WO2015006262A1

WO2015006262A1 - Tumor tissue analysis techniques

Info

Publication number: WO2015006262A1
Application number: PCT/US2014/045649
Authority: WO
Inventors: Cristiano Ferlini; Marisa MARIANI; Paul Fiedler; Shohreh SHAHABI
Original assignee: Western Connecticut Health Network, Inc.
Priority date: 2013-07-10
Filing date: 2014-07-08
Publication date: 2015-01-15

Abstract

In some aspects the present invention provides, compositions and methods for the analysis tumor tissue samples. In some aspects, samples are analyzed for the correlation of two individual parameters (e.g., expressed markers) in order to accurately assess tumor prognosis and/or therapeutic response. Automated methods for performing the techniques of the embodiments are likewise provided.

Description

DESCRIPTION

TUMOR TISSUE ANALYSIS TECHNIQUES

[0001] This application claims the benefit of United States Provisional Patent Application No 61/844,734, filed July, 10, 2013, the entirety of which is incorporated herein by reference.

BACKGROUND OF THE INVENTION

1. Field of the Invention

[0002] The present invention relates generally to the fields of medicine and cell biology. More particularly, it concerns a method for interpreting the level of expression of a target marker in a tissue or cell sample using, for example, image analysis and quantitative fluorescent immunohistochemistry.

2. Description of Related Art

[0003] The expression of a given antigen of interest is commonly interpreted as the average of the expression of the antigen across a population of cells. While this method works well in homogeneous samples, namely if cells exhibit similar expression of the antigen, the same method applied to a heterogeneous population is biased by the fact that the expression is averaged from a wide expression range that varies from no to high expression of the antigen of interest. As a consequence, the discriminator cutoff (the percentage value that is the threshold to determine whether a given sample is positive or negative) is inside the distribution range thus potentially biasing the patient selection method with false negative and false positive patients. Since then the expression is translated into a clinical decision in most of the cases with a binary choice (i.e., positive expression will lead to intention to treat) such bias may be responsible for clinical inefficacy of the used drug (false positive) or for not treating a potentially treatable patient (false negative). SUMMARY OF THE INVENTION

[0004] Embodiments of the present invention provide a method for analysis of a tissue sample using a correlation matrix among two or more markers. As demonstrated herein, such analysis techniques provide far more accurate characterization of tissue samples than analyses that rely on averaging the percentage of cells exhibiting a particular marker. [0005] Thus, in a first embodiment, the present invention provides a method of characterizing a tissue sample comprising a) obtaining a tissue sample from a subject; b) quantifying the expression of two or more markers in a plurality of cells in the sample; c) determining a correlation coefficient between the expression the two or more markers in the plurality of cells and calculating a confidence interval over said plurality of cells in the sample; and d) comparing the calculated confidence interval to a reference value to characterize the sample. For example, in some aspects, step c), determining a correlation coefficient, comprises creating a new synthetic variable obtained with a correlation coefficient between the expression the two or more markers in the plurality of cells and calculating a confidence interval over said plurality of cells in the sample. In preferred aspects, the tissue is a tissue that is, or is suspected of being, diseased (e.g., a sample suspected of being associated with a cancer, an infection or an autoimmune disorder). For example, the tissue sample may be a cancer tissue sample, such as a tumor biopsy sample or a section from a resected tumor. In some aspects, characterizing a cancer according to the embodiments may comprise determining the aggressiveness or drug susceptibility of the cancer.

[0006] In a further embodiment, the present invention provides a method of selecting a candidate anti-cancer therapy for a subject comprising: a) obtaining a cancer tissue sample from a subject; b) quantifying the expression of two or more markers in a plurality of cells in the sample; c) determining a correlation coefficient between the expression the two or more markers in the plurality of cells and calculating a confidence interval over said plurality of cells in the sample; and d) comparing the calculated confidence interval to a reference value to select a candidate anti-cancer therapy for the patient. For example, in some aspects, step c), determining a correlation coefficient, comprises creating a new synthetic variable obtained with a correlation coefficient between the expression the two or more markers in the plurality of cells and calculating a confidence interval over said plurality of cells in the sample. In one aspect, the reference value may correspond to a correlation coefficient for a patient that does not respond to the candidate anti-cancer therapy. Thus, in some aspects, a reference value may correspond to a correlation coefficient for a patient with a favorable (or unfavorable) response to the candidate anti-cancer therapy.

[0007] As used herein a "marker" refers to a distinct parameter in a sample that can be quantified. For example, the marker may comprise expression of a protein, lipid or nucleic acid. In some aspects, distinct markers can comprise expression levels of the same molecule (e.g., a protein or nucleic acid) having different localization in the sample or a different modification (such as different phosphorylation). For example, the different localization can comprise localization in different tissue types (e.g., tumor versus stroma), different cell types, or different cellular compartments (e.g., nuclear versus cytoplasmic localization or extracellular versus intracellular localization).

[0008] In certain aspects, steps (b), (c) and/or (d) of a method of the embodiments may be automated or computer controlled. In some aspects, each of steps (b)-(d) may be automated. Thus, in some aspects, step (b) of the instant methods may comprise obtaining a digital image of the sample and performing quantitative analysis of the image.

[0009] Thus, in some aspects, quantifying the expression of the markers may comprise quantifying a level of protein (such as a secreted or cell surface protein) or nucleic acid expression. In some aspects, a marker for quantification is marker associated with cancer or with cell proliferation. Protein (or other antigen) expression levels may, in some aspects, be quantified by measuring antibody binding (e.g., by antibody staining or immunofluorescence). In some aspects, measuring antibody binding to an antigen may comprise performing an immunohistochemistry assay. Likewise, in certain aspects, nucleic acid levels may be quantified using an in situ hybridization assay.

[0010] Tissue samples for use according to the embodiments can be from any tissue for which analysis is desired. For example, the sample may be an ovarian, prostate, pancreas, muscle, neuronal (e.g., brain), heart, liver spleen, lung, lymph node, stomach, intestinal, colon, skin, blood, testicular or bone (or bone marrow) sample. In one aspect, the tissue sample is a tumor tissue sample, such as a solid tissue section. In some aspect, the plurality of cells for analysis may be comprised in a plurality samples from a patient, such as from a plurality of tissue sections. Thus, in aspects wherein the tissue sample is a solid tumor tissue, the plurality of cells may be from discrete parts of the tumor.

[0011] In certain specific aspects, a method of the embodiments may further comprise identifying the cancer cells in the sample, prior to the quantifying step. For example, tumor cells can be identified by the morphological analysis or by expression of tumor markers. Thus, in some aspects, the expression of the two or more markers may be quantified in the identified cancer cells. [0012] As detailed above, in some aspects, a marker may be an antigen or an antigen localized to a particular compartment. In certain aspects, a marker may be a RNA or a protein such as a kinase, a transcription factor, a cell surface protein or secreted protein. In some specific aspects, two or more markers for use according to the embodiments may be comprised in the same metabolic pathway. For example, the two or more markers may comprise: a receptor and a ligand for the receptor; a kinase and a target for the kinase; or protein components of a multiprotein complex. In a further aspects, the two or more markers can comprise a transcription factor and at least a first target gene of the transcription factor. In yet a further aspect, the two or more markers comprise a miRNA and a target gene of the miRNA.

[0013] In certain aspect, determining a correlation coefficient may comprise performing a Pearson correlation analysis, such as performing a Pearson correlation analysis for a plurality of markers in individual cells in a sample. In some aspect, determining a confidence interval may comprise performing a calculation according to the formula:

€<w (.A, Y ) E[{X - μ_χ) {Υ ^■■■■ μ_γ } \

Ρχ,γ = corr( ^')

(Ι (Ty <7χ(7γ where p is the new synthetic variable, σ is the standard deviation and μ the average of the expression of the markers X or Y.

[0014] Thus, in still a further aspect, a reference value for use according to the embodiments may be a confidence interval corresponding to a known diseased or a non- diseased tissue. Thus, a reference value may be a confidence interval corresponding to a known cancer or non-cancer sample (or a cancer sample known to have certain prognosis or know to respond to a particular therapy).

