MINERVA MEDICA COPYRIGHT

Transcription

1 This document is protected by international copyright laws. No additional reproduction is authorized. It is permitted for personal use to download and save only one file and print only one copy of this Article. It is not permitted to make additional copies The aim of this review is to critically analyze the current state of research in selected biomarkers and genomic-based tests for prostate cancer (PCa) diagnosis, staging, prognostication, and monitoring. Although in Western societies, PCa is the most common solid malignancy and the second leading cause of cancer death in men, the vast majority of men with PCa are diagnosed with clinically localized disease. The widespread use of prostate-specific antigen (PSA) testing, on one hand, has resulted in earlier PCa detection at a potentially more curable stage, but on the other hand has led to an increase in the rate of negative biopsies, as well as overdetection and overtreatment of potentially indolent tumors that would not have become life-threatening to a patient. A multitude of molecular tests and algorithms has been developed to enhance diagnostic accuracy, improve pretreatment and post-treatment patient risk stratification, and identify aggressive versus indolent disease to facilitate therapeutic decision-making. PSA and derivatives (PSA kinetics, PSA density, percentage of free PSA) as well as algorithms based on PSA and PSA isoforms measurements (prostate health index, four-kallikrein score), urinary molecular biomarkers-based tests (Prostate Cancer Antigen 3, and the Michigan Health System Prostate Score) and selected genomic/proteomic tests now commercially available for disease prognostication (such as Confirm MDx, Prostate Core Mi- *These authors contributed equally. Corresponding author: C. Magi-Galluzzi, Pathology and Laboratory Medicine Institute, Cleveland Clinic, 9500 Euclid Avenue, L25, Cleveland, OH 44195, USA. magic@ccf.org MINERVA UROL NEFROL 2015;67: Novel biomarkers and genomic tests in prostate cancer: a critical analysis S. M. 1 *, M. Ferro 2 *, E. Bollito 3, E. A. Klein 4, G. Carrieri 5, C. Magi-Galluzzi 1, 4 1Robert J. Tomsich Pathology and Laboratory Medicine Institute Cleveland Clinic, Cleveland, OH, USA 2European Institute of Oncology Division of Urology, Milan, Italy 3Department of Pathology, San Luigi, Turin, Italy 4Glickman Urological and Kidney Institute Cleveland Clinic, Cleveland, OH, USA 5Department of Urology and Renal Transplantation University of Foggia, Foggia, Italy tomic Test, Oncotype DX, Prolaris, ProMark, and Decipher) are herein discussed to inform the readers about current and future clinical applications and their limitations. Finally, we briefly touch upon potential biomarkers predictive of response to therapy, such as androgen receptor splice variant AR-V7, and detection and quantification of circulating tumor cells in the blood stream. Key words: Prostatic neoplasms - Biological markers - Genomics. Although in Western societies, prostate cancer (PCa) is the most common solid malignancy and the second leading cause of cancer death in men, 1 most newly diagnosed PCa are biologically indolent. Meanwhile men with aggressive PCa benefit from treatment, patients with indolent PCa may suffer the side effects of treatment, with compromised quality of life and potential significant morbidity without long term ben- Vol No. 3 MINERVA UROLOGICA E NEFROLOGICA 211

2 efit. The vast majority of men with PCa are diagnosed with clinically localized disease, but most are treated aggressively, mainly because of the uncertain malignant potential of their cancer. Although approximately 15-20% of patients with PCa have high-risk disease at presentation, their recognition has notoriously been somewhat hampered by the lack of reliable tests to distinguish aggressive from indolent disease. A decline in PCa mortality has been observed since the Food and Drug Administration (FDA) approved the widespread use of the total prostate-specific antigen (PSA, also known as human kallikrein 3, hk3) for PCa screening. In addition, PSA testing has resulted in earlier PCa detection at a potentially more curable stage. 2, 3 However, the ubiquitous application of PSA screening has also led to overdetection and overtreatment, since PSA is neither cancer-specific nor a surrogate for the biologic behavior of PCa. This has led to the publication of a level D recommendation against PSA screening by the United States (US) Preventive Services Task Force in 2012, 4 followed by variable, age- and patient-tailored recommendations endorsed by the American Urological Association (AUA). 5 Many pretreatment risk assessment models have been developed based on clinicopathologic factors, such as biopsy Gleason Score (GS), PSA or PSA density, clinical T- stage (ct), and cancer burden at biopsy, to predict the presence of small moderately differentiated, organ-confined disease and to better counsel patients regarding appropriate management decision tailored to their individual risk. Since the preferred course of treatment in many cases of early-stage PCa is uncertain and given the different options available, more accurate prognostic markers are needed to allow for individualized risk assessment. There has been a concerted effort to discover and validate novel PCa-specific biomarkers and prognostic tests that could grant better stratification of PCa patients based on their disease-specific risk of mortality, determine their likelihood of disease progression, and distinguish more aggres- sive disease that need to be treated more appropriately from indolent disease that can be treated conservatively. The detection of somatic mutations and epigenetic changes in precancerous lesions or in morphological normal-appearing tissue close to PCa (field effect phenomenon) can offer an important advantage to overcome the limitations related to prostate biopsy localization and false negative findings due to tissue sampling errors. Under-treatment of men who harbor aggressive PCa remains a significant problem in a subset of patients who may benefit from multimodality therapy which may result in decreased disease-specific mortality. In an attempt to increase the prognostic/ predictive power of single genes, the nonrandom clustering of genomic alterations has been explored to identify gene panels that may better characterize prognostically relevant PCa subtypes. A number of novel biomarkers and genomic tests are becoming available to guide clinicians through challenging clinical presentations. This review aims to critically analyze the current state of research in biomarkers and genomic-based tests for PCa diagnosis, staging, prognostication, and monitoring, to inform the readers about important current and future clinical applications and their limitations. Biomarkers for PCa testing in blood and urine Typically a biomarker refers to a characteristic that is objectively measured and evaluated as an indicator of normal biological processes, pathogenic processes, or pharmacologic responses to a therapeutic intervention. 6 Tumor diagnostic/biomarker capabilities from an accessible compartment (blood/ serum, urine, or saliva) would be a major advantage for oncology, especially since the measurement of concentrations of circulating and/or urine molecules is an attractive strategy, as the assays are inexpensive, and analytes may be readily and quantitatively measured by automated methods. 212 MINERVA UROLOGICA E NEFROLOGICA September 2015

