WO2018098241A1 - Methods of assessing risk of recurrent prostate cancer - Google Patents

Abstract

Described herein are compositions and methods relating to biomarkers used to assess the risk of recurrent prostate cancer in a subject. The biomarkers can be used to establish or effect treatment regimens based upon the assessed risk for recurrent prostate cancer.

Description

TITLE OF THE INVENTION

METHODS OF ASSESSING RISK OF RECURRENT PROSTATE CANCER

CROSS-REFERENCE TO RELATED APPLICATIONS

This application claims the benefit of priority under 35 U.S.C. § 119(e) to U.S. Provisional Patent Application No. 62/425,348 filed November 22, 2016, the contents of which are incorporated by reference herein in their entirety.

STATEMENT REGARDING FEDERALLY SPONSORED RESEARCH OR

DEVELOPMENT

This invention was made with government support under CA120317, CA138249, CA184687, HG006853, CA077739, CA159981, and CA016056 awarded by the National Institute of Health. The government has certain rights in the invention.

BACKGROUND OF THE INVENTION

Prostate cancer is the most common non-dermatological cancer among men in the United States, with an estimated 220,000 new diagnoses in 2015 (Siegel et al., 2015, CA Cancer J Clin, 65: 5-29). Prostate cancer is often indolent and may not require immediate treatment upon diagnosis. On the other hand, prostate cancer can adopt a locally aggressive and rapidly metastatic phenotype that is fatal without intervention. The aggressiveness of prostate cancer can be assessed via clinical staging, levels of prostate- specific antigen (PSA) and the Gleason score, a histological assessment of tumor architecture. Patients with intermediate or high-risk prostate cancer are often treated by radical prostatectomy, an invasive surgical procedure that removes the prostate in its entirety. Radical prostatectomy is potentially curative, but approximately 33% of patients will experience biochemical persistence or recurrence, as defined by a non-zero serum PSA level (National Comprehensive Cancer Network; Ward et al., 2013, J Urol, 170(5): 1872-1876). These patients may receive salvage radiotherapy, which has demonstrated modest benefits to survival (Boorjian et al., 2009, J Urol., 182(6): 2708-2714). Salvage radiotherapy substantially decreases the rate of local recurrence, but many of these patients will develop metastatic disease (Moul et al., 2013, Rising serum PSA following local therapy for prostate cancer: Definition, natural history, and risk stratification. B. D.S. (Ed.)). Adjuvant radiotherapy has been found to significantly decrease the rate of biochemical recurrence and increase cancer-free survival (Bolla et al., 2012, Lancet., 380(9858): 2018-2027). However, to prevent over-treatment, adjuvant radiotherapy is typically reserved for patients who have diffusely positive surgical margins or tumor invasion of the seminal vesicles (Ward et al., 2013, Prostate cancer: Pathologic stage T3 disease, positive surgical margins, or microscopic lymph node involvement following radical prostatectomy. B. D.S. (Ed.)). Predictive methods for identifying patients likely to develop recurrent disease are critical for selecting proper treatment.

Most prediction models for prostate cancer recurrence are based on clinical features alone. Gene expression signatures predictive of patient outcomes are in clinical use for many other cancers, and therefore prediction models for prostate cancer may be improved by the inclusion of a molecular component. Previous efforts made to identify genes predictive of prostate cancer recurrence have utilized microarray data, which were comprehensive, but had limited reproducibility (Cheville et al., 2008, J Clin Oncol., 26(24): 3930-3936; Cooperberg et al., 2013, J Clin Oncol., 31(11): 1428-1434; Donovan et al., 2008, J Clin Oncol., 26(24): 3923-3929; Long et al., 2011, Am J Pathol., 179(1): 46-54; Michiels et al., 2005, Lancet, 365(9458): 488-492; Nakagawa et al., 2008, PLoS One, 3(5): e2318; Sun et al., 2009, Prostate, 69(10): 1119-1127).

Thus, there is a need in the art for improved methods for the detection of risk for prostate cancer recurrence. The present invention satisfies this unmet need.

SUMMARY OF THE INVENTION

In one aspect, the present invention provides a method of assessing the risk of recurrence of prostate cancer in a subject. The method comprises detecting the level of at least one biomarker in a biological sample obtained from the subject, wherein the at least one biomarker is selected from the group consisting of HBEGF, HOXC13, IGFBP2, and SATBl; comparing the level of the at least one biomarker in the biological sample to a control level of the at least one biomarker; and determining that the subject is at increased risk for recurrence of prostate cancer when the at least one biomarker is differentially expressed in the biological sample as compared to the control level. In one embodiment, the biological sample comprises prostate tissue of the subject. In one embodiment, the prostate tissue is obtained during surgical resection of at least a portion of the prostate of the subject.

In one embodiment, the at least one biomarker comprises two or more of the group consisting of HBEGF, HOXC13, IGFBP2, and SATB1. In one embodiment, the at least one biomarker comprises three or more of the group consisting of HBEGF, HOXC13, IGFBP2, and SATB1. In one embodiment, the at least one biomarker comprises HBEGF, HOXC13, IGFBP2, and SATB 1.

In one embodiment, the subject is determined to be at increased risk for recurrence of prostate cancer when HOXC13 in the biological sample is increased as compared to the control level. In one embodiment, the subject is determined to be at increased risk for recurrence of prostate cancer when HOXC13 in the biological sample is increased by greater than about 6.5 fold as compared to the control level.

In one embodiment, the subject is determined to be at increased risk for recurrence of prostate cancer when at least one biomarker selected from the group consisting of HBEGF, IGFBP2, and SATB 1 in the biological sample is decreased as compared to the control level. In one embodiment, the subject is determined to be at increased risk for recurrence of prostate cancer when HBEGF in the biological sample is decreased by greater than 2 fold as compared to the control level. In one embodiment, the subject is determined to be at increased risk for recurrence of prostate cancer when IGFBP2 in the biological sample is decreased by greater than 1.2 fold as compared to the control level. In one embodiment, the subject is determined to be at increased risk for recurrence of prostate cancer when SATB 1 in the biological sample is decreased by greater than 3 fold as compared to the control level.

In one embodiment, the method comprises the use of a multi-dimensional non-linear algorithm to determine if the at least one biomarker is differentially expressed.

In one embodiment, the at least one biomarker is an RNA biomarker. In one embodiment, the at least one biomarker is a protein biomarker.

In one embodiment, the method further comprises using information from the surgical resection of at least a portion of the prostate of the subject. In one

embodiment, the information comprises at least one selected from the group consisting of primary tumor staging, lymph node staging, and metastatic staging of a primary tumor of the subject.

In one embodiment, the method further comprises effectuating a treatment of the subject. In one embodiment the method comprises administering adjuvant radiotherapy to the subject determined to be at risk for recurrence of prostate cancer.

BRIEF DESCRIPTION OF THE DRAWINGS

The following detailed description of preferred embodiments of the invention will be better understood when read in conjunction with the appended drawings. For the purpose of illustrating the invention, there are shown in the drawings embodiments which are presently preferred. It should be understood, however, that the invention is not limited to the precise arrangements and instrumentalities of the embodiments shown in the drawings.

Figure 1 depicts the hierarchical clustering of samples on highly significantly differentially expressed cooperation response genes (CRGs). Genes were selected via two-tailed t-test between prostate cancer and benign tissue specimens with p- value < .01. Gene expression values were determined following normalization of qPCR data and imputation of missing values using the R package "nondetects".

Figure 2, comprising Figure 2A through Figure 2C, depicts the results of experiments determining the ideal prediction conditions using the training set. Figure 2A: Prediction accuracy (% correct predictions) on the training set comparing imputed vs unimputed data. Predictions were made on either imputed or unimputed data via Leave- One-Out Cross Validation (LOOCV) using all three prediction algorithms (centroid, distance, and nearest neighbor) at multiple gene signature p-value cutoffs. The arithmetic mean for accuracy under all three algorithms was displayed for each p-value cutoff. Area under the curve of arithmetic means was computed. Predictions were significantly better across all p-value cutoffs and all algorithms tested using the imputed data. Figure 2B: Prediction accuracy on the training set comparing the three prediction algorithms.

Predictions were made using each algorithm via Leave-One-Out Cross Validation (LOOCV) using both imputed and unimputed data at multiple gene signature p-value cutoffs. The arithmetic mean for accuracy under both data-handling techniques was displayed for each p-value cutoff. Area under the curve of arithmetic means was computed. The centroid and distance algorithms were superior to the nearest neighbor algorithm. The centroid algorithm was most effective at reducing the impact of outliers. Figure 2C: Prediction accuracy on the training set comparing p-value cutoffs. Predictions were made using the optimal data handling strategy (normalization to BECN1 and imputation via R package "nondetects") and prediction algorithm (centroid) at multiple gene signature p-value cutoffs. Prediction accuracy was highest at p-value cutoff <0.01, which produced a 4-gene signature. This strategy gave 100% sensitivity for the training set.

Figure 3 depicts the results of experiments using test set prediction data.

Predictions were made using different prediction strategies. Strategy 1 (CRG): Use centroid prediction algorithm and 4-gene CRG signature comparing samples to training set data normalized to BECN1 and imputed. Strategy 2 (Gleason 1): Samples were predicted to be recurrent if Gleason score was 7-10 and non-recurrent otherwise. Strategy 3 (Gleason 2): Samples were predicted to be recurrent if Gleason score was 7 (4+3), or above, and non-recurrent if Gleason score was 7 (3+4) or below. Strategy 4 (Surgery): Samples were predicted to be non-recurrent if primary tumor stage was <T2c, lymph node stage was 0 and metastatic stage was 0 and recurrent otherwise. Strategy 5 (CRG + Surgery): Samples were predicted to be non -recurrent if primary tumor stage was <T2c, lymph node stage was 0 and metastatic stage was 0. Other samples were evaluated using strategy 1. Strategy 6 (CRG + Gleason): Samples were predicted to be recurrent if Gleason score was 9-10 and non-recurrent if Gleason grade was 2-6. Other samples were evaluated using strategy 1. Strategy 7 (Gleason 1 + Surgery): Samples were predicted to be non-recurrent if primary tumor stage was <T2c, lymph node stage was 0 and metastatic stage was 0. Other samples were evaluated using strategy 2. Strategy 8

(Gleason 2 + Surgery): Samples were predicted to be non-recurrent if primary tumor stage was <T2c, lymph node stage was 0 and metastatic stage was 0. Other samples were evaluated using strategy 3. Accuracy = % predictions correct. Sensitivity = (1 - %False Negatives). Specificity = (1 - %False Positives). PPV = (# Recurrent tumors/# Total Recurrent Predictions). NPV = (# Non-recurrent tumors/# Total Non-Recurrent Predictions). Bars from left to right: CRG, Gleason grade 1, Gleason grade 2, Surgery, CRG + Surgery, CRG + Gleason, Gleason 1 + Surgery, Gleason 2 + Surgery.

Figure 4 is a graph depicting ROC curves for predictions using the validation set. ROC curves were constructed with all samples considered (blue; AUC =0.669) and with those best handled by surgical predictions removed compared to predictions made by chance (yellow; AUC = 0.75). Arrows indicate the points at which recurrence score and nonrecurrence score were given equal weight and predictions were made.

Figure 5 is a graph depicting survival curves for predictions using the validation set. Kaplan-Meier survival curves were constructed for the subset of samples in the validation set that were predicted to recur (n = 15) and the subset of samples predicted not to recur (n = 8). Statistical analysis was performed using log-rank test and the R package "survdif '.

Figure 6 is a waterfall plot of CRG dysregulation in prostate cancer vs matched benign tissue. Two-tailed t-tests were performed on each CRG and t-statistics were plotted. 58 of 91 CRGs (64%) were significantly dysregulated in prostate cancer (p < .05). Of these, 37 CRGs were highly significantly dysregulated (p < .01).

