WO2021228888A1

WO2021228888A1 - Biomarkers for hyperprogressive disease and therapy response after immunotherapy

Info

Publication number: WO2021228888A1
Application number: PCT/EP2021/062527
Authority: WO
Inventors: Andrey KHMELEVSKIY
Original assignee: Asylia Diagnostics
Current assignee: Asylia Diagnostics
Priority date: 2020-05-12
Filing date: 2021-05-11
Publication date: 2021-11-18
Anticipated expiration: 2022-11-12

Abstract

The invention relates to the field of diagnostics to guide cancer immunotherapy. More specifically, the invention provides a method for determining the likelihood of hyperprogression and/or progressive disease of cancer cells in response to immunotherapy, such as immune checkpoint inhibitor treatment. Furthermore, the invention also provides a method for determining the likelihood of a positive response to immunotherapy, such as immune checkpoint inhibitor treatment.

Description

BIOMARKERS FOR HYPERPROGRESSIVE DISEASE AND THERAPY RESPONSE AFTER

IMMUNOTHERAPY

FIELD OF THE INVENTION

BACKGROUND TO THE INVENTION

Cancer is a major public health problem. In 2015, 1.3 million people died from cancer in the European Union, which equated to more than one quarter (25.4%) of the total number of deaths. Many treatments have been devised for various cancers.

Immune checkpoint inhibitors (ICIs) have changed the treatment landscape for many tumor types, particularly in the metastatic setting. The development of these ICIs targeting cytotoxic T- lymphocyte antigen-4 (CTLA-4) and programmed cell death-1 (PD-1/PD-L1) has significantly improved the treatment of a variety of cancers, like melanoma and non-small cell lung carcinoma (NSCLC). ICIs enhance antitumor immunity by blocking the prototypical immune checkpoint receptors that exist both on immune cells and on tumor cells and exhibit negative regulation of T- cell function.

Although these inhibitors can lead to remarkable responses, patients with cancer respond differently to an ICI treatment. Some patients quickly improve and are able to overcome the disease, counting for about 25% to 30% of the treated patients, but 50% of the treated patients see no durable benefits. Even more alarming, a growing number of studies shows that immunotherapy may accelerate tumor progression in a significant subset of patients ranging from 4% to 29% across multiple histologies and lead to so-called hyperprogression.

Currently, no diagnostics are available to prognostically identify which patients are going to experience hyperprogressive disease or be non-responders to the ICI therapy. There are also ongoing efforts to utilize and bring into clinic biomarkers based on tumor mutation burden (TMB) and microsatellite instability (MSI), but both of them target general efficacy of the ICIs and do not attempt to predict hyperprogressive disease syndrome or non-response per se. Notably, at best the accuracy of such biomarkers falls below 63% and it does not provide much insight into the biological mechanism underlying the non-responsiveness to ICIs. Despite the pivotal interest of recognizing subjects at increased risk of hyperprogressive disease, only MDM2 amplification and EGFR aberrations and changes in Gc/GcR fragments have been described as potential biomarkers and require further validation (Kato et al. 2017). W02014151006 provides biomarkers for patient selection and prognosis in cancer. However, this patent application is limited to predicting the responsiveness of an individual with a disease to treatment with a PD-L1 axis binding antagonist.

W02019012147 proposes a radiomics based biomarker for detecting the presence and density of tumor infiltrating CD8 T-cells to prognose the survival and/or the treatment efficiency of cancer patients treated with immunotherapy such as anti-PD-1/PD-L1 monotherapy.

US20180107786 discloses a method for generating an immune score based on tumor infiltrating lymphocytes, T-cell receptor signaling and mutation burden.

Initially, the definition of hyperprogressive disease (HPD) varied between clinicians and research groups. In some studies, a hyperprogressive disease was measured as a time-to-treatment failure (TTF) <2 months, >50% increase in tumor burden compared to pre-immunotherapy imaging (Kato et al., 2017). In other studies, it was defined as a >2-fold increase in tumor progression pace or TGKR > 2-29%, where TGKR is the ratio of the slope of tumor growth before treatment and the slope of tumor growth on treatment. To this end, parameters such as tumor growth rate (TGR), tumor growth kinetics (TGK), and time to treatment failure (TTG) have been proposed (Lo Russo et al., 2019). Yet, this inconsistency was recently addressed by a clinical consortium (Matos et al. 2020) which defined a clear way to capture HPD using RECIST 1.1 criteria that is intuitive and easy to use in daily clinical practice. The HPD is now defined based on RECIST as PD in the first 8 weeks after treatment initiation and minimum increase in the measurable lesions of 10 mm plus: a) increase of >40% in sum of target lesions compared to baseline (which represents doubling in unidimensional target lesions compared to classical RECIST PD criterion [20%]); and/or b) increase of >20% in sum of target lesions compared to baseline (the classical RECIST PD criterion) plus the appearance of new lesions in at least 2 different organs. Unlike pseudoprogression, patients displaying hyper progression present worse survival outcomes.

At present, the occurrence rate of HPD varies depending on the cancer subtype. Still, overall its incidence rate is ~15% (9-27%), making it a considerable risk for patients, clinical oncologists, and drug development companies (Borcoman et al. 2019; Champiat et al., 2018). Unlike pseudoprogression, patients displaying HPD present worse survival outcomes (Fuentes-Antras et al. 2018).

Accordingly, there is a need for assays capable of characterizing an immunological tumor microenvironment for companion diagnostics development and therapeutic decisions. In particular, there is a high need for biomarkers to identify hyperprogressive disease and to identify responders and non-responders to immunotherapy treatment.

SUMMARY OF THE INVENTION

The inventors have addressed the challenges for determining whether a subject is predisposed for generating hyperprogressive disease when receiving treatment with an immunotherapeutic agent by identifying biomarkers that are able to predict whether a subject will develop hyperprogressive disease after receiving treatment with an immunotherapeutic agent.

In a second aspect of the application, the inventors identified biomarkers that are able to predict whether a subject will respond or not respond to treatment with an immunotherapeutic agent.

Accordingly, a first aspect of the present invention relates to methods for determining whether a subject is predisposed for generating hyperprogressive disease when receiving treatment with an immunotherapeutic agent or for deciding whether a subject is eligible for an treatment with an immunotherapeutic agent. Said methods are based on the analysis of the expression level of one or more biomarkers, more specifically based on the analysis of the RNA expression level of one or more protein coding genes, and wherein on the basis of said expression level it is predicted whether the subject will develop a predisposition for generating hyperprogressive disease and/or whether the subject is eligible for treatment with an immunotherapeutic agent.

In one embodiment, the invention thus relates to a method for determining whether a subject is predisposed for generating hyperprogressive disease when receiving treatment with an immunotherapeutic agent or for deciding whether a subject is eligible for treatment with an immunotherapeutic agent, such as an immune checkpoint inhibitor, said method comprising analyzing the expression level of one or more of the biomarkers, preferably protein coding genes, selected from IL-17A, IFN1 , IFNB1 , IL-6, ICOSLG, CCL20, ID01 , FGF2, CXCL8, NRP1 , MX1 , VEGFA, CXCL12 and TGFB1 in a sample of the subject. In a further embodiment, the present invention relates to a method for determining whether a subject is predisposed for generating hyperprogressive disease when receiving treatment with an immunotherapeutic agent, or for deciding whether a subject is eligible for treatment with an immunotherapeutic agent, said method comprising analyzing the expression level of the biomarkers IL-17A, IFN1 , IFNB1 , IL-6 in a sample of the subject. In a further embodiment, the method further comprises the analysis of the expression level of one or more of the biomarkers selected from the group consisting of ICOSLG, CCL20, ID01 , FGF2, CXCL8, NRP1 , MX1 , VEGFA, CXCL12 and TGFB1 in a sample of the subject. Thus, in this embodiment, the expression level of the biomarkers IL-17A, IFN1 , IFNB1 , IL- 6 is analyzed, in combination with the expression level of one or more of the biomarkers selected from the group consisting of ICOSLG, CCL20, ID01 , FGF2, CXCL8, NRP1 , MX1 , VEGFA, CXCL12 and TGFB1 in order to determine whether the subject is predisposed for generating hyperprogressive disease when receiving treatment with an immunotherapeutic agent or for deciding whether the subject is eligible for treatment with an immunotherapeutic agent. In a further embodiment, the expression level of the biomarkers IL-17A, IFN1 , IFNB1 , IL-6 is analyzed, in combination with the expression level of at least 3, at least 4, at least 5, at least 6, at least 7, at least 8 or at least 9, of the biomarkers selected from the group consisting of ICOSLG, CCL20, ID01 , FGF2, CXCL8, NRP1 , MX1 , VEGFA, CXCL12 and TGFB1 . In yet another embodiment, a method is provided for determining whether a subject is predisposed for generating hyperprogressive disease when receiving treatment with an immunotherapeutic agent, or for deciding whether a subject is eligible for treatment with an immunotherapeutic agent, said method comprising analyzing the expression level of the biomarkers IL-17A, IFN1 , IFNB1 , IL-6, ICOSLG, CCL20, ID01 , FGF2, CXCL8, NRP1 , MX1 , VEGFA, CXCL12 and TGFB1 in a sample of the subject. In a further embodiment, the expression level of the biomarkers is compared to a corresponding reference value or threshold value that is characteristic of a subject with a known response to the treatment with the immunotherapeutic agent.

In some embodiments ofthe method, the expression level of the analyzed biomarkers is compared to a corresponding reference value or threshold value that is characteristic of a subject with a known response to treatment with the immunotherapeutic agent.

In some embodiments, a risk score is obtained on the basis ofthe expression level ofthe analyzed biomarkers, said risk score representing the likelihood for developing hyperprogressive disease in the subject or for responding to the treatment with the immunotherapeutic agent by the subject.

In another aspect, a kit for determining whether a subject is predisposed for generating hyperprogressive disease when receiving treatment with an immunotherapeutic agent or for deciding whether a subject is eligible for treatment with an immunotherapeutic agent is provided. Said kit comprises means for measuring the expression level of one or more of the biomarkers, preferably protein coding genes, IL-17A, IFNA1 , IFNB and IL-6 in a sample of the subject; optionally further complemented with means for measuring the expression level of one or more of the biomarkers, preferably protein coding genes, ICOSLG, CCL20, ID01 , FGF2, CXCL8, NRP1 , MX1 , VEGFA, CXCL12 and TGFB1 ; and/or a reference value or threshold value for each of the biomarkers, preferably protein coding genes. Also the use of said kit in a method for determining whether a subject is predisposed for generating hyperprogressive disease when receiving treatment with an immunotherapeutic agent or for deciding whether a subject is eligible for treatment with an immunotherapeutic agent is disclosed.

