Landmarks in prostate cancer diagnosis: the biomarkers


Professor Walter Artibani, Department of Urology, University of Verona, Policlinico Borgo Roma, P.le L.A. Scuro 10, 37135 Verona, Italy. e-mail:


  • • The main diagnostic biomarker in current use is prostate-specific antigen (PSA) and it is one of the recommended diagnostic tools from the European Association of Urology Guidelines on prostate cancer.
  • • One of the challenges with PSA is that men with very low levels of PSA can harbour prostate cancer, making it difficult to set a lower limit.
  • • Several modifications to PSA biomarker detection have been suggested to improve its sensitivity and selectivity including PSA density, free:total PSA, PSA velocity/doubling time and different PSA isoforms.
  • • However, there remains a need to improve accuracy of diagnosis and this has led to research in to a number of promising new biomarkers.
  • • These include genetic and blood or urine based biomarkers. The most advanced of these is prostate cancer gene 3 found in urine and developed into a commercial test in 2006.
  • • Other promising markers include circulating tumour cells (CTC) in blood, which have been correlated with survival in castration-resistant prostate cancer. A system for evaluating CTC was approved by the USA Food and Drug Administration in 2008.

α-methylacyl-CoA racemase


circulating tumour cells


European Association of Urology


erythroblastosis virus E26 transforming sequence (family)

%free PSA

percentage of free PSA, i.e. free/total PSA × 100


glutathione-S-transferase P


human kallikrein type 2


prostate cancer gene 3


[-2]proenzyme PSA


PSA density


PSA doubling time


PSA velocity


prostate-specific membrane antigen


single nucleotide polymorphism


transmembrane protease, serine2 (gene)


urokinase plasminogen activator


The recommendations of the European Association of Urology (EAU) Guidelines on Prostate Cancer state that the main diagnostic tools for prostate cancer include DRE, serum PSA level and TRUS [1]. It has been shown that PSA is a better predictor of prostate cancer than suspect findings on DRE or TRUS [2]. Increasing levels of PSA are associated with an increased risk of the disease but presently there is no upper or lower threshold limit [3]. Notably, men with very low levels of PSA can harbour prostate cancer; a rate of 6.6% has been indicated in men with a PSA level of ≤0.5 ng/ml [3]. Several modifications to PSA biomarker detection have been suggested to improve its sensitivity and selectivity including PSA density, free:total PSA, PSA velocity (PSAV)/doubling time (PSADT) and different PSA isoforms. This paper reviews the use of PSA-based biomarkers and the broad spectrum of new markers that are currently being researched (Table 1).

Table 1.  Biomarkers in prostate cancer
PSA basedBiomarkers in urineBiomarkers in bloodCellular markersGenetic markers
PSAPCA3KallikreinsCytokeratinNucleotide polymorphisms
Free:total PSAProteinsuPAp63Gene fusions
PSAD Interleukin-6PSMA 
PSA isoforms CTCKi-67 



Most PSA in the blood is bound to serum proteins and the small amount that is not bound is termed free PSA. The percentage of free PSA (%free PSA, i.e. free/total PSA x 100) has been used to stratify the risk of prostate cancer in men with total PSA levels of 4–10 ng/mL and a negative DRE. With a free:total PSA threshold of <0.1, prostate cancer was detected in 56% of men on biopsy compared with 8% of men with a free:total PSA level >0.25 [4]. A large scale study of 14 453 has recently reported the value of %free PSA to reduce the risk of over diagnosis [5]. The indications for a prostate biopsy were suspicious DRE, total PSA of >10 ng/mL, total PSA of ≤2.5 ng/mL with %free PSA <15%, total PSA of 2.6–4 ng/mL with %free PSA of <20% and total PSA of 4.1–10 ng/mL with %free PSA <25%. The cancer detection rate was 28.8% in men with a total PSA of ≤10 ng/mL and there was a low incidence of pathologically insignificant prostate cancer. The diagnostic performance of %free PSA among men with a total PSA of 2–10 ng/mL compared with total PSA alone was shown to be improved in one meta-analysis [6]. The diagnostic performance of the free:total PSA was significantly higher in the total PSA range of 4–10 ng/mL compared with a total PSA range of 2–4 ng/mL; at a sensitivity of 95%, the specificity was 18% in the 4–10 ng/mL PSA range and 6% in the 2–4 ng/mL PSA range. There are limitations to the use of free PSA in that it is unstable at 4 °C and room temperature [7] and can produce conflicting results in men with BPH and large prostates [8]. A comparison of 10 different free PSA assay kits has been reported showing that there was variability in the values recorded that impacted the free:total PSA measure [9]. Consequently, it is important that combinations of free and total PSA assays be carefully selected based on validated diagnostic performance.


