DIAGNOSTIC MARKERS OF INDOLENT PROSTATE CANCER

CROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This application claims benefit of Provisional Appln. 61/684,029, filed 8/16/2012, and Provisional Appln. 61/718,468, filed 10/25/2012, and Provisional Appln. 61/745,207, filed 12/21/2012, the entire contents of which are hereby incorporated by reference as if fully set forth herein, under 35 U.S.C. § 119(e).

STATEMENT OF GOVERNMENT SUPPORT

[0002] This invention was made with government support under Grant Nos. R01CA076501, CA154293, CA084294 and CA121852 awarded by the National Cancer Institute, and a Silico Research Centre of Excellence NCI-caBIG, SAIC 29XS192 grant awarded by the National Cancer Institute. Thus, the United States Government has certain rights in the present invention.

BACKGROUND OF THE INVENTION

[0003] With over 200,000 new diagnoses per year (7), prostate cancer is one of the most prevalent forms of cancer in aged men. Several factors, including an increase in the aging population and widespread screening for prostate specific antigen (PSA), have contributed to a substantial rise in diagnoses of early-stage prostate tumors, many of which require no immediate therapeutic intervention (2-4). Indeed, the primary means of determining the appropriate treatment course for men diagnosed with prostate cancer still relies on Gleason grading, a histopathological evaluation that lacks a precise molecular correlate (5). While patients presenting with biopsies of high Gleason score (Gleason >8) tumors are recommended to undergo immediate treatment, determining the appropriate treatment for those with biopsies of low (Gleason 6) or even intermediate (Gleason 7) Gleason score tumors can be more ambiguous.

[0004] Currently, there is the potential for overtreatment of patients who have indolent prostate cancer (e.g., low-risk, non-aggressive or non-invasive cancers) who would not have died of the disease if left untreated (4, 6-8). Consequently, the practice of "watchful waiting" (9) or, more recently, "active surveillance" (10-12) has emerged as an alternative for monitoring men with potentially low risk prostate cancer, with the intention of avoiding treatment unless there is evidence of disease progression. The advantage is to minimize overtreatment; however, the obvious risk is that active surveillance may miss the opportunity for early intervention of tumors that are seemingly low risk but are actually aggressive. Thus, there is a critical need to identify biomarker panels that distinguish the majority of low Gleason score tumors that will remain indolent from the few that are truly aggressive. Unfortunately, so far prostate cancer, unlike many other cancer types, has proven remarkably resilient to classification into molecular subtypes associated with distinct disease outcomes (13, 14). Additionally, an inherent lack of understanding of the biological processes that distinguish indolence from aggressiveness has represented a considerable limitation for identifying relevant biomarkers.

SUMMARY OF THE INVENTION

[0005] Certain embodiments are directed to methods for determining if an indolent epithelial cancer is at a high risk of progressing to an aggressive cancer. More specifically, the method comprises (a) identifying a subject having indolent epithelial cancer, (b) obtaining a test biological sample of the epithelial cancer from the subject and a control sample of benign noncancerous prostate tissue from the subject or from a normal subject, (c) detecting a level of expression of a prognostic mRNA or protein encoded by each of three prognostic genes selected from the group consisting of FGFR1, PMP22, and CDKN1A in the test sample, as compared to the level of expression in the control sample, and (d) if the level of expression of the mRNA or a protein or both is the same or higher than the corresponding level in the control, then

determining that the epithelial cancer is indolent, and if there is about a two-fold or greater decrease in the level of expression of the mRNA or protein compared to the control then determining that the epithelial cancer is at high risk of progressing to an aggressive form. In some embodiments the epithelial cancer is prostate cancer with a Gleason score of 7 or less, breast cancer or lung cancer. In another embodiment the method further includes (e) treating the subject if it is determined that the indolent cancer is at a high risk of progressing toward an aggressive form, the biological sample is blood, plasma, urine or cerebrospinal fluid [0006] Another embodiment is directed to a method for determining if a subject who has an indolent cancer has progressed or is progressing to an aggressive form of cancer by

(a) identifying a subject having indolent epithelial cancer, (b) obtaining a first biological sample of the indolent cancer from the subject at a first time point and a second biological sample at a second time point; (c) determining a level of expression of a prognostic mRNA or protein or both encoded by each of three prognostic genes selected from the group consisting of FGFR1, PMP22, and CDKN1A in the first and second samples at the respective first and second time points, (d) comparing the expression levels of the prognostic mRNA or protein at the first time point to the expression levels at the second time point, and (e) determining that the indolent cancer is not progressing to an aggressive form if the level of expression of the prognostic mRNA or the protein or both at the second time point is the same or greater than at the first time point, and determining that the indolent cancer is at a high risk of progressing toward an aggressive form if there is about a two-fold or greater decrease in the level of expression of the prognostic mRNA or a protein at the second time point compared to the levels at the first time point. In an embodiment the subject is treated if it is determined that the indolent cancer is at a high risk of progressing toward an aggressive form.

[0007] Other embodiments are directed to various diagnostic kits for detecting the expression levels of a prognostic mRNA or a protein encoded or both by each of three prognostic genes selected from the group consisting of FGFR1, PMP22, and CDKN1A in a biological sample, the kit comprising oligonucleotides that specifically hybridize to each of the respective mRNAs or one or more agents that specifically bind to each of the respective proteins, or both, optionally having a forward primer and a reverse primer specific for each mRNA encoded by each of the prognostic genes for use n a qRT-PCR assay to specifically quantify the expression level of each mRNA. In another embodiment this diagnostic further includes one or more antibodies or antibody fragments that specifically bind to each of the respective proteins.

[0008] Other embodiments are directed to a-microarray comprising a plurality of

oligonucleotides that specifically hybridize to an mRNA encoded by each of three prognostic genes selected from the group consisting of FGFR1, PMP22, and CDKN1A, which cDNAs or oligonucleotides are fixed on the microarray; in which the oligonucleotides are optionally labeled to facilitate detection of hybridization to the mRNAs. In some embodiments the oligonucleotides are RNA or DNAs. In other embodiments the microarrays have a plurality of antibodies or antibody fragments that specifically bind to a prognostic protein or variant or fragment thereof encoded by each of three prognostic genes selected from the group consisting of FGFR1, PMP22, and CDKN1A, which antibodies or antibody fragments are fixed on the microarray. An immunoassay for detecting whether epithelial cancer in a biological sample taken for a subject is indolent or is at high risk of progressing to an aggressive form, wherein the immunoassay comprises a plurality of antibodies or antibody fragments that specifically bind to prognostic proteins encoded by each of three prognostic genes selected from the group consisting of

FGFR1, PMP22, and CDKN1A.

[0009] Another embodiment is directed to the method where determining expression level of a prognostic protein comprises immunohistochemistry using one or more antibodies or fragments thereof that specifically binds to the proteins or Western Blot. In some embodiments mRNA expression is quantified by qRT-PCR.

BRIEF DESCRIPTION OF THE DRAWINGS

[0010] The present invention is illustrated by way of example, and not by way of limitation, in the figures of the accompanying drawings.

[0011] Figure 1: Study design

[0012] Step 1 : Assembly of a 377-gene signature enriched for cellular processes associated with aging and senescence (Table 1).

[0013] Step 2: Gene set enrichment analyses (GSEA) using the 377-gene signature to query: (i) aggressive human prostate tumors from Yu et al. (ii) aggressive cancers from lung and breast followed by meta-analyses with the human prostate dataset. (Hi) cross-species analysis on indolent mouse prostate lesions from Ouyang et al. The intersection of the leading edge from mouse prostate lesions and the lagging edge from the meta-analyses of human aggressive cancers led to identification of 19-gene "indolence signature" (Table 5). The indolence signature was validated on human prostate tumors from Taylor et al.

[0014] Step 3: Decision tree-learning to classify the 19-gene indolence signature to identify a 3- gene prognostic panel of indolent prostate cancer using Sboner et al.

[0015] Step 4: Validation of the 3-gene prognostic panel at the mRNA and protein levels. [0016] Step 5: Validation of the 3-gene prognostic panel on biopsies from Gleason Grade 6 patients.

[0017] Figure 2: A gene signature of aging and senescence stratifies human prostate cancer (A-Q Identification of an indolence signature: (A, B) GSEA analyses using the 377- gene signature to query expression profiles from aggressive prostate tumors (in A; from Yu et al.) and mouse indolent prostate cancer (in B; from Ouyang et al.). (C) Intersection from the lagging edge in the meta-analyses of aggressive tumors and the leading edge in the mouse indolent lesions to identify the 19-gene indolence signature. (D-F) Validation of an indolence signature: (D, F) GSEA analyses on aggressive (i.e., Gleason score 8,9) or low Gleason score (Gleason score 6 and 7(3 + 4)) prostate tumors Taylor et al. separated by short time to biochemical recurrence (BCR < 35 months; n = 5) or a long time with no evidence of recurrence (BCR > 100 months; n = 5). (E) Summary of the enrichment score from GSEA analyses done on all Gleason 6 prostate tumors (n = 44) partitioned by interval free of biochemical recurrence. Leading and lagging edge genes from each of GSEA plot are provided in Table 3; genes in indolence signature are provided in Table 5.

[0018] Figure 3: A decision tree-learning model identifies a 3-gene prognostic panel (A)

Schematic representation of the decision tree-learning model. The decision tree algorithm systematically samples the expression states of all combinations of the 19-gene indolence signature to identify combinations most effective in segregating patients into indolent and lethal groups. The decision tree-learning model was performed using Sboner et al (Table 22). (B) Summary of the top 3-gene combinations from the decision tree-learning model. The first column shows combinations ranked by cross validation error (Table 6). The next two columns show independent validation using: (1) the odds ratio for each of the 3-gene combinations to accurately predict patient outcome (i.e., indolence or lethality) using confusion matrices (Figure 8); and (2) Kaplan Meier analyses of low Gleason score patients using the Taylor dataset; log- rank p values are summarized here and Kaplan Meier plots shown in Panel C and Figure 9. (C) Kaplan-Meier analysis of patients with low Gleason score (Gleason 6 and 7(3+4); n = 95) from Taylor et al. showing stratification of FGFR1, PMP22, and CDKN1A for fast-recurring versus slow-recurring patients. The Log-Rank p value is indicated. (D, E) C-statistical analysis and Cox proportional hazard model on Gleason 6 and 7(3+4) patients comparing the performance of FGFRl, PMP22 and CDKNIA expression levels with the D'Amico classification or with Gleason score alone.

[0019] Figure 4: Predictive accuracy of the 3-gene predictive panel at the protein expression level (A) Analyses of a tissue microarrays immunostained for FGFRl, PMP22 and CDKNIA showing representative cases of Gleason grade 6 tumors that were indolent or lethal. (B) Kaplan-Meier analysis for patients with Gleason 6 and 7(3+4) included in the TMA (n = 44) separated into high-risk versus low-risk cancers by protein expression of FGFRl, PMP22 and CDKNIA can. The Log-Rank P value is indicated. (C) C-statistical analysis and Cox proportional hazard models for Gleason 6 and 7(3 + 4) patients from the TMA comparing the performance of protein expression levels of FGFRl, PMP22 and CDKNIA with Gleason score. (D) Representative immunohistochemical results from the "non-failed" and "failed" biopsy groups of Gleason 6 patients monitored by surveillance (see Table 1) showing expression levels of FGFRl, PMP22 and CDKNIA. (E) Summary of analyses of initial biopsy samples using all the "failed" cases (n = 14) in the cohort compared to non-failed cases (n = 19) and validated with a second group of non- failed cases (n = 10).

[0020] Figure 5: Supplementary GSEA data for human cancer (A) GSEA analyses showing results using the 377 gene-set of aging and senescence to query gene expression data from a lung and breast cancer dataset. (B) GSEA analyses using the 377 gene-set of aging and senescence to query gene expression data from Gleason grade 6 cancers from Taylor et al.

partitioned according to time to biochemical recurrence. Leading and lagging edge genes are listed in Table 3.

[0021] Figure 6: Phenotypic analysis of a mouse model of indolent prostate cancer A-D. Representative H&E images the anterior prostate of NL·3.1 null mutant mice at the indicated ages. Note that the mice develop prostatic intraepithelial neoplasia (PIN) by 15 months of age. E-H. Analyses of senescence associated β-galactasidase (SA β-gal) activity in the mouse prostate tissues. I. Summary of proliferation rate in the mouse prostate tissues as measured by expression of Ki67 staining. J. Western blot analyses of mouse prostate tissues for analyses of growth arrest (Gadd45alpha), autophagy (Beclin) and senescence-associated (HPlgamma and PML) proteins using the indicated antibodies. [0022] Figure 7: Supplementary data for the decision tree-learning model and K-means clustering (A) Summary of the range of cross validation error for all possible 3-gene combinations identified from the decision tree-learning model. A list of 3-gene combinations from the decision tree-learning model ranked by their cross validation error is provided in 6. (B) K-means clustering analyses showing fast-recurring (aggressive, red) and slow-recurring (indolent; blue) Gleason grade 6 and 7(3+4) prostate tumors from Taylor et al segregated by expression levels of FGFR1, PMP22 and CDKN1A. (C) K-means clustering analyses showing segregation of predicted aggressive (red) and indolent (blue) patients from the Gleason grade 6 and 7(3 + 4) prostate tumors from the TMA by protein expression levels of FGFR1, PMP22 and CDKN1A.

[0023] Figure 8: Confusion matrices for top-ranked 3-gene combinations from the decision tree-learning model Confusion matrices showing the predicted versus actual calls for indolence versus lethality for the indicated 3-gene combinations using the gene expression and clinical outcome data from Sboner et al. (N=8 lethal and N=28 indolent); only cases excluded the training set used for the decision tree analyses (Table 2D). Odds ratios indicate the predictive accuracy for each 3-gene combination.

[0024] Figure 9: Supplementary Kaplan-Meier analyses comparing the 19-gene indolence signature and the top 3-gene combinations from the decision tree-learning model Kaplan Meier analyses were calculated using gene expression values in K-means clustering and correlated to clinical outcome data provided in the Taylor dataset using all Gleason Grade primary tumors (n = 131) or only the Gleason 6 and 7(3 + 4) (n = 95) patients as indicated.

[0025] Figure 10: Supplementary Kaplan-Meier analyses for the single genes in the 3-gene prognostic panel Kaplan Meier analyses were calculated using gene expression values in K- means clustering and correlated to clinical outcome data provided in (A) the Taylor dataset using the Gleason 6 and 7(3 + 4) (n = 95) patients and in (B) the HICCC TMA using the Gleason 6 and 7(3 + 4) (n = 44) patients.

[0026] Figure 11: Immunostaining of 3-gene prognostic panel comparing biopsies and primary tumors A. Negative and positive controls for immunostaining with each antibody on biopsy samples showing low and high power images. B. Controls showing analogous staining on biopsy and whole prostate tissue. [0027] Figure 12: Kaplan-Meier analyses comparing the 3-gene prognostic panel with biomarkers from Ding et al. and Cuzick et al. Kaplan Meier analyses were calculated using gene expression values in K-means clustering and correlated to clinical outcome data provided in the Taylor dataset using the Gleason 6 and 7(3 + 4) (n = 95) patients.

[0028] Figure 13: Provides the sequence information for certain genes, protein, and mRNAs, which are all publically available.

TABLES

[0029] Table 1: Description of the 377 gene-set of aging and senescence [0030] Table 2: Description of patient samples used in this study

A. Yu et al. Training set

B. Taylor et al. Test set

C. Sboner et al., Training set

D. Sboner et al., Test set

[0031] Table 3: Leading/lagging edge genes from the GSEA analyses

A. Yu et al (human prostate cancer) lagging edge genes, Figure 2A.

B. Lung cancer (human) lagging edge genes, Figure 5A

C. Breast Cancer (human) lagging edge genes, Figure 51 A

D. Ouyang et al (mouse) leading edge genes, Figure 2B

E. Taylor et al (human) Gleason grade 6 and 7(3 + 4) (BCR < 35) lagging edge genes, Figure 2F

F. Taylor et al (human) Gleason grade 6 and 7(3 + 4) (BCR > 100) leading edge genes, Figure 2F

[0032] Table 4: Integrative analyses of the 377 gene-set

A. Meta-analyses of human prostate, lung and breast, Integrative analyses of extreme Gleason Grade 6 groups from Taylor et al

B. Integrative analyses of all Gleason Score 6 patients from Taylor et al. Fig. 5b.

[0033] Table 5: Description of the 19-gene indolence signature [0034] Table 6: 3-gene combinations from the decision tree learning model

A. Top 3-gene combinations with 25% cross-validation error

B. All 3-gene combinations

DETAILED DESCRIPTION

Definitions

[0035] 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 invention, the preferred methods and materials are now described. All publications mentioned herein are incorporated herein by reference.

[0036] Generally, nomenclatures used in connection with, and techniques of, cell and tissue culture, molecular biology, immunology, microbiology, genetics, protein, and nucleic acid chemistry and hybridization described herein are those well-known and commonly used in the art. The methods and techniques of the present invention are generally performed according to conventional methods well known in the art and as described in various general and more specific references that are cited and discussed throughout the present specification unless otherwise indicated. See, e.g., Sambrook et ah Molecular Cloning: A Laboratory Manual, 2d ed., Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y. (1989); Ausubel et ah, Current Protocols in Molecular Biology, Greene Publishing Associates (1992, and Supplements to 2002); Harlow and Lan, Antibodies: A Laboratory Manual, Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y. (1990); Principles of Neural Science, 4th ed., Eric R. Kandel, James H. Schwart, Thomas M. Jessell editors. McGraw-Hill/ Appleton & Lange: New York, N. Y. (2000). 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.

[0037] The following terms as used herein have the corresponding meanings given here. 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 invention, the example methods and materials are now described, including the currently preferred embodiments. All publications mentioned herein are incorporated herein by reference.