Another aspect of the present application relates to a method for determining whether a subject is predisposed for responding to treatment with an immunotherapeutic agent. Said method comprises the analysis ofthe expression level of one or more biomarkers, preferably protein coding genes selected from TLR9, IL23A, CLEC4C, CCR4, SIGLEC6, IL3RA, IL12A, TLR7, CXCR3, AXL, TLR8, SIGLEC1 , TCF4, CD22, ITGAX, and FCGR3A, in a sample of the subject. In a further embodiment, the expression level of at least three, at least four, at least five, or at least six, of the biomarkers, preferably protein coding genes, selected from TLR9, IL23A, CLEC4C, CCR4, SIGLEC6, IL3RA, IL12A, TLR7, CXCR3, AXL, TLR8, SIGLEC1 , TCF4, CD22, ITGAX, and FCGR3A, is analyzed in a sample of the subject and on the basis of said analysis a risk score is obtained, said risk score representing the likelihood for responding to the treatment with the immunotherapeutic agent in the subject. In a further embodiment, the method comprises analyzing the expression level of the biomarkers, preferably protein coding genes, TLR9, IL-23A, CLEC4C, and CCR4, in a sample of the subject. In a further embodiment, the method further comprises analyzing the expression level of the biomarkers, preferably protein coding genes, TLR9, IL-23A, CLEC4C, and CCR4, in combination with the analysis of the expression level of one or more of the biomarkers, preferably protein coding genes, selected from the group consisting of SIGLEC6, IL3RA, IL12A, TLR7, CXCR3, AXL, TLR8, SIGLEC1 , TCF4, CD22, ITGAX, and FCGR3A, in a sample of the subject in order to determine whether a subject is predisposed for responding to treatment with an immunotherapeutic agent. In a further embodiment, the expression level of the biomarkers, preferably protein coding genes, TLR9, IL-23A, CLEC4C, and CCR4, is analyzed, in combination with the expression level of at least 3, at least 4, at least 5, at least 6, at least 7, at least 8, at least 10, or at least 11 , of the biomarkers, preferably protein coding genes, selected from the group consisting of SIGLEC6, IL3RA, IL12A, TLR7, CXCR3, AXL, TLR8, SIGLEC1 , TCF4, CD22, ITGAX, and FCGR3A.

In yet another embodiment, a method is provided for determining whether a subject is predisposed for responding to treatment with an immunotherapeutic agent, said method comprising analyzing the expression level of the biomarkers, preferably protein coding genes, TLR9, IL-23A, CLEC4C, CCR4, SIGLEC6, IL3RA, IL12A, TLR7, CXCR3, AXL, TLR8, SIGLEC1 , TCF4, CD22, ITGAX, and FCGR3A, in a sample of the subject.

In some embodiments of the method, the expression level of the analyzed biomarkers is compared to a corresponding reference value or threshold value that is characteristic of a subject with a known response to treatment with the immunotherapeutic agent.

In some embodiments, a risk score is obtained on the basis of the expression level of the analyzed biomarkers, said risk score representing the likelihood for responding to the treatment with the immunotherapeutic agent in the subject.

In another aspect, a kit for determining whether a subject is predisposed for responding to treatment with an immunotherapeutic agent is provided. Said kit comprises means for measuring the expression level of one or more of the biomarkers, preferably protein coding genes, TLR9, IL- 23A, CLEC4C, and CCR4 in a sample of the subject; optionally further complemented with means for measuring the expression level of one or more of the biomarkers, preferably protein coding genes, SIGLEC6, IL3RA, IL12A, TLR7, CXCR3, AXL, TLR8, SIGLEC1 , TCF4, CD22, ITGAX, and FCGR3A; and/ora reference value orthreshold value for each of the biomarkers, preferably protein coding genes. Also the use of said kit in a method for determining whether a subject is predisposed for responding to treatment with an immunotherapeutic agent is disclosed.

In some embodiments of the methods of the present application, the risk score is further improved by immunohistochemistry data of the tumor.

In some embodiments, the methods of the application are further characterized in that the risk score is obtained and calculated using a pre-trained machine learning model and using the expression level of the biomarkers as input values. In particular, the risk score is produced by a machine learning model that uses the normalized expression levels of the biomarker signature as input data. More specifically, the machine learning model uses a pre-trained architecture to calculate the risk score from 0 to 1 . In some embodiments, the methods according to the present application are characterized in that the immunotherapeutic agent is an immune checkpoint inhibitor or a combination of several different immune checkpoint inhibitors. In some further preferred embodiments, the immunotherapeutic agent is selected from the group consisting of a PD-1 targeting agent, a PD- L1 targeting agent, and a CTLA-4 targeting agent, or a combination thereof. In some further preferred embodiments, a PD-1 targeting agent or a PD-L1 targeting agent is combined with a CTLA-4 targeting agent as immunotherapeutic treatment.

As outlined above, in the methods of the present application, a sample of a subject is analyzed to quantify the expression level of one or more biomarkers; in particular one or more protein coding genes. Said sample is preferably a tumor tissue sample, such as a biopsy sample, derived from the subject; preferably the subject diagnosed with cancer. In an even more preferred embodiment, said sample is a tumor tissue sample derived from cancer patients diagnosed with bladder cancer, kidney cancer, liver cancer, lung cancer, pancreatic cancer, prostate cancer, thyroid cancer, uterine cancer, ovarian cancer, colorectal cancer, breast cancer, head and neck cancer, skin cancer; even more preferably cancer patients diagnosed with skin cancer or melanoma, lung cancer such as lung adenocarcinoma, lung squamous cell carcinoma, non-small cell lung carcinoma (NSCLC).

BRIEF DESCRIPTION OF THE DRAWINGS

Figure 1 . Boxplots for RNA expression for various membrane proteins, interleukins, and cytokines expressed by myeloid cells split by different response types.

Figure 2. Bootstrap performance for each gene signature with 95% Cl intervals on label randomization for HPD vs non-HPD signature. Blue dashed lines represent Cl boundaries and red line is a real fitted value of the signature.

Figure 3. Bootstrap performance for each gene signature with 95% Cl intervals on label randomization for responders vs non-responders signature. Blue dashed lines represent Cl boundaries and red line is real fitted value of the signature.

Figure 4. Bootstrap performance for each gene signature with 95% Cl intervals on gene randomization for HPD vs non-HPD signature. Blue dashed lines represent Cl boundaries and red line is real fitted value of the signature.

Figure 5. Bootstrap performance for each gene signature with 95% Cl intervals on gene randomization for responders vs non-responders signature. Blue dashed lines represent Cl boundaries and the red line is real fitted value of the signature.

Figure 6. Feature importance (absolute values) for the model highlighting the most important features for the target classes. (A) - R vs NR; (B) - HP vs non HPD.

Figure 7. Heatmap of differentially expressed genes showing the separation of HPD vs. non-HPD cases and clustering of patients according to the biomarker signature.

Figure 8. Individual boxplots and ANOVA tests demonstrating separation of HPD vs. non-HPD cases for PD1/PD-L1 treated patient cohort for gene expression of the top 4 markers (IL-17A, IFNA1 , IFNB1 and IL-6).

Figure 9. Discriminatory power of predictive signature for HPD vs non-HPD classes. Gene expression of the top 4 markers (IL-17A, IFNA1 , IFNB1 and IL-6) contribute over 90% of explanatory power.

Figure 10. Significantly enriched pathways based on top differentially expressed biomarkers (downregulated).

DETAILED DESCRIPTION OF THE INVENTION

Unless otherwise defined, all terms used in disclosing the invention, including technical and scientific terms, have the meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. By means of further guidance, term definitions are included to better appreciate the teaching of the present invention.

As used herein, the following terms have the following meanings:

“A”, “an”, and “the” as used herein refers to both singular and plural referents unless the context clearly dictates otherwise. By way of example, “a compartment” refers to one or more than one compartment.

“About” as used herein referring to a measurable value such as a parameter, an amount, a temporal duration, and the like, is meant to encompass variations of +/-20% or less, preferably +/- 10% or less, more preferably +/-5% or less, even more preferably +/-1% or less, and still more preferably +/-0.1 % or less of and from the specified value, in so far such variations are appropriate to perform in the disclosed invention. However, it is to be understood that the value to which the modifier “about” refers is itself also specifically disclosed.

“Comprise”, “comprising”, and “comprises” and “comprised of as used herein are synonymous with “include”, “including”, “includes” or “contain”, “containing”, “contains” and are inclusive or open-ended terms that specifies the presence of what follows e.g. component and do not exclude or preclude the presence of additional, non-recited components, features, elements, members, steps, known in the art or disclosed therein.

Furthermore, the terms first, second, third and the like in the description and in the claims, are used for distinguishing between similar elements and not necessarily for describing a sequential or chronological order, unless specified. It is to be understood that the terms so used are interchangeable under appropriate circumstances and that the embodiments of the invention described herein are capable of operation in other sequences than described or illustrated herein. The recitation of numerical ranges by endpoints includes all numbers and fractions subsumed within that range, as well as the recited endpoints.

The expression “% by weight”, “weight percent”, “%wf or “wt%”, here and throughout the description unless otherwise defined, refers to the relative weight of the respective component based on the overall weight of the formulation.

Whereas the terms “one or more” or “at least one”, such as one or more or at least one member(s) of a group of members, is clear perse, by means of further exemplification, the term encompasses inter alia a reference to any one of said members, or to any two or more of said members, such as, e.g., any >3, >4, >5, >6 or >7 etc. of said members, and up to all said members.

"Gene" is generally used herein to encompass a polynucleotide that encodes a gene product, e.g., a nucleic acid sequence defining an open reading frame.

"Diagnosis" as used herein generally includes determination of a subject's susceptibility to a disease or disorder, determination as to whether a subject is presently affected by a disease or disorder, as well as to the prognosis of a subject affected by a disease or disorder.

The terms "individual," "subject," "host," and "patient," used interchangeably herein and refer to any mammalian subject for whom diagnosis, treatment, or therapy is desired, particularly humans. The term "sample" or "biological sample" encompasses a variety of sample types obtained from an organism and can be used in a diagnostic or monitoring assay. The term encompasses blood and other liquid samples of biological origin, solid tissue samples, such as a biopsy specimen or tissue cultures or cells derived therefrom and the progeny thereof. The term encompasses samples that have been manipulated in any way after their procurement, such as by treatment with reagents, solubilization, or enrichment for certain components. The term encompasses a clinical sample, and also includes cells in cell culture, cell supernatants, cell lysates, serum, plasma, biological fluids, and tissue samples.

The terms "cancer", "neoplasm", "tumor", and "carcinoma", are used interchangeably herein to refer to cells which exhibit relatively autonomous growth, so that they exhibit an aberrant growth phenotype characterized by a significant loss of control of cell proliferation. In general, cells of interest for detection in the present application include immune cells in the tumor microenvironment.

The term "differentially expressed gene" is intended to encompass a polynucleotide that represents or corresponds to a gene that is differentially expressed in a tumour cell when compared with another cell of the same cell type. Such differentially expressed gene may include an open reading frame encoding a gene product (e.g., a polypeptide), as well as introns of such genes and adjacent 5' and 3' non-coding nucleotide sequences involved in the regulation of expression, up to about 20 kb beyond the coding region, but possibly further in either direction. In general, a difference in expression level associated with a decrease in expression level of at least about 25%, usually at least about 50% to 75%, more usually at least about 90% or more is indicative of a differentially expressed gene of interest, i.e., a gene that is downregulated in the test sample relative to a control sample. Furthermore, a difference in expression level associated with an increase in expression of at least about 25%, usually at least about 50% to 75%, more usually at least about 90% and may be at least about 1 ,5-fold, usually at least about 2-fold to about 10-fold, and may be about 100-fold to about 1 ,000-fold increase relative to a control sample is indicative of a differentially expressed gene of interest, i.e., an overexpressed or upregulated gene. "Differentially expressed polynucleotide" as used herein means a nucleic acid molecule (RNA or DNA) comprising a sequence that represents a differentially expressed gene, e.g., the differentially expressed polynucleotide comprises a sequence (e.g., an open reading frame encoding a gene product) that uniquely identifies a differentially expressed gene so that detection of the differentially expressed polynucleotide in a sample is correlated with the presence of a differentially expressed gene or gene product of a differentially expressed gene in a sample. "Differentially expressed polynucleotides" is also meant to encompass fragments of the disclosed polynucleotides, e.g., fragments retaining biological activity, as well as nucleic acids that are homologous, substantially similar, or substantially identical (e.g., having about 90% sequence identity) to the disclosed polynucleotides.

