Abstract
Purpose:
We sought to determine whether combining prostate health index (Phi) with urinary prostate cancer antigen 3 (PCA3) and TMPRSS2:ERG (T2:ERG) could improve selection of men for prostate biopsy. These biomarkers have been validated in prostate cancer (PCa) detection separately, but their combination has not previously been developed.
Materials and Methods:
Prebiopsy blood and post–digital rectal examination urine specimens were assayed to predict subsequent biopsy outcomes from training and validation cohorts (1073 participants across 11 academic centers). Clinical algorithms for combining Phi and PCA3-T2:ERG to predict Grade Group ≥ 2 (GG ≥ 2) PCa were formulated using the training cohort (N = 512). Prediction rules and hypotheses were locked before validation using biopsy-naïve men from the NCI Early Detection Research Network urinary PCA3 trial (N = 561). Rules were compared in weighted sum of specificity and sensitivity with weights specified a priori, and P values were obtained through bootstrap in the validation study.
Results:
Primary validation analysis showed that Phi combined with urinary PCA3 outperformed Phi alone (P = .002). Furthermore, serum Phi combined with urinary PCA3-T2:ERG outperformed urinary PCA3-T2:ERG in each of the 3 algorithms reflecting different potential clinical workflows: (1) serum Phi and urine PCA3-T2:ERG tested simultaneously, either exceeding its own threshold (P = .04); (2) urine PCA3-T2:ERG first and those in the grey zone resolved by subsequent serum Phi (P = .03); and (3) serum Phi first and those in the grey zone resolved by subsequent urine PCA3-T2:ERG (P = .002).
Conclusions:
Combining serum Phi with urinary PCA3 RNA alone or together with urinary T2:ERG RNA, simultaneously or sequentially, improves selection of men for initial prostate biopsy and represents an avenue to improve early detection of aggressive PCa.
Key Words: prostate cancer, PSA, biopsy, biomarkers
PSA screening has been shown to reduce prostate cancer (PCa) disease progression and mortality, however, not without increased morbidity and potential overtreatment that accompanies unnecessary prostate biopsies.1-3 Risk stratification of biopsy-naïve men to optimize detection of clinically significant PCa (Grade Group [GG] ≥ 2) while safely ruling out the need for biopsy remains a significant clinical challenge.
Serum and urine biomarkers for PCa provide 2 potential noninvasive avenues to improve patient selection for biopsy.4 Serum protein biomarkers for PCa detection include the prostate health index (Phi), a multiplex serum protein assay that combines 3 protein biomarkers derived from transcription of the KLK3 kallikrein gene: total serum PSA, free PSA (fPSA), and [-2]pro-PSA (p2PSA).5 Phi has been associated with improved detection of GG ≥ 2 PCa across different PSA ranges while reducing detection of indolent disease6,7 and impacting decision to biopsy.8 Urinary RNA biomarkers for PCa detection include prostate cancer antigen 3 (PCA3—a long noncoding RNA associated with PCa) and the TMPRSS2:ERG gene fusion (T2:ERG). Urinary PCA3 and T2:ERG have been shown to be effective both individually and together for GG ≥ 2 PCa detection,9-12 culminating in 2 commercially available clinical assays that leverage urinary PCA3 and ERG RNA to improve selecting men for prostate biopsy.9,13
Previous work evaluating the potential to combine serum kallikrein and urinary RNA testing to improve selection of men for initial prostate biopsy has been limited to one single institution study which showed that combining PCA3 and Phi improved overall PCa detection.14 However, that study did not evaluate initial diagnosis of aggressive cancers with GG ≥ 2. Moreover, incorporating serum Phi and urinary PCA3 RNA together with urinary T2:ERG RNA, for detection of GG ≥ 2 PCa has not, to our knowledge, been previously reported.
Therefore, we sought to determine whether combining these 3 markers could improve selection of men undergoing initial prostate biopsy for detection of GG ≥ 2 cancer.
