Molecular Biomarkers in the Clinical Management of Prostate Cancer

Abstract

Prostate cancer, one of the most common noncutaneous malignancies in men, is a heterogeneous disease with variable clinical outcome. Although the majority of patients harbor indolent tumors that are essentially cured by local therapy, subsets of patients present with aggressive disease or recur/progress after primary treatment. With this in mind, modern clinical approaches to prostate cancer emphasize the need to reduce overdiagnosis and overtreatment via personalized medicine. Advances in our understanding of prostate cancer pathogenesis, coupled with recent technologic innovations, have facilitated the development and validation of numerous molecular biomarkers, representing a range of macromolecules assayed from a variety of patient sample types, to help guide the clinical management of prostate cancer, including early detection, diagnosis, prognostication, and targeted therapeutic selection. Herein, we review the current state of the art regarding prostate cancer molecular biomarkers, emphasizing those with demonstrated utility in clinical practice.

From early detection through diagnosis and treatment, the clinical management of prostate cancer is a multistep process that necessitates a collaborative, multidisciplinary approach, and, at each step, established and emerging molecular biomarkers are available to help guide clinicians. Although prostate cancer is the most common noncutaneous malignancy in American men, for the majority of patients, it is an indolent disease with excellent long-term survival rates (Thompson et al. 2013; Siegel et al. 2017). Thus, recent clinical paradigms have focused on decreasing the usage of prostate biopsies in clinically asymptomatic men and deferring definitive treatment of men with low-grade tumors until the disease progresses to a more aggressive form (i.e., active surveillance) (Shieh et al. 2016; Tosoian et al. 2016); these efforts aim to reduce the identification (overdiagnosis) and overtreatment of patients with indolent disease, respectively. On the other hand, patients who either present with or subsequently develop metastatic prostate cancer are not typically treated with curative intent, and few therapeutic options are available to relieve clinical symptoms (i.e., bone pain, spinal cord compression, hypercalcemia, etc.) related to increasing tumor burden. Thus, there are clear opportunities for improvement in prostate cancer early detection, diagnosis, and prognostication, as well as a potential role for targeted therapeutic approaches in patients with advanced disease.

Advances in our understanding of the oncogenesis of prostate cancer, as well as recent technological innovations, have spurred the development of numerous proposed biomarkers, which represent a broad spectrum of macromolecules, including DNA, RNA, and protein, and are assayed in a variety of patient samples, including blood, urine, and tissue. Despite this multitude of potential biomarkers, however, relatively few have been successfully translated to clinical practice, in part because the clinical utility of a given biomarker depends on its estimated net benefit over existing clinicopathologic models. For example, histopathologic review of prostate core biopsy and radical prostatectomy material remains a fundamental aspect of prostate cancer clinical management; in particular, histologic grading via assignment of a Gleason score (GS) and clinical and/or pathologic staging via the American Joint Committee on Cancer (AJCC) tumor/node/metastasis (TNM) system are effective for risk-stratifying patients for clinical decision-making purposes. Thus, for a proposed new prostate cancer molecular biomarker, the most important issue for clinical translation is whether its inclusion significantly improves the performance of existing optimized models that incorporate routinely available clinicopathologic information. Importantly, this is different than just showing statistically significant independence as a predictive/prognostic factor in a multivariate model and, correspondingly, is a much higher bar for a potential biomarker to clear (Kattan 2003). Likewise, molecular biomarkers much show clear improvement over clinicopathological parameter improvement, which likely can be implemented with little to no cost. For example, Sauter et al. (2016) recently showed through assessment of more than 10,000 cases that, although the GS is a strong prognostic parameter for predicting biochemical recurrence, further stratification of GS 7 tumors by % pattern 4 markedly improves the prognostic performance. Hence, to truly assess the benefit of molecular prognostic biomarkers, such biomarkers should be added to models including quantitative Gleason scoring to truly assess their ability to improve on optimized clinicopathological models. With this in mind, this article focuses on current and emerging biomarkers likely to be used in the clinical management of prostate cancer.

EARLY DETECTION

Current early detection programs for prostate cancer have two main goals: (1) identify asymptomatic men with prostate cancer for prostate core biopsy, and (2) determine which men with prostate cancer would benefit from definitive treatment. Indeed, for the majority of patients, these are the central issues in the evaluation and clinical management of prostate cancer and, as such, much of the effort in the development, validation, and application of prostate cancer molecular biomarkers has been focused on improving these clinical objectives.

