Biomarkers and personalized risk stratification for patients with clinically localized prostate cancer

Abstract

Prostate cancer (PCa) is the most common neoplasia among men in developed countries and a leading cause of cancer-related morbidity and mortality. PCa is a very heterogeneous disease, both clinically and biologically. Currently, it is difficult to stratify patients into risk groups that entail different disease management. Therefore, a personalized view of this disease is mandatory, through the development of new and more accurate biomarkers that may help clinicians to stratify patients according to threat that PCa poses for each patient. Hence, this review focuses on recent developments of molecular and immunohistochemical biomarkers for PCa risk stratification that might enable a personalized approach to PCa patients. However, despite the increasing amount of available data, there is also an urgent need to translate the most promising biomarkers for clinical use through large multicenter validation trials. Ultimately, these will contribute for an improved clinical management of PCa patients.

Keywords::

• biomarkers
• epigenetics
• genetic alterations
• personalized management
• prostate cancer
• risk stratification

Prostate cancer (PCa) is a major burden on human health worldwide. In 2014, this neoplasia is predicted to be the most incident cancer in men and the second leading cause of cancer-related death [1]. PCa is well known for its heterogeneity, ranging from clinically indolent to extremely aggressive tumors [2,3]. Such feature poses a challenge for clinicians concerning patient’s risk stratification [4]. As PCa is mostly asymptomatic until extra-prostatic invasion occurs, there was an urge to develop and implement accurate screening methods on a population basis [5]. Thus, serum prostate-specific antigen (PSA) testing and digital rectal examination have been commonly used in clinical practice to screen and assist clinicians on therapeutic decision [5]. Consequently, international guidelines, regarding the application of these tests and precise actions to be taken depending on results, have been developed [6,7]. Since the serum PSA test implementation in the 1980s, PCa incidence rates in the USA and elsewhere skyrocketed followed by a significant decrease [8]. PSA is prostate-specific but not PCa-specific, thus leading to reduced accuracy in diagnosis of PCa, but also on the identification of clinically significant tumors. Ultimately, this lack of specificity has led not only to overdiagnosis and overtreatment of indolent tumors, on the one hand, but also to underdiagnosis and undertreatment of life-threatening PCa, on the other [9]. Thus, an attentive look at PCa incidence and mortality rates worldwide clearly indicates that there is an urgent need for a more patient-directed approach for risk stratification. More important than to identify individuals with PCa is to determine how life threatening is that condition.

The current lack of optimal tools to accurately stratify PCa patients into risk groups might be overcome through the development of novel biomarkers. Biomarkers are, typically, molecules whose detection or quantitation provides additional insight on disease beyond clinical features observed by physicians [10]. Proteins, metabolites, RNA transcripts, DNA and epigenetic alterations are all candidates to become disease biomarkers [10]. Imaging techniques are under development but mostly for cancer localization and guidance of diagnostic procedures and are also limited due to their cost and availability [10]. Moreover, biomarkers might fall into several categories according to their potential to assist in disease predisposition assessment, screening, diagnosis, prognostication, prediction of response to treatment, disease monitoring and pharmacogenomics [10]. There are several established criteria that define an ideal biomarker: be harmless and preferably non-invasive, highly accurate (i.e., high specificity, sensitivity, positive and negative predictive values) for its intent and improve decision-making abilities together with clinicopathological data [10]. Hence, it is extremely unlikely that a single molecule or alteration can meet all these features, and a panel of biomarkers might thus provide a superior and more reliable approach. In PCa, three major areas are left to be filled with reliable biomarkers: distinguish low PSA tumors from benign prostatic hyperplasia; discriminate indolent from aggressive disease and identify metastatic disease [10–12].

