or expression in a particular cancer type). When a crude extract is found to modulate the activity or expression of a cancer-associated molecule, further fractionation of the positive core extract can be performed to isolate the chemical component responsible for the observed effect using methods known in the art.

[799] Exemplary tests and animal models for testing therapies

[800] If desired, one or more treatments provided herein may be tested for their effect on a disease or disorder, such as cancer, using a cell line (such as a cell line with one or more mutations identified in a subject who has been diagnosed with cancer or an increased risk of cancer using the methods of the invention) or in an animal model of the disease or disorder, such as the SCID mouse model (Jain et al., Tumor Models In Cancer Research, ed. Teicher, Humana Press Inc., Totowa , NJ, pp. 647-671, 2001, incorporated herein by reference in its entirety). In addition, there are a variety of standard assays and animal models that can be used to determine the effectiveness of specific treatments to stabilize, treat, or prevent a disease or disorder, such as cancer, or an increased risk of a disease or disorder, such as cancer. Therapies can also be tested in standard human clinical trials.

[801] To select the preferred therapy for a particular subject, compounds can be tested for their effect on the expression or activity of one or more genes mutated in the subject. For example, the ability of a compound to modulate the expression of specific mRNA molecules or proteins can be detected using standard Northern blotting, Western blotting, or microarray analysis. In some embodiments, one or more compounds are selected that (i) inhibit the expression or activity of cancer-promoting mRNA molecules or proteins that are expressed at a higher than normal level or have a higher than normal level of activity in a subject ( for example, in a sample from a subject) or (ii) promote the expression or activity of cancer-inhibiting mRNA molecules or proteins that are expressed at a lower than normal level or have a lower than normal level of activity in the subject. A single or combination therapy that (i) modulates the largest number of mRNA molecules or proteins that have mutations associated with a subject's cancer, and (ii) modulates the smallest number of mRNA molecules or proteins that do not have mutations associated with a subject's cancer. In some embodiments, the selected single or combination therapy has high medicinal efficacy and produces few, if any, side effects.

[802] As an alternative to the subject-specific analysis presented above, DNA chips can be used to compare the expression of mRNA molecules in a particular type of early or late stage cancer (eg, breast cancer cells) with expression in normal tissue (Marrack et al ., Current Opinion in Immunology 12, 206-209, 2000; Harkin, Oncologist. 5:501-507, 2000; Pelizzari et al. , Nucleic Acids Res. 28(22):4577-4581, 2000, each of these documents are incorporated herein by reference in their entirety). Based on this analysis, single or combination therapies can be selected for subjects with this type of cancer to modulate the expression of mRNAs or proteins that are altered in expression in this type of cancer.

[803] In addition to being used to select therapy for a particular subject or group of subjects, expression profiles can be used to monitor changes in mRNA and/or protein expression that occur during treatment. For example, expression profiling can be used to determine whether the expression of cancer-related genes has returned to normal levels. Alternatively, the dose of one or more compounds in the therapy may be adjusted to increase or decrease the effect of the therapy on the expression levels of the corresponding gene(s) associated with cancer. In addition, this assay can be used to determine whether therapy affects the expression of other genes (for example, genes associated with adverse side effects). If necessary, the dose or composition of therapy may be changed to prevent or reduce unwanted side effects.

[804] Exemplary Formulations and Routes of Administration

[805] To stabilize, treat, or prevent a disease or disorder, such as cancer, or an increased risk of a disease or disorder, such as cancer, the composition may be formulated and administered using any method known to those skilled in the art (see, e.g. , US Patent Nos. 8389578 and 8389557, each of which is incorporated herein by reference in its entirety). General methods of formulation and administration are found in “Remington: The Science and Practice of Pharmacy,” 21st Edition, Ed. David Troy, 2006, Lippincott Williams & Wilkins, Philadelphia, Pa., incorporated herein by reference in its entirety. Examples of such formulations are liquids, suspensions, tablets, capsules, pills, powders, granules, gels, ointments, suppositories, injections, inhalants and aerosols. As an example, a modified or extended release oral formulation can be prepared using additional methods known in the art technique. For example, a suitable sustained release form of the active ingredient may be a matrix tablet or capsule composition. Suitable matrix forming materials include, for example, waxes (for example, carnauba, beeswax, paraffin wax, ceresin, shellac wax, fatty acids and alcohols), oils, hardened oils or fats (for example, hardened canola oil, castor oil, beef tallow, palm oil and soybean oil) and polymers (for example, hydroxypropylcellulose, polyvinylpyrrolidone, hydroxypropylmethylcellulose and polyethylene glycol). Other suitable materials for matrix tabletting are microcrystalline cellulose, powdered cellulose, hydroxypropyl cellulose, ethyl cellulose with other carriers and fillers. Tablets may also contain granules, coated powders or pellets. Tablets can also be multilayered. Optionally, the finished tablet may be coated or uncoated.

[806] Typical routes of administration of such compositions include, but are not limited to, oral, sublingual, buccal, topical, transdermal, inhalation, parenteral (eg, subcutaneous, intravenous, intramuscular, intrasternal injection or infusion techniques), rectal, vaginal, and intranasal. In preferred embodiments, therapy is administered using a sustained release device.

The compositions of the invention are formulated in such a way that the active ingredient(s) contained therein are/are bioavailable after administration of the composition. The compositions may take the form of one or more dosage units. The compositions may contain 1, 2, 3, 4 or more active ingredients and may optionally contain 1, 2, 3, 4 or more inactive ingredients.

[807] Alternative Embodiments

[808] Any of the methods presented herein may involve outputting data in a physical format, such as on a computer screen or on a paper printout. Any of the methods of the invention can be combined to output actionable data in a format that can be used by a physician. Some of the embodiments presented herein for determining genetic data related to a target individual may be combined with notifying a healthcare professional of a potential chromosomal abnormality (eg, deletion or duplication) or lack thereof. Some of the embodiments presented herein may be combined with outputting actionable data and making a clinical decision that results in clinical treatment, or making a clinical decision to take no action.

[809] In some embodiments, this document discloses a method for generating a report disclosing the result of any method of the invention (eg, the presence or absence of a deletion or duplication). As a result of the method of the invention, a report may be generated and it may be sent to the physician electronically, displayed on an output device (e.g., a digital report), or may be delivered to the physician as a written report (e.g., a hard copy report ). In addition, the presented methods may be combined with the actual execution of a clinical decision that results in clinical treatment, or the execution of a clinical decision to take no action.

[810] In some embodiments, the present invention provides reagents, kits and methods, as well as computer systems and computer environments with encoded instructions for performing such methods to detect both ICH and CNV in a single sample using the multiplex PCR methods provided herein . In some preferred embodiments, the sample is a single cell sample or a plasma sample suspected of containing circulating tumor DNA. These embodiments take advantage of the discovery that by testing DNA samples from single cells or plasma for the presence of ICH and NNV using the highly sensitive multiplex PCR methods disclosed herein, improved cancer detection can be achieved compared to testing for ICH or NNV alone. , especially for cancers exhibiting ICH, such as breast, ovarian, and lung cancers. Methods in some illustrative embodiments examine 50 to 100,000, or 50 to 10,000, or 50 to 1,000 SNPs for the analysis of ICH, and 50 to 1,000 SNPs, or 50 to 500 SNPs, or 50 to 250 SNPs, for SNVs. Methods provided herein for detecting ICH and/or NNV in the plasma of subjects with suspected cancer, including, for example, cancers known to exhibit ICH and NVC, such as breast, lung, and ovarian cancer, provide the advantage of detecting ICH and/or CNVs from tumors, which often consist of heterogeneous populations of cancer cells in terms of genetic composition. Thus, traditional methods that focus on analyzing only certain areas of the tumor can often miss ICHs or ONVs that are present in cells in other areas of the tumor. Plasma samples act as liquid biopsies that can be examined to detect any ICHs and/or NNVs that are present only in subpopulations of tumor cells.

[811] The following examples are presented to provide those of ordinary skill in the art with complete disclosure and description of how to use the embodiments presented herein and are not intended to limit the scope of the invention, nor are they intended to indicate that that the examples below are all or only experiments performed. Efforts have been made to ensure the accuracy of the numbers used (e.g. quantity, temperature, etc.), but some experimental errors and deviations should be taken into account. Unless otherwise noted, parts are parts by volume and temperatures are given in degrees Celsius. It should be understood that changes to the presented methods may be made without changing the fundamental aspects that these examples purport to illustrate.

EXAMPLES

[812] EXAMPLE 1: Personalized analysis of circulating tumor DNA for monitoring colorectal cancer

[813] Early detection of disease recurrence has been shown to improve survival in patients with colorectal cancer (CRC). Detection of circulating tumor DNA (ctDNA) after surgery identifies a subgroup of CRC patients at very high risk of relapse. Previous studies have performed ctDNA analysis to monitor tumor burden in early stage CRC using small gene sequencing or digital droplet PCR.

[814] The purpose of this case study was to use a personalized multiplex PCR SSP platform targeting 16 tumor-specific mutations per patient to assess minimal residual disease after surgery and monitor response to CRC treatment.

[815] The study included 130 patients with stage I–IV colorectal cancer treated with curative surgery and optionally adjuvant chemotherapy (see Table 1). Plasma samples were collected over time at baseline before surgery and at scheduled follow-up visits after surgery (FIG. 20A). Whole exome sequencing revealed somatic mutations; Following Signatera's standard workflow, patient-specific multiplex PCR tests targeting 16 somatic single nucleotide and insertion/deletion variants were analyzed by massively parallel sequencing in plasma samples collected pre- and post-operatively and during adjuvant therapy (FIG. 20B).

[816] Table 1. Patient Characteristics and Demographics (N=130)

Patients, N = 130, Age (years), median (range) = 69.9 (43.3-91), Imaging follow-up, months, median = 12.3 Gender, n (%) Histological grade of malignancy, n (%) Women 57 (43.1) Moderately differentiated 100 (76.9) Men 74 (56.9) Poorly differentiated
BUT* 20 (15.5)
10 (7.7) Localization, n (%) Adjuvant treatment, n (%) I
II 0 (0)
6 (14.3) Colon 124 (95.4) III 71 (88.8) Rectum 6 (4.6) IV 2 (100) Total 79 (60.8) Pathological stage, n (%) Relapse, n (%) I 0 (0) I 6 (4.7) II 3 (7.1) II 42 (3.2) III 19 (23.8) III 80 (62.5) IV 0 (0) IV 2 (1.6) Total 22 (16.9) Histological type, n (%) Microsatellite stability (MSS) / microsatellite instability (MSI) Adenocarcinoma 120 (92.3) M.S.S. 104 (80) Mucinous carcinoma 10 (7.7) MSI
BUT* 17 (13.1)
9 (6.9) *N/A = Not found

[817] FIG. 22 is a schematic overview of the results of ctDNA profiling of over 800 plasma samples from 128 of 130 patients. FIGS. 23A-B show the risk of relapse stratified by postoperative ctDNA status. ctDNA status was established based on the first postoperative blood sample taken at 6 weeks and before the start of chemotherapy. FIG. 23 (A) shows Kaplan-Meier analysis of disease-free survival stratified by ctDNA status. Patients without an event were assessed at the end of the follow-up period. FIG. 23 (B) shows the relapse rate according to ctDNA status (without patient assessment). Adjuvant chemotherapy was administered to 58 patients, which likely influenced the relapse rate in ctDNA-positive patients.

[818] FIG. 24(A) shows the risk of relapse after post-juvant chemotherapy, stratified by post-juvant ctDNA status. Patients were considered ctDNA positive if the status was positive at any time point after adjuvant therapy, and negative if the status was negative at all post-adjuvant time points. FIG. 24(B) shows ctDNA profiling of a representative patient during adjuvant and post-adjuvant treatment.

[819] FIG. 25 shows the relapse rate in 10 patients who were postoperative ctDNA positive before treatment with adjuvant chemotherapy, and how ctDNA was cleared by AChT in two of the three relapse-free ctDNA positive patients.

[820] FIG. 26A-B shows a comparison of time to relapse (TTR) using ctDNA and computed tomography: (A) comparison of TTR using ctDNA and CT imaging for 12 patients with relapse in whom relapse was detected by both methods; (b) serial ctDNA profiling of a representative patient with relapse with a ctDNA study period of 10.2 months.

[821] FIGS. 27A-D show serial ctDNA profiling of four representative patients.

[822] In conclusion, Signatera RUO's approach to massively parallel sequencing of personalized multiplex PCR tests targeting tumor-specific mutations provides a highly sensitive and specific platform for the detection and quantification of ctDNA. Postoperative ctDNA analysis allows dividing patients with colorectal cancer into subgroups with a very high or very low risk of relapse, both before and after chemotherapy. Both before and after adjuvant chemotherapy. Analysis of ctDNA over time provides effective postoperative monitoring of treatment and early detection of relapses. ctDNA analysis is very useful in making treatment decisions in both the adjuvant and post-adjuvant settings.

[823] EXAMPLE 2: ScDNA sequencing of plasma from patients with locally advanced bladder cancer for surveillance and monitoring of therapeutic efficacy

[824] Studies in various types of cancer have shown that circulating tumor DNA (ctDNA) levels can be effectively used to monitor response to neoadjuvant therapy and/or detect disease recurrence before clinical and radiological detection. In bladder cancer, mutations in plasma have previously been used to monitor response during treatment and detect early signs of metastatic disease. Recently, the detection of ctDNA over time in patients with NSCLC has been described, and a personalized circulating tumor DNA (ctDNA) detection test (Signatera™ RUO) has been developed.

[825] The purpose of the study was to use patient-specific mutations identified in the primary tumor to detect metastatic recurrence, assess prognosis, and monitor treatment response using ctDNA from plasma samples collected over time.

[826] Clinical protocol. Patients diagnosed with locally advanced muscle-invasive bladder cancer (MIBC) scheduled for chemotherapy were prospectively recruited between 2013 and 2017. All patients received neoadjuvant chemotherapy or first-line chemotherapy before cystectomy (CE) and were followed for 2 years (FIG. 28). Plasma samples were collected over time before and after systemic therapy and at scheduled follow-up visits after CE.

[827] Molecular protocol. Patient-specific somatic mutations were identified by whole exome sequencing (WES) of the tumor and compared with normal samples. Personalized multiplex PCR assays were used to detect patient-specific tumor DNA in plasma using cDNA from plasma samples collected over time. For each patient, 16 tumor-specific targets were sequenced and the data were analyzed in a clinically blinded manner for the presence of ctDNA. Samples were considered ctDNA positive if and only if they identified at least two positive patient-specific targets that met the required confidence threshold. Clinical results (radiographic images and response to treatment) were not blinded and were compared directly with the results of the Signatera plasma assay.

[828] A total of 50 patients were included in the study (Table 2).

[829] Table 2. Patient Characteristics and Demographics (N=50)

Age (years), ^a mean (range) 65.5 (43-77) Gender, n (%) Tumor stage in CE, ^b n (%) Women 9 (18) T0/Cis/Ta/T1 36 (72) Men 41 (82) T2/T3/T4a 12 (24) Tumor stage in TUP-MP, n (%) Metastases, ^With n (%) T1/T2
T4a/b 43 (86)
7 (14) Local recurrence (pelvis, rectum, urethra) 6 (12) Distant metastases (bones, lungs, liver, skin) 6 (12) Stage N before treatment, n (%) Clinical observation, days (range) Without signs of disease (CE; n=38) 434 (119-778) N0 43 (86) Clinical relapse (CE; n=10) 347 (65-973) N1/N2 7 (14) Progression (CE ineffective; n=2) 280 (105-455) ^a When collecting samples; ^b n=49; ^c n=12

[830] On-line sequencing quality control was carried out in 5 cycles of HiSeq PE 2x50; the average of the target read depth and background error rate is shown in FIG. 29.

[831] Early detection of relapse. Clinical relapse after CE was diagnosed in ten patients, and ctDNA was detected in plasma in nine of these patients, with a median of 128 days to clinical relapse. (Table 3). In one patient, the last control sample 4 months after CE had not yet been analyzed.

[832] Table 3. Comparison of molecular and clinical relapse (n=9).

Patient
No. Stage at diagnosis Stage at cystectomy Duration of ctDNA detection (days) Wedge. relapse after cystectomy (days) Localization of metastases 1 T2 N0 T2 N0 0 973 bones 2 T2 N0 T0 N0 265 624 local (urethra) 3 T4a N0 T4a N0 0 119 local (pelvis) 4 T2 N0 T3 N0 152 298 LU. lungs, bones, local (pelvis) 5 T2 N0 T2 N3 128 241 Liver, bones 6 T2 N0 T4a N0 96 379 local (pelvis) 7 T2 N0 T1 N3 245 309 local 8 T4b N0 T3 N0 186 324 local (rectum) 9 T4b N0 T4a N2 50 65 liver

[833] FIGS. 30A-F depict 6 patients who showed early relapse. ctDNA was detected in FNA as low as 0.02% during molecular relapse, which had a lead time of up to 265 days before clinical relapse (FIGS. 30D, 30E and 30F). In most patients, the response to chemotherapy was consistent with a decrease in ctDNA FNA (Table 4).

[834] Table 4. Prediction of response to treatment after chemotherapy

Patient no. Stage at diagnosis ctDNA FNA levels (%) Stage at cystectomy ctDNA FNA levels (%) 10a T4a 3.6 T2, N0 0.02 eleven T2 0.1 T0, N0 BUT 12 ^b,c T1 BUT T4a, N0 0.05 13 T2 0.1 T0, N0 BUT 14 T2 1.2 CIS, N0 BUT 15 ^a T1 0.3 T2, N0 BUT

^a Received first-line chemotherapy before CE. ^b Had TUR-P before TUR-B, T1 tumor from urethra, prostatic part. ^c Treatment response was defined as progression from unresectable to resectable status after chemotherapy (despite change in cancer stage). VVA, variant allele frequency; CE, cystectomy; N/A, not found.

[835] FIGS. 32A-B depict two patients who responded to ctDNA treatment; ctDNA initially detected at diagnosis decreased with neoadjuvant treatment and remained undetected after CE.

[836] In conclusion, these data demonstrate that ctDNA analysis (eg, with Signatera) can provide information about treatment response and detect disease relapse up to 265 days earlier than radiography. Survival analysis revealed significantly lower disease-free survival in patients with ctDNA at diagnosis or after cystectomy. Ultimately, ctDNA analysis may be included in routine surveillance for early detection of relapse and therefore potentially earlier initiation of alternative treatment such as immunotherapy. The beneficial impact on overall survival achieved through detection of ctDNA relapse should be assessed in randomized clinical trials.

[837] Example 3: Highly Sensitive Assay for Non-Invasive Cancer Recurrence Detection and Therapy Monitoring Based on Patient-Specific Multiplex PCR Based on Next Generation Sequencing (NGS)

[838] Identification of tumor mutations in circulating cell-free DNA has great potential for non-invasive detection of cancer recurrence before clinical manifestation, detection of minimal residual disease after curative treatment, and detection of therapeutically relevant mutations. Analyze and report results of analytical validation to identify tumor-specific variants using the current version of the assay.

[839] Signatera RUO. The Signatera™ (RUO) process begins with the identification and prioritization of somatic mutations based on whole exome sequencing of the tumor and corresponding normal samples. Patient-specific multiplex PCR assays targeting 16 somatic single nucleotide variants and insertion/deletion variants are then analyzed through massively parallel sequencing of plasma samples collected throughout the patient's disease course to help detect and monitor circulating tumor DNA.

[840] Analytical validation. Analytical validation of the current version of the Signatera assay (RUO) was performed on two breast cancer cell lines (HCC2218, HCC1395), one lung cancer cell line (NCI-H1395) and their corresponding normal counterparts (HCC2218-BL, HCC1395-BL, and NCI-H1395-BL). Various amounts of tumor cell line DNA (0; 0.005; 0.01; 0.03; 0.05; 0.1; 0.3; 0.5; 1%) were titrated to match the corresponding normal cell line DNA. Primer pools for multiplex PCR analyzes (each consisting of 16 runs of primer pairs specific for high-confidence somatic mutations) were designed using whole-exome data from matched tumor cell line DNA and matched normal cell line DNA. The initial total input to library preparation for each reaction was 20 thousand genome equivalents; ONV targets and insertions/deletions from corresponding tumor DNA samples were amplified using the multiplex PCR primer pools mentioned above. The mPCR products were barcoded, then pooled with other mPCR products, and subsequently sequenced on an Illumina HiSeq 2500 Rapid Run with 50 cycles of paired reads using the Illumina Paired End kit v2 with an average read depth of ~100,000/assay.

[841] FIG. 34 depicts a sequence of process quality control checks across 4 HiSeq PE 2x50 runs, including background transition and conversion error rates, and the average per depth-of-reads (DOR) of approximately 100K. (where any targets receiving fewer than 5000 reads were considered failures and were not considered for recognition).

[842] With Signatera (RUO), an analytical sensitivity of ~60% for detection was achieved with 0.03% spiked tumor DNA (Table 5).

[843] Table 5. Analytical sensitivity results for the current version of Signatera (RUO)

Tumor DNA concentration (%) Average number of mutant copies ONV detected Total number of NVGs used Sensitivity (%) 95% CI 0.005 1 120 794 15.1 12.7 - 17.8 0.010 2 246 635 38.7 34.9 - 42.7 0.030 6 291 433 67.2 62.6 - 71.6 0.050 10 208 244 85.3 80.2 - 89.5 0.100 20 227 228 99.6 97.6 - 100 0.300 60 177 177 100 98 - 100 0.500 100 120 120 100 97 - 100 1,000 200 76 76 100 93 - 100 An assay specificity of 99.9% was achieved for normal (0% spiked) cell line DNA (Table 1, Figure 3).

[844] For a given set of SNVs, the number of expected input percentages versus detected variant allele frequency (VAF) percentages for various added mutant DNA concentrations is shown for 6 targets, demonstrating high sensitivity above 0.03% tumor DNA concentration (FIG. 36) . False negatives, shown with inputs of no more than 0.01% OHB (starting with less than 2 mutant copies), include mutant molecules lost as a result of sampling (FIG. 36).

[845] The estimated sample-level sensitivity for Signatera (RUO) for recognizing at least two NVGs from a set of 16 NVGs with high confidence is shown in Table 6.

[846] Table 6. Sample-level calculated sensitivity

Tumor DNA concentration (%) Sensitivity (%) 0.005 49.97 - 72.02 0.010 94.55 - 99.56 0.030 99.97 - 99.99 0.050 100 0.100 100 0.300 100 0.500 100 1,000 100

[847] In conclusion. The Signatera RUO assay is a new way to non-invasively detect recurrence of personalized cancer signatures in plasma through ultra-deep multiplex sequencing of customized PCR assays (selected for the patient's tumor) with high sensitivity, high specificity and low error rates. Based on analytical validation results, the Signatera RUO assay at the NBC level has a specificity of 99.9% and a sensitivity of more than 65% for tumor fractions greater than 0.03%, and a sensitivity of 100% for tumor fractions greater than 0.1%. At the sample level, the Signatera RUO assay has greater than 95% sensitivity at 0.01% tumor fraction, nearly 100% sensitivity at 0.03% tumor fraction, and 100% sensitivity at 0.05% tumor fraction and above. These data demonstrate high detection rates at the single-molecule mutant level; they also suggest that lower plasma volumes can be used to achieve the same single-molecule detection with high specificity. The performance of the Signatera assay suggests that it can be used to determine the effectiveness of chemotherapy treatments.

[848] Example 4. Dynamic evaluation of multiplex individual ctDNA biomarkers in bladder cancer for diagnosis, monitoring and detection of relapse

[849] Justification. The use of circulating tumor DNA (ctDNA) as a biomarker to determine disease stage at diagnosis (DX), treatment response, and relapse monitoring is an emerging area for many cancer types. In bladder cancer, the use of ctDNA has shown promising results. A highly sensitive and specific approach to monitoring ctDNA based on SSP is revealed.

[850] Methods. A cohort of 50 patients with locally advanced muscle-invasive bladder cancer treated with neoadjuvant chemotherapy was prospectively enrolled. A panel of 16 tumor-specific mutations (SignateraTM RUO) was developed for each patient based on whole exome sequencing of tumor and germline DNA. In total, we analyzed ctDNA from longitudinally collected plasma samples from 386 time points obtained at diagnosis, during treatment, at cystectomy (Cx), and during monitoring until disease relapse or up to 2 years of follow-up. The results of ctDNA analysis were compared with radiographic findings and clinical outcomes. ctDNA from urine samples collected over time can also be analyzed for treatment response and disease relapse.