[0038] "Biological sample" refers to a body sample in which the prognostic biomarkers can be detected. In some embodiments, the sample refers to biopsy tissues collected from an individual having epithelial cancer and to benign or noncancerous control tissue from the subject or a normal control. In other embodiments the biological sample is urine, blood, csf or any other tissue where the prognostic protein and mRNA biomarkers can be detected. Biological samples of cancerous cells can also come from urine of the subject, and the prognostic biomarker mRNA and protein can be found in blood, plasma and cerebrospinal fluid.

[0039] "Indolent, or low-risk, or non-aggressive or non-invasive cancer" means cancer that is unlikely to become symptomatic during life.

[0040] "Aggressive cancer" means prostate cancer or other epithelial cancer that is symptomatic and likely to be lethal. For prostate cancer, aggressive forms typically have a Gleason score >8.

[0041] "High Gleason score " means a Gleason score >8 on the prostate cancer biopsy. Such patients are recommended to undergo immediate treatment.

[0042] "Intermediate Gleason score" means a Gleason score >7 on the prostate cancer biopsy.

[0043] "Low Gleason score" means a Gleason score less than or equal to 6 on the prostate cancer biopsy.

[0044] "At High Risk of Progressing to Aggressive Prostate Cancer" means prostate cancer that is not indolent as is determined by a two-fold decrease in expression of mRNA or protein encoded by the 3-gene prognostic panel compared to normal controls.

[0045] "3-gene prognostic panel" means the genes: FGFR1, PMP22 and CDKN1A, the simultaneous expression of which identifies prostate cancer tumors that are indolent as opposed to at risk of becoming aggressive.

[0046] "Proteins encoded by the 3 gene prognostic panel" and "prognostic biomarker proteins" are used interchangeably and mean the proteins encoded by the 3-gene prognostic panel and their variants and fragments. [0047] "mRNA encoded by the 3 gene prognostic panel" means mRNA transcribed from each of the genes in the 3 gene panel, which mRNAs are translated the prognostic biomarker proteins.

[0048] "Prognostic biomarker mRNA" means mRNA encoded by the genes in the 3 gene prognostic panel.

[0049] "Detect" "detection" or "detecting" refer to the quantification of a given prognostic biomarker mRNA or protein.

[0050] "Treatment" includes any process, action, application, therapy, or the like, wherein a subject (or patient), including a human being, is provided medical aid with the object of improving the subject's condition, directly or indirectly, or slowing the progression of a condition or disorder in the subject, or ameliorating at least one symptom of the disease or disorder under treatment.

[0051] "Indolence signature" means a group of 19 "PCIG" genes associated with cellular processes of aging and senescence that are enriched in indolent prostate tumors identified using Gene Set Enrichment Analysis (GSEA). The 19 genes are either enriched in down-regulated in aggressive human prostate cancer or conversely up-regulated in indolent prostate cancer {i.e., the leading edge).

[0052] "PCIG" is an abbreviation for "Prostate Cancer Indolence Genes" (used

interchangeably) and refers to any single one or any combination of the following 19 genes: B2M, CAT, CDKN1A, CFH, CLIC4, CLU, CTSH, CX3CL1, FGFRl, GPX3, IGF1, ITM2A, LGALS3, MECP2, MSN, NFE2L2, PMP22, SERPINGl, TXNIP: which genes are spelled out below. Beta-2 microglobulin (B2M), Cyclin-dependent kinase inhibitor 1 A (p21 or Cipl) (CDKN1 A), Chloride intracellular channel 4 (CLIC4), Clusterin (CLU), Cathepsin H (CTSH), Chemokine (C-X3-C motif) ligand 1 (CX3CL1), Fibroblast growth factor receptor 1 (FGFRl), Glutathione peroxidase 3 (plasma) (GPX3), Insulin-like growth factor 1 (somatomedin C) (IGF1), integral membrane protein 2A (ITM2A), Lectin, galactose-binding, soluble, 3

(LGALS3), Methyl CpG binding protein 2 (Rett syndrome) (MECP2), Moesin (MSN), Nuclear factor (erythroid-derived 2)-like 2 (NFE2L2), Peripheral myelin protein 22 (PMP22), Serpin peptidase inhibitor, clade G (CI inhibitor), member 1 (SERPINGl) and Thioredoxin interacting protein (TXNIP). [0053] The term "probe" refers to any molecule which is capable of selectively binding to a specifically intended target molecule, for example, an oligonucleotide probe that specifically hybridizes to a prognostic biomarker mRNA, or an antibody that specifically binds a biomarker protein encoded by the 3 gene prognostic panel. Probes can be either synthesized by one skilled in the art, or derived from appropriate biological preparations. For purposes of detection of the target molecule, probes may be specifically designed to be labeled, as described herein.

Examples of molecules that can be utilized as probes include, but are not limited to, RNA, DNA, proteins, antibodies, and organic molecules.

[0054] "Epithelial, prostate, breast and lung cancer" refer to a cancerous tumor. For purposes of this application, cancer is not intended to be limited to cancer of any specific types and instead broadly includes many types of epithelial cancers.

[0055] As used herein, the term "expression level" refers to expression of protein as measured quantitatively by methods such as Western blot, immunohistochemistry or ELISA and expression of mRNA encoding the three prognostic biomarkers as measured quantitatively by methods including but not limited to, for example, qRT-PCR. Methods for quantifying expression levels of mRNA are further described below in references.

[0056] As used herein, the term "detect an expression level" refers to measuring or quantifying either protein expression or mRNA expression.

[0057] "An increased or decreased expression level" refers to increased or decreased protein expression level or mRNA expression level relative to a normal or control value. For purposes of this application, increased or decreased protein or mRNA expression refers to expression in the cancerous biological sample compared to either the corresponding level in a control subject (free of cancer) or in normal tissue adjacent to the cancer.

[0058] Unless otherwise specified, the terms "antibody" and "antibodies" broadly encompass naturally-occurring forms of antibodies (e.g., IgG, IgA, IgM, IgE) and recombinant antibodies such as single-chain antibodies, chimeric and humanized antibodies and multi-specific antibodies, as well as fragments and derivatives of all of the foregoing, which fragments and derivatives have at least an antigenic binding site. Antibody derivatives may comprise a protein or chemical moiety conjugated to an antibody moiety. [0059] An "isolated" nucleic acid molecule is one which is separated from other nucleic acid molecules which are present in the natural source of the nucleic acid molecule, namely cancerous or noncancerous biological samples. Preferably, an "isolated" nucleic acid molecule is free of sequences (preferably protein-encoding sequences) which naturally flank the nucleic acid (i.e., sequences located at the 5' and 3' ends of the nucleic acid) in the genomic DNA of the organism from which the nucleic acid is derived. Moreover, an "isolated" nucleic acid molecule, such as a cDNA molecule, can be substantially free of other cellular material or culture medium when produced by recombinant techniques, or substantially free of chemical precursors or other chemicals when chemically synthesized. All prognostic biomarkers and mRNA in the present embodiments are isolated.

[0060] As used herein, the term "about" is used to mean approximately, roughly, around, or in the region of. When the term "about" is used in conjunction with a numerical range, it modifies that range by extending the boundaries above and below the numerical values set forth. In general, the term "about" is used herein to modify a numerical value above and below the stated value by a variance of 20 percent up or down (higher or lower).

Summary of Results

[0061] Many newly diagnosed prostate cancers present as low (G 6 or less) or high (G 8 or higher) Gleason score tumors that require no treatment intervention. However, distinguishing the many indolent tumors from the minority of lethal ones remains a major clinical challenge. It has now been discovered that Gleason score 7 or less prostate tumors can be distinguished as truly indolent or aggressive subgroups based on their expression of a 3-gene prognostic panel:

FGFR1, PMP22, and CDKN1A. The embodiments described here also apply to epithelial tumor classification and prognosis, including lung and breast cancers.

[0062] One of the most significant risk factors associated with prostate cancer is aging (73), which represents a balance of anti- tumori genie and pro-tumorigenic signals. One of the principal anti-tumorigenic signals is cellular senescence (15-18). Indeed, it is now widely appreciated that senescence plays a critical role in tumor suppression in general, and has been associated with benign prostate lesions in humans (19, 20), as well as mouse models (21). Thus, it was hypothesized that prostate tumors destined to remain indolent versus aggressive could be distinguished based on their enrichment for cellular processes associated with aging and senescence.

[0063] 1. To test the hypothesis, a bioinformatics approach was used to identify a 19-gene group (hereafter "an indolence signature" See Table 5) that is enriched in indolent prostate tumors compared to aggressive tumors was identified using Gene Set Enrichment Analysis.

[0064] 2. The 19-gene group indolence signature was further classified using a decision tree learning model leading to the identification of a 3-gene prognostic panel: FGFRl, PMP22, and CDKN1A, which together accurately predicted the outcome of low Gleason score tumors as either truly indolent or at a high risk of becoming aggressive. Validation of this 3-gene prognostic panel on independent cohorts confirmed its independent prognostic value, as well as its ability to improve prognosis with currently used clinical nomograms. Expression of the 3- gene prognostic panel was determined by quantifying mRNA or protein encoded by the panel (collectively referred to as "prognostic biomarkers"). The prognostic biomarkers were discovered to be up-regulated in indolent tumors and down-regulated in aggressive forms of prostate cancer (Fig. 1).

19 "PCIG" gene Indolence Panel also the Indolence Signature

Entrez ID Gene Symbol Hyperlink Gene Description [0065] 3. In various embodiments, the level of expression of the prognostic biomarkers in biopsy samples is used to identify and distinguish truly indolent forms of prostate cancer (and other epithelial cancers) from aggressive forms. In particular it has been discovered that prostate cancer from Gleason 7 or less patients will not progress to malignant disease if their prostate cancers show normal or elevated levels of expression of the 3-gene prognostic panel (mRNA or protein) compared to benign or noncancerous prostate tissue can be identified as indolent prostate cancer. By contrast, prostate cancer of Gleason 7 or less that shows significantly reduced levels (about a 2-fold reduction) of expression of the 3-gene prognostic panel compared to benign or noncancerous prostate tissue can be identified as aggressive Prostate cancer. Other embodiments are directed to the same methods as applied to epithelial cancers generally, and lung and breast cancers specifically.

[0066] 4. The prognostic accuracy expression of this 3-gene panel was tested on biopsies from patients monitored by active surveillance and therefore has clinical utility. A previously identified 4-gene signature of aggressive tumors that includes Pten, Smad4, Cyclin Dl and SPPl, do not overlap with the present new 3-gene panel of indolence. Notably, this 4-gene biomarker panel, which was identified on the basis of its ability to stratify advanced prostate tumors, was not effective for stratifying low Gleason score prostate tumors (Fig. 12).

[0067] 5. Lung and breast cancer also showed significant enrichment of the indolence signature among genes down-regulated in aggressive tumors (NES = -1.90 and -1.52,

respectively; p < 0.001 in both cases) (Fig. 5 A; Table 3B,C). Meta-analysis of the down- regulated {i.e., lagging-edge) genes from the prostate, lung, and breast tumors led to the refinement of the original 377 gene signature to a subset of 68 genes that were most significantly enriched in aggressive tumors (Table 4A). In some embodiments expression of prognostic biomarkers is extended to distinguish forms of indolent vs aggressive epithelial cancers, including lung and breast cancer.

[0068] Details of experiments and description of their significance are set forth in the

Examples.

Sample Preparation: Protein and Nucleic Acid Extraction

[0069] All of the gene, protein and mRNA sequences of the respective genes, proteins and mRNA used in the Examples are set forth herein. [0070] Biological samples of the epithelial cancer in humans (such as prostate, breast and lung) can be conveniently collected by methods known in the art. Usually, the cancerous tissue can be harvested by trained medical staffs or physicians under sterile environment. Biopsies often are taken, for example, by endoscopic means. After harvested from patients, biological samples may be immediately frozen (under liquid nitrogen) or put into a storage, or transportation solution to preserve sample integrity. Such solutions are known in the art and commercially available, for example, UTM-RT transport medium (Copan Diagnostic, Inc, Corona, Calif), Multitrans Culture Collection and Transport System (Starplex Scientific, Ontario, CN), ThinPrep® Paptest Preservcyt® Solution (Cytyc Corp., Boxborough, Mass.) and the like. Biological samples of cancerous cells can also come from urine of the subject, and the prognostic biomarker mRNA and protein can be found in blood, plasma and csf.

[0071] After collection, biological samples are prepared prior to detection of biomarkers.

Sample preparation typically includes isolation of protein or nucleic acids (e.g., mRNA). These isolation procedures involve separation of cellular protein or nucleic acids from insoluble components (e.g., cytoskeleton) and cellular membranes. In situ immunostaining of prognostic biomarker proteins can also be done.

[0072] In one embodiment, the tissues in the biological samples are treated with a lysis buffer solution prior to isolation of protein or nucleic acids. A lysis buffer solution is designed to lyse tissues, cells, lipids and other biomolecules potentially present in the raw tissue samples.

Generally, a lysis buffer of the present invention may contain one or more of the following ingredients: (i) chaotropic agents (e.g., urea, guanidine thiocyanide, or formamide); (ii) anionic detergents (e.g., SDS, N-lauryl sarcosine, sodium deoxycholate, olefine sulphates and sulphonates, alkyl isethionates, or sucrose esters); (iii) cationic detergents (e.g., cetyl trimethylammonium chloride); (iv) non-ionic detergents (e.g., Tween.RTM.-20, polyethylene glycol sorbitan monolaurate, nonidet P-40, Triton.RTM. X-100, NP-40, N-octyl-glucoside); (v) amphoteric detergents (e.g., CHAPS, 3-dodecyl-dimethylammonio-propane-l -sulfonate, lauryldimethylamine oxide); or (vi) alkali hydroxides (e.g., sodium hydroxide or potassium hydroxide). Suitable liquids that can solubilize the cellular components of biological samples are regarded as a lysis buffer for purposes of this application. [0073] In another embodiment, a lysis buffer may contain additional substances to enhance the properties of the solvent in a lysis buffer (e.g., prevent degradation of protein or nucleic acid components within the raw biological samples). Such components may include proteinase inhibitors, RNase inhibitors, DNase inhibitors, and the like. Proteinase inhibitors include but not limited to inhibitors against serine proteinases, cysteine proteinases, aspartic proteinases, metallic proteinases, acidic proteinases, alkaline proteinases or neutral proteinases. RNase inhibitors include common commercially available inhibitors such as SUPERase.In.TM.

(Ambion, Inc. Austin, Tex.), RNase Zap.RTM. (Ambion, Inc. Austin, Tex.), Qiagen RNase inhibitor (Valencia, Calif.), and the like.

Quantification of proteins

[0074] One of ordinary skill in the art will appreciate that proteins frequently exist in a biological sample in a plurality of different forms. These forms can result from either or both of pre- and post-translational modification. Pre-translational modified forms include allelic variants, splice variants and RNA editing forms. Post-translationally modified forms include forms resulting from proteolytic cleavage (e.g., cleavage of a signal sequence or fragments of a parent protein), glycosylation, phosphorylation, lipidation, oxidation, methylation, cysteinylation, sulphonation and acetylation. When detecting or measuring a prognostic protein biomarker of the invention in a sample, the ability to differentiate between different forms of a protein biomarker depends upon the nature of the difference and the method used to detect or measure. For example, an immunoassay using a monoclonal antibody will detect all forms of a protein containing the epitope and will not distinguish between them. However, a sandwich

immunoassay that uses two antibodies directed against different epitopes on a protein will detect all forms of the protein that contain both epitopes and will not detect those forms that contain only one of the epitopes. The embodiments of the invention for determining protein levels include adaptations that permit detection of various forms of the protein.

[0075] The 3 prognostic protein (or mRNA) markers may be combined into one test for efficient processing of a multiple of samples. In addition, one skilled in the art would recognize the value of testing multiple samples (for example, at successive time points) from the same individual. Such testing of serial samples will allow the identification of changes in marker levels over time. Increases or decreases in marker levels, as well as the absence of change in marker levels, provide useful information as described herein to distinguish indolent from aggressive epithelial cancers as well as to determine the appropriateness of drug therapies, and identification of the patient's outcome, including risk of future events.

[0076] In diagnostic assays, the inability to distinguish different forms of a biomarker protein has little impact when the forms detected by the particular method used are equally good biomarkers as any other particular form. However, when a particular form (or a subset of particular forms) of a protein is a better biomarker than the collection of different forms detected together by a particular method, the power of the assay may suffer. In this case, it may be useful to employ an assay method that distinguishes between forms of a protein and that specifically detects and measures a desired form or forms of the protein. Distinguishing different forms of an analyte (e.g., a biomarker) or specifically detecting a particular form of an analyte is referred to as "resolving" the analyte.

[0077] Mass spectrometry is a particularly powerful methodology to resolve different forms of a protein because the different forms typically have different masses that can be resolved by mass spectrometry. Accordingly, if one form of a protein is a superior biomarker for a disease than another form of the biomarker, mass spectrometry may be able to specifically detect and measure the useful form where traditional immunoassay fails to distinguish the forms and fails to specifically detect to useful biomarker. A useful methodology combines mass spectrometry with immunoassay. Additionally, certain methods and devices, such as biosensors and optical immunoassays, may be employed to determine the presence or amount of analytes without the need for a labeled molecule. See, e.g., U.S. Pat. Nos. 5,631,171; and 5,955,377.

[0078] In embodiments where the three prognostic biomarker proteins are extracted from the biological samples for quantification, expression level can be determined using standard assays that are known in the art. These assays include, but not limited to Western blot analysis, ELISA, radioimmunoassay, competitive binding assays, immune-histochemistry assay and the like. A common assay for the prognostic protein biomarkers is an immunoassay, although other methods are well known to those skilled in the art. The presence or amount of a marker is generally determined using antibodies specific for each marker and detecting specific binding. Specific immunological binding of the antibody to the marker can be detected directly or indirectly. Direct labels include fluorescent or luminescent tags, metals, dyes, radionuclides, and the like, attached to the antibody. Indirect labels include various enzymes well known in the art, such as alkaline phosphatase, horseradish peroxidase and the like. In a preferred embodiment, expression level of the prognostic protein biomarkers may be detected by Western blot analysis.