Besides well-established technologies such as immunohistochemistry (IHC) and FISH analysis, recent developments have occurred on the characterization of epigenetic alterations in PCa as promising biomarkers. Due to a better comprehension and description of these alterations, data concerning their usefulness as biomarkers have been accumulating. Bearing in mind their stability, frequency, reversibility and accessibility in body fluids, epigenetic biomarkers might be considered at the top of the list for candidates to next-generation biomarkers that may answer most, if not all, the questions raised above [13].

This review aims to compile and critically analyze the most relevant data on emerging biomarkers for PCa risk stratification, with a special emphasis on those related with epigenetic alterations. For that purpose, PubMed was searched for publications concerning PCa biomarkers, between 1998 and 2014. Reports were selected based on the detail of analysis and/or potential clinical applications and/or clinical translational studies (Table 1). Biomarkers hereafter reported were selected based on substantial clinical translational studies or attempts to bring them to daily clinical routine.

Table 1. Best performing biomarkers for diagnosis and prognosis assessment of prostate cancer.

Biomarker	Significance in prostate cancer	Ref.
Immunohistochemical
Ki-67	Overexpression has prognostic value for: PCa-related death in watchful waiting patients; biochemical relapse upon radical prostatectomy; disease progression in patients submitted to radiation therapy	[15–20]
AMACR	Diagnostic marker in tissue biopsy due to its overexpression in PCa	[21,22]
Genetic
TMPRSS2:ERG	Present in nearly 50% of PCa and in up to 25% of HGPIN lesions but no prognosis value yet confirmed	[25–27,30,31]
8q amplification	Detected in 12.8–54% of PCa and able to discriminate patients with poor outcome as a marker of aggressiveness	[36,37]
Non-coding RNAs
PCA3	Overexpressed in PCa and is useful to predict outcome after a first biopsy and correlates with disease aggressiveness	[11,42,46,47]
miR-141	Elevated blood levels are associated with castration-resistant metastatic PCa and high Gleason score	[50]
miR-375
miR-21	Expression in serum from PCa patients showed an association to docetaxel resistance	[57]
Histone modifications and modifiers
Global levels of H3K27Me3	Overrepresented in metastatic tumors compared with non-malignant prostate tissue, using immunohistochemistry analysis. In plasma samples of PCa patients, by ELISA assay, a decrease in those levels was found in metastatic disease compared with localized disease	[63,64]
Global levels of H3K18Ac	Associated with disease recurrence in low-grade PCa and independent predictors of PCa progression, irrespective of tumor grade	[61,62]
Global levels of H3K4Me2
Enhancer of zeste homolog 2	Higher expression in higher Gleason score PCa, correlates with TNM stage and disease progression, associated with shorter time to PCa biochemical relapse	[12,68,70]
SMYD3	Upregulated and associated with shorter time to relapse in clinically localized disease upon radical prostatectomy	[70]
Methylation levels
GSTP1	Increased methylation levels correlate with higher Gleason score and tumor stage and might be detected in body fluids, including urine and serum. Increased risk of biochemical relapse in one study	[80,85,92]
APC	Higher promoter methylation levels were independently associated with worse disease-free and disease specific survival, with superior prognostic performance compared with serum PSA and Gleason score	[95]
ProCaM (GSTP1/APC/RARB2)	A positive result in the ProCaM assay associated with increased risk of finding high Gleason score (≥7) PCa in prostate biopsy, also is an independent predictor of PCa risk to guide decision for repeat biopsy	[89,90]
PTGS2	Increased risk for biochemical recurrence was determined for PCa patients with promoter hypermethylation	[93]
CD44

HGPIN: High-grade prostate intraepithelial neoplasia; PCa: Prostate cancer; ProCaM: Prostate cancer methylation; PSA: Prostate-specific antigen.