[851] Results. In DX, plasma ctDNA status was a strong predictor of disease-free survival. Specifically, 62% (8/13) of ctDNA+ patients on DX relapsed after neoadjuvant treatment and Cx; conversely, none (0/22) of the ctDNA- patients experienced relapse (log rank; p less than 0.0001). In addition, a strong correlation was also observed between the presence of ctDNA after Cx and disease relapse. Specifically, relapse after Cx was found in 100% (10/10) of ctDNA+ patients approximately 120 days (0–245 days) before radiography, while 0% (0/38) of ctDNA− patients had relapse noted (log rank); p less than 0.0001).

[852] Conclusions. Significant prognostic potential of ctDNA in bladder cancer at the time of DX was demonstrated, suggesting a role for ctDNA in bladder cancer staging. In addition, it has been shown that ctDNA is detected in all patients with disease relapse after Cx. Including ctDNA analysis in routine monitoring for early detection of relapse allows for earlier initiation of alternative treatment methods.

[853] Example 5: Serial Analysis of Circulating Tumor DNA to Detect Residual Disease, Evaluate the Effectiveness of Adjuvant Therapy, and Early Detection of Recurrent Colorectal Cancer

[854] Justification. Early detection of disease recurrence has been shown to improve survival in patients with colorectal cancer (CRC). Previous studies have analyzed circulating tumor DNA (ctDNA) to monitor tumor burden in CRC using small gene panels and digital droplet PCR. Here, a personalized multiplex PCR and SSP platform (Signatera™RUO) was used to detect ctDNA in serially collected plasma samples to evaluate whether ctDNA detection identifies a subgroup of patients at high risk of relapse both before and after adjuvant chemotherapy (ACT).

[855] Methods. A cohort of 130 patients with stage I-IV colorectal cancer treated according to the standard of care was analyzed. Tumor-specific panels of 16 mutations were developed for each patient using somatic mutation signatures derived from the RPE. Plasma samples (n=829) collected before and after surgery, as well as during chemotherapy, were analyzed. Relapse-free survival was calculated for patients stratified by ctDNA status after surgery (n=91) and after chemotherapy (n=58).

[856] Results. ctDNA status after surgery but before chemotherapy was assessed in 91 patients. Relapse was observed in 75% (6/8) of patients with ctDNA+ and only in 13% (11/83) of patients with ctDNA-. Effective treatment with AChT was observed in 30% (3/10) of postoperative ctDNA+ patients. They consistently had ctDNA- status in serial blood samples taken after chemotherapy and, accordingly, had no relapses at the end of the observation period. The ctDNA status after chemotherapy was assessed in 58 patients. Radiologically confirmed relapse was observed in 77% (10/13) of patients with ctDNA+ and in 4% (2/45) of patients with ctDNA-. On average, ctDNA detects relapse 9.13 months earlier than standard CT scanning.

[857] Conclusions. Serial postoperative ctDNA analysis allows dividing patients into subgroups with high or low risk of relapse, assessing the effectiveness of AChT treatment and ensuring early detection of relapse. Importantly, this also indicates that ACT can eliminate residual disease in 30% of postoperative ctDNA+ patients and may therefore be a treatment option for ctDNA+ patients. Thus, ctDNA analysis has great potential to guide treatment decisions in both adjuvant and post-adjuvant settings.

[858] Example 6: Early detection of residual breast cancer (BC) using reliable, scalable and personalized circulating tumor DNA (ctDNA) analysis precedes significant metastatic recurrence

[859] Justification. Many patients with breast cancer experience recurrence after primary treatment, but there are no reliable tests to detect distant metastases before they become apparent. Our variant demonstrated earlier identification of relapsed patients using personalized ctDNA analysis. The method is applicable to all patients and is not limited to hot spot mutations that are typically detected using gene panels.

[860] Methods. Forty-nine patients with non-metastatic breast cancer after surgery and adjuvant therapy were recruited. Plasma samples (n=208) were collected serially every six months. Using the analytically validated SignateraTM workflow, mutational signatures were determined from whole-exome data from the primary tumor and through ultra-deep sequencing (on average greater than 100,000X) and highly sensitive personalized assays were developed targeting 16 variants. A patient-by-patient assay was used to determine whether a mutational signature could be detected in plasma. All but 5 patients received chemotherapy, and the remaining 5 received radiation therapy.

[861] Results. In 16 of 18 (89%) patients with clinical relapse, ctDNA was detected before metastatic relapse diagnosed by clinical examination and biochemical measurements (CA15-3), and ctDNA remained positive at follow-up. Of the 2 patients with undetected ctDNA, one had a local recurrence and the other had two primary tumors. None of the 33 disease-free patients were ctDNA positive at any time point (n=142). Metastatic recurrence was predicted with high accuracy by Signatera and a lead time of up to 2 years (median = 9.5 months). A summary of these results is presented in FIG. 37. Detailed results are shown in FIG. 38-59.

[862] Conclusions. Using a scalable, validated, patient-specific ctDNA workflow allows for earlier identification of patients who will relapse. Accurate and earlier prognosis using ctDNA analysis may provide a means of monitoring breast cancer patients requiring second-line adjuvant therapy to prevent advanced, life-threatening metastatic progression.

[863] The results presented herein demonstrate that methods based on the detection of a personalized cancer marker/circulating tumor DNA in a patient's blood sample are very sensitive in predicting breast cancer recurrence. For example, 16 of 18 cases of relapse were correctly predicted, and no false positives were noted. The methods are also very consistent. Once a positive result is detected, a subsequent blood sample from the same patient remains consistently positive.

[864] The methods allow detection of a personalized cancer marker/circulating tumor DNA and prediction of relapse, for example, 27-610 days before detection of relapse using standard methods (eg, imaging). The median time to detect relapse is 9 months before relapse is detected by standard methods.

[865 ] Example 7: Personalized Serial Analysis of Circulating Tumor DNA (ctDNA) in High-Risk Early Stage Breast Cancer Patients to Monitor and Predict Response to Neoadjuvant Therapy (NAT) and Outcome in the I-SPY2 TRIAL Study

[866] Justification. ctDNA analysis offers a noninvasive approach to monitoring treatment response and resistance. Serial ctDNA testing during NAT may provide early indicators of emerging resistance and disease progression. In this study, ctDNA was analyzed in patients with high-risk early breast cancer who received NAT and radical surgery in the I-SPY2 TRIAL (NCT01042379). Data collected in this case study will be used to: (1) determine the relationship between ctDNA levels during early treatment and pathological complete response/residual cancer burden/long-term recurrence-free survival; (2) comparing the effectiveness of ctDNA and MRI in predicting tumor response to therapy; and (3) studying the relationship of ctDNA levels before and after NAT with 3-year event-free survival (EFS).

[867] Methods. ctDNA analysis was performed in 84 patients with high-risk stage II and III breast cancer randomized to neoadjuvant investigational agent (n = 57), the AKT inhibitor MK-2206 (M) in combination with paclitaxel (T), and then to doxorubicin and cyclophosphamide (AC). (M+T->AC) or standard of care (T->AC) (n = 27). HER2+ patients received trastuzumab (H) in addition to T or M+T.

[868] Serial plasma was collected before NAT, at the start of treatment (3 weeks), between regimens (12 weeks) and after NAT before surgery. Mutational profiles obtained from pre-treatment tumor biopsies and whole-exome germline DNA sequences were used to develop personalized assays targeting 16 patient tumor-specific variants to detect ctDNA in plasma. In the subgroup of patients who did not experience a complete pathological response (n = 18–22), mutations in residual cancer were compared with mutations detected in the tumor before treatment.

[869] Analysis: Of the 84 patients in this analysis, 15-25% were HR-HER2-, 40-60% H+ HER2- and 35-35% were HER2+. Of these, 20-25% and 30-42% achieved pathological complete response, respectively, in the control and treatment groups. Data are currently being collected to: (1) determine the relationship between ctDNA levels during early treatment and pathological complete response/residual cancer burden/long-term recurrence-free survival; (2) comparing the effectiveness of ctDNA and MRI in predicting tumor response to therapy; (3) studying the relationship of pre- and post-NAT ctDNA levels with 3-year event-free survival (EFS).

[870] Conclusions. This study provides a platform to evaluate the clinical relevance of ctDNA for serial monitoring of response to NAT. Accurate and early prediction of response using highly sensitive ctDNA analysis can facilitate timely and intelligent treatment changes to improve patients' chances of achieving pathological complete response. Finally, personalized ctDNA testing can complement imaging and pathological assessment of tumor response to optimize pathological complete response as a surrogate endpoint for improving EFS.

[871] Example 8: Early detection of residual breast cancer using scalable and personalized circulating tumor DNA (ctDNA) analysis precedes significant metastatic recurrence.

[872] INTRODUCTION

[873] Breast cancer is one of the most commonly diagnosed cancers worldwide and the second leading cause of cancer death in women. The current standard of care for women with nonmetastatic breast cancer is surgery, often followed by adjuvant therapy to eliminate microscopic residual disease that may lead to recurrence or further disease progression. Unfortunately, up to 30% of women who remain disease-free after curative treatment eventually relapse and die from metastatic breast cancer due to micrometastases. Current disease monitoring tools, including imaging and/or biochemical methods (including serum cancer antigen 15-3 (CA15-3) levels), have limited sensitivity and accuracy in detecting micrometastases. Late detection of metastases is associated with poor outcomes in many patients, highlighting the need to develop earlier and more sensitive measures of minimal residual disease (MRD).

[874] Circulating tumor DNA (ctDNA) released by apoptotic and necrotic cancer cells has been found to reflect tumor mutational signatures and is emerging as a potential non-invasive biomarker for monitoring tumor progression in various cancer types. In breast cancer, the use of ctDNA to detect minimal residual disease after surgery and/or adjuvant therapy and to monitor metastatic disease has shown promising results. In particular, plasma ctDNA levels have been shown to correlate with changes in tumor burden, thereby providing earlier assessment of response to treatment and allowing discrimination between patients with and without clinical relapse after surgery. Although numerous studies have documented the potential use of ctDNA analysis in breast cancer, to date there are no scalable tests that can reliably detect minimal residual disease in all patients.

[875] A personalized tumor-specific approach examining single nucleotide and INSERT/DELETION variants in plasma cDNA predicted relapse in patients with non-small cell lung cancer before clinical detection, suggesting that it may also be a suitable method for monitoring minimal residual disease in patients with cancer mammary gland. Here, using an expanded and scalable version of this approach, we sought to determine the use of serial ctDNA analysis for monitoring breast cancer recurrence after surgery and adjuvant therapy compared with traditional monitoring methods. The primary objective was to determine the “lead interval” between detection of ctDNA in plasma and clinical detection of advanced metastatic disease in patients with primary breast cancer.

[876] METHODS

[877] Patients and samples

[878] EBLIS is a multicentre prospective cohort study (NIHR REC number 13/LO/1152; IRAS: 126462) funded by Cancer Research UK and the National Institute for Health Research (NIHR). All patients provided written informed consent before the start of the study. The study protocol was approved by the Riverside Research Ethics Committee (REC): 13/LO/115; IRAS: 126462. All research and technical staff were blinded to patient outcomes.

[879] We recruited a total of 197 patients from 3 UK centers. Nine patients did not meet study inclusion criteria; Thus, a cohort of 188 patients was followed up with 6-month ctDNA blood sampling along with concomitant clinical examination and biochemical measurements including CA153 (FIG. 60). Eligible patients had to be 18 years or older, have no clinical evidence of metastatic disease, and therefore be considered healthy after surgery and adjuvant chemotherapy. All of them received adjuvant chemotherapy within 5 years after inclusion in the study. All had high-risk breast cancer (risk of death greater than 50% after 10 years without treatment, corresponding to a recurrence rate of 65% after 10 years without treatment).

[880] At the midpoint of the study (2 years), after the interim analysis and given that 50% of the predicted events were observed, we decided to perform a whole-exome analysis of the primary tumor in the first 50 patients.

[881] Blood samples were collected in K2-EDTA tubes. Samples were processed within 2 hours of collection by double centrifugation of blood, first for 10 minutes at 1000 g, then plasma for 10 minutes at 2000 g. Plasma was stored in 1 ml aliquots at −80°C.

[882] Signatera™ RUO platform. All research and technical personnel were blinded to the patient outcomes, and the analyzes were performed in a blinded manner. A strictly controlled semi-automated laboratory process was performed by trained personnel with signed SOPs and witnessing. All reagents and equipment have been tested and qualified before entering the Signatera™ RUO process. Process and reagent/equipment information was captured electronically and loaded into a database with built-in integrity checks. Quality control was carried out at each stage of the work process (FIG. 67A-D). Samples and amplicons that failed quality control were excluded from analysis. For each patient, a set of 45 SNPs were genotyped in PES and plasma sequencing to ensure sample matching. The PES data were used to develop a customized somatic mutation panel for all 49 patients. A total of 215 plasma samples were analyzed for ctDNA determination. For each target variant, a confidence score was calculated based on the read depth of the mutant and reference alleles. A plasma sample with two or more highly significant variants was considered ctDNA positive. Details of the Signatera workflow steps are provided below.

[883] Statistical analyzes

[884] All data were presented descriptively as means, medians or proportions. Disease-free survival from the date of inclusion in the study was determined using the Kaplan-Meier method. Cox proportional hazards regression analysis was used to model time to disease relapse. All statistical analyzes were performed using Stata, release 12.0 (Stata Corp., College Station, Texas, USA), and survival plots were generated using R version 3.5.1 (package “survminer” version 0.4.2.99).

[885] A conservative strategy for handling missing continuous data was used. If individual data points were missing, the last observation was carried forward or, if previous and subsequent data were available, then the average of these two values served as the missing data estimate.

[886] Whole exome sequencing

[887] Diagnostic FFPE blocks were checked by visual inspection and the block with the largest residual tumor was used for DNA extraction. A single hematoxylin and eosin section of the tumor was examined by a consultant histopathologist (D M) and at least 2 areas of the tumor were grossly dissected using a 1 mm microarray needle biopsy. DNA was extracted from FFPE tumor cores using the Gene Read kit (Qiagen) according to the manufacturer's instructions, and DNA concentration was measured as previously described.

[888] Illumina HiSeq was used to perform whole-exome sequencing on 200–500 ng of pooled tumor DNA from 1–3 regions generated from each FFPE block of the primary tumor (sequencing was performed by Novogen as a fee-for-service, at an average target read depth with 150x deduplication for all 49 tumor DNAs and 50x for 49 corresponding germline samples.All sequencing data are deposited in the European Genome-Phenome Archive.

[889] Custom panel design. Somatic variants specific to each patient were identified by analysis of the primary tumor and corresponded to normal PES for all 49 patients. The clonality of variants was inferred from the estimated proportion of cancer cells harboring the variant. Inferred clonality and variant types were used for the prioritized somatic SNVs and short insertions/deletions identified for each tumor. Standard Signatera assay processing was used to generate PCR primers for this set of variants. For each patient, 16 high-ranking compatible assays were selected for the individual patient-specific panel. Patient-specific 16-plex PCR assays were ordered from Integrated DNA Technologies.

[890] Extraction and quantification of cDNA. Up to 8 mL of plasma per case was available for this study (range, 1–8 mL; median, 5 mL). The entire volume of plasma was used for scDNA extraction. scDNA was extracted using the QIAamp Circulating Nucleic Acid kit (Qiagen) and eluted in 50 μl of DNA suspension buffer (Sigma). Each cDNA sample was quantified using the Quant-iT High Sensitivity dsDNA Assay Kit (Invitrogen). In 49 patients, cDNA was isolated from 215 serial plasma samples.

[891]Preparation of scDNA library.Up to 66 ng (20,000 genome equivalents) of scDNA from each plasma sample was used as input for custom library preparation. Cell-free DNA was subjected to end repair, A-tail extension, and ligation using special adapters.The purified ligation product was amplified for 20 cycles, purified using Ampure XP beads microbeads (Agencourt/Beckman Coulter).

[892] SSP workflow in multiplex PCR. An aliquot of each library was used as input for a patient-specific 16-plex PCR reaction. Samples were amplified using the Signatera tumor-specific assay and barcoded and then pooled . Sequencing was performed on an Illumina HiSeq 2500 Rapid Run with 50 paired-end read cycles using the Illumina Paired Ends v2 kit with an average read depth of >100,000X per amplicon.

[893] Bioinformatics Processing Process. All paired-end reads were merged using Pear software. Bases that did not match forward and reverse reads or had a low quality score were filtered out to minimize sequencing errors. The merged reads were aligned to the hg19 reference genome using Novoalign version 2.3.4. Amplicons with fewer than 5000 high-quality reads were considered to fail quality control. Quality control was performed using a proprietary program that checked a comprehensive list of statistics for each sample, which included total reads, mapped reads, targeted reads, number of failed targets, and average error rate (FIG. 67A-D)

[894] Recognition of plasma variants. To construct a variant-specific background error model, a large set of negative control samples (~1000) were preprocessed. Confidence scores were calculated for each target variant using mutant and reference alleles based on an error model and are incorporated herein by reference in their entirety. A plasma sample containing at least 2 variants with a confidence score above a predefined threshold (0.97) is called ctDNA positive.

[895] Analytical validation . Analytical validation was performed using titrations of mononucleosomal DNA from three cancer cell lines and their corresponding normal counterparts (ATCC). Two breast cancer cell lines (HCC2218 and HCC1395), one lung cancer cell line (NCI-H1395) and their corresponding matched normal B lymphoblast-derived cell lines (HCC2218-BL, HCC1395-BL and NCI-BL1395) were used. . DNA was exome-sequenced for each pair of cell lines, targeted variants were selected, and two primer pools were designed for multiplex PCR using standard Signatera processing. Tumor titration to normal mononucleosomal DNA was performed at an average NNA (based on DNA input) of 1.0; 0.5; 0.3; 0.1; 0.05; 0.03; 0.01; 0.005, 0% (replicate numbers were taken from two to nine - increasing with dilution factor). Due to the possibility of heterogeneity and aneuploidy in tumor cell lines, the NVA of individual targets may differ from the average administered NVA. To accurately calculate the nominal NVA of each target at each titration step, a separate experiment was conducted with mixtures of 10% NVA. The observed NVA from this experiment was then used to calculate input correction factors (observed NVA/10%). Correction factors were applied to the appropriate targets in the dilution series. Additional negative samples were analyzed using scDNA isolated from 16 individual plasma samples (approximately 8 ml each). For the titration series, 66 ng of scDNA (corresponding to 20,000 haploid genome equivalents) was used as input for Signatera library preparation; for plasma scDNA samples, all extracted DNA (ranging from 13 to 55 ng) was used as input. These libraries were then run through the Signatera plasma workflow (two primer pools for each titration sample and five primer pools for each cDNA sample), sequenced, and analyzed using the Signatera assay processing workflow. FIG. 68A shows our calculated sensitivity for detecting targets in plasma at various concentration levels. For negative samples, a target specificity of greater than 99.6% was achieved. By making the assumption that the custom panel has 10 to 16 clonal variants, sample-level sensitivity can be obtained as indicated in FIG. 68B. Sample-level specificity is estimated to be greater than 99.8%.

[896] RESULTS

[897] Here we present an analysis of the first 50 patients enrolled in the EBLIS study (FIG. 60). One tumor sample was not suitable for exome sequencing, so we continued the study with 49 patients. As of the reporting date (June 30, 2018), 18 patients had relapsed and 31 patients remained disease-free. All but 7 of the 49 patients received adjuvant chemotherapy or NAT with an anthracycline/taxane regimen. (See FIG. 69 and Table A). Forty-one patients were receiving adjuvant endocrine therapy throughout the duration of blood sampling. (Table B1-B3). Although repeat scanning was not required before study entry, all but 3 patients had imaging studies at diagnosis or at study entry, and all were within normal limits (Table B1-B3).

[898] Patients without relapse were consecutive patients recruited over the same period of time compared to those who relapsed, had sufficient tumor DNA extracted from the FFPE block of the primary tumor for exome profiling, and were followed for at least 2 years with serial blood sampling. We analyzed serial plasma samples in a blinded manner using an optimized Signatera workflow.

[899] Of the 18 patients who relapsed, 10 relapses were detected by CT scan, 3 by bone scan, and 1 each by mammography, MRI, liver enzyme elevation, and ultrasound. One patient died of unknown cause.

[900] Detection of circulating tumor DNA and preemptive interval

[901] We collected FFPE tumor samples from all patients: 39 did not receive systemic therapy before biopsy; 10 people received neoadjuvant chemotherapy (NACT) before resection of breast cancer (details of all systemic therapy, including timing of blood sampling, are contained in FIG. 60 and Tables B1-B3).

[902] To assess the presence of circulating tumor DNA for each patient, we developed individual tests targeting 16 somatic SNVs and INSERT/DELETION variants from each patient's tumor somatic mutation profiles (FIG. 64A-C). We then applied 49 corresponding personalized assays to each of 208 plasma samples (range: 1–8 time points) from 49 patients.

[903] Circulating tumor DNA was detected in 89% (16 of 18) of patients who relapsed; detection rates were 82%, 100%, and 100% for HR+/HER2-, HER2+, and triple negative breast cancer (TNBC), respectively (FIG. 61A and B). Of the two patients with relapse who were not detected by ctDNA, one (1018) had three primary cancers, and the other (1019) had a small local recurrence in the sternum (subsequently removed) (FIG. 61A, Table A). Notably, one patient (1072) initially had a normal blood profile but was ctDNA positive at the final visit; the patient had no evidence of disease at the time of the blood test, but the patient subsequently developed distant metastases shortly before the date of collection (FIG. 61A and FIG. 65A). We detected disease relapse 2 years before clinical relapse with a median of 266 days (8.9 months; FIG. 61B). When stratified by subtype, median lead times were 301, 164, and 258 days, respectively, for HR+/HER2-, HER2+, and triple negative breast cancer (TNBC) (FIG. 61B and C). Strikingly, two ER+ PR+ HER2- patients (1031 and 1051, FIG. 61A) had up to four time points that were ctDNA positive before clinical relapse, implying a lead interval of almost 2 years.

[904] We did not detect ctDNA in any of the 156 plasma samples from 31 patients who did not have disease relapse (FIG. 61A). The presence of ctDNA was associated with a significantly worse prognosis, both when ctDNA was detected in the first postoperative plasma sample (HR = 11.8 (95%, CI 4.3–32.5)) and in subsequent postoperative plasma samples (HR = 35 .8 (95% CI 8.0-161.3)) (FIG. 62A-B). All patients who tested positive for ctDNA relapsed within 50 months after surgery.

[905] For all patients, radiographic and scanning findings were negative for distant metastases until the date of significant recurrence, which was usually preceded by the patient's symptoms. Many patients had follow-up scans in addition to the scans taken at presentation, which were also negative (Table B1-B3). In seven patients (1004, 1055, 1072, 1091, 3018, 3019, 3048), scans were performed within 4 months of first testing positive for ctDNA, and all results were negative.

[906] We also monitored CA15-3 in 43 of 49 patients throughout the study. Thirty-nine of these patients showed normal results during the follow-up period. Twelve of the 18 patients who relapsed had normal CA15-3 levels during the follow-up period. Two patients (1051 and 1088) had a progressive increase in CA15-3 levels, but ctDNA was more sensitive than CA15-3, with ctDNA detected before the increase in CA15-3 levels earlier at 224 and 212 days, respectively (FIG. 63A and FIG. 65B). Patients 1111 and 1018 had one blood sample taken before clinical relapse and had elevated CA15-3 values. Patient 1111 tested positive for ctDNA and patient 1018 tested negative for ctDNA. Notably, six other patients (3 with relapse and 3 without relapse) had periodic blood samples taken with slightly elevated CA15-3, but this value fluctuated and did not reflect disease progression (Table B1-B3).

