WO2011146928A2 - Methods for calibrating protein levels in cells and tissue specimens - Google Patents

Abstract

A method is described for the use of an internal calibrator marker for normalization of quantifiable protein detection in cells. This internal calibrator will ostensibly serve as an enumerator of epitope immune-integrity. Optimally, it will be a robust biomarker analyte of the same class (i.e protein) to that for which it is being held as a control, with consistent levels of expression across tissues but with a dynamic range commensurate with assay and treatment conditions. While stable expression in a biological context is a requirement, it will exhibit a suitable dynamic level of expression due to technical variation relative to a test marker to which it will serve as a surrogate.

Description

METHODS FOR CALIBRATING PROTEIN LEVELS IN CELLS AND TISSUE

SPECIMENS

RELATED APPLICATIONS

[0001] This application claims the benefits of U.S. provisional application No. 61/346,968, filed May 21, 2010, which is incorporated herein by reference in its entirety.

FIELD OF THE INVENTION

[0002] The invention relates to quantifying biomarker expression from tissue samples where there is a need to monitor the non-biomarker changes of the specimen that may influence the ability to accurately determine the amount of the biomarker.

BACKGROUND OF THE INVENTION

[0003] Molecular diagnostic tests have significant utility in cancer management applications. Biomarker expression is already known to provide meaningful data relative to diagnosis, prognosis, as well as offer valuable insights towards predicting therapeutic response for the individual patient. While diagnosis and prognostic information should not be discounted, information leading to correct, actionable decisions in patient management should prove most beneficial.

[0004] Protein biomarkers are measurable in terms of abundance so that quantitation will be expressed as a continuous score. For large numbers of biomarker measurements, the overall protein level may be used for normalization. For small numbers of biomarker assessment, additional strategies are required to determine a means to calibrate the general effect on a tissue sample.

[0005] To accurately measure the biomarkers of interest across several patients to accurately correlate expression with prediction, there must be a means to standardize measurements. One way to standardize measurements is through the use of calibrators. Thus a need exists for the identification of calibrators useful in normalizing biomarker expression across a patient population. BRIEF DESCRIPTION OF THE DRAWINGS

[0006] Figure 1 - SnRNP70 Expression Optimization and Characterization. Panels A and B depicts optimization of staining for a rabbit polyclonal Ab for SnRNP70. Four separate breast tumor -derived patient samples were stained with different Ab concentration and then scored for staining intensity. Staining is fairly consistent with some sample -to- sample variability; 2 out of the 4 samples are depicted in Panel B with 2 different Ab concentration. Small differences in QIM scores between samples in Panel A are possibly indicative of technical differences between the samples employed. Panel C indicates that the Ab is specific for identifying a single protein of appropriate size upon Western analysis.

[0007] Figure 2 - PABPN1 Expression Optimization and Characterization. Panels A, B, and C depict optimization of staining for a rabbit monoclonal Ab EP3000Y and Atlas Polyclonal Ab for PABPN1. Four separate breast tumor -derived patient samples were stained (Panels B and C) with different Ab concentration and then scored for staining intensity (Panel A). Staining very consistent among the 4 samples tested for both of these Abs and the same sample is slightly lower in QIM scoring for both Abs, which offers some level of cross-validation. Staining is nuclear specific and very clean for both Abs. Panel D indicates that rabbit MAb EP3000Y is specific for identifying a single protein of appropriate size upon Western analysis.

[0008] Figure 3 - SNRPA Expression Optimization and Characterization. Panels A, B, and C depict optimization of staining for a mouse monoclonal 3F9-1F7 and Genway Polyclonal Ab for SNRPA. Four separate breast tumor -derived patient samples were stained (Panels B and C) with different Ab concentration and then scored for staining intensity (Panel A). Similar to staining patterns among the same tissues employed in PABPN1 optimization, staining is very consistent among the 4 samples tested for both of these Abs and the same sample is slightly lower in QIM scoring for both Abs, which offers some level of cross-validation. Staining is nuclear specific and very clean for both Abs. Panels D and E indicate that both Abs are specific for identifying a single protein of appropriate size upon Western analysis.

[0009] Figure 4 - ZNF207 Expression Optimization and Characterization. Panels A, B, and C depict optimization of staining for a mouse monoclonal Ab 6D7 and Atlas Polyclonal Ab for ZNF207. Four separate breast tumor -derived patient samples were stained (Panels B and C) with different Ab cone and then scored for staining intensity (Panel A). Staining is somewhat consistent for the Atlas polyclonal Ab but was weak and variable between samples with the 6D7 mouse clone. Panels D and E both indicate that each Ab is specific for identifying a single protein of appropriate size upon Western analysis.

[00010] Figure 5 - HNRPM Expression Optimization and Characterization. Panels A, B, and C depict optimization of staining for a mouse monoclonal Abs for HNPRM, 2B6 and 3F7. Four separate breast tumor -derived patient samples were stained with different Ab

concentration and then scored for staining intensity. Staining is somewhat consistent, albeit with more sample-to- sample variability than was observed with some of the other Abs and calibrator biomarker candidates discussed here for breast cancers; 2 out of the 4 samples are depicted in Panels B and C with 2 different Ab cone. Additionally, the 2 Abs do not co-validate one another with respect to ordering of samples after QIM scoring; clone 2B6 would be considered less reliable than 3F7 as it had not been validated previously in either IHC or western analyses. Panel D indicates that clone 3F7 derived Ab is specific for identifying a single protein of appropriate size upon Western analysis.

[00011] Figure 6 - Staining of Candidate Calibrators in Human Tumor-derived

Xenografts. PAR Biomarker expression from three independent xenografts with nil, moderate, and low expressions is depicted on the right-hand column. QIM scores can be generated for PAR and normalized to QIM scores generated from serial sections of these same xenografts stained for candidate calibrators SNRPA (center column) and/or PABPN1 (left column).

[00012] Figure 7 - Candidate Calibrator to Normalize Technical Variation in Biomarker Staining in Human Tumor-derived Xenografts: PAR staining normalization to calibrator SNRPA. Panel A represents a graph of a continuum of QIM scores generated for a panel of 30 xenograft samples for the Calibrator marker SNRPA and test biomarker PAR ordered by increasing QIM score (lowest to highest). Panel B shows broken out for each individual xenograft the raw QIM score for SNRPA and PAR, as well as a normalized score derived from a ratio gained by QIMtest/QIMcalibrator. Within a continuum of scoring, the ordering for individual samples within this population ( tile rank) changes for many of the samples when calibrator normalization strategies are applied (panel C).

[00013] Figure 8 - Establishing Levels of Expression Variation in Across a Patient Population for Candidate Calibrator Markers. Serial sections from a commercial TMA comprising breast cancer patients was stained and scored for canidate calibrator biomarkers of interest.

[00014] Figure 9 - A series of plots to depict relative patient order rank change. In these figures the rank ordering of patients from non-calibrated patient Q- score data is plotted against patient rank order ratiometrics generated from a Q- score of the test biomarker of interest normalized to a candidate calibrator of interest, here SnRNP70. R values indicate an estimate on the amount of rank ordering rearrangement of patients occurs by normalized patient data relative to raw Q-scores.

DETAILED DESCRIPTION

[00015] Calibration of immunohistochemistry (IHC) of cell and tissue samples has become increasingly important due to the emergence of the field molecular diagnostics. Calibration of IHC has been a complex issue to approach and solve due to a variety of factors. Chief among these are variable conditions for fixation and tissue processing for samples newly processed as well as archived material used in discovery and clinical studies. This topic has been recognized as a critical issue since 1977 at the First National Cancer Institute Workshop on the

standardization of IHC reagents (DeLellis et al. 1979). While efforts have been undertaken to adopt better standard protocols, user variability, reagent consistency, sample age and storage and tissue thickness are all variables that may influence quantification of a given antigen. The variation due to the delay, as well as the duration of fixation is also a key component that can only be controlled for in a limited fashion (Shi et al., 2006). The absolute quantification of a biomarker of interest (such as a phosphoprotein) may be prone to technical variation that will also contribute to expression differences. Thus, there is a need to be able to properly modify the score by calibration or normalization so that the amended score primarily reflects only the true expression (or biologic) differences among samples devoid of variation due to technical issues.

[00016] In profiling and quantifying biomarker expression, establishment of a scoring strategy to incorporate the greatest diagnostic sensitivity while maintaining high specificity is paramount. It is therefore important to establish that detectable variation in biomarker expression patterns is solely due to biological criteria of the cell or tissue sample relative to a continuum of scores for this same marker or set of markers from a patient population. The invention described herein provides composition and methods to calibrate the overall biomarker(s) score to correct for technical variations including but not limited to age of tissue, section thickness, and tissue fixation, so that samples can be more properly placed in a continuum of responders and non- responders in the sole context of biological differences of the patient cells/tissue under interrogation.

[00017] In particular, tissue cancer specimens when measured with biomarkers will exhibit a wide dynamic range of expression values. Thus in order to provide key information in the ability to determine whether a single specimen from one patient is a member of a selected patient group for treatment response and/or survival, it is valuable to use calibration and normalization schemas. The present invention provides a means to attenuate the placement of patient samples for a given biomarker(s) along a reference curve, and to refine cut-points in the process of forming a patient ranking strategy. The present invention also provides a method to have every patient specimen individually analyzable along this reference curve.

[00018] Accordingly, the invention provides calibrator markers useful for normalizing expression levels of clinically relevant biomarkers in cell and tissue samples obtained from a subject. The normalization method may be broadly applied to any setting where biomarker expression is evaluated. The methods of the invention described herein can be used in any method that requires evaluation of biomarker expression levels of one or more proteins.