Immunohistochemical biomarkers

The ability of IHC to simultaneously evaluate protein expression and correlate with morphology was decisive to make it the most widely used ancillary tool in pathology, with a substantial impact in clinical practice. Therefore, development of new biomarkers based on IHC is a constantly evolving field of research, including for PCa [14]. Notwithstanding the vast array of candidate IHC biomarkers, published to date, for PCa risk assessment, only Ki-67 has been extensively validated, displaying the potential to stratify patients according to disease aggressiveness [14]. Ki-67 has been described as possessing prognostic value for PCa-related mortality in patients ascribed to watchful waiting cohorts [15,16], biochemical relapse in PCa patients submitted to radical prostatectomy [17,18] and for disease progression in patients undergoing radiation therapy [19,20]. As Ki-67 is a marker of cell proliferation, it is not surprising that this biomarker carries independent prognostic value in most studies, adding important information to the Gleason score and serum PSA levels [14]. However, differences in study design and cohort composition led to different results regarding Ki-67 usefulness as a prognostic biomarker for PCa. Unsolved issues relate with the method for assessing Ki-67 immunostain, including the minimum number of cells to be counted, how to count them and which cutoff value to be used for patient’s stratification. Therefore, additional studies, preferentially enrolling patients from multiple centers, are mandatory to determine the true relevance of Ki-67 in clinical practice.

Although a prognostic value has not been established for α-methylacyl-coenzyme A racemase (AMACR) expression in PCa, this enzyme is currently used as a diagnostic ancillary tool. Indeed, AMACR is frequently overexpressed in PCa [21], displaying high sensitivity and specificity (>90%) when used as a diagnostic marker in tissue biopsy [22]. AMACR, in conjunction with basal cell markers (e.g., P63) is reported to help avoiding unnecessary biopsies [23]. Its value alone is, however, questionable because AMACR is significantly expressed also in PCa precursor lesions such as high-grade prostate intraepithelial neoplasia (HGPIN), prompting some authors to suggest AMACR as a marker for HGPIN and not PCa [24]. It has not been elucidated whether AMACR expression levels might correlate with disease outcome or might identify HGPIN lesions more prone to progress to invasive adenocarcinoma.

Genetic biomarkers

Fusion genes

Over the last years, fusion genes have unexpectedly emerged as a common feature of PCa. Among these, the most common is TMPRSS2:ERG, which is PCa-specific, accounting for approximately 90% of all currently known fusion events [10], and detected in nearly 50% of prostate adenocarcinomas [25]. These characteristics suggest that TMPRSS2:ERG is an early, and probably founding, event in many PCa, a postulate that is also supported by the observation of this fusion gene in up to 25% of HGPIN lesions [26,27].

In a cohort of PCa patients on watchful waiting, Demichelis et al. found that the presence of TMPRSS2:ERG fusion gene was associated with Gleason score >7 tumors, increased mortality rates due to PCa and progression to metastatic disease [28]. Using real-time PCR, a different study found a positive correlation between fusion transcripts in urine after prostate massage and a high serum PSA, pathological stage and Gleason score [29]. Despite its apparent prognostic utility, these results have been challenged by other researchers, which have not found any correlation between fusion genes and PCa aggressiveness [30,31]. This is most likely due to PCa heterogeneity and differences in patients’ cohorts [11]. Nevertheless, this lack of consistency did not discourage researchers from evaluating its detection in urine samples. A study found that TMPRSS2:ERG displayed 93% specificity and 94% positive predictive value in post-digital rectal examination urine samples, for prediction of PCa upon prostate biopsy [32]. However, this fusion gene was present in only 50% of the cases, thus compromising sensitivity. Therefore, to augment the sensitivity of the assay, assessment of prostate cancer antigen 3 (PCA3) mRNA was added to the test (normalized to urine PSA mRNA), resulting in enhanced utility over PSA serum alone, both for PCa diagnosis and prediction of disease aggressiveness [33]. Nonetheless, this test is useless when PSA mRNA is low or untraceable in urine [11].

As PCa often display multiple tumor foci within the same gland, the distribution of neoplastic cells harboring the TMPRSS2:ERG fusion gene might be heterogeneous [34]. This might limit its usefulness as a single biomarker and a combined approach (e.g., with PCA3) seems to be more efficient. Notwithstanding these constraints, published data on TMPRSS2:ERG suggest a possible role as PCa biomarker both for diagnosis and risk stratification.