[907] Characterization of circulating tumor DNA

[908] The Signatera assay is designed to target 16 patient-specific somatic SNVs and INSERT/DELETION variants that provide the highest probability of detection. All 16 patients with relapses identified by ctDNA are shown in FIG. 63 and FIG. 65. In 10 plasma samples from eight patients with relapse, variant allele frequencies (VAFs) ranging from 0.01-0.02% were detected. The lowest variant allele frequency of 0.01% corresponds to the detection of one mutant molecule in the plasma sample (FIG. 66). This level of sensitivity can be seen in four patients - 1004, 01055, 1072 and 1096 (FIG. 63, FIG. 65, Table C). Test specificity greater than 99.5% is achieved by requiring that two or more variants measure above the selected confidence threshold of the recognition algorithm in order to have confidence that the ctDNA is present in the plasma. The specificity is emphasized by the fact that none of the plasma samples from patients without relapse were found to be positive.

[909] Regarding changes in ctDNA profiles over time, of the 16 relapses identified by ctDNA, seven patients had a positive ctDNA test at all time points analyzed and showed an increase in both the number of detected variants and the percentage of PTA with the passage of time; six patients who were initially ctDNA negative later became positive, and three patients with relapse had only one plasma time point available for analysis, all of whom were ctDNA positive (FIG. 63 and FIG. 65).

[910] Five of these patients are highlighted in FIG. 63, where at least one of each subtype is represented - HR+/HER2- (1031 and 1051), HER2+ (1096) and TNBC (1055 and 1074). Patients 1031, 1051, 1055 were initially ctDNA negative, but ctDNA became detectable at later time points. (FIG. 63A-C). Two of these patients, both HR+ (1031 and 1051), had the longest lead time between ctDNA-detected molecular relapse and clinical relapse, respectively, 721 and 611 days (FIG. 63A-B). In patients 1031, 1055, 1074 and 1096, variant allele frequencies were found in the range of 0.01 to 0.02%, and they showed a progressive increase in NVA, correlating with disease progression (FIG. 63B-E). After detection of ctDNA, all patients remained positive during the observation period (FIG. 63 and FIG. 65).

[911] In general, disease progression can be monitored by both variant allele frequency and the number of variants detected, as shown in FIG. 63F. Because the number of time points varied between patients, differences between the first and last time points in a series of plasma samples from the same patient are highlighted. The median variant allele frequency increased from 0.092% at the first time point (range: 0.01 to 9.2%) to 3.9% (range: 0.05 to 64.4%), while the median number of variants detected at the first time point was 5 (range: 2-12) compared to 12 variants (range: 5-15) at the last time point. The low number of variants detected at early time points and the fact that they are present in very low copy numbers indicate the importance of testing multiple mutations present in a patient's tumor to provide a highly sensitive test for the presence of ctDNA in the patient's plasma.

[912] Discussion

[913] This report describes a reliable and reproducible method for monitoring minimal residual disease in patients with breast cancer across major subtypes. The approach uses tumor genomic data to develop patient-specific assays. The plasma scDNA is then sequenced to very high depth, averaging over 100,000 reads per target, to achieve sensitivity down to a single mutant molecule. This new transformative technology is robust and scalable for implementation in modern healthcare, and has already been developed for use in research. This is particularly timely given, for example, the recent announcement that all cancers in the UK will receive genomic profiling from autumn 2018. The Signatera platform is capable of providing a personalized way to detect micrometastases, which is clearly supported by the study results. In this study, all but one of the breast cancer patients who relapsed with distant metastases had a positive blood test before significant recurrence, and in some cases showed a nearly 2-year lead time.

[914] A previous prototype study of this technology showed promising results in patients with non-small cell lung cancer (Abbosh et al., 2017, incorporated herein by reference in its entirety), and the workflow has since been refined to achieve higher detection sensitivity rates ctDNA and in order to be more economical. In this case, we demonstrate the excellent repeatability and accuracy of this system. Focusing on SNVs unique to the patient, rather than known driver genes, represents the most accurate and sensitive way to detect MRD in patients with breast cancer.

[915] The clinical application of ctDNA measurements in the treatment of early cancer remains highly controversial, and a recent joint review by ASCO and ACP concluded that there is no evidence of its clinical utility, and there is little evidence of its clinical validity for monitoring treatment or early detection of residual disease. diseases in the early stages of cancer. This finding may be partly due to the fact that all previous studies were conducted without comparison with conventional markers, as was done here.

[916] Several important points emerge from this study. First, relapse can now be predicted with a high degree of accuracy in patients without evidence of disease on imaging. Second, most patients are currently followed in US and UK centers with a combination of annual mammography and, in some cases, CA15-3 and liver function tests. All of these, with the exception of two cases of progressively elevated CA15-3 in the absence of detectable metastases, were normal until significant metastatic recurrence in our cohort, demonstrating the limitations of current approaches used by clinicians. Seven of the patients who relapsed also had scans around the time of first ctDNA detection, and all results were negative. Of note is the observation that some patients have transiently elevated CA15-3 levels and remain healthy, whereas the Signatera test was always positive upon detection and there were no false positives. Third, although NACT shrunk the primary cancer (on which exome sequencing was performed), the mutation signature of the residual cancer reflects residual metastatic disease in these patients, suggesting that a personalized test developed based on the mutation profile of the primary tumor is not only possible, but also effective.

[917] However, there is a requirement that the ctDNA be released into the plasma, and thus the test may be limited to those with fairly aggressive disease to begin with, and therefore may not be applicable to patients with smaller, less aggressive breast cancers. which often have a good prognosis. Because the limit of detection in our assay is limited to a single molecule, the lack of detection is likely due to tumor biology, in which less aggressive tumors may secrete fewer ctDNA molecules. An example of this is one patient (1018) who experienced recurrence of locally resectable disease while being ctDNA negative. Additionally, in patients with multiple primary tumors, all tumors should be profiled to monitor disease progression. This is observed in patient 1019, who had three primary tumors; unfortunately, only one of them was subjected to exome sequencing due to the limited availability of available tissue. In this case, recurrent disease was not detected by ctDNA, and our hypothesis is that the metastases did not originate from the same tumor that was sequenced. Finally, the test is not suitable for detecting a second primary breast cancer unless it is a recurrence of the original tumor; An example of this is patient 1044, who had a contralateral second primary cancer detected on routine mammography, but plasma remained ctDNA negative throughout (Table A).

[918] The analytical platform described herein is not intended for the identification of detectable targets from plasma. Most of the ONVs selected for use in detecting tumor DNA were unique to each patient, and they were selected as a reflection of tumor burden rather than as representative of driver mutations that often contribute to cancer progression. However, this is also an advantage of the Signatera assay - selection of passenger and clonal mutations is imperative for monitoring disease burden, as driver mutations often provide a selective advantage leading to alteration of tumor heterogeneity. Although 16-plex assays do not provide meaningful targets, tumor RPE may provide such targets, and plasma libraries can also be tested with other assays that identify detectable mutations.

[919] Our results complement others in which significant mutations, for example in PIK3CA , are observed during neoadjuvant and adjuvant therapy. While driver mutation targeting can track the progression of some tumors (links), studies have shown that they are not beneficial in detecting early metastatic recurrence in all patients because not all patients have the same target driver mutations in tumors (links). In one previous study, only 78% (43 of 55) of cases had one or more somatic mutations that were identified and subsequently monitored by digital droplet PCR. We also analyzed an off-the-shelf breast cancer panel (Oncomine™ Breast cfDNA Assay) targeting over 150 hotspots in 10 breast cancer genes, and this panel detected ctDNA in only 73% of breast cancer patients. Because gene panels cannot reflect the heterogeneity of all breast cancer cases, they do not include all patients and are not applicable to all patients with breast cancer, as shown in the above two examples. Therefore, tumor exome profiling and Signatera's personalized approach should be used to detect minimal residual disease in all patients with metastatic breast cancer; in case of a positive result, it is necessary to conduct a secondary analysis for the presence of significant mutations.

[920] There are several important implications of our study. Thus far, systemic treatment with targeted or cytotoxic therapy has been shown to be curative only when administered in the adjuvant setting; Treatment of advanced metastatic disease is rarely, if ever, curative (link). The approach presented here offers an alternative: attempting to save ctDNA patients with second-line therapy. Another application could be to help evaluate new drugs, especially those whose mechanism is to enhance the immune response. Until now, an indirect measure of success—time to tumor progression—has been used; Signatera's method of detecting ctDNA will now allow the use of another criterion for success - reduction or removal of ctDNA - as a measure of success.

[921] In conclusion, the Signatera platform is capable of detecting MRD in breast cancer patients with high sensitivity. It is superior to traditional monitoring methods and shows promise for monitoring patients from a precision medicine perspective. For the first time, a blood test is being offered that reassures patients that their disease is under control.

[922] Summary

[923] Many patients with breast cancer relapse after primary treatment, but there are no reliable tests to detect distant metastases before they become evident. Here, we demonstrate earlier identification of breast cancer recurrence using scalable, personalized analysis of circulating tumor DNA (ctDNA). The method is applicable to all patients and is not limited to hot spot mutations that are typically detected using gene panels.

[924] Forty-nine patients with non-metastatic breast cancer after surgery and adjuvant therapy were recruited. Plasma samples (n = 208) were collected serially every six months. Using the analytically validated Signatera™ workflow, we identified mutational signatures from primary tumor whole-exome data and developed personalized tests with high sensitivity using ultra-deep sequencing (averaging over 100,000X) targeting 16 variants. A patient-specific assay was used to determine the presence of ctDNA in plasma.

[925] In 16 of 18 (89%) patients with clinical relapse, ctDNA was detected before metastatic relapse diagnosed by clinical examination, radiological examination, and CA15-3 measurement, and the patients remained ctDNA positive at follow-up. Of the 2 patients in whom no ctDNA was detected, one had only a small local recurrence (currently removed) and the other had three primary tumors. None of the 31 disease-free patients were ctDNA positive at any time point (n = 156). Metastatic recurrence was predicted by Signatera with a lead time of up to 2 years (median = 8.9 months, HR: 35.84 (95% CI 7.9626 - 161.32)).

[926] Using a scalable, validated, patient-specific ctDNA-based workflow detects minimal residual disease before metastatic recurrence of breast cancer, outpacing clinical detection of major breast cancer subtypes. Accurate and earlier prognosis by ctDNA analysis may provide a means of monitoring breast cancer patients requiring second-line adjuvant therapy in an attempt to prevent significant life-threatening metastatic recurrence.

[927] Table A1 . Clinical characteristics of all 49 patients. Patients marked with an asterisk* have died.

Case no. Age at diagnosis Tumor grade ER status Her2 status Stage
TNM Time from first sample collection to relapse (days) Date of relapse Localization of distant metastasis E001 48 3 - + T2 N1 M0 - absent absent E002 56 3 + + T2 N2 M0 - absent absent E003 65 2 + - T3 N2 M0 133 06/25/2014 Pleural effusion E004 50 2 + - T2 N3 M0 - absent absent E005 69 2 + - T3 N3 M0 263 01/02/2015 Involvement of the lymph nodes of the root of the right lung and mediastinum E006 48 2 + - T2 N2 M0 - absent absent E007 67 2 + ₊ T2 N3 M0 - absent absent E008 47 2 + - T3 N3 M0 - absent absent E009 44 3 + - T2 N1 M0 125 10/28/2014 Sternum, pelvis, spine* E010 38 2 + - T2 N2 M0 561 01/08/2016 absent E011 69 2 + - T2 N2 M0 - absent absent E012 66 3 + - T2 N1 M0 - absent absent E013 80 2 + - T3 N2 M0 - absent absent E014 57 2 + - T2 N3 M0 - absent absent E015 81 3 - - T2 N1 M0 - absent absent E016 80 3 - + T2 N2 M0 - absent absent E017 67 3 + - T3 N3 M0 1281 01/31/2018 Sternoclavicular joint, skin, lung E018 55 2 + - T3 N3 M0 - absent absent E019 53 3 + - T2 N1 M0 - absent absent E020 80 2 + - T2 N2 M0 - absent absent E021 44 2 + - T2 N3 M0 - absent absent E022 54 2 + - T2 N2 M0 - absent absent E023 54 2 + - T3 N2 M0 - absent absent E024 78 3 - - T2 N1 M0 - absent absent E025 46 3 - - T1 N2 M0 - absent absent E026 73 2 + - T2 N3 M0 793 03/17/2017 Spine E027 60 2 + - T2 N2 M0 - absent absent E028 76 3 + + T1 N2 M0 - absent absent E029 57 3 - - T2 N1 M0 442 11/30/2016 Lung E030 70 3 + - T2 N1 M0 - absent absent E031 77 2 + - T2 N3 M0 1087 05/23/2018 Skin on the lower back on the right E032 39 3 + - T1 N1 M0 - absent absent E033 57 2 - - T1 N0 M0 570 12/30/2016 Intraclavicular fossa and sentinel lymph nodes E034 66 2 + - T1 N3 M0 - absent absent E035 59 2 + - T2 N3 M0 - absent absent E036 46 2 + - T1 N3 M0 415 10/13/2016 Bone and bladder E037 78 3 - + T2 N3 M0 610 06/07/2017 Lung E038 63 1 + - T2 N2 M0 - absent absent E039 48 3 + - T2 N1 M0 - absent absent E040 55 3 - + T2 N1 M0 313 09/05/2016 Bone E041 67 3 + - T2 N1 M0 - absent absent E042 49 2 + - T2 N1 M0 - absent absent E043 41 2 + - T3 N1 M0 68 05/23/2016 Liver, lung, bone* E044 46 3 + - T2 N1 M0 323 02/04/2016 Local lymph nodes E045 46 3 + + T2 N1 M0 - absent absent E046 57 3 - - T2 N2 M0 269 02/08/2016 Bone, liver and pleura E047 62 3 + - T3 N1 M0 117 11/18/2015 Local nodes and intrapulmonary nodes E048 52 3 - - T2 N1 M0 202 12/31/2015 Bone and pleura E049 45 3 - - T2 N1 M0 148 08.10.2016 Unknown*

[928] Table B1 . Temporary points for plasma collection from patients.

Case ID Temporary point of plasma collection (date) №0 No. 1 No. 2 No. 3 No. 4 No. 5 No. 6 No. 7 No.EoS E001 01/15/2014 10/08/2014 01/07/2015 06/24/2015 01/20/2016 07/13/2016 01/04/2017 07/05/2017 E002 02/01/2014 09/08/2014 02/09/2015 09/01/2015 02/18/2016 08/25/2016 03/16/2017 E003 02/12/2014 E004 04/14/2014 10/27/2014 04/27/2015 10/19/2015 06/09/2016 11/28/2016 E005 04/14/2014 10/16/2014 E006 05/07/2014 09.12.2015 08/23/2016 03/01/2017 E007 05/12/2014 11/11/2014 11/17/2015 05/26/2016 12/08/2016 07/16/2017 E008 05/16/2014 03.11.2014 04/20/2015 10/19/2015 12/12/2016 E009 06/25/2014 E010 06/26/2014 12/10/2014 01/20/2016 E011 07/09/2014 01/28/2015 08/19/2015 01/06/2016 07.202016 01/25/2017 09/06/2017 E012 07/16/2014 01/14/2015 07/08/2015 01/27/2016 07/272016 01/25/2017 07/26/2017 E013 07/16/2014 01/14/2015 07/15/2015 01/13/2016 07/13/2016 01/25/2017 08/09/2017 E014 07/16/2014 01/21/2015 02/03/2016 07/12/2016 01/31/2017 E015 07/21/2014 01/14/2015 07/15/2015 01/20/2016 07/20/2016 02/08/2017 07/26/2017 E016 07/23/2014 01/21/2015 10/29/2015 06/08/2016 01/11/2017 07/05/2017 E017 07/30/2014 01/14/2015 07/08/2015 02/10/2016 08/31/2016 02/22/2017 07/04/2017 E018 08/11/2014 02/16/2015 04/04/2016 11/28/2016 05/24/2017 E019 08/27/2014 02/11/2015 06.08.2015 03/16/2016 10/18/2016 04/21/2017 E020 09/04/2014 04/20/2015 10/15/2015 01/14/2016 03.11.2016 05/03/2017 E021 09.09.2014 03/10/2015 09.15.2015 03/24/2016 09/22/2016 03/30/2017 E022 09/10/2014 03/11/2015 08/04/2015 E023 09/11/2014 02/05/2015 08/03/2015 03/08/2016 08/25/2016 03/09/2017 E024 09/24/2014 03/18/2015 09.23.2015 05/18/2016 11/30/2016 06/28/2017 E025 05.11.2014 05/13/2015 12/23/2015 06/22/2016 01/18/2017 09/06/2017 E026 01/14/2015 07/15/2015 02/24/2016 09/13/2016 02/28/2017 E027 02/03/2015 08/04/2015 02/04/2016 E028 02/11/2015 08/30/2016 02/21/2017 09/05/2017 E029 09/15/2015 03/17/2016 09.29.2016 01/12/2017 E030 05/20/2015 11/18/2015 05/18/2016 07/19/2017 E031 06/01/2015 10/29/2015 04/28/2016 01/26/2017 07/26/2017 E032 06/03/2015 10/14/2015 06/08/2016 12/21/2016 E033 06/09/2015 12/22/2015 07/21/2016 01/09/2017 E034 06/10/2015 09.12.2015 06/07/2016 07/04/2017 E035 06/18/2015 01/21/2016 09.09.2016 01/30/2017 06/12/2017 E036 08/25/2015 03/24/2016 09/22/2016 E037 10/06/2015 05/24/2017 05/24/2017 E038 10/07/2015 04/13/2016 10/19/2016 04/19/2017 E039 10.20.2015 05/18/2016 11/16/2016 05/31/2017 E040 10/28/2015 04/13/2016 10/19/2016 E041 01/20/2016 07/20/2016 02/08/2017 08/04/2017 E042 02/03/2016 08/16/2016 01/24/2017 E043 03/16/2016 E044 03/18/2015 09.23.2015 02/17/2016 E045 12/15/2015 06/15/2016 12/14/2016 E046 05/15/2015 06.11.2015 02/12/2016 E047 07/24/2015 12/16/2016 E048 06/12/2015 04.12.2015 01/08/2016 E049 05/13/2016

[929] Table B2 . Patient CA15-3 levels.

Case ID CA15-3 levels №0 No. 1 No. 2 No. 3 No. 4 No. 5 No. 6 No. 7 No.EoS E001 7 20 19 22 20 18 17 20 E002 8 9 8 10 9 9 7 E003 17 E004 18 18 N/O 18 18 19 17 E005 22 26 E006 15 eleven* 14* eleven 13 13 E007 15 14 14 13 14 14 E008 17 17 18 17 21 15 E009 51 thirty* E010 13 12 16 16 E011 9 8 7 8 9 9 E012 9 8 9 8 10 9 10 E013 31 33 23 21 19 57 28 E014 18 17 18 18 20 E015 20 20 22 22 20 23 20 E016 15 13 15 17 16 15 E017 13 13 13 14 15 17 17 E018 5 6 5 7 7 E019 29 23 29 26 29 29 E020 26 51 32 28 29 36 E021 14 14 14 14 14 14 E022 10 10 12 E023 eleven 13 13 12 13 14 E024 23 23 28 24 25 22 E025 20 20 20 17 19 21 E026 24 31 46 58 104 138 E027 24 23 20 E028 35 33 29 E029 10 13 10 12 12 E030 13 13 10 14 E031 34 27 31 32 BUT E032 12 10 eleven eleven 9* E033 18 21 20 22 E034 26 25 23 27 E035 21 19 24 22 21 E036 25 48 131 E037 20 22* 21 E038 16 17 19 16 E039 14 12 eleven 13 E040 33 35 36 E041 26 23 26 22 E042 5 5 5 E043 65 E044 E045 E046 E047 E048 E049

[930] Table B3 . Endocrine therapy for patients.

Case ID Previous chemotherapy Endocrine therapy №0 No. 1 No. 2 No. 3 No. 4 No. 5 No. 6 No. 7 No.EoS E001 Yes absent absent absent absent absent absent absent absent absent E002 Yes Letrozole Letrozole Letrozole Letrozole Letrozole Letrozole Letrozole Letrozole Letrozole E003 Yes Letrozole Letrozole E004 Yes Tamoxifen Tamoxifen Tamoxifen Tamoxifen Tamoxifen Tamoxifen E005 No Anastrozole Anastrozole Anastrozole E006 Yes Letrozole Letrozole Letrozole Exemestane Letrozole Letrozole E007 Yes Anastrozole Tamoxifen No visit Tamoxifen Tamoxifen Tamoxifen Tamoxifen E008 Yes Anastrozole/Zoladex Anastrozole/Zoladex Anastrazole Exemestane Exemestane Exemestane E009 Yes absent E010 Yes Tamoxifen/Zoladex Tamoxifen/Zoladex E011 Yes Letrozole Letrozole Letrozole Letrozole Letrozole Letrozole E012 Yes Anastrozole Anastrozole Anastrozole Anastrozole Anastrozole Anastrozole Anastrozole E013 No Letrozole Letrozole Letrozole Letrozole Letrozole Letrozole Letrozole E014 Yes Letrozole Letrozole omitted Letrozole Letrozole Letrozole E015 No absent absent absent absent absent absent absent E016 No Anastrozole Anastrozole Anastrozole Anastrozole Anastrozole Anastrozole E017 Yes Tamoxifen Tamoxifen Tamoxifen Tamoxifen Tamoxifen Tamoxifen Tamoxifen E018 Yes Exemestane Exemestane omitted absent absent absent E019 Yes absent Tamoxifen Tamoxifen Tamoxifen Letrozole Letrozole E020 No Tamoxifen Tamoxifen Tamoxifen Tamoxifen Tamoxifen Tamoxifen E021 Yes Tamoxifen Tamoxifen Tamoxifen Tamoxifen Tamoxifen Tamoxifen E022 Yes Anastrozole Anastrozole E023 Yes Letrozole Letrozole Letrozole Letrozole Letrozole Letrozole E024 Yes absent absent absent absent absent absent E025 Yes Tamoxifen Tamoxifen Tamoxifen Tamoxifen Tamoxifen Tamoxifen E026 Yes Letrozole Letrozole Letrozole Letrozole Letrozole E027 Yes Anastrozole Anastrozole Anastrozole E028 Yes Letrozole omitted omitted Letrozole Letrozole Letrozole E029 Yes absent absent absent absent E030 Yes Letrozole Letrozole Letrozole Letrozole E031 No Anastrozole Anastrozole Anastrozole Anastrozole Anastrozole E032 Yes absent absent Absent absent E033 Yes absent absent absent E034 Yes Letrozole Letrozole Letrozole Letrozole Letrozole E035 Yes Anastrozole Anastrozole Anastrozole Anastrozole Anastrozole E036 Yes Tamoxifen Tamoxifen Tamoxifen E037 No absent absent E038 Yes Letrozole Letrozole Letrozole Letrozole E039 Yes absent absent Letrozole Letrozole E040 Yes Letrozole Letrozole E041 Yes absent Tamoxifen Tamoxifen Tamoxifen E042 Yes Tamoxifen Tamoxifen Tamoxifen E043 Yes Tamoxifen E044 Yes Tamoxifen Tamoxifen Tamoxifen E045 Yes Tamoxifen No entry No entry E046 Yes Absent Absent Absent E047 Yes Endocrine (to be clarified) Endocrine (to be clarified) E048 Yes Absent Absent Absent E049 Yes Absent Absent Absent

[931] Table C . Sample level summary and NVA.