The term “microenvironment,” as used herein, may refer to the tumor microenvironment as a whole or to an individual subset of cells within the microenvironment. Immune cells found within the tumor microenvironment are macrophages, monocytes, mast cells, helper T cells, cytotoxic T cells, regulatory T cells, natural killer cells, myeloid-derived suppressor cells, regulatory B cells, neutrophils, dendritic cells, and fibroblasts. The tumor microenvironment is generally defined as a complex mixture of cells, soluble factors, signaling molecules, extracellular matrices, and mechanical cues that promote neoplastic transformation, support tumor growth and invasion, and protect the tumor from host immunity.

As used herein, the terms “marker(s)”, “biomarker(s)” refer to a gene or genes or a protein, polypeptide, or peptide expressed by the gene or genes whose expression level, alone or in combination with other genes, is correlated with the risk of developing hyperprogression or be non- responsive to an immunotherapeutic treatment. The correlation can relate to either an increased or decreased expression of the gene (e.g. increased or decreased levels of mRNA orthe peptide encoded by the gene).

As used herein, the term “in vitro” refers to an artificial environment and to processes or reactions that occur within an artificial environment. In vitro environments can consist of, but are not limited to, test tubes and cell culture. The term “in vivo” refers to the natural environment (e.g., an animal or a cell) and to processes or reaction that occur within a natural environment. The term “in silico” refer to artificial environment, wherein a procedure is performed or modelled using a computer system, thereby partly or entirely avoiding the need of physically manipulating the data (e.g. genes, polynucleotides, proteins, ...).

“Hyperprogressive”, “hyperprogressor”, “hyperprogression” is defined as a syndrome, which is measured as: a time-to-treatment failure (TTF) < 2 months; or/and > 50% increase in tumor burden compared to pre-immunotherapy imaging; or/and a > 2-fold increase in tumor progression pace; or/and tumor growth kinetics (TGK) ratio > 2 - 29%, where TGKR is the ratio of the slope of tumor growth before treatment and the slope of tumor growth on treatment; or/and as overall survival less than 3 months after commencing treatment.

All references cited in the present specification are hereby incorporated by reference in their entirety. In particular, the teachings of all references herein specifically referred to are incorporated by reference.

Unless otherwise defined, all terms used in disclosing the invention, including technical and scientific terms, have the meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. By means of further guidance, definitions for the terms used in the description are included to better appreciate the teaching of the present invention. The terms or definitions used herein are provided solely to aid in the understanding of the invention. Reference throughout this specification to "one embodiment" or "an embodiment" means that a particular feature, structure or characteristic described in connection with the embodiment is included in at least one embodiment of the present invention. Thus, the appearance of the phrases "in one embodiment" or "in an embodiment" in various places throughout this specification are not necessarily all referring to the same embodiment, but may. Furthermore, the particular features, structures or characteristics may be combined in any suitable manner, as would be apparent to a person skilled in the art from this disclosure, in one or more embodiments. Furthermore, while some embodiments described herein include some but not other features included in other embodiments, combinations of features of different embodiments are meant to be within the scope of the invention, and form different embodiments, as would be understood by those in the art. For example, in the following claims, any ofthe claimed embodiments can be used in any combination.

With the present invention, the inventors identified biomarkers for determining whether a subject is predisposed for generating hyperprogressive disease when receiving an immunotherapeutic treatment, or for deciding whether a subject is eligible for an immunotherapeutic treatment. Also biomarkers for determining whether a subject is predisposed of responding to an immunotherapeutic treatment were identified.

Immunotherapy using immune checkpoint inhibitors has changed the treatment landscape for many tumor types, particularly in the metastatic setting. Though, while meaningful, durable responses are achieved in some patients, a majority of patients do not respond, even worrisome, some patients develop hyperprogressive disease, a phenomenon clinically defines as an unexpected acceleration of the tumor kinetics measured on imaging with dynamic parameters. Hyperprogressive disease (HPD) following treatment with immunotherapy such as immune checkpoint inhibitors (ICIs) often leads to patient death within 24-65 days after the ICI treatment (Ferrara et al., 2020).

To date, the search for biomarkers to predict the response to immunotherapeutic treatment, and even to predict whether a patient would develop HPD, has been challenging by the dynamic interplay between the ICIs and the immune microenvironment and the heterogeneity ofthe immune milieu in different tumor types (Davis and Patel, 2019). With the present application, the inventors identified several biomarkers that predict whether a subject would develop HPD after receiving an immunotherapeutic treatment, or that facilitate to decide whether a subject is eligible for an immunotherapeutic treatment, or that predict whether a subject is predisposed for responding to an immunotherapeutic treatment.

In a first aspect, the current application provides a method for determining whether a subject is predisposed for generating hyperprogressive disease when receiving treatment with an immunotherapeutic agent, or for deciding whether a subject is eligible for treatment with an immunotherapeutic agent, said method comprising the analysis of the expression level of one or more biomarkers, preferably one or more protein coding genes, in a sample of the subject. More in particular, the expression level of one or more of the following biomarkers, preferably protein coding genes, will be assessed: IL-17A, IFN1 , IFNB1 , IL-6, ICOSLG, CCL20, ID01 , FGF2, CXCL8, NRP1 , MX1 , VEGFA, CXCL12 and/or TGFB1 , in a sample of the subject. In a preferred embodiment, the expression level of the biomarkers, preferably protein encoding genes, IL-17A, IFN1 , IFNB1 , IL-6 will be assessed, whether or not in combination with one or more of the following biomarkers, preferably protein coding genes, ICOSLG, CCL20, ID01 , FGF2, CXCL8, NRP1 , MX1 , VEGFA, CXCL12 and/or TGFB1. In another embodiment, the expression level of the biomarkers IL-17A, IFN1 , IFNB1 , IL-6 will be assessed, whether or not in combination with at least 3, at least 4, at least 5, at least 6, at least 7, at least 8 or at least 9 of the following biomarkers, preferably protein coding genes, ICOSLG, CCL20, ID01 , FGF2, CXCL8, NRP1 , MX1 , VEGFA, CXCL12 and/or TGFB1. In still another embodiment, the expression level of the biomarkers, preferably protein coding genes, IL-17A, IFN1 , IFNB1 , IL-6, ICOSLG, CCL20, ID01 , FGF2, CXCL8, NRP1 , MX1 , VEGFA, CXCL12 and TGFB1 will be assessed.

In some embodiments, the expression level of the biomarkers is compared to a corresponding reference value or threshold value that is characteristic of a subject with a known response to treatment with an immunotherapeutic agent, in particular an immune checkpoint inhibitor.

In some embodiments, on the basis of one of said expression analyses, a risk score is obtained and said risk score representing the likelihood for developing a hyperprogressive disease in the subject, or for responding or not responding to the immunotherapeutic treatment by the subject. In some further embodiments, the risk score is compared with a threshold score that is calculated based on the expression analysis of the biomarkers in a sample obtained from a subject with a known response to treatment with an immunotherapeutic agent.

The list of biomarkers, preferably protein coding genes, including IL-17A, IFN1 , IFNB1 , IL-6, ICOSLG, CCL20, ID01 , FGF2, CXCL8, NRP1 , MX1 , VEGFA, CXCL12 and TGFB1 , thus comprises the predictive biomarkers of one aspect of the current application. Preferably, these predictive biomarkers are used for determining whether a subject is predisposed for generating hyperprogression and/or progressive disease when receiving treatment with an immunotherapeutic agent. More preferably, this list of predictive biomarkers is used for classifying a subject into a responder, non-responder or hyperprogressor toward treatment with an immunotherapeutic agent. These predictive biomarkers are further used for deciding whether a subject is eligible for treatment with an immunotherapeutic agent.

In a another aspect, the current invention provides a method for determining whether a subject will respond to treatment with an immunotherapeutic agent, said method comprising the analysis of the expression level of one or more biomarkers, preferably protein coding genes, selected from TLR9, IL23A, CLEC4C, CCR4, SIGLEC6, IL3RA, IL12A, TLR7, CXCR3, AXL, TLR8, SIGLEC1 , TCF4, CD22, ITGAX, and FCGR3A, in a sample of the subject. In a further embodiment, the expression level of TLR9, IL23A, CLEC4C, CCR4 will be assessed, whether or not in combination with one or more of the following biomarkers, preferably protein coding genes: SIGLEC6, IL3RA, IL12A, TLR7, CXCR3, AXL, TLR8, SIGLEC1 , TCF4, CD22, ITGAX, and/or FCGR3A. In another embodiment, the expression level of TLR9, IL23A, CLEC4C, CCR4 will be assessed, whether or not in combination with at least 3, at least 4, at least 5, at least 6, at least 7, at least 8 or at least 9 of the following biomarkers, preferably protein coding genes, SIGLEC6, IL3RA, IL12A, TLR7, CXCR3, AXL, TLR8, SIGLEC1 , TCF4, CD22, ITGAX, and/or FCGR3A. In still another preferred embodiment, the expression level of the biomarkers, preferably protein encoding genes, TLR9, IL23A, CLEC4C, CCR4, SIGLEC6, IL3RA, IL12A, TLR7, CXCR3, AXL, TLR8, SIGLEC1 , TCF4, CD22, ITGAX, and FCGR3A will be assessed. In some embodiments, the expression level of the biomarkers will be compared to a corresponding reference value or threshold value that is characteristic of a subject with a known response to the immunotherapeutic treatment.

In some embodiments, on the basis of one of said expression analyses, a risk score is obtained and said risk score representing the likelihood for responding to treatment with the immunotherapeutic agent in the subject. Even further, said risk score may be compared with a threshold score that is calculated based on the expression analysis of the biomarkers in a sample obtained from a subject with a known response to treatment with the immunotherapeutic agent. The list of 16 biomarkers, including TLR9, IL23A, CLEC4C, CCR4, SIGLEC6, IL3RA, IL12A, TLR7, CXCR3, AXL, TLR8, SIGLEC1 , TCF4, CD22, ITGAX, FCGR3A thus comprises the predictive biomarkers of one aspect of the current application. Preferably, these predictive biomarkers are used for determining whether a subject will respond to treatment with an immunotherapeutic agent. More preferably, this list of predictive biomarkers is used for classifying a subject into a responder, non-responder or hyperprogressor toward treatment with an immunotherapeutic agent.

In some embodiments, the expression level analysis of the different biomarkers to generate a risk score representing the likelihood for developing hyperprogressive disease in a subject or for representing whether a subject is eligible for treatment with an immunotherapeutic agent, or for responding to treatment with an immunotherapeutic agent in the subject is based on the multifactorial character of cancer disease. In particular, the tumor microenvironment modifies the malignancy of tumors. This environment is populated by many macrophages and dendritic cells, which promote tumor progression to malignancy and increase metastatic potential. Within this tumor microenvironment different phenotypes of tumor-associated macrophages (TAMs) and tumor-associated dendritic cells (TADCs) are indicated due to polarization, wherein these cells modulate from an inflammatory phenotype towards an immunosuppressive phenotype. More in particular, the biomarker signature associated with the risk score representing the likelihood for developing a hyperprogressive disease is mostly associated with genes and/or proteins related to the cell surface, cell adhesion and mediation of intercellular interactions. On the other hand, the biomarker signature associated with the risk score representing the likelihood for responding to immunotherapeutic treatment is mostly associated with genes and/or proteins related to interferon- beta production and genes participating in effector immune response.