Chromosome 8q amplification

A few studies have provided whole-genome information on PCa from material of prostate biopsy and that has been shown to be largely consistent with data derived from radical prostatectomy specimens [35]. Among chromosome-level alterations, relative 8q gain has been proposed as a biomarker of aggressive disease, being detected in 12.8–54% of patients [36]. A comparative genomic hybridization study, using a large series of prospectively collected prostate biopsies, found that patients with relative gains of 8q had worse prognosis [37]. Moreover, when patients were stratified according to tumor grade or clinical stage, relative gains of 8q were able to discriminate those with poorer outcome [37]. These results are in accordance with previous published studies that used FISH analysis to determine 8q status. In those studies, 8q gain was associated with disease-related mortality after an average 35 months of follow-up [38] and also with lower disease-specific survival at 10 years of follow-up [36]. The fact that 8q status might help to predict outcome in prostate biopsy material, that is, before any therapeutic option has been decided, is of key importance for its potential use in patients’ management and risk stratification.

Whole-transcriptome markers

The recent development of genome-wide characterization techniques prompted researchers to apply this knowledge to obtain specific transcript profiles of PCa patients that might correlate with disease aggressiveness and patient outcome. Several studies have been conducted to achieve this goal and all have identified new candidate biomarkers that could better stratify patients [39–41]. There is, however, a lack of consistency in the data generated by those studies, which probably derives from differences in patient’s characteristics, sampling procedures, methodologies and data analysis [14]. Therefore, more robust results and independent validation studies in larger cohorts are needed to ascertain whether a specific prognostic signature for PCa is likely to become a clinically useful tool.

Epigenetic biomarkers

Non-coding RNAs

PCA3

PCA3 has become the most prominent biomarker for PCa since the emergence of serum PSA in the 1990s [10]. PCA3 is a long non-coding RNA that has been found to be highly expressed in 95% of prostate adenocarcinomas but only slightly expressed in normal prostate tissues, which has been estimated as a 66-fold upregulation in PCa compared with normal adjacent tissue [11,42]. Importantly, PCA3 expression is independent of tumor size [43]. These characteristics created great expectations about its usefulness as a biomarker, not only in tissue but also in urine samples, providing a non-invasive test for PCa [44]. The most prominent role for PCA3 has been the evaluation of the chance for a man to have a positive repeat prostate biopsy [45]. Consequently, the US FDA recently approved urine PCA3 expression levels as a diagnostic test for men suspected to have PCa but with a negative prostate biopsy, which still might be at risk and require a second intervention, that is, re-biopsy [11].

Besides this diagnostic utility, PCA3 has been also proposed as a prognostic tool since it might be able to forecast outcome after a first biopsy and to be correlated with PCa aggressiveness [46,47]. Moreover, a correlation between PCA3 score, radical prostatectomy outcome and positive margins risk on prostatectomy specimens has been reported, as well [48]. However, these data are still controversial because several studies neither found any correlation between PCA3 levels and clinicopathological data nor any prognostic value for this non-coding RNA [25]. Nevertheless, PCA3 might be a good candidate biomarker both for distinguishing low PSA tumors from benign prostate hyperplasia and, eventually, to stratify patients according to their risk of progressive disease before and after radical prostatectomy, in a non-invasive manner.

miRNAs

miRNAs are small non-coding RNAs, of approximately 22 nucleotides in length, involved in gene expression regulation. They regulate several key cellular pathways and have been found to be deregulated in numerous diseases, including cancer [49]. Development of more accurate technologies to evaluate miRNAs’ presence and expression levels led to enhanced interest in translating current knowledge about miRNAs into clinically useful cancer biomarkers. Recent research has focused on the identification of specific miRNAs signatures in PCa and on how they might correlate with clinicopathological variables, such as Gleason score, stage and metastatic spread. Assessment of miRNAs is also feasible in body fluids, such as urine and plasma/serum, which makes them potentially useful for non-invasive evaluation and stratification of PCa patients [50]. Additionally, these molecules also show remarkable levels of stability in different clinical samples, as their small size and distinctive biochemical structure grant them resistance to RNase-mediated degradation [51].