[0015] In still further aspects, a method of characterizing a cancer according to the embodiments may comprise determining a prognosis of the cancer or determining whether the cancer is predicted to respond to a given therapy. For example, a method may be used to determine whether the cancer is predicted to respond to particular radiation, surgical or chemotherapy agent. In yet further aspects, a method of the embodiments may comprise reporting the characterization of tissue or of a cancer. For example, reporting may comprise preparing an oral, written or electronic report. In some aspects, a report is provided to a patient, a doctor, a hospital or an insurance company.

[0016] In certain specific aspects, the two or more markers for analysis according to the embodiments comprise HGF and/or cMet expression. In still further aspects, the two or more markers may comprise nuclear c-Met and/or cytoplasmic c-Met expression. In certain aspects HGF and/or c-Met (e.g., nuclear and/or cytoplasmic c-Met) is analyzed in an ovarian tumor sample. Thus, in some aspects, a method for characterizing an ovarian tumor tissues sample is provided. In particular, a method is provided for determining the prognosis of an ovarian cancer and/or for determining whether the cancer is predicted to respond to a therapy (such as a c-Met targeted therapy).

[0017] In further aspects, the two or more markers for analysis in a method of the embodiments may comprise OSBPL3 and/or IGFBP3 expression. In certain aspects OSBPL3 and/or IGFBP3 is analyzed in a colorectal cancer tumor sample. Thus, in some aspects, a method for characterizing colorectal cancer tumor sample is provided. In particular, a method is provided for determining the prognosis of a colorectal cancer and/or for determining whether the cancer is predicted to respond to a therapy.

[0018] In yet further aspects, the two or more markers for analysis according to the embodiments comprise a Hypoxia-inducible factor 1 -alpha (HIFIA) gene product and a product of gene regulated by HIFIA (e.g., NIMA (Never In Mitosis Gene A)-Related Kinase 6 (NEK6)). In still further aspects, the two or more markers may comprise HIFIA protein and/or NEK6 RNA expression. In certain aspects, HIFIA protein and/or NEK6 RNA expression is analyzed in situ in a tumor sample (e.g., an ovarian tumor sample). Thus, in some aspects, a method for characterizing an ovarian tumor tissues sample is provided. In particular, a method is provided for determining the prognosis of an ovarian cancer and/or for determining whether the cancer is predicted to respond to a therapy (such as a HIFIA targeted therapy).

[0019] In still further aspects, the two or more markers for analysis according to the embodiments comprise micro RNA miR-301a-5p and a product of gene regulated by miR- 301a-5p (e.g., Laminin-Binding Protein (LBP)). In still further aspects, the two or more markers may comprise miR-301a-5p RNA expression and/or LBP protein expression. In certain aspects, miR-301a-5p and/or LBP is analyzed in situ in a tumor sample (e.g., a colorectal tumor sample). Thus, in some aspects, a method for characterizing a colorectal tumor tissues sample is provided. In particular, a method is provided for determining the prognosis of colorectal cancer and/or for determining whether the cancer is predicted to respond to an anticancer therapy. [0020] In a further embodiment, the present invention provides a method of treating a cancer patient comprising a) selecting a candidate anticancer therapy for a patient (by characterizing the cancer in accordance with the embodiments); and b) administering the selected therapy to the patient.

[0021] In yet a further embodiment, the present invention provides a tangible computer-readable medium comprising computer-readable code that, when executed by a computer, causes the computer to perform operations comprising a) receiving information corresponding to an expression level of two or more markers in a plurality of cells in a sample from a cancer patient; and b) determining a correlation coefficient between the expression of the two or more markers in the plurality of cells and calculating a confidence interval over said plurality of cells in the sample. In some aspects, the media may further comprise computer-readable code that, when executed by a computer, causes the computer to perform operations comprising: c) comparing the calculated confidence interval to a reference value to characterize the cancer; and/or c) providing at least a first candidate agent predicted to be effective to inhibiting cancer cells in the patient based on the calculated confidence interval. In further aspects, the media may further comprise computer-readable code that, when executed by a computer, causes the computer to perform one or more additional operations comprising: sending information corresponding to at least a first candidate agent predicted to be effective to inhibiting cancer cells in the patient to a tangible data storage device. In a further aspect, the receiving of information comprises receiving from a tangible data storage device information corresponding to an expression level of two or more markers in a plurality of cells in a sample from a cancer patient. In a further aspect, receiving information comprises receiving a digitized image of the sample from the cancer patient.

[0022] In certain aspects, methods of the embodiments concern quantifying the expression of a marker in a sample. In some cases such quantifying comprises selectively quantifying the expression or two or more markers. As used herein the phrase "selectively quantifying" refers to methods wherein only a finite number of markers (e.g., proteins, phosphoprotein or nucleic acid markers) are quantified rather than assaying essentially all markers (e.g., proteins or nucleic acids) in a sample. For example, in some aspects "selectively quantifying" nucleic acid or protein markers can refer to measuring no more than 100, 75, 50, 25, 10 or 5 different markers. [0023] In yet a further aspect, a media of the embodiments may further comprise computer-readable code that, when executed by a computer, causes the computer to perform one or more additional operations comprising sending information corresponding to the calculated confidence interval to a tangible data storage device.

[0024] As used herein the specification, "a" or "an" may mean one or more. As used herein in the claim(s), when used in conjunction with the word "comprising", the words "a" or "an" may mean one or more than one.

[0025] The use of the term "or" in the claims is used to mean "and/or" unless explicitly indicated to refer to alternatives only or the alternatives are mutually exclusive, although the disclosure supports a definition that refers to only alternatives and "and/or." As used herein "another" may mean at least a second or more.

[0026] Throughout this application, the term "about" is used to indicate that a value includes the inherent variation of error for the device, the method being employed to determine the value, or the variation that exists among the study subjects.

[0027] Other objects, features and advantages of the present invention will become apparent from the following detailed description. It should be understood, however, that the detailed description and the specific examples, while indicating preferred embodiments of the invention, are given by way of illustration only, since various changes and modifications within the spirit and scope of the invention will become apparent to those skilled in the art from this detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] The following drawings form part of the present specification and are included to further demonstrate certain aspects of the present invention. The invention may be better understood by reference to one or more of these drawings in combination with the detailed description of specific embodiments presented herein. [0029] FIG. 1: Diamond chart depicting results of AQUA® analysis according to the group of clinical interest for HGF (A) and c-Met (B) in 109 ovarian cancer patients. Each dot represents the quantification in a single electronic segment. The top and bottom of each diamond represent the confidence interval for each group mean. The mean line across the middle of each diamond represents the group mean.

[0030] FIG. 2: Distribution range of the results after pooling the same data reported in FIG. 1 for group B-D. Panels A and B depict the distribution range of HGF and c-Met, respectively. There is a significant increase of both HGF and c-Met expression in group A (t- test). The confidence interval (box-plot) is shown in the images. Confidence interval is overlapping across groups A and B-D. The top and bottom of each diamond represents the confidence interval for each group mean. The mean line across the middle of each diamond represents the group mean.

[0031] FIG. 3: ROC (Receiver Operating Characteristic) curve to optimize a discriminator cutoff to identify patients belonging to group A from the data represented in FIG. 2. Accuracy is poor for the overlap between the ranges in both HGF and c-Met, which are depicted in panels A and B, respectively.

[0032] FIG. 4: Using the same data reported in FIGs. 1 and 2, the system object of this invention is used. Calculation of the correlation with the Pearson method produced non- overlapping results and the possibility to identify patients belonging to group A (refractory patients) by the absence of overlapping ranges.

[0033] FIG. 5: (A) Distribution range of the results after traditional analysis of stromal c-Met expression in group A vs. group B-D. There is a significant increase of c-Met expression in group A (t-test). The confidence interval (box-plot) is shown in the images and is overlapping across groups A and B-D. The top and bottom of each diamond represent the confidence interval for each group mean. The mean line across the middle of each diamond represents the group mean. (B) ROC curve after optimization of the cutoff using data shown in panel A. Accuracy is poor for the overlap between the range of stromal c-Met.

