Prostate cancer is the most diagnosed, non-cutaneous malignancy, and the third leading cause of cancer-related deaths in American men (behind lung and colorectal cancer). Approximately 165,000 men are expected to be diagnosed with prostate cancer in 2018, with approximately 18% dying of the disease (2). However, prostate cancer is a heterogeneous disease with a clinical course ranging from indolent to life-threatening. Prostate-specific antigen (PSA) screening, introduced around 1990 (3), resulted in a marked drop in the incidence of metastatic disease at diagnosis, and probably, but not definitely, reduced prostate cancer-specific mortality (1,4). However, a concomitant over-diagnosis (via prostate biopsy) and over-treatment of early-stage and indolent disease occurred as well (5-9). Only about 25% of men with PSA in the 4-10 ng/mL range have prostate cancer on biopsy, and of those, about 20-50% are indolent, disease that would not be a problem if undetected or untreated (10,11). As noted in a recent editorial, “in a biological sense, of course, screening does not cause prostate cancer, but in a practical sense, it does (12).” The results of three major PSA screening trials involving hundreds of thousands of men, the US Prostate, Lung, Colorectal and Ovarian Cancer Screening (PCLO) trial, the European Randomized Study of Screening for Prostate Cancer (ERSPC) trial, and the Cluster Randomized Trial of PSA Testing or Prostate Cancer (CAP), all show the risk-benefit ratio to be a close call (6,13,14). Because screening seemed to be doing more harm than good, the US Preventive Services Task Force (USPSTF) advised against PSA testing in 2012, though this is under reconsideration for men aged 55 to 69 years (10,15).
However, ample evidence has shown survival benefits associated with treatment of intermediate- and high-risk, early-stage prostate cancer (6,16). A Gleason score of 2 to 5 is regarded as normal prostate tissue; 6 is low-grade prostate cancer that usually grows slowly; 7 is an intermediate grade; 8 to 10 is high-grade cancer that grows more quickly. Ten-year survival rates stratified by Gleason score have been estimated from the Surveillance, Epidemiology, and End Results registry to be about 98% for scores 2 through 6, 92% for a score of 7 with primary pattern 3 and secondary pattern 4 (3+4), 77% for a score of 7 (4+3), and 70% for scores between 8 and 10 (17).
Unfortunately, PSA is not even specific to prostate cancer, much less clinically significant prostate cancer. Therefore, the current focus is on finding a more nuanced approach (beyond PSA and digital rectal exam (DRE)), by reserving biopsy and treatment for men with clinically significant, higher-grade (Gleason ≥ 7) prostate cancer (HGPC). Such testing would represent secondary or reflex testing, rather than screening, effectively allowing for risk stratification and a more targeted response to PSA screening results (18). This strategy has the potential to not only decrease biopsies (and associated risks), but also of reducing detection of indolent disease (and the attendant risks of overtreatment). The primary aim is to increase specificity compared with PSA without decreasing the sensitivity to diagnose high-risk prostate cancer.
This Local Coverage Determination (LCD) will focus on biomarker testing used to refine selection of patients for initial (not repeat) biopsy.
Percent Free PSA (%fPSA)
Unbound or free PSA (fPSA), expressed as a ratio of total PSA (tPSA), is significantly lower in men with prostate cancer. In 1998, a large, prospective, multicenter study showed that a 25% %fPSA cutoff detected 95% of prostate cancers (of any grade) while avoiding 20% of unnecessary biopsies (11).
In 2003, the FDA approved %fPSA for use “as an aid in distinguishing prostate cancer from benign prostate conditions in men 50 years or older with total PSA 4-10 ng/mL, and DRE findings not suspicious for cancer (19).”
In a recent publication, among 417 men with a PSA>2, using a 25% cutoff, 83 (20%) would have avoided a biopsy (20). %fPSA has since become widely adopted more for active surveillance post prostate biopsy (1).