In a study using serum and plasma samples from PCa patients, as well as normal prostate tissues, miR-141 expression was found to be increased in metastatic PCa compared with localized disease. MiR-375 expression levels were also associated with higher Gleason score and positive lymph node status [52]. Moreover, Schaefer et al. found that higher miR-96 expression was related to earlier biochemical relapse in PCa, using a training set/testing set methodology [53]. In addition, miR-100, miR-145 and miR-191 were reported to bear prognostic value in PCa patients following radical prostatectomy [54] and miR-221 downregulation has been associated with cancer progression in a cohort of 92 PCa patients [55].

Furthermore, elevated blood levels of miR-141, miR-200b and miR-375 have been associated with castration-resistant metastatic PCa and high Gleason score, discriminating these high-risk patients from those with low-risk, clinically localized disease [50]. Likewise, expression of miR-21 and miR-221 has been also associated with biochemical relapse [56]. Expression of miR-21 in serum collected from PCa patients showed, also, an association with docetaxel resistance [57]. Concerning miRNAs as biomarkers for PCa in urine, miR-107 and miR-574-3p were found to be upregulated in men with PCa compared with normal controls. Strikingly, a panel comprising both these miRNAs was even more accurate to identify PCa than PCA3 normalized to PSA [50]. Despite being accurate biomarkers in serum and plasma samples, miR-141 and miR-375 were not able to distinguish between PCa patients and healthy donors in urine samples.

Taken together, these results suggest that miRNAs detection and quantitation in blood and urine might be useful for diagnosis and risk stratification of PCa patients. Nevertheless, it should be recalled that most of those studies were conducted in small cohorts, thus requiring validation in larger datasets, with standardization of protocols, biological material selection and patients’ characteristics [58].

Histone modifications & chromatin remodeling

Histone marks

Histone modifications are very dynamic and, consequently, more difficult to scrutinize than the remaining epigenetic alterations. This class of alterations includes methylation, acetylation and phosphorylation of specific residues mostly on N-terminal tails of histones [59,60], thus altering chromatin conformation and gene transcription [59,60]. Currently, our knowledge about histone modifications and their role in gene expression regulation is still rather imperfect, impairing its use as potential cancer biomarkers.

Nevertheless, the first study suggesting histone modifications as potential cancer biomarkers was performed in PCa patients [61]. It was found that global levels of H4K12Ac, H3K18Ac, H3K4Me2 and H4R3Me2, assessed by IHC, correlated with PCa stage, and that H3K18Ac and H3K4Me2 levels were associated with disease recurrence in low-grade PCa [61]. Subsequently, global levels of these two marks were found to be independent predictors of PCa progression, irrespective of tumor grade [62].

Another histone mark recognized as altered in PCa is H3K27Me3, which is overrepresented in metastatic tumors compared with non-malignant prostate tissue, using IHC analysis [63]. This finding indicates a potential role for H3K27Me3 overexpression as a prognostic biomarker in PCa patients, which might be able to stratify patients according to their risk of developing PCa. Intriguingly, the only study that assessed H3K27Me3 levels in plasma samples of PCa patients, using ELISA, found a decrease in those levels in metastatic disease compared with localized disease [64]. These controversial results emphasize the difficulties of using histone modification marks as potential PCa biomarkers. Therefore, more accurate techniques to evaluate histone marks in body fluids, IHC protocol standardization and testing in large cohorts of PCa are mandatory to effectively determine the role of histone marks as PCa biomarkers.