Serum total prostate-specific antigen (PSA) is the canonical prostate cancer biomarker, and despite recent debate about its role in early detection programs, it remains the most commonly used biomarker in the clinical management of prostate cancer (Fig. 1A). PSA, which is encoded by KLK3 and regulated by androgen receptor signaling, is a highly prostate-specific protein macromolecule that is expressed predominantly in prostatic tissue, serum, and seminal fluid (Balk et al. 2003; Lilja et al. 2008; Thorek et al. 2013). Similar to other serine proteases in the kallikrein family, PSA is translated as a preproenzyme (preproPSA), which undergoes posttranslational processing into the proenzyme (proPSA) and final active forms via trypsin-like proteases (including human kallikrein 2 [hK2], another prostatic-specific member of the kallikrein family encoded by KLK2), although a subset of proPSA molecules are instead processed into catalytically inactive forms (e.g., [-2]proPSA). In serum, PSA molecules circulate either in the active form complexed to protease inhibitors (e.g., α-1 antitrypsin) or uncomplexed as one of several inactive forms (collectively known as “free PSA”); these free PSA molecules include both intact and nicked or multichain forms. Thus, serum total PSA represents the sum of complexed active PSA and complexed free PSA molecules.

Figure 1.

Molecular biomarkers for early detection of prostate cancer. (A) Serum-based molecular biomarkers for early detection of prostate cancer include total prostate-specific antigen (PSA), free PSA, intact PSA, [-2]proPSA, human kallikrein 2 (hK2), MIC1, and MSMB. % free PSA is calculated as the percentage of free PSA (including intact PSA and [-2]proPSA) compared with total PSA (which includes complexed PSA). The Prostate Health Index (phi) test combines assessment of total PSA, free PSA, and [-2]proPSA. The 4K panel measures total PSA, free PSA, intact PSA, and hK2. The STHLM3 model includes total PSA, free PSA, intact PSA, hK2, MIC1, and MSMB. (B) Urine-based molecular biomarkers for early detection of prostate cancer include PCA3, ERG, PSA, TMPRSS2:ERG, DLX1, HOXC6, TDRD1, and SPDEF. The Food and Drug Administration (FDA)-approved Progensa PCA3 assay only measures PCA3, whereas the Mi-Prostate Score (MiPS) also includes assessment of serum total PSA and urine TMPRSS2:ERG. In addition to serum total PSA, the model proposed by Leyten et al. measures urine DLX1, HOXC6, and TDRD1 (which are normalized to urine PSA) (Leyten et al. 2015). The ExoDx Prostate IntelliScore (EPI) combines serum total PSA with urine PCA3 and ERG (which are normalized to urine SPDEF).

For many patients who undergo biomarker testing for prostate cancer, serum total PSA is the only value assayed. In general, increasing serum total PSA is associated with an increasing likelihood of prostate cancer and risk of clinically significant disease (GS ≥ 7); however, serum total PSA is also positively correlated with increasing age and prostate gland volume (i.e., benign prostatic hypertrophy), and serum total PSA may also be elevated in patients with other benign conditions, such as prostatitis (Balk et al. 2003; Lilja et al. 2008; Thorek et al. 2013). Therefore, in clinically asymptomatic men undergoing early detection for prostate cancer, an isolated elevated serum total PSA value needs to be interpreted with caution, and it is now recommended that, before initiating a PSA-based early detection program for prostate cancer, an informed discussion with the patient regarding the probability of prostate cancer, risk of clinically significant disease, available treatment options, and long-term clinical prognosis should be undertaken (Mottet et al. 2017).

There is no universally accepted lower threshold for serum total PSA in prostate cancer early detection programs, and most modern programs use age-adjusted values based on distributions in presumed cancer-free populations (Mottet et al. 2017). As indicated above, however, with increasing levels of serum total PSA, there is a higher likelihood of prostate cancer and risk of clinically significant disease (Balk et al. 2003; Lilja et al. 2008; Thorek et al. 2013). For example, for the majority of patients with serum total PSA ≥10 ng/mL, no additional biomarker tests are typically necessary for risk-stratification within prostate cancer early detection programs (Mottet et al. 2017). Indeed, patients with serum total PSA ≥10 ng/mL and a negative prostate biopsy typically undergo repeat biopsy and/or close clinical follow-up to exclude clinically significant prostate cancer; similarly, patients with serum total PSA ≥20 ng/mL have an increased risk of high-grade (GS ≥ 8) prostate cancer and, regardless of the biopsy results, may undergo radiographic workup (i.e., bone scan) to evaluate for possible metastatic disease (Mottet et al. 2017). On the other hand, for patients with serum total PSA <10 ng/mL, the false positive rate is high (with a corresponding low positive predictive value for prostate cancer), and many of the detected tumors are relatively low grade (GS6) (Balk et al. 2003; Lilja et al. 2008; Thorek et al. 2013).