In particular embodiments, the methods as taught herein may comprise comparing the biomarker expression level to a corresponding reference value or threshold value that is characteristic of a subject with a known response to treatment with an immunotherapeutic agent, in particular immune checkpoint inhibitor. Said reference or threshold value may represent the expression level of the one or more biomarkers of a subject with a known prognosis after treatment with the immunotherapeutic agent. Said known prognosis can be a response to the treatment, a partial response to the treatment, no response to the treatment, or even the development of hyperprogressive disease in response to the treatment; In other embodiments, said reference or threshold value may represent the expression level of the one or more biomarkers of a healthy subject. Said comparison may generally include any means to determine the presence or absence of at least one difference and optionally of the size of such difference between values or profiles being compared. A comparison may include a visual inspection, an arithmetical or statistical comparison of measurements. Such statistical comparisons include, but are not limited to, applying an algorithm. If the values or biomarker expression profiles comprise at least one standard, the comparison to determine a difference in said values or biomarker expression profiles may also include measurements of these standard, such that measurement of the biomarker are correlated to measurements of the internal standards.

Reference values or threshold values for the expression level of any of the biomarker may be established according to known procedures previously employed for other biomarkers.

For example, a reference value of the expression level of a biomarker for determining whether a subject is predisposed of generating hyperprogressive disease when receiving an immunotherapeutic treatment as taught herein may be established by determining the quantity or expression of said biomarker in a sample(s) from one individual or from a population (e.g., group) of individuals characterized by a known predisposition for generating hyperprogressive disease. In another example, a reference value of the expression level of a biomarker for determining whether a subject is predisposed of responding to an immunotherapeutic treatment as taught herein may be established by determining the quantity or expression of said biomarker in a sample(s) from one individual or from a population (e.g., group) of individuals characterized by a known response to treatment with the immunotherapeutic agent. Such population may comprise without limitation >2, >10, >100, or even several hundred of individuals or more.

In an embodiment, reference value(s) or threshold value(s) as intended herein may convey absolute quantities of the biomarker as intended herein. In another embodiment, the quantity of the biomarker in a sample from a test subject may be determined directly relative to the reference value (e.g., in terms of increase or decrease, or fold-increase or fold-decrease). Advantageously, this may allow the comparison of the quantity or expression level of the biomarker in the sample from the subject with the reference value (in other words to measure the relative quantity of the biomarker in the sample from the subject vis-a-vis the reference value) without the need first to determine the respective absolute quantities of the biomarker.

As explained, the present methods, uses, or products may involve finding a deviation or no deviation between the expression level of the one or more biomarkers in a sample of the subject as taught herein and a given reference value or threshold value. A ’’deviation” of a first value from a second value or a “difference” between a first value and a second value may generally encompass any direction (e.g., increase: first value > second value; or decrease: first value < second value) and any extent of alteration.

For example, a deviation or a difference may encompass a decrease in a first value by, without limitation, at least about 10% (about 0.9-fold or less), or by at least about 20% (about 0.8-fold or less), or by at least about 30% (about 0.7-fold or less), or by at least about 40% (about 0.6-fold or less), or by at least about 50% (about 0.5-fold or less), or by at least about 60% (about 0.4-fold or less), or by at least about 70% (about 0.3-fold or less), or by at least about 80% (about 0.2-fold or less), or by at least about 90% (about 0.1-fold or less), relative to a second value with which a comparison is being made.

For example, a deviation or a difference may encompass an increase of a first value by, without limitation, at least about 10% (about 1 .1-fold or more), or by at least about 20% (about 1 .2-fold or more), or by at least about 30% (about 1 .3-fold or more), or by at least about 40% (about 1 .4-fold or more), or by at least about 50% (about 1 .5-fold or more), or by at least about 60% (about 1 .6- fold or more), or by at least about 70% (about 1 .7-fold or more), or by at least about 80% (about 1.8-fold or more), or by at least about 90% (about 1.9-fold or more), or by at least about 100% (about 2-fold or more), or by at least about 150% (about 2.5-fold or more), or by at least about 200% (about 3-fold or more), or by at least about 500% (about 6-fold or more), or by at least about 700% (about 8-fold or more), or like, relative to a second value with which a comparison is being made.

Preferably, a deviation or a difference may refer to a statistically significant observed alteration. For example, a deviation or a difference may refer to an observed alteration which falls outside of error margins of reference values in a given population (as expressed, for example, by standard deviation or standard error, or by a predetermined multiple thereof, e.g., +1xSD or +2xSD or +3xSD, or +1xSE or +2xSE or +3xSE). Deviation or a difference may also refer to a value falling outside of a reference range defined by values in a given population (for example, outside of a range which comprises >40%, > 50%, >60%, >70%, >75% or >80% or >85% or >90% or >95% or even >100% of values in said population).

In a further embodiment, a deviation or a difference may be concluded if an observed alteration is beyond a given threshold or cut-off. Such threshold or cut-off may be selected as generally known in the art to provide for a chosen accuracy, sensitivity and/or specificity of the prediction methods, e.g., accuracy, sensitivity and/or specificity of at least 50%, or at least 60%, or at least 70%, or at least 80%, or at least 85%, or at least 90%, or at least 95%.

For example, receiver-operating characteristic (ROC) curve analysis can be used to select an optimal threshold or cut-off value of the quantity of a given biomarker for clinical use of the present diagnostic tests, based on acceptable global accuracy, sensitivity and/or specificity, or related performance measures which are well-known per se, such as positive predictive value (PPV), negative predictive value (NPV), positive likelihood ratio (LR+), negative likelihood ratio (LR-), Youden index, or similar. For example, an optimal threshold or cut-off value may be selected for each individual biomarker as a local extremum of the receiver operating characteristic (ROC) curve, i.e. a point of local maximum distance to the diagonal line, as described in Robin X., PanelomiX: a threshold-based algorithm to create panels of biomarkers, 2013, Translational Proteomics, 1 (1):57-64.

The person skilled in the art will understand that it is not relevant to give an exact threshold or cutoff value. A relevant threshold or cut-off value can be obtained by correlating the sensitivity and specificity and the sensitivity/specificity for any threshold or cut-off value.

It is to the diagnostic engineers to determine which level of positive predictive value/negative predictive value/sensitivity/specificity is desirable and how much loss in positive or negative predictive value is tolerable. The chosen threshold or cut-off level could be dependent on other diagnostic parameters used in combination with the present method by the diagnostic engineers.

In some embodiments, in the methods, uses or products as taught herein, a risk score will be generated. The risk scores that are generated represent the likelihood for developing hyperprogressive disease or represent the likelihood for responding to immunotherapeutic treatment in the subject. In some embodiments, said risk score ranges from 0 to 1 and represents the likelihood for developing hyperprogressive disease in a subject, with 1 being 100% likelihood of developing hyperprogressive disease and 0 having no risk of developing hyperprogressive disease. In another embodiment, the risk score ranges from 0 to 1 and represents the likelihood for responding to immunotherapeutic treatment in the subject, with 1 being 100% likelihood for responding to treatment with an immunotherapeutic agent in the subject, and 0 being 0% likelihood for responding to treatment with an immunotherapeutic agent.

In some embodiments, the likelihood values are produced by a machine learning model that uses normalized expression levels of one of the biomarker signatures as input data. In a further embodiment, the machine learning model is a pre-trained machine learning model that uses a pretrained architecture to calculate the risk score from 0 to 1 . This pre-trained architecture is trained and updated based on the pool of retrospective and prospective samples with already known annotation towards the development of hyperprogressive disease or towards the response to immunotherapeutic treatment, in particular this pre-trained architecture is trained and updated based on the pool of reference values or threshold values obtained from samples of subjects with already known annotation towards the development of hyperprogressive disease or towards the response to immunotherapeutic treatment.

In some embodiments, the machine learning model is selected from (1) linear mixed-effects model with random effect, (2) differential expression analysis, (3) random forest machine learning model, (4) gradient boosting machine model and (5) novel deep regression model based pre-trained architectures such as VGG-16 and ResNet-50 architectures. In some embodiments, a combination of machine learning models can be applied wherein each model is run independently on each dataset and then collectively on aggregated data to ensure that the signature replicates well on the independent trials. In some further embodiments, gene and label bootstrapping and independent cohort comparisons can be applied to exclude the probability of random fit of these models and to ensure that the one or more biomarkers are unlikely to appear in the data by chance alone.

In a particularly preferred embodiment, the analysis, in particular the biomarker expression analysis and calculation of the risk score, is thus trained and updated using machine learning, for example using the gradient boosting machine (GBM) classification methodology to learn the specific pattern of biomarker expression features responsible for the subject’s response to one or more immunotherapy or for the subject to develop hyperprogressive disease.

According to an embodiment, expression levels of one or more biomarkers selected from IL-17A, IFN1 , IFNB1 , IL-6, ICOSLG, CCL20, ID01 , FGF2, CXCL8, NRP1 , MX1 , VEGFA, CXCL12 and TGFB1 are used in the gradient boosting machine (GBM) classification methodology to learn the specific pattern of said biomarker expression features responsible for a subject to develop hyperprogressive disease. The response variable in those GBM models is further optimized for RECIST criteria and hyperprogression.

In another embodiment, expression levels of one or more biomarkers selected from TLR9, IL-23A, CLEC4C, CCR4, SIGLEC6, IL3RA, IL12A, TLR7, CXCR3, AXL, TLR8, SIGLEC1 , TCF4, CD22, ITGAX, and FCGR3A are used in the gradient boosting machine (GBM) classification methodology to learn the specific pattern of said biomarker expression features responsible for the subject response to one or more therapy. The response variable in those GBM models is further optimized for RECIST criteria and response to immunotherapy.

In some embodiments, the methods of the current invention, wherein a risk score is obtained, are further supplemented with the use of other cancer diagnostics known by a person skilled in the art.

The methods, kits, or uses as taught herein, are typically characterized by the expression analysis of one or more biomarkers in a sample of subject. As used herein, a “biomarker” is widespread in the art and may broadly denote a biological molecule and/or detectable portion thereof whose qualitative and/or quantitative evaluation in a subject is predictive or informative (e.g., predictive, diagnostic and/or prognostic) with respect to one or more aspects of the subject’s phenotype and/or genotype, such as, for example, with respect to the status of the subject as to a given disease or condition.

In preferred embodiments, the biomarkers as taught herein are protein coding genes and the expression level of said protein coding genes is evaluated. In even more preferred embodiments, RNA-based expression levels of the one or more protein coding genes are evaluated. The expression level of the protein coding genes disclosed herein may be assessed by any method known in the art suitable for determination of specific gene expression levels in a sample. Such methods are well-known and routinely implemented in the art. Gene expression analysis can for example be performed by reverse transcriptase real-time quantitative PCR, gene expression arrays, TruSeq gene expression analysis, in-situ hybridization, dye sequencing, pyrosequencing, CRISPR-Cas-based or any other form of transcriptome sequencing (total RNA-Seq, mRNA-Seq, gene expression profiling).

Additionally, a customized version of TruSeq Targeted RNA expression Kit is used for measuring the expression levels of a plurality of genes as listed herein. TruSeq Targeted RNA Expression Kit enables highly customizable mid- to high-plex gene expression profiling studies which allow defining panels of 12-1 ,000 assays to target individual exons, isoforms, splice junctions, coding SNPs (cSNPs), gene fusions, and non-coding RNA transcripts, plus multiplex up to 384 samples. Also, a customized version of TruSeq Custom Amplicon Kit Dx can be used for the qualitative and/or quantitative assessment of a plurality of genes as listed herein. TruSeq Custom Amplicon Kit Dx is an FDA-approved regulated and CE-IVD-marked amplicon sequencing kit that enables clinical labs to develop their own next-generation sequencing (NGS) assays for use on the FDA- regulated and CE-IVD-marked MiSeqDx and NextSeq 550dx instruments.