Histone modifiers

Enhancer of zeste homolog 2

The enhancer of zeste homolog 2 (EZH2) gene encodes for a polycomb group protein that is directly involved in transcription regulation, usually promoting gene silencing [65,66]. EZH2 has been found overexpressed in several tumors, including PCa, in which it correlates with increased levels of H3K27Me3 [12]. Because EZH2 has been pointed out as a mediator of E-cadherin silencing in cancer cells, it might constitute a potential biomarker for disease progression [67]. Increased expression of both EZH2 mRNA and protein has been found in PCa tissues compared with benign prostate hyperplasia and HGPIN [68]. Moreover, significantly higher EZH2 expression was observed in higher Gleason score tumors, correlating also with TNM stage and disease progression [68]. In a different study, EZH2 performed better than serum PSA and Gleason score in the prediction of tumor progression [65]. Furthermore, immunohistochemical assessment of EZH2 protein in PCa tissues was able to accurately predict tumor recurrence after therapy [69]. Finally, higher EZH2 transcript levels have been associated with shorter time to PCa biochemical relapse, although it did not disclose independent prognostic value [70].

Thus, EZH2 stands as a potentially useful biomarker of aggressive PCa, amenable to be assessed in prostate biopsies using IHC but validation in large datasets has not been accomplished, yet. Moreover, the potential usefulness of EZH2 assessment in body fluids as PCa biomarker is still largely unexplored.

SMYD3

SMYD3 is a histone methyltransferase responsible for establishing several histone marks in lysine residues, such as H3K4, H4K5 and H4K20 [71,72]. This enzyme was found overexpressed in multiple cancers, including those of colorectum, breast, hepatocellular and prostate [70,73,74].

Recently, it has been reported that higher SMYD3 transcript levels in PCa tissue samples were associated with shorter time to relapse in a series of 150 patients with clinically localized disease submitted to radical prostatectomy [70]. Importantly, SMYD3 levels retained their statistical significance as independent prognostic biomarker in multivariate analysis, in addition to Gleason score and pathological stage [70]. These results require further validation to ascertain SMYD3 potential for risk stratification in PCa. Furthermore, the development of a test that might detect and quantify SMYD3 expression in body fluids might provide a novel non-invasive approach for PCa detection and assessment of disease aggressiveness.

DNA methylation

DNA methylation is the most frequently studied epigenetic alteration in cancer, including PCa. Therefore, the majority of candidate epigenetic cancer biomarkers are generally based on aberrant DNA methylation. Indeed, this is a very common, stable and accessible alteration that might be detected in tissues and body fluids of PCa patients. Global genome hypomethylation appears to be correlated with tumor recurrence in PCa patients [75]. However, because the assessment of global hypomethylation is rather cumbersome, most research efforts have focused in DNA, and especially gene promoter, hypermethylation, instead.

In PCa, several gene promoters have been reported to be hypermethylated, but the best established as a biomarker is the glutathione-S-transferase P1 (GSTP1) gene. Several reasons account for its prominent role as PCa biomarker: GSTP1 methylation is more PCa specific than serum PSA [76]; GSTP1 methylation levels can discriminate PCa from benign prostate lesions [77–79]; increased GSTP1 methylation levels correlate with higher Gleason score and tumor stage [80], and, importantly, it might be detected in body fluids, including urine and serum [81]. Remarkably, quantitative GSTP1 methylation assays display 93–100% specificity and 21.4–38.9% sensitivity for PCa detection in urine samples [77,82,83], which might be increased 75% by prostatic massage prior to urine collection [84].