[00019] Suitable calibrator marker would have consistent level of expression relative to biomarkers under consideration as well as across a population of specimens within an evaluation group. Chief among macromolecules that are consistent in their levels of expression are the protein products of uniformly expressed genes, such as housekeeping genes. RNA expression transcriptome data may be evaluated to identify genes where there are uniform or consistent mRNA levels amongst specimens (Popovici et al., 2009; Kwon et al., 2009). Further, it may also be evaluated for this subset of genes with uniform expression whether the protein levels are also uniform. In principle, these features of consistent RNA and protein expression levels may be independent of the overall expression level.

[00020] In particular, the invention is based, at least in part, on the discovery of calibrators used in conjunction with DNA Response and Repair biomarkers. One area where biomarkers are relevant to clinical decisions about cancer therapies is in the DNA repair pathways. Table 1, list exemplary DNA Repair and DNA Damage Response Markers (DNARMARKERS). There are many markers from DNA repair and DNA damage pathways having wide expression ranges against which to judge response. However, other DNA repair and DNA damage markers display a more consistent level of expression. Markers that display a more consistent level of expression would not be considered as informative based on the ability to predict effectiveness of treatment regimens because they do not vary. However these markers would be suited as calibrators for normalizing dynamic and informative biomarkers expressions. For example, gene products derived from the Non-homologous end joining repair pathway (NHEJ) are quite stable in their expression levels across a wide variety of tumors types. These enzymes include but are not limited to Ku70, Ku80, XRCC4, and DNA PK. Other suitable candidate calibrator markers include those listed on Table 2. Additionally, other characteristics possessed by DNA repair enzymes that are stably expressed favorable to their use as calibrators include nucleus localization (identical to the enzyme for which it serves as calibrator), similar levels of protein stability, and orthologous point of action at the level of DNA binding and base modification.

[00021] In silico analyses of genes and expression patterns listed in Table 2, identified five candidate calibrator markers on the basis of consistent expression patterns across a wide range of tumor tissues and cell lines. These candidate calibrator markers include, SnRNP70, PABPN1, and SNRPA, ZNF207 and HNPRM. Additional criteria that were applied in prioritizing candidates are the availability of well-characterized antibodies with clean IHC expression profiling data, and/or single band of appropriate size recognition upon Western analyses and the absence of information associating of aberrant expression of the candidate gene product with disease. Figures 1 through 5 depict optimization and characterization of antibody staining for the five calibrator candidates. Consistent patterns of strong nuclear specific expression are observed for several of the biomarkers for this set of breast cancer specimens analyzed including

SnRNP70, PABPN1, and SNRPA (Figures 1, 2, and 3) while expression variation among the set of breast samples is more variable for ZNF207 and HNPRM (Figures 4 and 5). These observations are confirmed on a larger set (n=75) of breast cancer samples in Figure 8 with SNRPA, PABPN1, and SnRNP70 displaying relatively consistent levels of expression and ZNF207 and HNPRM exhibiting sample-to-sample variability.

[00022] Additionally, in the cases of SNRPA and PABPN1, the expression patterns are confirmed with 2 available antibodies, serving as cross-validation for staining specificity. The specificity of several of the antibodies has also been established through Western analyses. In the case of PABPN1, the human NSCLC-derived cell line A549 is a known to underexpress this protein. In the course of optimization we confirmed the specificity of this observation with a QIM score of 40 relative to the remainder of the set carrying a mean QIM of approx 280. In applying normalization strategies from an analyte of consistent expression as a calibrator, the QIM score generated from a test biomarker serves as the numerator while the score generated from the candidate calibrator will serves as the denominator.

[00023] Definitions

[00024] "Accuracy" refers to the degree of conformity of a measured or calculated quantity (a test reported value) to its actual (or true) value. Clinical accuracy relates to the proportion of true outcomes (true positives (TP) or true negatives (TN) versus misclassified outcomes (false positives (FP) or false negatives (FN)), and may be stated as a sensitivity, specificity, positive predictive values (PPV) or negative predictive values (NPV), or as a likelihood, odds ratio, among other measures.

[00025] "Biological state" of a subject is the condition of the subject, as with respect to circumstances or attributes of the biological condition.

[00026] "Biological condition" of a subject is the condition of the subject in a pertinent realm that is under observation, and such realm may include any aspect of the subject capable of being monitored for change in condition, such as health; disease including cancer; trauma; aging; infection; tissue degeneration; developmental steps; physical fitness; obesity; and mental state. As can be seen, a condition in this context may be chronic or acute or simply transient.

Moreover, a targeted biological condition may be manifest throughout the organism or population of cells or may be restricted to a specific organ (such as skin, heart, eye or blood) but in either case, the condition may be monitored directly by a sample of the affected population of cells or indirectly by a sample derived elsewhere from the subject. The term "biological condition" includes a "physiological condition". For example, the biological condition is cancer such as prostate cancer, ovarian cancer, lung cancer, breast cancer, skin cancer, colon cancer, or cervical cancer.

[00027] "Biomarker" in the context of the present invention encompasses, without limitation, proteins, nucleic acids, and metabolites, together with their polymorphisms, mutations, variants, modifications, subunits, fragments, protein-ligand complexes, and degradation products, protein- ligand complexes, elements, related metabolites, and other analytes or sample-derived measures. Biomarkers can also include mutated proteins or mutated nucleic acids. Biomarkers also encompass non-blood borne factors or non-analyte physiological markers of health status, such as "clinical parameters" defined herein, as well as "traditional laboratory risk factors", also defined herein. Biomarkers also include any calculated indices created mathematically or combinations of any one or more of the foregoing measurements, including temporal trends and differences. Where available, and unless otherwise described herein, determinants which are gene products are identified based on the official letter abbreviation or gene symbol assigned by the international Human Genome Organization Naming Committee (HGNC) and listed at the date of this filing at the US National Center for Biotechnology Information (NCBI) web site.

[00028] "Calibrator Marker" is an analyte that permits the determination of a quantity of a biomarker present in a sample. The expression level of the calibrator marker being constant across a plurality of samples and (ii) from the expression being specific for said one or more chosen cell type(s). It is sufficient that the calibrator markers are constantly expressed across the set of samples under consideration. Nevertheless, it is envisaged that the calibrator markers are constantly expressed one or more chosen cell type(s) under most or all conceivable conditions. The term "constant per cell" means that each cell of one or more chosen cell type(s) expresses the same or substantially the same amount of transcript and/or protein of the calibrator marker. The term "specific for one or more chosen cell type(s)" in relation to expression designates calibrator markers whose detectable expression is confined or substantially confined to one or more chosen cell type(s). The term "chosen cell type(s)" may refer to a subset of the cell types present in the sample. Alternatively, the chosen cell types may embrace all cell types present in the sample. In both cases, the chosen cell type(s) is/are (a) cell types for which calibrator markers are known. In case of a plurality of chosen cell types, these calibrator markers are constantly expressed in all chosen cell types, preferably at identical or substantially identical levels across the different cell types comprised in the set of chosen cell types.

[00029] "Clinical parameters" encompasses all non-sample or non-analyte biomarkers of subject health status or other characteristics, such as, without limitation, age (Age), ethnicity (RACE), gender (Sex), or family history (FamHX).

[00030] Expression data" and "gene expression data" refer to quantitative data characterizing the RNA expression level and/or the protein expression level of one or more genes. In other words, the term "gene expression data" as used herein comprises both "RNA expression data" and "protein expression data". The methods of the invention can be performed irrespective of the specific type of expression data. The skilled person is aware of methods for the quantitation of RNA and proteins. Expression levels of purified protein in solution can be determined by physical methods, e.g. photometry. Methods of determining the expression level of a particular protein in a mixture rely on specific binding, e.g of antibodies. Specific detection and quantitation methods exploiting the specificity of antibodies comprise immunohistochemistry (in situ) and surface plasmon resonance. Western blotting combines separation of a mixture of proteins by electrophoresis and specific detection with antibodies. Other means of determining protein expression data include two-dimensional gel-electrophoresis, preferably in combination with mass spectrometry. With the advent of microarray technology, measurement of protein expression levels in array format became increasingly widespread. Protein arrays for determining protein expression data exploit interactions such as protein-antibody, protein-protein, protein- ligand, protein-drug and protein- small molecule interactions or any combination thereof. Protein expression data reflect, in addition to regulation at the transcriptional level, regulation at the translational level as well as the average lifetime of a protein prior to degradation.

[00031] "FN" is false negative, which for a disease state test means classifying a disease subject incorrectly as non-disease or normal.

[00032] "FP" is false positive, which for a disease state test means classifying a normal subject incorrectly as having disease.

[00033] A "formula," "algorithm," or "model" is any mathematical equation, algorithmic, analytical or programmed process, or statistical technique that takes one or more continuous or categorical inputs (herein called "parameters") and calculates an output value, sometimes referred to as an "index" or "index value." Non-limiting examples of "formulas" include sums, ratios, and regression operators, such as coefficients or exponents, biomarker value