In certain other embodiments, a biomarker as taught herein, may be peptide-, polypeptide- and/or protein-based. The reference to any marker, including any gene, peptide, polypeptide, protein, corresponds to the marker, gene, peptide, polypeptide, protein, commonly known under the respective designations in the art. The terms encompass such markers, genes, peptides, polypeptides, proteins of any organism where found, and particularly of animals, preferably warmblooded animals, more preferably vertebrates, yet more preferably mammals, including humans and non-human mammals, still more preferably of humans. The terms particularly encompass such markers, genes, peptides, polypeptides, proteins with a native sequence, i.e. , ones of which the primary sequence is the same as that of the markers, genes, peptides, polypeptides, proteins found in or derived from nature. A skilled person understands that native sequences may differ between different species due to genetic divergence between such species. Moreover, native sequences may differ between or within different individuals of the same species due to normal genetic diversity (variation) within a given species. Also, native sequences may differ between or even within different individuals of the same species due to genetic alterations, or post- transcriptional or posttranslational modifications. Any such variants or isoforms of markers, genes, peptides, polypeptides, proteins are intended herein. Accordingly, all sequences of markers, genes, peptides, polypeptides, or proteins found in or derived from nature are considered “native”. The terms encompass the markers, genes, peptides, polypeptides, or proteins when forming a part of a living organism, organ, tissue or cell, when forming a part of a biological sample, as well as when at least partly isolated from such sources. The terms also encompass markers, genes, peptides, polypeptides, or proteins when produced by recombinant or synthetic means.

In certain embodiments, the biomarkers as taught herein, may be a human biomarker, such as in particular any of the biomarkers selected from IL17A, IFNA1 , IFNB1 , IL-6, ICOSLG, CCL20, ID01 , FGF2, CXCL8, NRP1 , MX1 , VEGFA, CXCL12 and TGFB1 , or selected from TLR9, IL23A, CLEC4C, CCR4, SIGLEC6, IL3RA, IL12A, TLR7, CXCR3, AXL, TLR8, SIGLEC1 , TCF4, CD22, ITGAX, and FCGR3A.

Unless otherwise apparent from the context, reference herein to any marker, gene, peptide, polypeptide or protein, or fragment thereof may generally also encompass modified forms of said marker, gene, peptide, polypeptide, or protein, or fragment thereof, such as genetic alterations, such as mutations, or bearing post-expression modifications including, for example, phosphorylation, glycosylation, lipidation, methylation, cysteinylation, sulphonation, glutathionylation, acetylation, oxidation, and the like.

The reference herein to any marker, gene, peptide, polypeptide or protein also encompasses fragments thereof. Hence, the reference herein to measuring (or measuring the quantity of), determining the presence, or determining the expression level of any one marker, gene, peptide, polypeptide or protein may encompass measuring, determining the presence or determining the expression level of the marker, gene, peptide, polypeptide, or protein, such as, e.g. measuring any mature and/or processed soluble/secreted form(s) thereof (e.g., plasma circulating form(s)) and/or measuring one or more fragments thereof.

For example, any marker, gene, peptide, polypeptide or protein, and/or one or more fragments thereof may be measured collectively, such that the measured quantity corresponds to the sum amounts of the collectively measured species. In a further example, any marker, gene, peptide, polypeptide or protein and/or one or more fragments thereof may be measured each individually.

The term “fragment” with reference to a peptide, polypeptide, or protein generally denotes a N- and/or C-terminally truncated form of the peptide, polypeptide, or protein. Preferably, a fragment may comprise at least about 30%, e.g., at least about 50% or at least about 70%, preferably at least about 80%, e.g., at least about 85%, more preferably at least about 90%, and yet more preferably at least about 95% or even about 99% of the amino acid sequence length of said peptide, polypeptide, or 10 protein. For example, insofar not exceeding the length of the full-length peptide, polypeptide, or protein, a fragment may include a sequence of > 5 consecutive amino acids, or > 10 consecutive amino acids, or > 20 consecutive amino acids, or > 30 consecutive amino acids, e.g., > 40 consecutive amino acids, such as for example > 50 consecutive amino acids, e.g., > 60, > 70, > 80, > 90, > 100, > 200, > 300 or > 400 consecutive amino acids of the corresponding full-length peptide, polypeptide, or protein.

The reference herein to any protein, polypeptide or peptide may also encompass variants thereof. The term “variant” of a protein, polypeptide or peptide refers to proteins, polypeptides or peptides the sequence (i.e. amino acid sequence) of which is substantially identical (i.e. largely but not wholly identical) to the sequence of said recited protein or polypeptide, e.g. at least about 80% identical or at least about 85% identical, e.g. preferably at least about 90% identical, e.g., at least 91% identical, 92% identical, more preferably at least about 93% identical, e.g. at least 94% identical, even more preferably at least about 95% identical, e.g., at least 96% identical, yet more preferably at least about 97% identical, e.g. at least 98% identical, and most preferably at least 99% identical. Preferably, a variant may display such degrees of identity to a recited protein, polypeptide or peptide when the whole sequence of the recited protein, polypeptide or peptide is querried in the sequence alignment (i.e, overall sequence identity).

A variant of a protein, polypeptide or peptide may be a homologue (e.g, orthologue or paralogue) of said protein, polypeptide or peptide. As used herein, the term “homology” generally denotes structural similarities between macromolecules, particularly between two proteins or polypeptides, from same or different taxons, wherein said similarity is due to shared ancestry.

Where the present specification refers to or encompasses fragments and/or variants of proteins, polypeptides or peptides, this preferably denotes variants and/or fragments which are “functional”, i.e., which at least partly retain the biological activity or intended functionality of the respective proteins, polypeptides or peptides. Preferably, a functional fragment and/or variant may retain at least about 20%, e.g., at least 30%, or at least about 40%, or at least about 50%, e.g., at least 60%, more preferably at least about 70%, e.g., at least 80%, yet more preferably at least about 85%, still more preferably at least about 90%, and most preferably at least about 95% or even about 100% or higher of the intended biological activity or functionality compared to the corresponding protein, polypeptide or peptide.

In some embodiments, the risk score obtained in any of the methods of the application can further be improved by immunohistochemistry or protein expression data of the tumor of the subject. For example, the Cytokine 30-Plex Human Panel and immunohistochemistry stainings can be used for additional qualitative and/or quantitative assessment of the sample of the subject.

The current invention also provides a computer-implemented method for determining whether a subject is predisposed for generating hyperprogressive disease when receiving treatment with an immunotherapeutic agent, wherein a sample obtained from said subject is analyzed, the method comprising: (a) providing the quantified expression level of one or more biomarkers, preferably protein encoding genes, from the group consisting of IL-17A, IFN1 , IFNB1 , IL-6; even more preferably consisting of IL-17A, IFN1 , IFNB1 , IL-6, said biomarkers being quantified in a sample of the subject; (b) optionally, providing the quantified expression level of one or more biomarkers selected from ICOSLG, CCL20, ID01, FGF2, CXCL8, NRP1 , MX1 , VEGFA, CXCL12 and TGFB1 , said biomarkers being quantified in a sample of the subject; (c) normalizing the quantified expression levels whereby normalization occurs via comparison with data obtained from corresponding assessments and expression levels from a reference set, (d) classifying, whether the normalized values of step c exceed a predetermined threshold, (e) obtaining a risk score of said normalized values, wherein said score is calculated using a predictive algorithm or machine learning model, and wherein the risk score represents the likelihood for developing hyperprogressive disease in said subject.

In another embodiment, a computer-implemented method is provided for determining whether a subject will respond to treatment with an immunotherapeutic agent, wherein a sample obtained from said subject is analyzed, the method comprising: (a) providing the quantified expression level of one or more biomarkers, preferably protein encoding genes from the group consisting of TLR9, IL-23A, CLEC4C, CCR4; even more preferably consisting of TLR9, IL-23A, CLEC4C, CCR4, said biomarkers being quantified in a sample of the subject; (b) optionally providing the quantified expression level of one or more biomarkers selected from SIGLEC6, IL3RA, IL12A, TLR7, CXCR3, AXL, TLR8, SIGLEC1 , TCF4, CD22, ITGAX, and FCGR3A, said biomarkers being quantified in a sample of the subject; (c) normalizing the quantified expression levels whereby normalization occurs via comparison with data obtained from corresponding assessments and expression levels from a reference set, (d) classifying, whether the normalized values of step c exceed a predetermined threshold, (e) obtaining a risk score of said normalized values, wherein said score is calculated using a predictive algorithm, and wherein the risk score represents a likelihood for responding to the immunotherapeutic agent.

In some embodiments, a kit for determining whether a subject is predisposed for generating hyperprogressive diseases when receiving treatment with an immunotherapeutic agent or for deciding whether a subject is eligible for treatment with an immunotherapeutic agent is disclosed. Said kit comprises means for measuring the expression level of the biomarkers IL-17A, IFNA1 , IFNB and IL-6 in a sample of the subject. Optionally said kit further comprises means for measuring the expression level of one or more of the biomarkers selected from ICOSLG, CCL20, ID01 , FGF2, CXCL8, NRP1 , MX1 , VEGFA, CXCL12 and TGFB1 . Said kit may further also comprise a reference value or threshold value for the biomarkers IL-17A, IFNA1 , IFNB and IL-6. The kit may further also comprise a reference value or threshold value for one or more of the biomarkers selected from the group consisting of ICOSLG, CCL20, ID01 , FGF2, CXCL8, NRP1 , MX1 , VEGFA, CXCL12 and TGFB1 .

In some embodiments, a kit for determining whether a subject is predisposed for responding to treatment with an immunotherapeutic agent is disclosed. Said kit comprises means for measuring the expression level of the biomarkers TLR9, IL-23A, CLEC4C, CCR4. Optionally, said kit comprises means for measuring the expression level of one or more of SIGLEC6, IL3RA, IL12A, TLR7, CXCR3, AXL, TLR8, SIGLEC1 , TCF4, CD22, ITGAX, and FCGR3A. Said kit may further comprise a reference value or threshold value for the biomarkers TLR9, IL-23A, CLEC4C, CCR4. In a further embodiment, the kit may also comprise a reference value or threshold value for one or more of the biomarkers selected from the group consisting of SIGLEC6, IL3RA, IL12A, TLR7, CXCR3, AXL, TLR8, SIGLEC1 , TCF4, CD22, ITGAX, and FCGR3A.

In still a further embodiment, the kit of the present application may comprise ready-to-use substrate solutions, wash solutions, dilution buffers and instructions. The kit may also comprise positive and/or negative control samples. In some embodiments, the kit comprises instructions for use. Preferably, the instructions for use included in the kit are unambiguous, concise and comprehensible to those skilled in the art. The instructions typically provide information on kit contents, how to collect the sample, methodology, experimental read-outs and interpretation thereof and cautions and warnings.

The terms “kit of parts” and “kit” as used throughout this specification refer to a product containing components necessary for carrying out the specified methods (e.g., method for determining whether a subject is predisposed for generating hyperprogressive disease, method for deciding whether a subject is eligible for an immunotherapeutic treatment or method for determining whether a subject is predisposed for responding to an immunotherapeutic treatment), packed so as to allow their transport and storage. Materials suitable for packing the components comprised in a kit include crystal, plastic (e.g., polyethylene, polypropylene, polycarbonate), bottles, flasks, vials, ampules, paper, envelopes, or other types of containers, carriers or supports. Where a kit comprises a plurality of components, at least a subset of the components (e.g., two or more of the plurality of components) or all of the components may be physically separated, e.g. comprises in or on separate containers, carriers or supports. The components comprised in a kit may be sufficient or may not be sufficient for carrying out the specified methods, such that external reagents or substances may not be necessary or may be necessary for performing the methods, respectively. Typically, kits are employed in conjunction with standard laboratory equipment, such as liquid handling equipment, environment (e.g., temperature) controlling equipment, analytical instruments, etc. The kits may also include some or all of solvents, buffers (such as for example without limitation histidine-buffers, citrate-buffers, succinate-buffers, acetate-buffers, phosphate- buffers, formate buffers, benzoate buffers, TRIS ((Tris(hydroxymethyl)-aminomethan) buffers or maleate buffers, or mixtures thereof), enzymes (such as for example but without limitation thermostable DNA polymerase), detectable labels, detection reagents, and control formulations (positive and/or negative), useful in the specified methods.