Prostate Health Index (PHI)
The PHI adds [-2]proPSA (p2PSA) (a fPSA isoform associated with PCa) to tPSA and fPSA in an algorithm that calculates a score that has demonstrated a correlation with HGPC. The test was initially validated in 2011 in a multi-institutional, prospective trial evaluating 892 men (with no history of PCa, a normal DRE, and PSA of 2-10 ng/mL) for the presence of Gleason ≥4+3 prostate cancer (21). The receiver-operating characteristic curve (AUC) for PHI (0.724) exceeded that of %fPSA (0.670) in discriminating between Gleason ≥4+3 vs. lower Gleason grade PCa or negative biopsies. No optimal PHI cutoff was defined, however. Shortly thereafter, in 2012, PHI was FDA approved with identical indications to %fPSA (22).
A subsequent, prospective multi-center study of 658 men with a PSA of 4-10 ng/mL and normal DRE showed AUCs of 0.707 for PHI, 0.661 for %fPSA, and 0.551 for PSA, potentially sparing 30.1% of unnecessary biopsies with a PHI cutpoint of 28.6 (23). Another prospective multi-center studied two independent cohorts of 561 (primary) and 395 (validation) biopsy naive men (24). The primary and validation cohorts demonstrated an AUC for the detection of high-grade (GS ≥7) prostate cancer of 0.815 and 0.783, respectively. The study determined the optimal cutoff of PHI to be a score of 24, reducing unnecessary biopsies by 36% and only missing 2.5% of high-grade cancers. A subsequent publication by the same group showed that PHI significantly improved the performance of the Prostate Cancer Prevention Trial (PCPT) and ERSPC risk calculators in men with a PSA 2-10 ng/mL for predicting HGPC (25).
In a prospective, single-center observational study in 188 men with elevated PSA (> 2ng/mL) and negative DRE, PHI had a higher discriminative ability for clinically significant PCa (AUC 0.76) compared to PSA (AUC 0.52) or %fPSA (AUC 0.75%) (51). However, the authors focus seems to be on something called PHI density rather than PHI.
In another prospective observational study, 506 men over 50 years old, with PSA in the 4-10 ng/mL range, a negative DRE, and receiving a PHI test, were compared with a historical control group of 683 similar men (52). Men receiving a PHI test showed nearly a 24% reduction in biopsy procedures performed compared to the historical control group (36.4% versus 60.3%, respectively, p<0.0001). “Based on questionnaire responses, the phi score impacted the physician’s patient management plan in 73% of cases, including biopsy deferrals when the phi score was low, and decisions to perform biopsies when the phi score indicated an intermediate or high probability of prostate cancer (phi ≥36).” The authors conclude that the study “supports the routine use of PHI testing for men presenting with elevated serum total PSA and non-suspicious DRE findings.”
Of note, there is conflicting data on the optimal PHI cutoff, with another study claiming poor results using a cutoff of 25-30 (26). The study, comparing 4K and PHI, is described in more detail in the 4Kscore section.
The 4Kscore combines data from serum levels of four kallikrein proteins (fPSA, tPSA, iPSA, human kalilkrein 2 (hK2)), along with clinical information (age, DRE, prior negative biopsy) to estimate the percent likelihood of HGPC on biopsy using a proprietary algorithm.
In a population-based study involving 2914 men with elevated PSA >3 ng/mL, addition of free PSA, intact PSA, and hK2 to a model containing total PSA and age improved the AUC from 0.64 to 0.76 and 0.70 to 0.78 for models without and with digital rectal examination results, respectively (P <.001 for both) (27). The authors claim application of the panel could reduce biopsies by 51.3% and miss 12% of high-grade cancers.
In a multi-center prospective study of 1012 subjects scheduled for prostate biopsy, the predictive accuracy of the 4Kscore was compared to a modified Prostate Cancer Prevention Trial (PCPT) Risk Calculator 2.0 and showed superior discrimination in detecting HGPC (AUC 0.82 versus 0.74, p < 0.0001) (28). It was estimated that, depending on the 4Kscore cutoff (≥6% to ≥15%), biopsies avoided would range from 30% to 58%, and the number of HGPC missed would range from 1.3% to 4.7%.