[0034] FIG. 6: Using the same data reported in FIG. 5, the system object of this invention is used. Calculation of the correlation of c-Met expression in stromal and cancer cells with the Pearson method produces non-overlapping results and the possibility to identify patients belonging to group A (refractory patients) by the absence of overlapping ranges. [0035] FIG. 7: Distribution range of the results after traditional analysis of OSBPL3 (A) and IGFBP3 (B) expression in a panel of colorectal cancer patients with benign (n=110) or aggressive disease (n=71). A significant increase of OSBPL3 was noticed in patients with aggressive disease (t-test). (C-D) ROC curves generated for the same data shown in panels A and B for OSBPL3 (C) and IGFBP3 (D). After optimization of the cutoff both antigens cannot be used as predictive biomarker in colorectal cancer patients for the low values of AUC.

[0036] FIG. 8: Using the same data shown in FIG. 7, the system object of this invention is used. Calculation of the correlation between OSBPL3 and IGFBP3 with the Pearson method produces non-overlapping results and the possibility to identify patients with aggressive disease by the absence of overlapping ranges.

[0037] FIG. 9: Diamond chart depicting results of AQUA® analysis according to the group of clinical interest for nuclear (A) and cytoplasmic (B) c-Met in 109 ovarian cancer patients. Each dot represents the quantification in a single electronic segment. The top and bottom of each diamond represent the confidence interval for each group mean. The mean line across the middle of each diamond represents the group mean.

[0038] FIG. 10: Distribution range of the results after pooling the same data reported in FIG. 9 for group B-D. Panels A and B depict the distribution range of nuclear and cytoplasmic c-Met, respectively. There is a significant increase of both nuclear and cytoplasmic c-Met expression in group A (t-test). The confidence interval (box-plot) is shown in the images. Confidence interval is overlapping across groups A and B-D. The top and bottom of each diamond represents the confidence interval for each group mean. The mean line across the middle of each diamond represents the group mean. Panels C-D: ROC curve to optimize a discriminator cutoff to identify patients belonging to group A from the data represented in FIGs. 10 A and B. Accuracy is poor for the overlap between the ranges in both nuclear and cytoplasmic c-Met, which are depicted in panels C and D, respectively.

[0039] FIG. 11: Using the same data reported in FIGs. 9 and 10, the system object of this invention is used. Calculation of the correlation with the Pearson method produced non- overlapping results and the possibility to identify patients belonging to group A (refractory patients) by the absence of overlapping ranges. [0040] FIG 12: Diamond chart depicting results of AQUA®, in situ protein expression, analysis according to the group of clinical interest for NEK6 (A) and Hif-loc (B) in 348 serous ovarian cancer patients. Each dot represents the quantification in a single electronic segment. The top and bottom of each diamond represent the confidence interval for each group mean. The mean line across the middle of each diamond represents the group mean.

[0041] FIG 13: Distribution range of the results after pooling the same data reported in FIG. 13 for group B-D. Panels A and B depict the distribution range of NEK6 and Hif-loc, respectively. There is no significant difference between expressions of the two markers in group A (t-test). The confidence interval (box-plot) is shown in the images. Confidence interval is overlapping across groups A and B-D. The top and bottom of each diamond represents the confidence interval for each group mean. The mean line across the middle of each diamond represents the group mean. Panels C-D: ROC curve to optimize a discriminator cutoff to identify patients belonging to group A from the data represented in FIGs. 13A and B. Accuracy is poor for the overlap between the ranges in both NEK6 and Hif-loc, which are depicted in panels C and D, respectively.

[0042] FIG 14: Using the same data reported in FIGs. 12 and 13, the correlation of the markers is calculated, in accordance with embodiments of the invention, using the Pearson method to produced non-overlapping results. As shown, this method allows for highly accurate identification of patients belonging to group A (refractory patients) by the absence of overlapping ranges.

[0043] FIG. 15: Kaplan-Meier analysis (A) for 358 colorectal cancer patients with high and low expression of miR-301a-5p. Patients were stratified using as a cutoff the 75* percentile. Continuous and dotted lined were used to show low (<75* percentile) and high (>75^ώ percentile) miR-301a-5p expression. High expression of miR-301a-5p was correlated with poor outcome (p<0.001, Wilcoxon test). Diamond charts depicting the expression range of LBP (B) and miR-301a-5p in the same clinical cohort of 358 patients. Each dot represents the quantification in a single electronic segment. The top and bottom of each diamond represent the confidence interval for each group mean. The mean line across the middle of each diamond represents the group mean. [0044] FIG. 16: ROC curve to optimize a discriminator cutoff to identify patients belonging to the group with tumor from the data represented in FIG. 15B and C. Data are presented for miR-301a-5p (A) and LBP (B). Accuracy is suboptimal for the overlap between the ranges in both factors. [0045] FIG. 17: Using the same data reported in FIGs. 15 and 16, the correlation of the markers is calculated, in accordance with embodiments of the invention, using the Pearson method to produce non-overlapping results. As shown, this method allows for highly accurate identification of patients belonging to the group who will recur within five years from diagnosis due to the not overlapping range of distributions. DESCRIPTION OF ILLUSTRATIVE EMBODIMENTS

[0046] The present invention provides, in part, compositions and methods for the analysis and characterization of tumor samples. Previously tumor tissue analysis has relied upon a combination of morphological characterization and, in some cases, immunohistochemistry staining for a particular antigen. However, such techniques have proven completely inadequate to properly characterize tumor tissue in a way that can accurately predict tumor prognosis and/or therapeutic response. In contrast the new analysis methods detailed herein provide for highly specific and accurate tissue analysis that can be used to guide individual patient care.

[0047] Tissue analysis techniques previously involved mere qualitative assessments of marker (e.g., antigen) expression. Previously cells were enumerated and characterized as bearing an antigen of interest or note. The sample was then defined as positive or negative for a given antigen if the expression is higher than a given threshold. However, in tumor tissues, such enumeration of cells is made difficult by the problem of clonal heterogeneity. In some cancers, the expression of a given antigen is restricted to only a few cancer cells inside the analyzed population. Either with the use of a standard microscope or with an image- analysis system the pathologist counts the cells positive for the antigen of interest across multiple fields. Then, the number of positive cells for each field is calculated as a percentage of the total cells. Percentages of positive cells are finally averaged to represent the expression of a given antigen inside a cancer specimen. [0048] In contrast to previous techniques, embodiments of the present invention provide a new system to interpret the expression of an antigen of interest in a heterogeneous population of cells. Instead of using the average of the expression of the antigen of interest, these new methodologies use a new synthetic variable obtained with the correlation matrix among the expression of two parameters. These two parameters (e.g., two markers) can be two different antigens inside the same specimen (i.e., a ligand and its receptor) or the same antigen in two different cell populations (cancer cells vs. stromal cells) or two different compartments (expression in the nucleus vs. expression in the cytoplasm).

[0049] Thus, the methods of the present embodiments are useful to accurately interpret the clinical significance of the expression of an antigen of interest in a sample containing a heterogeneous population of cells. A growing number of targeted agents are clinically used only if a patient expresses the target of a given drug. For this reason in oncology several diagnostic tests have been developed to monitor the expression of an antigen and then decide if a patient can benefit or not from a given treatment. However, even such tests often do not adequately predict responders versus non-responders. Using the analysis techniques detailed here markers for drug response can be more accurately assessed in tumor tissues and patients can be properly selected for the therapeutic most likely to benefit their disease state.

I. Biomarker detection

[0050] The expression of biomarkers or gene products may be measured by a variety of techniques that are well known in the art. In perfered aspects of the embodiments such measurements are quantitative and performed in situ in a tissue sample. In some aspects, quantifying the levels of the messenger RNA (mRNA) of a gene may be used to measure the expression of the biomarker. Alternatively, quantifying the levels of the protein product of a biomarker (or the phosphorylation status or localization of a protein) may be used to measure the expression of the biomarker. One skilled in the art will know which parameters may be manipulated to optimize detection of the RNA or protein of interest.