MATERIALS AND METHODS
Study Population and Protocol
The training cohort (n = 512) was comprised of subjects prospectively enrolled at 3 academic medical centers (Cornell University, University of Michigan, and Beth Israel-Deaconess Medical Center) for participation in the Prostate Cancer Biomarker Clinical Center of the National Cancer Institute's Early Detection Research Network (NCI-EDRN) after respective Institutional Review Board (IRB) approval and informed consent; patients with a history of prior biopsy or cancer diagnosis/treatment were excluded from this analysis. The validation cohort (n = 561) consisted of biopsy-naïve men referred for biopsy as part of the NCI-EDRN PCA3 trial.15 Full inclusion criteria for both training and validation cohorts have been previously described.10
Biospecimen Collection and Assays
First-catch post–digital rectal examination urine samples were prospectively collected, preserved, and stored according to established protocols.9,10,16 PCA3 and T2:ERG assays were conducted through transcription-mediated amplification as previously described.9,10 Serum specimens for Phi measurement were collected per NCI-EDRN standard protocols17 and analyzed for PSA, fPSA, and p2PSA on the Beckman Coulter Access 2 Immunoassay Analyzer. Phi was calculated as (p2PSA/fPSA) × PSA1/2. Phi assays were conducted by the NCI EDRN Biomarker Reference Laboratory at Johns Hopkins University.
Statistical Analysis
The primary outcome of interest was GG ≥ 2 PCa on initial prostate biopsy. Using the current training cohort, 2 sets of prediction algorithms for clinically significant PCa (GG ≥ 2) were then developed to combine (i) Phi and PCA3 and (ii) Phi, PCA3, and T2:ERG. Three approaches were adopted for each combination: (i) either-or logic of the biomarkers exceeding their respective thresholds; (ii) sequential testing of blood (for Phi) and urine (for PCA3 and T2:ERG) with the allowance for the declaration of result based on the first sample only, whereas the second sample is only necessary if the first is in the gray zone; and (iii) linear combination through logistic regression. To elaborate on approach (ii), first-blood-then-urine and first-urine-then-blood rules were developed. In the former case, Phi has 2 thresholds and when the value is in the gray zone (ie, between the thresholds), the prediction is based on either-or logic of PCA3 and T2:ERG exceeding their respective thresholds. Conversely, in the case of first-urine-then-blood rule, PCA3 and T2:ERG each has 2 thresholds, and the declaration of result can be based on the urine sample only if either urine biomarker exceeds its upper threshold (positive result) or both urine biomarkers fall below their respective lower thresholds (negative result); further testing on Phi becomes only necessary otherwise (ie, the gray zone). With approaches (i) and (ii), the combination algorithms were developed to empirically optimize specificity at 95% sensitivity for GG ≥ 2 PCa. Meanwhile, with the third approach, the biomarkers were combined through logistic regression and then the threshold for the estimated risk score was calculated to result in empirical 95% sensitivity. Favorable multiplex prediction rules determined from the training study, as judged by empirical specificity at 95% sensitivity, were locked down. The algorithms developed from the third approach had lower specificity at 95% sensitivity in the training study, and they were not selected for evaluation in the validation study. The validation cohort was then used to test our primary hypothesis that combining serum Phi with urinary PCA3 improves GG ≥ 2 PCa detection compared with Phi alone. Our secondary hypothesis was that incorporating serum Phi with urinary PCA3/T2:ERG could further improve detection of clinically significant disease and reduce otherwise unnecessary biopsies compared with urinary PCA3/T2:ERG alone. Of note, the proportion of GG ≥ 2 cancers missed based on the testing approach is 100% − sensitivity. Furthermore, specificity represents the proportion of unnecessary biopsies (benign/GG1 pathology) that would have been avoided using a given test. Testing algorithms were compared in weighted sum of specificity and sensitivity with weights specified a priori, and P values were obtained through bootstrap in the validation cohort. These weights were computed such that the developed algorithms approximately maximized the weighted sum of specificity and sensitivity empirically in the training cohort. Analyses were performed using R version 4.1.1 (The R Foundation), and P < .05 was considered statistically significant.