Figure 7, comprising Figure 7A and Figure 7B, depicts data demonstrating the expression of stromal markers in validation set samples. Figure 7A: Non-normalized expression of smooth muscle alpha-actin (Acta), vimentin (vim), and beclin-1 (Becnl). Expression was quantified using RT-PCR. Figure 7B: Relative expression of smooth muscle alpha-actin and vimentin normalized to beclin-1.

Figure 8 depicts data demonstrating the expression of reference genes in validation set samples. Expression of reference genes beclin-1 (Becnl), Ras homolog gene family member A (RhoA), and 18S ribosomal RNA (18S) was quantified using RT- PCR. Pearson's correlational coefficients were calculated among the three genes using expression across all samples in the validation set.

DETAILED DESCRIPTION

The present invention provides compositions and methods relating to biomarkers that can be used for the diagnosing or assessing the risk of prostate cancer recurrence in a subject. The markers of the invention can be used to establish and evaluate treatment plans for a subject at risk for prostate cancer recurrence and for a subject not at risk for prostate cancer recurrence.

In certain embodiments, the method comprises examining relevant biomarkers and their expression. In one embodiment, biomarker expression includes transcription into messenger RNA (mRNA) and translation into protein. In certain embodiments, the method comprises determining if the expression levels of the relevant biomarkers are differentially expressed as compared to a control. In certain embodiments, the control may be at the level of the relevant biomarkers in a subject not having prostate cancer, a subject not at risk for prostate cancer recurrence, a population not having prostate cancer, or a population not having a risk for prostate cancer recurrence. In certain embodiments, the method comprises determining if the expression levels of the relevant biomarkers in a sample obtained from the subject are differentially expressed as compared to the expression levels of the relevant biomarkers in a subject or population where prostate cancer has not recurred. In certain embodiments, the method comprises using a non-linear prediction algorithm to assess whether a set of relevant biomarkers is differentially expressed as compared to a subject or population where prostate cancer has not recurred. In certain embodiments, the method comprises the use of the expression pattern of the relevant biomarkers in combination with surgical data or assessment of the primary tumor to assess the risk for prostate cancer recurrence.

In one embodiment, the invention provides a biomarker for assessing the risk for prostate cancer recurrence in a subject. In one embodiment, the biomarker for assessing the risk for prostate cancer recurrence is selected from the group of HBEGF, HOXC13, IGFBP2, and SATB1. In one embodiment, the invention provides a set or panel of biomarkers for assessing the risk for prostate cancer recurrence in a subject, wherein the set of biomarkers comprises two or more of HBEGF, HOXC13, IGFBP2, and SATB1.

The present invention is based in part on the discovery of a gene signature differentially expressed in prostate cancer that later recurred. Further, various predictive algorithms were evaluated using this signature. It was further observed that predictive power was increased when the algorithms were combined with a surgical nomogram. This was capable of predicting clinical outcomes of an independent blinded validation set with 83% accuracy, outperforming previous methods.

In some embodiments of the invention, the methods comprise a) providing a biological sample from the subject; b) analyzing the biological sample with an assay that specifically detects at least one biomarker of the invention in the biological sample; c) comparing the level of the at least one biomarker in the sample with the level in a control sample, wherein a statistically significant difference between the level of the at least one biomarker in the sample with the level in a control sample or earlier obtained biological sample is indicative of enhanced risk for prostate cancer recurrence in the subject. In some embodiments, the methods further comprise the step of d) effectuating a treatment regimen based thereon.

In one embodiment, the biomarker types comprise mRNA biomarkers. In various embodiments, the mRNA is detected by at least one of mass spectroscopy, PCR microarray, thermal sequencing, capillary array sequencing, solid phase sequencing, and the like.

In another embodiment, the biomarker types comprise polypeptide biomarkers. In various embodiments, the polypeptide is detected by at least one of ELISA, Western blot, flow cytometry, immunofluorescence, immunohistochemistry, mass spectroscopy, and the like.

Definitions

Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. Although any methods and materials similar or equivalent to those described herein can be used in the practice or testing of the present invention, the preferred methods and materials are described.

As used herein, each of the following terms has the meaning associated with it in this section.

The articles "a" and "an" are used herein to refer to one or to more than one (i.e., to at least one) of the grammatical object of the article. By way of example, "an element" means one element or more than one element. "About" as used herein when referring to a measurable value such as an amount, a temporal duration, and the like, is meant to encompass variations of ±20%, ±10%, ±5%), ±1%), or ±0.1%) from the specified value, as such variations are appropriate to perform the disclosed methods.

The term "abnormal" when used in the context of organisms, tissues, cells or components thereof, refers to those organisms, tissues, cells or components thereof that differ in at least one observable or detectable characteristic (e.g., age, treatment, time of day, etc.) from those organisms, tissues, cells or components thereof that display the "normal" (expected) respective characteristic. Characteristics which are normal or expected for one cell or tissue type, might be abnormal for a different cell or tissue type.

The term "amplification" refers to the operation by which the number of copies of a target nucleotide sequence present in a sample is multiplied.

The term "antibody," as used herein, refers to an immunoglobulin molecule which is able to specifically bind to a specific epitope on an antigen. Antibodies can be intact immunoglobulins derived from natural sources or from recombinant sources and can be immunoreactive portions of intact immunoglobulins. The antibodies in the present invention may exist in a variety of forms including, for example, polyclonal antibodies, monoclonal antibodies, intracellular antibodies ("intrabodies"), Fv, Fab and F(ab)2, as well as single chain antibodies (scFv), heavy chain antibodies, such as camelid antibodies, synthetic antibodies, chimeric antibodies, and humanized antibodies (Harlow et al., 1999, Using Antibodies: A Laboratory Manual, Cold Spring Harbor Laboratory Press, NY; Harlow et al., 1989, Antibodies: A Laboratory Manual, Cold Spring Harbor, New York; Houston et al., 1988, Proc. Natl. Acad. Sci. USA 85:5879-5883; Bird et al., 1988, Science 242:423-426).

An "antibody heavy chain," as used herein, refers to the larger of the two types of polypeptide chains present in all antibody molecules in their naturally occurring conformations.

An "antibody light chain," as used herein, refers to the smaller of the two types of polypeptide chains present in all antibody molecules in their naturally occurring conformations, κ and λ light chains refer to the two major antibody light chain isotypes. By the term "synthetic antibody" as used herein, is meant an antibody which is generated using recombinant DNA technology, such as, for example, an antibody expressed by a bacteriophage as described herein. The term should also be construed to mean an antibody which has been generated by the synthesis of a DNA molecule encoding the antibody and which DNA molecule expresses an antibody protein, or an amino acid sequence specifying the antibody, wherein the DNA or amino acid sequence has been obtained using synthetic DNA or amino acid sequence technology which is available and well known in the art.

As used herein, an "immunoassay" refers to any binding assay that uses an antibody capable of binding specifically to a target molecule to detect and quantify the target molecule.

By the term "specifically binds," as used herein with respect to an antibody, is meant an antibody which recognizes a specific antigen, but does not substantially recognize or bind other molecules in a sample. For example, an antibody that specifically binds to an antigen from one species may also bind to that antigen from one or more species. But, such cross-species reactivity does not itself alter the

classification of an antibody as specific. In another example, an antibody that specifically binds to an antigen may also bind to different allelic forms of the antigen. However, such cross reactivity does not itself alter the classification of an antibody as specific. In some instances, the terms "specific binding" or "specifically binding," can be used in reference to the interaction of an antibody, a protein, or a peptide with a second chemical species, to mean that the interaction is dependent upon the presence of a particular structure (e.g., an antigenic determinant or epitope) on the chemical species; for example, an antibody recognizes and binds to a specific protein structure rather than to proteins generally. If an antibody is specific for epitope "A", the presence of a molecule containing epitope A (or free, unlabeled A), in a reaction containing labeled "A" and the antibody, will reduce the amount of labeled A bound to the antibody.

The term "coding sequence," as used herein, means a sequence of a nucleic acid or its complement, or a part thereof, that can be transcribed and/or translated to produce the mRNA and/or the polypeptide or a fragment thereof. Coding sequences include exons in a genomic DNA or immature primary RNA transcripts, which are joined together by the cell's biochemical machinery to provide a mature mRNA. The anti-sense strand is the complement of such a nucleic acid, and the coding sequence can be deduced therefrom. In contrast, the term "non-coding sequence," as used herein, means a sequence of a nucleic acid or its complement, or a part thereof, that is not translated into amino acid in vivo, or where tRNA does not interact to place or attempt to place an amino acid. Non-coding sequences include both intron sequences in genomic DNA or immature primary RNA transcripts, and gene-associated sequences such as promoters, enhancers, silencers, and the like.

As used herein, the terms "complementary" or "complementarity" are used in reference to polynucleotides (i.e., a sequence of nucleotides) related by the base- pairing rules. For example, the sequence "A-G-T," is complementary to the sequence "T- C-A." Complementarity may be "partial," in which only some of the nucleic acids' bases are matched according to the base pairing rules. Or, there may be "complete" or "total" complementarity between the nucleic acids. The degree of complementarity between nucleic acid strands has significant effects on the efficiency and strength of hybridization between nucleic acid strands. This is of particular importance in amplification reactions, as well as detection methods that depend upon binding between nucleic acids.

As used herein, the term "diagnosis" refers to the determination of the presence of a disease or disorder. In some embodiments of the present invention, methods for making a diagnosis are provided which permit determination of the presence of a particular disease or disorder.

A "disease" is a state of health of an animal wherein the animal cannot maintain homeostasis, and wherein if the disease is not ameliorated then the animal's health continues to deteriorate. In contrast, a "disorder" in an animal is a state of health in which the animal is able to maintain homeostasis, but in which the animal's state of health is less favorable than it would be in the absence of the disorder. Left untreated, a disorder does not necessarily cause a further decrease in the animal's state of health.

"Encoding" refers to the inherent property of specific sequences of nucleotides in a polynucleotide, such as a gene, a cDNA, or an mRNA, to serve as templates for synthesis of other polymers and macromolecules in biological processes having either a defined sequence of nucleotides (i.e., rRNA, tRNA and mRNA) or a defined sequence of amino acids and the biological properties resulting therefrom. Thus, a gene encodes a protein if transcription and translation of mRNA corresponding to that gene produces the protein in a cell or other biological system. Both the coding strand, the nucleotide sequence of which is identical to the mRNA sequence and is usually provided in sequence listings, and the non-coding strand, used as the template for transcription of a gene or cDNA, can be referred to as encoding the protein or other product of that gene or cDNA.

As used herein, the term "hybridization" is used in reference to the pairing of complementary nucleic acids. Hybridization and the strength of hybridization (i.e., the strength of the association between the nucleic acids) is impacted by such factors as the degree of complementarity between the nucleic acids, stringency of the conditions involved, the T_m of the formed hybrid, and the G:C ratio within the nucleic acids. A single molecule that contains pairing of complementary nucleic acids within its structure is said to be "self-hybridized." A single DNA molecule with internal complementarity could assume a variety of secondary structures including loops, kinks or, for long stretches of base pairs, coils.

"Instructional material," as that term is used herein, includes a publication, a recording, a diagram, or any other medium of expression which can be used to communicate the usefulness of the nucleic acid, peptide, and/or compound of the invention in the kit for identifying, diagnosing or alleviating or treating the various diseases or disorders recited herein. Optionally, or alternately, the instructional material may describe one or more methods of identifying, diagnosing or alleviating the diseases or disorders in a cell or a tissue of a subject. The instructional material of the kit may, for example, be affixed to a container that contains one or more components of the invention or be shipped together with a container that contains the one or more components of the invention. Alternatively, the instructional material may be shipped separately from the container with the intention that the recipient uses the instructional material and the components cooperatively.