[0051] In some embodiments, said obtaining expression information may comprise RNA quantification, e.g. , cDNA microarray, quantitative RT-PCR, in situ hybridization, Northern blotting or nuclease protection. Said obtaining expression information may comprise protein quantification, e.g., protein quantification comprises immunohistochemistry, an ELISA, a radioimmunoassay (RIA), an immunoradiometric assay, a fluoroimmunoassay, a chemiluminescent assay, a bioluminescent assay, a gel electrophoresis, a Western blot analysis, a mass spectrometry analysis, or a protein microarray.

[0052] In a further embodiment, a level of expression may be compared to a control. The control may comprise data obtained at the same time (e.g. , in the same hybridization experiment) as the patient's individual data, or may be a stored value or set of values, e.g. , stored on a computer, or on computer-readable media. If the latter is used, new patient data for the selected marker(s), obtained from initial or follow-up samples, can be compared to the stored data for the same marker(s) without the need for additional control experiments. A. Methods of protein detection

[0053] In some aspects, measuring the expression of said genes comprises measuring protein expression levels. Measuring protein expression levels may comprise, for example, performing an ELISA, Western blot, immunohistochemistry, or binding to an antibody array. In certain aspects, determining a level of a phosphoprotein in a sample comprises contacting the sample with a phosphorylation specific antibody to the indicated phosphoprotein.

[0054] Immunohistochemical staining may also be used to measure the differential expression of a plurality of biomarkers. This method enables the localization of a protein in the cells of a tissue section by interaction of the protein with a specific antibody. For this, the tissue may be fixed in formaldehyde or another suitable fixative, embedded in wax or plastic, and cut into thin sections (from about 0.1 mm to several mm thick) using a microtome. Alternatively, the tissue may be frozen and cut into thin sections using a cryostat. The sections of tissue may be arrayed onto and affixed to a solid surface (i.e. , a tissue microarray). The sections of tissue are incubated with a primary antibody against the antigen of interest, followed by washes to remove the unbound antibodies. The primary antibody may be coupled to a detection system, or the primary antibody may be detected with a secondary antibody that is coupled to a detection system. The detection system may be a fluorophore or it may be an enzyme, such as horseradish peroxidase or alkaline phosphatase, which can convert a substrate into a colorimetric, fluorescent, or chemiluminescent product. The stained tissue sections are generally scanned under a microscope. Because a sample of tissue from a subject with cancer may be heterogeneous, i.e. , some cells may be normal and other cells may be cancerous, the percentage of positively stained cells in the tissue may be determined. This measurement, along with a quantification of the intensity of staining, may be used to generate an expression value for the biomarker.

[0055] Thus, a preferred aspect of the present invention involves immunohistochemistry techniques for evaluating tumor tissues samples of a cancer patient. Specifically, methods can involve comprises antibody staining of biomarkers within a tissue sample, more particularly a tumor sample, that are indicative of prognosis. One of skill in the art will recognize that the immunohistochemistry methods described herein below may be performed manually or in an automated fashion using, for example, the Autostainer Universal Staining System (Dako). [0056] In one immunohistochemistry method, a patient tissue sample is collected by, for example, biopsy techniques known in the art. Samples may be frozen for later preparation or immediately placed in a fixative solution. Tissue samples may be fixed by treatment with a reagent such as formalin, glutaraldehyde, methanol, or the like and embedded in paraffin. Methods for preparing slides for immunohistochemical analysis from formalin-fixed, paraffin-embedded tissue samples are well known in the art.

[0057] In some embodiments, particularly the immunohistochemistry methods of the invention, samples may need to be modified in order to make the biomarker antigens accessible to antibody binding. For example, formalin fixation of tissue samples results in extensive cross-linking of proteins that can lead to the masking or destruction of antigen sites and, subsequently, poor antibody staining. As used herein, "antigen retrieval" or "antigen unmasking" refers to methods for increasing antigen accessibility or recovering antigenicity in, for example, formalin-fixed, paraffin- embedded tissue samples. Any method for making antigens more accessible for antibody binding may be used in the practice of the invention, including those antigen retrieval methods known in the art. [0058] Antigen retrieval methods include but are not limited to treatment with proteolytic enzymes (e.g., trypsin, chymoptrypsin, pepsin, pronase, etc.) or antigen retrieval solutions. Antigen retrieval solutions of interest include, for example, citrate buffer, pH 6.0 (Dako), Tris buffer, pH 9.5 (Biocare), EDTA, pH 8.0 (Biocare), L.A.B. ("Liberate Antibody Binding Solution;" Polysciences), antigen retrieval Glyca solution (Biogenex), citrate buffer solution, pH 4.0 (Zymed), Dawn® detergent (Proctor & Gamble), deionized water, and 2% glacial acetic acid. In some embodiments, antigen retrieval comprises applying the antigen retrieval solution to a formalin-fixed tissue sample and then heating the sample in an oven (e.g., 60°C), steamer (e.g., 95°C), or pressure cooker (e.g., 120°C) at specified temperatures for defined time periods. In other aspects of the invention, antigen retrieval may be performed at room temperature. Incubation times will vary with the particular antigen retrieval solution selected and with the incubation temperature. For example, an antigen retrieval solution may be applied to a sample for as little as 5, 10, 20, or 30 minutes or up to overnight. The design of assays to determine the appropriate antigen retrieval solution and optimal incubation times and temperatures is standard and well within the routine capabilities of those of ordinary skill in the art. [0059] Following antigen retrieval, samples are blocked using an appropriate blocking agent, e.g., hydrogen peroxide. An antibody directed to a biomarker of interest is then incubated with the sample for a time sufficient to permit antigen-antibody binding. As noted above, one of skill in the art will appreciate that a more accurate cancer prognosis may be obtained in some cases by detecting overexpression of more than one biomarker in a patient sample. Therefore, in particular embodiments, at least two antibodies directed to two distinct biomarkers are used to evaluate the prognosis of a cancer patient. Where more than one antibody is used, these antibodies may be added to a single sample sequentially as individual antibody reagents or simultaneously as an antibody cocktail. Alternatively, each individual antibody may be added to a separate tissue section from a single patient sample and the resulting data pooled.

[0060] Techniques for detecting antibody binding are well known in the art. Antibody binding to a biomarker of interest may be detected through the use of chemical reagents that generate a detectable signal that corresponds to the level of antibody binding and, accordingly, to the level of biomarker protein expression. For example, antibody binding can be detected through the use of a secondary antibody that is conjugated to a labeled polymer. Examples of labeled polymers include but are not limited to polymer-enzyme conjugates. The enzymes in these complexes are typically used to catalyze the deposition of a chromogen at the antigen-antibody binding site, thereby resulting in cell staining that corresponds to expression level of the biomarker of interest. Enzymes of particular interest include horseradish peroxidase (HRP) and alkaline phosphatase (AP). Commercial antibody detection systems, such as, for example the Dako Envision+ system and Biocare Medical's Mach 3 system, may be used to practice the present invention. [0061] Detection of antibody binding can be facilitated by coupling the antibody to a detectable substance. Examples of detectable substances include various enzymes, prosthetic groups, fluorescent materials, luminescent materials, bioluminescent materials, and radioactive materials. Examples of suitable enzymes include horseradish peroxidase, alkaline phosphatase, β-galactosidase, or acetylcholinesterase; examples of suitable prosthetic group complexes include streptavidin/biotin and avidin/biotin; examples of suitable fluorescent materials include umbelliferone, fluorescein, fluorescein isothiocyanate, rhodamine, dichlorotriazinylamine fluorescein, dansyl chloride or phycoerythrin; an example of a luminescent material includes luminol; examples of bioluminescent materials include luciferase, luciferin, and aequorin; and examples of suitable radioactive material include 125 I, ¹³¹1, ³⁵S, ³²P or ³H.