3 Table I. Novel blood and urine-based prostate cancer biomarkers to help decide who to biopsy. phi 4K score* Test Type of Assay Markers Type of sample Application Strengths Limitations Progensa (PCA3 score) MiPS* (and MiPS high-grade) Multi-analyte immunoassay Multi-analyte immunoassay In vitro RNA TMA assay In vitro RNA TMA and Hybrid Protection Assay (HPA) PSA, free PSA, p2psa total PSA, free PSA, intact PSA, hk2 PSA and PCA3 mrnas PSA, PCA3, and TMPRSS2: ERG mrnas Serum Blood plasma anticoagulated with EDTA Post-DRE first catch urine Post-DRE first catch urine Distinguishing PCa from benign prostatic conditions in men 50 year old, with total PSA results in the 4 10 ng/ ml range and negative DRE to help decide whether biopsy is indicated Algorithm calculates the risk probability for finding high grade PCa (GS 7) if biopsy/s were to be performed (1 st biopsy or rebiopsy if initial biopsy negative >6 months prior to test) Reduces Limited clinical negative biopsies settings (age and (increased PCa PSA range) detection) RP features used Improves as surrogate specificity for for aggressive detection of disease high-grade/ high stage PCa (GS 7/pT3) Extensive validation FDA-approved Reduces Not FDAapproved negative biopsies (increased PCa Possible cost detection) Improves specificity for detection of aggressive PCa (metastatic disease) Extensive validation Applicable regardless of age, PSA, or clinical findings Indicated in Reduces rebiopsapproved Only FDA- 50 year old when men who Extensive repeat biopsy is have had prior validation considered negative prostate FDA-approved Not validated biopsy/s and for in a first whom a repeat biopsy setting biopsy would to distinguish be clinically indolent from recommended aggressive to increase disease specificity of PCa Not performed detection on non-dre urine specimens Refines patient Reduces Not FDAapproved risk stratification negative biopsies of detecting PCa and high grade PCa (GS >6) on diagnostic needle biopsy (increased PCa detection) Improves specificity for detection of high-grade PCa (GS>6) Extensive validation Possible cost Not performed on non-dre urine specimens RP features used as surrogate for aggressive disease *Clinical Laboratory Improvement Amendments (CLIA)-based clinical laboratory-developed test (LDT) phi: Prostate Health Index; FDA: United States Food and Drug Administration; PSA: prostate specific antigen; TMA: transcription mediated amplification; p2psa (aka [-2]proPSA): isoform of free PSA; PCa: prostate cancer; DRE: digital rectal examination; 4K: four prostate-specific kallikreins; hk2: human kallikrein-related peptide 2; PCA3: Prostate Cancer Antigen 3; MiPs: Mi prostate score; GS: Gleason Score; pt pathologic tumor stage; RP: radical prostatectomy. Vol No. 3 MINERVA UROLOGICA E NEFROLOGICA 213

4 The identification of new molecular markers not only may allow a more reliable prediction of the pathological stage of the disease, but can also be useful in the process of selecting therapeutic targets: the ideal scenario would be a panel of molecular biomarkers. New biomarkers, including urine prostate cancer antigen 3 (PCA3) score, prostate health index (phi), and the four-kallikrein panel (4K), have been widely investigated during recent years especially with the aim of detecting aggressive PCa (Table I). Some of these results suggest the ability of these novel biomarkers to improve PSA specificity in the detection of PCa, although there are not enough studies directly comparing different biomarkers to know their complementarity. The relationship with PCa aggressiveness seems to be confirmed for phi and for the four-kallikrein panel, but not for PCA3 score. PSA PSA is an organ-specific marker, and not a cancer-specific marker. Normal, hyperplastic, and neoplastic prostate epithelial cells all produce PSA, with the highest levels found in the prostatic transition zone of patients with benign prostatic hyperplasia (BPH). Neoplastic cells produce somewhat lower and varying tissue levels of total PSA compared to benign hyperplastic epithelial cells although both conditions cause total PSA elevation in the blood. 7 PSA expression is tightly linked to the presence of prostatic cells: as such, increasing serum PSA levels in patients who had their prostate surgically removed indicates the presence of disseminated and eventually growing cells. For the same reason, increasing PSA levels during chemical or after surgical castration can indicate failure of the therapy. While PSA is recognized as a formidable follow-up marker, there is a longstanding debate on its value as a diagnostic marker for early detection of PCa. A major limitation of total PSA as a reliable marker in the diagnostic phase of the disease is that it cannot distinguish between indolent and aggressive disease: PSA is not a classic tu- mor marker, in the sense that levels are not directly correlated with increasing grade/ stage of PCa. Interestingly, PSA expression has been found to decrease with increasing Gleason grade in some studies. 8 PSA levels have been shown to be subjected to yearly fluctuations thereby affecting the interpretation of any single result, and making any single PSA result, with or without combined digital rectal examination (DRE), an unreliable measure for subjecting patients to prostate biopsy. Although more stable than the DRE, occurrences of reversed PSA cut point based decisions to biopsy one or more years later are not uncommon. 9 Variation in total PSA may be due to several factors, including both analytical (i.e., preanalytical sample handling, laboratory processing, assay performance, and standardization) and biological variation (i.e., metabolism, renal elimination, medication, physical and sexual activity, size and integrity of the prostate), but have been found not related to cancer aggressiveness. 10 Several strategies have been investigated to enhance the diagnostic accuracy of serum PSA measurement and improve specificity for detection of aggressive vs. indolent disease. One of the earliest has been the introduction of age-specific total PSA cut-offs 11 aimed to normalize PSA levels to a particular decade of life in an attempt to account for normal prostatic enlargement with age. For instance, this has led to the recommendation by AUA of discontinuing PSA screening among men over age 70 with a PSA below 3 ng/ml. 5 Estimates of PSA kinetics, including PSA velocity (PSAV) and PSA doubling time (PSADT), have been evaluated as a tool to predict the presence of PCa at biopsy and/ or a greater risk of PCa death after radical treatment in several studies with somehow discordant results. 12 A new variation on PSA velocity is called the risk count and implies counting the number of times in a row that the PSAV exceeded a specific threshold (e.g. 0.4 ng/ml/year) in men with serial PSA measurements. 13 The proportion of high-risk PCa increased significantly with 214 MINERVA UROLOGICA E NEFROLOGICA September 2015