[0079] For western blots, cellular proteins are extracted or isolated from the biological samples (e.g., cancerous tissues), and then separated using SDS-PAGE gel electrophoresis. The conditions for SDS-PAGE gel electrophoresis can be conveniently optimized by one skilled in the art. The three prognostic protein biomarkers in the gels can then be transferred onto a surface such as nitrocellulose paper, nylon membrane, PVDF membrane and the like. The conditions for protein transfer after SDS-PAGE gel electrophoresis may be optimized by one skilled in the art. Preferably, a PVDF membrane is used.

[0080] In some embodiments, biomarker proteins are detected using antibodies specific for each of the 3 biomarker proteins for example using immunohistochemical staining on a tissue microarray (TMA) comprised of primary prostate tumors. In the embodiments most of the tumors will have low (G 6 or less) or intermediate (G 7) Gleason scores (Fig. 4A, B; Table 1; Fig. 11). In some embodiments "first" antibodies that that specifically bind to each of the 3 prognostic protein biomarkers are used. Antibodies against the various protein biomarkers can be prepared using standard protocols or obtained from commercial sources. Techniques for preparing mouse monoclonal antibodies or goat or rabbit polyclonal antibodies (or fragments thereof) are well known in the art. See the Examples.

[0081] Direct detectable label or signal-generating systems are well known in the field of immunoassay. Labeling of a second antibody with a detectable label or a component of a signal- generating system may be carried out by techniques well known in the art. Examples of direct labels include radioactive labels, enzymes, fluorescent and chemiluminescent substances.

Radioactive labels include .sup.1241, .sup.1251, .sup.1281, .sup.1311, and the like. A fluorescent label includes fluorescein, rhodamine, rhodamine derivatives, and the like. Chemiluminescent substances include ECL chemiluminescent.

ELISA

[0082] In another embodiment, detection and quantification of biomarker protein levels is determined by ELISA, typically wherein a first antibody is immobilized onto a solid surface, for example an inert support useful in immunological assays. Examples of inert support include sephadex beads, polyethylene plates, polypropylene plates, polystyrene plates, and the like. In one embodiment, the first antibody is immobilized by coating the antibody on a microtiter plate.

Detection of mRNA Expression Level

All mRNA was studied using published values for each respective dataset described herein, and as such was retrospective. The methods used for RNA isolation and running of the microarrays are described in those studies and are standard prortocols that are well known in the art. Details of the methods are described in:

[0083] 1) Mouse: Ouyang et al. : Ouyang X, DeWeese TL, Nelson WG, Abate-Shen C. Loss-of- function of Nkx3.1 promotes increased oxidative damage in prostate carcinogenesis. Cancer Res 2005; 65: 6773-9.

[0084] 2) Human a) Yu et al.: Yu YP, Landsittel D, Jing L, et al. Gene expression alterations in prostate cancer predicting tumor aggression and preceding development of malignancy. Journal of clinical oncology : official journal of the American Society of Clinical Oncology 2004; 22: 2790-9.

[0085] b) Taylor et al.: Taylor BS, Schultz N, Hieronymus H, et al. Integrative genomic profiling of human prostate cancer. Cancer cell 2010; 18: 11-22.

[0086] c) Sboner et al.: Sboner A, Demichelis F, Calza S, et al. Molecular sampling of prostate cancer: a dilemma for predicting disease progression. BMC medical genomics 2010; 3: 8.

[0087] The Examples have the materials and methods used to isolate mRNA, protein and to select subjects for the Ouyang, Yu, Taylor and Sboner data sets.

[0088] Methods for isolating nucleic acids including mRNA from a cell are well-known in the art. Detection and quantification of mRNA expression levels includes standard mRNA quantitation assays that are also well-known. These assays include but not limited to qRT-PCR (quantitative reverse transcription-polymerase chain reaction), Northern blot analysis, RNase protection assay, and the like. qRT-PCR is preferable to quantify mRNA levels from much smaller samples.

[0089] Real-time polymerase chain reaction, also called quantitative real time polymerase chain reaction (Q-PCR/qPCR/qRT-PCR), is used to amplify and simultaneously quantify a targeted DNA molecule. It enables both detection and quantification (as absolute number of copies or relative amount when normalized to DNA input or additional normalizing genes) of one or more specific sequences in a DNA sample. Currently at least four (4) different chemistries, TaqMan®. (Applied Biosystems, Foster City, Calif), Molecular Beacons, Scorpions® and SYBR® Green (Molecular Probes), are available for real-time PCR.

[0090] All of these chemistries allow detection of PCR products via the generation of a fluorescent signal. TaqMan probes, Molecular Beacons and Scorpions depend on Forster Resonance Energy Transfer (FRET) to generate the fluorescence signal via the coupling of a fluorogenic dye molecule and a quencher moiety to the same or different oligonucleotide substrates. SYBR Green is a fluorogenic dye that exhibits little fluorescence when in solution, but emits a strong fluorescent signal upon binding to double- stranded DNA.

[0091] Two common methods for detection of products in real-time PCR are: (1) non-specific fluorescent dyes that intercalate with any double-stranded DNA, and (2) sequence-specific DNA probes consisting of oligonucleotides that are labeled with a fluorescent reporter which permits detection only after hybridization of the probe with its complementary DNA target.

[0092] Real-time PCR, when combined with reverse transcription, can be used to quantify messenger RNA (mRNA) in cells or tissues. An initial step in the reverse transcription PCR amplification is the synthesis of a DNA copy (i.e., cDNA) of the region to be amplified. Reverse transcription can be carried out as a separate step, or in a homogeneous reverse transcription- polymerase chain reaction (RT-PCR), a modification of the polymerase chain reaction for amplifying RNA. Reverse transcriptases suitable for synthesizing a cDNA from the RNA template are well known.

[0093] Following the cDNA synthesis, methods suitable for PCR amplification of ribonucleic acids are known in the art (See, Romero and Rotbart in Diagnostic Molecular Biology: Principles and Applications pp. 401-406). PCR reagents and protocols are also available from commercial vendors, such as Roche Molecular Systems. PCR can be performed using an automated process with a PCR machine.

[0094] Primer sets used in the present qRT-PCR reactions for various biomarkers may be prepared or obtained through commercial sources. [0095] The primers used in the PCR amplification preferably contain at least 15 nucleotides to 50 nucleotides in length. More preferably, the primers may contain 20 nucleotides to 30 nucleotides in length. One skilled in the art recognizes the optimization of the temperatures of the reaction mixture, number of cycles and number of extensions in the reaction. The amplified product (i.e., amplicons) can be identified by gel electrophoresis. In real-time PCR assay, a fluorometer and a thermal cycler for the detection of fluorescence during the cycling process is used. A computer that communicates with the real-time machine collects fluorescence data. This data is displayed in a graphical format through software developed for real-time analysis.

[0096] In addition to the forward primer and reverse primer (obtained via commercial sources), a single-stranded hybridization probe is also used. The hybridization probe may be a short oligonucleotide, usually 20-35 by in length, and is labeled with a fluorescent reporting dye attached to its 5 '-end as well as a quencher molecule attached to its 3 '-end. When a first fluorescent moiety is excited with light of a suitable wavelength, the absorbed energy is transferred to a second fluorescent moiety (i.e., quencher molecule) according to the principles of FRET. Because the probe is only 20-35 by long, the reporter dye and quencher are in close proximity to each other and little fluorescence is detected. During the annealing step of the PCR reaction, the labeled hybridization probe binds to the target DNA (i.e., the amplification product). At the same time, Taq DNA polymerase extends from each primer. Because of its 5' to 3' exonuclease activity, the DNA polymerase cleaves the downstream hybridization probe during the subsequent elongation phase. As a result, the excited fluorescent moiety and the quencher moiety become spatially separated from one another. As a consequence, upon excitation of the first fluorescent moiety in the absence of the quencher, the fluorescence emission from the first fluorescent moiety can be detected. By way of example, a Rotor-Gene System is used and is suitable for performing the methods described herein. Further information on PCR amplification and detection using a Rotor-Gene can conveniently be found on Corbett's website.

[0097] In another embodiment, suitable hybridization probes such as intercalating dye (e.g., Sybr-Green I) or molecular beacon probes can be used. Intercalating dyes can bind to the minor grove of DNA and yield fluorescence upon binding to double-strand DNA. Molecular beacon probes are based on a hairpin structure design with a reporter fluorescent dye on one end and a quencher molecule on the other. The hairpin structure causes the molecular beacon probe to fold when not hybridized. This brings the reporter and quencher molecules in close proximity with no fluorescence emitted. When the molecular beacon probe hybridizes to the template DNA, the hairpin structure is broken and the reporter dye is no long quenched and the real-time instrument detects fluorescence.

[0098] The range of the primer concentration can optimally be determined. The optimization involves performing a dilution series of the primer with a fixed amount of DNA template. The primer concentration may be between about 50 nM to 300 nM. An optimal primer concentration for a given reaction with a DNA template should result in a low Ct-(threshold concentration) value with a high increase in fluorescence (5 to 50 times) while the reaction without DNA template should give a high Ct-value.

[0099] The probes and primers of the invention can be synthesized and labeled using well- known techniques. Oligonucleotides for use as probes and primers may be chemically synthesized according to the solid phase phosphoramidite triester method first described by Beaucage, S. L. and Caruthers, M. H., 1981, Tetrahedron Letts., 22 (20): 1859-1862 using an automated synthesizer, as described in Needham-VanDevanter, D. R., et al. 1984, Nucleic Acids Res., 12: 6159-6168. Purification of oligonucleotides can be performed, e.g., by either native acrylamide gel electrophoresis or by anion-exchange HPLC as described in Pearson, J. D. and Regnier, F. E., 1983, J. Chrom., 255: 137-149.

Kits

[0100] The present invention provides a kit of manufacture, which may be used to perform detecting either the prognostic biomarker proteins (or fragments thereof) or the mRNA encoding them. In one embodiment, an article of manufacture (i.e., kit) according to the present invention includes a set of antibodies (i.e., a first antibody and a second antibody) specific for each of the 3 biomarker proteins. Antibodies against a house-keeper gene (e.g., GADPH) are provided as a control. In another embodiment, the present kit contains a set of primers (i.e., a forward primer and a reverse primer) (directed to a region of the gene specific to each of the 3 genes in the prognostic panel and optionally a hybridization probe (directed to the same genes, albeit a different region).

[0101] Kits provided herein may also include instructions, such as a package insert having instructions thereon, for using the reagents (e.g., antibodies or primers) to quantify the protein expression level of mRNA expression level of the epithelial cancer biomarkers in a biological sample. Such instructions may be for using the primer pairs and/or the hybridization probes to specifically detect mRNA of the prognostic genes. In an embodiment the kids may include oligonucleotides that specifically hybridize with each of the 3 prognostic mRNA biomarkers.

[0102] In another embodiment, the kit further comprises reagents used in the preparation of the sample to be tested for protein (e.g. lysis buffer). In another embodiment, the kit comprises reagents used in the preparation of the sample to be tested for mRNA (e.g., guanidinium thiocyanate or phenol-chloroform extraction).

[0103] The analysis of a plurality of biomarkers may be carried out separately or

simultaneously with one test sample. For separate or sequential assay of markers, suitable apparatuses include clinical laboratory analyzers such as the ELECSYS® (Roche), the

AXSYM® (Abbott), the ACCESS® (Beckman), the AD VIA® CENTAUR® (Bayer) immunoassay systems, the NICHOLS ADVANTAGE®. (Nichols Institute) immunoassay system, etc. Preferred apparatuses or protein chips perform simultaneous assays of a plurality of markers on a single surface. Particularly useful physical formats comprise surfaces having a plurality of discrete, addressable locations for the detection of a plurality of different analytes. Such formats include protein microarrays, or "protein chips" (see, e.g., Ng and Hag, J. Cell Mol. Med. 6: 329-340 (2002)) and certain capillary devices (see e.g., U.S. Pat. No. 6,019,944). In these embodiments each discrete surface location may comprise antibodies to immobilize one or more of the prognostic biomarker proteins in a sample for detection at each location.

[0104] Certain embodiments are directed to microarrays or DNA chips and the like that can be used to quantify or detect the presence of the three prognostic biomarker proteins or mRNA isolated from a biological sample. An embodiment of a microarray for determining if an epithelial tumor is indolent or aggressive includes antibodies or fragments thereof that specifically bind to each of the prognostic biomarker proteins (or variants or fragments thereof) fixed on the array. Another microarray embodiment has at least one oligonucleotide probe that specifically hybridizes to each of the three prognostic biomarker mRNAs fixed on the array.

[0105] Surfaces may alternatively comprise one or more discrete particles (e.g., microparticles or nanoparticles) immobilized at discrete locations of a surface, where the microparticles comprise antibodies to immobilize one analyte (e.g., a marker) for detection. As noted, many protein biochips are described in the art. These further include, for example, protein biochips produced by Ciphergen Biosystems, Inc. (Fremont, Calif.), Packard Bioscience Company (Meriden Conn.), Zyomyx (Hayward, Calif), Phylos (Lexington, Mass.) and Biacore (Uppsala, Sweden). Examples of such protein bio chips are described in the following patents or published patent applications: U.S. Pat. No. 6,225,047; PCT International Publication No. WO 99/51773; U.S. Pat. No. 6,329,209, PCT International Publication No. WO 00/56934 and U.S. Pat. No. 5,242,828.

[0106] The antibodies and oligonucleotides can be immobilized onto a variety of solid supports, such as magnetic or chromatographic matrix particles, the surface of an assay place (such as microtiter wells), pieces of a solid substrate material or membrane (such as plastic, nylon, paper), and the like. An assay strip could be prepared by coating the antibody or a plurality of antibodies in an array on solid support. This strip could then be dipped into the test sample and then processed quickly through washes and detection steps to generate a measurable signal, such as a colored spot.

[0107] The invention has been described in the foregoing specification with reference to specific embodiments. It will however be evident that various modifications and changes may be made to the embodiments without departing from the broader spirit and scope of the invention. The specification and drawings are to be regarded in an illustrative rather than a restrictive sense. The invention is illustrated herein by the experiments described by the following examples, which should not be construed as limiting. The contents of all references, pending patent applications and published patents, cited throughout this application are hereby expressly incorporated by reference. Those skilled in the art will understand that this invention may be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will fully convey the invention to those skilled in the art. Many modifications and other embodiments of the invention will come to mind in one skilled in the art to which this invention pertains having the benefit of the teachings presented in the foregoing description. Although specific terms are employed, they are used as in the art unless otherwise indicated. EXAMPLES Example 1

Materials and Methods

[0108] Study Design: The study design is shown in Fig. 1. The present study was designed to test the hypothesis that molecular processes of aging and senescence distinguish indolent versus aggressive prostate cancer (Fig. 1). This hypothesis was tested by first assembling a 377-gene signature of aging and cellular senescence, which was used to query human cancer profiles (Table 1), as well as using a mouse model of indolent prostate cancer using GSEA. This resulted in the identification of a 19-gene indolence signature, which was then used to perform decision- tree learning using an independent human cohort to identify a 3 -gene prognostic panel that was validated at the mRNA and protein levels using independent cohorts, and then validated on biopsies from patients on active surveillance.

[0109] Statistical methods: K-means clustering was done using the "kmeans" function from the Statistical toolbox in MATLAB. For confusion matrices, accurate predictions were calculated for indolent or lethal clusters and combined to calculate an Odds Ratio. Kaplan-Meier analyses were conducted using the MATLAB script; ^-values were computed using a log-rank test. The overall C-index (54), confidence intervals, and corresponding ^-values were calculated using the survcomp package of R. The predicted probability of survival for computing C-index was obtained through the multivariate Cox proportional hazards models.

[0110] Immunohistochemical analyses: All studies involving human subjects were approved by the Institutional Review Board of Columbia University Medical Center. Tissue microarrays (TMAs) were comprised of primary prostate tumors obtained from the Herbert Irving

Comprehensive Cancer Center Tissue Bank (Table 1). Biopsy samples were obtained from patients seen in the Department of Urology at Columbia University Medical Center from 1992 to 2012. Immunohistochemical analyses were performed using: anti-FGFRl (Abeam, Cat# abl0646); anti-PMP22 (Sigma, Cat# #P0078); and anti-CDKNlA (BD Pharmingen,

Cat#556431). The percentage of positive tumor cells (0% to 100%) and staining intensity (0-2) were assessed for each cores or biopsy, and composite scores were generated. [0111] Computational methods: The 377-gene signature of aging and cellular senescence was assembled from the following sources: (i) Meta-profile analyses (22); (ii) Ingenuity pathway analysis [http://www.ingenuity.com/]; and (iii) manual curation (50-52). A complete description of the 377-gene set is provided in Table 1. GSEA was performed described (53). Integrative p- values were calculated using Fisher's combined probability test. The decision-tree learning algorithm was run by selecting the "classification" method from the "classregtree" function (MATLAB, Statistical toolbox).

[0112] Curation of the aging and senescence signature: The following resources were used to compile a 377-gene set associated with biological processes of aging and cellular senescence:

(i) Meta-profile analyses of 27 datasets from mouse, rat, and human samples (336 genes) (22);

(ii) Ingenuity pathway analysis for senescence related genes (44 genes)

[http://www.ingenuity.com/]; and (iii) manual curation of senescence-related genes (3 genes) (50-52). A complete description of the 377-gene set is provided in Table 1.

[0113] Datasets used: Gene expression profile datasets used in this study are from: (i) Yu et ah. primary human prostatectomy samples (n = aggressive tumors used in this study) with adjacent normal tissue (n = 58), on a Affymetrix U95a, U95b and U95c microarray platform (25);

(ii) Taylor et ah. primary human prostatectomy samples with adjacent normal tissue (n = 131 tumor; 95 Gleason 6 and 7(3 +4); n = 23 adjacent normal), on a Affymetrix human Exon 1.0 ST microarray platform (14); (iii) Sboner et al (also called the Swedish cohort): primary human prostate tissue from transurethral resection of the prostate (TURP) (n = 281; Training set used was 25 indolent and 29 lethal; test set used was 28 indolent and 8 lethal), on a 6K DASL microarray platform (33); (iv) Ouyang et ah. prostate tissues from x3.7 homozygous null and wild- type mice (n = 9 total mice in each group), on a Affymetrix Mu74AV2 microarray platform (37); (v) TCGA breast cancer dataset: invasive breast carcinomas and normal breast tissue (n = 354), on an Agilent G4502A microarray (27); and (vi) Lung cancer dataset: lung tumors and normal lung tissue (n = 190), on an Affymetrix human U95A microarray platform (26).