Typically, the kits may also include instructions for use thereof, such as a printed insert or on a computer readable medium. The terms may be used interchangeably with the term “article of manufacture”, which broadly encompasses any man-made tangible structural product, when used in the present context.

In particular embodiments, the kit further comprises a computer readable storage medium having recorded thereon one or more programs for carrying out the methods as taught herein.

Further, also the use of said kits is provided, in particular for determining whether a subject is predisposed for generating hyperprogressive disease, for deciding whether a subject is eligible for an immunotherapeutic treatment or for determining whether a subject is predisposed for responding to an immunotherapeutic treatment.

The inventors of the present application identified several biomarkers for determining whether a subject is predisposed for generating hyperprogressive disease when receiving treatment with an immunotherapeutic agent, or for deciding whether a subject is eligible for treatment with an immunotherapeutic agent, or for predicting a subject’s response to immunotherapeutic agent. More specifically, the biomarkers were identified to be specifically suitable for predicting responses after treatment with immune checkpoint inhibitors. In some embodiments, the immunotherapeutic agent is therefore an immune checkpoint inhibitor or a combination of several immune checkpoint inhibitors. Even further, the immune checkpoint inhibitors can be selected from PD-1 targeting agents, PD-L1 targeting agents or CTLA-4 targeting agents, or a combination thereof. In some further preferred embodiments, the immunotherapeutic treatment is a combination of a PD-1 targeting agent or a PD-L1 targeting agent in combination with a CTLA-4 targeting agent. In some embodiments, the PD-1 targeting agents are monoclonal antibodies against PD-1 , such as for example pembrolizumab, nivolumab or cemiplimab. In some embodiments, the PD-L1 targeting agents are monoclonal antibodies against PD-L1 , such as for example Atezolizumab, Avelumab, Durvalumab. In some embodiments, the CTLA-4 targeting agents are monoclonal antibodies against CTLA-1 , such as for example ipilimumab.

Immunotherapeutic drugs, like immune checkpoint inhibitors, target certain immune cells that need to be activated or inactivated to start an immune response. Monoclonal antibodies that target either PD-1 or PD-L1 can block these proteins, which are present on respectively T-cells and normal (or cancer) cells, and boost the immune response against cancer cells. These drugs are helpful in treating different types of cancer, including bladder cancer, non-small cell lung cancer, and Merkel cell skin cancer. CTLA-4 is another protein on some T cells that, just like PD-1 , acts as a type of switch to keep the immune system in check. Anti-CTLA4 treatment is often used to treat melanoma of the skin and some other cancers.

The PD-1 targeting agents, PD-L1 targeting agents or CTLA-4 targeting agents can be any targeting agent known in the art, such as an antibody or a fragment thereof that recognizes and blocks PD-1 , PD-L1 or CTLA-4. In some embodiments, the PD-1 targeting agents are monoclonal antibodies against PD-1 , such as for example pembrolizumab, nivolumab or cemiplimab. In some embodiments, the PD-L1 targeting agents are monoclonal antibodies against PD-L1 , such as for example Atezolizumab, Avelumab, Durvalumab. In some embodiments, the CTLA-4 targeting agents are monoclonal antibodies against CTLA-1 , such as for example ipilimumab.

In a further aspect of the current invention, immune checkpoint inhibitors are provided for use in the treatment of cancer in a subject wherein the subject has been found not to be predisposed for generating hyperprogressive disease using any of the methods for analyzing the expression level of the one or more biomarkers in a sample of the subject as disclosed herein. In particular, immune checkpoint inhibitors are provided for use in the treatment of cancer, preferably melanoma and/or non-small cell lung cancer, in a subject wherein the subject has been found not to be predisposed for generating hyperprogressive disease using any of the methods for analyzing the expression level of the biomarkers IL-17A, IFNA1 , IFNB and IL-6, whether or not in combination with any of ICOSLG, CCL20, ID01 , FGF2, CXCL8, NRP1 , MX1 , VEGFA, CXCL12 and TGFB1 in a sample of the subject.

Depending on the type of cancer another ICI treatment is advised. A combination therapy or a monotherapy is possible. The set of predictive biomarkers of current invention are preferably used to determine whether a subject is predisposed for generating hyperprogression when receiving one or more of said immunotherapeutic treatments, in particular one or more immune checkpoint inhibitors. More preferably the set of predictive biomarkers of current invention is used to determine whether a subject is predisposed for generating hyperprogression when receiving an immunotherapeutic treatment, such as treatment with an immune checkpoint inhibitor. Most preferably the set of predictive biomarkers of current invention is used to determine whether a subject is predisposed for generating hyperprogression when receiving a monotherapy.

A further aspect of the current invention concerns a method for treating a subject diagnosed with cancer with an immunotherapeutic agent. The method comprises determining whether the subject is predisposed for generating hyperprogressive disease when receiving an immunotherapeutic agent, or for responding to treatment with an immunotherapeutic agent according to any of the methods as disclosed herein, and administering to the subject an immunotherapeutic agent when the subject is found to be responsive to immunotherapeutic treatment. In some preferred embodiments, the immunotherapeutic agent is an immune checkpoint inhibitor, such as a PD-1 targeting agent, a PD-L1 targeting agent or a CTLA-4 targeting agent. In some further preferred embodiments, the immunotherapeutic agent is a combination of different immune checkpoint inhibitors, such as a PD-1 targeting agent or a PD-L1 targeting agent in combination with a CTLA- 4 targeting agent.

Another embodiment of the current invention provides a tumor response report, comprising the steps of isolating one or more samples from a tumor of a patient; generating from one or more samples, expression data, preferably gene expression data, even more preferably R,NA sequencing data, comprising information about a plurality of biomarkers, preferably protein coding genes, in particular about a plurality of protein coding genes genes selected from IL-17A, IFN1 , IFNB1 , IL-6, ICOSLG, CCL20, ID01 , FGF2, CXCL8, NRP1 , MX1 , VEGFA, CXCL12 and TGFB1 ; and/or a plurality of protein coding genes selected from TLR9, IL-23A, CLEC4C, CCR4, SIGLEC6, IL3RA, IL12A, TLR7, CXCR3, AXL, TLR8, SIGLEC1 , TCF4, CD22, ITGAX, and FCGR3A; and assessing the likelihood of the patient’s tumor to develop hyperprogressive disease or to respond to a plurality of possible treatments. In a further embodiment, immunohistochemistry data or any other valuable patient or tumor data can be added to the tumor response report.

In some embodiments, the sample is a tumor tissue sample derived subjects diagnosed with cancer. Preferably, the expression level of a cell signature is assessed in a biological sample that contains or is suspected to contain cancer cells. The tissue sample may be, for example, a tissue resection, a tissue biopsy, or metastatic lesion obtained from a patient suffering from, suspected to suffer from, or diagnosed with cancer. Preferably, the sample is a sample of tissue, a resection or biopsy of a tumor, known or suspected metastatic cancer lesion or section, or a blood sample, e.g. a peripheral blood sample, known or suspected metastatic cancer lesion or section, or a blood sample, e.g. a peripheral blood sample, known or suspected to comprise circulating cancer cells. The sample may comprise both cancer cells, i.e. tumor cells, and non-cancerous cells, and, in certain embodiments, comprises both cancerous and non-cancerous cells.

Methods of obtaining biological samples including tissue resections, biopsies, and body fluids, e.g., blood samples comprising cancer/tumor cells, are well known in the art. In some embodiments, the sample obtained from the patient is collected prior to beginning any immunotherapy or other treatment regimen or therapy, e.g. chemotherapy, or radiation therapy for the treatment of cancer or the management or amelioration of a symptom thereof. Therefore, in some embodiments, the sample is collected before the administration of immunotherapeutic agents or other agents, or the start of immunotherapy or other treatment regimen.

In preferred embodiments, the sample is a tumor tissue sample, such as a fresh-frozen tumor tissue sample, a fresh tumor tissue sample, or a formalin-fixed paraffin-embedded (FFPE) tumor tissue sample.

In some embodiments the subject is a subject diagnosed with cancer. The cancer can be selected from bladder cancer, kidney cancer, liver cancer, lung cancer, pancreatic cancer, prostate cancer, thyroid cancer, uterine cancer, ovarian cancer, colorectal cancer, breast cancer, head and neck cancer, skin cancer; even more preferably cancer patients diagnosed with skin cancer or melanoma, lung cancer such as lung adenocarcinoma, lung squamous cell carcinoma, non-small cell lung carcinoma (NSCLC). In a preferred embodiment, the subject is a subject diagnosed with melanoma and/or lung cancer. In a further preferred embodiment, the subject is diagnosed with melanoma. In another further embodiment, the subject is diagnosed with lung cancer, preferably non-small cell lung carcinoma (NSCLC).

The current invention is preferably depicted for patients diagnosed with non-small cell lung cancer and/or melanoma. These patients may have already undergone an anti-PD-1 , anti-PD-L1 and/or anti-CTLA-4 treatment, are undergoing such a treatment or are planned to be treated with such drugs. The set of predictive biomarkers is optimized for such patients, and predicts with high specificity and sensitivity the response of the patient’s tumor towards such a treatment or the chance of developing hyperprogressive disease.

In a preferred embodiment, the different set of biomarkers of the current invention are used for the development of novel immunotherapeutic drugs.

One aspect of the current invention relates to a method for the development of predictive biomarkers involved in developing hyperprogression and/or progressive disease when receiving an immunotherapeutic treatment, wherein the method comprises the steps of (a) treating an animal model with an immunotherapeutic treatment; (b) analyzing the expression levels of the biomarkers, preferably protein coding genes, involved in developing hyperprogressive disease and/or responsiveness to the immunotherapeutic treatment; (c) selecting differentially expressed biomarkers, preferably protein coding genes, wherein said differentially expressed biomarkers are correlated with the predisposition of a subject being responsive, non-responsive, or hyperprogressive toward to the immunotherapeutic treatment.

Another aspect of the current invention provides a method for testing biomarkers involved in developing hyperprogressive disease or for predicting the responsiveness to immunotherapeutic treatment, wherein the method comprises the steps of (a) treating an animal model with an immunotherapeutic treatment; (b) analyzing the expression level of biomarkers, preferably protein coding genes involved in developing hyperprogressive disease and/or responsiveness to the immunotherapeutic treatment; (c) selecting differentially expressed biomarkers, preferably protein coding genes, wherein said differentially expressed biomarkers, which are correlated with the predisposition of the animal model being responsive, non-responsive, or hyperprogressive toward the immunotherapeutic treatment, are asserted.

Preferably, animal models are selected from the group consisting of mouse, rat, rabbit, cat, dog, frog. These animal models can be utilized for the development and testing of the predictive biomarkers. A preferred animal model is a mouse model.

In another preferred embodiment, the biomarkers of current invention are developed in an animal model. The predictive biomarkers of current invention are developed in a mouse model. Preferred mouse models are MISTRG, NSG-SGM3, NOG-EXL, BRG hlL-3 hGM-CSF. In another and further embodiment of current invention, the predictive biomarkers are tested in an animal model, preferably a mouse model.