Patient no. Time point Days after surgery ng DNA for preparation. librarian Quantity
discovered
targets ctDNA status Number of analyzes that passed quality control Average NVA (%) Median NVA (%) Maksim. NVA (%) Minimum NVA (%) 1002 1 468 9.99 0 Negative 16 absent absent absent absent 1002 2 734 6.84 0 Negative 16 absent absent absent absent 1002 3 825 7.81 0 Negative 16 absent absent absent absent 1002 4 993 6.36 0 Negative 16 absent absent absent absent 1002 5 1203 8.93 0 Negative 16 absent absent absent absent 1002 6 1378 11.83 0 Negative 16 absent absent absent absent 1002 7 1553 14.74 0 Negative 16 absent absent absent absent 1002 8 1735 10.68 0 Negative 16 absent absent absent absent 1003 1 530 16.28 0 Negative 16 absent absent absent absent 1003 2 740 21.77 0 Negative 16 absent absent absent absent 1003 3 894 7.36 0 Negative 16 absent absent absent absent 1003 5 1268 10.87 0 Negative 16 absent absent absent absent 1003 6 1457 28.57 0 Negative 16 absent absent absent absent 1003 7 1660 17.08 0 Negative 16 absent absent absent absent 1004 1 302 32.98 3 Positive 16 0.015 0.014 0.018 0.013 1009 1 1452 35.74 0 Negative 16 absent absent absent absent 1009 2 1648 18.31 0 Negative 16 absent absent absent absent 1009 3 1830 16.79 0 Negative 16 absent absent absent absent 1009 4 2005 22.25 0 Negative 16 absent absent absent absent 1009 5 2239 41.96 0 Negative 16 absent absent absent absent 1009 6 2411 21.3 0 Negative 16 absent absent absent absent 1011 1 920 17.14 12 Positive 16 2,902 1,442 9.219 0.697 1011 2 1105 66 12 Positive 16 2.39 0.988 8.31 0.691 1013 1 37 21.44 0 Negative 16 absent absent absent absent 1013 3 876 37.67 0 Negative 16 absent absent absent absent 1013 4 1066 43.23 0 Negative 16 absent absent absent absent 1014 1 440 22.19 0 Negative 16 absent absent absent absent 1014 2 623 18.16 0 Negative 16 absent absent absent absent 1014 3 994 17.33 0 Negative 16 absent absent absent absent 1014 4 1185 13.01 0 Negative 16 absent absent absent absent 1014 5 1381 27.12 0 Negative 16 absent absent absent absent 1014 6 1601 9.01 0 Negative 16 absent absent absent absent 1015 1 1330 18.83 0 Negative 16 absent absent absent absent 1015 2 1501 27.39 0 Negative 16 absent absent absent absent 1015 3 1669 9.04 0 Negative 16 absent absent absent absent 1015 4 1851 13.65 0 Negative 16 absent absent absent absent 1015 5 2271 23.95 0 Negative 16 absent absent absent absent 1018 1 131 66 0 Negative 16 absent absent absent absent 1019 1 296 35.69 0 Negative 16 absent absent absent absent 1019 2 463 32.73 0 Negative 16 absent absent absent absent 1019 3 869 66 0 Negative 16 absent absent absent absent 1021 1 405 30.37 0 Negative 16 absent absent absent absent 1021 2 608 51.4 0 Negative 16 absent absent absent absent 1021 3 811 31.59 0 Negative 16 absent absent absent absent 1021 4 951 35 0 Negative 16 absent absent absent absent 1021 5 1147 45.12 0 Negative 16 absent absent absent absent 1021 6 1336 66 0 Negative 16 absent absent absent absent 1021 7 1560 41.33 0 Negative 16 absent absent absent absent 1023 1 565 23.61 0 Negative 16 absent absent absent absent 1023 2 747 43.97 0 Negative 16 absent absent absent absent 1023 3 922 18.15 0 Negative 16 absent absent absent absent 1023 4 1125 53.07 0 Negative 16 absent absent absent absent 1023 5 1307 33.89 0 Negative 16 absent absent absent absent 1023 6 1489 66 0 Negative 16 absent absent absent absent 1023 7 1671 35.48 0 Negative 16 absent absent absent absent 1024 1 132 66 0 Negative 16 absent absent absent absent 1024 2 496 32.12 0 Negative 16 absent absent absent absent 1024 3 314 50.66 0 Negative 16 absent absent absent absent 1024 4 678 -1.41 0 Negative 16 absent absent absent absent 1024 5 860 39.77 0 Negative 16 absent absent absent absent 1024 6 1056 66 0 Negative 16 absent absent absent absent 1024 7 1252 66 0 Negative 16 absent absent absent absent 1025 1 300 40.77 0 Negative 16 absent absent absent absent 1025 2 489 66 0 Negative 16 absent absent absent absent 1025 3 867 66 0 Negative 15 absent absent absent absent 1025 4 1027 40.02 0 Negative 15 absent absent absent absent 1025 5 1230 44.77 0 Negative 15 absent absent absent absent 1028 1 581 19.24 0 Negative 16 absent absent absent absent 1028 2 758 12.8 0 Negative 16 absent absent absent absent 1028 3 940 11.26 0 Negative 16 absent absent absent absent 1028 4 1129 21.97 0 Negative 16 absent absent absent absent 1028 5 1311 23.56 0 Negative 16 absent absent absent absent 1028 6 1514 22.65 0 Negative 16 absent absent absent absent 1028 7 1682 23.79 0 Negative 16 absent absent absent absent 1029 1 311 66 0 Negative 16 absent absent absent absent 1029 2 493 66 0 Negative 16 absent absent absent absent 1029 3 774 46.18 0 Negative 16 absent absent absent absent 1029 4 631 66 0 Negative 16 absent absent absent absent 1029 5 1214 66 0 Negative 16 absent absent absent absent 1029 6 1389 66 0 Negative 16 absent absent absent absent 1031 1 330 20.3 0 Negative 15 absent absent absent absent 1031 2 673 8.32 0 Negative 15 absent absent absent absent 1031 3 498 27.97 0 Negative 15 absent absent absent absent 1031 4 890 14.11 4 Positive 15 0.121 0.072 0.303 0.037 1031 5 1093 18.35 9 Positive 15 0.074 0.068 0.14 0.018 1031 6 1268 50.46 8 Positive 15 0.045 0.027 0.168 0.023 1031 7 1400 22.65 10 Positive 15 0.133 0.075 0.424 0.048 1033 1 1680 34.17 0 Negative 16 absent absent absent absent 1033 2 1869 37.49 0 Negative 16 absent absent absent absent 1033 4 2520 66 0 Negative 16 absent absent absent absent 1033 5 2697 58.43 0 Negative 16 absent absent absent absent 1038 1 29 35.15 0 Negative 16 absent absent absent absent 1038 2 197 24.26 0 Negative 16 absent absent absent absent 1038 3 373 7.17 0 Negative 15 absent absent absent absent 1038 4 596 25.45 0 Negative 16 absent absent absent absent 1038 5 812 18.91 0 Negative 16 absent absent absent absent 1038 6 997 50.72 0 Negative 16 absent absent absent absent 1042 1 147 15.99 0 Negative 15 absent absent absent absent 1042 2 375 17.03 0 Negative 15 absent absent absent absent 1042 3 553 18.42 0 Negative 15 absent absent absent absent 1042 4 644 22.36 0 Negative 15 absent absent absent absent 1042 5 938 25.68 0 Negative 15 absent absent absent absent 1042 6 1119 24.15 0 Negative 15 absent absent absent absent 1043 1 442 22.81 0 Negative 15 absent absent absent absent 1043 2 624 31.97 0 Negative 15 absent absent absent absent 1043 3 1004 66 0 Negative 15 absent absent absent absent 1043 4 813 46.71 0 Negative 15 absent absent absent absent 1043 5 1186 37.77 0 Negative 15 absent absent absent absent 1043 6 1375 39.38 0 Negative 15 absent absent absent absent 1044 1 862 62.53 0 Negative 16 absent absent absent absent 1044 2 1044 66 0 Negative 16 absent absent absent absent 1044 3 1190 40.39 0 Negative 16 absent absent absent absent 1045 1 122 19.86 0 Negative 16 absent absent absent absent 1045 2 269 14.12 0 Negative 16 absent absent absent absent 1045 3 448 14.91 0 Negative 16 absent absent absent absent 1045 4 666 27.65 0 Negative 16 absent absent absent absent 1045 5 836 30.77 0 Negative 16 absent absent absent absent 1045 6 1032 23.83 0 Negative 16 absent absent absent absent 1047 1 698 18.3 0 Negative 16 absent absent absent absent 1047 2 873 15.25 0 Negative 16 absent absent absent absent 1047 3 1062 14.44 0 Negative 16 absent absent absent absent 1047 4 1300 29.89 0 Negative 16 absent absent absent absent 1047 5 1496 9.68 0 Negative 16 absent absent absent absent 1047 6 1706 15.97 0 Negative 16 absent absent absent absent 1048 1 166 19.21 0 Negative 14 absent absent absent absent 1048 2 355 21.9 0 Negative 15 absent absent absent absent 1048 3 579 16.01 0 Negative 14 absent absent absent absent 1048 4 761 15.09 0 Negative 15 absent absent absent absent 1048 5 971 24.91 0 Negative 15 absent absent absent absent 1048 6 1202 33.49 0 Negative 15 absent absent absent absent 1051 1 470 22.73 0 Negative 14 absent absent absent absent 1051 2 652 7.86 2 Positive 13 0.093 0.093 0.094 0.092 1051 3 1078 21.04 14 Positive 14 0.272 0.222 0.573 0.119 1051 4 876 36.72 eleven Positive 14 0.125 0.11 0.321 0.046 1051 5 1246 27.55 14 Positive 14 2,737 2.37 5.617 1.111 1052 1 574 20.03 0 Negative 14 absent absent absent absent 1052 2 756 17.54 0 Negative 14 absent absent absent absent 1052 3 940 16.79 0 Negative 14 absent absent absent absent 1054 1 443 22.25 0 Negative 14 absent absent absent absent 1054 2 1009 22.74 0 Negative 14 absent absent absent absent 1054 3 1184 24.73 0 Negative 14 absent absent absent absent 1054 4 1380 15.71 0 Negative 14 absent absent absent absent 1055 1 476 25.53 0 Negative 14 absent absent absent absent 1055 2 660 19.15 9 Positive 14 0.312 0.256 0.711 0.022 1055 3 856 19.38 9 Positive 14 1.257 1,144 3,267 0.014 1055 4 961 17.58 8 Positive 14 3,006 2,844 5,871 1.69 1069 1 462 10.59 0 Negative 10 absent absent absent absent 1069 2 644 27.48 0 Negative 16 absent absent absent absent 1069 3 826 24.27 0 Negative 16 absent absent absent absent 1069 4 1253 22.57 0 Negative 15 absent absent absent absent 1072 1 340 13.83 0 Negative 13 absent absent absent absent 1072 2 672 15 0 Negative 16 absent absent absent absent 1072 3 490 21.88 0 Negative 16 absent absent absent absent 1072 4 945 39.24 0 Negative 16 absent absent absent absent 1072 5 1126 42.97 2 Positive 16 0.036 0.036 0.059 0.013 1073 1 610 15.91 0 Negative 15 absent absent absent absent 1073 2 743 37.19 0 Negative 15 absent absent absent absent 1073 3 981 25.62 0 Negative 15 absent absent absent absent 1073 4 1177 35.93 0 Negative 15 absent absent absent absent 1074 1 110 15.36 8 Positive 16 0.072 0.059 0.151 0.026 1074 2 306 26.05 12 Positive 16 0.054 0.063 0.105 0.017 1074 3 518 25.02 15 Positive 16 5.81 5,364 11,181 0.08 1074 4 690 35.86 15 Positive 16 16,936 16.54 29,448 0.261 1077 1 1044 16.64 0 Negative 13 absent absent absent absent 1077 3 1407 14.63 0 Negative 13 absent absent absent absent 1077 4 1799 34.01 0 Negative 13 absent absent absent absent 1082 1 835 19.52 0 Negative 16 absent absent absent absent 1082 2 1052 28.22 0 Negative 16 absent absent absent absent 1082 3 1284 30.24 0 Negative 16 absent absent absent absent 1082 4 1427 40.32 0 Negative 16 absent absent absent absent 1082 5 1560 30.03 0 Negative 16 absent absent absent absent 1088 1 546 35.77 12 Positive 16 0.065 0.062 0.154 0.016 1088 3 940 66 12 Positive 16 4.42 3.877 7,651 2,687 1091 1 107 4.91 0 Negative 16 absent absent absent absent 1091 2 703 47.32 16 Positive 16 0.452 0.386 1.08 0.151 1091 3 N.A. 64.15 16 Positive 16 0.136 0.119 0.405 0.016 1092 1 590 25.27 0 Negative 16 absent absent absent absent 1092 2 779 40.63 0 Negative 16 absent absent absent absent 1092 4 1150 37.21 0 Negative 16 absent absent absent absent 1093 1 243 57 0 Negative 16 absent absent absent absent 1093 2 454 34.04 0 Negative 16 absent absent absent absent 1093 3 636 66 0 Negative 16 absent absent absent absent 1093 4 832 35.68 0 Negative 16 absent absent absent absent 1096 1 1258 46.72 3 Positive 16 0.03 0.032 0.047 0.011 1096 2 1426 51.28 5 Positive 16 0.493 0.52 0.805 0.231 1096 3 1615 66 5 Positive 16 20,829 25,272 26,994 10,804 1104 1 183 39.36 0 Negative 15 absent absent absent absent 1104 2 365 22.87 0 Negative 15 absent absent absent absent 1104 3 568 47.98 0 Negative 15 absent absent absent absent 1104 4 745 13.9 0 Negative 15 absent absent absent absent 1107 1 373 18.46 0 Negative 16 absent absent absent absent 1107 2 568 36.06 0 Negative 16 absent absent absent absent 1107 3 729 25.94 0 Negative 16 absent absent absent absent 1111 1 467 66 eleven Positive 16 6,388 6,206 11,209 0.415 2004 1 645 2.98 2 Positive 16 0.247 0.247 0.295 0.199 2004 3 981 18.36 9 Positive 16 6,148 5,801 10.105 3,842 2013 1 505 26.97 0 Negative 16 absent absent absent absent 2013 2 688 4.3 0 Negative 16 absent absent absent absent 2013 3 870 16.33 0 Negative 16 absent absent absent absent 3018 1 170 16.2 3 Positive 16 0.088 0.055 0.192 0.016 3018 2 345 23.3 15 Positive 16 19.35 21,092 25,391 13,484 3018 3 443 66 15 Positive 16 53,703 57,854 64,446 34,164 3019 1 185 3.64 5 Positive 16 0.357 0.351 0.607 0.207 3019 2 696 23.18 14 Positive 16 10,495 10,841 19,652 0.074 3023 1 170 30.52 0 Negative 16 absent absent absent absent 3023 2 345 26.79 10 Positive 16 3.295 2,917 6,581 1,949 3023 3 380 16.26 10 Positive 16 3,608 2,907 6.72 1.855 3048 1 255 30.79 eleven Positive 16 0.157 0.148 0.292 0.039

[932] Example 9: Early detection of metastatic relapse and monitoring of therapeutic efficacy using ultra-deep sequencing of serial plasma free-cell DNA in patients with urothelial carcinoma of the bladder.

[933] INTRODUCTION

[934] Bladder cancer is the most common malignancy of the urinary tract, and approximately 20–25% of patients with newly diagnosed urothelial carcinoma develop muscle-invasive bladder cancer (MIBC), and 10–30% of patients diagnosed with non-MIBC "(NMIBC) will progress to MIBC. Currently, the standard treatment for MIBC is radical cystectomy. Unfortunately, 20% of node-negative patients and 80% of node-positive patients at the time of surgery experience metastatic recurrence, and overall survival (OS) averages 50% at 5 years.

[935] Neoadjuvant chemotherapy (NACT) improves survival in patients with MIBC, and treatment with gemcitabine and cisplatin (GC) is the most commonly used neoadjuvant chemotherapy (NACT) for MIBC. Currently, treatment with gemcitabine and cisplatin results in significant downstaging (pT less than 2 N0 at cystectomy) in approximately 40-50% of patients.

[936] Early detection of metastatic recurrence in patients with bladder cancer may offer new therapeutic approaches to improve survival. Detection of metastatic recurrence after cystectomy at an early stage, when recurrence is undetectable by radiography, can significantly improve the identification of patients who may benefit from early/adjuvant treatment and improve survival outcome for this group of patients. In addition, early detection of recurrence and metastasis may help prevent unnecessary and potentially harmful long-term treatment in patients who do not respond to treatment.

[937] Currently, standard computed tomography (CT) scanning at predetermined intervals is used to detect recurrence, metastasis, and monitor response to treatment. While imaging modalities offer an assessment of tumor burden, monitoring capabilities are limited by suboptimal detection limits and inherent measurement variability, so early detection of metastatic recurrence and/or progression and assessment of treatment response remain a major clinical challenge.

[938] The full potential of using circulating tumor DNA (ctDNA) as a biomarker for disease staging at diagnosis, determination of tumor burden, early detection of metastatic relapse, and therapeutic response to treatment remains unrealized. Recent promising studies have shown that free-cell DNA (scDNA) can be used to monitor the evolution of early-stage lung cancer and subclonal development in metastatic disease (Abbosh, et al., Nature 545, 446-451 (2017) (“Abbosh et al. . 2017”), included in this document in its entirety). In bladder cancer, it has been shown that ctDNA is detectable in plasma and urine, and that high levels of ctDNA are associated with later clinical disease progression and metastasis (Birkenkamp-Demtröderet al., Eur. Urol. 70, 75-82 (2016); Christensenet al., Eur. Urol. 71, 961-969 (2017); Birkenkamp-Demtröder, K. et al. Eur. Urol. 73, 535-540 (2018); Patel, et al., Sci. Rep. 7, 5554 (2017)). However, these previous studies of bladder cancer were based on small selected cohorts and used digital droplet PCR assays, which have relatively limited sensitivity compared with the next-generation sequencing (NGS)-based method presented here.

[939] Here we report the results of a prospective study involving whole exome sequencing (WES) of primary tumors and corresponding germline DNA in 68 patients who received neoadjuvant chemotherapy (n = 56) or first-line chemotherapy (n = 12) before cystectomy. Sensitive, personalized multiplex PCR-based SSP assays specific to each individual patient's tumor mutational signature have been developed and used to monitor somatic mutations in plasma samples obtained over time, before, during, and after chemotherapy. The main goal of this study was to develop a ctDNA detection method allowing the use of ctDNA as a powerful biomarker for prognosis, early detection of metastatic disease, and as a predictor of response to chemotherapy.

[940] METHODS

[941] Patients and clinical specimens

[942] Patients diagnosed with MIBC who received neoadjuvant chemotherapy before cystectomy, as well as patients who received chemotherapy for metastatic disease with or without previous cystectomy, were enrolled between 2013 and 2017 at a single tertiary university hospital (Aarhus University Hospital , Denmark). Radical cystectomy was performed as open cystectomy or robotic-assisted depending on patient criteria and robot availability. All patients underwent extended lymph node dissection to the level of the aortic bifurcation.

[943] Patients were treated according to Danish national guidelines. Neoadjuvant chemotherapy was administered in the form of 4 series of gemcitabine and cisplatin (GC) at three-week intervals. Patients with metastatic or cT4b tumors at diagnosis received up to 6 series of GCs. Pathological downstaging after chemotherapy administered before CE was defined as less than T1N0 after treatment. Cystectomy patients were followed radiographically with pre-therapy PET/CT and CT of the chest and abdomen at planned follow-ups at 4, 12 and 24 months post-cystectomy for patients diagnosed with pT2N0 and additional controls at 8 and 18 months for patients diagnosed more than pT2 and/or N+. Patients treated for advanced disease were followed up at 3-4 month intervals using CT. Detailed follow-up data were available for all patients, and clinical endpoints were last recorded visit or time of death obtained from a national personal registry. Patients were selected for whole exome sequencing based on the following criteria: 1) neoadjuvant chemotherapy/first-line chemotherapy for localized MIBC 2) number of visits with plasma samples collected before and during chemotherapy, before and after CE 3) presence of tumor biopsy. All patients provided written informed consent and the study was approved by the National Medical Research Ethics Committee (No. 1302183).

[944] Sample collection and DNA extraction

[945] We analyzed material from blood, tumor biopsies, and over time collected plasma samples. Tissue biopsies were taken from TUR-B at the time of diagnosis. DNA was extracted as previously described, either from TISSUE-TEK ® OCT Tissue-embedded (Sakura) or formalin-fixed paraffin-embedded (FFPE) sections from a biopsy sample taken from the most representative location with a high percentage of carcinoma cells. 40 ml of EDTA blood was collected at each visit or before each chemotherapy series and processed immediately. Samples were centrifuged at 3000 g for 10 minutes at room temperature, and plasma and buffy coat were stored separately at −80°C. Germline DNA was extracted from buffy coat leukocytes and the concentration was measured by the QUBIT ® fluorometric quantitation assay. Plasma was stored at minus 80 °C. This study used up to 9 mL of plasma per case (range 4–9 mL; mean X mL). The entire volume of plasma was used for scDNA extraction using the QIAAMP ® Circulating Nucleic Acid kit (Qiagen) and eluted in 50 μl of DNA suspension buffer (Sigma). Each scDNA sample was quantified using the QUANT-IT ® High Sensitivity dsDNA Assay Kit (Invitrogen).

[946] Exome sequencing and bioinformatics analysis

[947] Libraries of tumor DNA and matching germline DNA were prepared using 100-500 ng of DNA and captured using the SEQCAPEZ ® MedExomeV1_hg19 or MedExomePlusV1_hg19 panel (Roche). Exome sequencing metrics, percentage of carcinoma cells, and tissue type are shown.

[948] All variants that passed the applied filters were analyzed for the activity of mutation signatures. Variants were initially loaded into a VRanges object and then sequence context was extracted using the SomaticSignatures R package (Obenchainet al., Bioinformatics 30, 2076-2078 (2014); Gehring et al., Bioinformatics 31, 3673-3675 (2015)). De novo mutation signature extraction was not used due to cohort size. Instead, the mutational profiles identified in our samples were projected onto known COSMIC signatures using the MutationalPatterns R package (see cancer.sanger.ac.uk/cosmic/signatures and Blokzijl et al., Genome Med . 10, 33 (2018)) . Mutation signatures 1, 2, 5, and 13 identified in Robertson et al., Cell 171, 540-556.e25 (2017) for bladder cancer were prioritized for analyses.

[949] Mutations associated with DNA damage response

[950] Selected DNA damage response genes have been described in Teo et al., Clin. Cancer Res . 3610-3618 (2017). Mutations identified in these genes were analyzed to determine whether they were deleterious or benign. All loss-of-function mutations were considered damaging. Missense mutations were further analyzed using PolyPhen2 and MutationAssessor as described in Reva et al., Nucleic Acids Res. 39, e118 (2011) and Adzhubei et al., Nature Methods 7, 248-249 (2010). Variants identified as likely damaging/possibly damaging or medium/high, respectively, in PolyPhen2 and MutationAssessor were considered damaging.

[951] Free Cell DNA Library Preparation

[952] Preparation of cell-free DNA (cfDNA) was previously described as input for library preparation using up to 66 ng (20,000 genome equivalents) of cell-free DNA (cfDNA) from each plasma sample. The scDNA was subjected to end repair, A-tail extension, and ligation using special adapters. The purified ligation product was amplified for 20 cycles and purified using AMPURE ® XP microbeads (Agencourt/Beckman Coulter).

[953] Next generation sequencing (NGS) workflow for plasma in multiplex PCR

[954] An aliquot of each library was used as input for a patient-related 16-plex PCR reaction. Samples were amplified using the tumor-specific SIGNATERA ® assay and barcoded and pooled. Sequencing was performed on an Illumina HISEQ ® 2500 Rapid Run with 50 paired-end read cycles using the Illumina PAIRED END ® v2 kit with an average read depth of more than 100,000X per amplicon, as described in Abbosh et al., Phylogenetic ctDNA analysis depicts early-stage lung cancer evolution , Nature 545, 446-451 (2017).

[955] Recognition of plasma variants

[956] To build the background error model, a large set of negative control samples (~1000) were preprocessed. For each target variant, using the read depth of mutant and reference alleles, a confidence score was calculated based on the error model as described in Abbosh et al. 2017. A ctDNA positive plasma sample was defined as having at least 2 variants with a confidence score above the predefined algorithm threshold (0.97) as described by Abbosh et al. 2017.

[957] Whole exome sequencing of plasma

[958] An aliquot of each plasma library was barcoded and captured using the SEQCAP ® EZ MedExome Target Enrichment Kit. Sequencing was performed on an Illumina HISEQ ® 2500 with 200 cycles of single-end reads using the Illumina TRUESEQ ® v1 kit. Data were demultiplexed, adapter trimmed, and mapping was performed using the Burrows-Wheeler Alignment tool (BWA-mem) using hg19 as the reference sequence. For sequencing data manipulation, duplicates were labeled using the Picard tool. Bam files generated from this mapping have been processed according to GATK best practices (link). Variants were recognized using MuTect2, and variants passed the built-in filters were used to determine genomic positions, with modifications for each patient. Recognized variants in both exomes for each patient were subsequently used to manually analyze all included positions using BAM-readcount (link). Only bases and reads with a quality greater than 20 were included, and positions with fewer than 10 reads in both exomes were not included in the comparison.

