The role of biomarkers in undiagnosed men

Abstract

Purpose of review

This article intends to review biomarkers derived from blood, urine, and tissue that can aid in the diagnosis of prostate cancer.

Recent findings

Prostate cancer (PCa) screening requires tools that complement prostate-specific antigen (PSA) with a higher specificity for clinically significant disease. Novel blood biomarkers, such as the prostate health index and 4Kscore, utilize isoforms of PSA to more accurately predict high-grade PCa than traditional tools such as PSA and the percentage free-to-total PSA. Several gene products associated with PCa can be detected in the urine through commercially available assays. Prostate cancer antigen 3 (PCA3), though approved for repeat biopsy decisions, appears inferior to other biomarkers such as phi for identifying aggressive disease. However, combinations of PCA3 with other urine assays have shown promising results. One tissue-based hypermethylation test, named ConfirmMDx, can also be used to determine the need for repeat biopsy in men with a prior negative biopsy.

Summary

Several biomarkers have been developed to aid in the screening and diagnosis of PCa. Such tests are often indicated in men with moderately elevated PSA or history of a prior negative biopsy. Their use facilitates reduction of unnecessary biopsies without sacrificing the early diagnosis of clinically significant PCa.

Introduction

Although most prostate cancer (PCa) is currently diagnosed through prostate-specific antigen (PSA) screening, this test is not specific for clinically significant PCa. False positives trigger unnecessary biopsies, and biopsy risks have increased. Furthermore, PCa represents a wide spectrum of disease, ranging from clinically indolent to aggressive, high-grade cancers. PSA-based screening leads to overdiagnosis, and better markers are needed for significant cancer. Several PCa marker tests are now available, derived from blood, urine and tissue. Our objective is to review the evidence on currently available biomarkers for prostate biopsy decisions.

Blood Biomarkers

Free PSA(fPSA)

Serum PSA exists in two forms, either bound to proteins (complexed PSA) or unbound (free PSA). The majority of total PSA (tPSA) is complexed to α₁-antichymotrypsin, α₂-macroglobulin, or α₁-protease inhibitor.(1) Men with PCa have a lower fraction of fPSA (%fPSA) compared PCa-free men.(2)

The FDA approved the use of %fPSA for screening in men with PSA’s of 4–10ng/mL. The test is also included in professional guidelines on early detection of PCa from the National Comprehensive Cancer Network (NCCN) and European Association of Urology (EAU). The NCCN suggests %fPSA in men with no prior biopsy and PSA >3ng/mL, or prior negative biopsy with continued suspicion of cancer.(3) The EAU recommends its use in men with PSA 2–10ng/mL prior to prostate biopsy.(4) There is currently no standardized threshold of %fPSA for clinical use.

The utility of %fPSA is predicated on improved specificity for PCa and clinically significant PCa in men with moderately elevated PSA.(5, 6) A prospective, multicenter study of 773 men with PSA 4–10ng/mL demonstrated that a 25% fPSA cutoff detects 95% of PCa and avoids 20% of unnecessary biopsies (6). Even men with PSA <4ng/mL can harbor significant disease (7), and %fPSA has further proven valuable in this group (8, 9).

Ankerst et al recently evaluated serial measurements of PSA and %fPSA in 2,183 men from a screening study (10). In men that developed PCa, %fPSA was an earlier indicator of cancer than PSA in 71% and 34% using %fPSA thresholds of 25% and 15%, respectively. The study also demonstrated that %fPSA values were prone to fluctuations, particularly in patients never diagnosed with PCa. Therefore, repeat measurements might be informative when initial values are near the cutoff.

Though fPSA has demonstrated significant value as a standalone test, it is also an important part of newer tests such as the Prostate Health Index and 4Kscore.

Prostate Health Index (phi)

Incomplete processing of precursor PSA yields an isoform of fPSA named [-2]proPSA. This marker is elevated in men with PCa and has higher specificity than tPSA (11). Phi combines tPSA, fPSA, and [-2]proPSA as follows: ([−2]proPSA/fPSA)x√PSA, to predict the probability of PCa and high-grade disease on biopsy.