transformations and normalizations (including, without limitation, those normalization schemes based on clinical parameters, such as gender, age, or ethnicity), rules and guidelines, statistical classification models, and neural networks trained on historical populations. Of particular use in combining biomarkers are linear and non-linear equations and statistical classification analyses to determine the relationship between levels of biomarkers detected in a subject sample and the subject's responsiveness to chemotherapy. In panel and combination construction, of particular interest are structural and synactic statistical classification algorithms, and methods of risk index construction, utilizing pattern recognition features, including established techniques such as cross-correlation, Principal Components Analysis (PCA), factor rotation, Logistic Regression (LogReg), Linear Discriminant Analysis (LDA), Eigengene Linear Discriminant Analysis (ELD A), Support Vector Machines (SVM), Random Forest (RF), Recursive Partitioning Tree (RPART), as well as other related decision tree classification techniques, Shrunken Centroids (SC), StepAIC, Kth-Nearest Neighbor, Boosting, Decision Trees, Neural Networks, Bayesian Networks, Support Vector Machines, and Hidden Markov Models, among others. Other techniques may be used in survival and time to event hazard analysis, including Cox, Weibull, Kaplan-Meier and Greenwood models well known to those of skill in the art. Many of these techniques are useful either combined with a biomarker selection technique, such as forward selection, backwards selection, or stepwise selection, complete enumeration of all potential panels of a given size, genetic algorithms, or they may themselves include biomarker selection methodologies in their own technique. These may be coupled with information criteria, such as Akaike's Information Criterion (AIC) or Bayes Information Criterion (BIC), in order to quantify the tradeoff between additional biomarkers and model improvement, and to aid in minimizing overfit. The resulting predictive models may be validated in other studies, or cross-validated in the study they were originally trained in, using such techniques as Bootstrap, Leave-One-Out (LOO) and 10-Fold cross-validation (10-Fold CV). At various steps, false discovery rates may be estimated by value permutation according to techniques known in the art. A "health economic utility function" is a formula that is derived from a combination of the expected probability of a range of clinical outcomes in an idealized applicable patient population, both before and after the introduction of a diagnostic or therapeutic intervention into the standard of care. It encompasses estimates of the accuracy, effectiveness and performance characteristics of such intervention, and a cost and/or value measurement (a utility) associated with each outcome, which may be derived from actual health system costs of care (services, supplies, devices and drugs, etc.) and/or as an estimated acceptable value per quality adjusted life year (QALY) resulting in each outcome. The sum, across all predicted outcomes, of the product of the predicted population size for an outcome multiplied by the respective outcomes expected utility is the total health economic utility of a given standard of care. The difference between (i) the total health economic utility calculated for the standard of care with the intervention versus (ii) the total health economic utility for the standard of care without the intervention results in an overall measure of the health economic cost or value of the intervention. This may itself be divided amongst the entire patient group being analyzed (or solely amongst the intervention group) to arrive at a cost per unit intervention, and to guide such decisions as market positioning, pricing, and assumptions of health system acceptance. Such health economic utility functions are commonly used to compare the cost-effectiveness of the intervention, but may also be transformed to estimate the acceptable value per QALY the health care system is willing to pay, or the acceptable cost- effective clinical performance characteristics required of a new intervention.

[00034] For diagnostic (or prognostic) interventions of the invention, as each outcome (which in a disease classifying diagnostic test may be a TP, FP, TN, or FN) bears a different cost, a health economic utility function may preferentially favor sensitivity over specificity, or PPV over NPV based on the clinical situation and individual outcome costs and value, and thus provides another measure of health economic performance and value which may be different from more direct clinical or analytical performance measures. These different measurements and relative trade-offs generally will converge only in the case of a perfect test, with zero error rate (a.k.a., zero predicted subject outcome misclassifications or FP and FN), which all performance measures will favor over imperfection, but to differing degrees.

[00035] "Measuring" or "measurement," or alternatively "detecting" or "detection," means assessing the presence, absence, quantity or amount (which can be an effective amount) of either a given substance within a clinical or subject-derived sample, including the derivation of qualitative or quantitative concentration levels of such substances, or otherwise evaluating the values or categorization of a subject's non-analyte clinical parameters.

[00036] "Negative predictive value" or "NPV" is calculated by TN/(TN + FN) or the true negative fraction of all negative test results. It also is inherently impacted by the prevalence of the disease and pre-test probability of the population intended to be tested.

[00037] See, e.g., O'Marcaigh AS, Jacobson RM, "Estimating The Predictive Value Of A Diagnostic Test, How To Prevent Misleading Or Confusing Results," Clin. Ped. 1993, 32(8): 485-491, which discusses specificity, sensitivity, and positive and negative predictive values of a test, e.g., a clinical diagnostic test. Often, for binary disease state classification approaches using a continuous diagnostic test measurement, the sensitivity and specificity is summarized by Receiver Operating Characteristics (ROC) curves according to Pepe et al, "Limitations of the Odds Ratio in Gauging the Performance of a Diagnostic, Prognostic, or Screening Marker," Am. J. Epidemiol 2004, 159 (9): 882-890, and summarized by the Area Under the Curve (AUC) or c- statistic, an indicator that allows representation of the sensitivity and specificity of a test, assay, or method over the entire range of test (or assay) cut points with just a single value. See also, e.g., Shultz, "Clinical Interpretation Of Laboratory Procedures," chapter 14 in Teitz,

Fundamentals of Clinical Chemistry, Burtis and Ashwood (eds.), 4^th edition 1996, W.B.

Saunders Company, pages 192-199; and Zweig et al., "ROC Curve Analysis: An Example Showing The Relationships Among Serum Lipid And Apolipoprotein Concentrations In Identifying Subjects With Coronory Artery Disease," Clin. Chem., 1992, 38(8): 1425-1428. An alternative approach using likelihood functions, odds ratios, information theory, predictive values, calibration (including goodness-of-fit), and reclassification measurements is summarized according to Cook, "Use and Misuse of the Receiver Operating Characteristic Curve in Risk Prediction," Circulation 2007, 115: 928-935.

[00038] Finally, hazard ratios and absolute and relative risk ratios within subject cohorts defined by a test are a further measurement of clinical accuracy and utility. Multiple methods are frequently used to defining abnormal or disease values, including reference limits, discrimination limits, and risk thresholds.

[00039] "Normalizing" in relation to expression data is common in the art and relates to a processing step of the raw expression data which renders the signal intensities of each gene comparable across multiple measurements. Expression levels of a particular gene/protein may differ between samples for a variety of reasons. Reasons of particular relevance are different amounts of cells n the samples analyzed on the one side and different transcriptional activity of the gene(s) under consideration on the other side. While the former is generally not indicative of a distinct biological state of the samples being compared, the latter generally is. In case protein expression levels are monitored instead of or in addition to RNA expression levels, different transcriptional and/or translational activity may contribute to different protein expression levels. Meaningful analysis of expression data requires the two possible contributions to changes in expression levels-amount of cells and/or RNA vs. transcriptional and/or translational activity-to be disentangled. Normalization is a method for disentangling said contributions. Practically speaking, normalization is a transformation of the raw expression data such that the effect of different amounts of cells and/or of RNA is removed or substantially removed. Global normalization, a procedure well known in the art, for example involves (i) the determination of the average signal intensity across all genes whose expression is being measured and (ii) subsequent division of raw signal intensities by the average signal intensity obtained in step (i).

[00040] "Analytical accuracy" refers to the reproducibility and predictability of the

measurement process itself, and may be summarized in such measurements as coefficients of variation, and tests of concordance and calibration of the same samples or controls with different times, users, equipment and/or reagents. These and other considerations in evaluating new biomarkers are also summarized in Vasan, 2006.

[00041] "Performance" is a term that relates to the overall usefulness and quality of a diagnostic or prognostic test, including, among others, clinical and analytical accuracy, other analytical and process characteristics, such as use characteristics (e.g., stability, ease of use), health economic value, and relative costs of components of the test. Any of these factors may be the source of superior performance and thus usefulness of the test, and may be measured by appropriate "performance metrics," such as AUC, time to result, shelf life, etc. as relevant.

[00042] "Positive predictive value" or "PPV" is calculated by TP/(TP+FP) or the true positive fraction of all positive test results. It is inherently impacted by the prevalence of the disease and pre-test probability of the population intended to be tested.

[00043] "Raw expression data" specifically refers to expression data prior to normalization. For example, and in the case of expression data obtained from microarrays, raw expression data are the data obtained from the image processing of the scanned hybridized microarray.

[00044] "Risk" in the context of the present invention, relates to the probability that an event will occur over a specific time period, as in the responsiveness to treatment, and can mean a subject's "absolute" risk or "relative" risk. Absolute risk can be measured with reference to either actual observation post-measurement for the relevant time cohort, or with reference to index values developed from statistically valid historical cohorts that have been followed for the relevant time period. Relative risk refers to the ratio of absolute risks of a subject compared either to the absolute risks of low risk cohorts or an average population risk, which can vary by how clinical risk factors are assessed. Odds ratios, the proportion of positive events to negative events for a given test result, are also commonly used (odds are according to the formula p/(l-p) where p is the probability of event and (1- p) is the probability of no event) to no-conversion.

[00045] "Risk evaluation," or "evaluation of risk" in the context of the present invention encompasses making a prediction of the probability, odds, or likelihood that an event or disease state may occur, the rate of occurrence of the event or conversion from one disease state. Risk evaluation can also comprise prediction of future clinical parameters, traditional laboratory risk factor values, or other indices of cancer, either in absolute or relative terms in reference to a previously measured population. The methods of the present invention may be used to make continuous or categorical measurements of the responsiveness to treatment thus diagnosing and defining the risk spectrum of a category of subjects defined as being at responders or non- responders. In the categorical scenario, the invention can be used to discriminate between normal and other subject cohorts at higher risk for responding. Such differing use may require different biomarker/calibrator marker combinations and individualized panels, mathematical algorithms, and/or cut-off points, but be subject to the same aforementioned measurements of accuracy and performance for the respective intended use.

[00046] A "sample" in the context of the present invention is a biological sample isolated from a subject and can include, by way of example and not limitation, tissue biopies, whole blood, serum, plasma, blood cells, endothelial cells, lymphatic fluid, ascites fluid, interstitital fluid (also known as "extracellular fluid" and encompasses the fluid found in spaces between cells, including, inter alia, gingival crevicular fluid), bone marrow, cerebrospinal fluid (CSF), saliva, mucous, sputum, sweat, urine, or any other secretion, excretion, or other bodily fluids.

[00047] "Sensitivity" is calculated by TP/(TP+FN) or the true positive fraction of disease subjects.

[00048] "Specificity" is calculated by TN/(TN+FP) or the true negative fraction of non- disease or normal subjects.

[00049] "Signal intensity" as used herein refers to a measured quantity indicative of the expression level of a gene. Preferably, the signal intensity is proportional to the amount of transcript or the amount of protein translated from a gene. Depending on the label used, which is further detailed below, the signal may be light emitted by a fluorescence or luminescent process, or radiation and/or particles emitted by a radioactive label or dye (quantum dots).