[959] RNA sequencing, data processing and analysis

[960] RNA sequencing was performed using QUANTSEQ ® 3' mRNA-Seq Library Prep (Lexogen) with RNA input ranging from 50-250 ng. Libraries were prepared according to the manufacturer's recommendations. Sequencing was performed with a single read of 70 bp. on the Illumina NEXTSEQ ® 500 platform. Sequence reads were aligned to the GRCh38 transcriptome (cDNA + ncRNA) using Salmon (Patro et al., Nature Methods , 14, 417-419 (2017)) without gene length correction, and TPM gene expression data were normalized using edgeR (Robinson and Oshlack, Genome Biology , 11,R25 (2010)). Samples were classified according to consensus MIBC subtypes (manuscript in preparation).

[961] Statistical analyzes

[962] Survival analysis was performed in R statistics using the Survminer and Survival packages. Statistical significance was assessed using the Wilcoxon rank sum test for continuous variables and the Fisher exact test for categorical variables.

[963] RESULTS

[964] Patient characteristics and primary tumor analysis

[965] We included patients with localized MIBC who received chemotherapy before cystectomy between 2014 and 2017 at Aarhus University Hospital, Denmark (FIG. 70). A total of 68 patients met all inclusion criteria (see FIG. 70, FIG. 71A-G and Table IA below).

Table IA: Patient characteristics and demographics.

Patients, N = 68 Age ^a (years), median (range) = 65.2 (43-79) Gender, n (%) Tumor stage in CE, ^b n (%) Women 12 (17.6) T0/Cis/Ta/T1 47 (69.1) Men 56 (82.4) T2/T3/T4a 19 (27.9) Tumor stage at TUR-MP, n (%) Metastases, ^With n (%) T1/T2
T4a/b 58 (85.3)
10 (14.7) Local recurrence (pelvis, rectum, urethra) 6 (17.6) Distant metastases (bones, lungs, liver, skin) 8 (11.8) Stage N before treatment, n (%) Clinical observation, days, (range) Without signs of disease (CE; n=38) 498 (103-991) N0 58 (85.3) Clinical relapse (CE; n=10) 321 (57-1119) N1/N2 10 (14.7) Progression (CE ineffective; n=2) 263 ^a When collecting samples; ^b n=66; ^with n=14; CE, cystectomy

[966] Whole exome sequencing (WES) of tumor and corresponding germline DNA was performed at an average target coverage of 104X (31X-251X) for tumor samples and 66X (35X-120X) for germline samples, identifying an average of 488 (11-3536) mutations per patient. In addition, RNA sequencing of 46 tumors was performed to determine bladder cancer subtypes, immune signatures, and cellular composition. A summary of the molecular characteristics and clinical data for all patients is shown in FIG. 71A-G. The clinical protocol design and sampling schedule for this study are shown in FIG. 72.

[967] Monitoring ctDNA using next generation sequencing (NGS) based on ultra-deep multiplex PCR.

[968] A customized SSP multiplex PCR approach was used to detect ctDNA. Somatic SNVs and short INSERT/DELETIONS were prioritized based on whole exome sequencing (WES) data based on observed variant allele frequency (VAF) in tissue and sequence context. Unique patient-specific assays were developed and synthesized for sixteen high-ranking somatic mutations, as indicated in FIG. 73. SSP in multiplex PCR was performed on plasma cDNA. A sample was considered ctDNA positive only if at least two target variants were detected based on a previously developed recognition algorithm presented in Abbosh et.al 2017. Sample-level analytical sensitivity (when 2 or more variants out of 16 are detected) was defined as more than 95% with variant allele frequencies of 0.01%.

[969] Using this approach, ctDNA status was analyzed in 618 plasma samples from 68 patients included in the study (FIG. 70). Quality control was carried out throughout the entire work process. Samples and amplicons that failed quality control were excluded from further analysis. For each patient, a set of 45 SNPs were genotyped by whole exome sequencing (WES) and plasma sequencing to ensure sample matching. The median target coverage was 120,000X.

[970] Detection of ctDNA for prognosis and detection of relapse

[971] Throughout the course of the disease, the presence or absence of ctDNA was highly correlated with disease outcome (FIG. 74, FIG. 75A-C). Of significant interest was the ctDNA status at the following three time points. The first time point of interest is ctDNA status before NACT administration, and this time point has been found to be highly predictive of outcome. For bladder cancer, the first intervention is TURB (transurethral resection of bladder tumor), and therefore this first time point has been found to serve as a proxy for measuring minimal residual disease. Strikingly, 94% (34/36) of patients with a negative ctDNA result at this first time point did not relapse throughout the study. In contrast, 44% (11/25) of patients who were ctDNA positive at this point before NACT had a recurrence after cystectomy. Thus, detection of ctDNA at this early time point is a very strong prognostic factor for long-term clinical outcome after NACT and cystectomy (CE).

[972] The second time point was after NACT and before cystectomy, ctDNA status at this time point was also predictive of patient outcome. Among the ctDNA-negative patients, there were only 7% (4/55; 4 were post-cystectomy (CE) positive and had recurrence) patients who relapsed compared with 70% (7/10) of ctDNA positive patients who had a relapse. Pre-cystectomy (CE) ctDNA status was associated with CE pathology because 100% of ctDNA-positive patients at this time point had residual T2+ tumor and/or lymph node metastases identified at cystectomy (FIG. 75B).

[973] The third time point was after cystectomy. Stratification of patients by ctDNA status after CE showed a significantly worse outcome for ctDNA positive patients (FIG. 75C). The majority, 96% of patients with a negative ctDNA result (50/52) did not relapse, but 2/52 patients relapsed more than 1.5 years after CE (ctDNA testing was not performed closer to relapse). In contrast, 92% (12/13) of ctDNA-positive patients relapsed. Notably, one patient died before clinical evaluation (FIG. 75D). Post-CE ctDNA status was found to be highly predictive of disease recurrence and a stronger prognostic factor than any other prognostic factor such as pre-cystectomy N stage and response to chemotherapy (FIG. 75E).

[974] Serial ctDNA measurements for disease surveillance

[975] It is disclosed here that serial ctDNA measurement can be used to both monitor therapeutic response and detect relapse. In our study, to evaluate the value of ctDNA in a surveillance setting, plasma was collected after cystectomy, at serial time points when patients were disease-free. When we included analysis of serial ctDNA measurements throughout the course of the disease, we observed a sensitivity of 92% with 100% specificity (FIG. 76). On average, ctDNA detection was observed 96 days (0–245 days) before radiographic detection. For example, patient 4265 had a drop in ctDNA during NACT, and then ctDNA was detected 138 days after CE; whereas clinical relapse was detected after 186 days (324 days after CE). Similarly, patient 4189 tested positive for ctDNA before CE and was subsequently negative, and then 273 days after CE, ctDNA was detected again; however, clinical relapse was detected on day 369 or 96 days later (see FIG. 76). For 12 patients with metastatic relapse, we found that ctDNA analysis had a median lead time of 103 days (range; p=0.019) compared with conventional imaging (FIG. 77). The lead time may be biased by analysis of more frequent plasma sampling compared with imaging. Restricting our analysis to patients with simultaneous plasma and radiographic studies identified eight patients, five of whom showed advanced time to relapse detection for ctDNA analyses. The remaining three patients showed concurrent recurrence (FIG. 77), and the mean lead time for all eight patients was 106 days.

[976] Serial sampling of ctDNA measurements to monitor response to therapy

[977] Bladder cancer is treated with NACT, but clinically useful predictive biomarkers of treatment response have previously been and are currently unavailable, and downstaging in CE is used as an indicator of treatment response (ref). In our series, disease relapse was significantly associated with expected response to chemotherapy (FIG. 78A), but only 44% (x/y) of non-responder patients experienced disease relapse, indicating that downstaging was a suboptimal marker for evaluation effectiveness of treatment. Here, we found that ctDNA measurements over time during NACT showed very different distributions between responders and nonresponders (p = xx; FIG. 78F-G). A total of 83% (34/41) of ctDNA-negative patients responded to chemotherapy, and 53% (9/17) of patients who had ctDNA clearance from initially positive tests responded to therapy, indicating that that ctDNA levels may serve as better indicators of treatment efficacy during and after treatment. Patients with positive ctDNA after NACT did not respond to chemotherapy. Overall, ctDNA levels reflected the disease characteristics observed in the cohort (FIG. 78G).

[978] Analysis of the molecular characteristics of treatment-naive tumors showed a significantly higher contribution of mutational signature 5 (p = 0.01) in patients responding to chemotherapy (FIG. 78B). The high contribution of signature 5 was significantly associated with ERCC2 mutation status (FIG. 80), indicating a correlation with DNA damage response (DDR) mechanisms, as previously reported. ERCC2 mutations in responding patients, however, were not significantly more common (FIG. 78C). Overall, DDR mutation status was not a predictive biomarker of response to chemotherapy (FIG. 78D). By stratifying tumors by molecular subtype (FIG. 78E), we found a significant correlation with survival (FIG. 78F), but these subtypes did not predict response to NACT (FIG. 78G). Tumors classified as "infiltrated" showed a higher response rate to chemotherapy compared with other subtypes (FIG. 78G). This contrasts with earlier reports, which emphasized that basal-like tumors were most associated with response to NACT treatment.

[979] Whole exome sequencing of plasma ctDNA from patients with metastatic disease

[980] We performed whole exome sequencing (WES) of scDNA from plasma samples with ctDNA targets measured at allele frequencies of 10% or higher in multiplex PCR SSP assays. Four samples from three patients were sequenced to an average target coverage of 307X (272X-340X), and 508-1294 mutations were identified. We compared all mutations identified in the plasma RPE data with the corresponding primary tumor RPE data to evaluate mutational changes acquired during metastatic evolution (FIG. 79A-D). We found high similarity between the mutational landscapes of primary tumors and metastatic lesions, indicating limited clonal evolution during the disease course of the selected patients. On average, we identified 62 ctDNA mutations present at the time of metastasis that were not detected in primary tumors. Interestingly, we identified two CYP2C19 mutations in the plasma of patient 4119, both affecting codon 214. The resulting amino acid is in a channel involved in directing compounds to the active site of the protein. These mutations may have occurred during chemotherapy and may explain the lack of pathological downstaging observed in this patient.

[981] Discussion

[982] This report describes a reliable and reproducible method for early detection and improved treatment of metastatic disease based on detection and monitoring of ctDNA before, during and after treatment. In patients undergoing radical surgery, detection of ctDNA provides direct evidence of occult cancer cells and therefore residual disease.

[983] Interestingly, in this case it was found that detection of ctDNA often preceded detection of metastases by imaging modalities. Specifically, we found that all patients who were ultimately diagnosed with metastatic bladder cancer were ctDNA positive after cystectomy, with a median lead time of 103 days compared with imaging modalities. Thus, the ctDNA analysis method presented here offers a unique opportunity to initiate treatment of metastatic relapse at an earlier point in time. Importantly, initiating treatment for small amounts of metastatic disease may improve response rates and positively impact survival.

[984] This study also describes a reliable and reproducible method for identifying patients at low risk of metastatic recurrence. It was found that all patients who did not have disease after cystectomy were ctDNA negative (100% specificity). The 100% specificity of the post-cystectomy test can be used to identify patients at low risk of metastatic recurrence and therefore reduce the need for continuous surveillance with costly radiographic imaging and associated patient anxiety. Accordingly, the present description provides a better association between ctDNA status and outcome compared with conventional methods, thereby bringing ctDNA analysis closer to clinical relevance.

[985] The present invention also resulted from the discovery that ctDNA dynamics can help identify patients who respond to NACT already during treatment. Patients with MIBC received NACT to reduce the primary tumor burden and potentially eliminate micrometastases before cystectomy, and most patients experienced a reduction in pathological stage after NACT. No reduction in pathological stage, i.e. residual primary tumor or lymph node infiltration at the time of cystectomy, or previous lymph node infiltration, are risk factors associated with disease recurrence after cystectomy. In our study, ctDNA was detected in 37% (25/68) of patients before or during NACT, and 53% of patients with ctDNA reduction below our detection threshold had a reduction in pathological stage. It is important to note that none of the patients with persistently detectable ctDNA experienced a decrease in pathological stage. However, we observed a subgroup of patients with pathological downstaging who experienced disease relapse and vice versa. These results indicate that although clinical and histopathological parameters serve as prognostic risk factors, risk stratification of patients based on pathology is less than ideal.

[986] The present invention also proposes that ctDNA status before and during chemotherapy may be a powerful new clinical risk factor and may potentially aid in the selection of patients with early metastatic spread who may benefit from long-term regimens of NACT and enhanced surveillance. It was demonstrated that none of the patients who were ctDNA negative before or during chemotherapy had disease recurrence after cystectomy, while 44% (11/25) of patients who were ctDNA positive before or during chemotherapy, there was a relapse of the disease. Thus, the presence of ctDNA even at an early stage may indicate metastatic spread and facilitate higher risk stratification of patients compared with currently available risk factors for this group of patients, which, to our knowledge, have not previously been demonstrated at a similar level of pattern. A subset of patients with initially detectable ctDNA but no evidence of possible disease relapse may represent cases in which chemotherapy or cystectomy has effectively eradicated the disease. Therefore, patients without evidence of early metastatic spread (ctDNA negative) and pathological downstaging after NACT may be eligible for bladder-sparing treatment.

[987] Notably, for many patients, ctDNA frequencies were quite low (few mutated copies), and CSP-based ultra-deep sequencing approaches are required to reliably detect these rare variants. Selection of clonal mutations based on primary tumor PES allows for ultra-deep sequencing of patient-specific mutations in plasma ctDNA. Earlier work has shown genetic heterogeneity between primary tumors and metastases, indicating that gene panels of frequently mutated genes may be required. When using PES scDNA from plasma, we still observed heterogeneity between primary tumors and metastases, but importantly, all clonal mutations selected from primary tumors were found in metastases.

[988] In conclusion, the invention presented herein has shown that determination of ctDNA levels in patients with bladder cancer is a highly predictive marker of metastatic recurrence and response to treatment. In particular, the inventors of the present invention have provided herein accurate and reliable methods for monitoring a patient's response to cancer treatment to reassure the patient that the disease is under control and detecting recurrence significantly earlier than conventional methods can improve survival outcomes.

[989] Example 10. Dynamic analysis of plasma free-cell DNA by ultra-deep sequencing in patients with stage I-III colorectal cancer.

[990] The purpose of this case study was to demonstrate that postoperative analysis of circulating tumor DNA (ctDNA) dynamics can identify and monitor residual tumor burden in asymptomatic patients. In particular, this example showed that ctDNA analysis makes it possible to personalize and stratify the risk of postoperative treatment of patients with stage I-III colorectal cancer.

[991] INTRODUCTION

[992] Colorectal cancer (CRC) is the third most common cancer in the world and the second leading cause of cancer-related deaths, with 1.3 million newly diagnosed cases annually. Despite improvements in surgical procedures, screening, and advances in treatment regimens, the 5-year mortality rate for patients with CRC remains high at approximately 40%, representing a significant healthcare burden worldwide.

[993] The current standard of care for patients with CRC involves surgical resection of the tumor followed by adjuvant chemotherapy (ACT) in selected patients. Most stage II patients do not receive chemotherapy, but approximately 10-15% have residual disease after surgery. If they could be identified, treatment with ACT could potentially reduce the risk of relapse. On the contrary, most stage III patients receive chemotherapy. Despite this, more than 50% are already cured by surgery. In addition, approximately 30% of patients treated with stage III ACT experience relapse, making them candidates for additional therapy. Therefore, improved tools are urgently needed to identify the patient population that will benefit from ACT.

[994] Early diagnosis of recurrent disease is another significant clinical unmet need in CRC. After completion of definitive treatment, surveillance for recurrence is recommended to detect recurrence early enough for potentially curative surgery. Despite surveillance, many relapses are detected late, and only 10-20% of metachronous metastases are treated with curative intent. Therefore, there is a need for better biomarkers that can earlier identify patients at high risk of relapse, allowing appropriate follow-up and treatment strategies to improve patient survival.

[995] METHODS

[996] Patients

[997] Stage I to III CRC patients were recruited from the surgical departments of Aarhus University Hospital, Randers Hospital and Herning Hospital between 2014 and 2018. Tumor tissue was collected during surgery. Blood samples were collected preoperatively (14 days before surgery) and postoperatively on day 30 (with the option of collecting a sample 14 days before or after), and then every three months until death, patient withdrawal from the study, or monthly 36, whichever comes first. Seventy-five patients provided serial blood samples (ranging from 3 to 14 samples per patient), whereas the remaining 50 patients provided only two blood samples (preoperative and postoperative at day 30). Patient characteristics and demographics are summarized below in Table 12.1.

Table 12.1: Characteristics and demographics of eligible patients. Patients, n 125 Cancers, n 129* Age (years), median (range) 69.9, (43.3-91) Gender, n (%) Women 52 (41.6) Men 73 (58.4) Follow-up imaging (months), median (range) 12.5, 1.4-38.5 Localization, n (%) Colon 119 (95.2) Rectum 6 (4.8) Pathological stage according to UICC, n (%) I 5 (4) II 39 (31.2) III 81 (64.8) Histological type, n (%) Adenocarcinoma 115 (92) Mucinous carcinoma 10 (8.7) Histological grade of malignancy, n (%) Moderately differentiated 96 (76.8) Poorly differentiated 19 (15.2) BUT 10 (8) Adjuvant therapy by UICC stage, n (%) I 0 (0.0) II 6 (15.3) III 71 (87.7) Total 77 (61.6) Relapse by UICC stage, n (%) I 0 (0.0) II 4 (10.2) III 20 (24.7) IV 0 (0.0) Total 24 (19.2) MSS/MSI status, n (%) M.S.S. 109 (87.2) MSI 20 (16) Smoking, n (%) Never 52 (41.6) Previously 57 (45.6) Now 16 (12.8) *Four patients with synchronous cancers

[998] Information on postoperative clinical intervention and other clinicopathological information was collected for all patients, as shown in Table 12.2 below. All patients in the study underwent resection of the primary tumor.

Table 12.2 Clinicopathological information. IN
patient Age Floor Localization
primary tumor Diameter
primary
tumors, mm Stage
By
UICC TNMV Adjuvant
therapy 1 68 M sigmoid colon 52 II T3N0MxV0 No 2 70 M sigmoid colon 54 II T3N0MxV0 No 3 67 AND sigmoid colon 32 II T3N0MxV0 No 4 64 M right bend 35 III T3N1MxV0 Yes 5 77 M sigmoid colon 70 II T3N0MxV0 No 6 75 M cecum 65 II T3N0MxV0 No 7 70 M ascending colon 29 II T3N0M0V0 No 8 65 AND sigmoid colon 42 III T3N1M0V0 Yes 9 50 M sigmoid colon 100 II T3N0M0V0 Yes 10 50 AND sigmoid colon 15 III T3N1M0V0 Yes eleven 70 M right bend 40 III T3N2M0V1 Yes 12 67 AND transverse colon 50 II T3N0M0V0 No 13 66 AND sigmoid colon 40 II T4N0MxV0 Yes 14 68 AND cecum 63 II T3N0MxV0 No 15 48 M sigmoid colon 42 III T3N2MxV0 Yes 16 68 AND sigmoid colon 50 II T3N0MxV0 No 18 67 M cecum 45 III T2N2MxV0 Yes 19 69 AND cecum 58 II T3N0M0V0 No 20 73 AND middle part of the rectum,
5-10 cm 55 III T3N1M0V2 Yes 21 68 M sigmoid colon 55 III T3N1M0V0 Yes 22 67 AND upper part of the rectum
10-15 cm 60 III T3N1M0V1 Yes 23 69 M upper part of the rectum
10-15 cm 53 III T4N1M0V1 Yes 24 82 M right bend 77 III T3N1M0V1 Yes 25 72 AND right bend 95 II T3N0M0V0 No 26 66 M sigmoid colon 82 III T4N1M0V0 Yes 27 68 M upper part of the rectum
10-15 cm 37 III T3N1M0V0 Yes 28 46 AND sigmoid colon 66 III T4N2M0V1 Yes 29 65 AND upper part of the rectum
10-15 cm 37 III T3N1M0V1 Yes thirty 52 AND sigmoid colon 47 III T3N2M0V2 Yes 31 54 M sigmoid colon 25 III T3N1M0V0 Yes 33 47 M transverse colon (1) and
ascending colon (2) 170 (1) and
40 (2) III T3N1M0V1 (1) and
T3N1M0V1 (2) Yes 34 69 M left bend 37 II T3N0M0V1 Yes 35 68 M ascending colon 38 III T3N1M0V0 Yes 36 70 M sigmoid colon 47 III T4N2M0V1 Yes 37 75 M sigmoid colon 46 II T3N0M0V0 No 38 49 AND ascending colon 41 II T3N0M0V0 No 39 74 M cecum 115 III T4N2M0V1 Yes 40 69 M cecum 85 III T3N2M0V1 Yes 41 70 M sigmoid colon 67 II T3N0M0V0 No 42 60 M left bend 73 III T4N2MxV1 Yes 43 48 M upper part of the rectum
10-15 cm 51 II T3N0MxV0 No 44 76 AND sigmoid colon 70 II T4N0M0V0 Yes 45 83 AND transverse colon 40 II T3N0MxV0 No 46 72 AND ascending colon 70 III T4N2MxV1 Yes 47 73 AND cecum 7 II T3N0MxV0 No 48 91 AND ascending colon (1)
and transverse colon (2) 55 (1) and
13 (2) III T3N1M0V1 (1) and
TxN1M0Vx (2) No 49 80 AND ascending colon 20 III T4N1M0V0 Yes 50 64 M sigmoid colon 83 II T4N0M0V2 Yes 51 60 M sigmoid colon 70 II T3N0M0V0 No 52 81 M sigmoid colon thirty II T3N0M0V0 No 53 80 AND left bend 32 III T3N1M0V0 No 54 66 M sigmoid colon 40 III T3N1M0V0 No 55 78 M sigmoid colon 40 II T3N0M0V0 No 57 50 M ascending colon 45 II T3N0M0V0 No 58 77 M ascending colon 52 II T4N0M0V0 No 59 51 AND sigmoid colon 20 I T2N0M0V0 No 60 70 AND sigmoid colon 32 II T3N0M0V0 No 61 67 M sigmoid colon 24 II T3N0M0V0 No 62 55 M descending colon thirty III T4N1M0V1 Yes 63 82 M right bend 40 III T3N1MxV0 No 64 71 AND sigmoid colon 75 III T3N1MxV1 Yes 65 72 M right bend 55 III T4N1MxV1 Yes 66 72 M cecum 90 III T3N2MxV0 Yes 67 71 M sigmoid colon 35 III T4N1MxV0 Yes 68 83 AND sigmoid colon 75 III T3N2MxV1 No 69 56 AND sigmoid colon 25 III T3N1MxV1 Yes 70 66 M sigmoid colon 20 III T3N2MxV0 Yes 71 77 M cecum 90 III T3N1MxV0 Yes 72 49 AND cecum 41 III T3N1MxV0 Yes 73 43 AND ascending colon 64 III T4N2MxV1 Yes 74 72 M descending colon 18 III T3N2MxV0 Yes 75 48 AND ascending colon 74 III T3N1MxV0 Yes 76 75 M sigmoid colon 73 III T3N2MxV0 Yes 77 71 AND cecum 52 III T3N1MxV1 Yes 78 64 AND descending colon 50 III T4N2MxV0 Yes 79 50 AND left bend 40 III T3N1MxV0 Yes 80 73 AND cecum 82 III T4N1MxV0 Yes 81 58 AND sigmoid colon 23 III T3N2MxV0 Yes 82 50 AND sigmoid colon thirty III T3N2M0V1 Yes 83 62 AND sigmoid colon 70 III T3N2MxV0 Yes 84 65 AND descending colon 100 III T3N1MxV2 Yes 85 61 AND transverse colon 35 III T3N2MxV0 Yes 86 79 M cecum 90 III T3N2MxV1 Yes 87 70 M sigmoid colon 145 III T3N1MxV2 Yes 88 64 M sigmoid colon 60 III T3N2MxV1 Yes 89 75 M ascending colon 40 III T3N2MxV0 No 90 87 AND left bend (1) and
left bend (2) 62 (1) and
N/A (2) III T3N1MxV0 (1) and
T3N1M0V0 (2) No 91 69 M sigmoid colon 70 III T3N1MxV0 Yes 92 73 AND cecum 100 III T4N2MxV0 Yes 93 81 M ascending colon 80 III T4N2MxV1 Yes 95 70 M sigmoid colon thirty I T2N0MxV0 No 96 64 M left bend 40 III T4N2M0V1 Yes 97 67 AND transverse colon 60 II T4N0MxV0 Yes 98 78 M transverse colon thirty II T3N0MxV0 No 99 79 M ascending colon 40 III T4N2MxV0 No 100 59 AND cecum 100 II T4N0MxV1 No 101 72 M sigmoid colon 40 II T3N0MxV2 No 102 69 M sigmoid colon 35 I T2N0MxV0 No 103 70 M cecum 100 III T3N1MxV1 Yes 104 50 M sigmoid colon thirty III T3N1MxV2 Yes 105 72 M right bend 70 II T3N0MxV1 No 106 70 AND ascending colon 25 I T2N0MxV0 No 107 75 M descending colon thirty III T3N2MxV0 Yes 108 72 AND cecum 61 II T3N0MxV0 No 109 74 M sigmoid colon 50 II T3N0MxV0 No 110 67 M sigmoid colon 35 I T2N0MxV0 No 111 52 M descending colon 45 III T3N2M0V1 Yes 112 61 M cecum 40 III T3N1MxV1 Yes 113 78 M sigmoid colon 70 III T3N1MxV1 No 114 79 M cecum (1) and
ascending colon (2) 60 (1) and
35 (2) III T3N1MxV0 (1) and
T3N1MxV0 (2) Yes 115 78 M cecum 35 III T3N2M0Vx Yes 116 75 AND sigmoid colon 70 III T4N1MxV0 Yes 117 84 AND cecum 56 III T3N1MxV0 Yes 118 69 M transverse colon 80 III T4N2MxV0 Yes 119 69 AND right bend 60 III T4N1MxV1 Yes 120 61 AND transverse colon 50 III T3N2MxV1 Yes 121 81 AND cecum 20 III T3N1MxV0 No 122 76 M ascending colon 50 III T3N1MxV1 Yes 123 59 M left bend 50 III T4N1MxV0 Yes 124 71 AND cecum 47 III T3N2MxV2 Yes 125 78 AND cecum 40 II T4N0MxV0 No 126 71 M sigmoid colon 60 II T3N0MxV0 No 127 67 M ascending colon thirty II T3N0MxV0 No 128 51 M descending colon 8 III T4N1M0V0 Yes 130 66 M ascending colon 60 III T4N2M0V0 Yes