RESULTS
Developing Models to Combine Serum Phi and Urinary PCA3-T2:ERG
The training cohort consisted of 512 patients (Table 1) with median (IQR) for PSA = 4.8 ng/mL (3.6-6.5), Phi = 34.6 (25.8-47.4), PCA3 score = 24.7 (11.7-54.5), and T2:ERG score = 6.3 (0.1-37.9). Using the training cohort, Phi, PCA3, and combination algorithms were developed and optimized for specificity (ie, proportion of unnecessary biopsies) at 95% sensitivity for GG ≥ 2 PCa (Table 2): When evaluating the biomarkers individually, a PCA3 > 7.7 coincided with specificity of 17%, Phi > 23.2 with specificity of 27%, and PCA3 + Phi (PCA3 > 28.7 or Phi > 30.9, ie, testing was considered positive if either test exceeded its threshold) with specificity of 25%. These established thresholds were defined as the primary rules later used to address our primary hypothesis in the validation cohort (vide infra).
Table 1.
Subject Characteristics and Distribution of Phi, Prostate Cancer Antigen 3, and T2:ERG in the Training Cohort
| Participant characteristics | Developmental cohort | |||
| N (%) | Phi valuea | PCA3 scorea | T2:ERG scorea | |
| Age in yrb,c | ||||
| <65 | 306 (60) | 33.2 (24.6-44.7) | 18.5 (9.7-42.9) | 6.1 (0.1-35.4) |
| ≥65 | 206 (40) | 36.5 (26.9-53.1) | 39.0 (16.9-75.2) | 6.4 (0.1-47.0) |
| Raceb,c | ||||
| White | 444 (87) | 34.4 (25.2-46.9) | 23.8 (11.4-52.7) | 7.2 (0.2-41.7) |
| Black | 50 (9.8) | 39.0 (27.0-63.8) | 47.4 (17.4-76.8) | 2.8 (0.0-30.0) |
| Other | 18 (3.5) | 32.6 (24.4-35.9) | 16.1 (9.5-31.5) | 2.4 (0.0-16.4) |
| Hispanic/Latino ethnicity | ||||
| Yes | 26 (5.1) | 33.1 (25.2-52.0) | 23.6 (11.6-63.5) | 8.5 (0.0-93.1) |
| No | 486 (95) | 34.7 (25.8-47.3) | 25.0 (11.8-54.1) | 6.1 (0.1-36.9) |
| DRE results | ||||
| Normal | 144 (28) | 34.5 (26.3-47.0) | 23.0 (11.4-54.5) | 6.0 (0.3-30.5) |
| Enlarged/benign | 285 (56) | 34.1 (25.7-44.9) | 22.9 (11.1-53.2) | 4.7 (0.1-37.1) |
| Abnormal/suspect | 79 (16) | 38.9 (22.4-62.5) | 28.5 (14.2-59.1) | 13.0 (0.8-59.2) |
| Biopsy findingb,c,d | ||||
| No PCa | 249 (49) | 29.0 (21.8-37.7) | 14.4 (8.6-27.8) | 1.6 (0.0-16.0) |
| PCa-GS6 | 108 (21) | 34.5 (25.7-44.9) | 40.4 (19.3-66.8) | 18.0 (2.0-62.0) |
| PCa-GS7+ | 155 (30) | 48.5 (37.6-69.6) | 45.0 (20.7-78.5) | 21.2 (0.5-72.2) |
DRE, digital rectal examination; GS, Gleason score; PCa, prostate cancer; PCA3, prostate cancer antigen 3; Phi, prostate health index; T2:ERG, TMPRSS2:ERG.
Median (IQR) ^ 4 participants refused DRE.
Statistically significant difference for Phi values with nonparametric tests.
Statistically significant difference for PCA3 values with nonparametric tests.
Statistically significant difference for T2:ERG values with nonparametric tests.
Table 2.