"Isolated" means altered or removed from the natural state. For example, a nucleic acid or a peptide naturally present in a living animal is not "isolated," but the same nucleic acid or peptide partially or completely separated from the coexisting materials of its natural state is "isolated." An isolated nucleic acid or protein can exist in substantially purified form, or can exist in a non-native environment such as, for example, a host cell.

The term "label" when used herein refers to a detectable compound or composition that is conjugated directly or indirectly to a probe to generate a "labeled" probe. The label may be detectable by itself (e.g. radioisotope labels or fluorescent labels) or, in the case of an enzymatic label, may catalyze chemical alteration of a substrate compound or composition that is detectable (e.g., avidin-biotin). In some instances, primers can be labeled to detect a PCR product.

The terms "microarray" and "array" refers broadly to "DNA microarrays,"

"DNA chip(s)," "protein microarrays" and "protein chip(s)" and encompasses all art- recognized solid supports, and all art-recognized methods for affixing nucleic acid, peptide, and polypeptide molecules thereto. Preferred arrays typically comprise a plurality of different nucleic acid or peptide probes that are coupled to a surface of a substrate in different, known locations. These arrays, also described as "microarrays" or colloquially "chips" have been generally described in the art, for example, U.S. Pat. Nos. 5, 143,854, 5,445,934, 5,744,305, 5,677,195, 5,800,992, 6,040,193, 5,424,186 and Fodor et al., 1991, Science, 251 :767-777, each of which is incorporated by reference in its entirety for all purposes. Arrays may generally be produced using a variety of techniques, such as mechanical synthesis methods or light directed synthesis methods that incorporate a combination of photolithographic methods and solid phase synthesis methods. Techniques for the synthesis of these arrays using mechanical synthesis methods are described in, e.g., U.S. Pat. Nos. 5,384,261, and 6,040,193, which are incorporated herein by reference in their entirety for all purposes. Although a planar array surface is preferred, the array may be fabricated on a surface of virtually any shape or even a multiplicity of surfaces. Arrays may be nucleic acids on beads, gels, polymeric surfaces, fibers such as fiber optics, glass or any other appropriate substrate. (See U.S. Pat. Nos. 5,770,358, 5,789,162, 5,708,153, 6,040,193 and 5,800,992, which are hereby incorporated by reference in their entirety for all purposes.) Arrays may be packaged in such a manner as to allow for diagnostic use or can be an all-inclusive device; e.g., U.S. Pat. Nos. 5,856, 174 and 5,922,591 incorporated in their entirety by reference for all purposes. Arrays are commercially available from, for example, Affymetrix (Santa Clara, Calif.) and Applied Biosystems (Foster City, Calif), and are directed to a variety of purposes, including genotyping, diagnostics, mutation analysis, marker expression, and gene expression monitoring for a variety of eukaryotic and prokaryotic organisms. The number of probes on a solid support may be varied by changing the size of the individual features. In one embodiment the feature size is 20 by 25 microns square, in other embodiments features may be, for example, 8 by 8, 5 by 5 or 3 by 3 microns square, resulting in about 2,600,000, 6,600,000 or 18,000,000 individual probe features.

Assays for amplification of the known sequence are also disclosed. For example primers for PCR may be designed to amplify regions of the sequence. For RNA, a first reverse transcriptase step may be used to generate double stranded DNA from the single stranded RNA. The array may be designed to detect sequences from an entire genome; or one or more regions of a genome, for example, selected regions of a genome such as those coding for a protein or RNA of interest; or a conserved region from multiple genomes; or multiple genomes, arrays and methods of genetic analysis using arrays is described in Cutler, et al., 2001, Genome Res. 11(11): 1913-1925 and

Warrington, et al., 2002, Hum Mutat 19:402-409 and in US Patent Pub No 20030124539, each of which is incorporated herein by reference in its entirety.

A "nucleic acid" refers to a polynucleotide and includes poly- ribonucleotides and poly-deoxyribonucleotides. Nucleic acids according to the present invention may include any polymer or oligomer of pyrimidine and purine bases, preferably cytosine, thymine, and uracil, and adenine and guanine, respectively. (See Albert L. Lehninger, Principles of Biochemistry, at 793-800 (Worth Pub. 1982) which is herein incorporated in its entirety for all purposes). Indeed, the present invention contemplates any deoxyribonucleotide, ribonucleotide or peptide nucleic acid component, and any chemical variants thereof, such as methylated, hydroxymethylated or

glucosylated forms of these bases, and the like. The polymers or oligomers may be heterogeneous or homogeneous in composition, and may be isolated from naturally occurring sources or may be artificially or synthetically produced. In addition, the nucleic acids may be DNA or RNA, or a mixture thereof, and may exist permanently or transitionally in single-stranded or double-stranded form, including homoduplex, heteroduplex, and hybrid states.

An "oligonucleotide" or "polynucleotide" is a nucleic acid ranging from at least 2, preferably at least 8, 15 or 25 nucleotides in length, but may be up to 50, 100, 1000, or 5000 nucleotides long or a compound that specifically hybridizes to a polynucleotide. Polynucleotides include sequences of deoxyribonucleic acid (DNA) or ribonucleic acid (RNA) or mimetics thereof which may be isolated from natural sources, recombinantly produced or artificially synthesized. A further example of a polynucleotide of the present invention may be a peptide nucleic acid (PNA). (See U. S. Pat. No.

6, 156,501 which is hereby incorporated by reference in its entirety.) The invention also encompasses situations in which there is a nontraditional base pairing such as Hoogsteen base pairing which has been identified in certain tRNA molecules and postulated to exist in a triple helix. "Polynucleotide" and "oligonucleotide" are used interchangeably in this disclosure. It will be understood that when a nucleotide sequence is represented herein by a DNA sequence (e.g., A, T, G, and C), this also includes the corresponding RNA sequence (e.g., A, U, G, C) in which "U" replaces "T".

The terms "patient," "subject," "individual," and the like are used interchangeably herein, and refer to any animal, or cells thereof whether in vitro or in situ, amenable to the methods described herein. In certain non-limiting embodiments, the patient, subject or individual is a human.

As used herein, the term "polymerase chain reaction" ("PCR") refers to the method of K. B. Mullis (U. S. Pat. Nos. 4,683, 195 4,683,202, and 4,965, 188, hereby incorporated by reference), which describe a method for increasing the concentration of a segment of a target sequence in a mixture of genomic DNA without cloning or purification. This process for amplifying the target sequence consists of introducing a large excess of two oligonucleotide primers to the DNA mixture containing the desired target sequence, followed by a precise sequence of thermal cycling in the presence of a DNA polymerase. The two primers are complementary to their respective strands of the double stranded target sequence. To effect amplification, the mixture is denatured and the primers then annealed to their complementary sequences within the target molecule. Following annealing, the primers are extended with a polymerase so as to form a new pair of complementary strands. The steps of denaturation, primer annealing and polymerase extension can be repeated many times (i.e., denaturation, annealing and extension constitute one "cycle"; there can be numerous "cycles") to obtain a high concentration of an amplified segment of the desired target sequence. The length of the amplified segment of the desired target sequence is determined by the relative positions of the primers with respect to each other, and therefore, this length is a controllable parameter. By virtue of the repeating aspect of the process, the method is referred to as the "polymerase chain reaction" (hereinafter "PCR"). Because the desired amplified segments of the target sequence become the predominant sequences (in terms of concentration) in the mixture, they are said to be "PCR amplified". As used herein, the terms "PCR product," "PCR fragment," "amplification product" or "amplicon" refer to the resultant mixture of compounds after two or more cycles of the PCR steps of denaturation, annealing and extension are complete. These terms encompass the case where there has been amplification of one or more segments of one or more target sequences.

As used herein, the term "probe" refers to an oligonucleotide (i.e., a sequence of nucleotides), whether occurring naturally as in a purified restriction digest or produced synthetically, recombinantly or by PCR amplification, that is capable of hybridizing to another oligonucleotide of interest. A probe may be single-stranded or double-stranded. Probes are useful in the detection, identification and isolation of particular gene sequences.

As used herein, the terms "peptide," "polypeptide," and "protein" are used interchangeably, and refer to a compound comprised of amino acid residues covalently linked by peptide bonds. A protein or peptide must contain at least two amino acids, and no limitation is placed on the maximum number of amino acids that can comprise a protein's or peptide's sequence. Polypeptides include any peptide or protein comprising two or more amino acids joined to each other by peptide bonds. As used herein, the term refers to both short chains, which also commonly are referred to in the art as peptides, oligopeptides and oligomers, for example, and to longer chains, which generally are referred to in the art as proteins, of which there are many types. "Polypeptides" include, for example, biologically active fragments, substantially homologous polypeptides, oligopeptides, homodimers, heterodimers, variants of polypeptides, modified

polypeptides, derivatives, analogs, fusion proteins, among others. The polypeptides include natural peptides, recombinant peptides, synthetic peptides, or a combination thereof.

As used herein, "polynucleotide" includes cDNA, RNA, DNA/RNA hybrid, antisense RNA, ribozyme, genomic DNA, synthetic forms, and mixed polymers, both sense and antisense strands, and may be chemically or biochemically modified to contain non-natural or derivatized, synthetic, or semi-synthetic nucleotide bases. Also, contemplated are alterations of a wild type or synthetic gene, including but not limited to deletion, insertion, substitution of one or more nucleotides, or fusion to other

polynucleotide sequences.

The term "primer" refers to an oligonucleotide capable of acting as a point of initiation of synthesis along a complementary strand when conditions are suitable for synthesis of a primer extension product. The synthesizing conditions include the presence of four different deoxyribonucleotide triphosphates and at least one polymerization- inducing agent such as reverse transcriptase or DNA polymerase. These are present in a suitable buffer, which may include constituents which are co-factors or which affect conditions such as pH and the like at various suitable temperatures. A primer is preferably a single strand sequence, such that amplification efficiency is optimized, but double stranded sequences can be utilized.

Ranges: throughout this disclosure, various aspects of the invention can be presented in a range format. It should be understood that the description in range format is merely for convenience and brevity and should not be construed as an inflexible limitation on the scope of the invention. Accordingly, the description of a range should be considered to have specifically disclosed all the possible subranges as well as individual numerical values within that range. For example, description of a range such as from 1 to 6 should be considered to have specifically disclosed subranges such as from 1 to 3, from 1 to 4, from 1 to 5, from 2 to 4, from 2 to 6, from 3 to 6 etc., as well as individual numbers within that range, for example, 1, 2, 2.7, 3, 4, 5, 5.3, and 6. This applies regardless of the breadth of the range. Description

The present invention relates to compositions and methods for assessing the risk of prostate cancer recurrence in a subject. In one embodiment, the invention provides a biomarker for assessing the risk of prostate cancer recurrence in a subject. In one embodiment, the biomarker is at least one of HBEGF, HOXC13, IGFBP2, and

SATB. In one embodiment, the invention provides a biomarker set or panel comprising two or more of HBEGF, HOXC13, IGFBP2, and SATB 1.

In one embodiment, the present invention provides a method for assessing the risk for prostate cancer recurrence in a subject, comprising detecting a differentially expression level of the biomarker in a biological sample obtained from the subject, as compared to the expression level in a control sample. In certain embodiments, the biological sample obtained from the subject comprises prostate tissue of the subject, including prostate tissue excised during biopsy or prostatectomy.

In one embodiment, a subject is identified as having an enhanced risk for prostate cancer recurrence when the expression level of HOXC13 is increased in the biological sample as compared to a control. In one embodiment, a subject is identified as having an enhanced risk for prostate cancer recurrence when the expression level of one or more of HBEGF, IGFBP2, and SATB 1 is decreased in the biological sample as compared to a control. In one embodiment, the method comprises using a multi- dimensional non-linear algorithm to determine if the expression level of a set of biomarkers in the biological sample is statistically different than the expression level in a control sample.