Available clinicopathological information for the specific patients/samples used in this study is provided in Table 1 and Table 2.

[0114] Data normalization: Normalized data was available for the Taylor et al, Sboner et al, breast cancer, and lung cancer datasets. For the Ouyang et al and Yu et al datasets, expression intensities were background-corrected, normalized, and summarized using the Gene Chip Robust Multiarray Algorithm (GC-RMA) (55) in the R/Bioconductor GCRMA package (5(5).

[0115] Differential expression: Differentially expressed genes were identified using Student's t-test by running "ttest2" command in MATLAB ^® . For comparing across platforms genes rather than probes were evaluated; if multiple probesets were present for a gene, the probe with the highest absolute differential expression between tumor and normal was selected. For cross- species comparison, mouse genes were first mapped to their human orthologs using the sequence-based method available from NCBI HomoloGene

(http://www.ncbi.nlm.nih.gov/pubmed/21097890 and

http ://www.ncbi.nlm.nih.gov/books/NBK21083/#A866) .

[0116] Gene set enrichment analysis: For Gene Set Enrichment Analysis (GSEA) (53, 57) genes were ranked by computing their differential expression in the tumor versus normal samples using the Student's t-test method. Sample shuffling (human datasets) or gene shuffling (mouse dataset) with 1,000 shuffles allowed estimation of p-values with an accuracy of up to 1 X 10 ^" . A list of the leading and lagging edge genes is provided in Table 3.

[0117] Integrative p-value analyses: To compute the integrative p-value for the metaanalyses, first GSEA WAS performed on each of the datasets individually. Then the Fisher's combined probability test (also known as Fisher's method) was used to integrate p-values.

Fisher's method is computed as follows:

where n is the number of p-values pi and X ² is a variable that follows a chi-squared distribution with 2n degrees of freedom under the hypothesis of no enrichment. Genes with integrated p- values below 0.05 are listed in Table 4. The criteria for inclusion of a given gene in the metaanalyses of the human cancers were as follows: (i) must be present in the lagging edge of prostate cancer dataset; (ii) must be present in the lagging edge of at least one of the other human datasets (i.e., lung or breast); and (Hi) must have an integrative p-value <0.05.

[0118] The 19 Gene Indolence signature: The 19-gene indolence signature was generated from the intersection of genes from the meta-analyses of human cancers (68 genes) and those in the leading edge from the GSEA of the indolent prostate cancer mouse model (73 genes). A description of the 19-gene indolence signature is provided in Table 5.

[0119] Decision-tree learning model: The decision-tree learning algorithm was run by selecting the "classification" method from the "classregtree" function (MATLAB, Statistical toolbox). The expression of each gene was discredited into 3 states (up, normal, and down) by comparing the expression in each sample to the average expression across all samples. Genes whose expression in a sample was e,- > μ + σ/2 where μ is the average expression and σ is the standard deviation were assigned an "up" value, while those whose expression was e,- < μ + σ/2, were assigned a "down" value; the remaining samples were assigned a "normal" value. In the first step, individual genes were identified whose expression state was significantly predictive of the relative covariate (i.e., indolence or lethality) (p < 0.05). The expression state of these genes was used to partition the patients. Then, 2-gene combinations were formed by combining each gene from the previous step (e.g., A) with any additional gene (e.g., B) from the remaining 18 genes in the signature (or the same gene with a different expression state). The 2-gene combinations that significantly improved predicted outcome classification over the

corresponding single gene classifier were selected (i.e., AB should predict outcome better than A alone; p < 0.05, to be selected). This process was repeated iteratively by adding more genes, one at the time, to the predictive combinations (i.e., a new branch in the classification tree), up to a maximum of 4 genes. However, the tree pruning method revealed that more than 3 gene combination leads to over fitting suggesting that 3 genes is the optimal number with highest predictive value.

[0120] The combinations from the decision tree were verified using a 5-fold cross-validation procedure using the Sboner et al training set (Table 2). For the 5-fold cross-validation, 44 patient samples (i.e., 4/5 ^th of test set) were chosen at random for training and the remaining 11 samples (1/5 ^Λ of the test set) were used to test the trained classifier performance. Gene combinations were ranked based on those with the minimum cross-validation error. A summary of the top combinations from the decision tree is provided in Table 6.

[0121] Computational methods are documented in a S WEAVE documents . Statistical methods for validation

[0122] K-means Clustering: X-means clustering algorithm (58) was run using the "kmeans" function from the Statistical toolbox in MATLAB with n = 2 clusters and default values for the remaining parameters.

[0123] Confusion matrices: Accuracy of predictions were calculated by identifying patients within a test set from Sboner et al, which were correctly assigned to indolent (n = 26) or lethal (n = 9) clusters, as well as the number of incorrect predictions. These numbers were combined to calculate an Odds Ratio to assess the predictive accuracies.

[0124] Kaplan-Meier: Kaplan-Meier analyses for survival difference of patient clusters, partitioned using X-means clustering, were conducted using the MATLAB script; ^-values were computed using a log-rank test.

[0125] Prognostic models: The overall C-index (54), confidence intervals, and corresponding p- values were calculated using the survcomp package of R. The predicted probability of survival for computing C-index was obtained through the multivariate Cox proportional hazards models. Statistical methods are documented in a SWEAVE documents.

Example 2

Methods for isolating protein and mRNA for the Ouyang, et al. dataset: Cancer Res 2005; 65: (15). August 1, 2005.

[0126] To further minimize variability from individual specimens, prostate tissues from three independent animals were pooled to generate RNA for each array and a minimum of three arrays were probed for the wild-type and mutant mice (thus allowing comparison of a total of nine mice for each). RNA was extracted using Trizol (Invitrogen, Carlsbad, CA) and purified using an RNeasy kit (Qiagen, Chatsworth, CA). cDNA was labeled using a Bio Array High- Yield RNA transcript labeling kit (Enzo Life Sciences, Farmingdale, NY) and hybridized to Affymetrix GeneChips (Mu74AV2). For statistical analyses, initial data acquisition and normalization was done using Affymetrix Microarray Suite 5.0 software followed by an ANOVA test. Validation of gene expression changes by quantitative reverse transcription- PCR was done using an Mx4000 Multiplex Quantitative PCR system (Stratagene, La Jolla, CA). Validation to tissue sections was done by in situ hybridization or immunohistochemistry as described, depending on the availability of antisera. For Western blot analyses, anterior prostate tissues were snap-frozen on liquid nitrogen and protein extracts were made by sonication in buffer containing 10 mmol/L Tris-HCl (pH 7.5), 0.15 mol/L NaCl, 1 mmol/L EDTA, 0.1% SDS, 1% deoxycholate (sodium salt), 1% Triton X-100, with freshly added protease inhibitor and phosphatase inhibitor cocktail (Sigma, St. Louis, MO). For in situ hybridization, sequence-verified expressed sequence tag clones were purchased from Invitrogen.

Example 3

Methods for isolating protein and mRNA for the Yu dataset:

[0127] A comprehensive gene expression analysis was performed on 152 human prostate samples, including prostate cancer (PC), prostate tissues adjacent to (AT) cancer, and donor (OD) prostate tissue totally free of disease, using the Affymetrix (Santa Clara, CA) U95a, U95b, and U95c chip sets. A set of 671 genes were identified whose expression levels were

significantly altered in PCs compared with normal tissues. Interestingly, the expression patterns of histological benign prostate tissues were significantly overlapped with those of PC, and were distinctly different than donor prostate tissue. Separately, a "70-gene" model was developed to predict the aggressiveness of the disease. Collectively, these data suggest that genetic alterations in a gland with PC are not limited to the malignant cells, and these patterns of alteration may predict the population both at risk for the disease and for disease progression.

[0128] Sample Preparation: Fresh prostate tissues, recovered immediately from the operating room after removal, were dissected and trimmed to obtain pure tumor (completely free of normal prostate acinar cells) or normal prostate (free of tumor cells) tissues. Microdissection was coupled with sandwich frozen and permanent section analyses to confirm the purity and homogeneity of the samples: gross and microscopic analyses were performed by board-certified genitourinary pathologists. For tumor tissues, only samples with less than 30% of stromal components were selected. For donor prostate tissues, obtained at the time of organ donation in brain-dead men, samples from peripheral zone of the prostate gland with at least 60% glandular components and free of any pathological alteration were selected For prostate tissues adjacent to cancer, samples free of cancer cells, high-grade prostatic neoplasia, or any obvious neoplastic alterations, containing at least 60% glandular cells, were selected. Whenever possible, all tissues were processed and frozen within 30 minutes after removal. These tissues were then homogenized. All patients with PCs have at least a 4-year follow-up, with regular evaluations for relapse or the presence of metastasis. Protocols for tissue banking, tissue anonymization, and tissue processing, were approved by the institutional review board.

[0129] Affymetrix Chip Analysis cRNA preparation: Total RNA was extracted and purified with Qiagen RNeasy kit (Qiagen, San Diego, CA). Five micrograms of total RNA were used in the first strand cDNA synthesis with T7-day(T)24 primer

(GGCCAGTGAATTGTAATACGACTCACTATAGGGAGGCGG-(dT)24) by Superscript II (GIBCO-BRL, Rockville, MD). The second strand cDNA synthesis was carried out at 16°C by adding Escherichia coli DNA ligase, E coli DNA polymerase I, and RnaseH in the reaction. This was followed by the addition of T4 DNA polymerase to blunt the ends of newly synthesized cDNA. The cDNA was purified through phenol/chloroform and ethanol precipitation. The purified cDNA were then incubated at 37°C for 4 hours in an in vitro transcription reaction to produce cRNA labeled with biotin using MEGAscript system (Ambion Inc, Austin, TX).

Affymetrix chip hybridization. Between 15 and 20 _g of cRNA were fragmented by incubating in a buffer containing 200 mmol/L Tris-acetate, pH8.1, 500 mmol/L KOAc, and 150 mmol/L MgOAc at 95 °C for 35 minutes. The fragmented cRNA were then hybridized with a pre- equilibrated Affymetrix chip at 45°C for 14 to 16 hours. After the hybridization cocktails were removed, the chips were then washed in a fluidic station with low-stringency buffer (6_ sodium chloride, sodium phosphate dibasic, and EDTA; 0.01% Tween 20; 0.005% antifoam) for 10 cycles (two mixes/cycle), and stringent buffer (100 mmol/L MES, O.lMNaCl and 0.01% Tween 20) for four cycles (15 mixes/cycle), and stained with Strepto-avidin Phycoerythrin (SAPE; Molecular Probe, Eugene, OR). This was followed by incubation with biotinylated mouse antiavidin antibody, and restained with SAPE. The chips were scanned in aHPChipScanner (Affymetrix Inc) to detect hybridization signals. For quality assurance, all samples were run on Affymetrix test-3 chips to evaluate the integrity of RNA; samples with RNA 3_/5_ ratios less than 2.5 were accepted for further analysis.

[0130] Data analysis: Hybridization data were normalized to an average target intensity of 500 per chip, and were converted to Microsoft Excel spreadsheet text file (Redmond, WA). The primary comparison of OD to PC was conducted through the following steps: (1) Two sample t tests of log-transformed gene expression values, (2) adjustment of P values through the

Benjamini and Hochberg procedure, (3) selection of genes that meet both the critical P value and show at least a two-fold change in PC, (4) reduction of dimensionality through principal component analysis, (5) prediction of case status (ie, normal v cancer tissue) through logistic regression, and (6) evaluation of the classification rate using 10- fold cross-validation. Regarding the second step, the Benjamini and Hochberg procedure calculates a conservative P value to minimize the expected number of falsely significant results. For tests between PC and AT, the paired t test (of log-transformed expressions) was utilized to account for the matching. A sufficient number of principle components (in the fourth step) were retained to quantify at least 90% of the variability in these genes. For the cross validation procedure (sixth step), a separate logistic model is fit for each of the ten subsets used for training, and then used to predict the outcome for the remaining subset of validation data. After this process is implemented for classifying donors versus PC, the resulting model parameters (using the entire data set) were saved and utilized to predict case status of adjacent to tumor normal tissue. The fitted logistic model (again using the entire data set) was also used to classify separate validation data sets collected from other institutions. These analyses were all conducted using S-PLUS statistical software (Insightful Corp, Seattle, WA).

Example 4

Methods for isolating protein and mRNA for the Sboner dataset, BMC Medical Genomics 2010, 3:8:

[0131] Patient population: This present study is nested in a cohort of men with localized prostate cancer diagnosed in the Orebro (1977 to 1994) and South East (1987 to 1999) Health Care Regions of Sweden. Eligible patients were identified through population- based prostate cancer quality databases maintained in these regions (described in Johansson et al., Aus et al., and Andren et al. and included men who were diagnosed with incidental prostate cancer through (TURP) or adenoma enucleation, i.e. stage Tla-b tumors. In accordance with standard treatment protocols at the time, patients with early stage/localized prostate cancer were followed expectantly ("watchful waiting"). No PSA screening programs were in place at the time. The study cohort was followed for cancer-specific and all cause mortality until March 1, 2006 through record linkages to the essentially complete Swedish Death Register, which provided date of death or migration. Information on causes of death was obtained through a complete review of medical records by a study end-point committee. Deaths were classified as cancer-specific when prostate cancer was the primary cause of death. Tumor tissue specimens were traced from 92% (1256/1367) of all potentially eligible cases. In order to provide complete and consistent information, available hematoxylin and eosin (H&E) slides from each case were reviewed to identify all tissue specimens with tumor tissue. Slides and corresponding paraffin-embedded formalin- fixed blocks were subsequently retrieved and rereviewed to confirm cancer status and to assess Gleason score and other notable histopathologic features. The reviewers were blinded with regard to disease outcome. Gleason score was evaluated according to Epstein et al. All patients gave informed consent for the study. Since our overarching aim was to identify signatures predicting a lethal or an indolent course of prostate cancer, efficiency was maximized by devising a study design that included men who either died from prostate cancer during follow up (lethal prostate cancer cases) or who survived at least 10 years after their diagnosis (men with indolent prostate cancer). Thus men with non-informative outcomes were excluded, namely those who died from other causes within ten years of their prostate cancer diagnosis or had been followed for less than 10 years with no disease progression (n = 595). All men with samples in which high-density tumor regions (defined as more than 90% tumor cells) could be identified were included (n = 381). Men who had received any type of androgen deprivation treatment during follow up (n = 79) were excluded from the indolent group, since some of these had potentially lethal disease that was deferred by therapy. Twenty-one men were further excluded due to poor sample quality. In total, 281 men (116 with indolent disease and 165 with lethal prostate cancer) were included in the analyses. The study design was approved by the Ethical Review Boards in Orebro and Linkoping. The clinical and pathologic demographics of these of 281 men with prostate cancer are presented. In addition to the standard pathology evaluation each case was also characterized with respect to ERG gene rearrangement, since it appears that this event is an indicator of poor prognosis .

[0132] Complementary DNA-mediated annealing, selection, ligation, and extension array design: An array of 6100 genes (6K DASL) was designed for the discovery of molecular signatures relevant to prostate cancer by using four complementary DNA (cDNA)-mediated annealing, selection, ligation, and extension (DASL) assay panels (DAPs) See Gene Expression Omnibus (GEO: http:// www.ncbi.nlm.nih.gov/geo/ with platform accession number: GPL5474. This data set is also available at GEO with accession number: GSE16560. Example 5

Taylor dataset: Cancer Cell 18, 11-22, July 13, 2010 Cancer Cell 18, 11-22, July 13, 2010

[0133] Specimen collection and annotation: A total of 218 tumor samples and 149 matched normal samples were obtained from patients treated by radical prostatectomy at Memorial Sloan- Kettering Cancer Center. All patients provided informed consent and samples were procured and the study was conducted under Memorial Sloan-Kettering Cancer Center Institutional Review Board approval. Clinical and pathologic data were entered and maintained in our prospective prostate cancer database. Following radical prostatectomy, patients were followed with history, physical exam, and serum PSA testing every 3 months for the first year, 6 months for the second year, and annually thereafter. For all analyses described here, biochemical recurrence (BCR) was defined as PSA R0.2 ng/ml on two occasions. At the time of data analysis, patient follow-up was completed through December 2008.

[0134] Analyte extraction and microarray hybridization: DNA and RNA were extracted from dissected tissue containing greater than 70% tumor cell content as well as from seven cell lines and seven xenografts (see Supplemental Information). Resulting DNA and RNA were hybridized to Agilent 244K array comparative genomic hybridization (aCGH) microarrays, Affymetrix Human Exon 1.0 ST arrays, and/or Agilent microRNA V2 arrays, respectively. The normalization and statistical analysis of both DNA copy-number and expression array data are available in the Supplemental Information.

[0135] DNA sequencing: In total, 251 million bases in coding exons and adjacent intronic sequences of 138 cancer-related genes in 91 samples were PCR-amplified and sequenced by Sanger capillary sequencing. Ninety- five sites of known mutation in 22 genes were also genotyped using the iPLEX Sequenom platform. The details of whole-genome amplification, sequencing, mutation detection pipelines, mutation validation, background mutation rate analysis, and Sequenom genotyping are described in the Supplemental Information.

[0136] Outlier expression analysis: Outlier profiles for all transcripts and outlier assignments in all tumors were determined from normalized expression data as previously described (Ghosh and Chinnaiyan, 2009). In brief, in this nonparametric approach an empirical distribution function generated from transcript expression in the 29 normal prostate tissues was used to transform expression in the tumor samples, from which outliers were determined with the criteria described in the Benjamini and Hochberg algorithm (Benjamini and Hochberg, 1995) at an error rate (a) = 0.01.