To augment the clinical utility of serum total PSA, several additional PSA-based metrics have been devised, including age- and race-adjusted PSA cutoffs, PSA trajectory, and PSA density (utilizing an estimated prostate gland volume from transrectal ultrasonagraphy); although, in general, these measurements only incrementally add to the positive predictive value for prostate cancer in patients with serum total PSA <10 ng/mL (Balk et al. 2003; Lilja et al. 2008; Thorek et al. 2013). Thus, additional serum-based protein biomarkers, including serum free PSA, intact PSA, [-2]proPSA, and hK2, have been developed to improve prostate cancer detection and risk stratification in these patients (Fig. 1A) (Thorek et al. 2013). Indeed, increasing serum free PSA, intact PSA, [-2]proPSA, and hK2 are all independently associated with a higher risk of prostate cancer in clinically asymptomatic men, and higher serum hK2 levels are correlated with an increased likelihood of clinically significant disease (Recker et al. 2000; Mikolajczyk et al. 2001; Steuber et al. 2002, 2007). In addition to absolute serum levels, the serum free-to-total PSA and intact-to-free PSA ratios are associated with an increased risk of prostate cancer in patients with serum total PSA <10 ng/mL (Bjork et al. 1996; Steuber et al. 2002).

Two combinatorial serum protein biomarker assays have recently been validated for prostate cancer early detection programs (in particular, patients with serum total PSA <10 ng/mL): the Prostate Health Index (phi) test and four kallikrein (4K) panel (Fig. 1A). The phi test combines serum total PSA, free PSA, and [-2]proPSA into a single score, and an increasing phi score is associated with increased likelihood of prostate cancer and risk of clinically significant disease (Loeb et al. 2014; de la Calle et al. 2015; Tosoian et al. 2017a,b). The 4K panel, on the other hand, combines serum total PSA, free PSA, intact PSA, and hK2 into a single score, and an increasing 4K score is associated with a higher risk of prostate cancer and likelihood of clinically significant disease (Vickers et al. 2008, 2010a,b, 2011; Loeb et al. 2014; Parekh et al. 2014; Bryant et al. 2015; Stattin et al. 2015). Importantly, both the phi test and 4K panel outperform their individual components (including total PSA), although they show similar clinical value when compared with one another (Nordström et al. 2014). Likewise, Gronberg et al. (2015) showed the additive performance of the serum-based protein biomarkers MSMB, MIC1, free PSA, intact PSA, and hK2 to serum total PSA for predicting clinically significant prostate cancer in a large prospective Swedish early detection cohort (Fig. 1A).

In addition to new serum-based protein biomarkers, several novel urine-based RNA biomarkers have been developed for prostate cancer early detection programs, including prostate cancer–specific molecules such as the long noncoding RNA (lncRNA) PCA3 and TMPRSS2:ERG gene fusion transcripts (Fig. 1B). Increasing urine PCA3 scores are associated with an increased likelihood of prostate cancer in clinically asymptomatic men, and this association is independent of serum total PSA (Bussemakers et al. 1999; de Kok et al. 2002; Hessels et al. 2003, 2007; Ankerst et al. 2008; Deras et al. 2008; Chun et al. 2009; Auprich et al. 2010). Thus, urine PCA3, assessed by the Progensa PCA3 assay, has been approved by the Food and Drug Administration (FDA) for prostate cancer risk assessment in patients with a prior negative prostate core biopsy who are being considered for a repeat biopsy; for these patients, a high PCA3 score indicates a high risk of clinically occult prostate cancer (and, therefore, need for repeat biopsy), whereas a low PCA3 score signifies a low risk of prostate cancer (for which repeat biopsy can be deferred). Similar to urine PCA3, urine TMPRSS2:ERG shows high specificity for prostate cancer (Laxman et al. 2006, 2008); however, because the TMPRSS2:ERG gene fusion is only identified in approximately 40%–50% of prostate cancers (Tomlins et al. 2005; Mehra et al. 2007), its use as a stand-alone assay is limited owing to lack of sufficient sensitivity. A combinatorial assay (Mi-Prostate Score [MiPS]) that includes serum total PSA, urine PCA3, and urine TMPRSS2:ERG, however, shows incremental improvement in detecting both prostate cancer and clinically significant disease in patients with serum total PSA <10 ng/mL compared with serum total PSA alone (Fig. 1B) (Tomlins et al. 2011, 2016; Salami et al. 2013; Merdan et al. 2015; Sanda et al. 2017). More recently, assessment of DLX1, HOXC6, and TDRD1 urine messenger RNA (mRNA) transcripts, normalized to PSA, has been shown to improve detection of clinically significant prostate cancer when combined with serum total PSA (Leyten et al. 2015; Van Neste et al. 2016); likewise, an assay that detects PCA3 and ERG urine transcripts from non-DRE collected whole urine exosomes (normalized to SPDEF expression) to detect clinically significant prostate cancer when combined with serum total PSA has also recently been validated (Fig. 1B) (McKiernan et al. 2016).