In certain embodiments, the method comprises using surgical data in combination with the detection of the relevant biomarkers described herein to assess the risk for prostate cancer recurrence. For example, in certain embodiments, the method comprises assessing the extent of primary tumor (T category), spread of cancer to the lymph node (N category), or spread of cancer to other parts of the body (metastatic stage) (M category). In certain embodiments, the method comprises an assessment of the Gleason score of tumor.

In certain embodiments, the method comprises effectuating a treatment of the subject, based upon the assessment of risk for prostate cancer recurrence. For example, in one embodiment, the subject is treated with radiotherapy when an enhanced risk of prostate cancer recurrence is identified. In one embodiment, the subject is not treated with radiotherapy when the subject is determined to not have an enhanced risk for prostate cancer recurrence.

Identifying a Marker or Biomarker

The invention includes methods for assessing the risk of prostate cancer recurrence by detecting differentially expressed biomarkers in a biological sample obtained from a subject. For example, in certain embodiments, the method comprises assessing the risk of prostate cancer recurrence by detecting differentially expressed biomarkers in a biological sample obtained from a subject having prostate cancer or being treated for prostate cancer. In one embodiment, the method comprises assessing the risk of prostate cancer recurrence by detecting differentially expressed biomarkers in prostate tissue excised from the subject during biopsy or prostatectomy.

The invention contemplates the detection of differentially expressed markers using nucleic acid microarray. The invention further contemplates using methods known to those skilled in the art to detect and to measure the level of differentially expressed marker expression products, such as RNA and protein, to measure the level of one or more differentially expressed marker expression products.

Methods of detecting or measuring gene expression may utilize methods that focus on cellular components (cellular examination), or methods that focus on examining extracellular components (fluid examination). Because gene expression involves the ordered production of a number of different molecules, a cellular or fluid examination may be used to detect or measure a variety of molecules including RNA, protein, and a number of molecules that may be modified as a result of the protein's function. Typical diagnostic methods focusing on nucleic acids include amplification techniques such as PCR and RT-PCR (including quantitative variants), and hybridization techniques such as in situ hybridization, microarrays, blots, and others. Typical diagnostic methods focusing on proteins include binding techniques such as ELISA, immunohistochemistry, microarray and functional techniques such as enzymatic assays. The genes identified as being differentially expressed may be assessed in a variety of nucleic acid detection assays to detect or quantify the expression level of a gene or multiple genes in a given sample. For example, traditional Northern blotting, nuclease protection, RT-PCR, microarray, and differential display methods may be used for detecting gene expression levels. Methods for assaying for mRNA include Northern blots, slot blots, dot blots, and hybridization to an ordered array of oligonucleotides. Any method for specifically and quantitatively measuring a specific protein or mRNA or DNA product can be used. However, methods and assays are most efficiently designed with array or chip hybridization-based methods for detecting the expression of a large number of genes. Any hybridization assay format may be used, including solution-based and solid support-based assay formats.

The protein products of the genes identified herein can also be assayed to determine the amount of expression. Methods for assaying for a protein include Western blot, immunoprecipitation, and radioimmunoassay. The proteins analyzed may be localized intracellularly (most commonly an application of immunohistochemistry) or extracellularly (most commonly an application of immunoassays such as ELISA).

Biological samples may be of any biological tissue or fluid. Frequently the sample will be a "clinical sample" which is a sample derived from a patient. The biological sample may contain any biological material suitable for detecting the desired biomarkers, and may comprise cellular and/or non-cellular material obtained from the individual. A biological sample can be obtained by appropriate methods, such as, by way of examples, blood draw, fluid draw, biopsy, or surgical resection. Examples of such samples include but are not limited to blood, lymph, urine, prostate fluid, semen, and biopsies. Samples that are liquid in nature are referred to herein as "bodily fluids." Body samples may be obtained from a patient by a variety of techniques including, for example, by scraping or swabbing an area or by using a needle to aspirate bodily fluids. Methods for collecting various body samples are well known in the art. Frequently, a sample will be a "clinical sample," i.e., a sample derived from a patient. Such samples include, but are not limited to, bodily fluids which may or may not contain cells, e.g., blood (e.g., whole blood, serum or plasma), urine, saliva, tissue or fine needle biopsy samples, tissue sample obtained during surgical resection, and archival samples with known diagnosis, treatment and/or outcome history. In certain embodiments, the biological sample comprises prostate tissue. In certain embodiments, the biological sample comprises prostate tissue of a subject having prostate cancer.

Control group samples may either be from a normal subject, samples from subjects with a known recurrence of prostate cancer, or samples from subjects with no known recurrence of prostate cancer. As described below, comparison of the expression patterns of the sample to be tested with those of the controls can be used to assess the risk for prostate cancer recurrence in the subject. In some instances, the control groups are only for the purposes of establishing initial cutoffs or thresholds for the assays of the invention. Therefore, in some instances, the systems and methods of the invention can assess the risk of prostate cancer recurrence without the need to compare with a control group.

Methods

The present invention provides methods for assessing the risk for prostate cancer recurrence in a subject. The present invention includes methods for identifying subjects who have an increased or enhanced risk for prostate cancer recurrence and subjects who do not have an enhanced risk for prostate cancer recurrence by detection of the biomarkers disclosed herein.

These biomarkers also are useful for monitoring subjects undergoing treatments and therapies for prostate cancer, subjects who have had prostate cancer, and subjects who are in remission. These biomarkers also are useful for selecting or modifying therapies and treatments that would be efficacious in subjects having prostate cancer, subjects who have had prostate cancer, and subjects who are in remission.

The invention provides improved methods for assessing the risk for prostate cancer recurrence. The risk for prostate cancer recurrence can be assessed by measuring one or more of the biomarkers described herein, and comparing the measured values to comparator values, reference values, or index values. Such a comparison can be undertaken with mathematical algorithms or formula in order to combine information from results of multiple individual biomarkers and other parameters into a single measurement or index. For example, in certain embodiments, the comparison is undertaken with a multi-dimensional non-linear algorithm, as described elsewhere herein.

Subjects identified as having an enhanced risk for recurrence can optionally be selected to receive treatment regimens, such as radiotherapy or

administration of therapeutic compounds to prevent, treat or delay the recurrence of prostate cancer.

Identifying a subject as having an enhanced risk for prostate cancer recurrence after prostatectomy allows for the selection and initiation of various therapeutic interventions or treatment regimens in order to delay, reduce or prevent recurrence in those at risk. Further, identifying a subject with a low risk, or those who do not have an enhanced risk, for prostate cancer recurrence allows for the sparing of unneeded additional therapy administered to the subject.

Monitoring the levels of at least one biomarker also allows for the course of treatment to be monitored. For example, a sample can be provided from a subject undergoing treatment regimens or therapeutic interventions. Such treatment regimens or therapeutic interventions can include surgery, radiation, chemotherapy, and the like.

The biomarkers of the present invention can thus be used to generate a biomarker profile or signature of the subjects: (i) who have an increased risk for prostate cancer recurrence, (ii) who do not have an increased risk for prostate cancer recurrence, and/or (iii) who have a low risk for prostate cancer recurrence. The biomarker profile of a subject can be compared to a predetermined or comparator biomarker profile or reference biomarker profile to assess the risk for prostate cancer recurrence. Data concerning the biomarkers of the present invention can also be combined or correlated with other data or test results, such as, without limitation, measurements of clinical parameters or other algorithms for prostate cancer recurrence. Other data includes age, ethnicity, PSA level, Gleason score, primary tumor staging, lymph node staging, metastasis staging, and other genomic data, specifically expression values of other gene signatures relevant to prostate cancer outcomes (including but not limited to BRCA1 and BRCA2), and the like. The data may also comprise subject information such as medical history and any relevant family history. The present invention also provides methods for identifying agents for treating prostate cancer that are appropriate or otherwise customized for a specific subject. In this regard, a test sample from a subject, exposed to a therapeutic agent or a drug, can be taken and the level of one or more biomarkers can be determined. The level of one or more biomarkers can be compared to a sample derived from the subject before and after treatment, or can be compared to samples derived from one or more subjects who have shown improvements in risk factors as a result of such treatment or exposure.

In one embodiment, the invention provides at least one biomarker for the assessment of risk for prostate cancer recurrence. In one embodiment, the biomarker is at least one of HBEGF, HOXC13, IGFBP2, and SATB. In one embodiment, the invention provides a biomarker set or panel comprising two or more of HBEGF, HOXC13, IGFBP2, and SATB1. In one embodiment, the invention provides a biomarker set or panel comprising HBEGF, HOXC13, IGFBP2, and SATB 1.

In one embodiment, a subject is identified as having an enhanced risk for prostate cancer recurrence when the expression level of HOXC13 is increased in the biological sample as compared to a control.

In various embodiments of the methods of the invention, the level of HOXC13 is determined to be increased when the level of HOXC13 in the biological sample is increased by at least 10%, by at least 20%, by at least 30%, by at least 40%, by at least 50%, by at least 60%, by at least 70%, by at least 80%, by at least 90%, by at least 100%, by at least 125%, by at least 150%, by at least 175%, by at least 200%, by at least 250%, by at least 300%, by at least 400%, by at least 500%, by at least 600%, by at least 700%, by at least 800%, by at least 900%, by at least 1000%, by at least 1500%, by at least 2000%, by at least 2500%, by at least 3000%, by at least 4000%, or by at least 5000%), when compared with a comparator.

In various embodiments of the methods of the invention, the level of HOXC13 is determined to be increased when the level of HOXC13 in the biological sample is increased by at least 1 fold, at least 1.1 fold, at least 1.2 fold, at least 1.3 fold, at least 1.4 fold, at least 1.5 fold, at least 1.6 fold, at least 1.7 fold, at least 1.8 fold, at least 1.9 fold, at least 2 fold, at least 2.1 fold, at least 2.2 fold, at least 2.3 fold, at least 2.4 fold, at least 2.5 fold, at least 2.6 fold, at least 2.7 fold, at least 2.8 fold, at least 2.9 fold, at least 3 fold, at least 3.5 fold, at least 4 fold, at least 4.5 fold, at least 5 fold, at least 5.5 fold, at least 6 fold, at least 6.5 fold, at least 7 fold, at least 7.5 fold, at least 8 fold, at least 8.5 fold, at least 9 fold, at least 9.5 fold, at least 10 fold, at least 11 fold, at least 12 fold, at least 13 fold, at least 14 fold, at least 15 fold, at least 20 fold, at least 25 fold, at least 30 fold, at least 40 fold, at least 50 fold, at least 75 fold, at least 100 fold, at least 200 fold, at least 250 fold, at least 500 fold, or at least 1000 fold, when compared with a comparator.

In one embodiment, a subject is identified as having an enhanced risk for prostate cancer recurrence when the expression level of one or more of HBEGF, IGFBP2, and SATB 1 is decreased in the biological sample as compared to a control.

In various embodiments of the methods of the invention, the level of one or more of HBEGF, IGFBP2, and SATB 1 is determined to be decreased when the level of one or more of HBEGF, IGFBP2, and SATB 1 in the biological sample is decreased by at least 10%, by at least 20%, by at least 30%, by at least 40%, by at least 50%, by at least 60%, by at least 70%, by at least 80%, by at least 90%, by at least 100%, by at least

125%, by at least 150%, by at least 175%, by at least 200%, by at least 250%, by at least 300%, by at least 400%, by at least 500%, by at least 600%, by at least 700%, by at least 800%, by at least 900%, by at least 1000%, by at least 1500%, by at least 2000%, by at least 2500%, by at least 3000%, by at least 4000%, or by at least 5000%, when compared with a comparator.