In 2012, the FDA approved phi for men with PSA levels of 4–10ng/mL. The NCCN and EAU suggest phi before initial or repeat biopsy.(3, 4)

Numerous studies have demonstrated improved accuracy of phi compared with tPSA and fPSA (12–15). Catalona et al examined 892 men with PSA 2–10ng/mL and found that phi improved specificity of cancer detection and was significantly associated with higher Gleason score.(12) A multicenter, prospective study of 658 men with PSA between 4–10ng/mL investigated phi specifically for detection of clinically significant cancer.(13) Phi was most accurate for detection of all cancers (AUCs phi 0.708, %fPSA 0.648, [-2]proPSA 0.550, PSA 0.516), Gleason ≥ 7 (AUCs phi 0.707, %fPSA 0.661, [-2]proPSA 0.558, PSA 0.551), and Epstein significant cancer (AUCs phi 0.698, %fPSA 0.654, [-2]proPSA 0.550, PSA 0.549). Using a phi cutoff of 28.6, 10.1% of Epstein significant cancers, 4.8% of Gleason ≥ 3+4, and 1.2% of Gleason ≥ 4+3 would be missed. However, 30.1% of men with benign or insignificant disease could avoid an unnecessary biopsy using phi compared to only 21.7% avoided with %fPSA. By reducing unnecessary biopsies, phi as a reflex test after PSA may improve cost-effectiveness of screening (16).

The value of phi is evident in patient populations from the US, Europe and Asia. It has demonstrated utility for both initial and repeat biopsy (17, 18). Although phi is most commonly utilized in men with a “borderline” PSA of 4–10ng/ml, Lazzeri et al also showed its utility in men with PSA >10ng/mL.(19) Furthermore, phi has been validated in men with a family history of PCa, African-American ethnicity, age <60, and obesity (20–23).

Multiple guidelines now recommend a multivariable, risk-adapted approach to biopsy decisions, and biomarkers like phi can be considered in the context of other risk factors. Several multivariable tools incorporating phi have been developed to predict an individual patient’s risk of PCa. Lughezzani et al developed a nomogram using age, prostate volume, digital rectal examination (DRE), biopsy history, and phi that outperformed the base model with tPSA.(24, 25) The European Randomized Study of Screening for Prostate Cancer Risk Calculator (ERSPC-RC) developed another tool based on age, family history, DRE, prior biopsy results, tPSA, and prostate volume. Addition of phi improved prediction of clinically significant PCa (AUC 0.78 vs 0.72, p=0.04).(26) This calculator is available as a smartphone application for clinical use. More recently, Loeb et al developed a new nomogram using phi with other variables to predict clinically significant PCa.(27)

Phi also appears to provide prognostic information. Several studies have demonstrated that preoperative phi can predict pathological stage, grade, tumor volume, and biochemical recurrence following radical prostatectomy (RP) (28–31). In patients on active surveillance (AS), baseline and longitudinal phi values are predictive of biopsy reclassification (32). Thus, phi can be useful for both selecting and monitoring patients for AS.

4Kscore

The 4Kscore incorporates both serum biomarkers (tPSA, fPSA, intact PSA, and human kallikrein 2) and clinical variables (age, DRE, and prior biopsy results) to predict risk of high-grade PCa on biopsy. The NCCN and EAU recommend its use as a reflex test in men with moderately elevated PSA (3, 4). The 4Kscore is commercially available, but unlike %fPSA and phi, it is not currently FDA-approved.

Many studies have established that that the 4-kallikrein panel improves accuracy of PCa detection compared with other multivariable tools, including the Prostate Cancer Prevention Trial Risk Calculator (PCPTRC)(33–35). It has been validated for both initial and repeat biopsy (36, 37). The largest of these trials examined 6,129 men from the ProtecT study with elevated PSA who underwent prostate biopsy (34). The base model for PCa detection incorporating age and tPSA was compared to the base model plus %fPSA or the 4-kallikrein panel. The 4-kallikrein model most accurately predicted high-grade cancer (AUCs 4-kallikrein 0.820, %fPSA 0.799, tPSA 0.738; p<0.001). Using one particular threshold for the 4-kallikrein panel, 43% of unnecessary biopsies would be avoided while missing 11% of high-grade cancers. In a comparative study, Nordström et al demonstrated that the 4-kallikrein panel and phi had similar accuracy for detecting any PCa and high-grade PCa (38).

The 4Kscore was recently evaluated in the Canary AS cohort. In this setting, longitudinal monitoring of the 4K score improved prediction of reclassification on the initial AS biopsy, but was no better than PSA for monitoring patients subsequent to that (39).