Example 6 Validation of 3-gene prognostic panel by immunohistochemistry

[0137] Immunohistochemical analyses: All studies involving human subjects were approved by the Institutional Review Board of Columbia University Medical Center. Tissue microarrays (TMAs) were comprised of primary prostate tumors obtained from the Herbert Irving

Comprehensive Cancer Center Tissue Bank from 121 radical prostatectomy specimens

(including 44 that were Gleason 6 or Gleason 7 (3+4)) with 102 adjacent normal tissues as controls (Table 1). The TMA was constructed (Beecher Instruments, MD, USA) by punching triplicate cores of 1 mm for each sample.

[0138] Immunohistochemical analyses were performed using: anti-FGFRl (Abeam, Cat# abl0646); anti-PMP22 (Sigma, Cat# #P0078); and anti-CDKNIA (BD Pharmingen,

Cat#556431). The percentage of positive tumor cells (0% to 100%) and staining intensity (0-2) were assessed for each cores or biopsy, and composite scores were generated.

[0139] A cohort of retrospective biopsy samples were obtained from patients enrolled in a surveillance protocol in the Department of Urology at Columbia University Medical Center from 1992 to 2012. Patients included in the surveillance protocol presented with low risk prostate cancer with the following essential criteria: normal digital rectal exam (DRE), serum PSA <10 ng/ml, biopsy Gleason score <6 in no more than 2 cores, and cancer involving no more than 50% of any core on at least a 12-core biopsy. The current protocol to monitor these patients includes DRE and serum PSA testing every three months, and repeat biopsy every 12 or 18 months, or "for-cause biopsy" if any sign of progression (abnormal DRE, increasing PSA) becomes evident. Biopsy samples were immunostained and scored using the protocol outlined above.

[0140] Immunohistochemical analyses were performed using a rabbit polyclonal anti-FGFRl antibody (Abeam, Cat# ab 10646) at a concentration of 1 μ§/ηι1; a rabbit polyclonal anti-PMP22 (Sigma, Cat# #P0078) at 1 μ§/ηι1; and a mouse monoclonal CDKN1 A (BD Pharmingen, Cat#556431) at 500 μ§/ηι1. Controls for antibody specificity are shown in Fig. 11. Slides were deparaffinized in xylene, followed by antigen retrieval through boiling for 37 minutes at 100°C in Decloaking Solution (Citrate buffer, pH 6.0, Biocare Medical) in a pressure cooker. Following cooling, slides were incubated in 3% H ₂ 0 ₂ and then blocked in 10% goat serum for rabbit primary antibodies or 10% horse serum for mouse primary antibodies. Following overnight incubation in primary antibody, slides were washed in PBS containing 0.05% Triton X-100 and then incubated for 1 hour at room temperature with biotinylated anti-rabbit or anti-mouse secondary antibody (Vector Laboratories.) The signal was amplified by Vectastain ABC system (Vector Laboratories, PK6200) and visualized with the NovaRed Substrate Kit (Vector

Laboratories, SK4800). Slides were counterstained with Harris Modified Hematoxylin (1 :4 diluted in H20) (Fisher Scientific) and coverslipped with Clearmount (American Master*Tech Scientific). Negative and positive controls for each of the antibodies were used in parallel to assure antibody specificity (Fig. 11).

[0141] Stained slides were scanned using an Olympus BX61Whole Slide scanner. For

CDKN1 A nuclear expression was evaluated; for FGFR1 and PMP22 both nuclear and cytoplasmic/cell surface expression were analyzed. Scoring was performed without knowledge of the clinico-pathological variables. The percentage of positive tumor cells (from 0% to 100%), as well as staining intensity was assessed for each of the cores. For intensity, values were assigned on a three-point scale: 0 represents no staining, 1 represents a mild to moderate positivity and 2 represents an intense immunoreaction. Composite scores were generated by multiplying the percentage of positive cells and staining intensity; the mean score for each patient from the triplicate cores was used for X-means clustering to identify low-risk and high- risk groups based on the three proteins in classifier.

Example 7

Methods for phenotypic analyses of Nkx3.1 mutant mice:

[0142] The NL·3.1 germline mutant mice have been described previously (28). Wild- type and null littermates were sacrificed for analyses at 4-month intervals from 3 to 24 months of age. For histological and immunohistochemical analyses, tissues were fixed in 10% formalin and analyses done as described previously (59). For SA- -Gal analysis, freshly dissected (unfixed) prostatic tissues were cryopreserved in Optimal Cutting Temperature (OCT) compound and stained using the Chemicon SA- -GAL kit (KAA002) following the manufacturer's instructions. For protein extraction, tissues were snap-frozen in liquid nitrogen, and processed for western blot analyses as described (59). Antibodies used in the mouse analyses were as follows: mouse monoclonal ΗΡΙγ (clone 2MOD-1G6) (EMD Millipore, Cat no. MAB3450); rabbit polyclonal Ab Ki67 (Novacastra/Leica, Cat no. NCL-Ki67p); rabbit polyclonal Ab GADD45alpha (Cell Signaling Technology, Cat no. 3518S); mouse monoclonal Ab PML clone 36.1-104 (Millipore, Cat no. 05- 718), rabbit polyclonal Ab BECN1 (H-300) (Santa Cruz, Cat no. sc-11427) and rabbit monoclonal Ab B-Actin (13E5) (Cell Signaling Cat no 4970).

[0143] Level of Evidence: The current study falls into the Level of Evidence category D as it is a retrospective, observational study that involves multiple independent datasets. A REMARK

Example 8 An "indolence gene signature" of aging and senescence distinguishes indolent versus aggressive prostate cancer

A. Identification of a gene signature for prostate cancer that is associated with aging and senescence

[0144] A first step was the generation of a literature-, pathway-, and manually-curated 377 gene signature associated with aging and senescence (Fig. 1, Step 1; Table 1). This gene signature was primarily assembled from a meta-analyses of aging-related genes (22), and accordingly was enriched for biological pathways associated with various aging-associated diseases, while it had limited enrichment for pro-tumorigenic pathways such as those associated with cellular proliferation. Notably, the 377-gene signature had virtually no overlap with previously identified signatures associated with cellular proliferation (23, 24).

[0145] Gene set enrichment analyses (GSEA) was next done to evaluate whether this signature of aging and senescence was enriched in genes down-regulated in aggressive human prostate cancer and, conversely, up-regulated in indolent prostate cancer (Fig. 1, Step 2). These analyses were extended to infer that the intersection of the genes enriched among those down-regulated in aggressive human prostate cancer (i.e., the lagging edge) and up-regulated in indolent prostate cancer (i.e., the leading edge) would identify those most closely associated with indolence (i.e., an "indolence signature", Fig. 1, Step 2). For these and subsequent analyses, published expression profiling datasets were used, either to discover or refine genes for classification purposes (training sets), or to validate their statistical power and performance (test/validation sets), but never for both purposes (Fig. 1, Table 1). [0146] To evaluate the expression of the 377-gene signature of aging and senescence in aggressive prostate cancer, GSEA analyses using the Yu et al dataset was performed, which includes a subset of aggressive, locally invasive prostate tumors (n = 29) with adjacent normal prostate tissue (n = 58) as controls (25) (Table 1; Table 2A). Consistent with the hypothesis, the 377-gene signature was enriched among genes down-regulated in these aggressive prostate tumors compared with the normal controls (NES = -1.87; p < 0.001) (Fig. 2A; Table 3A).

Interestingly, additional epithelial cancers, lung and breast (the references for the gene sets used for lung and breast are published datasets described in 26, 27) also showed significant enrichment of this indolence signature among genes down-regulated in aggressive tumors (NES = -1.90 and -1.52, respectively; p < 0.001 in both cases) (Fig. 5A; Table 3B,C). Meta-analysis of the down-regulated {i.e., lagging-edge) genes from the prostate, lung, and breast tumors led to the refinement of the original 377 gene signature to a subset of 68 genes that were most significantly enriched in aggressive tumors (Table 4A). These findings support the hypothesis that genes associated with aging and senescence are enriched among down-regulated genes in aggressive prostate cancer, as well as other epithelial cancers.

B. Cross-species analysis identifies a 19-gene "indolence signature;" Nkx3.1 homozygous mutant mice are a relevant model of indolent prostate cancer.

[0147] Since the 377-gene set is enriched for genes down-regulated in aggressive prostate cancers (Fig. 2A), it was expected that the most informative genes in this signature should be up- regulated in indolent prostate tumors. However, independent human datasets containing purely indolent prostate tumors were not available to evaluate this hypothesis. Therefore, as a source of purely indolent prostate lesions, cross-species analyses was performed using a well-characterized mouse model of pre-invasive prostate cancer, which is based on germline loss-of-function of the x3.7 homeobox gene (28, 29). Notably, this cross-species approach, which uses enrichment analyses of relatively homogenous mouse model to "filter" the characteristically heterogeneous human prostate tumors, also enabled identification of the most conserved and relevant genes among the signature.

[0148] Human NKX3.1 is localized to a chromosomal hotspot, 8p21, which is frequently lost in prostate intraepithelial neoplasia (PIN) and prostatic intraepithelial neoplasia (PIN). Down- regulation of Human NKX3.1 expression is associated with cancer initiation, although it is not sufficient for overt carcinoma (30). Targeted inactivation of NL·3.1 in mice leads to PIN, which does not progress to adenocarcinoma even in aged mice (28, 29) (Fig. 6A-D). Further, this age- associated arrest in cancer progression in the Nkx3.1 mutant mice is coincident with elevated cellular senescence and abrogation of cellular proliferation (Fig S2E-I). Since the Nkx3.l mutant mice develop pre-invasive prostate lesions with an aging-associated halt in tumor progression that is coincident with cellular senescence, it was hypothesized that they would provide a relevant model of indolent prostate cancer.

[0149] GSEA was performed using expression profiles from aged Nkx3.l homozygous mutant and control (age-matched) wild-type mouse prostates (n=9/group) (37). Whereas the 377-gene signature was enriched for genes down-regulated in the aggressive prostate tumors (i.e., in the lagging edge) (see Fig. 2A), the indolent prostate lesions were enriched for the up-regulated genes (i.e., in the leading edge) (NES = 1.81; p < 0.001) (Fig. 2B; Table 3D). Therefore it was hypothesized that the intersection of genes down-regulated in aggressive human tumors (i.e., the 68 genes from the meta-analysis of human cancers) and those up-regulated in the indolent prostate lesions from the x3.7 mice (i.e., the 73 genes from the leading edge) would identify the most consistently regulated genes for an effective indolence classifier (Fig. 2C). As predicted, these analyses identified 19 genes that are significantly up-regulated in indolent prostate cancer and down-regulated in aggressive tumors; herein the 19-gene "indolence signature" (Fig. 2C; Table 5). This intersection is highly statistically significant (p < 0.001, by Fisher Exact Test), suggesting that these genes are under coordinated regulation in the aggressive and indolent tumors, and are thus well-suited for classification of these states. Taken together, these findings show that genes associated with aging and senescence can be used to distinguish prostate cancers according to aggressive versus indolent behavior.

C. Gene signature of aging and senescence distinguishes disease outcome of low Gleason score prostate cancer

[0150] The Taylor et al dataset was used to independently validate these observations; it is one of the few publicly available human datasets with extensive clinical outcome data (14) (Table 1). Taylor et al contains a substantial number of prostatectomy samples (n = 131) with adjacent normal controls (n =23) from patients that encompass a wide range of Gleason scores and times to biochemical recurrence (14) (Table 1; Table 2B). This dataset includes a significant number (n = 13) of aggressive prostate tumors (i.e., Gleason 8,9) with a short time to biochemical recurrence (< 22 months) (Table 1; Table 2B). GSEA analyses of these high Gleason grade tumors demonstrated their similar behavior to the aggressive tumors from Yu et al, since the 377- gene signature was significantly enriched for genes down-regulated in these aggressive prostate tumors (NES =— 2.60 and p < 0.001), including most (18/19) of the 19-gene indolence signature (Fig. 2D; Table 5). Therefore, both the behavior and specific enrichment of the 377-gene signature was conserved in an independent dataset of aggressive human prostate cancer.

[0151] The Taylor et al. dataset also contains a substantial number of low Gleason score tumors (i.e., Gleason 6; n = 41; and Gleason score 7(3 + 4); n = 54) with varying times of progression to biochemical recurrence (BCR) ranging from >100 months (i.e., indolent) to <35 months (i.e., aggressive) (Table 1; Table 2B). Experiments were conducted to recapitulate the differential enrichment of the 377-gene signature in the indolent versus aggressive tumors by limiting the sample to only to low Gleason score prostate tumors (Fig. 2E-F; Fig. 6B). These and most subsequent analyses focused primarily on Gleason score 6 tumors, but (for increased statistical power) the subset of Gleason score 7 tumors that were scored as 3+4 (refer to these combined Gleason 6 and Gleason 7(3+4) as "low Gleason score tumors") were also included. Interestingly, it was consistent in the molecular analyses that Gleason 7 tumors scored as 3+4 behaved more like the Gleason score 6 tumors, while those scored as 4+3 behaved more like the more advanced Gleason Score tumors, which is in agreement with a recent study by Balk and colleagues showing that Gleason 3 and 4 lesions have different molecular features and progressive potential (32).

[0152] First, GSEA was performed on the low Gleason Score prostate tumors to evaluate enrichment of the 377-gene signature of aging and senescence in the two extreme patient groups (i.e., the most lethal versus the most indolent). In particular, the first group included patients with a short time to biochemical recurrence (i.e., the aggressive group, Gleason score 6 and Gleason score 7(3 + 4) tumors having BCR < 35 months; n = 5) and the second included patients that did not recur within the considerable follow-up period of greater than 100 months (i.e., the indolent group, Gleason score 6 and Gleason score 7(3 + 4) tumors BCR > 100 months; n = 5) (Fig. 2F; Table 2B). GSEA analyses demonstrated that the 377-gene signature was enriched in genes up- regulated in the indolent group (BCR > 100 months), with a positive NES score (NES=1.52 p value<0.001), whereas it was enriched in genes down-regulated in the aggressive group (BCR < 35 months), with a negative NES score (NES= -1.85, p valueO.001 Fig. 2F; Table 3E,F).

[0153] Enrichment of the 377-gene signature was further assessed in indolent versus aggressive low Gleason score tumors focusing only on the Gleason score 6 patients. In particular, the Gleason score 6 patients were partitioned into subgroups representing varying interval to biochemical recurrence: > 0 months (n = 41); >35 months (n=32), >50 months (n=20), >65 months (n=8), >80 months (n=5), > 100 months (n=3), and then GSEA was performed on each of these sub-groups. Strikingly, while all of the sub-groups displayed enrichment of the 377-gene signature, the direction of the enrichment was dependent on the interval to biochemical recurrence (Fig. 2E). In particular, Gleason grade 6 tumors with a longer interval to biochemical recurrence (> 65, > 80, and > 100 months) were enriched in the leading edge (and had a positive NES score), while those with a shorter interval to recurrence (> 0, > 35, > 50 months) were enriched in the lagging edge (and had a negative NES score) (Fig. 2E; Fig. 5 IB).

[0154] Taken together, these GSEA show that differential enrichment of a signature of aging and senescence can distinguish low Gleason score tumors that are destined to remain indolent from those destined to become aggressive. Furthermore, meta-analyses of the leading and lagging edge genes in these indolent versus aggressive sub-groups of Gleason 6 tumors included a majority of the 19-gene "indolence signature" among those that were significant (14/19 genes; Table 5). Taken together, these findings demonstrate that low Gleason score prostate tumors can be distinguished as indolent or aggressive based on enrichment for a gene signature of aging and senescence and constitute an independent validation of the indolence signature.

D. A 3-gene prognostic biomarker panel low Gleason score prostate tumors

[0155] Notably, while the 19-gene indolence signature is differentially enriched in indolent versus aggressive sub-types, it was not sufficient to stratify patient patients using Kaplan Meier analyses (Fig. 9A). Thus, it was important to identify a minimal subset(s) of genes among those in the 19-gene indolence signature that most effectively predicts clinical outcome for low Gleason score prostate tumors. A decision-tree learning model was used to evaluate gene combinations among the 19-gene signature that best distinguish indolent versus lethal prostate tumors (Fig. 1, Step 3; Fig. 3 A). The decision-tree model iteratively partitions patients according to the expression state of the gene with the highest predictive value, considering both synergistic and antagonistic affects between genes, and terminating once further partitioning has no additional statistical predictive value. Each leaf node in the resulting predictive tree corresponds to a set of patients with predicted prognostic outcome; each branch corresponds to the expression state of a predictive gene, and a walk from the root of the tree to a leaf node reveals the expression state of the gene panel used to predict outcome at the leaf node.

[0156] Decision-tree analyses was done using an independent dataset, namely the Swedish "watchful waiting" cohort of Sboner et al., which includes expression profiles from transurethral resection of prostate (TURP) specimens from 281 patients with localized prostate cancer that were followed for up to 30 years (33). Notably, this dataset differs from the Taylor dataset in several important respects: (i) sample collection in Sboner predates the PSA screening era (tissues collected prior to 1996); (ii) expression profiles were obtained from TURP rather than prostatectomies; and (Hi) the primary endpoint in the Sboner cohort is death due to prostate cancer rather than time to biochemical recurrence, as in the Taylor et al (Table 1). Considering these important distinctions between the Taylor and Sboner cohorts, biomarkers that show consistent stratification power in both were expected to be robust.

[0157] To focus on genes that most effectively inform outcome, analysis was limited to the extreme outcome of cases in the Sboner dataset. Specifically, two groups were identified: an "indolent group" with long-term survival following initial diagnosis (t > 10 years; n= 26), and a "lethal group" in which patients died early from prostate cancer (t < 4 years; n=29) (Table 1 ; Table 2). Thus, the decision tree was constructed using these extreme patients groups in the Sboner et al. training set.