Table 12.3 Clinicopathological information, continued. IN
patient First relapse
after operation,
month Localization
Relapse Treatment of relapse MSS/MSI Perineural invasion Tumor perforation 1 0 M.S.S. No No 2 0 M.S.S. Yes No 3 0 M.S.S. No No 4 0 M.S.S. No No 5 0 M.S.S. No No 6 0 M.S.S. No No 7 0 M.S.S. No No 8 0 M.S.S. No No 9 0 M.S.S. No No 10 0 M.S.S. No No eleven 0 M.S.S. Yes No 12 0 M.S.S. No No 13 0 M.S.S. Yes No 14 0 MSI Yes No 15 0 M.S.S. Yes No 16 0 M.S.S. Yes No 18 12 Liver Partial hepatectomy M.S.S. Yes No 19 0 MSI Yes No 20 12 Lung Partial pulmonary lobectomy M.S.S. Yes No 21 0 M.S.S. Yes No 22 0 M.S.S. Yes No 23 0 M.S.S. Yes No 24 13 Liver RFA of the liver M.S.S. Yes No 25 0 MSI Yes No 26 0 M.S.S. Yes No 27 0 M.S.S. Yes No 28 31 Multiple 5FU and Oxaliplatin and
folic acid M.S.S. Yes No 29 12 Lung 5FU and Irinotecan and
folic acid M.S.S. Yes No thirty 9 Bone 5FU and Oxaliplatin and
folic acid M.S.S. Yes No 31 0 M.S.S. Yes No 33 0 MSI N/A No 34 31 Liver Palliative M.S.S. Yes No 35 0 M.S.S. Yes No 36 0 M.S.S. Yes Yes 37 6 Liver RFA of the liver M.S.S. Yes No 38 0 MSI Yes No 39 0 M.S.S. Yes No 40 0 M.S.S. Yes No 41 0 M.S.S. Yes No 42 14 Carcinosis 5FU and Irinotecan and
folic acid M.S.S. Yes No 43 0 M.S.S. No No 44 0 M.S.S. No No 45 0 MSI No No 46 0 MSI No No 47 0 M.S.S. No No 48 0 MSI No 49 0 M.S.S. Yes Yes 50 0 M.S.S. Yes No 51 0 M.S.S. Yes No 52 0 M.S.S. No No 53 0 M.S.S. Yes No 54 0 M.S.S. Yes No 55 0 M.S.S. Yes No 57 0 M.S.S. Yes No 58 0 MSI Yes No 59 0 M.S.S. No No 60 0 M.S.S. Yes No 61 0 M.S.S. Yes No 62 0 M.S.S. No No 63 0 MSI No No 64 0 M.S.S. No No 65 0 M.S.S. Yes No 66 0 M.S.S. No No 67 0 M.S.S. No No 68 13 Carcinosis 5-FU M.S.S. No No 69 0 M.S.S. No No 70 0 M.S.S. No No 71 0 MSI No No 72 0 M.S.S. No No 73 0 M.S.S. No No 74 0 M.S.S. No No 75 eleven Lung Partial hepatectomy M.S.S. No No 76 0 M.S.S. No No 77 13 Lung Partial pulmonary lobectomy M.S.S. Yes No 78 0 M.S.S. No No 79 15 Liver Operation and
RFA of the liver M.S.S. No No 80 0 M.S.S. No No 81 0 M.S.S. No No 82 eleven Lymph nodes Lymph node (throat) surgery M.S.S. Yes No 83 0 M.S.S. Yes No 84 0 M.S.S. No No 85 eleven Liver and lung No M.S.S. No No 86 0 M.S.S. No No 87 0 M.S.S. No No 88 0 M.S.S. No No 89 17 Lung Partial pulmonary lobectomy M.S.S. No No 90 0 M.S.S. No 91 0 M.S.S. No No 92 14 Local Operation of anastomosis, small intestine and mesentery MSI No No 93 0 M.S.S. No No 95 0 M.S.S. No No 96 0 M.S.S. Yes No 97 0 M.S.S. No No 98 0 MSI No No 99 15 Multiple Palliative M.S.S. No No 100 0 M.S.S. Yes No 101 0 M.S.S. No No 102 0 M.S.S. No No 103 12 Multiple No M.S.S. No No 104 12 Liver RFA of the liver M.S.S. No No 105 0 M.S.S. No No 106 0 MSI No No 107 0 M.S.S. No No 108 12 Liver RFA of the liver M.S.S. No No 109 0 M.S.S. Yes No 110 0 M.S.S. No No 111 0 M.S.S. Yes No 112 0 M.S.S. Yes No 113 0 M.S.S. No No 114 0 M.S.S. No 115 0 M.S.S. No No 116 0 M.S.S. No No 117 0 M.S.S. No No 118 0 MSI No No 119 12 Local Under consideration M.S.S. Yes Yes 120 0 MSI No No 121 0 M.S.S. No No 122 0 M.S.S. Yes No 123 0 M.S.S. Yes No 124 12 Multiple Palliative M.S.S. No No 125 12 Lung Palliative M.S.S. No No 126 0 M.S.S. No No 127 0 MSI No No 128 0 MSI Yes N/A 130 0 M.S.S. Yes No

Table 12.4 Clinicopathological information, continued. Patient ID Number of lymph nodes in the removed sample Macroscope.
radical resection Microscope.
radical resection Radical
resection Ileus Anastomotic leak Perforation Perforation
tumors WHO status 1 41 Yes Yes Yes Ileus absent. No anastomotic leak. No No 1 2 31 Yes Yes Yes Ileus absent. No anastomotic leak. No No 1 3 24 Yes Yes Yes Ileus absent. No anastomotic leak. No No 1 4 39 Yes Yes Yes N/A No anastomotic leak. No No 1 5 34 Yes Yes Yes Ileus absent. No anastomotic leak. No No 1 6 23 Yes Yes Yes Ileus absent. No anastomotic leak. No No 1 7 28 Yes Yes Yes Ileus absent. No anastomotic leak. No No 1 8 32 Yes Yes Yes N/A No anastomotic leak. No No 1 9 51 Yes Yes Yes Ileus absent. N/A No No 1 10 27 Yes Yes Yes N/A No anastomotic leak. No No 1 eleven 27 Yes Yes Yes N/A No anastomotic leak. No No 1 12 41 Yes Yes Yes Ileus absent. No anastomotic leak. No No 1 13 27 Yes Yes Yes Ileus absent. No anastomotic leak. No No 1 14 28 Yes Yes Yes Ileus absent. No anastomotic leak. No No 1 15 38 Yes Yes Yes N/A No anastomotic leak. No No 1 16 28 Yes Yes Yes Ileus absent. No anastomotic leak. No No 1 18 21 Yes No No N/A No anastomotic leak. No No 1 19 28 Yes Yes Yes Ileus absent. No anastomotic leak. No No 2 20 21 Yes Yes Yes N/A No anastomotic leak. No No 1 21 26 Yes Yes Yes N/A No anastomotic leak. No No 1 22 36 Yes Yes Yes N/A No anastomotic leak. No No 1 23 41 Yes Yes Yes N/A No anastomotic leak. No No 1 24 33 Yes Yes Yes N/A No anastomotic leak. No No 1 25 62 Yes Yes Yes Ileus absent. No anastomotic leak. No No 1 26 42 Yes Yes Yes N/A No anastomotic leak. No No 1 27 27 Yes Yes Yes N/A No anastomotic leak. No No 1 28 32 Yes Yes Yes N/A No anastomotic leak. No No 1 29 15 No No No N/A No anastomotic leak. No No 1 thirty 28 Yes No No N/A No anastomotic leak. No No 1 31 28 Yes Yes Yes N/A No anastomotic leak. No No 1 33 81 Yes Yes Yes N/A No anastomotic leak. No No 1 34 35 Yes Yes Yes Ileus absent. No anastomotic leak. No No 1 35 21 Yes Yes Yes N/A No anastomotic leak. No No 1 36 37 Yes Yes Yes N/A No anastomotic leak. No No 1 37 thirty Yes Yes Yes Ileus absent. No anastomotic leak. No No 1 38 27 Yes Yes Yes Ileus absent. No anastomotic leak. No No 1 39 27 Yes No No N/A No anastomotic leak. No No 1 40 23 Yes No No N/A No anastomotic leak. No No 1 41 35 Yes Yes Yes Ileus absent. No anastomotic leak. No No 1 42 36 Yes No No N/A No anastomotic leak. No No 1 43 40 Yes Yes Yes Ileus No anastomotic leak. No No 1 44 19 Yes Yes Yes Ileus absent. No anastomotic leak. No No 1 45 27 Yes Yes Yes Ileus absent. No anastomotic leak. No No 2 46 37 No No No N/A No anastomotic leak. Before surgery Yes 2 47 37 Yes Yes Yes Ileus absent. No anastomotic leak. No No 2 48 32 Yes Yes Yes N/A No anastomotic leak. No No 2 49 19 Yes No No N/A No anastomotic leak. No No 1 50 40 Yes Yes Yes Ileus absent. No anastomotic leak. No No 1 51 40 Yes Yes Yes Ileus absent. a leak No No 1 52 17 Yes Yes Yes Ileus absent. No anastomotic leak. No No 2 53 18 Yes No No N/A No anastomotic leak. No No 2 54 17 Yes Yes Yes N/A a leak No No n/a 55 18 Yes Yes Yes Ileus absent. No anastomotic leak. No No 2 57 32 Yes Yes Yes Ileus absent. No anastomotic leak. No No 1 58 24 Yes Yes Yes Ileus absent. No anastomotic leak. No No 1 59 16 Yes Yes Yes N/A No anastomotic leak. No No 1 60 14 Yes Yes Yes Ileus absent. No anastomotic leak. No No 1 61 20 Yes Yes Yes Ileus absent. No anastomotic leak. No No 1 62 19 Yes Yes Yes N/A No anastomotic leak. No No 1 63 27 Yes Yes Yes N/A No anastomotic leak. No No 1 64 19 Yes Yes Yes N/A No anastomotic leak. No No 1 65 31 Yes No No N/A No anastomotic leak. No No 1 66 49 Yes Yes Yes N/A No anastomotic leak. No No 1 67 31 Yes Yes Yes N/A No anastomotic leak. No No 2 68 24 Yes Yes Yes N/A No anastomotic leak. No No 1 69 18 Yes Yes Yes N/A No anastomotic leak. No No 2 70 37 Yes Yes Yes N/A No anastomotic leak. No No 1 71 26 Yes Yes Yes N/A No anastomotic leak. No No 2 72 16 Yes Yes Yes N/A No anastomotic leak. No No 1 73 thirty Yes Yes Yes N/A No anastomotic leak. No No 1 74 25 Yes Yes Yes N/A No anastomotic leak. No No 2 75 48 Yes Yes Yes N/A No anastomotic leak. No No 2 76 26 Yes Yes Yes N/A No anastomotic leak. No No 2 77 280 Yes Yes Yes N/A No anastomotic leak. No No 2 78 23 Yes Yes Yes N/A No anastomotic leak. No No 1 79 20 Yes Yes Yes N/A No anastomotic leak. No No 1 80 47 Yes Yes Yes N/A No anastomotic leak. No No 1 81 17 Yes Yes Yes N/A No anastomotic leak. No No 1 82 29 Yes Yes Yes N/A No anastomotic leak. No No n/a 83 23 Yes Yes Yes N/A No anastomotic leak. No No 2 84 24 Yes Yes Yes N/A No anastomotic leak. No No 1 85 27 Yes Yes Yes N/A No anastomotic leak. No No 1 86 thirty Yes Yes Yes N/A No anastomotic leak. No No 1 87 17 Yes Yes Yes N/A No anastomotic leak. No No 3 88 47 Yes Yes Yes N/A No anastomotic leak. No No 1 89 12 Yes Yes Yes N/A No anastomotic leak. No No 3 90 21 Yes No No Ileus No anastomotic leak. No No 3 91 15 Yes Yes Yes N/A No anastomotic leak. No No 2 92 18 Yes No No N/A No anastomotic leak. No No 2 93 28 Yes Yes Yes N/A No anastomotic leak. No No 2 95 27 Yes Yes Yes N/A No anastomotic leak. No No 1 96 33 Yes Yes Yes N/A No anastomotic leak. No No 1 97 15 Yes Yes Yes Ileus absent. No anastomotic leak. No No 1 98 15 Yes Yes Yes Ileus No anastomotic leak. No No 2 99 thirty Yes Yes Yes N/A No anastomotic leak. No No 1 100 12 Yes Yes Yes Ileus absent. No anastomotic leak. No No 2 101 34 Yes Yes Yes Ileus absent. No anastomotic leak. No No 2 102 13 Yes Yes Yes N/A No anastomotic leak. No No 1 103 38 Yes No No N/A No anastomotic leak. No No 3 104 24 N/A N/A N/A N/A No anastomotic leak. N/A No N/A 105 36 Yes Yes Yes Ileus absent. No anastomotic leak. No No 2 106 21 Yes Yes Yes N/A No anastomotic leak. No No 1 107 17 Yes Yes Yes N/A No anastomotic leak. No No 1 108 12 Yes Yes Yes Ileus absent. No anastomotic leak. No No 3 109 24 Yes Yes Yes Ileus absent. No anastomotic leak. No No 2 110 20 Yes Yes Yes N/A No anastomotic leak. No No 2 111 31 Yes No No N/A No anastomotic leak. Before surgery Yes 1 112 25 Yes No No N/A No anastomotic leak. No No 1 113 22 Yes Yes Yes N/A No anastomotic leak. No No 1 114 33 Yes Yes Yes N/A No anastomotic leak. No No 1 115 24 Yes No No N/A No anastomotic leak. No No 2 116 31 Yes Yes Yes N/A No anastomotic leak. No No 1 117 19 Yes No No N/A No anastomotic leak. No No 2 118 25 Yes Yes Yes N/A No anastomotic leak. No No 1 119 29 Yes No No N/A No anastomotic leak. Before surgery Yes 1 120 31 Yes Yes Yes N/A No anastomotic leak. No No 1 121 14 Yes Yes Yes N/A No anastomotic leak. No No 2 122 34 Yes Yes Yes N/A No anastomotic leak. No No 1 123 24 Yes Yes Yes N/A No anastomotic leak. No No 1 124 41 Yes N/A N/A N/A No anastomotic leak. No No 2 125 31 Yes Yes Yes Ileus absent. No anastomotic leak. No No 1 126 36 Yes Yes Yes Ileus absent. No anastomotic leak. No No 1 127 23 Yes Yes Yes Ileus absent. No anastomotic leak. No No 1 128 54 Yes N/A N/A N/A No anastomotic leak. Before surgery Yes 2 130 19 Yes Yes Yes N/A No anastomotic leak. No No 1

[999] All patients were treated and monitored in accordance with the Danish National Guidelines. The study was approved by the Biomedical Research Ethics Committee of the Central Region of Denmark (1-16-02-453-14) and was conducted in accordance with the Declaration of Helsinki. All participants provided written informed consent.

[1000] Carcinoembryonic antigen (CEA) assay.

[1001] The CEA assay was performed on the Cobas e601 platform (Roche) according to the manufacturer's recommendations using 500 μl of serum. Threshold levels were set at 4.0 μg/L and 6.0 μg/L for nonsmokers and smokers, respectively, as recommended by the reviewing hospital. A person who had not smoked for 8 weeks prior to sample collection was considered a former smoker.

[1002] Tissue collection and whole exome sequencing.

[1003] Tumor tissue was collected from all patients as fresh frozen (n = 102) or formalin-fixed paraffin-embedded (FFPE) tissue (n = 27). Four patients presented with synchronous colorectal cancer (CRC); Tissue from both tumors was collected from these patients. Metastatic tissue was collected from three patients with relapse. Constitutional DNA corresponding to all patients was extracted from peripheral blood leukocytes.

[1004] Primary fresh frozen or formalin-fixed paraffin-embedded (FFPE) tissue samples had a median pathological tumor cellularity of 50% (range 20–90%). (Appendix, Table 2). DNA was extracted using a Puregene ® DNA purification kit (Gentra Systems) or using a QiAamp ® DNA FFPE tissue kit (Qiagen).

[1005] Whole exome sequencing (WES) was performed on matched tumor DNA and buffy coat leukocyte DNA. A summary of the sample and PES information is given in Table 12.5 below. Cancer content was assessed by hematoxylin-eosin assessment of tissue sections cut before and after those used for extraction. Synchronous CRCs are marked in the table with the letters S1 and S2.

Table 12.5: Summary of sample and whole exome sequencing information. Tumor and metastasis tissue Buffy coat Patient ID Fabric type Source of DNA Percentage of tumor cells Input, ng PES coverage Input, ng PES coverage 1 Primary tumor FF 60 500 100.8 500 41.9 2 Primary tumor FF 60 500 94.3 500 36.5 3 Primary tumor FF 60 500 101.8 500 45.3 4 Primary tumor FF 60 500 98.4 500 45.5 5 Primary tumor FF 25 500 167.8 500 40.9 6 Primary tumor FF 50 500 93 500 46.1 7 Primary tumor FF 50 500 91.2 500 40.8 8 Primary tumor FF 85 500 89.9 500 43.2 9 Primary tumor FF 60 500 91.3 500 36.1 10 Primary tumor FF 55 500 140.4 500 41.8 eleven Primary tumor FF 60 500 104.3 500 39.4 12 Primary tumor FF 60 500 106 500 40.8 13 Primary tumor FF 45 500 97 500 47.2 14 Primary tumor FF 70 500 106.6 500 61.3 15 Primary tumor FF 40 500 105.2 500 39.2 16 Primary tumor FF 80 250 122.7 500 92.2 18 Primary tumor FF 80 500 98 500 49.4 19 Primary tumor FF 40 500 100.3 500 37 20 Primary tumor FF 80 500 112.2 500 44.7 21 Primary tumor FF 70 500 117 500 56.5 22 Primary tumor FF 82 500 102.2 500 54.9 23 Primary tumor FF 60 500 104.9 500 43.1 24 Primary tumor FF 90 500 104.2 500 46.5 25 Primary tumor FF 35 500 106.1 500 43.6 26 Primary tumor FF 80 500 150.9 500 46.5 27 Primary tumor FF 50 500 149.2 500 54.9 28 Primary tumor FF 70 500 129.2 500 60.9 29 Primary tumor FF 45 500 86.1 500 45.1 thirty Primary tumor FF 20 500 222 500 40 31 Primary tumor FFPE 50 500 123.6 500 56.8 33 (S1) Primary tumor FF 35 500 65.5 500 55.9 33 (S2) Primary tumor FFPE 60 500 87.7 500 55.9 34 Primary tumor FF 50 500 90.4 500 46.8 35 Primary tumor FF thirty 500 186.3 500 41.6 36 Primary tumor FF 25 500 184.4 500 36.8 37 Primary tumor FF 70 500 87.8 500 50.8 38 Primary tumor FF 35 500 86.9 500 54.2 39 Primary tumor FF 20 500 179.4 500 49.8 40 Primary tumor FF 70 500 86.4 500 62.7 41 Primary tumor FF 65 500 88.4 500 60.5 42 Primary tumor FF 60 500 94.9 500 35.9 43 Primary tumor FF 40 500 121.3 500 35.2 44 Primary tumor FF 90 500 103.8 500 45.2 45 Primary tumor FF 40 500 107.6 500 44.8 46 Primary tumor FF 70 500 99 500 40.6 47 Primary tumor FF 45 500 94.4 500 37.6 48 (S1) Primary tumor FF 50 500 108.5 500 52.7 48 (S2) Primary tumor FFPE 50 500 95 500 52.7 49 Primary tumor FF 20 200 127.9 500 57.7 50 Primary tumor FF 50 500 93.1 500 43.2 51 Primary tumor FFPE 50 500 130.7 500 36.4 52 Primary tumor FFPE 50 500 113 500 55.4 53 Primary tumor FF 40 500 93.7 500 56 54 Primary tumor FF 60 500 113.6 500 79.7 55 Primary tumor FF 20 500 161 500 69.1 57 Primary tumor FF 80 500 85 500 72.8 58 Primary tumor FF 55 500 79.4 500 53.7 59 Primary tumor FF 50 250 127.1 500 58.4 60 Primary tumor FFPE 40 500 119.1 500 57.4 61 Primary tumor FF 40 500 93.6 500 81.9 62 Primary tumor FFPE 20 500 120.2 500 40 63 Primary tumor FF 60 100 116.6 500 44.8 64 Primary tumor FF 55 500 109 500 41.3 65 Primary tumor FFPE 50 500 99.4 500 39.1 66 Primary tumor FF 70 500 105.2 500 47.7 67 Primary tumor FF 77 500 82.1 500 50.2 68 Primary tumor FF 50 500 102 500 42 69 Primary tumor FF 60 500 97.3 500 44.9 70 Primary tumor FFPE 20 500 150.4 500 39.8 71 Primary tumor FF 25 500 160.4 500 39.2 72 Primary tumor FF 50 500 126.8 500 57 73 Primary tumor FF 45 500 90.3 500 55.3 74 Primary tumor FF 65 500 127.3 500 44.7 75 Primary tumor FF 85 500 108.8 500 36.5 76 Primary tumor FF 85 250 111.8 500 55.6 77 Primary tumor FF 50 500 124.5 500 42.1 78 Primary tumor FF 53 500 94.2 500 53.5 79 Primary tumor FF 65 500 101.2 500 54.1 80 Primary tumor FF 80 500 93.8 500 44.8 81 Primary tumor FF 45 500 101.8 500 38.3 82 Primary tumor FF thirty 500 219.9 500 69.3 83 Primary tumor FFPE 50 200 38.6 500 52.2 84 Primary tumor FFPE 50 500 69.2 500 39.6 85 Primary tumor FF 60 500 107.6 500 63.3 86 Primary tumor FF 80 200 116.2 500 38.3 87 Primary tumor FF 40 500 120 500 39.7 88 Primary tumor FF 18 500 192.4 500 45.4 89 Primary tumor FF 63 500 91.7 500 56.5 90 (S1) Primary tumor FFPE 50 500 94.6 500 41.3 90 (S2) Primary tumor FFPE 50 500 70.8 500 41.3 91 Primary tumor FF 43 500 82.8 500 84 92 Primary tumor FF 55 500 132.1 500 54.1 93 Primary tumor FFPE 50 500 118.4 500 41.1 95 Primary tumor FF 80 100 148.6 500 52.3 96 Primary tumor FFPE 40 500 48.3 500 52.9 97 Primary tumor FFPE 50 500 73.2 500 33.1 98 Primary tumor FFPE 50 500 96.8 500 41.6 99 Primary tumor FFPE 50 500 125.1 500 38.4 100 Primary tumor FF 45 500 81.5 500 38.4 101 Primary tumor FF 40 500 88.1 500 43.3 102 Primary tumor FFPE 50 500 127.7 500 32.9 103 Primary tumor FF 20 500 169.3 500 53.5 104 Primary tumor FF 35 500 99.9 500 107.4 105 Primary tumor FF 45 100 383.4 500 80.5 106 Primary tumor FF 55 500 83.2 500 46.9 107 Primary tumor FFPE 50 500 99.5 500 45.6 108 Primary tumor FF 75 500 98.5 500 57.3 109 Primary tumor FF thirty 500 168.7 500 56.5 110 Primary tumor FF 60 500 85.8 500 35.6 111 Primary tumor FFPE 50 500 54.5 500 38.8 112 Primary tumor FF 50 500 86.5 500 54.3 113 Primary tumor FFPE 50 500 92.2 500 78.7 114 (S1) Primary tumor FF 75 500 80.7 500 100.3 114 (S2) Primary tumor FF 85 500 98.6 500 100.3 115 Primary tumor FF 20 500 179.6 500 38.2 116 Primary tumor FF 45 500 97.6 500 43.2 117 Primary tumor FF 68 500 91.5 500 74.2 118 Primary tumor FF 80 500 94.4 500 53 119 Primary tumor FFPE 40 500 66 500 54.9 120 Primary tumor FF 40 500 80.4 500 83 121 Primary tumor FFPE 50 500 94.1 500 127 122 Primary tumor FFPE thirty 500 61.2 500 37.3 123 Primary tumor FF 25 500 176.9 500 56.5 124 Primary tumor FF 35 250 137.2 500 93.2 125 Primary tumor FF 70 500 119.6 500 35 126 Primary tumor FF 60 500 102.6 500 48.1 127 Primary tumor FFPE 50 500 83.1 500 48.5 128 Primary tumor FFPE 50 500 72.3 500 105.8 130 Primary tumor FF 70 500 88.8 500 50.4

[1006] Library preparation, sequencing and data analysis were performed as described in Lamy et al . Paired Exome Analysis Reveals Clonal Evolution and Potential Therapeutic Targets in Urothelial Carcinoma. Cancer Res . 76(19):5894–5906 (2016).