[00050] By "statistically significant", it is meant that the alteration is greater than what might be expected to happen by chance alone (which could be a "false positive"). Statistical significance can be determined by any method known in the art. Commonly used measures of significance include the p-value, which presents the probability of obtaining a result at least as extreme as a given data point, assuming the data point was the result of chance alone. A result is often considered highly significant at a p-value of 0.05 or less.

[00051] A "subject" in the context of the present invention is preferably a mammal. The mammal can be a human, non-human primate, mouse, rat, dog, cat, horse, or cow, but are not limited to these examples. Mammals other than humans can be advantageously used as subjects that represent animal models of cancer. A subject can be male or female.

[00052] "TN" is true negative, which for a disease state test means classifying a non-disease or normal subject correctly.

[00053] "TP" is true positive, which for a disease state test means correctly classifying a disease subject.

[00054] METHODS OF THE INVENTION

[00055] The present invention provides methods of normalizing gene expression levels.

Expression levels are usually normalized per total amount of RNA or protein in the sample and an endogenous control gene, which is typically a house-keeping gene. The present invention is based in par on using an internal reference standard (i.e., calibrator marker) for normalization of quantifiable protein detection in cells. Optimally, the calibrator marker is a robust biomarker analyte of the same class (i.e., protein) to that for which it is being held as a control, with consistent levels of expression across tissues but with a dynamic range commensurate with the assay and treatment conditions.

[0001] The actual measurement of levels (i.e., expression levels) or amounts of the biomarkers and calibrator markers can be determined at the protein or nucleic acid level using any method known in the art. For example, at the nucleic acid level, Northern and Southern hybridization analysis, as well as ribonuclease protection assays using probes which specifically recognize one or more of these sequences can be used to determine gene expression.

Alternatively, amounts of biomarkers and calibrator markers can be measured using reverse- transcription-based PCR assays (RT-PCR), e.g., using primers specific for the differentially expressed sequence of genes or by branch-chain RNA amplification and detection methods by Panomics, Inc. Amounts of biomarkers and calibrator markers can also be determined at the protein level, e.g., by measuring the levels of peptides encoded by the gene products described herein, or subcellular localization or activities thereof using technological platform such as for example AQUA. Such methods are well known in the art and include, e.g., immunoassays based on antibodies to proteins encoded by the genes, aptamers or molecular imprints. Any biological material can be used for the detection/quantification of the protein or its activity. Alternatively, a suitable method can be selected to determine the activity of proteins encoded by the marker genes according to the activity of each protein analyzed.

[0002] The biomarkers and calibrator markers proteins, polypeptides, mutations, and polymorphisms thereof can be detected in any suitable manner, but is typically detected by contacting a sample from the subject with an antibody which binds the biomarkers and calibrator markers protein, polypeptide, mutation, or polymorphism and then detecting the presence or absence of a reaction product. The antibody may be monoclonal, polyclonal, chimeric, or a fragment of the foregoing, as discussed in detail above, and the step of detecting the reaction product may be carried out with any suitable immunoassay. The sample from the subject is typically a biological fluid as described above, and may be the same sample of biological fluid used to conduct the method described above.

[0003] Immunoassays carried out in accordance with the present invention may be homogeneous assays or heterogeneous assays. In a homogeneous assay the immunological reaction usually involves the specific antibody, a labeled analyte, and the sample of interest. The signal arising from the label is modified, directly or indirectly, upon the binding of the antibody to the labeled analyte. Both the immunological reaction and detection of the extent thereof can be carried out in a homogeneous solution. Immunochemical labels which may be employed include free radicals, radioisotopes, fluorescent dyes, enzymes, bacteriophages, or coenzymes.

[0004] In a heterogeneous assay approach, the reagents are usually the sample, the antibody, and means for producing a detectable signal. Samples as described above may be used. The antibody can be immobilized on a support, such as a bead (such as protein A and protein G agarose beads), plate or slide, and contacted with the specimen suspected of containing the antigen in a liquid phase. The support is then separated from the liquid phase and either the support phase or the liquid phase is examined for a detectable signal employing means for producing such signal. The signal is related to the presence of the analyte in the sample. Means for producing a detectable signal include the use of radioactive labels, fluorescent labels, or enzyme labels. For example, if the antigen to be detected contains a second binding site, an antibody which binds to that site can be conjugated to a detectable group and added to the liquid phase reaction solution before the separation step. The presence of the detectable group on the solid support indicates the presence of the antigen in the test sample. Examples of suitable immunoassays are oligonucleotides, immunoblotting, immunofluorescence methods,

immunoprecipitation, quantum dots, multiplex fluorochromes, chemiluminescence methods, electrochemiluminescence (ECL) or enzyme-linked immunoassays.

[0005] Those skilled in the art will be familiar with numerous specific immunoassay formats and variations thereof which may be useful for carrying out the method disclosed herein. See generally E. Maggio, Enzyme-Immunoassay, (1980) (CRC Press, Inc., Boca Raton, Fla.); see also U.S. Pat. No. 4,727,022 to Skold et al. titled "Methods for Modulating Ligand-Receptor Interactions and their Application," U.S. Pat. No. 4,659,678 to Forrest et al. titled "Immunoassay of Antigens," U.S. Pat. No. 4,376,110 to David et al., titled "Immunometric Assays Using Monoclonal Antibodies," U.S. Pat. No. 4,275,149 to Litman et al., titled "Macromolecular Environment Control in Specific Receptor Assays," U.S. Pat. No. 4,233,402 to Maggio et al., titled "Reagents and Method Employing Channeling," and U.S. Pat. No. 4,230,767 to Boguslaski et al., titled "Heterogenous Specific Binding Assay Employing a Coenzyme as Label."

[0006] Exemplary antibodies include monoclonal Ab 3F9-1F7 (SNRPA), monoclonal Ab E3000Y (PABPNl), monoclonal Ab 3F7 (HNPRM), monoclonal Ab 2B6 (HNPRM), monoclonal Ab 6D7 (ZNF207),polyclonal Ab F-21 (SNRPA), Genway anti-polyclonal Ab (SNRPA), polyclonal Ab to SnRNP70 (Genway), polyclonal Ab G-17 (PABPNl), polyclonal anti-PABP2 (Atlas), monoclonal Ab Y-39 (ZNF207)

[0007] Antibodies can be conjugated to a solid support suitable for a diagnostic assay (e.g., beads such as protein A or protein G agarose, microspheres, plates, slides or wells formed from materials such as latex or polystyrene) in accordance with known techniques, such as passive binding. Antibodies as described herein may likewise be conjugated to detectable labels or groups such as radiolabels (e.g., 35 S, 125 I, 131 I), enzyme labels (e.g., horseradish peroxidase, alkaline phosphatase), and fluorescent labels (e.g., fluorescein, Alexa, green fluorescent protein, rhodamine) in accordance with known techniques. Highly sensitivity antibody detection strategies may be used that allow for evaluation of the antigen-antibody binding in a non- amplified configuration. In addition, antibodies may be conjugated to oligonucleotides, andfollowed by Polymerase Chain Reaction and a variety of oligonucleotide detection methods.

[0008] Antibodies can also be useful for detecting post-translational modifications of proteins, polypeptides, mutations, and polymorphisms, such as tyrosine phosphorylation, threonine phosphorylation, serine phosphorylation, glycosylation (e.g., O-GlcNAc). Such antibodies specifically detect the phosphorylated amino acids in a protein or proteins of interest, and can be used in immunoblotting, immunofluorescence, and ELISA assays described herein. These antibodies are well-known to those skilled in the art, and commercially available. Post- translational modifications can also be determined using metastable ions in reflector matrix- assisted laser desorption ionization-time of flight mass spectrometry (MALDI-TOF) (Wirth, U. et al. (2002) Proteomics 2(10): 1445-51). In addition to post-translation modifications, these processes may be coupled to localization of the protein, such that a re-localization process is monitored, and the biomarker is evaluated in a relative fashion exhibited by the constancy or change to the ratio of the protein in different compartments. For biomarker and calibrator marker proteins, polypeptides, mutations, and polymorphisms known to have enzymatic activity, the activities can be determined in vitro using enzyme assays known in the art. Such assays include, without limitation, kinase assays, phosphatase assays, reductase assays, among many others. Modulation of the kinetics of enzyme activities can be determined by measuring the rate constant KM using known algorithms, such as the Hill plot, Michaelis-Menten equation, linear regression plots such as Lineweaver-Burk analysis, and Scatchard plot.

[0009] Using sequence information provided by the database entries for the biomarkers and calibrator markers sequences, expression of the biomarkers and calibrator markers equences can be detected (if present) and measured using techniques well known to one of ordinary skill in the art. For example, sequences within the sequence database entries corresponding to biomarkers and calibrator markers sequences, or within the sequences disclosed herein, can be used to construct probes for detecting biomarker and calibrator marker RNA sequences in, e.g., Northern blot hybridization analyses or methods which specifically, and, preferably, quantitatively amplify specific nucleic acid sequences. As another example, the sequences can be used to construct primers for specifically amplifying the sequences in, e.g., amplification-based detection methods such as reverse-transcription based polymerase chain reaction (RT-PCR). When alterations in gene expression are associated with gene amplification, deletion, polymorphisms, and mutations, sequence comparisons in test and reference populations can be made by comparing relative amounts of the examined DNA sequences in the test and reference cell populations.

[00010] Expression of the genes disclosed herein can be measured at the RNA level using any method known in the art. For example, Northern hybridization analysis using probes which specifically recognize one or more of these sequences can be used to determine gene expression. Alternatively, expression can be measured using reverse-transcription-based PCR assays (RT- PCR), e.g., using primers specific for the differentially expressed sequences. RNA can also be quantified using, for example, other target amplification methods (e.g., TMA, SDA, NASBA), or signal amplification methods (e.g., bDNA), and the like.