5 increasing PSAV risk count, in the original study, 13 and was significantly associated with overall PCa detection on biopsy in another study. 14 Moreover, a PSAV risk count of 2 was recently shown to indicate a significantly greater risk of high-grade disease at biopsy beyond PSA alone and age, 15 whereas in a surveillance cohort a risk count of 3 or 2 provided a significant increase in risk of biopsy reclassification to high-grade disease beyond age, race, PSA density, and presence of cancer at first biopsy. 16 The major limitation of PSA kinetics measurement is intrinsic to the physiologic fluctuations of PSA, which the PSAV risk count could at least in part overcome in determinate clinical settings. PSA density (PSAD) (PSA divided by prostate volume) is a derivative measure of serum PSA that has been promoted as a more specific indicator of PCa risk in patients with a PSA level of <10 ng/ml, and a strong predictor of adverse pathological features and biochemical recurrence after radical prostatectomy (RP). 17, 18 More recently, PSAD has been shown to predict GS upgrade in patients with biopsy Gleason 6 and 3+4=7 cancer on biopsy, but it appears to be less effective in predicting upgrade from GS 7 to 8-10 disease, where the presence of palpable disease and number of cores involved are more predictive. 8, 19 Total PSA circulates in the blood in protein complexes composed of protease inhibitors, such as alpha-1-antichimotripsin. A small fraction of previously inactivated PSA circulates as free PSA. Percentage of free PSA (%freepsa or fpsa) was shown to be a useful marker for the detection of PCa and was evaluated as predictor of biopsy reclassification in men undergoing active surveillance with promising results in several studies although subsequent studies have failed to show such a correlation. 24, 25 The discrepancy of the results may in part be due to differences in patients populations. 25 PHI which propsa (and its most stable isoform, p2psa) has been associated with cancer. 26 The prostate health index (phi) was developed by Beckman Coulter, Inc. in combination with the NCI Early Detection Research Network, and approved by the Food and Drug Administration (FDA) in It results from the combination of total PSA (tpsa), free PSA (fpsa) and [-2]proPSA (p2psa). 27 Preliminary studies have shown that p2p- SA, %p2psa and phi are higher in malignant than in benign prostatic conditions and their use can significantly improve cancer detection with respect to other biomarkers such as tpsa and f/t PSA ratio. 27 At present, two meta-analysis have been performed 28, 29 to assess the usefulness of %[-2]proPSA and phi in PCa detection in the overall PSA range and in the initial and subsequent biopsies. Bruzzese et al. 29 evaluated the diagnostic value of phi in the gray PSA range of 2-10 ng/ml in patients undergoing first biopsy and showed that phi can improve the detection of PCa compared with %fpsa. Phi is best used to distinguish PCa from benign prostatic conditions in men aged 50 years and older with a DRE non suspicious for cancer and a serum PSA between 4 and 10 ng/ml. 30 Lazzeri et al. reported that the utilization of p2psa and phi significantly improved the predictive accuracy for GS 7 PCa detection compared to PSA and fpsa in 650 men with PSA levels between 2 and 10 ng/ml from five European centers. 31 In a later study including a cohort of 150 men with positive family history for PCa, the same investigators demonstrated that phi outperformed tpsa and %fpsa for the detection of aggressive PCa (Table I). 32, 33 Several isoforms of free PSA have been described (BPSA, ipsa, and propsa), among 4Kscore The four prostate-specific kallikrein panel (4Kscore) is a promising serum-based biomarker that consists of total PSA, free PSA, intact PSA, and human kallikrein-related peptide 2 (hk2) The 4Kscore test combines four prostate-specific kallikrein assay results with clinical information, including history (or absence) of a prior negative bi- Vol No. 3 MINERVA UROLOGICA E NEFROLOGICA 215