Finally, although our understanding of germline variants associated with an increased prostate cancer risk is incomplete, it is clear that subsets of patients with germline mutations in specific genes (e.g., HOXB13, BRCA1, BRCA2) are at an increased risk of developing prostate cancer (Castro and Eeles 2012; Ewing et al. 2012), and germline mutations in DNA repair genes (i.e., BRCA1, BRCA2, ATM, CHEK2, RAD51D, and PALB2) are enriched in prostate cancer patients with metastatic disease (Pritchard et al. 2016). In addition, a number of single nucleotide polymorphisms (SNPs) identified via genome-wide association studies (GWAS) are associated with an increased prostate cancer risk (Al Olama et al. 2009, 2014; Eeles et al. 2009), although the effect size of these SNPs (either alone or in combination) is relatively modest (Amin Al Olama et al. 2015; Pashayan et al. 2015). Regardless, clinical assessment of these cancer-susceptibility genes and/or disease-associated SNPs may be useful for risk-stratifying clinically asymptomatic men in prostate cancer early detection programs.

DIAGNOSIS

Although the diagnosis of prostate cancer remains firmly rooted in histopathologic review of prostate core biopsy tissue, several tissue-based molecular biomarkers have recently emerged as ancillary tools for pathologists and/or clinicians. For the vast majority of patients, recognition of prostate cancer by surgical pathologists utilizing routine hematoxylin and eosin (H&E) slides of prostate biopsy tissue and established morphologic criteria is straightforward. Nonetheless, there are several specific instances in which ancillary tests such as immunohistochemistry (IHC) or in situ hybridization (ISH) may be helpful. Occasionally, foci of atypical prostatic glands suspicious for, but not diagnostic of small cell prostate cancer are identified in prostate core biopsies; these foci are typically termed atypical small acinar proliferation (ASAP) (Iczkowski et al. 1997). IHC using PIN-4 cocktail (composed of basal cell markers [p63 and high-molecular-weight cytokeratins] and the cancer-associated protein biomarker AMACR and the prostate cancer–specific protein biomarker ERG (corresponding to the expressed translated product of the TMPRSS2:ERG gene fusion) are useful in the clinical evaluation of ASAP foci and may help to make a definitive diagnosis of prostate cancer in a subset of cases (Fig. 2) (Jiang et al. 2005; Park et al. 2010; He et al. 2011; Tomlins et al. 2012). An ASAP focus that lacks basal cell marker staining and expresses AMACR and ERG is essentially diagnostic of prostate cancer.

Figure 2.

Molecular biomarkers for diagnosis of prostate cancer. Tissue-based molecular biomarkers for prostate cancer diagnosis include PTEN immunohistochemistry (IHC) for intraductal carcinoma of the prostate (IDC-P), PIN-4 cocktail and ERG IHC for atypical small acinar proliferation (ASAP), RB1 and cyclin D1 IHC for neuroendocrine prostate cancer (NEPC), and a polymerase chain reaction (PCR)-based methylation assay (ConfirmMDx) for negative core biopsies.

Other scenarios in which ancillary tests may be helpful in prostate cancer diagnosis include evaluation of intraductal carcinoma of the prostate (IDC-P) and high-grade neuroendocrine prostate carcinoma ([NEPC], i.e., “small cell carcinoma of the prostate”). Recent molecular data have shed new insight on IDC-P, an aggressive form of prostate cancer that spreads in situ within the existing ducts and acini of the prostate gland (Magers et al. 2015; Chua et al. 2017). Unlike its histopathologic mimic high-grade intraepithelial neoplasia—a benign, nonobligate precursor to invasive prostate cancer—IDC-P shows frequent PTEN gene alterations, which can be detected as loss of PTEN protein by IHC (Fig. 2) (Lotan et al. 2013; Morais et al. 2015; Hickman et al. 2017). Similarly, although the diagnosis of NEPC can typically be made on histomorphologic grounds (with or without corresponding IHC for neuroendocrine markers) (Epstein et al. 2014), recent studies have identified recurrent RB1 gene alterations in NEPC (Tan et al. 2014; Beltran et al. 2016), and, thus, IHC to detect dysregulated retinoblastoma pathway proteins (including RB1 and cyclin D1) may be helpful diagnostically in cases without neuroendocrine marker staining (Fig. 2) (Tan et al. 2014; Tsai et al. 2015). In addition, NEPC may be difficult to distinguish histopathologically from small cell carcinoma of the urinary tract (i.e., urinary bladder or urethra) or metastatic small cell carcinoma from other organs (i.e., lung). In these cases, fluorescent in situ hybridization (FISH) to detect ERG gene rearrangements may be helpful to confirm prostatic origin of the tumor, as ERG gene rearrangements are highly specific for prostate cancer (although, as previously discussed, the incidence of TMPRSS2:ERG gene fusions in prostate cancer limits the sensitivity of this assay, and a negative result does not exclude a primary prostatic tumor) (Han et al. 2009; Lotan et al. 2011a; Williamson et al. 2011; Schelling et al. 2013). Likewise, recurrent mutations in the promoter of TERT are much more frequent in urothelial carcinomas (or bladder small cell carcinoma) compared with prostatic adenocarcinoma (or prostatic NEPC) and, hence, may also have diagnostic utility for determining site of origin (Huang et al. 2013; Killela et al. 2013; Zheng et al. 2014).