[00011] Alternatively, protein and nucleic acid metabolites can be measured. The term "metabolite" includes any chemical or biochemical product of a metabolic process, such as any compound produced by the processing, cleavage or consumption of a biological molecule (e.g., a protein, nucleic acid, carbohydrate, or lipid). Metabolites can be detected in a variety of ways known to one of skill in the art, including the refractive index spectroscopy (RI), ultra-violet spectroscopy (UV), fluorescence analysis, radiochemical analysis, near-infrared spectroscopy (near-IR), nuclear magnetic resonance spectroscopy (NMR), light scattering analysis (LS), mass spectrometry, pyrolysis mass spectrometry, nephelometry, dispersive Raman spectroscopy, gas chromatography combined with mass spectrometry, liquid chromatography combined with mass spectrometry, matrix-assisted laser desorption ionization-time of flight (MALDI-TOF) combined with mass spectrometry, ion spray spectroscopy combined with mass spectrometry, capillary electrophoresis, NMR and IR detection. (See, WO 04/056456 and WO 04/088309, each of which are hereby incorporated by reference in their entireties) In this regard, other analytes can be measured using the above-mentioned detection methods, or other methods known to the skilled artisan. For example, circulating calcium ions (Ca²⁺) can be detected in a sample using fluorescent dyes such as the Fluo series, Fura-2A, Rhod-2, among others.

[00012] Methods of the invention involve analysis of gene expression levels in a biological sample. A biological sample may contain material obtained cells or tissues, e.g., a cell or tissue lysate or extract. Extract may contain material enriched in sub-cellular elements such as that from the Golgi complex, mitochondria, lysosomes, the endoplasmic reticulum, cell membrane, and cytoskeleton, etc. In some embodiments, the biological sample contains materials obtained from a single cell.

[00013] Biological samples can come from a variety of sources. For examples, biological samples may be obtained from whole organisms, organs, tissues, or cells from different stages of development, differentiation, or disease state, and from different species (human and non-human, including bacteria and virus). The samples may represent different treatment conditions (e.g., test compounds from a chemical library), tissue or cell types, or source (e.g., blood, urine, cerebrospinal fluid, seminal fluid, saliva, sputum, stool), etc.

[00014] Tissues may be obtained from a subject using any of the methods known in the art. As used herein, a "subject" refers to a human or animal, including all mammals such as primates (particularly higher primates), sheep, dog, rodents (e.g., mouse or rat), guinea pig, goat, pig, cat, rabbit, and cow. In a preferred embodiment, the subject is a human. In another embodiment, the subject is an experimental animal or animal suitable as a disease model. A "tissue" sample from a subject may be a biopsy specimen sample, a normal or benign tissue sample, a cancer or tumor tissue sample, a freshly prepared tissue sample, a frozen tissue sample, a formalin fixed paraffin embedded sample, a primary cancer or tumor sample, or a metastasis sample. Exemplary tissues include, but are not limited to, epithelial, connective, muscle, nervous, heart, lung, brain, eye, stomach, spleen, bone, pancreatic, kidney, gastrointestinal, skin, uterus, thymus, lymph node, colon, breast, prostate, ovarian, esophageal, head, neck, rectal, testis, throat, thyroid, intestinal, melanocytic, colorectal, liver, gastric, and bladder tissues. Cells may be obtained, e.g., from cell culture or breakdown of tissues.

[00015] In some embodiments, the biological sample is derived from a cell line, optionally, treated with an agent whose effect on gene expression is evaluated. In other embodiments, the sample is a tissue or a biological fluid of a subject (e.g., a mammal, (e.g., a rodent or a primate, e.g., human)).

[00016] In some embodiments, the biological sample is divided into replicates (e.g., duplicates, triplicates, etc.) in which the expression levels are measured. The sample may be derived from the same source and split into replicates just prior to measuring the expression levels. Replicate samples may be analyzed in a serial or parallel manner. Gene expression levels for the same gene may be measured in replicates, and the final gene expression level expressed as an average or a mean of the replicates, or an otherwise calculated level representing multiple samples. In some embodiments, expression levels of two or more genes are measured in separate replicates individually. Alternatively, or in addition, the expression levels of at least some genes may be measured in the same reaction volume, e.g., multiplexing.

[00056] In some embodiments, a plurality of genes being measured comprises at least one biomarker of a disease, including a disease type or subtype. As used herein, the term "disease" includes a pathologic or otherwise abnormal condition identifiable by altered gene expression levels. As used herein, a biomarker is a gene whose expression correlates with the presence of a specified disease or condition. Such a disease or condition may be due to a pathogen, e.g., virus, fungus, bacteria, or a toxin. A disease or condition may be of any type, e.g., malignancy, immunological disorder, cardiovascular, or neurological. For example, cancers being evaluated may include, for example, cancers of colon, breast, prostate, skin, bladder, or lung as well as lymphoma, leukemia, etc. Numerous biomarkers for various diseases and conditions are known (see, e.g., Biomarkers in Breast Cancer (Cancer Drug Discovery and Development), Humana Press; 1 edition, 2005); Biomarkers of Disease: An Evidence-Based Approach; Cambridge University Press; 1 edition, 2002). In illustrative embodiments, the cancer markers used are DNA Repair and DNA Damage Response Markers. Biomarkers associated with disease states include those biomarkers listed on Tables 3-5.

[00057] Thus, in some embodiments, methods of the invention are used for differentiation between disease types or subtypes by evaluating two or more biomarkers specific to one or more disease types or subtypes. For example, the methods may include evaluation of 2, 3, 4, 5, 10, 25, 50, 100 or more biomarkers of disease types or subtypes.

[00058] Image Analysis, Approaches and Systems

[00059] While image analysis of molecular labels can include a number of applications, the following section will be limited to the discussion of the problem of estimating abundance of stains in histological tissue, with an emphasis on IHC as opposed to immunofluorescence. The assumption is made that the signal on the slide is representative and in some way quantitatively related to the abundance of the biomarker in the tissue section, which in turn is related, albeit in ways unknown, to the absolute amount of the analyte in the original tissue. Factors that affect performance of the imaging system include the choice of camera and illumination source, the optical performance of the stains themselves, as well as the presence and degree of multiplexing. After image acquisition, it is then necessary to deploy appropriate mathematical techniques to extract quantitative intensity and area measurements from the imaging data.

[00060] Imaging Hardware: RGB Vs. Multispectral Approaches

[00061] There is a long history of the application of image processing to pathology samples. While some early automated imaging systems employed grayscale cameras and filter wheels to collect images, most current brightfield (transmitted light) pathology imaging systems rely on standard color cameras similar in many respects to consumer digital cameras. These typically employ a Bayer-pattern color mask over a CCD or CMOS detector, and use various algorithms to process the raw image data to generate color images that can be presented to the pathologist, and that are also used in downstream automated analysis. Single-chip, Bayer-pattern red-green- blue (RGB) cameras that are often employed, especially in many "home-grown" systems, can generate imaging artifacts, especially with respect to fine structures or edges, and have poorer spatial fidelity than more expensive 3-chip systems in which separate pixel-registered cameras are used to acquire simultaneously red, green and blue images. While the simple acquisition of good-looking color images is appealing, RGB detectors can introduce significant problems when one is trying to achieve quantification and inter-instrument precision. There are a number of ways that variation arises. For example, color values can vary significantly with the color temperature of the illumination source, different color-correction routines in camera firmware can play a role in the exact color values that are reported out, and different camera chips have differing spectral responsiveness. Some cameras employ automatic gain control or related circuitry designed to "optimize" image quality, with unpredictable effects on resulting images.

[00062] Even if an RGB imaging system is working perfectly, there are intrinsic limitations to its ability to distinguish between similar chromogens, and even more challengingly, to be able to "unmix" such signals if they overlap spatially. "Unmix" in this sense means to isolate the optical signal from each chromogen so that each can be measured quantitatively, and separately. Signal processing theory suggests that at least n if not n+1 measurements are needed to unmix n signals. In theory, therefore, it is impossible to unmix more than 3 chromogens with an RGB sensor. In practice, while it is possible to do a good job unmixing DAB (brown) from hematoxylin (blue), it has proven extremely difficult to unmix brown from red from blue (a typical combination of stains for a double-labeled sample), using only RGB measurements, due to the color-overlap of the spectral profiles. To accomplish such tasks properly, true multispectral imaging approaches may be necessary.

[00063] Spectral Imaging

[00064] Spectral imaging microscopy represents a technological advance over visual or RGB- camera-based analyses. By acquiring a stack of images at multiple wavelengths, spectral imaging allows the determination of precise optical spectra at every pixel location. With this spatially resolved spectral information in hand, it is possible to enhance the utility of IHC and ISH stains, and even the standard biologic stains used in surgical pathology. There are a number of ways to perform spectral imaging, reviewed in (24,35). The focus in this review is on the commercially available liquid crystal tunable filter-based system (Nuance™, CRi, Woburn, Mass.), from which all examples here will be drawn; this is not to imply that the Nuance system is the best or only approach, merely that it is the model with which the authors have had most experience. This system is suitable for both brightfield and fluorescence imaging. Under automatic control, a series of images (from 3 to as many as 20 or more) are taken from blue to the red (e.g., 420 nm to 700 nm) and the resulting image "stack" or "cube" is assembled in memory in such a way that a spectrum is associated with every pixel. The ability to sample the spectrum with many discrete wavelength regions spanning the visible wavelength range allows for accurate unmixing of multiple spatially co-localized chromogens, even if they are similar in color and have largely overlapping absorption spectra. Thus, it becomes straightforward to separate dark reds from light browns, or even varieties of blue stains (hematoxylin vs. NBT-BCIP) (36,37).