6 opsy, patient s age, and detection (or not) of nodules upon DRE, in an algorithm that calculates the individual patient s percent risk for aggressive PCa. It is now commercially available in the United States as a Clinical Laboratory Improvement Amendments (CLIA)-based clinical laboratory-developed test (LDT) offered by OPKO Lab, LLC, but has not been FDA approved yet. The clinical utility of 4Kscore test was validated based on a decade of clinical research in over 20,000 men in Europe and in the United States. 34, 35, Despite discrepant results, 44 there are compelling data suggesting a correlation of hk2 with high-grade PCa and extraprostatic extension. 45, 46 Some recent studies have indicated that 4Kscore has the potential to distinguish pathologically insignificant PCa from aggressive disease, and to significantly enhance the prediction of metastatic risk, compared with PSA alone, among men age years with PSA 2 ng/ ml. 37, 47 The kallikrein panel can aid biopsy decision making, reduce unnecessary biopsies, improve patients outcome and cut costs (Table I). 36 PCA3 Score In 1999, Bussemakers et al. 48 were the first to publish their findings regarding a new PCa related gene, DD3. Prostate Cancer Antigen 3 (PCA3), or DD3, is a non-coding RNA, which is produced exclusively in prostate tissue. 49 Neoplastic prostate tissue specifically overexpresses PCA3. 50 The assay compares the concentration of PCA3 mrna levels to PSA mrna levels in post-dre urine specimens to generate a PCA3 score. Gen- Probe, Inc obtained FDA approval (Progensa PCA3) in 2012 with the intended use for patients with a suspicion of PCa based on PSA level and/or DRE and/or one of more negative biopsy results. The rate of PCa detection increased from 0.63 for PSA alone to 0.71 for PSA combined with PCA3. 51 In several studies using large patient cohorts, PCA3 performance appeared superior to PSA PCA3 score might be used together with PSA and other risk factors in nomograms or other methods of risk stratification in making deci- sions regarding whether to perform a first or repeat prostate biopsy. The preferred indication for PCA3 is in patients who have already had a prostate biopsy with negative results; in such cases the sensitivity and specificity averaged 52.6% and 71.6%, respectively, with a positive predictive value (PPV) of ~40% and a negative predictive value (NPV) of ~80%. 55 In a study by Gittelman et al., 56 men with a PCA3 score of less than 25 were 4.56 times more likely to have a negative repeat biopsy than men with a score of 25 or greater.. Some investigators advise that patients should not receive a biopsy if the PCA3 score is lower than Since PCA3 screening has only been evaluated in men previously screened with PSA, the utility of PCA3 in isolation is unknown at the present time. In a study by Bollito et al., 58 PCA3 cutoffs between 39 and 50 had the highest accuracy in an Italian population undergoing repeat biopsy after negative initial biopsy: in this study, a cutoff of 39 could have avoided 51.9% negative repeat biopsies, eventually missing 7.8% of cancers (all low risk); a cutoff of 50 would have prevented 56.5% of negative repeat biopsies, missing 29 tumors (10.3%), 5 (17%) of which potentially aggressive. The PCA3 test performed poorly in the first biopsy group. 58 A relevant finding is that men with high-grade prostatic intraepithelial neoplasia (HGPIN) show comparable PCA3 scores as men with PCa, but PCA3 scores are significantly higher in patients with HG- PIN than in the PCa-negative men. 59 These data are in agreement with previous reports showing that the discriminative performance of PCA3 score is lower between HGPIN and PCa. 59 PCA3 score was recently reported to be a good predictor of low-volume, clinically insignificant PCa, 60 suggesting a promising application in selecting patients for active surveillance. Additional validation studies are necessary to confirm its value (Table I). TMPRSS2-ERG gene fusion combined with PCA3 score A recurring gene fusion between the TMPRSS2 (21q22.3) gene, regulated by androgens, and ERG, from the ETS family 216 MINERVA UROLOGICA E NEFROLOGICA September 2015

7 (21q22.2), has been identified in approximately 50% of PSA-screened PCa Population-based cohort studies have shown an overall lower prevalence of TMPRSS2 ERG gene fusion and an association with clinicopathological parameters of worse outcome; however retrospective studies evaluating the association between TM- PRSS2-ERG fusion and patients outcome following RP have reported mixed results, with some describing an association with worse outcome, 63, some finding no association, 74 and some reporting association with favorable parameters and good prognosis PCa is typically a multifocal disease with a high incidence (41% to 51%) of interfocal heterogeneity for the fusion status. 62, The limitation related to tumor heterogeneity can potentially be mastered by combining TMPRSS2-ERG detection with other markers such as PCA3 and PSA The detection of TMPRSS2-ERG fusion transcripts in urinary sediment obtained after DRE has a high specificity (93%), but a low sensitivity (37%) in predicting PCa diagnosis; the combination of TMPRSS2-ERG fusion transcripts with PCA3 has been reported to improve sensitivity (from 62% for PCA3 alone to 73% when combined), without compromising specificity. 84 The integration of TMPRSS2:ERG transcripts levels in urine with PCA3 score and serum PSA has been validated in ~2000 urine specimens and has shown to predict the diagnosis of PCa with 80% sensitivity and 90% specificity , 85 A commercially available test (the Mi-Prostate Score - MiPS) is offered through the University of Michigan Health System as an early detection test for PCa to provide a quantitative risk assessment of having PCa detected on a subsequent biopsy, and/or to produce a risk assessment of PCa that potentially indicates the likelihood of aggressive cancer (the MiPS high grade cancer risk score). The combination of these biomarkers could help develop practical algorithms to identify which patients need a biopsy after elevated serum PSA levels, leading to considerable reduction of the number of prostate biopsies. Tissue-based genomic/proteomic tests Several genomic/proteomic tests have become commercially available in recent years (Table II). It is important to keep in mind that although these are laboratorydeveloped tests (LDT), offered under CLIA certified laboratories, they have not been approved by the FDA and their validation is somewhat limited. The main concern is the potential variability among different laboratories. Test useful to decide who to rebiopsy and to reduce the number of unnecessary biopsies Confirm MDx Confirm MDx (MDxHealth, Inc.) is a quantitative methylation specific PCR assay detecting an epigenetic field effect associated with the cancerization process based on DNA methylation in cells adjacent to PCa foci Field effect can be present despite tissue normal microscopic appearance. An initial test cohort of 30 cancer-positive and 12 cancer-negative prostate tissue samples was utilized for the development and optimization of an epigenetic multiplex assay based on the GSTP1, APC and RASSF1 genes. 89 The assay has been validated in large European and US cohorts confirming that epigenetic changes can accurately predict if a prostate needle biopsy missed cancer. These recently published studies found that the methylation levels of GSTP1, APC and RASSF1 in non-neoplastic prostate tissue surrounding PCa can discriminate if truly there is no cancer nearby with a 88-90% NPV. 90, 91 The MATLOC study evaluated prostate needle core biopsy tissue samples of 483 patients from the United Kingdom and Belgium with histopathologically negative prostate biopsies, followed by positive (cases =87) or negative (controls =396) repeat biopsy within 30 months. 90 The test performed on the first negative biopsies resulted in 68% sensitivity, 64% Vol No. 3 MINERVA UROLOGICA E NEFROLOGICA 217