Although the specificity of GSTP1 methylation is very high, even in urine and serum, sensitivity is suboptimal [85]. This might be overcome by the addition of other gene promoters also known to be frequently and specifically hypermethylated in PCa, thus constituting gene panels [85]. Indeed, the combination of GSTP1 with APC increased sensitivity to 98.3%, keeping 100% specificity in tissue samples [80]. Other gene panels were shown to accurately detect PCa in urine (GSTP1/ARF/CDKN2A/MGMT and GSTP1/APC/RARB2/RASSF1A) and serum (GSTP1/PTGS2/RPRM/TIG1), with 86% sensitivity and 89% specificity in urine and 42–47% sensitivity and 92% specificity in serum [86–88]. These results have led to the development of a diagnostic-level assay, the Prostate Cancer Methylation (ProCaM) assay, which consists of the quantitation of promoter methylation levels of GSTP1, APC and RARB2, in urine. The test has been validated in multicenter prospective study, in which urine samples of men with serum PSA levels of 2.0–10.0 ng/ml were collected and analyzed with the ProCaM assay and its performance was compared with existing methods based on clinical workup and serum PSA levels [89]. Remarkably, the ProCaM assay was able to more accurately detect PCa than serum PSA or any of its derived parameters [89]. Importantly, a positive result in the ProCaM assay associated with increased risk of finding high Gleason score (≥7) PCa in prostate biopsy [89]. Recently, a multicenter trial validated the performance GSTP1, APC and RASSF1 promoter methylation as an independent predictor of PCa risk to guide decision for biopsy repetition [90]. This assay demonstrated an 88% negative predictive value and may, thus, contribute to avoid unnecessary repeat prostate biopsies [90].

Concerning risk stratification, quantitative promoter methylation of GSTP1, APC and MDR1 was able to discriminate clinically localized from advanced disease with 72.1% sensitivity and 67.8% specificity [91]. On the other hand, patients with hypermethylated GSTP1 in serum had a 4.4-fold increased risk of biochemical relapse in one study [92], although this has not been confirmed in a different cohort [93]. However, in the latter study, a ninefold increased risk for biochemical recurrence was determined for PCa patients with CD44 and PTGS2 promoter hypermethylation [93]. Furthermore, a multicenter study of PCa patients submitted to radical prostatectomy found that PITX2 promoter methylation levels were of prognostic significance following treatment, both uni- and multivariate analysis [94]. Indeed, higher levels of PITX2 promoter methylation associated with a fourfold increased risk for biochemical relapse [94]. Finally, in a series of prospectively collected prostate biopsies, high APC promoter methylation levels were independently associated with worse disease-free and disease-specific survival, with superior prognostic performance compared with serum PSA and Gleason score [95].

Thus, multigene DNA methylation-based assays might not only provide useful tools for PCa detection, but also for identification of clinically significant disease, thus helping clinicians in the stratification of risk for developing PCa and in the assessment of PCa aggressiveness on an individual basis.

Expert commentary

The boom of PCa detection following the widespread use of serum PSA testing as a screening tool has come to a halt when large cooperative studies demonstrated no gain in survival or limited increase in overall survival, but at the cost of increased and unnecessary morbidity and mortality [96–98]. This has led to the recommendation of the US Preventive Services Task Force against the disseminated utilization of serum PSA for PCa screening [99]. Thus, the focus was changed from PCa detection by itself to the assessment of PCa aggressiveness once it was diagnosed, to avoid overtreatment.

Because currently available tools (mostly clinical and pathological) are rather imperfect to predict the threat that a given PCa poses to the patient’s life, the development and validation of new biomarkers is mandatory. Concerning detection, the new assays have to compare favorably with serum PSA, adding sensitivity and specificity, as well as maintaining a low cost and none or minimal invasiveness for sample collection. Thus, urine-based assays are quite appealing and, in this regard, the PCA3 assay may be considered an important step forward. However, its usefulness is restricted and its cost is a limiting factor for generalized use. In this setting, epigenetic biomarkers, especially those based on DNA methylation, seem to be most promising candidates for a new generation of urine-based PCa biomarkers. They were shown not only to accurately detect PCa in clinical samples, but they may also identify the clinically significant tumors.