In a study of 6129 men with elevated PSA (≥ 3 ng/mL), the 4Kscore AUC for predicting HGPC was 0.820 (95% CI = 0.802 to 0.838) vs 0.738 (95% CI = 0.716 to 0.761) for PSA and age alone (29). Using a 6% risk of high-grade cancer as a cutoff, the model would reduce biopsies by 42.8% and delay diagnosis in 14 of 133 (10.5%) of HGPC.
A multi-institutional clinical utility study was performed to evaluate the effect of the 4Kscore test in lieu of prostate biopsy for males referred to urologists for atypical PSA and/or DRE results (30). The study involved 611 subjects in 35 United States academic and community settings. Results for the patients were stratified into low risk (< 7.5%), intermediate risk (7.5%-19.9%), and high risk (≥ 20%) for aggressive prostate cancer. Performing the 4Kscore Test resulted in a 64.6% reduction in prostate biopsies in patients; the actual percentage of cases not proceeding to biopsy were 94.0%, 52.9%, and 19.0% for men who had low-, intermediate-, and high-risk 4Kscore results, respectively.
When comparing PHI and the 4K score, the two tests appear to demonstrate similar discriminatory ability in predicting high-risk prostate cancer in men with a PSA between 3 and 15 ng/mL (AUC 4Kscore 0.718 vs. PHI 0.711); both tests had a higher AUC than PSA and age alone (p<0.0001 for both) (26). Of note, the 4K panel showed net benefit when the cutoff for biopsy exceeds 8% risk for HGPC. The clinical utility of PHI was also strongly dependent on the cut-off used. PHI cutoffs of 25–30 had poor clinical utility compared to higher cutoffs (30-40). According to the study authors, the tests save almost 30% of the biopsies to the cost of missing 10% high grade cancers if using 10% risk of high grade cancer as predicted by the 4K panel or a PHI cutoff of 39.
However, a subsequent meta-analysis of twelve 4Kscore studies (N = 11,134), yielded an overall AUC of 0.81 (0.79-0.83), and found the Nordstrom study to be an outlier (AUC 0.72 vs. around 0.8 for others) (45). Excluding Nordstrom (a possible methodological flaw was admitted to by the Nordstrom authors (46)) increased this to 0.82 (0.80-0.84), and heterogeneity was no longer significant (p = 0.08). Interestingly, the authors also cite the lack of a specific “arbitrary” cutoff as a positive, saying “a continuous risk score from <1% to >90% that allows the physician and patient to act according to their own desired risk threshold.”
A prospective, multi-institutional study of 366 men (56% African American) showed better discrimination (AUC 0.81 vs. 0.74, p <0.01) than standard-of-care (SOC) (age, PSA, DRE) (47). There was no significant AUC difference for detecting clinically significant prostate cancer between African American and non-African Americans. In a retrospective study of 749 men referred for biopsy due to elevated PSA (≥3 ng/mL), low %fPSA (<20%), or suspicious DRE, the use of the 4Kscore (in conjunction with age and DRE) improved discrimination compared with SOC (age and PSA) for high-grade cancer (0.78 vs. 0.72; p = 0.002) (48). At a threshold of > 8%, 24% of biopsies would have been avoided and 13 high-grade cancers missed.