In various embodiments of the methods of the invention, the level of one or more of HBEGF, IGFBP2, and SATB 1 is determined to be decreased when the level of one or more of HBEGF, IGFBP2, and SATB 1 in the biological sample is determined to be decreased by at least 1 fold, at least 1.1 fold, at least 1.2 fold, at least 1.3 fold, at least 1.4 fold, at least 1.5 fold, at least 1.6 fold, at least 1.7 fold, at least 1.8 fold, at least 1.9 fold, at least 2 fold, at least 2.1 fold, at least 2.2 fold, at least 2.3 fold, at least 2.4 fold, at least 2.5 fold, at least 2.6 fold, at least 2.7 fold, at least 2.8 fold, at least 2.9 fold, at least 3 fold, at least 3.5 fold, at least 4 fold, at least 4.5 fold, at least 5 fold, at least 5.5 fold, at least 6 fold, at least 6.5 fold, at least 7 fold, at least 7.5 fold, at least 8 fold, at least 8.5 fold, at least 9 fold, at least 9.5 fold, at least 10 fold, at least 11 fold, at least 12 fold, at least 13 fold, at least 14 fold, at least 15 fold, at least 20 fold, at least 25 fold, at least 30 fold, at least 40 fold, at least 50 fold, at least 75 fold, at least 100 fold, at least 200 fold, at least 250 fold, at least 500 fold, or at least 1000 fold, when compared with a comparator.

In one embodiment, a subject is identified as having an enhanced risk for prostate cancer recurrence when the expression level HBEGF is decreased in the biological sample as compared to a control. For example, in one embodiment, a subject is identified as having an enhanced risk for prostate cancer recurrence when the expression level FIBEGF is decreased by at least 1 fold, at least 1.2 fold, at least 1.4 fold, at least 1.6 fold, at least 1.8 fold, at least 2 fold, at least 2.2 fold, or at least 2.4 fold.

In one embodiment, a subject is identified as having an enhanced risk for prostate cancer recurrence when the expression level IGFBP2 is decreased in the biological sample as compared to a control. For example, in one embodiment, a subject is identified as having an enhanced risk for prostate cancer recurrence when the expression level IGFBP2 is decreased by at least 1 fold, at least 1.1 fold, at least 1.2 fold, at least 1.3 fold, at least 1.4 fold, or at least 1.5 fold.

In one embodiment, a subject is identified as having an enhanced risk for prostate cancer recurrence when the expression level SATB 1 is decreased in the biological sample as compared to a control. For example, in one embodiment, a subject is identified as having an enhanced risk for prostate cancer recurrence when the expression level SATB 1 is decreased by at least 1 fold, at least 1.5 fold, at least 2 fold, at least 2.5 fold, at least 3 fold, at least 3.1 fold, or at least 3.2 fold.

In one embodiment, the method comprises using a multi-dimensional nonlinear algorithm to determine if the expression level of a set of biomarkers in the biological sample is statistically different than the expression level in a control sample. In various embodiments, the algorithm is drawn from the group consisting essentially of: linear or nonlinear regression algorithms; linear or nonlinear classification algorithms; ANOVA; neural network algorithms; genetic algorithms; support vector machines algorithms; hierarchical analysis or clustering algorithms; hierarchical algorithms using decision trees; kernel based machine algorithms such as kernel partial least squares algorithms, kernel matching pursuit algorithms, kernel fisher discriminate analysis algorithms, or kernel principal components analysis algorithms; Bayesian probability function algorithms; Markov Blanket algorithms; a plurality of algorithms arranged in a committee network; and forward floating search or backward floating search algorithms.

In one embodiment, the method comprises detecting one or more markers in a biological sample of the subject. In various embodiments, the level of one or more of markers of the invention in the biological test sample of the subject is compared with the level of the biomarker in a comparator. Non-limiting examples of comparators include, but are not limited to, a negative control, a positive control, standard control, standard value, an expected normal background value of the subject, a historical normal background value of the subject, a reference standard, a reference level, an expected normal background value of a population that the subject is a member of, or a historical normal background value of a population that the subject is a member of. In one embodiment, the comparator is a level of the one or more biomarker in a sample obtained from a subject not having prostate cancer. In one embodiment, the comparator is a level of the one or more biomarker in a sample obtained from a subject known not to have recurrence of prostate cancer.

In various embodiments, the subject is a human subject, and may be of any race, sex and age.

Information obtained from the methods of the invention described herein can be used alone, or in combination with other information (e.g., age, family history, disease status, disease history, vital signs, blood chemistry, PSA level, Gleason score, primary tumor staging, lymph node staging, metastasis staging, expression of other gene signatures relevant to prostate cancer outcomes, etc.) from the subject or from the biological sample obtained from the subject.

In the methods of the invention, a biological sample from a subject is assessed for the level of one or more of the markers of the invention in the biological sample obtained from the patient. The level of one or more of the markers of the invention in the biological sample can be determined by assessing the amount of polypeptide of one or more of the biomarkers of the invention in the biological sample, the amount of mRNA of one or more of the biomarkers of the invention in the biological sample, the amount of enzymatic activity of one or more of the biomarkers of the invention in the biological sample, or a combination thereof. Detecting a biomarker

In one embodiment, the invention includes detecting one or more mRNA biomarkers, polypeptide biomarkers, or a combination thereof in a biological sample. Biomarkers generally can be measured and detected through a variety of assays, methods and detection systems known to one of skill in the art.

Various methods include but are not limited to immunoassays, microarray, PCR, RT-PCR, refractive index spectroscopy (RI), ultra-violet spectroscopy (UV), fluorescence analysis, electrochemical analysis, radiochemical analysis, near-infrared spectroscopy (near-IR), infrared (IR) spectroscopy, nuclear magnetic resonance spectroscopy ( MR), light scattering analysis (LS), mass spectrometry, pyrolysis mass spectrometry, nephelometry, dispersive Raman spectroscopy, gas chromatography, liquid chromatography, gas chromatography combined with mass spectrometry, liquid chromatography combined with mass spectrometry, matrix-assisted laser desorption ionizati on-time of flight (MALDI-TOF) combined with mass spectrometry, ion spray spectroscopy combined with mass spectrometry, capillary electrophoresis, colorimetry and surface plasmon resonance (such as according to systems provided by Biacore Life Sciences). See also PCT Publications WO/2004/056456 and WO/2004/088309. In this regard, biomarkers can be measured using the above-mentioned detection methods, or other methods known to the skilled artisan. Other biomarkers can be similarly detected using reagents that are specifically designed or tailored to detect them.

Different types of biomarkers and their measurements can be combined in the compositions and methods of the present invention. In various embodiments, the protein form of the biomarkers is measured. In various embodiments, the nucleic acid form of the biomarkers is measured. In exemplary embodiments, the nucleic acid form is mRNA. In various embodiments, measurements of protein biomarkers are used in conjunction with measurements of nucleic acid biomarkers.

In various embodiments of the invention, methods of measuring polypeptide levels in a biological sample obtained from a subject include, but are not limited to, an immunochromatography assay, an immunodot assay, a Luminex assay, an ELISA assay, an ELISPOT assay, a protein microarray assay, a ligand-receptor binding assay, displacement of a ligand from a receptor assay, displacement of a ligand from a shared receptor assay, an immunostaining assay, a Western blot assay, a mass spectrophotometry assay, a radioimmunoassay (RIA), a radioimmunodiffusion assay, a liquid chromatography -tandem mass spectrometry assay, an ouchterlony

immunodiffusion assay, reverse phase protein microarray, a rocket

Immunoelectrophoresis assay, an immunohistostaining assay, an immunoprecipitation assay, a complement fixation assay, FACS, an enzyme-substrate binding assay, an enzymatic assay, an enzymatic assay employing a detectable molecule, such as a chromophore, fluorophore, or radioactive substrate, a substrate binding assay employing such a substrate, a substrate displacement assay employing such a substrate, and a protein chip assay (see also, 2007, Van Emon, Immunoassay and Other Bioanalytical

Techniques, CRC Press; 2005, Wild, Immunoassay Handbook, Gulf Professional Publishing; 1996, Diamandis and Christopoulos, Immunoassay, Academic Press; 2005, Joos, Microarrays in Clinical Diagnosis, Humana Press; 2005, Hamdan and Righetti, Proteomics Today, John Wiley and Sons; 2007).

Methods for detecting a nucleic acid (e.g., mRNA), such as RT-PCR, real time PCR, microarray, branch DNA, NASBA and others, are well known in the art. Using sequence information provided by the database entries for the biomarker sequences, expression of the biomarker sequences can be detected (if present) and measured using techniques well known to one of ordinary skill in the art. For example, sequences in sequence database entries or sequences disclosed herein can be used to construct probes for detecting biomarker 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 biomarker 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. In addition to Northern blot and RT- PCR, RNA can also be measured using, for example, other target amplification methods (e.g., TMA, SDA, NASBA), signal amplification methods (e.g., bDNA), nuclease protection assays, in situ hybridization and the like.

In some embodiments, quantitative hybridization methods, such as Southern analysis, Northern analysis, or in situ hybridizations, can be used (see Current Protocols in Molecular Biology, Ausubel, F. et al., eds., John Wiley & Sons, including all supplements). A "nucleic acid probe," as used herein, can be a DNA probe or an RNA probe. The probe can be, for example, a gene, a gene fragment (e.g., one or more exons), a vector comprising the gene, a probe or primer, etc. For representative examples of use of nucleic acid probes, see, for example, U.S. Pat. Nos. 5,288,611 and 4,851,330. The nucleic acid probe can be, for example, a full-length nucleic acid molecule, or a portion thereof, such as an oligonucleotide of at least 15, 30, 50, 100, 250 or 500 nucleotides in length and sufficient to specifically hybridize under stringent conditions to appropriate target mRNA or cDNA. The hybridization sample is maintained under conditions which are sufficient to allow specific hybridization of the nucleic acid probe to mRNA or cDNA. Specific hybridization can be performed under high stringency conditions or moderate stringency conditions, as appropriate. In a preferred embodiment, the hybridization conditions for specific hybridization are high stringency. Specific hybridization, if present, is then detected using standard methods. If specific

hybridization occurs between the nucleic acid probe having a mRNA or cDNA in the test sample, the level of the mRNA or cDNA in the sample can be assessed. More than one nucleic acid probe can also be used concurrently in this method. Specific hybridization of any one of the nucleic acid probes is indicative of the presence of the mRNA or cDNA of interest, as described herein.

Alternatively, a peptide nucleic acid (PNA) probe can be used instead of a nucleic acid probe in the quantitative hybridization methods described herein. PNA is a DNA mimic having a peptide-like, inorganic backbone, such as N-(2-aminoethyl)glycine units, with an organic base (A, G, C, T or U) attached to the glycine nitrogen via a methylene carbonyl linker (see, for example, 1994, Nielsen et al., Bioconjugate

Chemistry 5: 1). The PNA probe can be designed to specifically hybridize to a target nucleic acid sequence. Hybridization of the PNA probe to a nucleic acid sequence is used to determine the level of the target nucleic acid in the biological sample. In another embodiment, arrays of oligonucleotide probes that are complementary to target nucleic acid sequences in the biological sample obtained from a subject can be used to determine the level of one or more biomarkers in the biological sample obtained from a subject. The array of oligonucleotide probes can be used to determine the level of one or more biomarkers alone, or the level of the one or more biomarkers in relation to the level of one or more other nucleic acids in the biological sample. Oligonucleotide arrays typically comprise a plurality of different oligonucleotide probes that are coupled to a surface of a substrate in different known locations. These oligonucleotide arrays, also known as "Genechips," have been generally described in the art, for example, U.S. Pat. No. 5,143,854 and PCT patent publication Nos. WO 90/15070 and 92/10092. These arrays can generally be produced using mechanical synthesis methods or light directed synthesis methods which incorporate a combination of photolithographic methods and solid phase oligonucleotide synthesis methods. See Fodor et al., Science, 251 :767-777 (1991), Pirrung et al., U.S. Pat. No. 5, 143,854 (see also PCT Application No. WO 90/15070) and Fodor et al., PCT Publication No. WO 92/10092 and U.S. Pat. No. 5,424,186. Techniques for the synthesis of these arrays using mechanical synthesis methods are described in, e.g., U.S. Pat. No. 5,384,261.