Algorithm Training and Performance From Developmental Cohort (N = 512 Overall; N = 155 [30%] With Grade Group ≥ 2 Prostate Cancer)
| Biomarker | Rule | Sensitivity (%) | Specificity (%) | NPV (%) | PPV (%) |
| Phi | Phi > 23.2 | 94.8 | 26.9 | 92.3 | 36.0 |
| PCA3 | PCA3 > 7.7 | 94.8 | 17.4 | 88.5 | 33.3 |
| PCA3 + Phi | PCA3 > 28.7 or Phi > 30.9 | 97.4 | 35.3 | 96.9 | 39.5 |
| PCA3 + T2:ERG | PCA3 > 19.1 or T2:ERG > 7.6 | 92.3 | 35.6 | 91.4 | 38.4 |
| PCA3 + T2:ERG + Phi | PCA3 > 34.1 or T2:ERG > 21.0 or Phi > 36.8 | 97.4 | 40.6 | 97.3 | 41.6 |
| First blood, then urine | Phi > 39.6 or Phi ≥ 16.0 + (PCA3 > 34.1 or T2:ERG > 21.0) | 96.8 | 45.1 | 97.0 | 43.3 |
| First urine, then blood | PCA3 > 34.1 or Phi > 29.2 + (PCA3 ≥ 9.5 or T2:ERG ≥ 27.7) | 95.5 | 46.8 | 96.0 | 43.8 |
NPV, negative predictive value; PCA3, prostate cancer antigen 3; Phi, prostate health index; PPV, positive predictive value; T2:ERG, TMPRSS2:ERG.
We next incorporated urinary T2:ERG RNA scores into the models combining serum Phi and urinary RNA markers and again calibrated for 95% sensitivity for GG ≥ 2 PCa (Table 2). Consistent with our previous study that evaluated only urinary biomarkers in this training cohort,10 combining PCA3 and T2:ERG urinary RNA (wherein the threshold for optimal specificity at 95% sensitivity for PCA3 was 19.1 and for T2:ERG was 7.6) led to 36% specificity at 95% sensitivity. Simultaneous combination of serum Phi together with PCA3 + T2:ERG urinary RNA (wherein the threshold for optimal specificity at 95% sensitivity for PCA3 was 34.1, for T2:ERG was 21.0 and for Phi was 36.8) improved the specificity in predicting GG ≥ 2 PCa to 41%.
Next, we developed models to sequentially combine serum Phi and urinary PCA3 + T2:Erg RNA testing, reflecting that in the clinical setting, urine tests could either precede serum Phi testing (ie, sufficiently favorable urine test could exclude biopsy, sufficiently adverse result could trigger biopsy, and intermediate grey zone results would lead to serum Phi testing to resolve decision-making about biopsy) or vice versa (Table 2). Evaluating serum Phi first (which would be followed by urine PCA3 + T2:ERG only if serum Phi results were in a “grey zone”), when calibrated to 95% sensitivity yielded a specificity of 45% for predicting GG ≥ 2 PCa. Conversely, starting with urinary PCA3 + T2:ERG RNA testing (followed by serum Phi to resolve biopsy decisions only if the urinary RNA test results were in an inconclusive “grey zone”), when calibrated to 95% sensitivity resulted in specificity of 47% for GG ≥ 2 PCa (Table 2).
Validation of Models Combining Serum Phi and Urinary PCA3-T2:ERG
We validated these models in a cohort of 561 biopsy-naïve patients (Table 3) from the prospective, multicenter NCI-EDRN urinary PCA3 trial.15 The median (IQR) for PSA = 5.0 ng/mL (3.8-6.8); Phi = 40.8 (29.9-54.8); PCA3 score = 26.3 (11.4-59.5); and T2:ERG score = 5.1 (0.3-33.6). PCA3 > 7.7 had 95% sensitivity and 21% specificity, Phi > 23.2 had 99% sensitivity and 14% specificity, and using Phi > 30.9 or PCA3 > 28.7 had 99% sensitivity and 24% specificity (Table 4). Once again, algorithms were compared in weighted sum of specificity/sensitivity (specified a priori) with P-values obtained through bootstrap, and weights were computed such that the developed algorithms maximized the weighted sum of specificity/sensitivity in the training cohort.
Table 3.