Two similarly designed retrospective, case-control studies address the question of whether biopsy outcome is a good proxy for long-term prostate cancer morbidity and mortality. A case-control studied 12,542 men enrolled at ages 40-60 and followed for >15yr; 1,423 developed incident PCa, 235 with distant metastasis (49). PSA and 4Kscore were measured in cryopreserved blood. Among those with a PSA ≥3ng/mL (2,432), 62% had a 4Kscore ≥7.5% and a 16.36% (95% CI: 12.44-20.74) risk at 20 years of distant metastasis, versus a 1.82% (95% CI: 0.47-4.99) risk among the 38% with a 4Kscore <7.5%. The authors conclude that the 4Kscore “can be used as a reflex test to aid biopsy decisions.” Another case-control studied 11,506 men enrolled at ages 45-73 and followed for >15yr; 1,223 developed incident PCa, 235 prostate cancer deaths (50). PSA and 4Kscore were measured in cryopreserved blood. Among men aged 60-73 with a PSA ≥2ng/mL (1,822), 54% had a 4Kscore ≥7.5% and a 24.21% (95% CI: 20.62-27.98) risk at 20 years of prostate cancer death, versus a 4.24% (95% CI: 2.64-6.40) risk among the 38% with a 4Kscore <7.5%. The authors conclude that “men with elevated PSA but low 4Kscores can be monitored rather than being subject to biopsy.”
The 4K score is not FDA approved, but rather a Laboratory Developed Test (LDT) through one CLIA-accredited testing laboratory in Nashville, TN.
ExoDx Prostate IntelliScore (EPI)
EPI is a urine-based 3-gene exosomal RNA expression assay. The EPI gene signature and score incorporates levels of PCA3 (PCa antigen 3), ERG (v-ets erythroblastosis virus E26 oncogene homologs) and SPDEF (SAM-pointed domain-containing Ets transcription factor). EPI uses a proprietary algorithm to translate the level of expression of these three genes into an individualized risk score that predicts the presence of HGPC, with a higher EPI score indicative of a higher probability of high-grade disease. EPI does not incorporate PSA and other SOC factors into the score, but is intended to be used in conjunction with SOC elements such as age, family history, PSA level and DRE results.
A study in men over 50 years without a prior biopsy and a PSA 2-10 ng/mL demonstrated a correlation with HGPC significantly better than SOC alone (31). In 255 men in the training target population (median age 62 years and median PSA level 5.0 ng/mL, and initial biopsy), EPI plus SOC was associated with improved discrimination of HGPC: AUC 0.77 (95%CI, 0.71-0.83) vs SOC AUC 0.66 (95%CI, 0.58-0.72) (P <.001). Results were similar in the independent validation of 519 patients; EPI plus SOC AUC 0.73 (95%CI, 0.68-0.77) was superior to SOC AUC 0.63 (95%CI, 0.58-0.68) (P <.001). Using a predefined cut point, 138 of 519 (27%) biopsies would have been avoided, missing 8% of HGPC cases, but only 5% of patients with dominant pattern 4 high-risk GS7 disease. An accompanying editorial indicates EPI has the advantage over other biomarkers in being the least invasive (requires neither DRE or phlebotomy) (32).
In a second prospective, nonrandomized, controlled clinical study of 503 men 50 years or older with a PSA between 2-10 ng/mL, EPI AUC 0.70 (95%CI, 0.65-0.75) was superior to both the Prostate Cancer Prevention Trial Risk Calculator (PCPTRC2.0) AUC 0.63 (95%CI, 0.58-0.68) and the European Randomized Study of Screening for Prostate Cancer (ERSPC) risk calculator AUC 0.59 (95%CI, 0.54-0.64) (53). Using the test’s predefined cut point (15.6), 101 of 503 (20%) biopsies would have been avoided, missing 7% of HGPC cases (11 ≥ GG2, of which 7 were ≥ GG3). The authors concluded “EPI is a noninvasive, easy-to-use, gene expression urine assay, which has now been successfully validated in over 1000 patients across two prospective validation trials to stratify risk of ≥GG2 from GG1 cancer and benign disease. The test improves identification of patients with higher grade disease and would reduce the total number of unnecessary biopsies.”