Finally, a tissue-based molecular biomarker assay has been recently developed to help prostate cancer risk assessment in men with a prior negative prostate core biopsy who are being considered for a repeat biopsy. The ConfirmMDx assay uses methylation-specific polymerase chain reaction (PCR) to detect methylation at multiple gene loci, including APC and GSTP1, which are more frequently methylated in prostate cancer than benign prostatic tissue (Fig. 2) (Enokida et al. 2005; Trock et al. 2012; Partin et al. 2014). This assay has a very high negative predictive value for prostate cancer, as the absence of hypermethylated gene loci in tissue from the previous negative biopsy is strongly correlated with a lack of prostate cancer on repeat biopsy.

PROGNOSIS

Current prostate cancer prognostication is based on readily available clinical information (i.e., age, serum total PSA, etc.) and histopathologic review of either prostate core biopsy or radical prostatectomy tissue (i.e., GS, clinical or pathologic stage, etc.). At the time of prostate core biopsy, the outcome of interest for clinicians is typically the likelihood that a patient harbors clinically significant disease (and, therefore, requires definitive treatment) but may also include an assessment of overall cancer-specific mortality; on the other hand, at the time of radical prostatectomy, the outcomes of interest are usually the risk of early biochemical recurrence (which may necessitate adjuvant therapy) and overall cancer-specific mortality. For many patients, current clinicopathologic variables provide adequate risk stratification; however, a subset of patients with presumed low-risk disease will be undertreated, whereas a group of patients with presumed high-risk disease with be overtreated. With this in mind, a multitude of tissue-based molecular biomarkers have been proposed to help refine these clinical prognostications; despite these efforts, however, only a few single-gene and multigene assays have emerged as likely to be clinically useful.

PTEN alterations, including focal deletion, somatic mutation, and promoter hypermethylation, are among the most frequent genomic alterations in prostate cancer (Phin et al. 2013; Wise et al. 2017), and loss of PTEN function through combinatorial inactivation of the PTEN gene locus can be detected via FISH or IHC (Fig. 3) (Verhagen et al. 2006; Yoshimoto et al. 2006; Lotan et al. 2011b, 2016; Sathyanarayana et al. 2015; Picanco-Albuquerque et al. 2016). Loss of PTEN function in prostate cancer at the time of prostate core biopsy is associated with GS upgrading and locally advanced disease at the time of radical prostatectomy as well as decreased recurrence-free survival after definitive treatment (Lotan et al. 2015; Picanco-Albuquerque et al. 2016; Trock et al. 2016; Guedes et al. 2017). Similarly, loss of PTEN function in prostate cancer at the time of radical prostatectomy is associated with high-risk disease and poor clinical outcome (i.e., early biochemical recurrence, metastasis, prostate cancer–specific mortality, etc.) (Lotan et al. 2011b; Krohn et al. 2012; Ahearn et al. 2016). Finally, loss of PTEN function is associated with prostate cancer-specific mortality in low-risk patients managed conservatively with transurethral resection of the prostate (TURP) alone (Cuzick et al. 2013).

Figure 3.

Molecular biomarkers for prognostication of prostate cancer. Tissue-based molecular biomarkers for prostate cancer prognostication include single and multiple biomarker assays. Single biomarker assays with clinical utility include SChLAP1 RNA microarray or in situ hybridization (ISH) and PTEN DNA fluorescent in situ hybridization (FISH) or immunohistochemistry (IHC). Multiple biomarker assays with clinical utility include multiplex reverse transcription polymerase chain reaction (RT-PCR) assays (Oncotype Dx Prostate and Prolaris), RNA microarray (GenomeDx Decipher), and quantitative multiplex proteomics imaging ([QMPI] as reported by Blume-Jensen et al. 2015).