The aforementioned studies validate the 4Kscore using biopsy outcomes. However, prostate biopsy, particularly non-targeted biopsy, is imperfect and thus may misrepresent true disease burden. Using a different approach for validation, one study demonstrated that the 4-kallikrein panel improved prediction of aggressive RP pathology.(40) Stattin et al reported that men with moderately elevated PSA but low-risk 4Kscore had a very low likelihood of subsequent metastatic disease and thus may safely avoid biopsy (41). For example, men aged 60 with PSA ≥3ng/mL and 4Kscore <5% had a 20-year risk of metastasis <0.5%.

Urine Biomarkers

Prostate Cancer Antigen 3(PCA3)

PCA3 is a noncoding mRNA that is overexpressed in PCa tissue and detectable in urine after vigorous DRE (42). The FDA approved the use of PCA3 utilizing a cutoff of 25 in men age ≥50 with ≥1 prior negative biopsy. The NCCN recommends using a cutoff of 35 in men with moderately elevated PSA for whom repeat biopsy is being considered, but the EAU makes no distinction regarding biopsy history (3, 4).

Numerous studies indicate that PCA3 has greater accuracy for overall PCa detection in the repeat biopsy setting compared with tPSA and %fPSA (43–46). Additionally, some evidence suggests that PCA3 improves cancer detection prior to initial biopsy (46–48). Wei et al studied both indications in 859 men undergoing prostate biopsy (49). A threshold of PCA3 <20 in the repeat setting would avoid 46% of biopsies while missing 3% of high-grade cancer. However, that same threshold would miss 13% of high-grade cancer on initial biopsy.

Data are conflicting regarding the association of PCA3 score with high-grade or clinically-significant cancer detected on biopsy (43, 44, 48, 50). Furthermore, studies suggest that PCA3 score is not associated with progression on AS or RP pathology.(51–54) Comparative studies have demonstrated that phi outperforms PCA3 for prediction of clinically significant PCa at biopsy and RP.(55, 56) In 120 men eligible for AS who opted for RP, pre-operative MRI was best predictor of clinically significant PCa (57); phi was also a significant predictor but PCA3 was not. As the current paradigm emphasizes detection of aggressive PCa, such evidence diminishes the value of PCA3.

Perhaps the most appropriate application of PCA3 is in conjunction with other clinical tools. PCA3 combined with either MRI or real-time elastography was shown to improve the detection of significant PCa (58, 59). Several PCA3-based nomograms have been developed and validated, including one incorporating PCA3 into the PCPTRC (49, 60–63). One such study estimated that its nomogram would avoid up to 55% of unnecessary biopsies while missing ≤2% of high-grade PCa (60).

TMPRSS2:ERG

TMPRSS2:ERG (T2:ERG) is a gene fusion product that is highly specific for PCa and detectable in urine after DRE (64, 65). In a multi-institutional study of 1312 men, T2:ERG demonstrated greater diagnostic accuracy than tPSA (66). Furthermore, T2:ERG levels were associated with clinically significant PCa detected on biopsy and RP specimens in a subset of patients undergoing surgery.

PCA3 and T2:ERG

Although T2:ERG has high specificity for PCa, its low sensitivity reduces its value as a standalone test. Several studies have shown that combining PCA3 with T2:ERG can improve the prediction of PCa on biopsy (66–68).

A commercially available test, termed the Mi-Prostate Score (MiPS), incorporates serum tPSA and urinary PCA3 and T2:ERG to predict risk of PCa and high-grade PCa. In 1244 men undergoing biopsy, Tomlins et al examined the added value of PCA3, T2:ERG, or MiPS to a base model of either tPSA or the PCPTRC (69). For the prediction of high-grade PCa, MiPS (AUC 0.772) outperformed PSA+PCA3 (AUC 0.747, p<0.001), PSA+T2:ERG (AUC 0.729, p<0.01), and tPSA alone (AUC 0.651, p<0.001). Similarly, the addition of MiPS to the PCPTRC was superior to the base models. These results held true in the setting of either initial biopsy or repeat biopsy. Using various cutoffs, the MiPS-PCPTRC model would avoid 35–47% of unnecessary biopsies while missing 1.0–2.3% of high-grade PCa. MiPS is another promising reflex test following PSA screening, but it has not yet been directly compared with other biomarkers such as phi and 4Kscore.