In a prospective, randomized, blinded study of 1048 patients considered for initial prostate biopsy based on an elevated PSA (2-10 ng/mL), while all patients had an EPI test, only those in the EPI arm received results for their biopsy decision (33). The sole objective primary metric was reduction of initial prostate biopsy by 15%. The EPI arm (N=458) had a median age of 64 years and PSA of 4.8 ng/mL, similar to the SOC control arm (N=484) with a median age of 65 years and PSA of 4.8 ng/mL. This was a “real-world” exercise in that the EPI score potentially influenced but did not necessarily determine the physician biopsy recommendation, and the recommendation likewise met with variable patient compliance. Unfortunately, control arm physician recommendations are not reported, precluding comparisons of patient compliance.
Ninety-three (20%) EPI group patients had a low risk EPI score of <15.6; of these, 59 (63%) were recommended to defer biopsy (54 (92%) complied) and 69 (74%) deferred overall. Three hundred sixty-five (80%) EPI group patients had a high risk EPI score of ≥15.6; of these, 318 (87%) were recommended for biopsy (229 (72%) complied) and 240 (66%) underwent biopsy overall. Of the 484 control patients, although physician recommendations are not detailed, ultimately only 190 (39%) underwent biopsy. Thus overall, the EPI arm had a higher (264/458; 57.6%) rather than lower biopsy rate than the control group (190/484; 39.3%). This translated to finding 30% more HGPC compared to the control arm (78 vs. 60). Among African-Americans, representing a significant 22% and 24% of patients in the EPI and control arms, respectively, 81% more HGPCs were found in the EPI arm (29 vs. 16). Using HGPC prevalence in the corresponding biopsied groups, the authors project that there were 46 presumed missed HGPC in the EPI cohort, and 94 presumed missed HGPC in the control cohort, a 48% reduction.
Unlike the validation studies which had no control group and assumed every low risk score was a biopsy deferred, the control group here demonstrated the focus perhaps should be on greater physician recommendation accuracy and patient compliance, rather than an absolute drop in biopsy rate. Compared to the control group biopsy rate of 39.3% (190/484), knowledge of EPI results stratified the biopsy rate between 65.8% (240/365) in the high risk group, and 25.8% (24/93) in the low risk group, resulting 18 more HGPC identified. Including both EPI and control groups, 10.3% (7/68) low EPI risk patients had HGPC, whereas 33.9% (131/386) high EPI risk patients had HGPC. The EPI arm physician recommendations experienced 73.8% (338/458) compliance; as noted previously, comparable control arm physician recommendations are not described. Overall 68% of urologists in the study reported that the EPI test influenced their biopsy decision, and 23% said they recommended biopsy deferral due to the EPI result.
Mi-Prostate Score (MiPS)
The MiPS assay measures tPSA and post-DRE urine expression of PCA3 and the TMPRSS2:ERG fusion gene. A validation study of 1244 men with planned biopsy (80% initial) found the AUC for the prediction of HGPC was 0.772 for MiPS compared with 0.651 for PSA alone (34). Using a predicted risk cutoff of ≥30%, 35% of biopsies could have been avoided, and only 1% of HGPC missed.
A multicenter prospective validation study of 561 men found a sensitivity and specificity for HGPC of 93% and 33%, respectively (35). The authors calculate that 42% of unnecessary biopsies could be avoided at the cost of only 7% missed HGPC.
SelectMDx is a gene expression assay measuring mRNA levels of homeobox C6 (HOXC6) and distal-less hemeobox 1 (DLX1) in post-DRE urine; both HOXC6 and DL1 may be involved in the onset of prostate cancer and are associated with HGPC (36). A prospective, multicenter study involved a training cohort (n=519) and validation cohort (n=386) in men scheduled for prostate biopsy (PSA ≥ 3ng/mL, abnormal DRE, or family history of prostate cancer) (37). Using a cut-off of 27.5 for the prediction of HGPC, the expression of DLX1 and HOXC6 alone resulted in an AUC of 0.76 and 0.73 for the two cohorts, respectively. Combined with other SOC risk factors, the AUC increased to 0.90 in the training set and 0.86 in the validation set.