Probably, prostate biopsy will remain the gold-standard for PCa diagnosis and prognostication. However, to more efficiently predict PCa aggressiveness, histopathological evaluation must be coupled with biomarkers. These might be based on IHC assays, which have become routine in most pathology laboratories, permitting its disseminated application. Ki-67 is undoubtedly in the frontline in this respect, but other biomarkers such as histone posttranslational modifications and modifiers will be probably required to fine-tune the assessment of prognosis. Prostate biopsies might also be used for 8q amplification analysis by FISH and DNA extracted from these tissue samples might also be used for quantitation of promoter methylation of genes that have been associated with worse disease survival and more aggressive PCa, such as APC. Indeed, cancer tissue samples are the most informative source for the search of alterations at proteomic, genomic, epigenomic or other -omic level and they might be maximized to yield the most relevant information for patient management. However, a major challenge lies in the widely recognized heterogeneity of PCa, which might prevent accurate risk stratification based on non-representative samples of tumorous tissue.

Five-year view

PCa will remain a major health concern, not only in western countries, but also in developing countries as improved life conditions will be followed by increased longevity and, necessarily, growing cancer incidence. Because this will occur mostly in countries of low economical resources, cost-effective means to detect and diagnose PCa will be an obvious challenge that must be met through improved technology. On the other hand, the growing concern about PCa overdiagnosis and overtreatment will lead inexorably to the development of more accurate biomarkers for PCa. New biomarkers are needed to avoid unnecessary biopsies and radical prostatectomies to distinguish benign from malignant lesions and to better discriminate localized from advanced disease. These new biomarkers might be used in tissue or body fluid samples and are expected to guide clinicians toward a more accurate and personalized treatment of PCa.

The recent development of wide-spectrum technologies might also be of great utility because it will permit genome-wide analysis of a single tumor in a fast and relatively inexpensive manner. This will also allow for an exponential increase of data that might be generated from discovery or validation studies, providing more robust data with personalized profiles being generated. Bearing in mind that a multigene or target approach will probably lead to a more accurate management of this disease, there is still the need to identify individual biomarkers that would ultimately be part of more complex and accurate assays. Likewise, a multicenter perspective for validation of potential biomarkers is crucial to consolidate new findings that eventually show up. The development of next-generation and third-generation sequencing is expected to change the landscape of the assessment of genomic information from cancer patients within the next 5 years.

Key issues

• Prostate cancer (PCa), one of the most incident tumors in men worldwide, is highly heterogeneous ranging from clinically indolent to extremely aggressive.

• There is a current lack of optimal biomarkers to accurately stratify PCa patients according to their risk of developing clinically significant disease.

• Prostate-specific antigen is prostate specific but not PCa-specific leading to low accuracy in diagnosis of PCa and also in the identification of clinically significant tumors. This lack of specificity has led to overdiagnosis and overtreatment of indolent tumors.

• There is an urgent need to develop non-invasive, accurate biomarkers for PCa that may: distinguish low prostate-specific antigen tumors from benign prostatic hyperplasia; discriminate indolent from aggressive disease and identify metastatic disease.

• Recent developments in wide-spectrum genomic techniques allowed for the identification of both genetic and epigenetic aberrant patterns that may characterize different subsets of PCa.

• Several genetic and epigenetic alterations have been identified as promising biomarkers for PCa both in tissue and body fluids.

• Most candidate genetic and epigenetic biomarkers require further validation in large patient cohorts.

• A broader characterization of the prostate cancer genome and epigenome may contribute to more personalized risk stratification and an optimized clinical approach.

Financial & competing interests disclosure

This work was supported by Federal funds through Programa Operacional Temático Factores de Competitividade (COMPETE) with co-participation from the European Community Fund (FEDER) and by national funds through Fundação para a Ciência e Tecnología (FCT) under the projects EXPL/BIM-ONC/0556/2012 (CJ) and Project Ciência 2008 (CJ). The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending or royalties.

No writing assistance was utilized in the production of this manuscript.