8 specificity, and 90% NPV. 90 The DOCU- MENT multicenter trial in the United States evaluated cancer negative prostate biopsy core tissue samples of 320 subjects from 5 urological centers in the United States (Johns Hopkins University, Cleveland Clinic, University of California-Los Angeles, Lahey Hospital and Medical Center, Eastern Virginia Medical School). 91 All subjects underwent repeat biopsy within 24 months with either a negative (controls =228) or positive (cases =92) histopathological result. The test resulted in a NPV of 88%. 91 This epigenetic assay uses Table II. Tissue-based genomic/proteomic tests to help decide who to rebiopsy and/or who to treat. Confirm MDx PCMT Test Type of assay Markers Type of sample Oncotype DX (aka GPS) Prolaris (aka CCP score) ProMark Decipher (aka GC) Quantitative methylation specific PCR assay Quantitative RT-PCR assay specific for mtdna Multi-gene RT-PCR assay RNA expression-based assay Fully automated immunofluorescent imaging platform Whole-transcriptome microarray assay GSTP1, APC, RASSF1 12-core PBx- FFPE tissue within 24 months mtdna deletions 17 genes (12 cancerrelated + 5 reference genes) 46 genes (31 cell cycle progression and 15 housekeeping genes) PBx FPE PBx (as little as 1 mm cancer) FFPE PBx or RP 8 proteins FFPE PBx 22 coding and noncoding RNAs FFPE RP tissue PBx: prostate needle core biopsy; FFPE: formalin-fixed paraffin embedded; FPE: fixed paraffin embedded; PCMT: Prostate Core Mitomic Test; mtdna: mitochondrial DNA; GPS: Genomic Prostate Score; CCP: cell cycle progression; RP: radical prostatectomy; GC: Genomic Classifier; NPV: negative predictive value. formalin-fixed paraffin embedded (FFPE) prostate tissue collected during a 12-core biopsy and is designed to help men with a negative biopsy confirm the absence of cancer in their prostate (true negatives) and avoid unnecessary repeat biopsies, and to identify patients at high-risk for occult disease (false negatives). This epigenetic assay could lead to a meaningful decrease in the number of repeat biopsies (Table II). If the test detects a nearby tumor, it does not determine its aggressiveness and repeat biopsies are required. This test is limited to tissue submitted 218 MINERVA UROLOGICA E NEFROLOGICA September 2015

9 Application Strengths Limitations Improves PCa detection/reduces rebiopsy Improves PCa detection/reduces rebiopsy Refines pretreatment risk assessment beyond conventional clinicopathological criteria to help active surveillance vs. treatment decision On PBx: refines pretreatment cancer risk beyond clinicopathological criteria to help active surveillance vs. treatment decision On RP: determines post-operative risk of adverse outcome to support adjuvant/more aggressive therapy Stratifies patients for active surveillance versus therapeutic intervention Refines post-operative risk in men with high-risk features at RP to help decide who will benefit from adjuvant/more aggressive therapy within 24 months of the biopsy. High NPV (true negative biopsy reduces the need for re-biopsy) Identifies latent disease (reduces false negative biopsy) Large number of samples tested (see text for references) High NPV (true negative biopsy reduces the need for re-biopsy) Identifies latent disease (reduces false negative biopsy) Distinguishes aggressive vs. indolent disease Small cancer tissue needed for testing Design study accounts for PCa heterogeneity Validated on several cohorts (see text for references) Prostate Core Mitomic Test The Prostate Core Mitomic Test (PCMT) is a quantitative RT-PCR assay based on the unique structural and functional characteristics of mitochondrial DNA (mtdna). MtDNA comprises a circular genome of approximately 16,500 bp, coding for 37 genes. The small size and the accelerated somatic mutation rate make mitochondrial genome a highly attractive system for biomarker discovery and disease detection. 92 Large-scale Distinguishes aggressive vs. indolent disease Validated on several cohorts (see text for references) Distinguishes favorable vs. nonfavorable pathology at RP Predicts the probability of metastatic disease after surgery Multiple validation studies Does not distinguish aggressive vs. indolent latent disease Does not distinguish aggressive vs. indolent latent disease Genes of aggressive disease identified in clinically localized PCa with clinical recurrence endpoint In pre-operative setting, accuracy of test depends on accuracy of biopsy RP features used as surrogate to evaluate test performance in identifying aggressive disease Only one validation study to date (see text for reference) deletions (3.4kb) in mtdna are indicative of cellular changes associated with the development of PCa; these type of deletions are absent in normal prostate epithelium, but extremely common in PCa and adjacent tissue In a pilot study of 38 benign prostate biopsies, 29 malignant (GS 6 and 7), and 41 histologically normal prostate samples in proximity to PCa, Maki et al. (Table II) reported a significant difference in mtdna deletion between benign and malignant biopsy specimens with a sensitivity and specificity of 80% and 71%, respectively. Histologically normal prostate tissue Vol No. 3 MINERVA UROLOGICA E NEFROLOGICA 219