lncRNAs are a unique class of RNA molecules with increasingly recognized roles in tumor biology (Prensner and Chinnaiyan 2011; Iyer et al. 2015). Recent transcriptomic profiling studies have identified a number of prostate cancer-specific lncRNAs, including SChLAP1, which shows outlier expression in a subset of prostate cancers and is strongly associated with aggressive, lethal disease (Fig. 3) (Prensner et al. 2011, 2013, 2014b). Although several lncRNAs are likely to have clinical value for prostate cancer prognostication, SChLAP1 is currently the best characterized, and its prognostic utility has been validated across a variety of platforms (i.e., RNA microarray, RNA ISH, etc.) and independent patient cohorts (Prensner et al. 2013, 2014b; Mehra et al. 2014, 2016; Bottcher et al. 2015). Indeed, high SChLAP1 expression is associated with high-grade and locally advanced disease at the time of radical prostatectomy, early biochemical recurrence and metastasis after radical prostatectomy, and prostate cancer–specific mortality, and, importantly, it identifies a subset of patients with presumed low-risk disease that have a very poor clinical outcome (Prensner et al. 2013, 2014b; Mehra et al. 2014, 2016; Bottcher et al. 2015).

Despite the relative dearth of single gene prognostic assays, a number of recent multigene assays have been developed for prostate cancer prognostication, including Oncotype DX Prostate, Prolaris, and GenomeDx Decipher (Fig. 3). Oncotype DX Prostate is a multiplex reverse transcription polymerase chain reaction (RT-PCR)-based assay for use with prostate core biopsy material that measures expression of 12 prostate cancer–related genes and five reference genes to generate the genomic prostate score (GPS) (Knezevic et al. 2013). In otherwise low-risk patients, a high GPS is associated with high-risk disease at the time of radical prostatectomy and an increased risk of biochemical recurrence after definitive treatment (Cullen et al. 2014; Klein et al. 2014; Brand et al. 2016). Similar to Oncotype DX Prostate, Prolaris is a multiplex RT-PCR-based gene-expression assay; however, the Polaris test integrates expression of 31 genes involved in cell-cycle progression (CCP) to generate a CCP score and is used for core biopsy, radical prostatectomy, or TURP material (Cuzick et al. 2011). In multiple, large independent cohorts, a high CCP score is associated with increased risk of biochemical recurrence after radical prostatectomy and prostate cancer–specific mortality in low-risk patients managed conservatively with TURP alone (Cuzick et al. 2011, 2012; Cooperberg et al. 2013; Bishoff et al. 2014; Sommariva et al. 2014; Tosoian et al. 2017c). In contrast to Oncotype DX Prostate and Prolaris, GenomeDx Decipher is an RNA microarray-based test for use with core biopsy or radical prostatectomy material that assays expression of 22 RNA molecules to calculate a genomic classifier (GC) score (Erho et al. 2013). In intermediate- and high-risk prostate cancer patients, a high GC score is associated with metastatic progression and cancer-specific mortality after radical prostatectomy and may identify patients who would benefit from early adjuvant therapy (Erho et al. 2013; Karnes et al. 2013, 2017; Cooperberg et al. 2014; Den et al. 2015; Klein et al. 2015; Ross et al. 2016; Dalela et al. 2017; Spratt et al. 2017). Furthermore, a high GC score at the time of prostate biopsy is associated with increased metastatic progression and cancer-specific mortality after primary therapy (Klein et al. 2016; Nguyen et al. 2017) and may predict metastasis after salvage radiation for recurrent prostate cancer after radical prostatectomy (Freedland et al. 2016).

Finally, an eight-biomarker multiplex immunofluorescence (IF)-based assay has been recently validated for prostate cancer prognostication, using prostate core biopsy material. This assay uses quantitative multiplex proteomics imaging (QMPI) to make automated, quantitative measurements of each of the eight biomarkers and generate a biomarker risk score (Fig. 3) (Blume-Jensen et al. 2015). A high biomarker risk score is associated with both clinically significant disease (GS ≥ 7 vs. GS6) and high-risk disease at radical prostatectomy.

TARGETED THERAPEUTICS

The current treatment paradigms for advanced/metastatic prostate cancer predominantly involve targeting the androgen signaling pathway but may include other types of conventional chemotherapy for patients with aggressive disease (i.e., NEPC). The theory underlying hormonal therapy for advanced prostate cancer is that, in the majority of patients, activation of the androgen signaling pathway drives growth of cancer cells, and thus these tumors are “addicted” to androgen receptor signaling for their survival (Antonarakis et al. 2016). Prostate cancer treated with hormonal therapy, however, may develop resistance via several mechanisms, including amplification, mutation, and/or differential splicing of the androgen receptor gene, leading to the development of castration-resistant prostate cancer (CRPC) (Watson et al. 2015; Antonarakis et al. 2016). As such, molecular analysis of the androgen receptor gene may have clinical significance for monitoring the development of resistance to hormonal therapy (Fig. 4). In particular, the detection of the V7 androgen receptor splice variant (either in tissue, circulating tumor cells, or cell-free nucleic acids) is associated with decreased response to specific types of hormonal therapy and poor progression-free and overall survival (Antonarakis et al. 2014, 2015; Onstenk et al. 2015; Scher et al. 2016, 2017; Del Re et al. 2017). In addition, secondary to sustained targeting of the androgen signaling pathway with hormonal therapy, a subset of patients with advanced prostate cancer will develop androgen-insensitive prostate cancer (AIPC) via acquired alterations in TP53, RB1, and/or PTEN (Watson et al. 2015). For these patients, cancer cells are no longer dependent on or sensitive to inhibition of the androgen receptor signaling pathway, and many of these tumors show overlapping clinical and pathologic features with NEPC.