[00065] Image Processing and Unmixing

[00066] The key process, either with RGB images or multispectral datasets, is to partition the overall signal in a given pixel correctly into its component species. Linear unmixing algorithms rely on the signals adding together linearly. This is true with fluorescent dyes (which emit light), but this is not the case with chromogens imaged in brightfield, since they absorb light.

Fortunately, the Lambert-Beer (or simply Beer's) law relating concentrations to absorbance indicates that when the transmission data is converted to optical density (absorbance) units, linearity is restored, and quantification and unmixing (39) can be successfully achieved. There are many benefits attendant on the conversion to optical density (OD), which is typically performed by taking the negative (base 10) log of the transmitted image divided by the illumination (usually a clear area on the microscope slide). First, absorbance values are an intrinsic property of the sample, and do not depend on vagaries of illumination or camera responsivities. This means that absorbance measurements of a given specimen performed on any appropriate system should, in theory, be comparable. Secondly, in the process of creating an absorbance image, flat-fielding is automatically performed, which removes the effects of uneven illumination and minor flaws in the optical train. Conversion to OD can be performed on monochrome, RGB or multispectral images.

[00067] OD (absorbance) units are dimensionless and logarithmic: so that zero absorbance means all photons transmitted; an OD of 1.0 absorbs 90% of all photons, and an OD of 2.0 absorbs 99% of all potentially detected photons. IHC stains can individually generate signals of 1 OD. Accordingly, having 2 or more dense and overlapping stains can result in virtually black deposits from which little or no useful spectral or quantitative data can be recovered. This, plus the lesser dynamic range achievable with IHC vs. fluorescence -based approaches may mean that immunofluorescence may be preferable or necessary for some applications. Nevertheless, IHC has some practical advantages over immunofluorescence, including the fact that pathologists prefer it largely because it allows integration of 'phenotypic' features in the IHC stain with the traditional morphologic features, long the 'gold standard' for diagnosis.

[00068] An important caveat is that the optical properties of the chromogens will affect the linearity and dynamic range of the assay. The Lambert-Beer law that underlies the unmixing approach applies only to pure absorbers. Some chromogens, most notably the popular brown DAB stain, exhibit scattering behavior similar to that of melanosomes. In fact, it can be impossible to separate DAB from melanin pigmentation spectrally, since their spectra arise from the same optical properties. However, in practice, this does not seem to pose insuperable problems, since linearity and reasonable dynamic range can be achieved using DAB approaches (41). Other chromogens, such as Vector Red, have been shown to have excellent linearity and dynamic range (42).

[00069] In addition to the specific molecular labeling procedure, a counterstain is almost always applied. Thus the challenge for quantitation begins with the unmixing of the chromogen (typically DAB) from the counterstain (typically hematoxylin). The latter pair can be

successfully unmixed using simple RGB imagery if conversion to OD is performed (39), but other pairs may not be so amenable. One of the challenges (see below) is the accurate

determination of the spectra of the chromogens as input values into the unmixing procedure. Small variations in the spectra chosen can have quite dramatic effects on the calculated abundance values. While in many cases it suffices to measure the spectrum of the isolated chromogens (single stain, no counterstain), we have found that it may be necessary to measure the spectrum of the chromogens in the actual sample, after all the staining procedures have been performed, since the spectra can be affected by the presence of other dyes and reagents.

[00070] Multiplexing

[00071] Typically, only a single IHC-chromogen-biomarker combination is used per slide; if more than one biomarker is to be analyzed, serial sections are made and a different antibody is applied to each. This procedure benefits from simplified protocols and quality control regimens compared to multicolor techniques, but generates more slides and possibly more preparation steps than if the reagents are 'multiplexed' on a single slide. Moreover, multiple molecular events cannot be evaluated on a per-cell basis when parallel sections are employed, and this capability is very important in establishing the phenotype of individual tumor cells (e.g., lymphoma cells) distributed in a mixed cell population. Multicolor immunohistochemistry is thus an important goal, but is challenging to achieve. The prerequisite to quantitative accuracy in a multiple labeled section is lack of interference between the labels. Not only can one label physically block the successful labeling of the next antigen due to steric hindrance, but the various labeling procedures can be chemically incompatible. Suffice it to say that the

performance of multiple labelings on a single specimen increases the demands for appropriate controls. Assuming that the labeling procedures have been performed satisfactorily, unmixing of 3 or more chromogens is entirely feasible. In addition, multiple chromogenic in situ

hybridization signals can be combined with IHC.

[00072] Performance and Accuracy Measures of the Invention

[00073] The performance and thus absolute and relative clinical usefulness of the invention may be assessed in multiple ways as noted above. Amongst the various assessments of performance, the invention is intended to provide accuracy in clinical diagnosis and prognosis and predicting response to treatment. . The accuracy of a diagnostic or prognostic test, assay, or method concerns the ability of the test, assay, or method to distinguish between subjects responsive to chemotherapeutic treatment and those that are not, is based on whether the subjects have an "effective amount" or a "significant alteration" in the levels of a BIOMARKER. By "effective amount" or "significant alteration," it is meant that the measurement of an appropriate number of BIOMARKERS (which may be one or more) is different than the predetermined cutoff point (or threshold value) for that BIOMARKER(S) and therefore indicates that the subject responsiveness to therapy for which the BIOMARKER(S) is a determinant. The difference in the level of BIOMARKER between normal and abnormal is preferably statistically significant. As noted below, and without any limitation of the invention, achieving statistical significance, and thus the preferred analytical and clinical accuracy, generally but not always requires that combinations of several BIOMARKERS be used together in panels and combined with mathematical algorithms in order to achieve a statistically significant BIOMARKER index. [00074] In the categorical diagnosis of a disease state, changing the cut point or threshold value of a test (or assay) usually changes the sensitivity and specificity, but in a qualitatively inverse relationship. Therefore, in assessing the accuracy and usefulness of a proposed medical test, assay, or method for assessing a subject's condition, one should always take both sensitivity and specificity into account and be mindful of what the cut point is at which the sensitivity and specificity are being reported because sensitivity and specificity may vary significantly over the range of cut points. Use of statistics such as AUC, encompassing all potential cut point values, is preferred for most categorical risk measures using the invention, while for continuous risk measures, statistics of goodness-of-fit and calibration to observed results or other gold standards, are preferred.

[00075] Using such statistics, an "acceptable degree of diagnostic accuracy", is herein defined as a test or assay (such as the test of the invention for determining the clinically significant presence of BIOMARKERS, which thereby indicates the presence of cancer and/or a risk of having a metastatic event) in which the AUC (area under the ROC curve for the test or assay) is at least 0.60, desirably at least 0.65, more desirably at least 0.70, preferably at least 0.75, more preferably at least 0.80, and most preferably at least 0.85.

[00076] By a "very high degree of diagnostic accuracy", it is meant a test or assay in which the AUC (area under the ROC curve for the test or assay) is at least 0.80, desirably at least 0.85, more desirably at least 0.875, preferably at least 0.90, more preferably at least 0.925, and most preferably at least 0.95.

[00077] The predictive value of any test depends on the sensitivity and specificity of the test, and on the prevalence of the condition in the population being tested. This notion, based on Bayes' theorem, provides that the greater the likelihood that the condition being screened for is present in an individual or in the population (pre-test probability), the greater the validity of a positive test and the greater the likelihood that the result is a true positive. Thus, the problem with using a test in any population where there is a low likelihood of the condition being present is that a positive result has limited value (i.e., more likely to be a false positive). Similarly, in populations at very high risk, a negative test result is more likely to be a false negative.

[00078] As a result, ROC and AUC can be misleading as to the clinical utility of a test in low disease prevalence tested populations (defined as those with less than 1% rate of occurrences (incidence) per annum, or less than 10% cumulative prevalence over a specified time horizon). Alternatively, absolute risk and relative risk ratios as defined elsewhere in this disclosure can be employed to determine the degree of clinical utility. Populations of subjects to be tested can also be categorized into quartiles by the test's measurement values, where the top quartile (25% of the population) comprises the group of subjects with the highest relative risk for therapeutic unresponsiveness, and the bottom quartile comprising the group of subjects having the lowest relative risk for therapeutic unresponsiveness Generally, values derived from tests or assays having over 2.5 times the relative risk from top to bottom quartile in a low prevalence population are considered to have a "high degree of diagnostic accuracy," and those with five to seven times the relative risk for each quartile are considered to have a "very high degree of diagnostic accuracy." Nonetheless, values derived from tests or assays having only 1.2 to 2.5 times the relative risk for each quartile remain clinically useful are widely used as risk factors for a disease; such is the case with total cholesterol and for many inflammatory biomarkers with respect to their prediction of future events. Often such lower diagnostic accuracy tests must be combined with additional parameters in order to derive meaningful clinical thresholds for therapeutic intervention, as is done with the aforementioned global risk assessment indices.

[00079] A health economic utility function is an yet another means of measuring the performance and clinical value of a given test, consisting of weighting the potential categorical test outcomes based on actual measures of clinical and economic value for each. Health economic performance is closely related to accuracy, as a health economic utility function specifically assigns an economic value for the benefits of correct classification and the costs of misclassification of tested subjects. As a performance measure, it is not unusual to require a test to achieve a level of performance which results in an increase in health economic value per test (prior to testing costs) in excess of the target price of the test.

[00080] In general, alternative methods of determining diagnostic accuracy are commonly used for continuous measures, when a disease category or risk category has not yet been clearly defined by the relevant medical societies and practice of medicine, where thresholds for therapeutic use are not yet established, or where there is no existing gold standard for diagnosis of the pre-disease. For continuous measures of risk, measures of diagnostic accuracy for a calculated index are typically based on curve fit and calibration between the predicted continuous value and the actual observed values (or a historical index calculated value) and utilize measures such as R squared, Hosmer- Lemeshow P-value statistics and confidence intervals. It is not unusual for predicted values using such algorithms to be reported including a confidence interval (usually 90% or 95% CI) based on a historical observed cohort's predictions.