[0062] The terms "antibody" and "antibodies" broadly encompass naturally occurring forms of antibodies, including polyclonal and monoclonal antibodies, and recombinant antibodies, such as single-chain antibodies, chimeric and humanized antibodies and multi-specific antibodies, as well as fragments and derivatives of all of the foregoing, which fragments and derivatives have at least an antigenic binding site. Antibody derivatives may comprise a protein or chemical moiety conjugated to the antibody. "Antibodies" and "immunoglobulins" (Igs) are glycoproteins having the same structural characteristics. While antibodies exhibit binding specificity to an antigen, immunoglobulins include both antibodies and other antibody-like molecules that lack antigen specificity. Polypeptides of the latter kind are, for example, produced at low levels by the lymph system and at increased levels by myelomas. The term "antibody" is used in the broadest sense and covers fully assembled antibodies, antibody fragments that can bind antigen (e.g., Fab', F'(ab)2, Fv, single chain antibodies, diabodies), and recombinant peptides comprising the foregoing. B. Methods of nucleic acid detection

[0063] In another aspect, measuring expression of gene products comprises measuring RNA expression levels. Measuring RNA expression levels may comprise performing RT-PCR, Northern blot or in situ hybridization.

[0064] Quantitative real-time PCR (qRT-PCR) may also be used to measure the differential expression of a plurality of biomarkers. In qRT-PCR, the RNA template is generally reverse transcribed into cDNA, which is then amplified via a PCR reaction. The amount of PCR product is followed cycle-by-cycle in real time, which allows for determination of the initial concentrations of mRNA. To measure the amount of PCR product, the reaction may be performed in the presence of a fluorescent dye, such as SYBR Green, which binds to double- stranded DNA. The reaction may also be performed with a fluorescent reporter probe that is specific for the DNA being amplified. [0065] A non-limiting example of a fluorescent reporter probe is a TaqMan® probe (Applied Biosystems, Foster City, CA). The fluorescent reporter probe fluoresces when the quencher is removed during the PCR extension cycle. Multiplex qRT-PCR may be performed by using multiple gene- specific reporter probes, each of which contains a different fluorophore. Fluorescence values are recorded during each cycle and represent the amount of product amplified to that point in the amplification reaction. To minimize errors and reduce any sample-to-sample variation, qRT-PCR may be performed using a reference standard. The ideal reference standard is expressed at a constant level among different tissues, and is unaffected by the experimental treatment. Suitable reference standards include, but are not limited to, mRNAs for the housekeeping genes glyceraldehyde-3-phosphate-dehydrogenase (GAPDH) and β-actin. The level of mRNA in the original sample or the fold change in expression of each biomarker may be determined using calculations well known in the art.

[0066] Luminex multiplexing microspheres may also be used to measure the differential expression of a plurality of biomarkers. These microscopic polystyrene beads are internally color-coded with fluorescent dyes, such that each bead has a unique spectral signature (of which there are up to 100). Beads with the same signature are tagged with a specific oligonucleotide or specific antibody that will bind the target of interest (i.e. , biomarker mRNA or protein, respectively). The target, in turn, is also tagged with a fluorescent reporter. Hence, there are two sources of color, one from the bead and the other from the reporter molecule on the target. The beads are then incubated with the sample containing the targets, of which up to 100 may be detected in one well. The small size/surface area of the beads and the three dimensional exposure of the beads to the targets allows for nearly solution-phase kinetics during the binding reaction. The captured targets are detected by high-tech fluidics based upon flow cytometry in which lasers excite the internal dyes that identify each bead and also any reporter dye captured during the assay. The data from the acquisition files may be converted into expression values using means known in the art.

[0067] In situ hybridization may also be used to measure the differential expression of a plurality of biomarkers. This method permits the localization of mRNAs of interest in the cells of a tissue section. For this method, the tissue may be frozen, or fixed and embedded, and then cut into thin sections, which are arrayed and affixed on a solid surface. The tissue sections are incubated with a labeled antisense probe that will hybridize with an mRNA of interest. The hybridization and washing steps are generally performed under highly stringent conditions. The probe may be labeled with a fluorophore or a small tag (such as biotin or digoxigenin) that may be detected by another protein or antibody, such that the labeled hybrid may be detected and visualized under a microscope. Multiple mRNAs may be detected simultaneously, provided each antisense probe has a distinguishable label. The hybridized tissue array is generally scanned under a microscope. Because a sample of tissue from a subject with cancer may be heterogeneous, i.e. , some cells may be normal and other cells may be cancerous, the percentage of positively stained cells in the tissue may be determined. This measurement, along with a quantification of the intensity of staining, may be used to generate an expression value for each biomarker.

II. Examples

[0068] The following examples are included to demonstrate preferred embodiments of the invention. It should be appreciated by those of skill in the art that the techniques disclosed in the examples which follow represent techniques discovered by the inventor to function well in the practice of the invention, and thus can be considered to constitute preferred modes for its practice. However, those of skill in the art should, in light of the present disclosure, appreciate that many changes can be made in the specific embodiments which are disclosed and still obtain a like or similar result without departing from the spirit and scope of the invention.

Example 1 - Prediction of ovarian cancer patient outcome

[0069] Ovarian cancer is the deadliest gynecologic malignancy. Due to its indolence, diagnosis is often made when the disease is advanced and complete surgical debulking is not possible. Therefore chemotherapy is needed to induce remission and decrease the rate of relapse (Kyrgiou et ah , 2006). HGF (hepatocyte growth factor) and its receptor c-Met are involved in cancer progression and several agents have been developed to target this pathway (Blumenschein et ah, 2012). Activation of the HGF/c-Met pathway is typical of an aggressive cancer with low tendency to respond to standard treatments. For such patients, a targeted treatment with inhibitors of the HGF/c-Met pathway could be beneficial (Tang et ah , 2010). In order to select patients eligible for treatment with an agent against HGF/c-Met, an analysis of the expression of at least one of these two proteins is required. In a retrospective study of a clinical setting of 109 ovarian cancer patients such analysis was performed using c-Met and HGF as antigens of interest and the traditional method of quantification.

[0070] Patients were divided into four clinical groups of interest according to platinum free interval (PFI). PFI is the time interval (measured in months) between the last cycle of standard chemotherapy and the relapse of the disease. Patients were divided into four groups (0-3 months, 4-6 months, 7-12 months, >13 months will be referred to as groups A, B, C and D, respectively). Group A is the group completely refractory to the standard treatment for ovarian cancer with a median overall survival of 12 months, while group D is responsive with a median overall survival of 96 months. The test was designed to identify the patients who benefit/do not benefit from the standard treatment. Group B and C exhibited median overall survival of 24 and 40 months, respectively.

[0071] Calculation of the expression of the two antigens was performed using the AQUA® system. Automated quantitative analysis (AQUA®) allows exact measurement of protein concentration within compartments of interest. In brief, a series of high-resolution digital images were captured by the ScanScope FL system (Aperio, Vista, CA). For each histospot, images were obtained using the signal from nuclei (4,6-Diamidino-2- phenylindole), CD68 (HGF staining) or vimentin (c-Met staining); cytokeratin was revealed with the use of Alexa 555. Tumor was distinguished from stromal or macrophage elements by creating an epithelial tumor "mask" from the cytokeratin signal and a stromal or macrophages "mask" from the vimentin or CD68 signals, respectively. The images were analyzed by the AQUA® analysis software (Genoptix, Carlsbad, CA) to determine target protein expression in the epithelial, stromal, and macrophage cell populations by calculating the sum of target pixel intensity divided by the compartment area and normalized for exposure time.