Figure 4.

Molecular biomarkers for selection of targeted therapeutics for prostate cancer. Molecular biomarkers for prostate cancer–targeted therapeutics include AR-V7 transcripts (androgen receptor signaling); somatic mutations, PTEN alterations, and gene fusions (recurrent molecular alterations); DNA repair mutations (PARP1 inhibition and immunotherapy); and PD-L1 immunohistochemistry (IHC) (immunotherapy).

Aside from hormonal therapy, there is currently very limited targeted therapeutic selection for the treatment of advanced prostate cancer, although several recurrent molecular alterations present attractive targets for future therapeutic options (Fig. 4). PTEN alterations are among the most common recurrent genomic alterations in prostate cancer, and loss of PTEN activity leads to dysregulation of the PI3K signaling pathway; thus, emerging small-molecule PI3K inhibitors may have therapeutic relevance in advanced tumors with PTEN alterations (Phin et al. 2013; Wise et al. 2017). Although genomic and transcriptomic profiling of many solid tumors (including lung, colon, and breast) has revealed high-frequency targetable molecular alterations in other common oncogenes (i.e., EGFR, ALK, ERRB2, etc.), only a small subset of prostate cancers (particularly those with ETS family gene rearrangements) show recurrent targetable alterations; for example, less than 1% of tumors harbor RAS/RAF family gene rearrangements, FGFR2 gene fusions, or IDH1 mutations (Palanisamy et al. 2010; Prensner et al. 2011; Barbieri et al. 2012; Beltran et al. 2013; Wu et al. 2013; Cancer Genome Atlas Research Network 2015). Despite being among the most frequent molecular alterations in prostate cancer, as of yet there are no specific therapeutics targeted against ETS family gene rearrangements, although small molecular inhibitors targeting PARP1, a mediator of ETS gene fusion-dependent transcription, have been investigated for use in advanced prostate cancers with ETS family gene rearrangements (Fig. 4) (Brenner et al. 2011). PARP1 also plays a role in the DNA damage-response pathway, and its inhibition triggers DNA damage-mediated cell death (Do and Chen 2013). This unique protein function suggests that small molecule inhibitors of PARP1 may be useful in the subset of advanced prostate cancers with perturbations in DNA repair pathway genes (Mateo et al. 2017), and, indeed, recent data from a small phase II clinical trial showed high response rates to the PARP1 inhibitor olaparib in patients with metastatic tumors harboring mutations in DNA repair genes (Fig. 4) (Mateo et al. 2015). Interestingly, overexpression of the novel prostate cancer–specific lncRNA PCAT-1 represses BRCA2 expression, which results in a functional BRCA-deficient phenotype (“BRCAness”) and suggests that PCAT-1 express may also be a predictive biomarker for PARP1 inhibition in advanced prostate cancer (Fig. 4) (Prensner et al. 2014a). Recent molecular dissection of advanced prostate cancer has also uncovered a role for the bromodomain and extraterminal (BET) proteins in epigenetic regulation of the androgen receptor signaling pathway, and novel small-molecule BET bromodomain inhibitors may have therapeutic value in advanced tumors with intact androgen receptor pathway activity (i.e., CRPC) (Wyce et al. 2013; Asangani et al. 2014, 2016).

Finally, no review of potential biomarkers for targeted therapeutics for advanced prostate cancer would be complete without briefly discussing the emerging field of immunotherapeutics, which includes therapies that target the CTLA-4 and PD-1/PD-L1 immune checkpoint pathways (Fig. 4) (Littman 2015). Inhibition of these pathways facilitates host immune-mediated tumor cell killing, and although clinical utility of immunotherapeutics for advanced prostate cancer is still under investigation, the success of these targeted therapeutics approaches in other solid tumors is encouraging (Fig. 4) (Hodi et al. 2010; Brahmer et al. 2012; Topalian et al. 2012). As in other tumor types, there is tremendous clinical interest in the development and validation of predictive biomarkers for immunotherapeutic response. For example, the ability of tumor cell PD-L1 expression to predict response to PD-1/PD-L1-based targeted immunotherapeutics has been studied in other tumor types with mixed results (Gibney et al. 2016). Although tumor cell PD-L1 expression has not been explicitly examined as a predictive biomarker in advanced prostate cancer, increased tumor cell PD-L1 expression is associated with a higher risk of biochemical recurrence after radical prostatectomy (Gevensleben et al. 2016). In other tumors, additional possible predictive biomarkers for immunotherapeutic response in other tumors include mutational burden, neoantigen burden, and DNA repair mutations (Fig. 4) (Rizvi et al. 2015). As several clinical trials are currently evaluating the efficacy of a variety of immunotherapies in advanced prostate cancer, there will surely be a focused effort to develop molecular biomarkers for treatment response. Nevertheless, given the failure of the CTLA-4 inhibitor ipilumimab to improve overall survival in two phase III trials (Kwon et al. 2014; Beer et al. 2017), clinical translation of immunotherapy in prostate cancer will likely be more challenging and may impact fewer patients than other genitourinary malignancies (i.e., urothelial carcinoma and renal cell carcinoma) (Motzer et al. 2015; Bellmunt et al. 2017).