[00081] In general, by defining the degree of diagnostic accuracy, i.e., cut points on a ROC curve, defining an acceptable AUC value, and determining the acceptable ranges in relative concentration of what constitutes an effective amount of the BIOMARKERS of the invention allows for one of skill in the art to use the BIOMARKERS to identify, diagnose, or prognose subjects with a pre-determined level of predictability and performance.

[00082] Construction of Biomarker Algorithms

[00083] Any formula may be used to combine BIOMARKER results into indices useful in the practice of the invention. As indicated above, and without limitation, such indices may indicate, among the various other indications, the probability, likelihood, absolute or relative chance of responding to chemotherapy. This may be for a specific time period or horizon, or for remaining lifetime risk, or simply be provided as an index relative to another reference subject population.

[00084] Although various preferred formula are described here, several other model and formula types beyond those mentioned herein and in the definitions above are well known to one skilled in the art. The actual model type or formula used may itself be selected from the field of potential models based on the performance and diagnostic accuracy characteristics of its results in a training population. The specifics of the formula itself may commonly be derived from BIOMARKER results in the relevant training population. Amongst other uses, such formula may be intended to map the feature space derived from one or more BIOMARKER inputs to a set of subject classes (e.g. useful in predicting class membership of subjects as normal, responders and non-responders), to derive an estimation of a probability function of risk using a Bayesian approach (e.g. the risk of cancer or a metastatic event), or to estimate the class- conditional probabilities, then use Bayes' rule to produce the class probability function as in the previous case.

[00085] Preferred formulas include the broad class of statistical classification algorithms, and in particular the use of discriminant analysis. The goal of discriminant analysis is to predict class membership from a previously identified set of features. In the case of linear discriminant analysis (LDA), the linear combination of features is identified that maximizes the separation among groups by some criteria. Features can be identified for LDA using an eigengene based approach with different thresholds (ELD A) or a stepping algorithm based on a multivariate analysis of variance (MANOVA). Forward, backward, and stepwise algorithms can be performed that minimize the probability of no separation based on the Hotelling-Lawley statistic.

[00086] Eigengene-based Linear Discriminant Analysis (ELD A) is a feature selection technique developed by Shen et al. (2006). The formula selects features (e.g. biomarkers) in a multivariate framework using a modified eigen analysis to identify features associated with the most important eigenvectors. "Important" is defined as those eigenvectors that explain the most variance in the differences among samples that are trying to be classified relative to some threshold.

[00087] A support vector machine (SVM) is a classification formula that attempts to find a hyperplane that separates two classes. This hyperplane contains support vectors, data points that are exactly the margin distance away from the hyperplane. In the likely event that no separating hyperplane exists in the current dimensions of the data, the dimensionality is expanded greatly by projecting the data into larger dimensions by taking non-linear functions of the original variables (Venables and Ripley, 2002). Although not required, filtering of features for SVM often improves prediction. Features (e.g., biomarkers) can be identified for a support vector machine using a non-parametric Kruskal-Wallis (KW) test to select the best univariate features. A random forest (RF, Breiman, 2001) or recursive partitioning (RPART, Breiman et al., 1984) can also be used separately or in combination to identify biomarker combinations that are most important. Both KW and RF require that a number of features be selected from the total.

RPART creates a single classification tree using a subset of available biomarkers.

[00088] Other formula may be used in order to pre-process the results of individual

BIOMARKER measurement into more valuable forms of information, prior to their presentation to the predictive formula. Most notably, normalization of biomarker results, using either common mathematical transformations such as logarithmic or logistic functions, as normal or other distribution positions, in reference to a population's mean values, etc. are all well known to those skilled in the art. Of particular interest are a set of normalizations based on Clinical Parameters such as age, gender, race, or sex, where specific formula are used solely on subjects within a class or continuously combining a Clinical Parameter as an input. In other cases, analyte-based biomarkers can be combined into calculated variables which are subsequently presented to a formula. [00089] In addition to the individual parameter values of one subject potentially being normalized, an overall predictive formula for all subjects, or any known class of subjects, may itself be recalibrated or otherwise adjusted based on adjustment for a population's expected prevalence and mean biomarker parameter values, according to the technique outlined in D'Agostino et al, (2001) JAMA 286: 180-187, or other similar normalization and recalibration techniques. Such epidemiological adjustment statistics may be captured, confirmed,

improved and updated continuously through a registry of past data presented to the model, which may be machine readable or otherwise, or occasionally through the retrospective query of stored samples or reference to historical studies of such parameters and statistics. Additional examples that may be the subject of formula recalibration or other adjustments include statistics used in studies by Pepe, M.S. et al, 2004 on the limitations of odds ratios; Cook, N.R., 2007 relating to ROC curves. Finally, the numeric result of a classifier formula itself may be transformed postprocessing by its reference to an actual clinical population and study results and observed endpoints, in order to calibrate to absolute risk and provide confidence intervals for varying numeric results of the classifier or risk formula. An example of this is the presentation of absolute risk, and confidence intervals for that risk, derived using an actual clinical study, chosen with reference to the output of the recurrence score formula in the Oncotype Dx product of Genomic Health, Inc. (Redwood City, CA). A further modification is to adjust for smaller sub- populations of the study based on the output of the classifier or risk formula and defined and selected by their Clinical Parameters, such as age or sex.

[00090] Example 1

[00091] We have conducted biomarker studies using the International Adjuvant Lung Therapy trial (IALT) comprising a large randomized study of NSCLC patient-derived specimens. The treatment groups were cisplatin versus observation. We have measured the expression of DNA repair enzyme levels by IHC staining of formalin-fixed paraffin-embedded sections and then rank ordered patients following digital imaging and application of user-defined macros. These algorithms are designed to yield a semi-quantitative assessment of the % of tumor cells displaying positive staining weighted by the intensity of cells staining as positive at 10 levels (Q- score range = 0 to 1000).

Q-score = 10*( of cells scoring as 10+) + 9*(% of cells scoring as 9+)... l*( of cells scoring as 1+) [00092] These DNA repair biomarker levels have been assessed as tools for predictive response to platin-based therapy relative to patient outcomes (5yr disease-free survival [DFS] and 5yr overall survival [OS]).

[00093] As the tumor specimens were collected from multiple centers and were undoubtedly prone to variations in pre- analytical processing, we sought to address what effects normalization of the semi-quantitative assessment of the specimens to a candidate calibrator might manifest upon further analysis of the biomarker expression data relative to patient outcomes. Figure 9 indicates a series of R plots to depict relative patient order rank change within the cohort as a whole for specific biomarkers. In these figures the rank ordering of patients from non-calibrated patient Q-score data is plotted against patient rank order ratiometrics generated from a Q-score of the test biomarker of interest normalized to a candidate calibrator of interest, here SnRNP70. R values indicate an estimate on the amount of rank ordering rearrangement of patients occurs by normalized patient data relative to raw Q-scores. For several biomarkers stained and assessed, the R values for normalized and non-normalized patient data are listed below. As we have already established that SnRNP70 is expressed consistently across patient specimens in our research and vetting of this candidate calibrator, we established that a Q-score <200 (0-1000 scale), indicates a sample that is likely to have been compromised and should be removed from further analyses. In the case of this study, this effectively led to the elimination of 11 patients (from 551 total [1.98%] with assessable SnRNP70).

[00094] The average R across the 7 biomarkers tested is calculated at 90.1. This indicates that the application of the calibrator has resulted in a slight shift in the patient biomarker data, without dramatically altering it. It would be our hope then that the statistical significance from analysis of non-calibrated sets would be retained and possibly improved upon slightly by analysis of the attenuated biomarker data.

[00095] Table 6 below shows representative data from 2 biomarkers, PARP1 and MSH2, and how the statistical significance is affected by calibration of the sample set compared to statistical analysis of non-calibrated biomarker data with regards to patient outcomes. Note that the analysis cohort for each biomarker is restricted to patients for which there are Q-scores generated for both test as well as calibrator marker. Therefore, the same patients and biomarker scores are being assessed on a pre-normalized and post-normalized basis. Cox proportional hazard modelings were conducted on patient data derived from binary cut-offs of Q-scores or normalized ratiometric Q-scores based on median values. Analyses were carried out without the inclusion of clinical variable and then by inclusion of significant variable and finally by inclusion of all adjustment variable. Data was analyzed with all NSCLC grouped as a single cohort and then by separation of all NSCLC into constituent histological subgroupings of squamous cell carcinoma (SCC) and adenocarcinoma.

[00096] It is apparent that for statistical analyses in which the entire population (N=533) was analyzed and found to display significance (p<0.05) or borderline significance (0.05<p<0.1), that when analyzed in histological subgroups, all significance correlates and co-segregates with SCC to the exclusion of adenocarcinoma.

[00097] Moreover, no differences occur in adenocarcinoma lack of significance for statistics between non-normalized and normalized biomarker assessments. However, for SCC and total population analyses, the significant p-values are retained and in many instances in which no significance or borderline significance was achieved in the non-normalized group, the p-values are brought into significance upon normalization. That normalization has no affect on biomarkers that are non- significant and yet seemingly improve the data for histological subgroups and biomarkers that are significant is indicative of the potential use of this candidate calibrator in IHC of FFPE-derived patient tumor specimens.

OTHER EMBODIMENTS

[00099] While the invention has been described in conjunction with the detailed description thereof, the foregoing description is intended to illustrate and not limit the scope of the invention, which is defined by the scope of the appended claims. Other aspects, advantages, and

modifications are within the scope of the following claims.

References

Wang, D. et al. (2009) APEl overexpression is associated with cisplatin resistance in non-small cell lung cancer and targeted inhibition of APEl enhances the activity of cisplatin in A549 cells. Lung Cancer, in press

(Elevated APEl correlates with Cisplatin resistance) Kang et al., 2009 The prognostic significance of ERCCl, BRCA1, XRCC1, and betalll-tubulin expression in patients with non-small cell lung cancer treated by platinum- and taxane-based neoadjuvant chemotherapy and surgical resection; Lung Cancer 2009 Aug.14

(Inverse correlation for DNA repair markers with cell viability indicating a value for chemotherapy responsiveness) Scartozzi, M. et al. (2006) Mismatch repair system (MMR) status correlates with response and survival in non-small cell lung cancer (NSCLC) patients. Lung Cancer 53, 103-109.