After an oligonucleotide array is prepared, a nucleic acid of interest is hybridized with the array and its level is quantified. Hybridization and quantification are generally carried out by methods described herein and also in, e.g., published PCT

Application Nos. WO 92/10092 and WO 95/11995, and U.S. Pat. No. 5,424, 186. In brief, a target nucleic acid sequence is amplified by well-known amplification techniques, e.g., PCR. Typically, this involves the use of primer sequences that are complementary to the target nucleic acid. Asymmetric PCR techniques may also be used. Amplified target, generally incorporating a label, is then hybridized with the array under appropriate conditions. Upon completion of hybridization and washing of the array, the array is scanned to determine the quantity of hybridized nucleic acid. The hybridization data obtained from the scan is typically in the form of fluorescence intensities as a function of quantity, or relative quantity, of the target nucleic acid in the biological sample. The target nucleic acid can be hybridized to the array in combination with one or more comparator controls (e.g., positive control, negative control, quantity control, etc.) to improve quantification of the target nucleic acid in the sample.

The probes and primers according to the invention can be labeled directly or indirectly with a radioactive or nonradioactive compound, by methods well known to those skilled in the art, in order to obtain a detectable and/or quantifiable signal; the labeling of the primers or of the probes according to the invention is carried out with radioactive elements or with nonradioactive molecules. Among the radioactive isotopes used, mention may be made of ³²P, ³³P, ³⁵S or ³H. The nonradioactive entities are selected from ligands such as biotin, avidin, streptavidin or digoxigenin, haptenes, dyes, and luminescent agents such as radioluminescent, chemoluminescent, bioluminescent, fluorescent or phosphorescent agents.

Nucleic acids can be obtained from the cells using known techniques. Nucleic acid herein refers to RNA, including mRNA, and DNA, including cDNA. The nucleic acid can be double-stranded or single-stranded (i.e., a sense or an antisense single strand) and can be complementary to a nucleic acid encoding a polypeptide. The nucleic acid content may also be an RNA or DNA extraction performed on a biological sample, including a biological fluid and fresh or fixed tissue sample.

There are many methods known in the art for the detection and quantification of specific nucleic acid sequences and new methods are continually reported. A great majority of the known specific nucleic acid detection and quantification methods utilize nucleic acid probes in specific hybridization reactions. Preferably, the detection of hybridization to the duplex form is a Southern blot technique. In the

Southern blot technique, a nucleic acid sample is separated in an agarose gel based on size (molecular weight) and affixed to a membrane, denatured, and exposed to (admixed with) the labeled nucleic acid probe under hybridizing conditions. If the labeled nucleic acid probe forms a hybrid with the nucleic acid on the blot, the label is bound to the membrane.

In the Southern blot, the nucleic acid probe is preferably labeled with a tag. That tag can be a radioactive isotope, a fluorescent dye or the other well-known materials. Another type of process for the specific detection of nucleic acids in a biological sample known in the art are the hybridization methods as exemplified by U.S. Pat. No. 6, 159,693 and No. 6,270,974, and related patents. To briefly summarize one of those methods, a nucleic acid probe of at least 10 nucleotides, preferably at least 15 nucleotides, more preferably at least 25 nucleotides, having a sequence complementary to a nucleic acid of interest is hybridized in a sample, subjected to depolymerizing conditions, and the sample is treated with an ATP/luciferase system, which will luminesce if the nucleic sequence is present. In quantitative Southern blotting, the level of the nucleic acid of interest can be compared with the level of a second nucleic acid of interest, and/or to one or more comparator control nucleic acids (e.g., positive control, negative control, quantity control, etc.).

Many methods useful for the detection and quantification of nucleic acid takes advantage of the polymerase chain reaction (PCR). The PCR process is well known in the art (U.S. Pat. No. 4,683,195, No. 4,683,202, and No. 4,800,159). To briefly summarize PCR, nucleic acid primers, complementary to opposite strands of a nucleic acid amplification target sequence, are permitted to anneal to the denatured sample. A DNA polymerase (typically heat stable) extends the DNA duplex from the hybridized primer. The process is repeated to amplify the nucleic acid target. If the nucleic acid primers do not hybridize to the sample, then there is no corresponding amplified PCR product. In this case, the PCR primer acts as a hybridization probe.

In PCR, the nucleic acid probe can be labeled with a tag as discussed elsewhere herein. Most preferably the detection of the duplex is done using at least one primer directed to the nucleic acid of interest. In yet another embodiment of PCR, the detection of the hybridized duplex comprises electrophoretic gel separation followed by dye-based visualization.

Typical hybridization and washing stringency conditions depend in part on the size (i.e., number of nucleotides in length) of the oligonucleotide probe, the base composition and monovalent and divalent cation concentrations (Ausubel et al., 1994, eds Current Protocols in Molecular Biology).

In one embodiment, the process for determining the quantitative and qualitative profile of the nucleic acid of interest according to the present invention is characterized in that the amplifications are real-time amplifications performed using a labeled probe, preferably a labeled hydrolysis-probe, capable of specifically hybridizing in stringent conditions with a segment of the nucleic acid of interest. The labeled probe is capable of emitting a detectable signal every time each amplification cycle occurs, allowing the signal obtained for each cycle to be measured.

The real-time amplification, such as real-time PCR, is well known in the art, and the various known techniques will be employed in the best way for the implementation of the present process. These techniques are performed using various categories of probes, such as hydrolysis probes, hybridization adjacent probes, or molecular beacons. The techniques employing hydrolysis probes or molecular beacons are based on the use of a fluorescence quencher/reporter system, and the hybridization adjacent probes are based on the use of fluorescence acceptor/donor molecules.

Hydrolysis probes with a fluorescence quencher/reporter system are available in the market, and are for example commercialized by the Applied Biosystems group (USA). Many fluorescent dyes may be employed, such as FAM dyes (6-carboxy- fluorescein), or any other dye phosphoramidite reagents.

Among the stringent conditions applied for any one of the hydrolysis- probes of the present invention is the Tm, which is in the range of about 65°C to 75°C. Preferably, the Tm for any one of the hydrolysis-probes of the present invention is in the range of about 67°C to about 70°C. Most preferably, the Tm applied for any one of the hydrolysis-probes of the present invention is about 67°C.

In one aspect, the invention includes a primer that is complementary to a nucleic acid of interest, and more particularly the primer includes 12 or more contiguous nucleotides substantially complementary to the nucleic acid of interest. Preferably, a primer featured in the invention includes a nucleotide sequence sufficiently

complementary to hybridize to a nucleic acid sequence of about 12 to 25 nucleotides. More preferably, the primer differs by no more than 1, 2, or 3 nucleotides from the target flanking nucleotide sequence In another aspect, the length of the primer can vary in length, preferably about 15 to 28 nucleotides in length (e.g., 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, or 27 nucleotides in length).

The concentration of the biomarker in a sample may be determined by any suitable assay. A suitable assay may include one or more of the following methods, an enzyme assay, an immunoassay, mass spectrometry, chromatography, electrophoresis or an antibody microarray, or any combination thereof. Thus, as would be understood by one skilled in the art, the system and methods of the invention may include any method known in the art to detect a biomarker in a sample.

The invention described herein also relates to methods for a multiplex analysis platform. In one embodiment, the method comprises an analytical method for multiplexing analytical measurements of markers.

Kits

The present invention also pertains to kits useful in the methods of the invention. Such kits comprise various combinations of components useful in any of the methods described elsewhere herein, including for example, materials for quantitatively analyzing a biomarker of the invention (e.g., polypeptide and/or nucleic acid), materials for assessing the activity of a biomarker of the invention (e.g., polypeptide and/or nucleic acid), and instructional material. For example, in one embodiment, the kit comprises components useful for the quantification of a desired nucleic acid in a biological sample. In another embodiment, the kit comprises components useful for the quantification of a desired polypeptide in a biological sample. In a further embodiment, the kit comprises components useful for the assessment of the activity (e.g., enzymatic activity, substrate binding activity, etc.) of a desired polypeptide in a biological sample.

In a further embodiment, the kit comprises the components of an assay for monitoring the effectiveness of a treatment administered to a subject in need thereof, containing instructional material and the components for determining whether the level of a biomarker of the invention in a biological sample obtained from the subject is modulated during or after administration of the treatment. In various embodiments, to determine whether the level of a biomarker of the invention is modulated in a biological sample obtained from the subject, the level of the biomarker is compared with the level of at least one comparator control contained in the kit, such as a positive control, a negative control, a historical control, a historical norm, or the level of another reference molecule in the biological sample. In certain embodiments, the ratio of the biomarker and a reference molecule is determined to aid in the monitoring of the treatment. Treatments

In certain embodiments, the method of the present invention comprises effecting a therapy based on the assessment of risk for prostate cancer recurrence. For example, in certain embodiments, the detection of a differential expression of one or more biomarkers, indicates that further treatment of the subject is desired. Such treatment can include surgery, radiation, chemotherapy, and the like. For example, in one embodiment, the method comprises administering radiotherapy to the subject when it is determined that the subject has an enhanced risk of prostate cancer recurrence.

Measurement of biomarker levels allow for the course of treatment of a disease to be monitored. The effectiveness of a treatment regimen for a disease can be monitored by detecting one or more biomarkers in an effective amount from samples obtained from a subject over time and comparing the amount of biomarkers detected. For example, a first sample can be obtained prior to the subject receiving treatment and one or more subsequent samples are taken after or during treatment of the subject. Changes in biomarker levels across the samples may provide an indication as to the effectiveness of the therapy.

To identify therapeutics or drugs that are appropriate for a specific subject, a test sample from the subject can also be exposed to a therapeutic agent or a drug, and the level of one or more biomarkers can be determined. Biomarker levels can be compared to a sample derived from the subject before and after treatment or exposure to a therapeutic agent or a drug, or can be compared to samples derived from one or more subjects who have shown improvements relative to a disease as a result of such treatment or exposure. Thus, in one aspect, the invention provides a method of assessing the efficacy of a therapy with respect to a subject comprising taking a first measurement of a biomarker panel in a first sample from the subject; effecting the therapy with respect to the subject; taking a second measurement of the biomarker panel in a second sample from the subject and comparing the first and second measurements to assess the efficacy of the therapy.

Additionally, therapeutic agents suitable for administration to a particular subject can be identified by detecting one or more biomarkers in an effective amount from a sample obtained from a subject and exposing the subject-derived sample to a test compound that determines the amount of the biomarker(s) in the subject-derived sample. Accordingly, treatments or therapeutic regimens for use in subjects having an enhanced risk for recurrent prostate cancer can be selected based on the amounts of biomarkers in samples obtained from the subjects and compared to a reference value. Two or more treatments or therapeutic regimens can be evaluated in parallel to determine which treatment or therapeutic regimen would be the most efficacious for use in a subject to delay onset, or slow progression of a disease. In various embodiments, a recommendation is made on whether to initiate or continue treatment of a disease.