10 proximal to cancer was indistinguishable from malignant samples. 93 In a confirmation study including 98 benign prostate biopsies, 75 malignant, and 123 normal biopsy specimens adjacent to a malignant needle core specimen, the 3.4 kb deletion was accurate in predicting the presence of a malignant focus in 67% of the patients. The performance of the deletion improved to 78% when the investigators examined 18 patients who had two or more previous samples. 93 Using prostate biopsy tissue, PCMT can determine the presence of malignant cells by detecting underlying molecular alterations commonly found in prostate tumor or in normal appearing tissue surrounding the tumor via a cancerization field effect. 94, 96, 97 PCMT can discriminate if a prostate biopsy is truly negative of if it missed the tumor and confirm the need for a repeat biopsy (Table II). If both the prostate biopsy and the PCMT are negative, the test predicts that there is no need for a repeat biopsy with high accuracy. If the test detects changes consistent with the presence of a nearby PCa, it cannot determine its aggressiveness and repeat biopsies will be required. Additional studies are necessary to further validate the performance of this test. Markers useful to distinguish aggressive from indolent tumors and decide who to surveil or treat Oncotype DX The Oncotype DX Prostate Cancer Assay (Genomic Health, Inc.) is a multi-gene RT-PCR expression assay that was developed for use with fixed paraffin-embedded (FPE) diagnostic prostate needle biopsies containing as little as 1 mm of prostate tumor in the greatest dimension. 98 The Oncotype DX Prostate Cancer Assay includes 5 reference genes and 12 cancer genes representing distinct biological pathways with a known role in prostate tumorigenesis: androgen pathway (AZGP1, KLK2, SRD5A2, and FAM13C), cellular organization (FLNC, GSN, TPM2, and GSTM2), proliferation (TPX2), and stromal response (BGN, CO- L1A1, and SFRP4). 98 Reference normalized expression of the 12 cancer-related genes are used to calculate the Genomic Prostate Score (GPS), which ranges from 0 (low) to 100 (high), with higher scores indicating more aggressive disease. 99 The GPS has been shown to predict adverse PCa pathology beyond conventional clinical/pathologic criteria commonly used in nomograms. 100 Three studies were conducted to identify this gene signature, and are referred to as the prostatectomy study, the biopsy study, and the validation study: 99 in the prostatectomy study, the expression of 732 candidate genes was analyzed in a stratified cohort of 501 men who underwent RP at the Cleveland Clinic between 1987 and 2004, including 127 with clinical recurrence and 374 without clinical recurrence, with a 1:3 ratio of recurrent to non-recurrent patients. Eighty-one candidate genes selected from the prostatectomy study were assayed using the same methods in the biopsy study. The biopsy study included FPE prostate needle biopsy specimens from a separate cohort of 167 patients who had a diagnostic biopsy and underwent RP within 6 months of diagnosis at the Cleveland Clinic between 1999 and Of the 81 genes evaluated in the biopsy study, quantitative RT-PCR analysis of tumor from FPE prostate needle biopsy tissue for each patient confirmed the association of 58 genes (72%) with adverse pathology at prostatectomy. From the 58 genes, 12 genes associated with PCa aggressiveness and 5 reference genes were selected for inclusion in the final signature, called the GPS. 98, 99 The third, independent, clinical validation study was conducted to determine whether the 17-gene signature could be measured in prostate biopsies to predict adverse pathology and improve risk stratification at diagnosis. The final evaluable population in the validation study was composed of 395 prostatectomy patients with a mix of low to low-intermediate clinical risk characteristics, 31% of which had high-grade or non organ confined (non- OC) disease at RP. 99 The prespecified primary end point was the ability of GPS to 220 MINERVA UROLOGICA E NEFROLOGICA September 2015

11 Figure 1. Genomic Prostate Score (GPS) biopsy result. In a patient with low risk clinicopathological features according to the National Comprehensive Cancer Network (NCCN) risk group (ct1-t2a, bx GS6, PSA 10 ng/ml), a low GPS could refine classification by assigning the patient an individualized percentage of risk that could result in a risk consistent with the clinical criteria alone grey line or even a more favorable light grey line, whereas a high GPS could reassign the patient to a less favorable category than clinical criteria alone, impacting on the active surveillance versus treatment decision. predict prostatectomy grade and stage adjusting for biopsy GS. In separate multivariable analyses adjusting for significant clinical covariates (such as cancer of the prostate risk assessment [CAPRA] score, National Comprehensive Cancer Network [NCCN] group, or, taken together, age, PSA, clinical stage, and biopsy GS) GPS was a consistent predictor of high-grade and/or non-oc pathology. In decision-curve analysis, the combination of clinical (CAPRA) and genomic (GPS) information yielded greater net benefit than clinical information alone. 99 Oncotype Dx was subsequently validated as a biopsy-based genomic assay for localized PCa to predict adverse pathology (distinguishing between clinically indolent and aggressive disease) in a racially diverse cohort of 402 men treated for NCCN very low, low or low-intermediate-risk PCa between 1990 and 2011 at two US military medical centers, including a high proportion of African-American men. 100 In univariate and multivariate analyses (adjusted for clinical and pathologic factors from the univariate analysis) GPS was strongly associated with time to metastasis and adverse pathology (including high-grade and non-oc disease) at RP, respectively. This study also concluded that GPS was a strong measure of PCa aggressiveness independent of clinical and pathological parameters including race. Incorporation of GPS would be expected to lead to fewer treatments of patients who Vol No. 3 MINERVA UROLOGICA E NEFROLOGICA 221