Some embodiments of the application relate to the use of the previously described methods for predicting hyperprogression and/or non-responsiveness of a subject undergoing one or more immunotherapeutic treatments. Preferably, for predicting hyperprogression and/or nonresponsiveness of a subject undergoing anti-PD-1/PD-L1 monotherapy, or in combination with anti-CTLA-4 therapy.

Prognostically identifying which patients are going to experience hyperprogression or be nonresponders to such immunotherapeutic treatments is of medical and economical interest. An optimal patient specific therapy can be determined which assures a better and safer therapy.

The present invention further provides a method for determining the response of a tumor to an immunotherapeutic treatment; the method comprising the steps of quantitative assessment of the expression level of a plurality of biomarkers, preferably protein coding genes involved in developing hyperprogressive disease and/or responsiveness to the immunotherapeutic treatment; and based on this quantitative assessment classifying the tumor as a responder or a nonresponder to the therapy, wherein the responder is further classified as being at risk for hyperprogression to said therapy.

In some embodiments, the method further comprises the step of determining immunohistochemistry data for the tumor.

Preferably immunohistochemistry data is included in the characterization of the tumor and tumor microenvironment. This extra dataset makes the identification of the tumor is a R, NR, or HP more reliable and valid, as repeated measurements generate a similar output.

In a more preferred embodiment, the method further comprises the step of determining one or more possible immunotherapeutic treatment therapies. Next to the targeted ICI treatment other immunotherapeutic treatment therapies could be addressed in the treatment of cancer patients leading to a positive outcome of the cancer patients treatment.

The application further provides methods for administering an activating or suppressing immunotherapeutic treatment to patients with cancer, if the patient is determined to have a change in the level of expression of one or more gene signatures as disclosed herein. In one embodiment, the method of the present invention comprises the step of informing the patient that they have an increased likelihood of being responsive to therapy. In another embodiment, the method of the present invention comprises the step of recommending a particular therapeutic treatment to the patient. In other embodiments, the method of the present invention further comprises the step of administering a therapy to the patient if it is determined that the patient may benefit from the therapy, in particular the immunotherapy.

In a further embodiment, the immunotherapeutic agent comprises a checkpoint inhibitor, a chimeric antigen receptor T-cell therapy, an oncolytic vaccine, a cytokine agonist or a cytokine antagonist, or a combination thereof, or any other immunotherapy available in the art. The activating immunotherapy may further comprise the use of checkpoint inhibitors. Checkpoint inhibitors are readily available in the art and include, but are not limited to, a PD-1 inhibitor, PD-L1 inhibitor, PD- L2 inhibitor, a CTLA-4 inhibitor, or a combination thereof.

EXAMPLES

The invention is further described by the following non-limiting examples which further illustrate the invention, and are not intended to, nor should they be interpreted to, limit the scope of the invention.

EXAMPLE 1

Materials and methods

Data acquisition

ICIs have only recently become used on a regular basis in a clinic. Therefore, the availability of reliable and thorough data on patients' health outcomes is still a challenge. We identified key studies which have undertaken observational studies on patients exposed to ICIs (Rizvi et al. 2015; Chan, Wolchok, and Snyder 2015; Hugo et al. 2017; Hellmann et al. 2018; Goodman et al. 2017) and compiled a pooled dataset of RNA-sequencing data for further investigation.

We used RNA-sequencing (transcriptomic) data from pretreatment biopsies (tumor samples and adjacent normal tissue or peripheral blood mononuclear cells samples) to identify tumor levels of gene expressions. Briefly, we mapped paired-end RNA-seq reads to a reference genome GRCh37 (hg19) using Tophat2 (Kim et al. 2013). Output BAM-files with the read alignment were then used as input to Cufflinks suite (Trapnell et al. 2010) to quantify normalized gene expression counts (FPKM). Because our main goal was to compare expression levels between samples, we then converted FPKM values to TPM following standard procedures.

To account for the potential bias associated with data being generated by different labs, we made sure that data pre-processing has several normalization steps. In order to calibrate expression between experiments we have used housekeeping genes (e.g. CHMP2A, EMC7, VPS29) which have a constant expression among different samples. We have made a procedure that transforms most of housekeeping genes to the same order between samples by multiplying all expressions in each sample by a correspondent coefficient. In total, this resulted in 75 matched sample pairs. Because the assumed mechanism behind hyperprogression is related to the interaction between myeloid cells (non-conventional DCs and macrophages) with different subsets of T-cells, promoting the relative exuberance of eTregs and subsequent accelerated tumor progression, we first did a detailed revision of the existing literature and identified important proteins and associated genes related to the polarization and differentiation of myeloid cells. This led us to the construction of the most probable gene signatures for macrophages/DC cell polarization. Six gene signatures (Table 1) were constructed with five signatures (#1-5) specifically focusing on the biological mechanism of HPD and one gene signature (#6) on the biological mechanism of NR.

Table 1. Biological signatures for discrimination between HPD/non-HPD and R/NR patients

Each gene from Table 1 and associated expression was extracted from the previous step of the bioinformatics pipeline for each sample. Two-sided Mann-Whitney test (R function) was used to check each gene for differential expression with respect to HPD vs non-HPD or R vs NR. Forthese tests, each expression was log2-transformed in order to comply with normality assumptions. Bonferroni correction was used to deal with the multiple comparisons problem. Two genes (MMP12, FCGR1C) had to be excluded from the further analysis because no reads mapped to them.

Each gene signature was tested for NR/R classification and HPD/non-HPD classification as a target variable. We applied two statistical models - logistics regression (LR) and gradient boosting machine classifier (GBMC). LR (R glm, family=binomial(link='logit')) and GBM (R gbm) were used to create classification models for immunotherapy response. In both models, we used the listed gene expressions as predictors. We used standardized (centered and scaled) predictors in order to estimate the importance of each gene by comparing their regression coefficients. The absolute values of regression coefficients and feature importances were used to create a barplot which shows predictor importance.

ROC AUC curves and various accuracy metrics were created for the LR and GBM models. To train the models we split our dataset into test (20%) and train (80%) sets. We also run 10-fold CV on both models to ensure that the signal we get is stable regardless of the train/test target balance. Each sample was assigned to one of three classes and two independent models were constructed - one for discrimination of HPD vs non-HPD and one for discrimination of R vs NR. Due to the absence of original TGR data, we classified as HPD any sample that had survival data of less than 2 months (time-to-treatment failure (TTF) <2 months). To verify the absence of random fit, we have performed two bootstrap validations. First, we randomized labels (ie. if patients belong to HPD, R, or NR group) and run 1000 bootstraps for each hypothesis. Second, we randomly sampled an equal number of genes from the original datasets and performed 1000 gene bootstraps. The confidence intervals were then constructed at a very conservative 0.95 threshold (p-value of 0.05) and compared to real fitted ROC AUC values. This is a conservative estimate given the industry stand of 1 standard deviation for gene randomization. Results

In the univariate analysis and after Bonferroni correction, expression levels of no single gene were able to significantly discriminate the patient responses (Figure 1). This is not surprising given the inter-correlated nature of the expressed genes and a strong confounding effect that they have on each other. Both models performed similarly, with GBM slightly outperforming LR model in both CV and test/train split.

From the six signatures of genes (Table 1) the strongest evidence came for gene signatures #2 (HPD is caused by pDC or interaction of pDC and T-cells) and gene signatures # 6 (More aggressive disease is associated with nc-DCs and the crosstalk of TLR/FCR). Interestingly, gene signatures #2 has been successful in classifying both HPD/non-HPD and R/NR, whereas gene signatures #6 was only successful in R/NR classification. Gene signatures #2 and #6 performed very well in both label bootstrapping (Figure 2 and Figure 3) and gene bootstrapping validations (Figure 4 and Figure 5).

Across 10-fold CV for signature #2, average prediction accuracy for the HPD/non-HPD model reached 0.78 [Cl: 0.36 - 0.96] (Table 2). The average sensitivity and specificity at a simple threshold of 0.5 were 0.82 and 0.74 respectively (Table 2). For NR/R model in signature #2 across 10-fold CV average prediction accuracy reached 0.77 [Cl: 0.35 - 0.97] (Table 3). The average sensitivity and specificity at a simple 0.5 cut-off were 0.66 and 0.82 respectively (Table 3). For NR/R model in signature #6 across 10-fold CV average prediction accuracy reached 0.74 [Cl: 0.0.32 - 0.95] (Table 4). The average sensitivity and specificity at a simple 0.5 cut-off were 0.67 and 0.78 respectively (Table 4).

Table 2. The cross-validation results of GBM model for predictive power for HPD/non-HPD on signature #2.

Table 3. The cross-validation results of GBM model for predictive power for NR/R on signature #2.

Table 4. The cross-validation results of GBM model for predictive power for R/NR on signature #6.

The strongest genes contributing to the model classification performance for NR vs R were CXCR3, CLEC4C, TLR7, TLR8, TLR9, IL12A, IL23A, FCGR3A (Figure 6a). Besides expected up- regulation of IL12A and IL23A, we also observed downregulation of IL3A, FCGR3A and up- regulation of CXCR3, TLR8 in the responders, and the reverse pattern in non-responders. The strongest gene contributing to the model classification performance for HPD vs non-HPD were IFNB1 , SIGLEC1 , VEGFA, ID01 , ICOSLG, TGFB1 , FGF2, IL6 (Figure 6b). The directionality was an down-regulation of CCL20, CXCL12, IFNB1 and up-regulation of VEGFA, FGF2, and ICOSLG in the HPD, and the reverse pattern in non-HPD.

EXAMPLE 2 Materials and methods

To further validate potential biomarkers of hyperprogressive disease and immunotherapy resistance, we have compiled a large dataset composed of 5 independent clinical trials and totaling 293 patients with metastatic melanoma. In all trials, one of the agents (PD1/PDL1 , CTLA4, or a combination of two) was used to treat metastatic melanoma patients. For all patients, necessary clinical assessment information (demography, treatment history, treatment responses, patient genetic mutations) was available together with pre-treatment biopsies.

The pre-treatment FFPE tumor biopsies were used to extract RNA and obtain whole transcriptomics sequencing data. HPD suspects were qualified following the RECIST1 1-based definition based on PFS and OS data. Transcriptomics data were processed using the following bioinformatics procedure. Raw bulk RNA-seq reads were processed using nf-core/rnaseq pipeline to transform the raw RNA reads and to obtain the gene expression matrices. The workflow included a series of steps of read quality control and read trimming (FastQC, Trim Galore!), aligning reads against the reference genome (STAR), calculating counts relative to genes (featureCounts), and quality control on the results (RSeQC, Qualimap, Preseq, edgeR, MultiQC). We have adjusted the pipeline to best suit our needs and available resources.

After normalized, combat-standardized log2 transformed counts data were obtained, we applied several well-established statistical methods and developed a novel deep learning framework to validate potential markers of HPD and treatment resistance: (1) linear mixed-effects model with random effect; (2) differential expression analysis (3) random forest machine learning model; (4) gradient boosting machine model and (5) novel deep regression model based on two pre-trained architectures, the VGG-16 and ResNet-50 architectures. These models were run independently on each dataset and then collectively on aggregated data to ensure that signature replicates well on independent trials. We used gene and label bootstrapping to exclude the probability of the random fit of these models. We also used the PD-L1 only expression model (measured as expression of CD274) as the baseline comparator model. All models were run train and holdout datasets to ensure that biomarkers replicate well. Methods 1-5 were used to discover differentially expressed transcriptomics biomarkers consistently visible across clinical trials and to construct the biomarker signature. Then bootstrapping and independent cohort comparisons were used to ensure that the biomarkers are unlikely to appear in the data by chance alone.