In various exemplary embodiments, effecting a therapy comprises administering a disease-modulating drug to the subject. The subject may be treated with one or more drugs until altered levels of the measured biomarkers return to a baseline value measured in a population not having prostate cancer, not having recurrent prostate cancer, or showing improvements in disease biomarkers as a result of treatment with a drug. In one embodiment, the subject may be treated with one or more drugs until altered levels of the measured biomarkers return to a baseline value measured in pre-head trauma sample obtained from the subject. Additionally, improvements related to a changed level of a biomarker or clinical parameter may be the result of treatment with a disease- modulating drug.

Any drug or combination of drugs disclosed herein may be administered to a subject to treat a disease. The drugs herein can be formulated in any number of ways, often according to various known formulations in the art or as disclosed or referenced herein.

In various embodiments, any drug or combination of drugs disclosed herein is not administered to a subject to treat a disease. In these embodiments, the practitioner may refrain from administering the drug or combination of drugs, may recommend that the subject not be administered the drug or combination of drugs or may prevent the subject from being administered the drug or combination of drugs.

In various embodiments, one or more additional drugs may be optionally administered in addition to those that are recommended or have been administered. An additional drug will typically not be any drug that is not recommended or that should be avoided. EXPERIMENTAL EXAMPLES

The invention is further described in detail by reference to the following experimental examples. These examples are provided for purposes of illustration only, and are not intended to be limiting unless otherwise specified. Thus, the invention should in no way be construed as being limited to the following examples, but rather, should be construed to encompass any and all variations which become evident as a result of the teaching provided herein.

Without further description, it is believed that one of ordinary skill in the art can, using the preceding description and the following illustrative examples, make and utilize the present invention and practice the claimed methods. The following working examples therefore, specifically point out the preferred embodiments of the present invention, and are not to be construed as limiting in any way the remainder of the disclosure.

Example 1 : A Four Gene Signature Predictive of Prostate Cancer Recurrence

Following Radical Prostatectomy

Prostate cancer is the most common form of non-dermatological cancer among US men, with an increasing incidence due to the aging population. Patients diagnosed with clinically localized disease identified as intermediate or high-risk are often treated by radical prostatectomy. Approximately 33% of these patients will suffer recurrence after surgery. Identifying patients likely to experience recurrence after radical prostatectomy would lead to improved clinical outcomes, as these patients could receive adjuvant radiotherapy.

Studies presented herein were conducted to identify a gene expression signature to identify those at risk for recurrence. To circumvent the issue of

reproducibility that occurs using microarrays, quantitative PCR analysis of carefully selected smaller gene sets was used. This also allows the identification of differentially expressed genes by the same method by which gene expression would be determined in the clinic. Reported herein is a new tool for prediction of prostate cancer recurrence based on the expression pattern of a small set of cooperation response genes (CRGs). CRGs are a group of genes synergistically dysregulated in response to cooperating oncogenic mutations. They are critical to the cancer phenotype at about 50% frequency (McMurray et al., 2008, Nature, 453(7198): 1112-1116). Due to their significant influence on the malignant phenotype, it was examined whether expression of a subset of CRGs might be predictive of prostate cancer recurrence. It is demonstrated herein that systemic dysregulation of CRGs is also found in prostate cancer, including a 4-gene signature (HBEGF, HOXC13, IGFBP2, and SATBl) capable of differentiating recurrent from non-recurrent prostate cancer. To develop a suitable diagnostic tool to predict disease outcomes in individual patients, multiple algorithms and data handling strategies were evaluated on a training set using LOOC V.

The best-performing algorithm, when used in combination with a predictive nomogram based on clinical staging, predicted recurrent and non-recurrent disease outcomes in a blinded validation set with 83% accuracy, outperforming previous methods. Disease-free survival times between the cohort of prostate cancers predicted to recur and predicted not to recur differed significantly (p = 1.38xl0^"6). This allows accurate identification of prostate cancer patients likely to experience future recurrent disease immediately following removal of the primary tumor.

Further information regarding the data presented herein can be found in

Komisarof et al., 2017, Oncotarget, 8(2): 3430-3440, which is incorporated by reference in its entirety.

The materials and methods employed in the experiments are now described.

Tissue specimen

All tissue samples were collected from radical prostatectomy specimens by the RPCI Pathology Resource Network with IRB approval. Patient demographic, clinical, pathology, and outcome data were collected through the Clinical Data Network, another shared resource at RPCI. All tissue specimens were immediately processed and snap frozen in liquid nitrogen within 30 minutes of radical prostatectomy (Morrison et al., 2009, Prostate, 69(7): 770-773). All tissue samples were reviewed by a board certified anatomic pathologist to verify the diagnosis of prostatic adenocarcinoma and estimate the percent neoplastic tumor nuclei. Data collected for each patient included standard prognostic variables, such as clinical and pathological Gleason score, clinical (cT M) and pathological (pT M) stage, and PSA. All patients had at least 3 years of follow-up data. Biochemical progression was defined by the AUA guidelines of serum PSA of 0.2 ng/mL or greater (obtained 6 weeks - 3 months postoperatively), with a second confirmatory level of PSA greater than 0.2 ng/mL.

CRG expression data

Total RNA was harvested from frozen tissue sections using a standard Trizol (Life Technologies, Carlsbad, CA) application. Tissues were homogenized with Trizol reagent. RNA was precipitated from the aqueous phase using isopropanol and rehydrated using DEPC water. An RNA aliquot was run on an Agilent 2100 bioanalyzer to confirm RNA integrity by generating a RNA Integrity Number (RIN) value. RNA was converted to cDNA and quantified via Taq-Man Low Density Array (TLDA) RT-PCR.

Data handling

Ct values were normalized to Becnl . Non-detects were imputed based on estimation of a non-random missing data mechanism using the nondetects

R/Bioconductor package (McCall et al., 2014, Bioinformatics, 30(16): 2310-2316). Statistical assessment of differential expression was performed using a t-test based on maximum likelihood estimates (MLEs) of the within group means and variances generated by the development version of the nondetects package.

Generation of predictions

Two-tailed t-tests were performed on normalized data to identify genes differentially regulated between biochemically recurrent and non-recurrent cohorts. Multiple p-value cutoffs were tested to assess the relative success of different sizes of gene signature. Three prediction algorithms based on clustering techniques were generated and software was written to implement each in R. Algorithms incorporated a gene signature of size N and evaluated samples as points in N-dimensional space. The "distance" algorithm generates recurrence and non-recurrence scores by comparing the Euclidian distance between the sample point and all points in the recurrent and non- recurrent groups respectively. The "centroid" algorithm generates recurrence and non- recurrence scores by comparing the distance between the sample point and the centroids of the recurrent and non-recurrent group. The "nearest neighbor" algorithm generates recurrence and non-recurrence scores by comparing the distance between the sample point and the closest member in both the recurrent and non-recurrent groups. Predictions are made by selecting the lower of either the recurrence or non-recurrence scores.

Algorithms

Distance

Non-recurrence score: d ⁼∑ J(^nr(¾ ^~ ω)^{2 +} (^nr(¾^{) ~ (2)})² + - + (^ηΓ( ;) - ^χωϊ

Recurrence score:

^: J fanJ - Hi))² + (^r(¾) - (2))² + - + (r_(n£) ^~ ΗθΫ

1

I = # genes in signature

N = # samples in training set

Rn(i) = expression of the 1^th gene in the signature in the n"¹ recurrent sample in the training set

Rn(i) = expression of the 1^th gene in the signature in the n"¹ non-recurrent sample in the training set

Xi = expression of the 1^th gene in the signature in the current sample in the test set

Centroid Non-recurrence score:

Recurrence score:

l = # genes in signature

N = # samples in training set

Rn(i) = average expression of the 1^th gene in the signature in the η^Λ recurrent sample in the training set

Rn(i) = average expression of the 1^th gene in the signature in the n"¹ non-recurrent sample in the training set

Xi = expression of the 1^th gene in the signature in the current sample in the test set Nearest-neighbor

Non-recurrence score: d = J(nr₍₁₎ - x₍₁₎)² + (nr₍₂₎ - x₍₂₎)² + ··· + (nr₍₀ - x₍ )² Recurrence score: d = J(r₍₁₎ - x₍₁₎)² + (r₍₂₎ - x₍₂₎)² + - + (r₍₀ - x₍₀)² I = # genes in signature

Rn(i) = expression of the 1^th gene in the signature in the closest recurrent sample in the training set

Rn(i) = expression of the 1^th gene in the signature in the closest non-recurrent sample in the training set

Xi = expression of the 1^th gene in the signature in the current sample in the test set Evaluation of algorithms on training set

Predictions were made on a 32-sample training set with 16 biochemically recurrent and 16 non-recurrent tumors. Each permutation of normalization method, p- value cutoff for gene signature, and prediction algorithm was evaluated using Leave-one- out cross-validation (LOOCV).

Evaluation of algorithms on validation set

The centroid algorithm was used to make predictions about a 23 -sample validation set using the imputed and BECN1 -normalized data with a CRG signature generated using a p-value cutoff of < 0.01.

Incorporation of clinical and pathological information and final predictions

Clinical and pathological information was incorporated into the prediction decision procedure to improve specificity. Tumors with pathological stage T2bN0M0 or below, with negative surgical margins were classified as non-recurrent. All other predictions were generated algorithmically as before.

ROC curve generation

ROC curves were created using predictions made by varying the discrimination threshold between a prediction of recurrence vs non-recurrence. The model is altered by adding a modifier to the recurrence score before making predictions. When a modifier of -10 is added, all samples were predicted to be non-recurrent; when a modifier of +10 is added, all samples were predicted to be recurrent. Sensitivity and specificity of predictions were measured and plotted for each value of modifier at which a prediction changes.

Kaplan-Meier survival curve generation

Kaplan-Meier survival curves were taken by plotting time from radical prostatectomy to biochemical recurrence for patients predicted to recur and patients predicted not to recur respectively. Statistical analysis was done using log-rank test and the R package "survdif '. The results of the experiments are now described.

Identification of differentially expressed CRGs among prostate cancer samples

Tissue from prostate cancer and benign prostate was collected from patients (n = 55) who underwent radical prostatectomy between 1990 and 2002. All patients had newly-diagnosed, clinically localized prostate cancer (Morrison et al., 2009, Prostate, 69(7): 770-773). Data collected for each patient included standard prognostic variables, such as clinical (cT M - a measure of tumor size, nodal involvement, and metastasis) and pathological (pTNM) stage, clinical and pathological Gleason score, and PSA.

Differences in CRG expression between prostate cancer and benign prostate were assessed post-normalization. 64% of CRGs were significantly dysregulated in prostate cancer compared to benign prostate (two-tailed t-test, p < .05) (Figure 6), and expression values of CRGs distinguished the majority of malignant from benign samples via hierarchical clustering (Figure 1).

The samples were separated into training (n = 32) and validation (n = 23) sets, and disease outcomes were blinded in the validation set. The strategy was to test multiple methods of making predictions using the training set, identify the method that generated the most accurate predictions via cross-validation, and use this method to make predictions using the validation set.

To make predictions on patient outcomes, the gene expression data was first normalized. In addition, a newly developed method for imputing missing gene expression values caused by PCR amplification failure was implemented (McCall et al., 2014, Bioinformatics, 30(16): 2310-2316). Two-tailed t-tests were performed on the normalized training set to identify the genes most differentially expressed between the recurrent and non-recurrent cohorts. Various p-value cutoffs were used to create gene signatures of varying sizes. Finally, three prediction algorithms based on standard clustering techniques were developed and software to implement them was written in R.