12 have favorable pathology at prostatectomy without increasing the number of patients with adverse pathology left untreated (Figure 1). Additional studies are necessary to determine whether GPS might help identify PCa patients with indolent tumors who may be ideal candidates for lower frequency of surveillance biopsies. 99 Potential limitations of the GPS are intrinsic to its development (genes associated with aggressive PCa were indeed identified in clinically localized tumors) and validation (original validation study included men with biopsy Gleason 3+4, whereas in the primary analyses, definition of high grade was restricted to primary Gleason pattern 4 or any pattern 5). 99 The major strength of GPS relies on the test being able to assess the underlying tumor biology/aggressiveness from very small tumor biopsy samples, addressing the issue of tumor heterogeneity and prostate biopsy under-sampling. The inclusion of GPS in patients management may enable more accurate identification of a larger group of men who can more confidently choose active surveillance as an initial management strategy. Prolaris The Prolaris cell cycle progression (CCP) test (Myriad Genetic Laboratories Inc.) is a novel RNA expression-based assay which uses a prognostic RNA signature that directly measures tumor cell growth characteristics in order to stratify patients with localized prostate cancer according to disease aggressiveness. The assay is based on determining and combining the expression levels of 31 cell cycle progression genes (FOXM1, CDC20, CDKN3, CDC2, KIF11, KIAA0101, NUSAP1, CENPF, ASPM, BUB1B, RRM2, DLGAP5, BIRC5, KIF20A, PLK1, TO- P2A, TK1, PBK, ASF1B, C18orf24, RAD54L, PTTG1, CDCA3, MCM10, PRC1, DTL, CEP55, RAD51, CENPM, CDCA8, and ORC6L) and 15 housekeeping genes (RPL38, UBA52, PSMC1, RPL4, RPL37, RPS29, SLC25A3, CLTC, TXNL1, PSMA1, RPL8, MMADHC, RPL13A, LOC723658, PPP2CA, MRFAP1) on FFPE tissue obtained either from prostate biopsy or RP specimens. 101 The genes were selected because of their correlation with PCa proliferation and were not intended to capture information related to other factors (such as invasive potential). 101 The gene-signature was tested in two separate cohorts, one represented by 366 men treated with RP between identified through the tumor registry at the Scott and White Clinic, Temple, TX, USA (RP-cohort); the other composed of 337 men diagnosed with clinically localized PCa on transurethral resection of the prostate (TURP-cohort) between 1990 and 1996 (inclusively), who were younger than 76 years at the time of diagnosis, had a baseline PSA measurement, and had not received hormonal therapy. 101 The CCP score was shown to be a strong prognosticator of biochemical recurrence and death after disease progression in the RP-cohort, as well as death in the TURP-cohort, independent of several clinicopathological parameters, including GS, stage, and PSA in the RP-cohort, and GS, PSA, cancer extent and Ki-67 in the TURP-cohort. 101 The prognostic value of the CCP signature was confirmed in subsequent studies, with a low expression being associated with low risk of disease progression Prolaris test has been validated in multiple cohorts 102, and provides a risk assessment of PCa-specific progression and disease-specific mortality when combined to standard clinicopathologic parameters, such as PSA and GS. 102 CCP score generated from prostate biopsy samples from 582 patients treated with RP was significantly associated with biochemical recurrence and metastasis. 105 A recent review and meta-analysis summarizes the evidence on the value of the CCP score in elucidating the aggressive potential of PCa in an individual patient and how the score results may influence clinicians treatment decisions. 107 When the test is performed on biopsy tissue, it may help identify both low-risk patients who can be managed conservatively with active surveillance and high-risk patients who may benefit from earlier definitive treatment (Figure 2); when it is performed on RP tissue, it helps determine the postoperative risk of adverse outcome and if postoperative treatment is 222 MINERVA UROLOGICA E NEFROLOGICA September 2015

13 YBX1) were identified that predicted both prostate pathology aggressiveness and lethal outcome despite sampling error, based on 380 prostatectomy tissue samples. The test has been further refined to the current 8-biomarker protein signature in a study of 381 patient biopsies with matched prostatectomy specimens, followed by a second blinded validation study of 276 cases to distinguish favorable versus non-favorable pathology at RP, independently and relative to current risk classification systems (NCCN and D Amico). 109 The development study included a multi-institutional cohort of 381 patients with biopsy and matched prostatectomy samples. Biopsy sample inclusion/ exclusion criteria were defined as those that would be in place during routine clinindicated (Table II). 104 For biopsy tissue, the accuracy of the test depends on the accuracy of the biopsy and may add unnecessary expense in following very low risk PCa patients. ProMark ProMark (Metamark Genetics) is a biopsy-based prognostic test, from a CLIA-certified laboratory, detecting eight protein biomarkers (CUL2, DERL1, FUS, HSPA9, PDSS2, ps6, SMAD4, YBX1) using a fully automated immunofluorescence imaging platform. In the original study, 108 starting with a large pool of potential candidates, 12 biomarkers (ACTN1, CUL2, DERL1, FUS, HSPA9, PDSS2, PLAG1, ps6, SMAD2, SMAD4, VDAC1, and Figure 2. Prolaris biopsy test result. Similarly to GPS, a low versus high Prolaris score in a patient within the intermediate risk category according to D Amico/AUA (PSA >10, <20 ng/ml, or bx GS 7, or ct2b not qualifying for high-risk) may refine the patient s risk of cancer aggressiveness and 10-year prostate cancer specific mortality as provided in the Prolaris biopsy test results, making the patient more suitable for active surveillance than clinical criteria alone would have. Vol No. 3 MINERVA UROLOGICA E NEFROLOGICA 223