[1007] Blood collection and plasma isolation

[1008] Blood samples were collected in 10 ml K2-EDTA tubes (BD 367525) at Aarhus University Hospital. All samples were processed within 2 hours of collection by centrifuging the blood twice at room temperature, first for 10 minutes at 3000 g and then centrifuging the plasma for 10 minutes at 3000 g. Plasma was aliquoted into 5 ml cryovials and stored at minus 80°C.

[1009] Cell-free DNA extraction, quantitation and library preparation.

[1010] In this study, up to 10 ml of plasma per case was used (range 2-10 ml; median 8.5 ml), and cell-free DNA (scDNA) was extracted using the QIAamp ® Circulating Nucleic Acid kit (Qiagen) and eluted in 50 µl DNA Suspension Buffer (Sigma). Each scDNA sample was quantified using the Quant-iT ® High Sensitivity dsDNA Assay Kit (Invitrogen). In 125 patients, cDNA was isolated from 795 serial plasma samples.

[1011] Up to 66 ng (20,000 genome equivalents) of cDNA from each plasma sample was used as input for library preparation. The scDNA was subjected to end repair, A-tailing, and ligation using specific adapters as described in Abbosh et al. Phylogenetic ctDNA analysis depicts early-stage lung cancer evolution. Nature 545(7655):446–451(2017). The purified ligation product was amplified for 20 cycles and purified using Ampure ® X microbeads (Agencourt/Beckman Coulter).

[1012] Multiplex PCR assay scheme

[1013] Patient-specific somatic variants were identified by analyzing the primary tumor and RPE of corresponding normal samples for all patients. The clonality of variants was inferred from the estimated proportion of cancer cells harboring the variant. Note that inference of clonality for samples with a low tumor cell fraction is limited due to the rather flat frequency distribution of the variant allele. Observed TNAs in tissue and in the context of variant sequences were used to prioritize the somatic CNAs and short INSERTIONS/DELETIONS identified for each tumor. The standard Signatera amplicon design process was used to generate PCR primer pairs for this set of variants. For each patient, 16 high-ranking compatible amplicons were selected for an individual patient panel. PCR primers were ordered from Integrated DNA Technologies.

[1014] Multiplex PCR Plasma Next Generation Sequencing Workflow

[1015] An aliquot of each library was used as input for a patient-related 16-plex PCR reaction. Samples were amplified using a patient-specific assay and barcoded, followed by pooling. Sequencing was performed on an Illumina HiSeq ® 2500 Rapid Run with 50 paired-end read cycles using the Illumina PAIRED END ® v2 kit with an average read depth of greater than 105,000X per amplicon.

[1016] Bioinformatics Processing Process

[1017] All paired-end reads were merged using Pear software as described in Zhang, Bioinformatics, 30(5): 614–620 (2014). Bases that did not match forward and reverse reads or had low quality scores were filtered out to minimize sequencing errors. The merged reads were aligned to the hg19 reference genome using Novoalign version 2.3.4 ( http://www.novocraft.com/ ). Amplicons with fewer than 5000 high-quality reads were considered to fail quality control (QC). QC was performed using a proprietary program that examined a broad list of statistics for each sample, which included total reads, mapped reads, targeted reads, number of failed targets, and average error rate.

[1018] Recognition of plasma variants

[1019] To build the background error model, a large set of negative control samples (~1000) were preprocessed. For each target variant, using the read depth of the mutant and reference alleles, a confidence score was calculated based on the error model as described in Abbosh et al. Phylogenetic ctDNA analysis depicts early-stage lung cancer evolution. Nature 545(7655):446–451 (2017). A plasma sample with at least 2 variants with a confidence score above the predefined algorithm threshold was defined as ctDNA positive, as described in Abbosh et al, 2017, above.

[1020] Statistical analysis

[1021] The primary outcome measure was time to recurrence (TTR) assessed by standard radiological criteria. TTR was measured from the date of surgery to the recorded first radiological recurrence (local or distant) or death due to colorectal cancer and checked at last follow-up or death not related to colorectal cancer. Survival analysis was performed using the Kaplan-Meier method. Cox proportional hazards regression analysis was used to evaluate the effect of ctDNA and CEA on TTR. Multivariate analysis was performed on clinical parameters that were statistically significant on univariate analysis. All P-values were based on two-sided testing, and differences were considered significant at P≤0.05. Statistical analysis was performed using STATA IC/12.1 and R Statistical software, version 2.4 for Windows.

[1022] RESULTS

[1023] From 2014 to 2016, 130 patients with UICC stage I–III colorectal cancer were included in the study. Five patients were subsequently excluded because they were either lost to follow-up (n = 3) or converted to stage IV. Whole exome sequencing (WES) of tumor and corresponding germline DNA was used to identify somatic mutations, as shown in FIG. 90A-B. Tumor-specific multiplex PCR assay panels targeting 16 mutations were developed for each patient. Ultra-deep multiplex PCR-based SSP was used to analyze and quantify circulating tumor DNA in 795 plasma samples from 125 patients with a median follow-up of 12.5 months (range, 1.4 to 38.5 months). The workflow of this study is shown in FIG. 83A-E. Quality control was carried out at every stage of the work process. The read depth for tests that passed the coating quality control was greater than 105,000X, as shown in FIG. 91. Details of the results and ctDNA dynamics for all 125 patients are listed in Table 12.6 and presented in FIG. 92. During the follow-up period, 24 (19.2%) patients had radiographic recurrences.

[1024] Preoperative detection of ctDNA

[1025] In initial preoperative plasma samples (n=122), we detected ctDNA in 89% of samples with sensitivities of 40%, 92%, and 90% for stages I, II, and III, respectively, as shown in FIG. 87A. Carcinoembryonic antigen (CEA) analysis performed on the same samples detected 43.3% of cancers, as shown in FIG. 88.

[1026] Postoperative ctDNA status at day 30 predicts relapse

[1027] To evaluate the ability to detect residual disease and predict future relapse, we performed ctDNA analysis in plasma samples collected after surgery. Plasma collected at day 30, before the start of adjuvant chemotherapy, was available for 94 patients. Interestingly, the vast majority (89.4%) of patients were ctDNA negative, and only 10.6% of patients were ctDNA positive after surgery, as shown in FIG. 89. These ctDNA-positive patients had a significantly higher relapse rate (70%, 7/10) compared with those who were ctDNA-negative after surgery (11.9%, 10/84), as shown in FIG. 87B. The presence of ctDNA was associated with a significant reduction in time to relapse (TTR) compared with ctDNA-negative patients (HR, 7.2; 95% CI, 2.7-19; P less than 0.0000), as shown in FIG. 87C. Compared with known prognostic factors such as stage and lymphovascular invasion, in the multivariate logistic regression model, ctDNA status was the only significant prognostic factor, as shown in Table 12.7 below.

Table 12.7: Analysis of time to relapse using clinicopathological variables and postoperative ctDNA status at day 30. Variable Univariate analysis Multivariate analysis* All patients with postoperative sample at day 30 (n = 94) HR (95% CI) P HR (95% CI) P Age < average vs ≥ average 1.4 (0.52-3.8) 0.508 Stage Stage II vs Stage III 5.3 (1.2-23) 0.028 3.2 (0.65-16) 0.152 Tumor localization Right vs Left 1.9 (0.72-5.3) 0.191 Lymphovascular invasion No against yes 2.7 (1.0-7.0) 0.044 1.6 (0.40-3.3) 0.785 MMR status Scarcity vs. Availability 0.45 (0.059-3.4) 0.432 Radical resection (micro) Yes vs no 2.3 (0.72-7.2) 0.160 Histology Adeno- versus mucinous carcinoma 1.6 (0.35-7.2) 0.531 Tumor differentiation Moderate/high versus low 0.86 (0.19-3.8) 0.847 Floor Woman vs man 0.24 (0.079-0.74) 0.013 ctDNA status up to 6 weeks after surgery (without ACT) ctDNA- vs ctDNA+ 7.2 (2.7-19) <0.0000 4.4 (1.5-12.8) 0.006 *Includes variables that were statistically significant on univariate analysis except gender.

[1028] A subgroup of patients received chemotherapy (n=52), which may have altered the predictive value of ctDNA for outcome. However, even for this subgroup, ctDNA positivity was associated with a high risk of relapse (HR, 7.1; 95% CI, 2.2-22; P=0.0008) as shown in FIG. 90A-B. The relapse rate in ctDNA-negative patients was 11.9% regardless of whether they were treated with ACT (5/42) or not (5/42). Thus, ctDNA status at 30 days after surgery is a strong predictor of future relapse, even for patients treated with ACT.

[1029] Adjuvant chemotherapy eliminates ctDNA in a subfraction of day 30 in ctDNA positive patients.

[1030] Although randomized trials have shown that adjuvant chemotherapy (ACT) can reduce the overall relapse rate of stage III CRC, 21–24 it is currently unknown whether ACT can specifically prevent relapse among the high-risk ctDNA-positive subgroup. Ten patients who were ctDNA positive at day 30 were subsequently treated with AChT as shown in FIG. 87D. Among these patients subsequently treated with chemotherapy, 70% (n = 7) relapsed, while 30% (n = 3) were still disease free at the end of the follow-up period. This treatment efficacy is similar to that assessed when ACT is given to all stage III colon cancers, as described in Upadhyay et al ., Chemotherapy use in stage III colon cancer: a National Cancer Database analysis. Ther Adv Med Oncol. 7(5):244-251 (2015); Andre et al. , Improved overall survival with oxaliplatin, fluorouracil, and leucovorin as adjuvant treatment in stage II or III colon cancer in the MOSAIC trial. J Clin Oncol. 27(19):3109-3116 (2009); Gill et al. , Pooled analysis of fluorouracil-based adjuvant therapy for stage II and III colon cancer: who benefits and by how much? J Clin Oncol. 22(10):1797-1806 (2004); Haller et al. , Capecitabine plus oxaliplatin compared with fluorouracil and folinic acid as adjuvant therapy for stage III colon cancer . J Clin Oncol. 29(11):1465–1471 (2011). Thus, the results presented herein and shown in FIG. 87D showed that ACT can eliminate residual disease in a subgroup of high-risk ctDNA-positive patients.

[1031] For two of the three patients who did not relapse, collected plasma samples were available over time. Consistent with ACT eliminating residual disease, these patients demonstrated complete ctDNA clearance during therapy and remained negative throughout the study. In contrast, six patients with relapse and plasma available over time either remained ctDNA-positive during ACT or became ctDNA-positive again shortly after completion of ACT.

[1032] Monitoring ctDNA over time measures the effectiveness of treatment with AChT.

[1033] Follow-up blood samples were available for 8/10 patients who were ctDNA positive before initiation of chemotherapy. These longitudinally collected blood samples were analyzed to monitor changes in ctDNA levels during treatment. The ctDNA status became negative in 50% of patients (n=4), as shown in FIG. 87D, while in the remaining four patients the ctDNA status remained positive throughout treatment. Strikingly, all four patients (100%) who did not clear ctDNA had disease recurrence, indicating that residual ctDNA predicted that ACT failed to clear residual disease. Of the four patients who cleared ctDNA during treatment, two remained ctDNA negative in all post-ACT samples and did not relapse at all times, while the other two patients regained ctDNA positivity shortly after treatment and ultimately a relapse developed, as shown in FIG. 87D.

[1034] Detection of ctDNA after chemotherapy identifies a subgroup of patients at very high risk of relapse.

[1035] Because 100% of patients who did not clear ctDNA during adjuvant chemotherapy (ACT) subsequently experienced disease relapse, we hypothesized that ctDNA analysis of the first blood sample drawn after ACT could be used to identify a subgroup of patients with ongoing residual disease that will benefit from further treatment. We found that of the 58 patients with blood samples after ACT, all ctDNA-positive patients (7/7) had a relapse. In comparison, the relapse rate was 13.7% (7/51) for ctDNA-negative patients (Fisher's exact test, P less than 0.0001), as shown in FIG. 87E. ctDNA status after ACT was a stronger predictor of the onset of relapse compared with other prognostic factors such as stage, lymphovascular invasion, microscopic radical resection status, and CEA, as shown in Table 12.8 below, and ctDNA status was a highly significant predictor of time to relapse (TTR ) (HR, 18.0; 95% CI, 5.4-57; P less than 0.0000), as shown in FIG. 87F.

Table 12.8: Analysis of time to relapse using clinicopathological variables, postoperative ctDNA status, and postoperative CEA status at the first time point after ACT. Variable Univariate analysis* HR (95% CI) P All patients with plasma and AChT collected over time (n=58) Age less than average vs. more than or equal to average 1.3 (0.44-3.8) 0.612 Stage Stage II vs Stage III 1.3 (0.17-10) 0.794 Tumor localization Right vs Left 1.8 (0.63-5.3) 0.265 Lymphovascular invasion No against yes 3.0 (0.93-9.5) 0.0664 MMR status Scarcity vs. Availability 1.1 (0.0-Undef.) 0.998 Radical resection (micro) Two against no 2.6 (0.77-9) 0.125 Histology Adeno- versus mucinous carcinoma 1.1 (,14-8.3) 0.943 Tumor differentiation Moderate/high versus low 1.2 (0.27-5.6) 0.779 Tumor perforation No against yes 0.91 (0.12-7) 0.929 Floor Woman vs man 0.2 (0.055-0.71) 0.013 CEA CEA- vs CEA+ 2.4 (0.74-7.9) 0.141 ctDNA ctDNA- vs ctDNA+ 18 (5.4-57) <0.0000

[1036] Analysis of ctDNA over time, including all blood samples after chemotherapy, was an even stronger predictor of time to relapse (RR, 29.0; 95% CI, 6.4-130; p < 0.0000), and identified 13 ctDNA-positive patients, of which 92.3% (12/13) had recurrence, as shown in FIG. 91. Although the carcinoembryonic antigen (CEA) assay over time was also a significant predictor of time to relapse, as shown in FIG. 92, after multivariate adjustment, ctDNA status over time was the only significant predictor of time to relapse (HR, 26.9; 95% CI, 5.11–142; P = 0.0001), as shown in Table 12.9 below.

Table 12.9: Analysis of time to relapse using clinicopathologic variables, postoperative ctDNA status, and postoperative CEA status at all time points after ACT. Variable Univariate analysis* Multivariate analysis * All patients with plasma collected over time and AChT (n=58) HR (95% CI) P HR (95% CI) P Age less than average versus greater than or equal to average 1.3 (0.44-3.8) 0.612 Stage Stage II vs Stage III 1.3 (0.17-10) 0.794 Tumor localization Right vs Left 1.8 (0.63-5.3) 0.265 Lymphovascular invasion No against yes 3.0 (0.93-9.5) 0.0664 MMR status Scarcity vs. Availability 1.1 (0.0-Inf) 0.998 Radical resection (micro) Yes vs no 2.6 (0.77-9) 0.125 Histology Adeno- versus mucinous carcinoma 1.1 (,14-8.3) 0.943 Tumor differentiation Moderate/high versus low 1.2 (0.27-5.6) 0.779 Tumor perforation No against yes 0.91 (0.12-7) 0.929 Floor Woman vs man 0.2 (0.055-0.71) 0.013 CEA CEA- vs CEA+ 5.5 (1.7-18) 0.00449 1.1 (0.31-4.2) 0.854 ctDNA ctDNA- vs ctDNA+ 29 (6.4-130) <0.0000 26.9 (5.11-142) <0.0000 *Includes variables that were statistically significant in univariate analysis except gender.

[1037] Analysis of ctDNA over time predicted patient outcome and made earlier detection of relapse possible.

[1038] Serial analysis of ctDNA during follow-up after radical treatment of 75 patients using plasma samples over time detected metastatic recurrence with a sensitivity of 87.5% (14/16) and specificity of 98.3% (58/59). Strikingly, 93.3% (14/15) of ctDNA-positive patients had relapses compared with a relapse rate of only 3.3% (2/60) for ctDNA-negative patients (Fisher's exact test, P less than 0.0001 ). Patients positive for ctDNA had a significantly reduced time to relapse (TTR) (HR, 44.0; 95% CI, 9.8-190; P < 0.0000), as shown in FIG. 93A-B. The course of the disease and ctDNA results over time for all 75 patients are shown in FIG. 94. Serial ctDNA analysis missed two relapses (patients 20 and 24, FIG. 94). However, whole exome sequencing of the two missed metastases confirmed the presence of mutations used for plasma screening, as shown in Table 12.10 below.

Table 12.10: Corresponding PES of tumor and metastases in two patients in whom ctDNA was not detected and a third patient in whom plasma analysis detected ctDNA after clinical relapse. Patient ID 20 24 77 ctDNA
not identified ctDNA not detected ctDNA detected after relapse Mutations present in the tumor are also present in the metastasis (%) 62.6 50.0 79.4 Mutations found in plasma are also present in metastasis (%) 100.0 87.5 93.8

[1039] For two patients with relapse (IN 20 and 24, Table 12.10), follow-up analysis after surgery did not detect ctDNA, as shown in FIG. 90A-B. The same amount of plasma was analyzed from these two patients as from the other patients. Potential sample replacement may have been rejected due to 45 common SNPs across all tumors and plasma samples, confirming that no samples were replaced. We then performed whole exome sequencing (WES) of metastatic relapses for two patients and confirmed that the mutations selected for plasma profiling were present in the metastases. Table 12.9. PES was also performed on metastases from patient 77. For this patient, follow-up ctDNA analysis did not detect ctDNA until recurrence was detected by radiological imaging, as shown in FIG. 90A-B. Again, PES confirmed that the mutations selected for plasma profiling were present in the metastases. Thus, the negative postoperative results were due to ctDNA levels below the detection level and not because the selected markers were uninformative.

[1040] Time course analysis CEA analysis of the same population revealed recurrence with a sensitivity of 68.8% (11/16) and specificity of 64.4% (38/59), as shown in FIG. 95. In multivariate analysis, ctDNA was the only significant predictor of time to relapse (TTR) (HR, 41; 95% CI, 8.5-199; P < 0.0000), as shown in Table 12.11 below.

Table 12.11: Analysis of time to relapse using clinicopathological variables and postoperative ctDNA and CEA status in follow-up samples. Variable Univariate analysis* Multivariate analysis * All patients with plasma collected over time (n=75) HR (95% CI) P HR (95% CI) P Age < average vs ≥ average 1.0 (0.39-2.8) 0.947 Stage Stage II vs Stage III 3.3 (0.74-15) 0.118 Tumor localization Right vs Left 1.2 (0.45-3.2) 0.722 Lymphovascular invasion No against yes 4.1 (1.4-12) 0.010 1.1 (0.35-3.4) 0.875 MMR status Scarcity vs. Availability 1.2 (0.0-Inf) 0.998 Radical resection (micro) Two against no 2.9 (0.88-9.3) 0.080 Histology Adeno- versus mucinous carcinoma 0.84 (.84-6.4) 0.868 Tumor differentiation Moderate/high versus low 0.92 (0.21-4.1) 0.911 Tumor perforation No against yes 1.1 (0.15-8.7) 0.893 Floor Woman vs man 0.25 (0.08-0.8) 0.018 CEA CEA- vs CEA+ 2.8 (0.95-8.0) 0.061 ctDNA ctDNA- vs ctDNA+ 44 (9.8-190) <0.0000 41 (8.5-199) <0.0000 *Includes variables that were statistically significant in univariate analysis except gender.

[1041] For patients with metastatic recurrence and detectable ctDNA, ctDNA analysis was found to have a mean lead time of 8.7 months (Wilcoxon signed rank test; P=0.0009) compared with standard CT images, as shown in FIG. 93C; whereas the lead time by carcinoembryonic antigen (CEA) assay could not be established as shown in FIG. 96. From the time of ctDNA detection until radiological detection of recurrence, plasma samples remained positive for ctDNA, and a 50-fold increase in the mean ctDNA ANA was observed, indicating a dramatic increase in tumor burden while patients were awaiting radiological detection of recurrence, as shown in FIG. 93D.

[1042] ctDNA analysis revealed clinically significant mutations

[1043] Having demonstrated that time-course ctDNA analysis allows for early detection of micrometastases, we next examined whether time-course ctDNA analysis could be used to provide information about potentially relevant mutations present in metastases.

[1044] From the available follow-up samples, 11 patients with metastatic recurrence were identified and clinically significant mutations were identified by performing whole exome sequencing (WES) of the primary tumor, as shown in Table 12.12 below.