PSAV and PSADT have been used to measure the change in PSA level with time, the former recording the change per year and the latter a specific value increase. PSAV has been shown on a multivariate analysis including age, date of diagnosis, and PSA, to significantly improved the ability to detect high-risk prostate cancer unlike PSADT [10]. Nevertheless, a systematic review published in 2009 of 87 papers suggested that there was scant evidence to show that PSAV or PSADT provided predictive information that was better than PSA level alone [11]. The current EAU guidelines state that PSAV and PSADT have limited use in the diagnosis of prostate cancer due to background noise (total volume of prostate, BPH), the variations in interval between PSA determinations, and acceleration/deceleration of PSAV and PSADT over time [1].


Retrospective studies have suggested that an isoform of proenzyme PSA called [-2]proenzyme PSA (p2PSA) may enhance the specificity of PSA-based screening [12,13]. Other studies suggest that p2PSA and its derivatives, namely %p2PSA and the Beckman Coulter Prostate Health Index (phi), a mathematical combination of total PSA, free PSA, and p2PSA, may significantly improve the accuracy of total PSA and %free PSA in predicting the presence of prostate cancer [14,15]. A recent prospective study has compared the diagnostic accuracy of serum total PSA, %free PSA, PSA density (PSAD), p2PSA, %p2PSA ([p2PSA/fPSA] x 100) and phi in a group of 268 patients of whom 107 (39.9%) were diagnosed with prostate cancer at prostate biopsy [16]. Univariate analysis indicated that phi and %p2PSA were the most accurate predictors of prostate cancer followed by PSAD, %free PSA and total PSA. In multivariate accuracy analyses, both phi and %p2PSA significantly improved the accuracy of established predictors in determining the presence of prostate cancer at biopsy.



Several new genomic-based biomarkers for prostate cancer have been identified and 35 single nucleotide polymorphisms (SNPs) have been independently validated as being associated with prostate cancer [17]. It is estimated that these markers explain 20% of the familial risk of prostate cancer. The SNPs associated with prostate cancer so far are not associated with disease stage or outcome. A large scale study evaluated seven SNPs in 7370 patients with prostate cancer and 5742 controls, and found no association with tumour grade [18]. Another study reported an association between two variants, rs10993994 and rs5945619 and Gleason score [19].


Gene fusions of two distinct gene transcripts can occur after chromosomal translocation or deletion of segments of the genome, and such rearrangements are frequently the trigger point in oncogenesis [20]. The most commonly studied gene fusion in prostate cancer involves the prostate-specific gene transmembrane protease, serine2 (TMPRSS2) and members of the erythroblastosis virus E26 transforming sequence (ETS) family of transcription factors. This fusion has been identified in most PSA-screened prostate cancers [21,22]. The gene fusion TMPRSS2:ETV1 is rare and occurs in 1–10% of prostate cancers [23], while the TMPRSS2:ERG fusion is present in up to 50% of prostate cancers [23]. A recently completed study has examined the change in prostate cancer gene 3 (PCA3) and TMPRSS2:ERG expression during hormonal therapy [24].



High-molecular-weight cytokeratin present in the basal cells of prostatic glandular epithelium can be labelled with antibodies. Invasive prostate cancer is lacking in a basal cell layer, although there is low incidence of incorrect labelling [25]. Adding to the inaccuracy is the fact that benign lesions may sometimes be negative for the marker [26]. Staining for the marker should be examined in conjunction with conventional morphology.


p63 is another marker for the basal cell layer and can be used in conjunction with high-molecular-weight cytokeratin [27]. This marker does have the same inaccuracies as cytokeratin and the sensitivities of the two are similar [28].


AMACR is differentially expressed in prostate cancer usually being negative in benign glands [29]. Staining for AMACR can improve the accuracy of needle biopsy samples with a sensitivity of 80–100% in small atypical foci [30]. A correlation between AMACR staining and increased Gleason score has been reported [31]. However, AMACR is also positive in prostatic intraepithelial neoplasia and occasionally in benign lesions, so care is required in interpreting results [31].


PSMA is a cell surface membrane protein that is localised to the prostate and overexpressed in all tumour stages [32]. The protein has been researched in recent years from the perspective of diagnosis as well as therapy of prostate cancer. Identification of the protein is through radioisotope labelled antibody. High levels of PMSA expression have been associated with advanced tumour stage [33], androgen-independent tumour growth [34], presence of metastases [35] and early PSA recurrence [36], which could have implications for treatment decisions.


Immunohistochemical methods have been used to quantify the expression of the Ki-67 antigen and indicate that it provides an estimate of the growth fraction [37]. Ki-67 staining index has consistently been a significant correlate of outcome for patients with prostate cancer treated definitively with radical prostatectomy or radiotherapy [38,39].