14 ical use of the assay, with further specification that, since they are usually not candidates for active surveillance, patients with biopsy Gleason 4+3 were excluded, except for a limited number (N.=28) of biopsies that had been discordantly graded as both 3+4 and 4+3 by two expert pathologists. The separate and independent validation cohort was represented by 276 biopsy samples with matched RP from patients managed at the University of Montreal Hospital Center, Canada: inclusion criteria were biopsies with a centralized GS 3+3 or 3+4 (17 biopsies with discordant grading by two expert pathologists of 3+4 and 4+3 were included as well), and matched prostatectomy with pathologic TNM staging, PSA level, and resulting surgical GS. 109 The development study goal was to define a model able to distinguish between prostate pathology usually recommended for active surveillance (surgical Gleason 3+3 and T3a) versus those more likely to require prostatectomy (surgical Gleason 3+4 or non-localized >T3a, N, or M). The validation study goals were expanded to include two co-primary endpoints, one defined as favorable pathology (surgical Gleason 3+4 and organ-confined disease [ T2]) versus non-favorable pathology (surgical Gleason 4+3 or non-organ-confined disease [T3a, T3b, N, or M]), and the second defined as GS 6 pathology (surgical Gleason of 3+3 and localized disease [ T3a]) versus non-gs 6 pathology (surgical Gleason 3+4 or non-localized disease [T3b, N, or M]). At a biomarker risk score (ranging from 0 to 1) of 0.33, predictive values for favorable pathology in very low- and lowrisk NCCN and low-risk D Amico groups were 95%, 81.5%, and 87.2%, respectively, higher than for these current risk classification groups alone (80.3%, 63.8%, and 70.6%, respectively). The predictive value for nonfavorable pathology was 76.9% at biomarker risk scores >0.8 across all risk groups. Increased biomarker risk scores correlated with decreased frequency of favorable cases across all risk groups. The validation study met its two co-primary endpoints, separating favorable from non-favorable pathol- ogy (AUC, 0.68, P<0.0001, odds ratio=20.9) and GS-6 versus non-gs-6 pathology (AUC, 0.65, P<0.0001, OR=12.95). Some of the limitations of this test are again intrinsic to its development and validation: GS and TNM staging of matched prostatectomy samples were used to evaluate performance of the 8-biomarker assay, instead of clinical progression (such as metastatic disease or PCa-specific death), underpowering the aim of the study. The relatively low grade and early stage of the target population would indeed make it necessary to have very long follow-up of very large and conservatively managed cohorts in order to be sufficiently powered. Including a significant proportion of high-risk cases (biopsies with GS 4+3) may have in part compensated for the power, although, in such cases, the clinical utility of a complementary molecular test is admittedly more limited. 109 Moreover, frequent genetic differences between regions of individual tumors and limited tumor sampling by needle biopsy pose challenges to molecular assays developed and intended for use in biopsy PCa tissue. 99 The 8-biomarker signature derives from the 12-biomarker set originally discovered in a study designed to overcome the sampling error issue by evaluating the predictive ability of a large number of protein biomarker candidates in paired low-grade and high-grade tissue microarray cores from the prostatectomy sample of each patient. 108 The volume of cancer tissue required for the analysis may also be a limiting factor for such a test to be effectively applicable to very low- or low-risk PCa patients (Table II). Decipher test (Genomic Prostate Cancer Classifier) Only a minority of PCa patients with adverse pathology at RP and even post-operative biochemical recurrence (BCR) experience metastatic progression and die of disease. 110 Improved post-surgical risk prediction models using genomic information may help clinicians to better weigh the risk of metastasis and the morbidity and costs of adjuvant or salvage treatments. 224 MINERVA UROLOGICA E NEFROLOGICA September 2015

15 Decipher (GenomeDx Biosciences, San Diego, CA, USA) is a postoperative genomic classifier (GC) test that uses a wholetranscriptome microarray assay from FFPE PCa specimens, developed and validated as a specific predictor of metastases and PCa-specific mortality following RP in men with adverse pathologic features, as defined by high GS, extraprostatic extension (EPE), seminal vesicle invasion (SVI), or lymph node positivity. 111, 112 The GC consists of 22 markers (LASP1, IQGAP3, NFIB, S1PR4, THBS2, ANO7, PCDH7, MYBPC1, EPPK1, TSBP, PBX1, NUSAP1, ZWILCH, UBE2C, CAMK2N1, RABGAP1, PCAT-32, GLYATL1P4/PCAT-80, TNFRSF19, and three other, labeled as intronic, coding antisense, and non-coding transcript ) corresponding to RNAs from coding and nonprotein coding regions of the genome. GC was developed from analysis of more than one million coding and non-coding RNAs in 545 patients who received RP for primary PCa as first line treatment at the Mayo Clinic Comprehensive Cancer Center between 1987 and 2001, including 192 cases (patients with metastatic disease in the follow up period) and 353 controls (patients with no evidence of disease or BCR only in the follow up period). 111 The GC outputs a continuous variable score ranging between 0 and 1, where a higher score indicates a higher probability of clinical metastasis. 111 The GC was compared against a Clinical Classifier (CC), including pathologic GS, preoperative PSA; positive surgical margin status (SM+), SVI, EPE and positive lymph node status (LN+), and a combined genomic and clinical classifier (GCC). 111 The 545 patients were divided in a matched training (N.=359) and validation (N.=186) set, using a method described in a previous study. 113 Analyses showed that both in the training set and in the validation set GC and GCC had the highest ROC area-under the curve (AUC) for predicting cases (training set: 0.90 and 0.91; validation set: 0.75 and 0.74, respectively). The clinical-only CC had an AUC of 0.69 in the validation set, which was only marginally better than pathological GS alone (0.65). When dichotomized into low ( 0.5) and high (>0.5) GC risk groups, the odds ratio for predicting cases was 6.79 (95% CI: ), more than twice the odds ratio of GS (OR: 3.02 [95% CI: ]). In multivariable analysis, after adjustment for post-rp treatment, GC remained the only significant prognostic variable (P<0.001). The more direct measure of tumor biology provided by the 22-marker expression signature adds independent significant prognostic information for prediction of early metastasis after rising PSA, which is not captured by the clinical variables available from pathological analysis. The GC has been validated in several cohorts of RP treated men with high-risk of recurrence, including those treated with post- operative radiation (post-rp RT). 111, 112, In a subsequent study from the same institution (Mayo Clinic), 114 the GC confirmed to improve performance over any individual clinicopathological variable or multivariable prediction model in predicting metastatic disease after RP, including the previously described CC and combined GCC, 111 as well as scores of two validated prediction models (the GPSM [Gleason, prostate specific antigen, seminal vesicle and margin status] 117 and the Stephenson nomogram). 118 In a separate study, 116 GC was found able to discriminate amongst patients with BCR to help distinguish those who will progress to clinical metastasis from those who will not, outperforming clinical-only models such as the Cancer of the prostate risk assessment post-surgical (CAPRA-S) and Stephenson nomograms, with greater net benefit across a wide range of risk. As is the case with pathologic features of prostate tumors at prostatectomy, molecular characteristics that define these tumors might augment the prediction of disease aggressiveness at the time of BCR and help guide the decision to employ salvage treatments and avoid overtreatment. 116 In a subsequent study from a different institution, 112 GC has also been shown to add independent prognostic value to post-surgical models of risk assessment including CAPRA- S and post-rp Stephenson nomogram in predicting metastatic disease within 5 years after surgery (rapid metastasis) in 169 patients Vol No. 3 MINERVA UROLOGICA E NEFROLOGICA 225