SelectMDx

SelectMDx is another commercially available test that measures urinary RNA levels of two genes (DLX1 and HOXC6) following DRE. Using an algorithm including tPSA, PSA density, DRE, age, and family history, it provides an individual patient’s likelihood of low- and high-grade PCa. Van Neste et al developed this tool in 519 men undergoing prostate biopsy and subsequently validated it in a cohort of 386 men (70). SelectMDx demonstrated an AUC of 0.86 (95% CI 0.80–0.92) for high-grade PCa and outperformed the base model without RNA markers and the PCPTRC. One model, which excluded DRE, was superior to the PCPTRC combined with PCA3. Decision curve analysis suggested that 42% of biopsies could be avoided while missing 2% of high-grade PCa.

Urine Exosomes

ExoDx Prostate IntelliScore is a novel test analyzing exosomal RNA in urine without a prior DRE. It examines three PCa-associated genes (ERG, PCA3, and SPDEF) to predict the risk of high-grade PCa in men undergoing initial biopsy. It is intended to be used alongside standard of care (SOC) clinical factors such as tPSA, age, race, and family history. A validation study compared SOC with the urine exosome gene expression assay plus SOC for the detection of high-grade PCa (71). In 519 men, it demonstrated greater diagnostic accuracy than SOC. Setting a predefined cutoff would avoid 27% of unnecessary biopsies while missing 5% of dominant Gleason pattern 4 disease. It is unknown how this test compares with other available urine biomarkers.

Tissue Biomarkers

ConfirmMDx

Since prostate biopsy represents only a sampling of prostatic tissue, it may miss some significant PCa, particularly when performed in a non-targeted fashion. ConfirmMDx is a commercially available test that examines the tissue from a negative prostate biopsy to predict the presence of missed cancer in adjacent tissue. This test capitalizes on the presence a field effect around tumors. Epigenetic changes that alter gene expression are found not only in prostate tumors, but also in benign tissue adjacent the tumor. ConfirmMDx examines prostatic tissue for epigenetic changes, specifically the hypermethylation of three markers previously associated with prostate cancer (GSTP1, APC, and RASSF1).

Unlike other commercially available markers that evaluate tumor tissue to guide prognosis and decisions about treatment (e.g. Prolaris, OncotypeDx, and Decipher), ConfirmMDx is the only tissue marker that examines benign tissue to predict the presence of occult cancer. The NCCN guidelines suggest ConfirmMDx as another optional test to aid in decisions about repeat prostate biopsy (3).

ConfirmMDx has been validated in clinical studies from multiple countries. The MATLOC study included 483 men from the UK and Belgium with prior negative biopsy (72). ConfirmMDx was a significant predictor of PCa on repeat biopsy (OR 3.17, 95% CI 1.81–5.53), with a negative predictive value (NPV) of 90%. A subsequent US validation study of 320 men similarly reported a 2.5-fold increased risk of PCa on repeat biopsy with positive methylation status and a NPV of 88% (73). These two previously published cohorts were combined in a subsequent study, which confirmed that ConfirmMDx was a significant predictor of high-grade PCa, with a NPV of 96% (74). These results affirm a low risk of detecting significant PCa in men with negative findings on ConfirmMDx, suggesting that these men can safely defer a repeat biopsy.

In contrast to serum and urine markers, ConfirmMDx offers a potential advantage by localizing where additional sampling is warranted on repeat biopsy. However, the cost-effectiveness of ConfirmMDx within the context of an experienced multiparametric MRI program is unknown.

Conclusion

PSA remains the only marker demonstrated to reduce prostate cancer mortality in randomized trials. Nevertheless, there are an abundance of novel biomarkers which are commercially available and being incorporated into guidelines to assist with prostate biopsy decisions. Though some are intended specifically for either initial or repeat biopsy, there is much overlap in their application. Studies comparing phi with other biomarkers have been published, but there remains a need for further studies that compare these biomarkers head-to-head. As more data on MRI continues to emerge and its use in prostate cancer detection expands, more studies are also needed examining the cost-effectiveness of integrating biomarkers with MRI in this setting. With that, we might elucidate how to best incorporate all of these new tools into a risk-adapted screening protocol that accurately detects clinically significant PCa while minimizing diagnosis and treatment of insignificant PCa.

Key Points.

• Several biomarkers, derived from blood, urine, and tissue, can aid in decisions regarding prostate biopsy.

• Phi and the 4Kscore more accurately detect high-grade PCa than do PSA and free PSA.

• PCA3 is another test for repeat biopsy but is inferior to phi for identifying high-grade disease

• ConfirmMDx is the only tissue biomarker to predict occult cancer in men with negative biopsies

• Such tests can selectively identify patients who should undergo prostate biopsy, thus reducing unnecessary biopsies and diagnosis of indolent PCa.