Each algorithmic combination was assessed using LOOCV and the training data set. Each combination consists of a data handling method, a p-value cutoff for inclusion in gene signature, and a prediction algorithm. Of the data-handling strategies tested, predictions made using data normalized to BECN1 and imputed to restore missing values caused by PCR amplification failure performed best, with accuracy of 86% averaged across the three algorithmic predictive methods and p-value cutoffs (Figure 2A). Predictions made with un-imputed data had an accuracy of 59% (Figure 2A). Of the three prediction algorithms tested, the centroid algorithm resulted in the most accurate predictions, with an accuracy of 75% averaged across all data handling conditions and p-value cutoffs (Figure 2B). The distance algorithm also performed well (accuracy = 74%) on the training set, while the nearest neighbor algorithm was inferior (accuracy = 69%) (Figure 2B). The highest overall accuracy of 90% was achieved using the lowest p-value cutoff (p < 0.01) (Figure 2C). This cutoff resulted in the generation of a 4-gene signature (Table 1). Predictions on the validation set were made using the centroid algorithm with a 4-gene signature derived from applying p-value cutoff <0.01 to the BECN1 -normalized and imputed data. Predictions made on the training set using these conditions had sensitivity of 100%> (Figure 2C). The lack of false negatives is important since they represent patients who could benefit from adjuvant radiotherapy but would not receive it based on the predictions.

Table 1 : CRG predictive gene signature with fold changes in recurrent samples

The chosen conditions accurately predicted 70% of the samples in the validation set, outperforming predictions made utilizing Gleason score (61%) and pathological stage (64%) (Figure 3). Sensitivity (92%) was significantly higher than specificity (40%), indicating recurrent tumors were successfully identified, but also that many non-recurrent tumors were misidentified as likely to recur. It was hypothesized that these errors may result from making predictions about prostate cancers with an unfavorable gene expression profile, but which nevertheless were small, located in only one half of the prostate, with no lymph node involvement, no distant metastases, and negative margins. These cancers were detected early and cured by surgery. Therefore, pathological information was incorporated into the prediction decision procedure to improve specificity. Using this modified decision procedure, prediction accuracy improved to 83% (Figure 3). Specificity improved to 70%, while sensitivity remained high at 92% and false negatives were few. In comparison, predictions based on the CRG signature were not improved when combined with predictions based on Gleason score (Figure 3). For the important metrics of accuracy and sensitivity, the CRG + surgery- based predictions outperformed all other prediction modalities.

Receiver operating characteristic (ROC) curves were created to evaluate the sensitivity and specificity of predictions made using different discrimination thresholds for recurrence or non-recurrence. ROC curves generated using CRG-based predictions resulted in area under the curve (AUC) of 0.67 (Figure 4). When samples best handled by predictions based on pathological information were removed from the sample set, the AUC increased to 0.75 (Figure 4).

Kaplan-Meier survival curves were created to visualize recurrences in the predicted high risk and low risk cohorts (Figure 5). Of the patients in the validation set predicted to recur (n = 15), 12 experienced biochemical recurrence at a median time of 40 months post-prostatectomy. Only one patient in the cohort predicted not to recur (n = 8) experienced biochemical recurrence 126 weeks post-prostatectomy, a highly significant result (p= 1.38xl0^"6, log-rank test).

To control for the possibility that the predictive power of the gene signature reported herein may be affected by variation in stromal content of the prostate cancer specimens, the mRNA expression of two stromal markers, smooth muscle alpha- actin and vimentin, was assessed in the validation set. The two mRNAs are highly expressed in smooth muscle, fibroblasts, and myofibroblasts. Neither vimentin nor alpha- actin was significantly differentially expressed in recurrent vs non-recurrent cohorts using both non-normalized expression values (p = .18, .95, two-tailed student's t test) (Figure 7A) and expression values normalized to Becnl mRNA (p = .88, .15) (Figure 7B).

Consistently, Becnl mRNA also was not significantly differentially expressed in the two cohorts (p = .69) (Figure 7A). In addition, expression levels for Becnl and RhoA mRNAs, and ribosomal 18S RNA were strongly correlated between all samples (r = .822, .658, .850, Pearson's correlation coefficient) (Figure 8), suggesting that Becnl expression qualifies as a reasonable reference for sample normalization. Taken together, these results suggest that the predictive power of the gene signature originates primarily from gene expression differences associated with cancer cells, rather than variance in the stromal content of tissue samples.

A four-gene signature based algorithm for identifying risk for prostate cancer recurrence

An estimated 50,000 radical prostatectomies are performed each year in the US, and about 15,000 of these patients will experience biochemical recurrence. It is demonstrated herein that a four-gene signature based on heparin binding EGF-like growth factor (HBEGF), homeobox protein hox-cl3 (HOXC13), insulin-like growth factor binding protein 2 (IGFBP2), and special AT -rich sequence-binding protein- 1 (SATB l) was able to identify patients whose prostate cancer recurred. The prediction algorithm that incorporated surgical information provided accurate prognostic

information on patient outcomes with 83% accuracy. Most patients undergoing radical prostatectomy receive no additional treatment beyond regular monitoring of PSA.

Adjuvant radiotherapy administered immediately after radical prostatectomy could improve outcomes for patients identified using the CRG signature to be at high risk for recurrence.

The four genes identified in the presently described signature, HBEGF,

HOXC13, IGFBP2, and SATBl, all play significant roles in the modification of cancer phenotypes. HOXC13, a homeobox-family transcription factor known to control cell proliferation and differentiation, was found to be upregulated in the presently studied recurrent prostate cancer samples. Furthermore, knockdown of HOXC13 has been reported to decrease viability of several cancer cell lines in vitro, including the prostate cancer line PC-3ML (Kasiri et al., RSC Adv. 2013; 3 :3260-3269). SATB l, a chromatin organizer responsible for the recruitment of chromatin remodeling proteins; IGFBP2; and HBEGF, a ligand of the EGF receptor that also binds heparin, were all found to be downregulated in the presently studied recurrent sample cohort. FIBEGF was

unexpectedly downregulated in the recurrent prostate cancer samples, which would appear at odds with its previously documented role in cancer. However, a recent study found that the expression of the growth factor receptor FGFRl was associated with indolent prostate cancer. While the particular mechanism by which FGFRl acts to drive this outcome is unknown, FIBEGF may function in a similar way (Irshad et al, 2013, Sci Transl Med, 5(202): 202ral22).

The focus for this study was biochemical recurrence after radical prostatectomy. Biochemical recurrence after radical prostatectomy almost always requires the primary tumor to already have escaped the prostatic capsule, either invading local tissue or metastasizing to regional lymph nodes or distant organs. These behaviors are hallmarks of tumor aggressiveness, and suggest that the CRG signature may provide valuable information for assessing patient outcomes even in patients who have not undergone radical prostatectomy. Currently the best test to determine risk of aggressive disease is the Gleason score, a test with limited predictive power in many cases. Several groups have already identified gene signatures that are hallmarks of either indolent or aggressive disease (Dhanasekaran et al., 2001, Nature, 412(6849): 822-826; Irshad et al, 2013, Sci Transl Med, 5(202): 202ral22; Penney et al., 2011, J Clin Oncol, 29(17): 2391- 2396). The CRG signature, which is predictive of biochemical recurrence of prostate cancer, could likewise provide useful prognostic information to patients at the time of diagnosis.

Irshad et al. (Irshad et al, 2013, Sci Transl Med, 5(202): 202ral22) identified a three-gene signature associated with aging and cellular senescence that is predictive of indolence in prostate cancer with low Gleason grades. These low-grade cancers are often managed with active surveillance, however, not all of them are indolent. Identifying patients who would typically not be treated aggressively who do in fact require an intervention is also the focus of this study. Likewise, Ross-Adams et al. (Ross- Adams et al., 2015, EbioMedicine, 2(9): 1133-1144) identified a 100-gene panel predictive of prostate cancer recurrence using a transcriptomics approach. The indolence signature and the transcriptomics signature may complement the recurrence signature reported herein; combining these different approaches may lead to better prediction of prostate cancer outcomes.

The success of this model to predict prostate cancer recurrence speaks to the importance of the CRGs in regulating human cancer behavior. As the CRGs were originally identified in a colon cancer background, it stands to reason that CRG expression may provide valuable prognostic information for colon cancer as well. Early- stage colon cancer is treated primarily with surgical resection, and assessing the likelihood of recurrence is an important clinical question. Future studies of the predictive potential of CRGs may provide greater insight into the likelihood of different disease outcomes, thus permitting better-informed decisions about treatment.

The disclosures of each and every patent, patent application, and publication cited herein are hereby incorporated herein by reference in their entirety. While this invention has been disclosed with reference to specific embodiments, it is apparent that other embodiments and variations of this invention may be devised by others skilled in the art without departing from the true spirit and scope of the invention. The appended claims are intended to be construed to include all such embodiments and equivalent variations.

Claims

CLAIMS What is claimed is:

1. A method of assessing the risk of recurrence of prostate cancer in a subject, the method comprising:

a. detecting the level of at least one biomarker in a biological sample obtained from the subject, wherein the at least one biomarker is selected from the group consisting of HBEGF, HOXC13, IGFBP2, and SATB1;

b. comparing the level of the at least one biomarker in the biological sample to a control level of the at least one biomarker; and

c. determining that the subject is at increased risk for recurrence of prostate cancer when the at least one biomarker is differentially expressed in the biological sample as compared to the control level.

2. The method of claim 1, wherein the biological sample comprises prostate tissue of the subject.

3. The method of any one of claims 1-2, wherein the prostate tissue is obtained during surgical resection of at least a portion of the prostate of the subject.

4. The method of any one of claims 1-3, wherein the at least one biomarker comprises two or more of the group consisting of HBEGF, HOXC13, IGFBP2, and SATB 1.

5. The method of any one of claims 1-4, wherein the at least one biomarker comprises three or more of the group consisting of HBEGF, HOXC13, IGFBP2, and SATB 1.

6. The method of any one of claims 1-5, wherein the at least one biomarker comprises HBEGF, HOXC13, IGFBP2, and SATB 1.

7. The method of any one of claims 1-6, wherein the subject is determined to be at increased risk for recurrence of prostate cancer when HOXC13 in the biological sample is increased as compared to the control level.

8. The method of claim 7, wherein the subject is determined to be at increased risk for recurrence of prostate cancer when HOXC13 in the biological sample is increased by greater than about 6.5 fold as compared to the control level.

9. The method of any one of claims 1-7, wherein the subject is determined to be at increased risk for recurrence of prostate cancer when at least one biomarker selected from the group consisting of HBEGF, IGFBP2, and SATB1 in the biological sample is decreased as compared to the control level.

10. The method of claim 9 wherein the subject is determined to be at increased risk for recurrence of prostate cancer when FIBEGF in the biological sample is decreased by greater than 2 fold as compared to the control level.

11. The method of claim 9 wherein the subject is determined to be at increased risk for recurrence of prostate cancer when IGFBP2 in the biological sample is decreased by greater than 1.2 fold as compared to the control level.

12 The method of claim 9 wherein the subject is determined to be at increased risk for recurrence of prostate cancer when SATB1 in the biological sample is decreased by greater than 3 fold as compared to the control level.

13. The method of any one of claims 1-12, wherein the method comprises the use of a multi-dimensional non-linear algorithm to determine if the at least one biomarker is differentially expressed.

14. The method of any one of claims 1-13, wherein the at least one biomarker is an RNA biomarker.

15. The method of any one of claims 1-13, wherein the at least one biomarker is a protein biomarker.

16. The method of any one of claims 1-15, wherein the method further comprises using information from the surgical resection of at least a portion of the prostate of the subject.

17. The method of claim 16, wherein the information comprises at least one selected from the group consisting of primary tumor staging, lymph node staging, and metastatic staging of a primary tumor of the subject.

18. The method of any one of claims 1-17, wherein the method further comprises effectuating a treatment of the subject.

19. The method of claim 18, wherein the method comprises administering adjuvant radiotherapy to the subject determined to be at risk for recurrence of prostate cancer.