[0072] For each patient, up to 9 cores were sampled in order to represent different areas of each individual cancer. Thereafter, at the stage of the electronic analysis, cores (2 mm of diameter) were electronically segmented in two equal parts. Therefore, for each patient, there was the possibility to have up to 18 values representing the expression of the antigen of interest inside the cancer. [0073] Analysis was first conducted using the traditional method. In this case the expression values coming from the patients belonging to each of the 4 groups of clinical interest was averaged (FIG. 1). Significant differences in the expression range among the four groups were noticed for both HGF and c-Met expression. In particular both HGF and c-Met expression were higher in group A as compared to the other three (FIG. 2). However, there was a full overlap in the distribution range and also after a procedure of optimization of the cutoff, known for the experts in the field as ROC curve, the method of prediction of patients with refractory disease (group A) calculated according to the traditional analysis was heavily biased with AUC values of 0.67 and 0.61 for HGF and c-Met, respectively (FIG. 3). The reason underlying this fact is dependent on clonal heterogeneity, namely in a given cancer the expression of the antigen in each individual segment is not constant but presents huge variations. Any system based on average will not represent correctly the biology of a cancer and will be biased by false positive and false negative results.

[0074] Using the same data shown in FIG. 1, the inventors performed the analysis using the system disclosed herein. Exploiting clonal heterogeneity as an advantage, the inventors did not perform any averaging and calculated instead the correlation between the expression of HGF and c-Met. The correlation was calculated using the Pearson method. Briefly, it was obtained by dividing the covariance of the two variables by the product of their standard deviations. The population correlation coefficient ρΧ,Υ between two random variables X and Y with expected values uX and μΥ and standard deviations σΧ and σΥ is defined as:

,. _ cov( Y) E[(X - μ_χ) (Υ - μ_γ}\

Ρχ γ = corrf A . r }— = — , where E is the expected value operator, cov means covariance, and corr is a widely used notation for the correlation coefficient. The Pearson correlation is +1 in the case of a perfect positive (increasing) linear relationship (correlation), -1 in the case of a perfect decreasing (negative) linear relationship (anticorrelation), and some value between -1 and 1 in all other cases, indicating the degree of linear dependence between the variables. As it approaches zero there is less of a relationship (closer to uncorrected). The closer the coefficient is to either -1 or 1, the stronger the correlation between the variables. Therefore we obtained a p value representing the correlation between HGF and c-Met in our clinical study. As depicted in FIG. 4, if in group A the range of p was 0.78 (confidence interval (CI) 0.66-0.87), then in the other groups the values were always outside the CI of group A (group B 0.30 (CI 0.08- 0.48), group C 0.20 (CI 0.0-0.37), group D 0.14 (CI -0.11-0.38)). This means that it is possible to correctly predict with absolute accuracy that a patient will be refractory (group A) to first line treatment of ovarian cancer since the ranges of the p values are non-overlapping. Therefore, a patient exhibiting a p value higher than 0.66 belongs to group A. The same patient will possibly also benefit from treatment with an inhibitor of the HGF/c-Met axis while the others will not, since they are at least partial responders to first line chemotherapy.

Example 2 - Prediction of outcome based on the expression of the same protein in two different cell populations [0075] The method of the present invention can be also applied to prediction based on the expression of the same protein in two different cell populations. In this example, the inventors used the same clinical set reported above. Due to the multiplexing the inventors are able to calculate the expression of a marker of interest in different populations of interest, such as cancer cells and stromal cells. The two populations are identified with the use of an appropriate mask in the analysis and based on the gating of the fluorescence in independent channels. In this specific case the fluorescence is gated with the use of cytokeratin and vimentin as specific markers of cancer and stromal cells, respectively. The antigen of interest is c-Met. As reported in FIGs. IB, 2B and 3B, using the traditional method, the expression of c-Met in cancer cells is not able to identify the clinical population of interest (group A) with an acceptable accuracy. Calculation of the expression of c-Met in the stromal cells does not change the result. Although the expression of c-Met in stromal cells is higher in group A than in group B-D (FIG. 5A), the presence of an overlapping range makes impossible to use such a difference to predict the clinical outcome of ovarian cancer patients with good accuracy (FIG. 5B). If the methods of the present invention are used instead of the traditional method, then the results are different. The correlation between the expression of c-Met in the cancer and stromal cells (FIG. 6) is outstandingly high in group A (p=0.92 CI (0.88-0.95)) while it is much lower in the other three clinical groups (group B 0.54 (CI 0.38-0.72), group C 0.51 (CI 0.36-0.61), group D 0.57 (CI -0.38-0.72)). Therefore, the calculation of the p value is more effective than the traditional method to accurately predict the outcome of ovarian cancer patients since there is no overlapping range between the group A and the other three groups of interest. Example 3 - Prediction of outcome using two different markers that are not

ligand/receptor

[0076] Colorectal cancer patients exhibit a disease with diverse clinical behavior. As an example, if diagnosed at an early stage, such as stage II, some patients exhibit aggressive disease and poor outcome not benefiting from standard treatments, while others need only surgery without addition of chemotherapy (Donada et ah , 2013). In this context, the inventors performed a retrospective analysis on 181 colorectal cancer patients. Patients were categorized as good or poor responders according to the outcome. Patients who relapsed and died within three years after surgery (n=71) were compared with patients who responded to treatment (n=110). Analysis was performed at the protein level using quantitative fluorescent IHC. For each patient up to 9 cores were sampled in order to represent different areas of each individual cancer. Thereafter, at the stage of the electronic analysis, cores (2 mm of diameter) were electronically segmented in two equal parts. Therefore, for each patient, there was the possibility to have up to 18 values representing the expression of the antigen of interest inside the cancer. Antigens of interest were OSBPL3, a factor involved in cholesterol metabolism (Jusakul et ah , 2011) and IGFBP3 (Georges et ah , 2011), a protein involved in metastasis formation in colorectal cancer. Patients were grouped as described above and the results were averaged as for the traditional method. When analyzed with the traditional method, OSBPL3 was found to be significantly more highly expressed in patients with aggressive disease (FIG. 7A), while IGFBP3 exhibited a non-statistically significant increase in expression (FIG. 7B). When used as biomarker to predict a poor outcome, both antigens exhibited an AUC not compatible with use as biomarker (FIGs. 7C and D). Using the system disclosed herein, analysis conducted on the same data led to a different result. The correlation noticed in patients with aggressive disease (p=0.36 CI 0.29-0.43) showed higher p values as compared with those (p=0.17 CI 0.10-0.23) in patients with a more favorable outcome (FIG. 8). Due to the non-overlapping range of the CI, the simple correlation analysis inside multiple specimens coming from the same patients will be sufficient to predict the outcome of a colorectal cancer patient.

Example 4 - Prediction of outcome using the same antigen expressed in two different subcellular compartments (nucleus vs. cytoplasm)

[0077] The method of the present invention can be applied also to the expression of the same antigen in two different subcellular compartments. Thus in this case, the antigen in each compartment serve as separate markers. In this example, the inventors used the same clinical set reported above. Due to the multiplexing the inventors were able to calculate the expression of a marker of interest in different subcellular compartments in cancer cells such as nucleus and cytoplasm. The two compartments were identified with the use of appropriate masks in the analysis and based on the gating of the fluorescence in independent channels. Using the traditional method of analysis, c-Met expression was analyzed in the nucleus and in the cytoplasm of cancer cells in the four groups of patients stratified according to PFI. In both nucleus and cytoplasm, c-Met expression resulted higher in group A (FIG. 9). The same difference was conserved also after pooling the group B-D (FIGs. 10A and B). Also after optimization of the cutoff with the use of ROC curve, both nuclear and cytoplasmic c-Met did not provide a satisfactory biomarker of refractory disease for the low value of AUC which were 0.6338 (FIG. IOC) and 0.6119 (FIG. 10D) for nuclear and cytoplasmic c-Met, respectively. The reason underlying the inefficiency of traditional analysis relies on the presence of overlapping ranges across the clinical group of interests (A vs. B-D). Conversely, the object of this invention provided an optimal discrimination. Once the correlation was calculated (FIG. 11) the p value in group A (p=0.9829 CI 0.9727-0.9893) showed a non- overlapping range as compared with group B (p=0.6739 CI 0.545-0.7717), C (p=0.7633 CI 0.6711-0.8322) and D (p=0.8786 CI 0.8092-0.9238). Therefore according to these results it is possible to predict with absolute accuracy that a patient belongs to group A if the p value is higher than 0.9727.