Table 12.12: Patients with significant mutations found in the primary tumor. IN pat. Chr Gene Position IN ONP Nucleotide Alto. nucleotide Changing position Amino acid change 42 chr14 AKT1 105246551 rs121434592 C T c.49G>A p.Glu17Lys 42 chr7 BRAF 140453136 rs113488022 A T c.1799T>A p.Val600Glu 18 chr12 KRAS 25398284 rs121913529 C T c.35G>A p.Gly12Asp 20 chr12 KRAS 25398284 rs121913529 C T c.35G>A p.Gly12Asp 20 chr10 PTEN 89692981 . T TG c.988dupG p.Glu330fs 68 chr3 PIK3CA 178936091 rs104886003 G A c.1633G>A p.E545K 68 chr12 KRAS 25398284 rs121913529 C T c.35G>A p.Gly12Asp 24 chr12 KRAS 25398284 rs121913529 C A c.35G>T p.Gly12Val 24 chr18 SMAD4 48604706 . G T c.1528G>T p.Gly510* 29 chr12 KRAS 25398281 rs112445441 C T c.38G>A p.Gly13Asp thirty chr3 PIK3CA 178936082 . G A c.1624G>A E542K thirty chr12 KRAS 25398284 rs121913529 C A c.35G>T p.Gly12Val thirty chr18 SMAD4 48591928 rs377767350 T G c.1091T>G p.Leu364Trp 75 chr17 ERBB2 37880261 . G T c.2305G>T p.Asp769Tyr 75 chr17 ERBB2 37881000 rs121913471 G T c.2329G>T p.Val777Leu 77 chr7 BRAF 140453136 rs113488022 A T c.1799T>A p.Val600Glu 37 chr12 KRAS 25398285 rs121913530 C T c.34G>A p.Gly12Ser 89 chr12 KRAS 25398285 rs121913530 C A c.34G>T p.Gly12Cys 125 chr12 KRAS 25398284 rs121913529 C A c.35G>T p.Gly12Val 85 chr3 PIK3CA 178938934 . G A c.2176G>A p.Glu726Lys 85 chr12 KRAS 25398285 rs121913530 C A c.34G>T p.Gly12Cys 119 chr12 KRAS 25398285 rs121913530 C T c.34G>A p.Gly12Ser 99 chr12 KRAS 25380278 . A T c.180T>A p.Gly60Gly 99 chr12 KRAS 25380277 rs121913238 G T c.181C>A p.Gln61Lys 104 chr18 SMAD4 48604701 . G C c.1523G>C p.Gly508Ala 124 chr7 BRAF 140453136 rs113488022 A T c.1799T>A p.Val600Glu 108 chr3 PIK3CA 178936094 rs121913286 C A c.1636C>A Q546K 108 chr12 KRAS 25398285 rs121913530 C A c.34G>T p.Gly12Cys

12.12 continued: Patients with significant mutations found in the primary tumor. IN pat. Position Type of cancer A drug Link Database ctDNA+ dynamics in plasma 42 105246551 Solid tumors AZD5363 AKT Inhibition in solid tumors with AKT1 mutations. https://www.mycancergenome.org/ 42 140453136 Colorectal cancer RTK inhibitor+Trametinib http://oncokb.org/#/actionableGenes 18 25398284 All tumors GDC-0994, KO-947, LY3214996 http://oncokb.org/#/actionableGenes Yes 20 25398284 All tumors GDC-0994, KO-947, LY3214996 http://oncokb.org/#/actionableGenes 20 89692981 Colorectal cancer GSK2636771 and AZD8186 http://oncokb.org/#/actionableGenes 68 178936091 All tumors AZD8186 Bendell et al. 2011). Clarke and Workman (2012) https://www.mycancergenome.org/ Yes 68 25398284 All tumors GDC-0994, KO-947, LY3214996 http://oncokb.org/#/actionableGenes Yes 24 25398284 Colorectal cancer GDC-0994, KO-947, LY3214996 http://oncokb.org/#/actionableGenes 24 48604706 All tumors Targeting the TGFβ pathway for cancer therapy https://www.mycancergenome.org/ 29 25398281 Colorectal cancer GDC-0994, KO-947, LY3214996 http://oncokb.org/#/actionableGenes Yes thirty 178936082 Colorectal cancer/Breast cancer Aspirin/PI3K inhibitors Bendell et al. 2011). Clarke and Workman (2012) https://www.mycancergenome.org/ Yes thirty 25398284 Colorectal cancer/Breast cancer GDC-0994, KO-947, LY3214996 http://oncokb.org/#/actionableGenes Yes thirty 48591928 Colorectal cancer Targeting the TGFβ pathway for cancer therapy https://www.mycancergenome.org/ Yes 75 37880261 All tumors Neratinib/afatinib HER2 activating mutations are targets for colorectal cancer treatment https://civicdb.org Yes 75 37881000 All tumors Neratinib/afatinib HER2 activating mutations are targets for colorectal cancer treatment https://civicdb.org Yes 77 140453136 Colorectal cancer RTK inhibitor+Trametinib http://oncokb.org/#/actionableGenes Yes 37 25398285 Colorectal cancer GDC-0994, KO-947, LY3214996 http://oncokb.org/#/actionableGenes Yes 89 25398285 Colorectal cancer GDC-0994, KO-947, LY3214996 http://oncokb.org/#/actionableGenes 125 25398284 Colorectal cancer GDC-0994, KO-947, LY3214996 http://oncokb.org/#/actionableGenes 85 178938934 Colorectal cancer Aspirin/PI3K inhibitors Bendell et al. 2011). Clarke and Workman (2012) https://www.mycancergenome.org/ Yes 85 25398285 All tumors GDC-0994, KO-947, LY3214996 http://oncokb.org/#/actionableGenes Yes 119 25398285 Colorectal cancer GDC-0994, KO-947, LY3214996 http://oncokb.org/#/actionableGenes Yes 99 25380278 Colorectal cancer GDC-0994, KO-947, LY3214997 http://oncokb.org/#/actionableGenes 99 25380277 Colorectal cancer GDC-0994, KO-947, LY3214998 http://oncokb.org/#/actionableGenes 104 48604701 Colorectal cancer Targeting the TGFβ pathway for cancer therapy https://www.mycancergenome.org/ Yes 124 140453136 Colorectal cancer RTK inhibitor+Trametinib http://oncokb.org/#/actionableGenes Yes 108 178936094 Colorectal cancer Aspirin/PI3K inhibitors Bendell et al. 2011). Clarke and Workman (2012) https://www.mycancergenome.org/ 108 25398285 Colorectal cancer GDC-0994, KO-947, LY3214996 http://oncokb.org/#/actionableGenes

[1045] As a proof of concept, an additional multiplex PCR panel targeting significant mutations was developed and applied to dynamic samples. A significant mutation was found in 82% (9/11) of patients, as shown in FIG. 97A. We observed a good correlation between the average ctDNA FNA and the FNA of significant mutations, as shown in FIG. 97B. Changes in significant variant allele frequencies (SVA) showed generally good correlation with treatment and very little intermutational variability, as shown in FIG. 97C.

[1046] DISCUSSION

[1047] This example demonstrated that ctDNA analysis over time in patients with stage I–III CRC can effectively detect and monitor changes in tumor formation throughout the clinical course of the disease. In particular, it was demonstrated that ctDNA served as a reliable biomarker for i) preoperative detection of CRC, ii) postoperative and post-ACT risk stratification, iii) monitoring the effectiveness of ACT, iv) identification of clinically significant mutations and v) early detection of relapses. These observations have important and potentially paradigm-changing implications for the future postoperative management of patients with CRC, and they lay the foundation for future interventional trials to explore the clinical benefits of ctDNA-guided treatment.

[1048] In the preoperative context, the utility of preoperative ctDNA measurements for disease detection has been demonstrated.

[1049] With regard to patient stratification, the ctDNA analysis used here stratified patients into high- and low-risk relapse groups, which could have a potential impact on the selection of patients for treatment with adjuvant chemotherapy (ACT) and for decisions about additional treatment after AHT. Previously, treatment decisions for ACT were based on stage and clinical risk factors. However, this invention has shown that ctDNA status is a stronger prognostic factor than stage, CEA and other high-risk features. Therefore, in the future, it may be possible to prescribe ctDNA-based treatment with ACT for ctDNA-positive but clinically low-risk (stages I and II) patients who would not receive ACT as standard of care today. The inventors of the present invention are currently conducting trials to evaluate the clinical utility of ctDNA-based patient selection in these settings (eg, IMPROVE-IT ClinicalTrials.gov: NCT03748680 and DYNAMIC Australian New Zealand Clinical Trials registry: ACTRN12615000381583 ).

[1050] The present invention also demonstrated that ctDNA-negative patients had a low risk of relapse, regardless of whether they were administered ACT (11.9%) or not (11.9%). Therefore, in the future, it may be possible to keep ctDNA-negative but clinically high-risk (stage III) patients off chemotherapy with minimal impact on their risk of relapse. This group of patients could be offered ctDNA-based active surveillance instead of ACT, which would spare many patients cured by surgery alone from the toxicity of chemotherapy. Additionally, in the post-ACT setting, where current prognostic markers are lacking, we demonstrate that ctDNA analysis identifies patients who still have residual disease. This population may benefit from enhanced therapeutic treatment.

[1051] The present invention also demonstrates that monitoring ctDNA over time before, during and after ACT can provide patient-level measures of ACT effectiveness. 30% of patients who had ctDNA clearance and remained ctDNA negative in all subsequent samples remained disease free throughout the study. In summary, this example provided the first line of evidence that ACT may reduce the risk of relapse in ctDNA-positive patients. The present invention also demonstrated that all patients who did not clear ctDNA relapsed within a year of completion of chemotherapy, and all patients who experienced temporary clearance also relapsed. Future clinical trials that incorporate ctDNA clearance into the study design may allow patient-level measurement of therapy efficacy in real time.

[1052] Postoperatively, ctDNA monitoring shows significant improvement in detecting recurrence compared with standard radiographic imaging, demonstrating a significant lead time of 8.7 months (P < 0.001). Importantly, while awaiting radiological detection, ctDNA levels increased an average of 50-fold, indicating a dramatic increase in tumor burden during the 8.7-month lead time. Current guidelines recommend surveillance after radical CRC surgery, but most recurrences are detected too late to be suitable for radical intervention. Early detection of residual disease by ctDNA analysis may enable earlier radiological detection. In addition to identifying residual disease, ctDNA analysis also identified clinically significant mutations. Therefore, ctDNA has the potential for both early detection and treatment decisions.

[1053] In conclusion, the invention presented in this example provides a potentially paradigm-changing clinical application of ctDNA in colorectal cancer. As mentioned above, additional clinical trials are being developed or already underway to explore the clinical benefits of ctDNA-guided treatments. The results presented here enable the use of circulating biomarkers for personalized risk stratification and therapy monitoring to ensure that the right treatment is given to the right patient at the right time and for the right duration.

[1054] Example 11: Whole exome profiling of plasma cDNA captures mutational signatures of preclinical relapse to monitor disease evolution.

[1055] The purpose of this case study was to evaluate the use of whole exome sequencing of free-cell DNA (PES-cDNA) of plasma to study mutational signatures and clonal evolution in patients with advanced cancers or patients with a high burden of circulating tumor DNA (ctDNA). Specifically, here we demonstrated the use of PES-scDNA profiling to identify preclinical metastases in patients with primary breast cancer.

[1056] METHODS

[1057] Forty-nine patients with primary breast cancer were recruited after surgery and adjuvant therapy. Serial plasma samples were collected every six months for ctDNA analysis by ultra-deep sequencing with Natera's Signatera workflow using patient-specific assays targeting 16 variants. Whole exome sequencing was performed on plasma cDNA from all 17 relapsed patients before, during, or after clinical relapse to determine the concordance between variants identified in plasma and tumor biopsies and to understand tumor evolution during disease progression.

[1058] RESULTS

[1059] Preliminary analysis of PES-scDNA profiles from 3 patients with relapse showed a high degree of agreement between patient-specific variants identified in tumor biopsy and in plasma. Plasma PES also identified 34 of the 35 variants detected by Signatera and demonstrated highly concordant variant allele frequencies (HAFs). One variant that was not detected by PES-scDNA was previously detected by Signatera at 0.2% NVA.

[1060] CONCLUSIONS

[1061] These examples demonstrated that plasma PES could detect molecular residual disease in patients with primary breast cancer. Plasma analysis of PES potentially provides evidence of cancer evolution, which may be important for treatment decisions.

[1062] Example 12: Use of circulating tumor DNA (ctDNA) as a molecular biomarker to assess treatment response in lymphoma.

[1063] Here, in a pilot biopharmaceutical study, we assessed the potential of a personalized, tumor-specific, multiplex SSP-PCR-based approach (Signatera™) to detect ctDNA during a patient's treatment regimen to correlate the presence of ctDNA with overall clinical response in a cohort with non-Hodgkin's lymphoma (NHL).

[1064] METHODS

[1065] Blood samples from 8 patients with non-Hodgkin's lymphoma (NHL) (6 diffuse large B-cell lymphomas and 2 follicular lymphomas) were available for ctDNA analysis. Patient-specific somatic variants were identified by analyzing whole exome sequencing (WES) data from primary tumor biopsies and matched normal samples. Plasma samples were then analyzed in a blinded manner using the corresponding individual 16-plex assays using the Signatera workflow. Samples were considered ctDNA positive if at least two patient-specific targets met the qualifying confidence threshold.

[1066] RESULTS

[1067] For non-Hodgkin's lymphoma (NHL), 14.9 ng (median) (range 2.25-685 ng) of cDNA were extracted from 2.5 ml (median) of plasma. ctDNA was detected at 5 plasma time points in 4 patients. Of the 5 ctDNA+ plasma samples, 4 plasma samples were correlated with either clinically progressive disease or partial response to therapy at blood draw. Four patients without ctDNA detection at any time during blood sampling showed a complete clinical response.

[1068] CONCLUSIONS

[1069] Scalable patient-specific ctDNA monitoring assays can be used to establish baseline, monitor therapy, and detect relapse. Signatera's highly sensitive ctDNA detection assay provides a non-invasive means of monitoring current standards of care.

Claims (56)

1. A method for monitoring and detecting minimal residual disease in breast cancer, comprising: selecting a set of at least 16 patient-specific single nucleotide variant loci based on somatic mutations found in the tumor sample of the patient who has been diagnosed with breast cancer; collecting one or more blood or urine samples over time from a patient after the patient has undergone chemotherapy, radiation therapy, immunotherapy, stem cell transplantation, bone marrow transplantation, photodynamic therapy, palliative care, surgery and/or first-line chemotherapy; obtaining a set of amplicons by performing a multiplex amplification reaction on nucleic acids isolated from each blood or urine sample or a fraction thereof, where each amplicon from the set of amplicons covers at least one single nucleotide variant locus from a set of patient-specific single nucleotide variant loci associated with breast cancer glands; And performing high-throughput sequencing to determine the sequence of at least a segment of each amplicon from the set of amplicons that contains a patient-specific single nucleotide variant locus, where high-throughput sequencing is performed with a read depth of at least 100,000 per amplicon and detection of two or more patient-specific single nucleotide variants above a confidence threshold of 0.97, wherein the detection of two or more patient-specific single nucleotide variants above the confidence threshold of 0.97 in a blood sample indicates circulating tumor DNA (ctDNA) and the presence of minimal residual disease, and the method has a specificity of at least 99.5% in detecting minimal residual disease breast cancer. 2. The method of claim 1, wherein the method detects patient-specific single nucleotide variants in at least 95% of patients with early relapse or metastasis of HER2+ breast cancer. 3. The method of claim 1, wherein the method detects patient-specific single nucleotide variants in at least 95% of patients with early relapse or metastasis of triple-negative breast cancer. 4. The method of claim 1, wherein the method detects patient-specific single nucleotide variants in at least 80% of patients with early relapse or metastasis of HR+/HER2- breast cancer. 5. The method of claim 1, wherein the method detects patient-specific single nucleotide variants in patients with early relapse or metastasis of breast cancer at least 200 days before clinical recurrence or metastasis of breast cancer detected by imaging , and/or at least 100 days before CA15-3 levels rise. 6. The method of claim 1, wherein the method detects patient-specific single nucleotide variants in patients with early relapse or metastasis of HER2+ breast cancer at least 150 days before clinical relapse or metastasis of HER2+ breast cancer detected by imaging , and/or at least 100 days before CA15-3 levels rise. 7. The method of claim 1, wherein the method detects patient-specific single nucleotide variants in patients with early relapse or metastasis of triple-negative breast cancer at least 200 days before clinical relapse or metastasis of triple-negative breast cancer detected with by imaging, and/or at least 100 days before CA15-3 levels rise. 8. The method of claim 1, wherein the method detects patient-specific single nucleotide variants in patients with early relapse or metastasis of HR+/HER2- breast cancer at least 250 days before clinical relapse or metastasis of HR+/HER2 breast cancer - determined by imaging and/or at least 100 days before the increase in CA15-3 levels. 9. The method according to claim 1, characterized in that the method detects the absence of patient-specific single nucleotide variants in at least 99% of patients without early relapse or metastasis of breast cancer. 10. The method according to claim 1, characterized in that the method detects the absence of patient-specific single nucleotide variants in at least 99% of patients without early relapse or metastasis of HER2+ breast cancer. 11. The method according to claim 1, characterized in that the method detects the absence of patient-specific single nucleotide variants in at least 99% of patients without early relapse or metastasis of triple-negative breast cancer. 12. The method according to claim 1, characterized in that the method detects the absence of patient-specific single nucleotide variants in at least 99% of patients without early relapse or metastasis of HR+/HER2- breast cancer. 13. The method of claim 1, wherein the method has a specificity of at least 99.5% in detecting early recurrence or metastasis of HER2+ breast cancer when two or more patient-specific single nucleotide variants are detected above the confidence threshold of 0.97. 14. The method of claim 1, wherein the method has a specificity of at least 99.5% for detecting early relapse or metastasis of triple-negative breast cancer when two or more patient-specific single nucleotide variants are detected above the confidence threshold of 0.97 . 15. The method according to claim 1, characterized in that it has a specificity of at least 99.5% in detecting early relapse or metastasis of HR+/HER2- breast cancer when two or more patient-specific single nucleotide variants are detected above the confidence threshold of 0, 97. 16. A method for monitoring and detecting early relapse or metastasis of colorectal cancer, comprising: collecting one or more blood samples over time from a patient who is undergoing treatment for colorectal cancer; obtaining a set of amplicons by performing a multiplex amplification reaction of nucleic acids isolated from a blood sample or fraction thereof from a patient treated for colorectal cancer, where each amplicon from the set of amplicons covers at least one single nucleotide variant locus of a set of patient-specific single nucleotide variant loci related with colorectal cancer; And performing high-throughput sequencing to determine the sequence of at least a segment of each amplicon from a set of amplicons that contains a patient-specific single nucleotide variant locus, wherein high-throughput sequencing is performed with a read depth of at least 100,000 per amplicon and detects two or more patient-specific single nucleotide variants above a threshold confidence level of 0.97, with the detection of two or more patient-specific single nucleotide variants above the 0.97 confidence threshold indicating circulating tumor DNA (ctDNA) and early relapse or metastasis of colorectal cancer, where the presence of two or more patient-specific single nucleotide variants detects early recurrence or metastasis of colorectal cancer in a patient with cancer with a lead time of at least 8.7 months compared with clinical recurrence or metastasis of cancer detected by radiographic imaging and/or elevated CEA levels. 17. The method of claim 16, wherein nucleic acids are isolated from the patient's tumor and somatic mutations are identified in the tumor for a set of patient-specific loci of single nucleotide variants before determining the sequence of at least a segment of each amplicon from the set of amplicons for a blood sample or a fraction thereof . 18. The method of claim 16, wherein at least 5 NNVs are detected and wherein the presence of at least 5 NNVs indicates circulating tumor DNA (ctDNA) and early relapse or metastasis of colorectal cancer. 19. The method according to claim 16, characterized in that the individual has undergone surgical treatment before the blood sample is isolated. 20. The method of claim 16, wherein the individual is treated with chemotherapy before the blood sample is isolated. 21. The method of claim 16, wherein the individual is treated with radiotherapy before the blood sample is isolated. 22. The method of claim 16, wherein the individual is treated with adjuvant therapy before the blood sample is isolated. 23. The method of claim 16, wherein the method further comprises administering to the individual a compound where the compound is known to be specifically effective in treating colorectal cancer having one or more of the specified single nucleotide variants. 24. The method of claim 16, wherein the method further includes determining the variant allele frequency for each of the single nucleotide variants based on the sequence determination. 25. The method of claim 1, wherein the treatment plan for breast cancer is determined based on determinations of the variant allele frequency. 26. The method of claim 16, wherein the method further comprises administering to the individual a compound where the compound is known to be particularly effective in the treatment of colorectal cancer having one of the single nucleotide variants with a variable allele frequency greater than at least half of the other specified single nucleotide variants . 27. The method according to claim 16, characterized in that the sequence is determined by high-throughput DNA sequencing of multiple loci of single-nucleotide variations. 28. The method of claim 16, wherein the method further includes detecting a clonal single nucleotide variant in colorectal cancer by determining the frequency of the variant allele for each of the SNV loci based on multiple copy sequencing of the amplicon series, where the higher relative frequency of the allele compared to other single nucleotide variants at multiple single nucleotide variant loci indicates a clonal single nucleotide variant in colorectal cancer. 29. The method of claim 28, wherein the method further comprises administering to the individual a compound that targets one or more clonal single nucleotide variants, but not other single nucleotide variants. 30. The method according to claim 28, characterized in that the frequency of the variant allele is more than 1.0% indicating a clonal single-nucleotide variant. 31. The method of claim 16, wherein the method further includes forming an amplification reaction mixture by combining polymerase, nucleotide triphosphates, nucleic acid fragments from a nucleic acid library generated from the sample, and a set of primers, each of which binds within 150 base pairs single nucleotide variant loci, or a set of primer pairs each covering a region of 160 base pairs or less containing single nucleotide variant loci, and subjecting the amplification reaction mixture to amplification conditions to create a set of amplicons. 32. Method according to any one of paragraphs. 16-31, wherein determining whether a single nucleotide variant is present in a sample includes determining a confidence value for each allele determination at each of the set of single nucleotide variation loci based at least in part on the read depth for the loci. 33. Method according to any one of paragraphs. 16-31, wherein the set of single nucleotide variation loci contains all single nucleotide variation loci identified in the TCGA and COSMIC data sets for colorectal cancer. 34. Method according to any one of paragraphs. 16-31, wherein the set of single nucleotide variation sites contains all single nucleotide variation sites identified in the TCGA and COSMIC data sets for colorectal cancer. 35. Method according to any one of paragraphs. 16-31, characterized in that the set of single nucleotide variant loci contains from 25 to 1000 single nucleotide variation loci that are known to be associated with colorectal cancer. 36. Method according to any one of paragraphs. 16-31, characterized in that the efficiency and error rate per cycle are determined for each amplification reaction of the multiplex reaction for amplification of single nucleotide variation loci, and the efficiency and error rate are used to determine the presence of a single nucleotide variant in a set of single nucleotide variant loci in the sample. 37. Method according to any one of paragraphs. 16-31, characterized in that the amplification reaction is a PCR reaction, and the annealing temperature is 1-15 °C higher than the melting temperature of at least 50% of the primers from the primer set. 38. Method according to any one of paragraphs. 16-31, characterized in that the amplification reaction is a PCR reaction, and the duration of the annealing stage in the PCR reaction ranges from 15 to 120 minutes. 39. Method according to any one of paragraphs. 16-31, characterized in that the amplification reaction is a PCR reaction, and the duration of the annealing stage in the PCR reaction ranges from 15 to 120 minutes. 40. Method according to any one of paragraphs. 16-31, characterized in that the primer concentration in the amplification reaction is from 1 to 10 nM. 41. Method according to any one of paragraphs. 16-31, characterized in that the primers in the primer set are designed to minimize the formation of a primer dimer. 42. Method according to any one of paragraphs. 16-31, characterized in that the amplification reaction is a PCR reaction, the annealing temperature is 1-15 °C higher than the melting temperature of at least 50% of the primers of the primer set, the duration of the annealing stage in the PCR reaction is from 15 to 120 minutes, The primer concentration in the amplification reaction ranges from 1 to 10 nM, and the primers in the primer set are designed to minimize primer dimer formation. 43. Method according to any one of paragraphs. 16-31, characterized in that the multiplex amplification reaction is carried out under the conditions of a limiting primer. 44. Method according to any one of paragraphs. 1-15, further comprising whole exome sequencing of the patient's tumor sample to detect somatic mutations associated with breast cancer, and selecting a set of at least 16 patient-specific single nucleotide variant loci corresponding to the somatic mutations found in the tumor sample. 45. Method according to any one of paragraphs. 16-43, comprising selecting a set of at least 16 patient-specific single nucleotide variant loci based on somatic mutations found in a tumor sample from a patient diagnosed with colorectal cancer. 46. The method of claim 45, further comprising whole exome sequencing of the patient's tumor sample to detect somatic mutations associated with colorectal cancer, and selecting a set of at least 16 patient-specific single nucleotide variant loci corresponding to the somatic mutations found in the tumor sample.