[0158] Among thousands of possible trees evaluated in the decision tree model only fourteen 3- gene prognostic panel combinations had cross-validation power greater than 0.25 (Fig. 7A; Table 6). Trees with significant predictive power repeatedly included CDNK1A, FGFR1, PMP22, Clusterin, and CLIC4 (Fig. 3B; Table 6A). The top-ranked combinations were tested for predictive accuracy using confusion matrices to "score" predicted versus actual indolent and lethal cases (Fig. 3B, Fig. 8). First, a test set was assembled from cases in Sboner et al that had not been used for decision tree learning (n = 28 indolent and 8 lethal; Table 1 ; Table 2). Then, each gene panel was used to classify patients based on survival. Interestingly, the best gene panel (odds ratio = 1.94) identified from confusion matrix analysis was also the top-ranked panel from the decision-tree model. This panel included FGFRl, PMP22 and CDKN1A (Fig. 3B, Fig. 8) and was selected as our candidate biomarker panel to further evaluate for stratifying low Gleason score prostate tumors.

E. Validation of the 3-gene prognostic panel at the mRNA and protein levels

[0159] The prognostic accuracy of the 3-gene prognostic panel {i.e., FGFRl, PMP22 and CDKNIA) at the mRNA expression level (Fig. 1, Step 4) was first evaluated, using the low Gleason score {i.e., Gleason score 6 and Gleason score 7(3+4)) tumors from Taylor et al. (n = 95; Table 1; Table 2). The ability of the 3-gene prognostic panel to segregate these low Gleason score tumors into low- and high-risk groups was evident in A:-means clustering (Fig. 7B), an unsupervised clustering approach that relies only similarity of gene expression in different samples without using any clinical information about the patients. As is evident by Kaplan-Meier analysis, the 3-gene prognostic panel {FGFRl, PMP22 and CDKNIA) robustly segregated the low Gleason score prostate tumors into high- and low-risk groups based on time to biochemical recurrence (n=95 cases; p = 0.005) (Fig. 3C).

[0160] Interestingly, in these and subsequent analyses, the 3-gene prognostic panel was were consistently more effective in stratification of low Gleason score tumors as compared with the entire patient population, including higher Gleason score tumors (n = 131; p = 0.047) (Fig. 9B). Furthermore, the 3-gene prognostic panel was significantly more effective in segregating patients than the 19-gene indolence signature (compare Fig. 3C with Fig. 9A, B), which further demonstrates the efficacy of the decision tree learning model for selecting the most clinically- relevant biomarkers among the 19-gene signature. Notably, only one of the other top six gene combinations from the decision tree model {FGFRl, B2M and CDKNIA) was significant {p = 0.02) in stratifying low Gleason score prostate tumors into high- and low-risk groups (Fig. 3B, Fig. 9C), and it is noteworthy that this combination shares two genes in common with the 3-gene prognostic panel. Finally, although certain of the individual genes {FGFRl, PMP22 and CDKNIA) had prognostic power in some assays, only the 3-gene prognostic panel was consistently observed to have prognostic potential in all of the models and cohorts evaluated (see Fig. 10).

[0161] The prognostic value of the 3-gene prognostic panel was further evident using C- statistics in comparison with pathological Gleason score or the D'Amico classification nomogram, which takes into account Gleason score, Clinical T stage, and PSA levels (34) (Fig. 3D). In particular, the 3-gene prognostic panel performed better (C-index 0.86; CI 0.65-1.0; p = 3.3 x 10 ) than either Gleason score alone (C-index 0.82; CI 0.54-1.0; p = 0.010) or the D'Amico classification alone (C-index 0.72; CI 0.52-0.90; p = 0.012), while the 3-gene prognostic panel significantly improved prognostic capability when combined either with Gleason or D'Amico (C-index= 0.89; CI 0.74-1.0; p=4.7 x 10 ^"8 and C-index= 0.83; CI 0.73-0.95; p=1.8 x 10 ^"9 , respectively) (Fig. 3D). Furthermore, multivariate Cox proportional hazard analysis showed that the 3-gene prognostic panel together with Gleason had statistically significant improved prognostic ability than using Gleason alone (p=0.04). For D'Amico classification, the improved prognostic ability was mostly due to additive effects of the 3-gene prognostic panel, which was significant (p=0.017). This improvement was diluted by the high degrees of freedom of the full interaction model between D'Amico covariates and the 3-gene prognostic panel prediction (p = 0.11) (Fig. 3E). Taken together, these findings demonstrate the independent prognostic value of the 3-gene prognostic panel at the mRNA level.

[0162] These findings were extended to evaluate whether the 3-gene prognostic panel was also prognostic at the protein level (Fig. 1, Step 4). Immunohistochemical staining was performed on a tissue microarray (TMA) comprised of primary prostate tumors that corresponded to a wide range of Gleason scores, although the focus was on the low Gleason score tumors (i.e., the Gleason 6 and Gleason 7 (3+4)) (Fig. 4A, B; Table 1; Fig. 11). The predictive accuracy of the 3- gene prognostic panel was supported by unsupervised A:-means clustering analyses, in which there was 2 to 4 fold higher staining intensity for tumors classified in the indolent versus the aggressive clusters (Fig. 7C). Moreover, Kaplan-Meier analyses revealed that the protein expression levels of FGFR1, PMP22 and CDKN1A effectively stratified the low Gleason score tumors into high- and low-risk groups (p= 0.015) (Fig. 4B).

[0163] C-statistic analyses of this cohort revealed that the 3-gene prognostic panel performed significantly better (C-index 0.95; CI 0.90-1.0; p=2.0 x 10 ^"54 ) than Gleason score alone, which in this cohort displayed a relatively low C-index (C-index = 0.62; CI 0.34-.89; p=0A98), while the 3-gene prognostic panel significantly improved the prognostic accuracy of the Gleason score (C- index=0.82; CI 0.70-0.94; p=1.0 X 10 ^"7 ) (Fig. 4C). Additionally, multivariate Cox proportional hazard analyses showed that the 3-gene prognostic panel together with Gleason had improved prognostic ability (p = 0.034) over using Gleason alone (Fig. 4C). Taken together, these findings demonstrate that the 3-gene prognostic panel (herein the "prognostic panel") can accurately stratify low Gleason score primary prostate tumors at both the mRNA and protein levels, and provides independent prognostic information that improves the predictions of widely-utilized clinical nomograms.

[0164] Specifically indolent prostate cancer expresses normal or elevated levels of the prognostic panel genes compared to normal prostate while aggressive prostate tumors express significantly lower levels (about 2-fold or less).

F. Prognostic capability of the 3-gene prognostic panel on biopsy samples from surveillance patients

[0165] Analyses of protein expression of the 3-gene prognostic panel was done to determine if it could be effectively incorporated into clinical diagnosis of patients with low Gleason score prostate cancer (Fig. 1, Step 5). Toward this end, a retrospective analyses was performed of biopsy specimens from patients who had been monitored by surveillance in the Department of Urology at Columbia University Medical Center from 1992 to 2012 (35). In particular, a cohort of patients was assembled that had presented with clinically-low risk prostate cancer as defined by: normal digital rectal exam (DRE), serum PSA <10 ng/ml, biopsy Gleason score < 6 in no more than 2 cores, and cancer involving no more than 50% of any core on at least a 12-core biopsy (35). The protocol to monitor these patients included DRE and serum PSA testing every three months, and repeat biopsy every 12 months for the first three years and every 18 months for the next three years, or a "for-cause" biopsy if there was any sign of progression (i.e., abnormal DRE, increasing PSA). As long as all parameters and biopsy findings remained stable, patients were advised to remain on the surveillance protocol (and are referred to here as "non- failed"). Patients were considered "failure" for surveillance if they showed increasing cancer grade or volume on biopsy. Notably, all patients included in the "failed" group herein had "failed" based on these defined clinical parameters and not, for example, those who opted to undergo treatment for other reasons such as anxiety about having an untreated cancer, etc.

[0166] From a consecutive series of 213 patients that strictly adhered to the above criteria, all patients were identified that "failed" surveillance for which the initial biopsy tissue was available (n = 14) (Table 1). For comparison, an equivalent group of patients was analyzed that did not fail surveillance for at least ten years for which initial biopsy tissue was available (n = 29) (Table 1). Note that in both cases the initial biopsies used to enroll the patients to surveillance monitoring were evaluated.

[0167] Immunohistochemical analyses of these "failed" and "non-failed" groups of biopsy samples showed a striking correlation between the expression of FGFRl, PMP22 and CDKN1A and outcome (Fig. 4D, E; Fig. 11). In particular, all of the biopsies from the Gleason 6 patients that did not fail surveillance had robust and fairly uniform levels of expression of FGFRl, PMP22 and CDKN1A (average composite staining score of 4.11 ± 1.0). In striking contrast, the biopsies from the Gleason 6 patients that had failed active surveillance had reduced staining overall, as well as much more variable levels of FGFRl, PMP22 and CDKN1A (average composite staining score of 1.71 ± 1.2). Notably, the difference in the protein expression levels of the 3-gene prognostic panel (FGFRl, PMP22 and CDKN1A) in these Gleason 6 biopsy samples from patients that had "failed" or had "not- failed" surveillance was highly significant (t test p value = 1.5 X 10 ^"5 ), showing that expression levels of this 3-gene prognostic panel can be used as a prognostic indicator for these low Gleason score prostate tumors.

[0168] In certain embodiments, detection of FGFRl, PMP22 and CDKN1A on biopsy samples is used, in conjunction with other clinical parameters, to identify the subset of patients with low Gleason score prostate tumors that are likely to progress to aggressive disease and to monitor indolent tumors on active surveillance protocols.

[0169] CDKN1A (p21) is a cell-cycle regulatory gene whose expression is closely linked to senescence, whose down-regulation has been associated with promoting cancer progression in general, including prostate cancer (37, 38). The findings showing that CDKN1A (p21) expression is associated with indolence are consistent with previous studies. In contrast, the findings showing that expression of FGFRl is associated with indolence was unexpected. FGFRl is the major receptor for FGF growth factor signaling in the prostate and known to play a critical role in prostate development as well as prostate tumorigenesis (39, 40). Based on previous analyses of its functional role in cancer, including a recent study that evaluated the functional consequences of FGFRl in a mutant mouse model of lethal prostate cancer (41), it might have been predicted that elevated expression of FGFRl should be associated with cancer progression, rather than indolence. However, the complexity of FGFRl status in prostate cancer is highlighted by the fact that while a subset of aggressive, castration-resistant prostate tumors have been shown to display amplification of the gene locus including FGFRl (42), in the Taylor dataset, the specific genomic region that includes FGFRl is frequently deleted, which is correlated with down-regulation of FGFRl gene expression (14).

References

A. Jemal, R. Siegel, J. Xu, E. Ward, Cancer statistics, 2010, CA: a cancer journal for clinicians 60, 277 (2010).

M. R. Cooperberg, J. M. Broering, P. W. Kantoff, P. R. Carroll, Contemporary trends in low risk prostate cancer: risk assessment and treatment, The Journal of urology 178, S14 (2007).

H. G. Welch, P. C. Albertsen, Prostate cancer diagnosis and treatment after the introduction of prostate-specific antigen screening: 1986-2005, Journal of the National Cancer Institute 101, 1325 (2009).

J. R. Prensner, M. A. Rubin, J. T. Wei, A. M. Chinnaiyan, Beyond PSA: the next generation of prostate cancer biomarkers, Science translational medicine 4, 127rv3 (2012).

D. F. Gleason, Histologic grading of prostate cancer: a perspective, Human pathology 23, 273 (1992).

T. J. Wilt, R. MacDonald, I. Rutks, T. A. Shamliyan, B. C. Taylor, R. L. Kane, Systematic review: comparative effectiveness and harms of treatments for clinically localized prostate cancer, Annals of internal medicine 148, 435 (2008).

T. J. Daskivich, K. Chamie, L. Kwan, J. Labo, R. Palvolgyi, A. Dash, S. Greenfield, M. S. Litwin, Overtreatment of men with low-risk prostate cancer and significant comorbidity, Cancer 117, 2058 (2011).

H. G. Welch, W. C. Black, Overdiagnosis in cancer, Journal of the National Cancer Institute 102, 605 (2010). B. B. Cantrell, D. P. DeKlerk, J. C. Eggleston, J. K. Boitnott, P. C. Walsh, Pathological factors that influence prognosis in stage A prostatic cancer: the influence of extent versus grade, The Journal of urology 125, 516 (1981).

M. R. Cooperberg, P. R. Carroll, L. Klotz, Active surveillance for prostate cancer: progress and promise, Journal of clinical oncology : official journal of the American Society of Clinical Oncology 29, 3669 (2011).

J. H. Hayes, D. A. Ollendorf, S. D. Pearson, M. J. Barry, P. W. Kantoff, S. T. Stewart, V. Bhatnagar, C. J. Sweeney, J. E. Stahl, P. M. McMahon, Active surveillance compared with initial treatment for men with low-risk prostate cancer: a decision analysis, JAMA : the journal of the American Medical Association 304, 2373 (2010).

J. J. Tosoian, B. J. Track, P. Landis, Z. Feng, J. I. Epstein, A. W. Partin, P. C. Walsh, H. B. Carter, Active surveillance program for prostate cancer: an update of the Johns Hopkins experience, Journal of clinical oncology : official journal of the American Society of Clinical Oncology 29, 2185 (2011).

M. M. Shen, C. Abate-Shen, Molecular genetics of prostate cancer: new prospects for old challenges, Genes Dev 24, 1967 (2010).

B. S. Taylor, N. Schultz, H. Hieronymus, A. Gopalan, Y. Xiao, B. S. Carver, V. K. Arora, P. Kaushik, E. Cerami, B. Reva, Y. Antipin, N. Mitsiades, T. Landers, I. Dolgalev, J. E. Major, M. Wilson, N. D. Socci, A. E. Lash, A. Heguy, J. A. Eastham, H. I. Scher, V. E. Reuter, P. T. Scardino, C. Sander, C. L. Sawyers, W. L. Gerald, Integrative genomic profiling of human prostate cancer, Cancer cell 18, 11 (2010).

M. Narita, S. W. Lowe, Senescence comes of age, Nature medicine 11, 920 (2005).

J. Campisi, Senescent cells, tumor suppression, and organismal aging: good citizens, bad neighbors, Cell 120, 513 (2005). J. Campisi, Aging, tumor suppression and cancer: high wire-act!, Mechanisms of ageing and development 126, 51 (2005).

M. Collado, M. A. Blasco, M. Serrano, Cellular senescence in cancer and aging, Cell 130, 223 (2007).

J. Choi, I. Shendrik, M. Peacocke, D. Peehl, R. Buttyan, E. F. Ikeguchi, A. E. Katz, M. C. Benson, Expression of senescence-associated beta-galactosidase in enlarged prostates from men with benign prostatic hyperplasia, Urology 56, 160 (2000).

P. Castro, D. Giri, D. Lamb, M. Ittmann, Cellular senescence in the pathogenesis of benign prostatic hyperplasia, The Prostate 55, 30 (2003).

Z. Chen, L. C. Trotman, D. Shaffer, H. K. Lin, Z. A. Dotan, M. Niki, J. A. Koutcher, H. I. Scher, T. Ludwig, W. Gerald, C. Cordon-Cardo, P. P. Pandolfi, Crucial role of p53- dependent cellular senescence in suppression of Pten-deficient tumorigenesis, Nature 436, 725 (2005).

J. P. de Magalhaes, J. Curado, G. M. Church, Meta-analysis of age-related gene expression profiles identifies common signatures of aging, Bioinformatics 25, 875 (2009).

P. Wirapati, C. Sotiriou, S. Kunkel, P. Farmer, S. Pradervand, B. Haibe-Kains, C. Desmedt, M. Ignatiadis, T. Sengstag, F. Schutz, D. R. Goldstein, M. Piccart, M. Delorenzi, Meta-analysis of gene expression profiles in breast cancer: toward a unified understanding of breast cancer subtyping and prognosis signatures, Breast cancer research : BCR 10, R65 (2008).

J. Cuzick, G. P. Swanson, G. Fisher, A. R. Brothman, D. M. Berney, J. E. Reid, D. Mesher, V. O. Speights, E. Stankiewicz, C. S. Foster, H. Moller, P. Scardino, J. D. Warren, J. Park, A. Younus, D. D. Flake, 2nd, S. Wagner, A. Gutin, J. S. Lanchbury, S. Stone, Prognostic value of an RNA expression signature derived from cell cycle proliferation genes in patients with prostate cancer: a retrospective study, The lancet oncology 12, 245 (2011).

Y. P. Yu, D. Landsittel, L. Jing, J. Nelson, B. Ren, L. Liu, C. McDonald, R. Thomas, R. Dhir, S. Finkelstein, G. Michalopoulos, M. Becich, J. H. Luo, Gene expression alterations in prostate cancer predicting tumor aggression and preceding development of malignancy, Journal of clinical oncology : official journal of the American Society of Clinical Oncology 22, 2790 (2004).

A. Bhattacharjee, W. G. Richards, J. Staunton, C. Li, S. Monti, P. Vasa, C. Ladd, J. Beheshti, R. Bueno, M. Gillette, M. Loda, G. Weber, E. J. Mark, E. S. Lander, W. Wong,

B. E. Johnson, T. R. Golub, D. J. Sugarbaker, M. Meyerson, Classification of human lung carcinomas by mRNA expression profiling reveals distinct adenocarcinoma subclasses, Proceedings of the National Academy of Sciences of the United States of America 98, 13790 (2001).

Comprehensive molecular portraits of human breast tumours, Nature 490, 61 (2012).

R. Bhatia-Gaur, A. A. Donjacour, P. J. Sciavolino, M. Kim, N. Desai, P. Young, C. R. Norton, T. Gridley, R. D. Cardiff, G. R. Cunha, C. Abate-Shen, M. M. Shen, Roles for Nkx3.1 in prostate development and cancer, Genes Dev 13, 966 (1999).