Characteristics of Participants and Distribution of Phi, Prostate Cancer Antigen 3, and T2:ERG in the Validation Cohort
| Participant characteristics | Validation cohort | |||
| N (%) | Phi valuea | PCA3 scorea | T2:ERG scorea | |
| Age in yrb | ||||
| <65 | 365 (65) | 40.0 (29.9-53.3) | 21.2 (9.5-50.8) | 5.1 (0.3-25.8) |
| ≥65 | 196 (35) | 43.6 (29.9-56.8) | 36.4 (18.1-74.1) | 5.2 (0.1-42.0) |
| Racec,d | ||||
| White | 441 (79) | 40.4 (29.6-53.4) | 24.1 (10.4-57.8) | 5.8 (0.4-37.5) |
| Black | 79 (14) | 46.1 (35.9-59.4) | 34.3 (16.0-65.5) | 1.6 (0.0-11.8) |
| Other | 41 (7.3) | 38.1 (28.3-54.7) | 28.0 (13.5-91.6) | 5.0 (0.5-24.7) |
| Hispanic/Latino ethnicityc | ||||
| Yes | 43 (7.7) | 35.3 (24.9-42.5) | 20.4 (12.2-50.4) | 6.1 (0.6-28.3) |
| No | 514 (92) | 41.4 (30.3-55.4) | 27.4 (11.1-60.6) | 5.2 (0.3-35.6) |
| DRE results | ||||
| Normal | 263 (47) | 40.3 (30.7-51.4) | 24.8 (10.5-59.4) | 5.0 (0.3-27.5) |
| Enlarged/benign | 167 (30) | 40.5 (30.5-52.9) | 23.8 (12.4-56.5) | 4.2 (0.2-22.9) |
| Abnormal/suspect | 111 (20) | 42.1 (27.6-68.1) | 27.1 (14.7-64.7) | 6.1 (0.4-81.8) |
| Biopsy findingc,b,d | ||||
| No PCa | 297 (53) | 34.3 (25.2-45.0) | 16.7 (7.1-32.0) | 2.4 (0.0-10.2) |
| PCa-GS6 | 116 (21) | 41.2 (31.2-49.9) | 52.0 (22.0-85.4) | 15.3 (1.9-64.5) |
| PCa-GS7+ | 148 (26) | 57.8 (45.7-88.5) | 48.1 (22.8-99.5) | 24.3 (1.5-102.0) |
DRE, digital rectal examination; GS, Gleason score; PCa, prostate cancer; PCA3, prostate cancer antigen 3; Phi, prostate health index; T2:ERG, TMPRSS2:ERG.
Median (IQR) ^ 20 participants refused DRE.
Statistically significant difference for PCA3 values with nonparametric tests.
Statistically significant difference for Phi values with nonparametric tests.
Statistically significant difference for T2:ERG values with nonparametric tests.
Table 4.
Performance of Algorithms in the Validation Cohort (N = 561 Overall; N = 148 [26%] With Grade Group ≥ 2 Prostate Cancer)
| Biomarker | Rule | Sensitivity (%) | Specificity (%) | NPV (%) | PPV (%) |
| Phi | Phi > 23.2 | 99.3 | 14.0 | 98.2 | 29.3 |
| PCA3 | PCA3 > 7.7 | 95.3 | 21.1 | 92.6 | 30.2 |
| Phi + PCA3 | Phi > 30.9 or PCA3 > 28.7 | 99.3 | 24.0 | 99.0 | 31.9 |
| PCA3 + T2:ERG | PCA3 > 19.1 or T2:ERG > 7.6 | 91.2 | 34.1 | 91.5 | 33.2 |
| PCA3 + T2:ERG + Phi | PCA3 > 34.1 or T2:ERG > 21.0 or Phi > 36.8 | 98.6 | 28.8 | 98.3 | 33.2 |
| First blood, then urine | Phi > 39.6 or Phi ≥ 16.0 + (PCA3 > 34.1 or T2:ERG > 21.0) | 98.6 | 34.9 | 98.6 | 35.2 |
| First urine, then blood | PCA3 > 34.1 or (PCA3 ≥ 9.5 or T2:ERG ≥ 27.7) + Phi > 29.2 | 93.9 | 38.5 | 94.6 | 35.4 |
NPV, negative predictive value; PCA3, prostate cancer antigen 3; Phi, prostate health index; PPV, positive predictive value; T2:ERG, TMPRSS2:ERG.