Importantly, however, genomic studies clearly show that approximately 1%–2% of CRPC harbor predominantly somatic mutations in DNA mismatch repair genes (most commonly MSH2 and MSH6), leading to hypermutation and microsatellite instability/mismatch repair deficiency (Kumar et al. 2011; Grasso et al. 2012; Pritchard et al. 2014; Robinson et al. 2015; Abida et al. 2017). The recent FDA approval for pembroluzumab, a PD-1 inhibitor, in any mismatch repair deficient cancer (regardless of site of origin) provides an immediate standard-of-care option for these patients.

CONCLUDING REMARKS

The past two decades have witnessed a revolution in molecular biomarkers available for the early detection, diagnosis, and treatment of prostate cancer. Given renewed focus on spending and efficacy in all aspects of modern health-care systems, however, in the coming years there will surely be increasing pressure to justify the clinical benefit of expensive molecular assays, particularly those that evaluate only a single or small group of biomarkers. With this in mind, recent technological advances have spurred the development of comprehensive molecular biomarker assays that assess hundreds or thousands of biomarkers simultaneously. For example, next-generation sequencing technology has facilitated the development of multiplex targeted DNA and RNA sequencing assays that can concurrently assess somatic variants, copy number alterations, and gene expression from minute routine clinical tissue samples, such as prostate core biopsies and minute metastatic tissue foci (Fig. 5) (Grasso et al. 2015; Hovelson et al. 2015; Abida et al. 2017). Not only will such assays provide a focused yet comprehensive overview of the molecular alterations within a given patient’s tumor, but simultaneous gene-expression data will facilitate enhanced prognostication and risk stratification. In addition, prospective application of such assays within the context of ongoing clinical trials will provide unique opportunities to evaluate the combinatorial effort of a large number of biomarkers to define the precise set of optimized molecular biomarkers for a given clinical outcome. Overall, in this unfolding era of personalized medicine, advances in molecular biomarkers will continue to influence oncologic care and clinical management of prostate cancer. To drive changes in clinical practice, however, continued rigorous studies are needed to show that such approaches influence treatment decisions and improve outcome in a cost-effective manner.

Figure 5.

Comprehensive assessment of prostate cancer molecular biomarkers via integrative next-generation DNA and RNA sequencing. Prioritized DNA and RNA alterations identified by multiplexed next-generation sequencing using the Oncomine Comprehensive Panel (OCP) assay from a cohort of formalin fixed paraffin embedded prostate cancer tissue specimens are shown. Samples are grouped according to treatment status (hormone-naïve [HN], prior androgen deprivation therapy [ADT], radiation therapy [RT], ADT plus RT, and/or chemotherapy [ADT⁺]; AR- or NEPC, no [or reduced] AR signaling as indicated by no/focal prostate-specific antigen [PSA] staining or prostatic neuroendocrine/small cell carcinoma [SCC]). Clinicopathological information (including Gleason score) and RNA sequencing results (detected fusions are indicated according to the legend) are shown in the header. (Figure adapted from Hovelson et al. 2015 under a Creative Commons license.)

COMPETING INTEREST STATEMENT

The University of Michigan has been issued a patent on ETS gene fusions in prostate cancer on which S.A.T. is a coinventor. The diagnostic field of use has been licensed to Hologic/Gen-Probe, which has sublicensed rights to Roche/Ventana Medical Systems. S.A.T has received travel support from, and had a sponsored research agreement with, Compendia Bioscience/Life Technologies/ThermoFisher Scientific. S.A.T. has sponsored research agreements with Astellas and GenomeDX. S.A.T. has served as a consultant for and received honoraria from Roche/Ventana Medical Systems, Almac Diagnostics, Janssen, AbbVie, Sanofi, and Astellas/Medivation. S.A.T. is a cofounder of, consultant for, and Laboratory Director of Strata Oncology. A.M.U. has no disclosures.