PCA3 is a noncoding RNA that is highly overexpressed in prostate cancer tissues [40]. Testing for the biomarker involves collection of urine samples after a DRE is conducted. Samples are centrifuged and the level of PSA mRNA and PCA3 mRNA measured; the ratio of the two normalises for the variable number of prostate cancer cells collected [41]. The sensitivity of the optimal PCA3/PSA ratio has been shown to be 67% and the specificity 83% [41]. The largest double-blind study to evaluate the performance of PCA3 to date was the REDUCE (REduction by DUtasteride of prostate Cancer Events) trial involving PCA3 analysis of 1140 participants [42]. The PCA3 score was shown to correlate with the percentage of biopsy positive men and again the sensitivity and specificities reported were 48% and 79%, respectively. Furthermore, a predictive model incorporating PCA3, serum PSA level and %free PSA improved the diagnostic accuracy compared with PSA level and %free PSA. Other research has indicated that PCA3 was correlated with prostatectomy Gleason score [43], although other studies dispute this [44]. Identifying a biomarker indicative of tumour aggressiveness is still one of the unmet needs in prostate cancer. A sensitive, urine-based, quantitative test for PCA3 was developed by Gen-Probe Incorporated and released commercially in Europe in 2006 under the brand name PROGENSA® PCA3.


Several protein biomarkers for prostate cancer in urine have been studied including annexin 3 [45], matrix metalloproteinases (MMPs) [46], delta-catenin [47], hepatocyte growth factor (c-met) [48] and thymosin β15 [49]. The calcium-binding protein annexin 3 shows decreased production in prostate cancer compared with BPH; improved sensitivities have been noted when it was used in conjunction with PSA [45]. MMPs have a role in growth, invasion and metastatic spread in a number of tumours including prostate cancer [46]. The detection of any MMP in urine as predictive of prostate cancer has an 82% specificity and a 74% sensitivity [50]. Further studies are required to fully assess the clinical value of protein-based urinary markers in prostate cancer.


Several DNA methylation markers have been investigated in prostate cancer. The most common one studied is the loss of glutathione-S-transferase P (GSTP1) expression as a result of promoter hypermethylation; however, results are conflicting [51]. More recently a panel of markers, GSTP1, retinoic acid receptor β2 (RARβ2) and adenomatous polyposis coli (APC) have been assessed but no increased predictive value was shown compared with total PSA level and DRE alone [52]. Again the clinical utility of DNA-based markers needs to be evaluated in large scale studies.



Kallikrein 3 is also known as PSA and forms one of the 15 genes on the KLK gene locus on chromosome 19q13-4 [53]. Another member of the group is human kallikrein type 2 (hK2), which is also overexpressed in prostate cancer [54]. It has been shown that the ratio of hK2/PSA mRNA increases with grade of tumour [54]. Other members of the family are being studied for their prognostic significance.


Urokinase plasminogen activator (uPA) is involved in various phases of tumour development and progression. The suggested mechanism involves first binding to the uPA receptor and then converting plasminogen to plasmin, which in turn activates proteases that degrade extracellular matrix proteins [55]. Aggressive prostate cancer recurrence has been associated with over expression of uPA and its inhibitor, plasminogen activator inhibitor 1 (PAI-1) [56]. Other studies show that uPA and its receptor are linked to prostate cancer stage and bone metastases [57,58]; further large sale studies are being conducted.


TGF-β1 is involved in proliferation immune response, differentiation and angiogenesis plus regulating other cellular mechanisms [59]. In prostate cancer, TGF-β1 promotes cell progression, and is linked with higher tumour grade, invasion and metastasis [60].


Interleukin-6 is involved in the regulation of various cellular functions, among them proliferation, apoptosis, angiogenesis, differentiation and regulation of the immune response [61]. Elevated levels of interleukin-6 and its receptor have been linked with metastatic and hormone-refractory prostate cancer [62], suggesting its potential as a predictor for progression and survival in the disease.


A correlation between the number of CTCs in blood and survival in castration-resistant prostate cancer has been reported [63]. The median overall survival was significantly shorter in patients with ≥5 CTCs/7.5 mL compared with those with a CTC level of <5/7.5 mL. The CellSearch System (Veridex LLC) to determine CTC count received clearance in 2008 from the USA Food and Drug Administration for evaluation in castration-resistant prostate cancer and subsequent studies have shown the utility of the system in clinical trials [64,65].


The questionable specificity and potential for the over diagnosis of nonthreatening prostate cancers with PSA screening has led to research on a range of biomarkers, some of which show potential in not only diagnosis but also prognosis after treatment. The requirement of a marker with both high sensitivity and specificity is still unmet at present and a promising direction is the use of multiple markers in combination with other clinical factors.


None declared.