(Low hMLHl and hMSH2 correlates with cisplatin + gemcitibine resistance)

Olaussen, K.A .et al. (2006) DNA Repair by ERCCl in Non-Small-Cell Lung Cancer and Cisplatin- Based Adjuvant Chemotherapy. NEJM 355, 983-991.

(Low ERCCl patients benefit from Cisplatin + etoposide or vinca alkaloid sensitivity) (High ERCCl patients benefit from NOT receiving CDDP + etoposide or vinca alkaloid)

Zheng, Z. et al. (2007) DNA Synthesis and Repair Genes RRMl and ERCCl in Lung Cancer. NEJM 356, 800-808.

(High ERCCl and RRMl as prognostic biomarkers in surgical specimens)

Takenaka, T. et al. (2007) Combined evaluation of Rad51 and ERCCl expressions for sensitivity to platinum agents in non-small cell lung cancer. Int. J. Cancer: 121, 895-900. (Combined high expressions or ERCCl and RAD51 correlate with cisplatin resistance)

Lee, K. -H., et al. (2008) ERCCl expression by immunohistochemistry and EGFR mutations in resected non-small cell lung cancer. Lung Cancer 60, 401—407.

(High ERCCl patients have better prognosis following surgery alone than low ERCCl)

Gazdar A.F. (2007) DNA Repair and Survival in Lung Cancer—The Two Faces of Janus. NEJM 356, 771-773.

(High levels of RRM1 and ERCCl lead to cisplatin resistance) Wang, et al., 2009; Med Oncol. 2009 June2. Positive expression of ERCCl predicts a poorer platinum-based treatment outcome in Chinese patients with advanced non-small-cell lung cancer

(Expression of ERCCl, RRM1 was negatively associated with tumor response)

Papay J et al., Pathol Oncol Res. 2009 Mar 2. [Epub ahead of print] Platinum-Based

Chemotherapy in Lung Cancer Affects the Expression of Certain Biomarkers Including ERCCl.

(From bronchoscopic biopsies with surgical resections after neoadjuvant chemotherapy: Platinum chemotherapy may select for tumor cells with aggressive phenotype, and affect expression of ERCCl that could have predictive value)

Takenaka, T. et al. (2007) Combined evaluation of Rad51 and ERCCl expressions for sensitivity to platinum agents in non-small cell lung cancer. Int. J. Cancer: 121, 895-900.

(Combined high expressions or ERCCl and RAD51 correlate with cisplatin resistance)

Taron, M. et al. (2004) BRCAl mRNA expression levels as an indicator of chemoresistance in lun cancer. Human Molecular Genetics, 13, 2443-2449.

(High BRCAl leads to cisplastin + gemcitibine resistance)

(Low BRCAl more sensitive to cisplatin + gemcitibine) Kim, H.T. et al. (2008) Effect of BRCAl haplotype on survival of non-small-cell lung cancer patients treated with platinum-based chemotherapy. J Clin Oncol. 26, 5972-5979.

(BRCAl haplotype predicts the level of cisplatin sensitivitiy) Ko, J. -C. et al. (2008) Role of repair protein Rad51 in regulating the response to gefitinib in human non-small cell lung cancer cells. Molecular Cancer Therapeutics 7, 3632-3641.

(High Rad51 leads to gefitinib resistance)

Ceppi, P. et al. (2009) Polymerase H mRNA Expression Predicts Survival of Non-Small Cell Lung Cancer Patients Treated with Platinum-Based Chemotherapy. Clinical Cancer Research 15; 1039- 1045. (Low PolH predicts greater sensivity to cisplatin than high PolH) Samimi, G. et al (2000) Analysis of MLH1 and MSH2 expression in ovarian cancer before and after platinum drug-based chemotherapy. Clinical Cancer Research 6, 1415-1421.

(hMLHl and hMSH2 expression levels decrease following exposure to cisplatin) Scartozzi, M. et al (2003) Loss of hMLHl expression correlates with improved survival in stage ll l-IV ovarian cancer patients. Eur J Cancer. 39, 1144-1149.

(MLH1 and MSH2 expression loss after cisplatin treatment correlates with improved survival) Materna, V. et al (2007) Expression of factors involved in regulation of DNA mismatch repair- and apoptosis pathways in ovarian cancer patients. Oncol Rep. 17, 505-516.

(MSH2 levels correlate with prognosis following chemotherapy treatment) Smith, S. et al. (2007) ERCC1 genotype and phenotype in epithelial ovarian cancer identify patients likely to benefit from paclitaxel treatment in addition to platinum-based therapy. J Clin Oncol. 25, 5172-5179.

(Patients with high ERCC1 may benefit from paclitaxel +cisplatin)

(Patients with low ERCC1 are likely to benefit from cisplatin alone) Lin, K. et al. (2008) Protein expression levels of excision repair cross-complementation group 1 and xeroderma pigmentosum D correlate with response to platinum-based chemotherapy in the patients with advanced epithelial ovarian cancer. Int J Gynecol Cancer.18, 1007-1012.

(ERCC1 and XPD expression associated with resistance to platinum-based chemotherapy) Stevens, E.V. et al. (2008) Predicting cisplatin and trabectedin drug sensitivity in ovarian and colon cancers. Mol Cancer Ther. 7, 10-18.

(XPF and XPG expression correlate with cisplatin resistance) Steffensen, K.D. et al, (2009). The relationship of platinum resistance and ERCC1 protein expression in epithelial ovarian cancer. Int J Gynecol Cancer. 19, 820-825.

(ERCC1 expression correlated with carboplatin + paclitaxel resistance) Swisher, E.M. et al. (2009) Methylation and protein expression of DNA repair genes: association with chemotherapy exposure and survival in sporadic ovarian and peritoneal carcinomas. Molecular Cancer 8, 1-11.

(Recurrent ovarian cancer displays BRCA1 elevation and cisplatin resistance) Husain A., et al.(1998) BRCA1 up-regulation is associated with repair-mediated resistance to cisdiamminedichloroplatinum (II). Cancer Res 58, 1120-1123.

(BRCA1 overexpressed in cisplatin-resistant cells) Quinn J.E., et al. (2003) BRCAl functions as a differential modulator of chemotherapy-induced apoptosis. Cancer Res 63, 6221-6228.

(BRCAl - resistance to etoposide and bleomycin, sensitizes to paclitaxel and vinorelbine) Zhou C. et al. (2003): Role of BRCAl in cellular resistance to paclitaxel and ionizing radiation in an ovarian cancer cell line carrying a defective BRCAl. Oncogene 22,2396-2404.

(Mutant BRCAl cells more sensitive to paclitaxel and radiation)

Taniguchi, T., et al. (2003): Disruption of the Fanconi anemia-BRCA pathway in cisplatin-sensitive ovarian tumors. Nat Med 2003, 9:568-574.

(Cisplatin resistance occurs via a functioning Fanc/BRCA pathway)

Shi, S.-R. et al. (2007): Standardization of immunohistochemistry for formalin-fixed, paraffin- embedded tissue sections based on the antigen-retrieval technique: from experiments to hypothesis. Journal of Histochemistry and Cytochemistry 55, 105-109.

DeLellis et al. (1977): Immunoperoxidase technics in diagnostic pathology. Report of a workshop sponsored by the National Cancer Institute. Am J Clin Pathol 71:483-488. Popovici, V. et al. (2009); Selecting control genes for RT-QPCR using public microarray data. BMC Bioinformatics 2009, 10:42.31. Kwon, M.J., et al. (2009): Identification of novel reference genes using multiplatform expression data and their validation for quantitative gene expression analysis. PLoS One. 4.

Claims

We claim: 1. A method of determining the change in a value of a biomarker attributed to a biological condition of a subject comprising:

a. providing a test value for a biomarker from a sample from the subject, wherein the biomarker is indicative of the biological condition of the subject;

b. providing a test value for a calibrator marker, wherein the calibrator marker

expression is specific for one or more cell or tissue types and is constant across a plurality of samples; and

c. adjusting the test value for the biomarker based upon the test value of the

calibrator marker to arrive at a change in value of the biomarker that is attributable to the biological condition of the subject.

2. The method of claim 1, wherein the test value is obtained by immunofluorescence, colorimetric staining, biosensors, or Surface Plasmon Resonance.

3. A method of normalizing biomarker expression in a biological sample comprising:

a. determining the expression level of the calibrator marker in the sample, wherein the calibrator marker expression is specific for one or more cell or tissue types and is constant across a plurality of samples; and

b. normalizing the expression level of one or more biomarkers in the sample using the expression level determined in step (a).

4. The method of claim 3, wherein the expression level is determined by

immunofluorescence, colorimetric staining, biosensors, or Surface Plasmon Resonance. 5. A method of quantifying a test biomarker in a sample:

a. determining the amount of a calibrator marker wherein the calibrator marker expression is specific for one or more cell or tissue types and is constant across a plurality of samples; and

b. calculating the amount of the test biomarker in the sample from the amount of the calibrator marker in the sample.

6. The method of any one of claim 1-5, wherein the sample is a formalin-fixed paraffin embedded tissue sample, a frozen tissue sample, a blood sample, a circulating tumor cell. The method of any one of claim 1-6, wherein the calibrator marker is selected from the group consisting of HNPRM, SNRNP70, PABPN1, SNRPA, and ZNF207.

The method of any one of claim 1-6, wherein the calibrator marker is selected form Table

2.

The method of any one of claim 1-6, wherein the biomarker is selected form Tables 1, 3, 4, or 5.