Our primary, prespecified hypothesis comparing sensitivity weighted/specificity-weighted averages of serum Phi vs serum Phi with urinary PCA3 for detection of GG ≥ 2 PCa was validated, as combining urinary PCA3 and serum Phi significantly outperformed Phi alone in prediction of GG ≥ 2 PCa (P = .002). Meanwhile, the combination of urinary PCA3 and serum Phi also significantly outperformed PCA3 alone in prediction of GG ≥ 2 PCa (P = .019).
We then performed secondary validation analyses combining urinary PCA3/T2:ERG with serum Phi for detecting GG ≥ 2 disease (Table 4). Serum Phi and urine PCA3/T2:ERG tested simultaneously outperformed urine PCA3/T2:ERG alone (P = .04). In addition, we analyzed sensitivity/specificity weighted averages specific to testing sequence (ie, serum Phi first which then informs subsequent urinary PCA3/T2:ERG testing and vice versa). Serum Phi first to guide subsequent urine PCA3/T2:ERG testing significantly outperformed urinary PCA3/T2:ERG (P = .002). Similarly, urine PCA3/T2:ERG first to guide subsequent serum Phi testing was also superior to urinary PCA3/T2:ERG alone (P = .03). Overall, combining serum Phi with urinary PCA3/T2ERG testing significantly outperformed urinary PCA3/T2:ERG alone for detecting GG ≥ 2 PCa in each of the 3 algorithms reflecting different potential clinical scenarios.
DISCUSSION
In biopsy-naïve men referred for prostate biopsy, Phi, PCA3, and T2:ERG scores have individually been shown to significantly improve detection of GG ≥ 2 PCa compared with PSA and PSA-based clinical prediction models.6,7,10,18 This study used data from 2 large, multi-institutional cohorts to develop and validate serum and urinary assay cutoffs, both individually and across different clinically relevant scenarios, to better identify GG ≥ 2 disease in men being considered for initial prostate biopsy. In the validation cohort, serum Phi combined with urinary PCA3 significantly outperformed Phi alone (P = .002), thus supporting our primary hypothesis that urine RNA testing combined with serum Phi testing can outperform either performed without the other. In the validation cohort, combining serum Phi and urine PCA3 testing would have safely avoided 24% of unnecessary biopsies while missing just 1 instance (0.7%) of GG ≥ 2 disease. Comparatively, serum Phi alone would have missed the same number of GG ≥ 2 cancers, however forgoing only 14% of unnecessary biopsies.
Moreover, when validating our secondary aims, we found that combining serum Phi with urinary PCA3-T2:ERG RNA testing significantly outperformed urinary PCA3-T2:ERG alone in all 3 proposed clinical workflows (simultaneously, first serum then urine, and first urine then serum). These 3 combinations of serum Phi and urinary PCA3-T2:ERG RNA testing would have avoided 29% to 39% of unnecessary biopsies while missing 1% to 6% of GG ≥ 2 cancers, whereas urinary PCA3-T2:ERG testing alone would have missed more GG ≥ 2 cancers (9%) in this range of specificity. Regarding overall clinical feasibility, simultaneous testing with serum Phi and urine PCA3-T2:ERG demonstrated equal/improved sensitivity than the other 2 proposed clinical workflows. Overall, these data highlight the potential role of sequentially incorporating serum and urine biomarkers to aid in patient selection for prostate biopsy.
Overdiagnosis and overtreatment of indolent disease remains a prominent issue with current PSA-based testing3,19,20 despite countless ongoing efforts aimed at addressing this. Minimally invasive methods such as advanced imaging and biomarkers have shown promise in this domain. Amidst commercially available biomarkers, Phi has been shown to outperform its individual molecular components21 and to accurately discriminate between GG1/benign and GG ≥ 2 pathology on biopsy.5,6 Two previous single-institution cohorts evaluated combining serum Phi and urinary PCA3 RNA testing to predict PCa,14,22 and each found that combining serum Phi and urinary PCA3 performed better at predicting PCa on biopsy than either biomarker alone. However, neither of these previous reports validated models predicting GG ≥ 2 cancer at initial biopsy (this being the target clinical endpoint in our study), and neither evaluated the incorporation of urinary T2:ERG RNA testing as we have incorporated herein. Furthermore, our approach of focusing on predictive performance at high sensitivity is particularly germane in developing screening/risk-stratification tools due to the responsibility to limit missing clinically significant cancers (ie, proportion of GG ≥ 2 cancers missed = 100% − sensitivity).4 Thus, this study underscores the value of combining urinary and serum biomarkers to improve performance and ultimately patient risk stratification.