M. J. Kim, R. Bhatia-Gaur, W. A. Banach-Petrosky, N. Desai, Y. Wang, S. W. Hayward, G. R. Cunha, R. D. Cardiff, M. M. Shen, C. Abate-Shen, Nkx3.1 mutant mice recapitulate early stages of prostate carcinogenesis, Cancer Res 62, 2999 (2002).

C. Abate-Shen, M. M. Shen, E. Gelmann, Integrating differentiation and cancer: the Nkx3.1 homeobox gene in prostate organogenesis and carcinogenesis, Differentiation 76, 717 (2008). X. Ouyang, T. L. DeWeese, W. G. Nelson, C. Abate-Shen, Loss-of-function of Nkx3.1 promotes increased oxidative damage in prostate carcinogenesis, Cancer Res 65, 6773 (2005).

A. G. Sowalsky, H. Ye, G. J. Bubley, S. P. Balk, Clonal progression of prostate cancers from Gleason grade 3 to grade 4, Cancer Res 73, 1050 (2013).

A. Sboner, F. Demichelis, S. Calza, Y. Pawitan, S. R. Setlur, Y. Hoshida, S. Perner, H. O. Adami, K. Fall, L. A. Mucci, P. W. Kantoff, M. Stampfer, S. O. Andersson, E. Varenhorst, J. E. Johansson, M. B. Gerstein, T. R. Golub, M. A. Rubin, O. Andren, Molecular sampling of prostate cancer: a dilemma for predicting disease progression, BMC medical genomics 3, 8 (2010).

A. V. D'Amico, R. Whittington, S. B. Malkowicz, D. Schultz, K. Blank, G. A. Broderick, J. E. Tomaszewski, A. A. Renshaw, I. Kaplan, C. J. Beard, A. Wein, Biochemical outcome after radical prostatectomy, external beam radiation therapy, or interstitial radiation therapy for clinically localized prostate cancer, JAMA : the journal of the American Medical Association 280, 969 (1998).

P. Motamedinia, J. L. Richard, J. M. McKiernan, G. J. Decastro, M. C. Benson, Role of immediate confirmatory prostate biopsy to ensure accurate eligibility for active surveillance, Urology 80, 1070 (2012).

J. Campisi, Cancer and ageing: rival demons?, Nature reviews. Cancer 3, 339 (2003).

S. Roy, R. P. Singh, C. Agarwal, S. Siriwardana, R. Sclafani, R. Agarwal, Downregulation of both p21/Cipl and p27/Kipl produces a more aggressive prostate cancer phenotype, Cell Cycle 7, 1828 (2008).

T. Abbas, A. Dutta, p21 in cancer: intricate networks and multiple activities, Nature reviews. Cancer 9, 400 (2009).

V. D. Acevedo, M. Ittmann, D. M. Spencer, Paths of FGFR-driven tumongenesis, Cell Cycle 8, 580 (2009). N. Turner, R. Grose, Fibroblast growth factor signalling: from development to cancer, Nature reviews. Cancer 10, 116 (2010).

F. Yang, Y. Zhang, S. J. Ressler, M. M. Ittmann, G. E. Ayala, T. D. Dang, F. Wang, D. R. Rowley, FGFR1 is Essential for Prostate Cancer Progression and Metastasis, Cancer Res, (2013).

J. Edwards, N. S. Krishna, C. J. Witton, J. M. Bartlett, Gene amplifications associated with the development of hormone-resistant prostate cancer, Clinical cancer research : an official journal of the American Association for Cancer Research 9, 5271 (2003).

G. Meyer Zu Horste, K. A. Nave, Animal models of inherited neuropathies, Current opinion in neurology 19, 464 (2006).

K. Adlkofer, R. Martini, A. Aguzzi, J. Zielasek, K. V. Toyka, U. Suter, Hypermyelination and demyelinating peripheral neuropathy in Pmp22-deficient mice, Nature genetics 11, 274 (1995).

U. Suter, G. J. Snipes, Peripheral myelin protein 22: facts and hypotheses, Journal of neuroscience research 40, 145 (1995).

Z. Ding, C. J. Wu, G. C. Chu, Y. Xiao, D. Ho, J. Zhang, S. R. Perry, E. S. Labrot, X. Wu, R. Lis, Y. Hoshida, D. Hiller, B. Hu, S. Jiang, H. Zheng, A. H. Stegh, K. L. Scott, S. Signoretti, N. Bardeesy, Y. A. Wang, D. E. Hill, T. R. Golub, M. J. Stampfer, W. H. Wong, M. Loda, L. Mucci, L. Chin, R. A. DePinho, SMAD4-dependent barrier constrains prostate cancer growth and metastatic progression, Nature 470, 269 (2011). E. K. Markert, H. Mizuno, A. Vazquez, A. J. Levine, Molecular classification of prostate cancer using curated expression signatures, Proceedings of the National Academy of Sciences of the United States of America 108, 21276 (2011).

S. A. Tomlins, S. M. Aubin, J. Siddiqui, R. J. Lonigro, L. Sefton-Miller, S. Miick, S. Williamsen, P. Hodge, J. Meinke, A. Blase, Y. Penabella, J. R. Day, R. Varambally, B. Han, D. Wood, L. Wang, M. G. Sanda, M. A. Rubin, D. R. Rhodes, B. Hollenbeck, K. Sakamoto, J. L. Silberstein, Y. Fradet, J. B. Amberson, S. Meyers, N. Palanisamy, H. Rittenhouse, J. T. Wei, J. Groskopf, A. M. Chinnaiyan, Urine TMPRSS2:ERG fusion transcript stratifies prostate cancer risk in men with elevated serum PSA, Science translational medicine 3, 94ra72 (2011).

D. Olmos, D. Brewer, J. Clark, D. C. Danila, C. Parker, G. Attard, M. Fleisher, A. H. Reid, E. Castro, S. K. Sandhu, L. Barwell, N. B. Oommen, S. Carreira, C. G. Drake, R. Jones, C. S. Cooper, H. I. Scher, J. S. de Bono, Prognostic value of blood mRNA expression signatures in castration-resistant prostate cancer: a prospective, two-stage study, The lancet oncology 13, 1114 (2012). M. Braig, S. Lee, C. Loddenkemper, C. Rudolph, A. H. Peters, B. Schlegelberger, H. Stein, B. Dorken, T. Jenuwein, C. A. Schmitt, Oncogene-induced senescence as an initial barrier in lymphoma development, Nature 436, 660 (2005).

M. Collado, J. Gil, A. Efeyan, C. Guerra, A. J. Schuhmacher, M. Barradas, A. Benguria, A. Zaballos, J. M. Flores, M. Barbacid, D. Beach, M. Serrano, Tumour biology: senescence in premalignant tumours, Nature 436, 642 (2005).

M. Malumbres, I. Perez De Castro, M. I. Hernandez, M. Jimenez, T. Corral, A. Pellicer, Cellular response to oncogenic ras involves induction of the Cdk4 and Cdk6 inhibitor pl5(INK4b), Mol Cell Biol 20, 2915 (2000).

A. Subramanian, P. Tamayo, V. K. Mootha, S. Mukherjee, B. L. Ebert, M. A. Gillette, A. Paulovich, S. L. Pomeroy, T. R. Golub, E. S. Lander, J. P. Mesirov, Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles, Proceedings of the National Academy of Sciences of the United States of America 102, 15545 (2005).

M. J. Pencina, R. B. D'Agostino, Overall C as a measure of discrimination in survival analysis: model specific population value and confidence interval estimation, Statistics in medicine 23, 2109 (2004).

R. A. Irizarry, B. Hobbs, F. Collin, Y. D. Beazer-Barclay, K. J. Antonellis, U. Scherf, T. P. Speed, Exploration, normalization, and summaries of high density oligonucleotide array probe level data, Biostatistics 4, 249 (2003).

Z. Wu, R. Irizarry, R. Gentleman, F. M. Murillo, F. Spencer, A Model Based Background Adjustment for Oligonucleotide Expression Arrays, Johns Hopkins University, Dept. of Biostatistics Working Papers, (2004).

V. K. Mootha, C. M. Lindgren, K. F. Eriksson, A. Subramanian, S. Sihag, J. Lehar, P. Puigserver, E. Carlsson, M. Ridderstrale, E. Laurila, N. Houstis, M. J. Daly, N. Patterson, J. P. Mesirov, T. R. Golub, P. Tamayo, B. Spiegelman, E. S. Lander, J. N. Hirschhorn, D. Altshuler, L. C. Groop, PGC-1 alpha-responsive genes involved in oxidative phosphorylation are coordinately downregulated in human diabetes, Nature genetics 34, 267 (2003).

G. A. F. Seber, Multivariate Observations. ( John Wiley & Sons, Inc., Hoboken, NJ, 1984).

C. W. Kinkade, M. Castillo-Martin, A. Puzio-Kuter, J. Yan, T. H. Foster, H. Gao, Y. Sun, X. Ouyang, W. L. Gerald, C. Cordon-Cardo, C. Abate-Shen, Targeting AKT/mTOR and ERK MAPK signaling inhibits hormone-refractory prostate cancer in a preclinical mouse model, The Journal of clinical investigation 118, 3051 (2008). 515 ATP5F1 ATP synthi / - -