Example 5 - Prediction of outcome using the expression of a transcription factor and its target gene

[0078] The method of the present embodiments can be applied also to the expression of a transcription factor and its target gene. The gene HIF1A encodes for the protein Hif-la (Hypoxia-inducible factor 1-alpha), one of the pivotal transcription factors induced by hypoxia (Semenza, 2013). Using the multiplexing the inventors analyzed the expression of the protein Hif-la and its target gene NEK6 in a clinical cohort of 348 serous ovarian cancer patients. NEK6 (NIMA (Never In Mitosis Gene A)-Related Kinase 6) is a serine-threonine protein kinase required for progression through the metaphase portion of mitosis. NEK6 is elevated in malignant tumors and human cancer cell lines as compared with normal tissue and fibroblast cells and is believed to play a role in tumorigenesis (Nassirpour et al., 2010). Expression of nuclear Hif-la was detected with a specific monoclonal antibody and quantified with the AQUA® in situ protein expression analysis technology. The expression of NEK6 gene was retrieved with a custom made probe, recognizing specifically a unique sequence for this gene. The probe was labeled with digoxigenin and detected with an anti- digoxigenin monoclonal antibody. Quantification of NEK6 was also performed with AQUA® technology. As described in previous examples, patients were grouped in four clinical groups according to PFI. Group A represented the refractory population, while group B, C and D were patients relapsing between 3 and 6 months, between 6 and 12 months and after 12 months, respectively. The obtained expression is visualized in Fig.12. In order to connect expression of both Hif-la and NEK6 and outcome, firstly the traditional analysis of the independent factors was performed. As shown in Fig. 13 using the traditional method, the expression of both NEK6 and Hif-la was not able to identify the clinical population of interest (group A) with an acceptable accuracy. The reason underlying the inefficiency of traditional analysis relies on the presence of overlapping ranges across the clinical group of interests (A vs. B-D). Using the system of the embodiments disclosed herein, analysis conducted on the same data led to a different result. The creation of a synthetic variable depicting the correlation between expressions of Hif-la and NEK6 (Fig. 14) is outstandingly high in group A (p=0.77 CI (0.61-0.87)) while it is much lower in the other three clinical groups (group B 0.12 (CI -0.16-0.38), group C 0.09 (CI -0.14-0.32), group D 0.57 (CI -0.10- 0.29)). Therefore, the calculation of the p value is more effective than the traditional method to accurately predict the outcome of ovarian cancer patients since there is no overlapping range between the group A and the other three groups of interest. In fact, a p value higher than 0.61 will accurately identify a patient with refractory disease.

Example 6 - Prediction of outcome using the expression of a microRNA and its target gene [0079] The method of the present invention can be applied also to the expression of a microRNA and its target gene. Micro-RNAs are small not coding RNA which are capable to regulate gene expression and translation (Bhayani et ah, 2012). Using the multiplexing the inventors investigated the expression of miR-301a-5p (Chen et ah, 2012) and of its target Colon Carcinoma Laminin-Binding Protein (LBP). LBP is a factor which is activated in hypoxia and seems to confer to colorectal cancer an increased aggressiveness (Hongo et ah, 2013). Analysis was conducted retrospectively in a clinical subset of 358 colorectal cancer patients. Patients were divided in two clinical categories. Patients who 5-years after treatment resulted free of disease (tumor-free n=180) and patients who relapsed within the same time interval (with tumor n=178). Expression of miR-301a-5p was assessed with a custom made probe conjugated with digoxigenin. An antibody anti-digoxigenin served to monitor then the microRNA expression. The target was instead analyzed at the protein level with the use of a polyclonal antibody anti-LBP. Both targets were quantified with AQUA® technology. The expression of miR-301a-5p was correlated with poor outcome (FIG. 15A). Expression levels of both miR-301a-5p (FIG. 15B) and LBP (FIG. 15C) were higher in patients with tumor as compared with patients tumor free. When used as predictor of relapse, accuracy of both biomarkers were suboptimal for clinical use (FIG. 16). In fact, in both cases the distribution ranges were overlapping and either miR-301a-5p or LBP were incapable to categorize patients at risk of recurrence without possibility of misclassification. Using the system disclosed herein, analysis conducted on the same data led to a different result. Firstly, the inventors generated the synthetic variable obtained with the Pearson correlation between expressions of miR-301a-5p and LBP (FIG. 17). The synthetic variable was lower in the group with tumor (p=0.17 CI (0.02-0.31) as compared with the group tumor free (p=0.51 CI (0.39-0.61). These results indicate the disruption of the regulatory circuit between the expression of miR-301a-5p and LBP in those patients who will have relapse within five years from diagnosis. Therefore, the calculation of the p value in accordance with the embodiments can be applied to predict if a patient is at risk of aggressive disease with no possibility of misclassification. In fact, a p value lower than 0.31 will accurately classify a patient who will recur within 5 years from diagnosis.

[0080] All of the methods disclosed and claimed herein can be made and executed without undue experimentation in light of the present disclosure. While the compositions and methods of this invention have been described in terms of preferred embodiments, it will be apparent to those of skill in the art that variations may be applied to the methods and in the steps or in the sequence of steps of the method described herein without departing from the concept, spirit and scope of the invention. More specifically, it will be apparent that certain agents which are both chemically and physiologically related may be substituted for the agents described herein while the same or similar results would be achieved. All such similar substitutes and modifications apparent to those skilled in the art are deemed to be within the spirit, scope and concept of the invention as defined by the appended claims. REFERENCES

The following references, to the extent that they provide exemplary procedural or other details supplementary to those set forth herein, are specifically incorporated herein by reference.

Bhayani, M. K., Calin, G. A., and Lai, S. Y. (2012). Functional relevance of miRNA sequences in human disease. Mutat Res 731, 14-19.

Blumenschein, G. R., Jr., Mills, G. B., & Gonzalez-Angulo, A. M. (2012). Targeting the hepatocyte growth factor-cMET axis in cancer therapy. J Clin Oncol, 30(26), 3287-

3296.

Chen, Z., Chen, L. Y., Dai, H. Y., Wang, P., Gao, S., and Wang, K. (2012). miR-301a promotes pancreatic cancer cell proliferation by directly inhibiting Bim expression.

Journal of cellular biochemistry 113, 3229-3235.

Donada, M., Bonin, S., Barbazza, R., Pettirosso, D., & Stanta, G. (2013). Management of stage II colon cancer - the use of molecular biomarkers for adjuvant therapy decision.

BMC Gastroenterol, 13, 36.

Georges, R. B., Adwan, H., Hamdi, H., Hielscher, T., Linnemann, U., & Berger, M. R.

(2011). The insulin-like growth factor binding proteins 3 and 7 are associated with colorectal cancer and liver metastasis. Cancer Biol Ther, 12(1), 69-79.

Hongo, K., Tsuno, N. H., Kawai, K., Sasaki, K., Kaneko, M., Hiyoshi, M., Murono, K., Tada,

N., Nirei, T., Sunami, E., et al. (2013). Hypoxia enhances colon cancer migration and invasion through promotion of epithelial-mesenchymal transition. The Journal of surgical research 182, 75-84.

Jusakul, A., Yongvanit, P., Loilome, W., Namwat, N., & Kuver, R. (2011). Mechanisms of oxysterol-induced carcinogenesis. Lipids Health Dis, 10, 44.

Kyrgiou, M., Salanti, G., Pavlidis, N., Paraskevaidis, E., & Ioannidis, J. P. (2006). Survival benefits with diverse chemotherapy regimens for ovarian cancer: meta-analysis of multiple treatments. J Natl Cancer Inst, 98(22), 1655-1663.

Nassirpour, R., Shao, L., Flanagan, P., Abrams, T., Jallal, B., Smeal, T., and Yin, M. J.

(2010). Nek6 mediates human cancer cell transformation and is a potential cancer therapeutic target. Mol Cancer Res 8, 717-728.