Results

In total, 293 patients with pre-treatment biopsies were analyzed. 47 patients were qualified as suspected HPD candidates. Five non-HPD patients were excluded from the downstream analysis due to unclear annotation. The breakdown by treatment type was 223 patients treated with PD1/PDL1 (Nivolumab, Pembrolizumab), 32 treated with PD1 combination (Ipilimumab and Nivolumab/ Pembrolizumab), and 41 treated with CTLA4 monotherapy (Ipilimumab); 60% of patients were males and 40% females. Nearly 80% of patients had no or one line of prior treatment lines, and a little over 20% had 1+ treatment lines. Almost 75% of patients were at M1c (distant metastatic spread) according to TNM classification. Of the genotyped patients, ~18% carried mutant BRAF and RAS genes.

Bioinformatics and ML methods confirmed previously identified markers of HPD vs. non-HPD in PD1/PDL1 treated cohort from a total of 20000+ mapped expressed tags (Figure 7 and Figure 8). Four core and supporting biomarkers were consistently visible across independent trials and on a combined dataset.

ML models with combined biomarkers achieved 81 % accuracy on a 50%-50% split of training/testing data (Figure 9 and Table 5). The baseline model that only included PD-L1 expression and clinical variables was only able to discriminate non-HPD vs HPD cases at the accuracy of 69%. The top predictive biomarkers were IL17A, IFNA1 , IFNB1 , and IL-6, and these markers were consistently identified in the top 10 discriminatory biomarkers in each trial separately. The pathways enriched for the significantly differentially expressed markers (Figure 10) highlighted enrichment for antigen processing and presentation, T cell receptor signaling, Th1 and Th2 cell differentiation and PD1/PDL1 checkpoint expression pathways.

Table 5. Predictive accuracy of the signature on a holdout part of the dataset.

REFERENCES

Borcoman et al. 2019. “Novel patterns of response under immunotherapy”. Ann. Oncol, 30(3): 385- 396.

Chan, Timothy A., Jedd D. Wolchok, and Alexandra Snyder. 2015. “Genetic Basis for Clinical Response to CTLA-4 Blockade in Melanoma.” The New England Journal of Medicine.

Champiat et al. 2018. “Hyperprogressive disease: recognizing a novel pattern to improve patient management”. Nat Rev Clin Oncol, 15(12): 748-762.

Davis AA, Patel VG 2019. “The role of PD-L1 expression as a predictive biomarker: an analysis of all US Food and Drug Administration (FDA) approvals of immune checkpoint inhibitors”. J. Immunother. Cancer 7(1): 278.

Ferrara et al. 2020. “Single or combined immune checkpoint inhibitors compared to first-line platinum-based chemotherapy with or without bevacizumab for people with advanced non-small cell lung cancer”. Cochrane Database Syst Rev; 12.

Fuentas-Antras et al., 2018. “Hyperprogression as a distinct outcome after immunotherapy”. Cancer Treat Rev. 70: 16-21.

Goodman, Aaron M., Shumei Kato, Lyudmila Bazhenova, Sandip P. Patel, Garrett M. Frampton, Vincent Miller, Philip J. Stephens, Gregory A. Daniels, and Razelle Kurzrock. 2017. “Tumor Mutational Burden as an Independent Predictor of Response to Immunotherapy in Diverse Cancers.” Molecular Cancer Therapeutics 16 (11): 2598-2608.

Hellmann, Matthew D., Tavi Nathanson, Hira Rizvi, Benjamin C. Creelan, Francisco Sanchez- Vega, Arun Ahuja, Ai Ni, et al. 2018. “Genomic Features of Response to Combination Immunotherapy in Patients with Advanced Non-Small-Cell Lung Cancer.” Cancer Cell 33 (5): 843- 52. e4.

Hugo, Willy, Jesse M. Zaretsky, Lu Sun, Chunying Song, Blanca Hornet Moreno, Siwen Hu- Lieskovan, Beata Berent-Maoz, et al. 2017. “Genomic and Transcriptomic Features of Response to Anti-PD-1 Therapy in Metastatic Melanoma.” Cell 168 (3): 542.

Kato, Shumei, Aaron Goodman, Vighnesh Walavalkar, Donald A. Barkauskas, Andrew Sharabi, and Razelle Kurzrock. 2017. “Hyperprogressors after Immunotherapy: Analysis of Genomic Alterations Associated with Accelerated Growth Rate.” Clinical Cancer Research: An Official Journal of the American Association for Cancer Research 23 (15): 4242-50.

Kim, Daehwan, Geo Pertea, Cole Trapnell, Harold Pimentel, Ryan Kelley, and Steven L. Salzberg. 2013. “TopHat2: Accurate Alignment of Transcriptomes in the Presence of Insertions, Deletions and Gene Fusions.” Genome Biology 14 (4): R36.

Russo, Giuseppe Lo, Massimo Moro, Michele Sommariva, Valeria Cancila, Mattia Boeri, Giovanni Centonze, Simona Ferro et al., 2019. “Antibody-Fc/FcR interaction on macrophages as a mechanism for hyperprogressive disease in non-small cell lung cancer subsequent to PD-1/PD- L1 blockade”. Clinical Cancer Research 25 (3): 989-999.

Matos, Ignacio, Juan Martin-Liberal, Alonso Garcia-Ruiz, Cinta Hierro, Maria Ochoa de Olza, Cristina Viaplana, Analia Azaro, et al. 2020. “Capturing Hyperprogressive Disease with Immune- Checkpoint Inhibitors Using RECIST 1.1 Criteria.” Clinical Cancer Research: An Official Journal of the American Association for Cancer Research 26 (8): 1846-55.

Rizvi, Naiyer A., Matthew D. Hellmann, Alexandra Snyder, Pia Kvistborg, Vladimir Makarov, Jonathan J. Havel, Wiliam Lee, et al. 2015. “Cancer Immunology. Mutational Landscape Determines Sensitivity to PD-1 Blockade in Non-Small Cell Lung Cancer.” Science 348 (6230): 124-28. Trapnell, Cole, Brian A. Williams, Geo Pertea, Ali Mortazavi, Gordon Kwan, Marijke J. van Baren, Steven L. Salzberg, Barbara J. Wold, and Lior Pachter. 2010. “Transcript Assembly and Quantification by RNA-Seq Reveals Unannotated Transcripts and Isoform Switching during Cell Differentiation.” Nature Biotechnology 28 (5): 511-15.

Claims

1. A method for determining whether a subject is predisposed for generating hyperprogressive disease when receiving treatment with an immunotherapeutic agent or for deciding whether a subject is eligible for treatment with an immunotherapeutic agent, the method comprising analyzing the expression level of the biomarkers IL-17A, IFNA1 , IFNB and IL-6 in a sample of the subject.

2. The method of claim 1 further comprising analyzing the expression level of one or more of the biomarkers selected from the group consisting of ICOSLG, CCL20, ID01 , FGF2, CXCL8, NRP1 , MX1 , VEGFA, CXCL12 and TGFB1 in a sample of the subject.

3. The method of claim 2, wherein the expression level of at least 2; preferably at least 3; of the biomarkers selected from the group consisting of ICOSLG, CCL20, ID01 , FGF2, CXCL8, NRP1 , MX1 , VEGFA, CXCL12 and TGFB1 are analyzed in a sample of the subject.

4. The method according to any of the preceding claims wherein the expression level of the biomarkers IL17A, IFNA1 , IFNB1 , IL-6, ICOSLG, CCL20, ID01 , FGF2, CXCL8, NRP1 , MX1 , VEGFA, CXCL12 and TGFB1 is analyzed in a sample of the subject.

5. The method according to any of the preceding claims wherein the expression level of the biomarkers is compared to a corresponding reference value or threshold value that is characteristic of a subject with a known response to treatment with the immunotherapeutic agent.

6. The method according to any of the preceding claims wherein on the basis of the expression level of the analyzed biomarkers a risk score is obtained, said risk score representing the likelihood for developing hyperprogressive disease in the subject, or for responding or not responding to treatment with an immunotherapeutic agent by the subject.

7. The method according to claim 6, wherein the risk score is obtained and calculated using a pre-trained machine learning model and the expression level of the biomarkers as input values.

8. The method according to any of the preceding claims 6 or 7, wherein the risk score is further improved by immunohistochemistry data of the tumor of the subject.

9. A kit for determining whether a subject is predisposed for generating hyperprogressive disease when receiving treatment with an immunotherapeutic agent or for deciding whether a subject is eligible for treatment with an immunotherapeutic agent, the kit comprising: means for measuring the expression level of the biomarkers IL-17A, IFNA1 , IFNB and IL- 6 in a sample of the subject; optionally, means for measuring the expression level of one or more of the biomarkers selected from the group consisting of ICOSLG, CCL20, ID01 , FGF2, CXCL8, NRP1 , MX1 , VEGFA, CXCL12 and TGFB1 ; optionally, a reference value or threshold value for the biomarkers IL-17A, IFNA1 , IFNB and IL-6; optionally, a reference value orthreshold value for one or more of the biomarkers selected from the group consisting of ICOSLG, CCL20, ID01 , FGF2, CXCL8, NRP1 , MX1 , VEGFA, CXCL12 and TGFB1.

10. Use of the kit according to claim 9 in a method according to any one of the claims 1 to 8.

11. A method of treating a subject diagnosed with cancer with an immunotherapeutic agent, the method comprising determining whether the subject is treatment with the immunotherapeutic agent by a method according to any one of the claims 1 to 8, and if the subject is found to be responsive to treatment with the immunotherapeutic agent, administering to the subject the immunotherapeutic agent.

12. A computer-implemented method for determining whether a subject is predisposed for generating hyperprogressive disease when receiving treatment with an immunotherapeutic agent, the method comprising:

(a) providing quantified expression levels of the biomarkers IL-17A, IFN1 , IFNB1 , IL-6, said biomarkers being quantified in a sample of the subject;

(b) optionally providing quantified expression levels of one or more biomarkers selected from ICOSLG, CCL20, ID01 , FGF2, CXCL8, NRP1 , MX1 , VEGFA, CXCL12 and TGFB1 , said biomarkers being quantified in a sample of the subject;

(c) normalizing the quantified expression levels whereby normalization occurs via comparison with data obtained from corresponding assessments and expression levels from a reference set;

(d) classifying, whether the normalized values of step (c) exceed a predetermined threshold;

(e) obtaining a risk score of said normalized values, wherein said risk score is calculated using a pre-trained machine learning model, and wherein the risk score represents the likelihood for developing hyperprogressive disease after treatment with the immunotherapeutic agent in said subject.

13. The method according to any of the preceding claims 1 to 8, or 11 to 12, or the kit of claim 9, or the use of claim 10 wherein the immunotherapeutic agent is an immune checkpoint inhibitor; preferably wherein the immunotherapeutic agent is selected from the group consisting of a PD-1 targeting agent, a PD-L1 targeting agent and a CTLA-4 targeting agent, or a combination thereof.

14. The method according to any of the preceding claims 1 to 8, or 11 to 13, or the kit of claim 9 or 13, or the use of claim 10 or 13 wherein the biomarker is a protein coding gene, in particular wherein the expression level of the biomarker is the RNA-based expression level of the protein coding gene.

15. The method according to any of the preceding claims 1 to 8, 11 to 14, or the kit of claim 9, 13 or 14, or the use of claim 10, 13 or 14, wherein the sample is a tumor tissue sample derived from the subject.

16. The method according to any of the preceding claims 1 to 8, or 11 to 15, or the kit of claim 9, 13, 14, or 15, or the use of claim 10, 13, 14 or 15 wherein the subject is a subject diagnosed with cancer, in particular wherein the subject is diagnosed with melanoma and/or lung cancer.