516 ATP5G1 ATP5G1 ATP synthi / - -

518 ATP5G3 TP5G3 ATP synthi / - -

522 ATP5J E ATP synthi / - -

545 ATR ATR ataxia telar - -

8313 AXIN2 XjN2 axin 2 / - -

567 B2M B2 beta-2-mici / - -

23786 BCL2L13 BCL2L13 BCL2-like ^' / - -

633 BGN BGN biglycan / - -

641 BLM Bloom sync - / -

648 BMI1 Mil BMI1 polyc - / -

653 BMP5 BMP5 bone morp - -

672 BRCA1 BRCA1 breast cane - -

55108 BSDC1 BSDC1 BSD doma - -

9184 BUB3 BUB3 BUB3 buck - -

79864 C11orf63 C11orf63 chromoson / - -

55196 C12orf35 C12orf35 chromoson / - -

79622 C16orf33 C16or§3 chromoson / - -

712 C1QA C1QA complemer / - -

713 C1QB C1Q complemer / - -

714 C1QC C1QC complemer / - -

715 C1 R C1E complemer / - -

716 C1S C S complemer / - -

116151 C20orf108 C20orf108 chromoson / - -

8209 C21orf33 C2 33 chromoson / - -

718 C3 complemer / - -

720 C4A CJA Compleme / - -

85438 C4orf35 G4orf35 chromoson / - -

9315 C5orf13 C5orf13 chromoson / - -

221545 C6orf 136 C6orf136 chromoson / - -

79017 C7orf24 CTor gamma-gli / - -

84302 C9orf125 C9orf125 chromoson / - -

79095 C9orf16 C9orfl6 chromoson / - -

762 CA4 CA4 carbonic ar / - -

23705 CADM1 CAQMl cell adhesk / - -

793 CALB1 CALB1 calbindin 1 / - -

794 CALB2 CALB2 calbindin 2 / - -

847 CAT CAT catalase - -

1235 CCR6 CCR6 chemokine - -

963 CD53 CD53 CD53 mole - - 967 CD63 CD63 CD63 mole / - -

968 CD68 CQ68 CD68 mole / - -

972 CD74 CUM CD74 mole / - -

3732 CD82 CD82 CD82 mole / - -

928 CD9 CD9 CD9 molec / - -

990 CDC6 CDC6 cell divisior / - -

999 CDH1 CDH1 cadherin 1 , / - -

1026 CDKN1A CDK 1A cyclin-depe / / -

1029 CDKN2A GDKN2A cyclin-depe - / -

1030 CDKN2B CDK 2B cyclin-depe - -

1051 CEBPB CEBPB CCAAT/en - -

3075 CFH CFH complemer - -

11200 CHEK2 CHEK2 CHK2 chec - -

1134 CHRNA1 CJHR Al cholinergic / - -

10462 CLEC10A CLEC10A C-type lect / - -

6320 CLEC11A CLEC11A C-type lect / - -

25932 CLIC4 chloride int / - -

1191 CLU clusterin / - -

1306 COL15A1 COL _; 1.5Aj. collagen, t / - -

80781 COL18A1 collagen, t / - -

1277 COL1A1 COL1A1 collagen, t / - -

1281 COL3A1 COL3A1 collagen, t / - -

1287 COL4A5 COL4A5 collagen, t / - -

1289 COL5A1 COL5A1 collagen, t / - -

1290 COL5A2 COL5A2 collagen, t / - -

1312 COMT COMT catechol-0 / - -

51004 COQ6 COQ6 coenzyme / - -

1351 COX8A COX8A cytochromi / - -

1356 CP CP ceruloplasr / - -

1393 CRHBP CRHBP corticotropi / - -

1410 CRYAB CRYAB crystallin, a / - -

1453 CSNK1 D CSNK1 D casein kina / - -

1465 CSRP1 CSRP1 cysteine ar / - -

1466 CSRP2 CSRP2 cysteine ar / - -

1509 CTSD CTSD cathepsin [ / - -

1512 CTSH CTSH cathepsin [ / - -

1520 CTSS CTSS cathepsin 5 / - -

1522 CTSZ CTSZ cathepsin Ί / - -

6376 CX3CL1 CX3CJJ. chemokine / - -

58191 CXCL16 CXCL16 chemokine / - - 1620 DBC1 DBCl deleted in t / - -

28960 DCPS DCPS decapping / - -

11258 DCTN3 QCIM3 dynactin 3 / - -

54541 DDIT4 DM! DNA-dama / - -

7913 DEK QEK DEK oncoc / - -

79139 DERL1 Q£BLi Deri -like d / - -

56616 DIABLO diablo horn / - -

3300 DNAJB2 DNAJB2 DnaJ (Hsp< / - -

29103 DNAJC15 DNAJC15 DnaJ (Hsp< / - -

113878 DTX2 0 2 Deltex horr / - -

151636 DTX3L 0IX3L deltex 3-lik / - -

1778 DYNC1 H1 DYNC1 H1 Dynein, cyt / - -

1869 E2F1 E2F1 E2F transc - -

1889 ECE1 ECEl endothelin / - -

2202 EFEMP1 EFEMP1 EGF-conta / - -

1958 EGR1 EGR1 early growt - -

30845 EHD3 EHD3 EH-domain / - -

2006 ELN EL elastin / - -

2033 EP300 EP300 E1A bindin - -

80314 EPC1 EPC1 enhancer c / - -

2160 F11 F11 coagulatior / - -

2170 FABP3 F BP3 fatty acid b / - -

11170 FAM107A E 1Q7A family with / - -

54463 FAM134B FA 134B family with / - -

404636 FAM45A ΕΔΜ45Α Family with / - -

137392 FAM92A1 family with / - -

25940 FAM98A E 98A family with / - -

2203 FBP1 FBP1 fructose-1 ,l / - -

2212 FCGR2A FCGR2A Fc fragmer / - -

2213 FCGR2B ECGR2B Fc fragmer / - -

2214 FCGR3A ECGR3A Fc fragmer / - -

83706 FERMT3 FER T3 fermitin fan / - -

2260 FGFR1 FGFRl fibroblast g - -

2271 FH FH fumarate h / - -

54621 FLJ20674 EU206Z hypothetic? / - -

64926 FLJ21438 EU21438 hypothetic? / - -

728772 FLJ77644 FLJ77644 hypothetic? / - -

2321 FLT1 FLT1 fms-relatec - -

2335 FN1 E fibronectin / - -

64838 FNDC4 FNDC4 fibronectin / - -

9987 HNRPDL HNEEDL heterogene / - -

3208 HPCA HPCA hippocalcin / - -

3265 HRAS H£ v-Ha-ras H - -

259217 HSPA12A HSPA12A heat shock / - -

3303 HSPA1A HSPA1A Heat shock / -

3315 HSPB1 HSPB1 heat shock / - -

3336 HSPE1 HSPEl heat shock / - -

3459 IFNGR1 IFNGR1 interferon c / - -

3479 IGF1 !GFl insulin-like / - -

3512 IGJ IGJ immunoglo / - -

90865 IL33 L33 interleukin / - -

3624 INHBA i !BA inhibin, bet / - -

8826 IQGAP1 IQGAP1 IQ motif co / - -

79191 IRX3 IRX3 iroquois ho / - -

3689 ITGB2 E¾B2 integrin, be / - -

3696 ITGB8 ITGB8 integrin, be / - -

9452 ITM2A ΠΜ2Α integral me / - -

152789 JAKMIP1 janus kinas / - -

3727 JUND jun D proto - -

9813 KIAA0494 KIAA0494 / - -

57650 KIAA1524 KIAA1524 KIAA1524 / - -

9314 KLF4 E i Kruppel-lik( / - -

8844 KSR1 KSRi. kinase sup / -

3916 LAMP1 LAMP1 lysosomal- / - -

7805 LAPTM5 LAPJM5 lysosomal ι / - -

84247 LD0C1 L LD0C1 L leucine zipi / - -

3958 LGALS3 LGALS3 lectin, gala / - -

22998 LIMCH1 UMCH1 LIM and ca / - -

9516 LITAF LITAF lipopolysac / - -

284194 LOC28419 Lj 2841& Lectin, gale / - -

4057 LTF LIE lactotransfi / - -

4069 LYZ LYZ lysozyme (i / - -

256691 MAMDC2 MAMQC2 MAM dom£ / - -

5604 MAP2K1 E2K1 mitogen-ac / - -

23118 MAP3K7IP ΑΡ3ΚΠΕ mitogen-ac / - -

1432 MAPK14 ΜΔΡΚ14 mitogen-ac - / -

64844 7-Mar MARCH? Membrane / -

4170 MCL1 MCLi myeloid ce / - -

4190 MDH1 MQ 1 malate deh / -

4204 MECP2 MECP2 methyl CpC / -

5213 PFKM PFKM phosphofn / - -

5305 PIP4K2A E1E4K2A phosphatid / - -

8502 PKP4 PKP4 plakophilin / - -

5331 PLCB3 P1CB3 phospholip / - -

5341 PLEK PLEK pleckstrin / - -

5371 PML EML promyeloc - -

5376 PMP22 EME22 peripheral ι / - -

5406 PNLIP PNLIP pancreatic / - -

9588 PRDX6 EED 6 peroxiredo: / - -

9588 PRDX6 EEQ 6 peroxiredo: / - -

5696 PSMB8 ES B8 proteasom< / - -

5717 PSMD11 ESM i proteasom< / - -

5717 PSMD11 PS D11 proteasom< / - -

5723 PSPH ESPH phosphose / - -

5728 PTEN PTEN phosphatai - -

10728 PTGES3 PTGES3 prostaglanc / - -

2185 PTK2B Γ 2Β PTK2B pro / - -

51495 PTPLAD1 EIEL D1 protein tyre / - -

5800 PTPRO EJIEO protein tyre / - -

29942 PURG E EG purine-rich / - -

54517 PUS7 PUS7 pseudourid / - -

25945 PVRL3 EVRL3 poliovirus r / - -

5828 PXMP3 E ME3 Peroxisom; / - -

10966 RAB40B RAB40B RAB40B, n / - -

8480 RAE1 BAE1 RAE1 RNA - -

22821 RASA3 BASA3 RAS p21 p - -

5925 RB1 EM retinoblastc - -

473 RERE SEEE arginine-gli / - -

162494 RHBDL3 RHBDL3 rhomboid, ¹ / - -

9912 RICH2 E1CH2 Rho-type G / - -

8780 RIOK3 E1QK3 RIO kinase / - -

8635 RNASET2 RNASET2 ribonucleas / - -

55298 RNF121 E FJ21 ring finger | / - -

57674 RNF213 ENF213 ring finger | / - -

6096 RORB EOEi RAR-relate / - -

6122 RPL3 EEL3 ribosomal [ / - -

6241 RRM2 RR 2 ribonucleot / - -

6281 S100A10 S100A10 S100 calcii / - -

6275 S100A4 S100A4 S100 calcii / - -

6277 S100A6 S100A6 S100 calcii / - -

55365 TMEM176/ Ί ΜΙ transmemb / - -

28959 TMEM176I I IMlIi transmemb / - -

7157 TP53 IP53 tumor prate - / -

8626 TP63 JEM tumor prate - / -

57761 TRIB3 IELB3 _. tribbles hor / - -

57570 TRMT5 IEMI5 TRM5 tRNi / - -

85480 TSLP ISLE thymic stro / - -

10078 TSSC4 TSSC4 tumor supp / - -

203068 TUBB lUBB tubulin, bet / - -

7295 TXN Ϊ Ν thioredoxin / - -

10628 TXNIP I !E thioredoxin / - -

7305 TYROBP KEQIE TYRO prat / - -

7307 U2AF1 U2AF1 U2 small ni / - -

6675 UAP1 UAPl UDP-N-act / - -

29796 UCRC UCRC ubiquinol-c / - -

7353 UFD1 L UFD1 L ubiquitin fu / - -

7386 UQCRFS1 UQCRFSl ubiquinol-c / -

27089 UQCRQ UQCRQ ubiquinol-c / - -

7390 UROS LEGS uroporphyr / - -

57602 USP36 USP36 ubiquitin sp / - -

10493 VAT1 VAT1 vesicle ami / - -

7422 VEGFA VEGFA vascular er - -

7436 VLDLR VLDLR very low de / - -

7450 VWF VWF von Willebr / - -

7486 WRN WRN Werner syr - -

7639 ZNF85 zinc finger / - -

223082 ZNRF2 zinc and rir / - -

2 A2M A2M alpha-2-me / - -

GSEA using 377 aging and senescence signature (Fig 2F); Validation of 3-gene combination Fig. 3 and Supplementary Fig 5, 6A, 8

GSEA using 377 aging and senescence signature (Fig 2D) and Supplementary Fig 5

3A: Lagging edge genes from GSEA <

847 CAT CAT

443 ASPA ASPA

1281 COL3A1 COL3A1

9452 ITM2A JIM2A

10914 PAPOLA PAPOLA

3135 HLA-G HLA-G

57602 USP36 USP36

23786 BCL2L13 BCL2U3

4738 NEDD8 NEDD8

3459 IFNGR1 jFNGRl

29796 UCRC UCRC

3122 HLA-DRA

4092 SMAD7 S AD7

10135 NAMPT AMEI

28959 TMEM176B M1Z6B

653 BMP5 B P5

5717 PSMD11 PSMDli

1026 CDKN1A CDKN1

2202 EFEMP1 EFE P1

4057 LTF LTF

7386 UQCRFS1 UQCRFS1

3043 HBB HBB

64114 TMBIM1 IMiiMl

4677 NARS AB!

1512 CTSH CTSH

3916 LAMP1 AME1

351 APP APP

10493 VAT1 VAT 1

30845 EHD3

11258 DCTN3 DCTN3

10972 TMED10 T ED10 2634 GBP2 GBP2

1466 CSRP2 CSBP2

2628 GATM GATM

79602 ADIPOR2 ADIP0R2

2341 1 SIRT1 SlRTl

3696 ITGB8

84883 AIFM2 ASF 2

25940 FAM98A FAM98A

2878 GPX3 GPX3

1051 CEBPB CEBPB

51421 AMOTL2 AM0TL2

5213 PFKM EEKM

10728 PTGES3 EIGE33

79026 AHNAK AljNAK

9516 LITAF UIAF

6392 SDHD SDHD

64981 MRPL34 MRPL34

7913 DEK il

522 ATP5J ATP5J

9315 C5orf13 C5orQ3

4714 NDUFB8 DUFB8

140609 NEK7 EKZ

567 B2M B2

648 BMI1 M

9813 KIAA0494 1Δ 494

1306 C0L15A1 COLJ5A1

967 CD63 CD63

9987 HNRPDL HNRPDL

2799 GNS GNS

4494 MT1 F MD£

6275 S100A4 S100A4

4493 MT1 E 4204 MECP2 ECP2

8626 TP63 I£63

2260 FGFR1 FGFRX

715 C1 R C1 R

8313 AXIN2 AXJ

84302 C9orf125 C9≤iI125

85480 TSLP TSLP

3315 HSPB1 HSPBi

9021 SOCS3 S0CS3

22998 LIMCH1 UMQHl

137392 FAM92A1 Ε 32Δ1

1287 COL4A5 COL4A5

84247 LDOC1 L iJ20ClL

4780 NFE2L2 F££L2

6376 CX3CL1 CX3CL1

90865 IL33 1133

5728 PTEN EIE

3479 IGF1 1GF1

6272 SORT1 S0RX1

307 ANXA4

8826 IQGAP1 IQGAP1

3958 LGALS3 LGALS3

5376 PMP22 E E22

716 C1 S CI S

4478 MSN MSN

710 SERPING1 SERPjNGl

9737 GPRASP1 G£13ASP1

51 100 SH3GLB1 SH3GLB1

2335 FN1 £N1

498 ATP5A1 ATP5A1

1410 CRYAB CRYAB 11170 FAM107A FA Ml 07A

5348 FXYD1 E Yfil

23022 PALLD PALLD

25932 CLIC4 CL.1C4

1191 CLU CLU

1465 CSRP1 CSRPl

128 ADH5 ADH5

C: Meta Analysis using Fisher combined mt

1465 CSRP1 CSRP1 cysteine and glycine-rich protein 1

4170 MCL1 myeloid cell leukemia sequence 1

(BCL2-related)

518 ATP5G3 ATP5G3 ATP synthase, H+ transporting, mitochondrial F0 complex, subunit

C3 (subunit 9)

10417 SPON2 SP0N2 spondin 2, extracellular matrix protein

5341 PLEK PLEK pleckstrin

967 CD63 CD63 CD63 molecule

1512 CTSH CTSH cathepsin H

1277 COL1 A1 CQklAi collagen, type I, alpha 1

3109 HLA-DMB major histocompatibility complex, class II, DM beta

3696 ITGB8 1TGB8 integrin, beta 8

7157 TP53 TP53 tumor protein p53

51228 GLTP GLIE glycolipid transfer protein

3732 CD82 CD82 CD82 molecule

2202 EFEMP1 EFE Pi EGF-containing fibulin-like extracellular matrix protein 1

5376 PMP22 EM 22 peripheral myelin protein 22

5046 PCSK6 BCSK6 proprotein convertase

subtilisin/kexin type 6

22998 LIMCH1 u ciii LIM and calponin homology domains 1

4714 NDUFB8 NDUFB8 NADH dehydrogenase (ubiquinone)

1 beta subcomplex, 8, 19kDa

4478 MSN MS moesin

7805 LAPTM5 LAPT 5 lysosomal multispanning membrane protein 5

84302 C9orf125 CSor| 25 chromosome 9 open reading frame

125

2260 FGFR1 EGfJil fibroblast growth factor receptor 1

51004 C0Q6 C0Q6 coenzyme Q6 homolog, monooxygenase (S. cerevisiae)

6376 CX3CL1 CX3CU chemokine (C-X3-C motif) ligand 1

25945 PVRL3 VRL3 poliovirus receptor-related 3

1281 C0L3A1 C0L3A1 collagen, type III, alpha 1

1958 EGR1 ^j ≡GQl early growth response 1

10135 NAMPT NAMPI nicotinamide

phosphoribosyltransferase 963 CD53 CD53 CD53 molecule

9021 SOCS3 SOCS3 suppressor of cytokine signaling 3

3958 LGALS3 LQALS3 lectin, galactoside-binding, soluble,

3

9314 KLF4 KLF4 Kruppel-like factor 4 (gut)

2335 FN1 fill fibronectin 1

713 C1 QB C1 QB complement component 1 , q subcomponent, B chain

23022 PALLD PALLD Palladin, cytoskeletal associated protein

6392 SDHD SDHD succinate dehydrogenase complex, subunit D, integral membrane protein

728772 FLJ77644 EU22I44 hypothetical protein FLJ77644

4864 NPC1 NPC1 Niemann-Pick disease, type C1

256691 MAMDC2 MAM domain containing 2

968 CD68 CD68 CD68 molecule

79026 AHNAK AHNAK AHNAK nucleoprotein

3039 HBA1 HBAl Hemoglobin, alpha 1

1778 DYNC1 H1 DY C lili Dynein, cytoplasmic 1 , heavy chain

1

57704 GBA2 QBA2 glucosidase, beta (bile acid) 2

9961 MVP MVP major vault protein

10966 RAB40B ΒΔ ΟΒ RAB40B, member RAS oncogene family

9315 C5orf13 C5orf13 chromosome 5 open reading frame

13

57650 KIAA1524 Κ1ΔΔ1524 KIAA1524

85480 TSLP TSLP thymic stromal lymphopoietin

4738 NEDD8 EDD8 neural precursor cell expressed, developmental^ down-regulated 8

9516 LITAF LITAF lipopolysaccharide-induced TNF factor

4507 MTAP IAP methylthioadenosine phosphorylase

80314 EPC1 EPC1 enhancer of polycomb homolog 1

(Drosophila)

3689 ITGB2 Ιΐί1Β2 integrin, beta 2 (complement component 3 receptor 3 and 4 subunit)

3043 HBB HBB Hemoglobin, beta

3075 CFH CFH complement factor H

347 APOD APOD apolipoprotein D 4722 NDUFS3 NDUFS3 NADH dehydrogenase (ubiquinone)

Fe-S protein 3, 30kDa (NADH- coenzyme Q reductase)

8480 RAE1 RAE1 RAE1 RNA export 1 homolog (S.

pombe)

3122 HLA-DRA major histocompatibility complex, class II, DR alpha

1889 ECE1 ECE1 endothelin converting enzyme 1

259217 HSPA12A HSPA12A heat shock 70kDa protein 12A

1 1214 AKAP13 A AP13 A kinase (PRKA) anchor protein 13

1 1258 DCTN3 DCTN3 dynactin 3 (p22)

5331 PLCB3 PLCB3 phospholipase C, beta 3

(phosphatidylinositol-specific)

51495 PTPLAD1 PTP AD1 protein tyrosine phosphatase-like A domain containing 1

1453 CSNK1 D CSNK1 D casein kinase 1 , delta

8313 AXIN2 AXIN2 axin 2

55902 ACSS2 ACSS2 acyl-CoA synthetase short-chain family member 2

382 ARF6 ARF6 ADP-ribosylation factor 6

10628 TXNIP E thioredoxin interacting protein

3300 DNAJB2 D AJB2 DnaJ (Hsp40) homolog, subfamily

B, member 2

5305 PIP4K2A PIP4K2A phosphatidylinositol-5-phosphate 4- kinase, type II, alpha

5138 PDE2A PDE2A phosphodiesterase 2A, cGMP- stimulated

24145 PANX1 ANX1 pannexin 1

90865 IL33 1L33 interleukin 33

4092 SMAD7 S AD7 SMAD family member 7

121536 AEBP2 AEBP2 AE binding protein 2

2212 FCGR2A FCGR2A Fc fragment of IgG, low affinity I la, receptor (CD32)

6272 SORT1 SQBIl sortilin 1

443 AS PA AS PA aspartoacylase (Canavan disease)

641 14 TMBIM1 IMilMi transmembrane BAX inhibitor motif containing 1

28973 MRPS18B MBBS18B mitochondrial ribosomal protein

S18B

10397 NDRG1 DjRGl N-myc downstream regulated gene

1 8826 IQGAP1 SQQAP1 IQ motif containing GTPase activating protein 1

55298 RNF121 E FJ 21 ring finger protein 121

1351 COX8A COX&A cytochrome c oxidase subunit 8A

(ubiquitous)

6558 SLC12A2 SLC12A2 solute carrier family 12

(sodium/potassium/chloride transporters), member 2

27069 GHITM GH!IM growth hormone inducible transmembrane protein

7018 TF TF transferrin

5348 FXYD1 FXYD1 FXYD domain containing ion transport regulator 1

9655 SOCS5 S0CS5 suppressor of cytokine signaling 5

4257 MGST1 MGST1 microsomal glutathione S- transferase 1

2634 GBP2 GBP2 guanylate binding protein 2, interferon-inducible

404636 FAM45A ΕΔ 5Α Family with sequence similarity 45, member A

25932 CLIC4 CLJC4 chloride intracellular channel 4

10457 GPNMB glycoprotein (transmembrane) nmb

57602 USP36 USP36 ubiquitin specific peptidase 36

7107 GPR137B 0£β137Β G protein-coupled receptor 137B

3916 LAMP1 LAM PI lysosomal-associated membrane protein 1

7037 TFRC IEBC transferrin receptor (p90, CD71 )

7436 VLDLR VLDLR very low density lipoprotein receptor

7040 TGFB1 TGFB1 transforming growth factor, beta 1

2805 G0T1 GQI1 glutamic-oxaloacetic transaminase

1 , soluble (aspartate aminotransferase 1 )

2203 FBP1 FBP 1 fructose-1 ,6-bisphosphatase 1

3727 JUND JUND jun D proto-oncogene

83706 FERMT3 fermitin family homolog 3

(Drosophila)

2896 GRN GE granulin

4717 NDUFC1 NDUFC1 NADH dehydrogenase (ubiquinone)

1 , subcomplex unknown, 1 , 6kDa

3135 HLA-G major histocompatibility complex, class I, G 972 CD74 CD74 CD74 molecule, major

histocompatibility complex, class II invariant chain

3157 HMGCS1 H GCS1 3-hydroxy-3-methylglutaryl- Coenzyme A synthase 1 (soluble)

3512 IGJ immunoglobulin J polypeptide, linker protein for immunoglobulin alpha and mu polypeptides

1312 COMT COMT catechol-O-methyltransferase

29103 DNAJC15 DNAJC15 DnaJ (Hsp40) homolog, subfamily

C, member 15

sthod for all Gleason score 6 patients

P-value o

o

0

0

7.77E-16

1.64E-14

1.88E-14

2.30E-14

2.82E-14

5.47E-14

5.50E-14

1.58E-13

3.93E-13

5.24E-13

1.27E-12

3.15E-12

9.07E-12

2.31 E-11

3.10E-11

1.03E-10

3.13E-10

3.27E-10

9.95E-10

1.04E-09

1.52E-09

2.88E-09

9.00E-09

3.83E-08 5.04E-08

5.98E-08

7.21 E-08

9.56E-08

1.15E-07

1.22E-07 1.26E-07 1.91E-07

4.16E-07

5.46E-07

7.46E-07 7.97E-07 1.13E-06

1.25E-06

1.28E-06 2.26E-06

3.37E-06

3.45E-06

3.70E-06 3.95E-06

5.74E-06

6.30E-06

7.81 E-06

8.32E-06

9.23E-06 1.20E-05 1.35E-05 1.54E-05 1.76E-05 2.13E-05

2.13E-05

2.18E-05 2.37E-05 4.55E-05

7.42E-05

7.56E-05

7.81 E-05

9.70E-05

1.23E-04 1.57E-04 1.79E-04

1.82E-04

1.84E-04

2.75E-04 3.40E-04 4.15E-04

4.33E-04

4.50E-04 4.77E-04 5.04E-04

5.06E-04

5.12E-04

6.28E-04

6.71 E-04

7.31 E-04 7.36E-04 1.03E-03 1.05E-03

1.27E-03

1.29E-03

1.30E-03 1.38E-03 1.47E-03

1.48E-03 1.79E-03

1.88E-03

1.96E-03 3.02E-03 3.08E-03

3.14E-03 3.43E-03 3.48E-03

3.55E-03

3.87E-03

3.96E-03

4.23E-03 4.23E-03 4.59E-03

5.25E-03

6.20E-03 6.69E-03

6.73E-03

6.93E-03

8.49E-03 9.45E-03

9.77E-03 1.11E-02

1.26E-02

1.35E-02

1.65E-02 1.65E-02

1.68E-02

1.80E-02

1.87E-02

1.96E-02

2.03E-02 2.11E-02

2.14E-02 2.28E-02 2.32E-02

2.77E-02 2.85E-02

2.99E-02

3.10E-02

3.22E-02 3.23E-02 3.25E-02

3.39E-02 4.05E-02

4.09E-02 4.11E-02

4.24E-02

4.70E-02

4.78E-02 4.87E-02