Semenza, G. L. (2013). HIF-1 mediates metabolic responses to intratumoral hypoxia and oncogenic mutations. J Clin Invest 123, 3664-3671. Tang, M. K., Zhou, H. Y., Yam, J. W., & Wong, A. S. (2010). c-Met overexpression contributes to the acquired apoptotic resistance of nonadherent ovarian cancer cells through a cross talk mediated by phosphatidylinositol 3-kinase and extracellular signal-regulated kinase 1/2. Neoplasia, 12(2), 128-138.

Claims

WHAT IS CLAIMED IS:

1. An in vitro method of characterizing a diseased tissue sample comprising:

a) quantifying the expression of two or more markers in a plurality of cells in the tissue sample;

b) determining a correlation coefficient between the expression the two or more markers in the plurality of cells and calculating a confidence interval over said plurality of cells in the sample; and

c) comparing the calculated confidence interval to a reference value to characterize the diseased tissue.

2. The method of claim 1, further comprising obtaining the tissue sample from a subject.

3. The method of claim 1, further comprising obtaining the tissue sample from a third party.

4. The method of claim 1, further defined as a method of characterizing a cancer wherein the tissue sample is a cancer tissue sample.

5. The method of claim 1, wherein quantifying the expression of said markers comprises quantifying protein expression.

6. The method of claim 1 , wherein quantifying the expression of said markers comprises quantifying RNA expression.

7. The method of claim 1, wherein quantifying the expression of said markers comprises performing in situ hybridization with an antibody or a nucleic acid probe.

8. The method of claim 5, wherein protein expression is quantified by measuring antibody binding.

9. The method of claim 8, wherein measuring antibody binding comprises performing an immunohistochemistry assay.

10. The method of claim 1, wherein the cancer tissue sample is a solid tissue section.

11. The method of claim 10, wherein the plurality of cells are comprised in a plurality of tissue sections.

12. The method of claim 4, wherein the cancer is a solid tumor.

13. The method of claim 12, wherein the plurality of cells are from discrete parts of the tumor.

14. The method of claim 4, further comprising identifying the cancer cells in the sample, prior to the quantifying step.

15. The method of claim 14, wherein expression of said two or more markers is quantified in the identified cancer cells.

16. The method of claim 1, wherein at least one of the markers is a protein.

17. The method of claim 1, wherein at least one of the markers is a RNA.

18. The method of claim 17, wherein at least one of the markers is mRNA or miRNA.

19. The method of claim 16, wherein the protein is a cell surface or secreted protein.

20. The method of claim 1, wherein said two or more markers are comprised in the same metabolic pathway.

21. The method of claim 20, wherein said two or more markers comprise a receptor and a ligand for the receptor.

22. The method of claim 20, wherein said two or more markers comprise a transcription factor and the product of a gene regulated by the transcription factor.

23. The method of claim 20, wherein said two or more markers comprise a miRNA and the product of a gene targeted by the miRNA.

24. The method of claim 1, wherein determining a correlation coefficient comprises performing a Pearson correlation analysis.

25. The method of claim 1, wherein determining the confidence interval comprises performing a calculation according to the formula: COY E[(X - μ_χ ) (Y - μγ)^'

Ρχ.γ = eorr( Y) =

( χ V γ ί7χ(7γ

26. The method of claim 4, wherein the reference value is a confidence interval from a non-cancer sample.

27. The method of claim 4, wherein the reference value is a confidence interval from a cancer that is known to respond to a given therapy.

28. The method of claim 4, wherein characterizing the cancer comprises determining whether the cancer is predicted to respond to a given therapy.

29. The method of claim 4, wherein the two or more markers comprise HGF or c-Met expression.

30. The method of claim 29, wherein the two or more markers comprise HGF and c-Met expression.

31. The method of claim 29, wherein the cancer is an ovarian cancer.

32. The method of claim 29, wherein characterizing the cancer comprises determining whether the cancer is predicted to respond to a therapy targeting the c-Met pathway.

33. The method of claim 4, wherein the two or more markers comprise OSBPL3 or IGFBP3 expression.

34. The method of claim 33, wherein the two or more markers comprise OSBPL3 and IGFBP3 expression.

35. The method of claim 33, wherein the cancer is colorectal cancer.

36. The method of claim 4, wherein the two or more markers comprise HIF1A or NEK6 expression.

37. The method of claim 36, wherein the two or more markers comprise HIF1A and NEK6 expression.

38. The method of claim 36, wherein the cancer is ovarian cancer.

39. The method of claim 4, wherein the two or more markers comprise miR-301a-5p or LBP expression.

40. The method of claim 39, wherein the two or more markers comprise miR-301a-5p and LBP expression.

41. The method of claim 39, wherein the cancer is colorectal cancer.

42. The method of claim 1, wherein steps (a), (b) and/or (c) is automated.

43. The method of claim 42, wherein steps (a)-(c) are automated.

44. The method of claim 1, wherein step (a) comprises obtaining a digital image of the sample and performing quantitative analysis of the image.

45. The method of claim 4, wherein characterizing the cancer comprises determining the aggressiveness of the cancer.

46. The method of claim 1, wherein the said two or more markers comprise the same antigen having different localization in the sample.

47. The method of claim 46, wherein the different localization comprises localization in different tissue types, different cell types, or different cellular compartments.

48. The method of claim 47, wherein the different localization comprises localization in the nucleus versus the cytoplasm.

49. The method of claim 4, wherein the two or more markers comprise nuclear c-Met or cytoplasmic c-Met expression.

50. The method of claim 49, wherein the two or more markers comprise nuclear c-Met and cytoplasmic c-Met expression.

51. A method of selecting a candidate anti-cancer therapy for a subject comprising:

a) obtaining a cancer tissue sample from a subject;

b) quantifying the expression of two or more markers in a plurality of cells in the sample; c) determining a correlation coefficient between the expression the two or more markers in the plurality of cells and calculating a confidence interval over said plurality of cells in the sample; and

d) comparing the calculated confidence interval to a reference value to select a candidate anti-cancer therapy for the patient.

52. The method of claim 51, wherein the reference value corresponds to a correlation coefficient for a patient that does not respond to the candidate anti-cancer therapy.

53. The method of claim 51, wherein the reference value corresponds to a correlation coefficient for a patient with a favorable response to the candidate anti-cancer therapy.

54. A method of treating a cancer patient comprising:

a) selecting a candidate anticancer therapy in accordance with claim 51 ; and b) administering the selected therapy to the patient.

55. A tangible computer-readable medium comprising computer-readable code that, when executed by a computer, causes the computer to perform operations comprising: a) receiving information corresponding to an expression level of two or more markers in a plurality of cells in a tissue sample from a patient; and

b) determining a correlation coefficient between the expression of the two or more markers in the plurality of cells and calculating a confidence interval over said plurality of cells in the sample.

56. The media of claim 55, further comprising computer-readable code that, when executed by a computer, causes the computer to perform operations comprising:

c) comparing the calculated confidence interval to a reference value to characterize the tissue sample.

57. The method of claim 55, wherein the tissue sample is a cancer tissue sample.

58. The media of claim 57, further comprising computer-readable code that, when executed by a computer, causes the computer to perform operations comprising:

c) providing at least a first candidate agent predicted to be effective for inhibiting cancer cells in the patient based on the calculated confidence interval.

59. The media of claim 55, wherein the receiving information comprises receiving from a tangible data storage device information corresponding to an expression level of two or more markers in a plurality of cells in a sample from the patient.

60. The media of claim 59, wherein receiving information comprises receiving a digitized image of the sample from the cancer patient.

61. The media of claim 55, further comprising computer-readable code that, when executed by a computer, causes the computer to perform one or more additional operations comprising: sending information corresponding to the calculated confidence interval to a tangible data storage device.

62. The media of claim 58, further comprising computer-readable code that, when executed by a computer, causes the computer to perform one or more additional operations comprising: sending information corresponding to at least a first candidate agent predicted to be effective to inhibiting cancer cells in the patient to a tangible data storage device.