Interestingly, there remains a paucity of validated tests incorporating both serum and urinary markers. One exception is the MyProstateScore (LynxDx, Inc.), a validated test consisting of urine PCA3 and T2:ERG normalized to serum PSA to render a single score.9 A threshold of 10 led to 97% sensitivity and 98% negative predictive value for GG ≥ 2 disease while preventing 33% of unnecessary biopsies and missing 3.0% of clinically significant cancers.16 Our stepwise clinical algorithm encompasses a different approach to the clinical conundrum of PCa overdiagnosis/overtreatment, but with strikingly similar results. In patients with either Phi > 39.6 or Phi 16 to 39.6 and then proceeding with using PCA3 > 34.1 or T2:ERG > 21.0 to confer a positive test result, we found 99% sensitivity for GG ≥ 2 PCa. This method would have avoided 35% of unnecessary biopsies while missing just 1% of GG ≥ 2 cancers. Such promising metrics further suggest that our understanding of sequential biomarker testing remains to be elucidated. Given confines of retrospective data, it would not be prudent to make clinical recommendations beyond our findings. However, it does raise the exciting possibility of combining existing, validated, liquid biomarkers as a pragmatic approach to risk stratifying patients referred for initial prostate biopsy.
This study does have limitations. For one, training and validation cohort biopsy specimens were obtained using systematic templates without targeted/fusion capabilities. Thus, patients with benign/GG1 pathology could have been harboring more aggressive disease more likely to go undetected.22 This is important to keep in mind when interpreting test performance given present-day advancements in biopsy/imaging techniques.23 In addition, PCA3 and T2:ERG are not commercially available as individual tests; however, available tests that incorporate both markers include MyProstateScore and ExoDx Prostate Intelliscore, thus supporting their relevance to contemporary practice. Moreover, our available data did not allow for investigating a role for biomarker use in combination with more widely available imaging modalities such as MRI and PSMA PET. The lack of MRI data in this study is certainly a limitation; however, this study may offer benefit and scientific insight despite the lack of MRI data. Similar to one study of MyProstateScore,24 it would be interesting to ascertain the role of sequentially incorporating serum/urine biomarkers within the context of equivocal (PI-RADS 3) MRI findings when selecting for prostate biopsy. Such studies would also corroborate available data supporting the use of Phi and MRI, which was shown to outperform Phi and MRI individually.25 Although not completely representative of contemporary workflow, MRI capabilities remain limited in rural practice and use of MRI was not the standard at the time this study. Ascertaining such a relationship, particularly in the case of Equivocal (PI-RADS 3) lesions on MRI, warrants further investigation and could aid in patient selection for biopsy.
CONCLUSIONS
In validating the combined use of urinary RNA and serum Phi testing in a cohort from a multicenter, nationwide prospectively enrolled clinical trial, we verified our primary hypothesis that combining serum Phi with urinary PCA3 significantly improves selection of men for initial prostate biopsy compared with serum Phi alone. Moreover, we also validated that secondary models combining serum Phi testing with urinary PCA3 and T2:ERG RNA testing outperformed urinary PCA3/T2ERG for detecting GG ≥ 2 cancer across 3 different potential clinical workflows (simultaneous testing, first serum then urine, and first urine then serum). As such, a combination of serum and urinary biomarkers appears to be a feasible, pragmatic approach to rule out the need for prostate biopsy in men harboring negative or otherwise indolent disease.
FUNDING
NIH U01 CA113913 (MGS PI), NIH U24 CA115102 and U2CCA271895 (DWC PI), and National Cancer Institute (NCI) Early Detection Research Network. Phi reagents provided by Beckman Coulter, Inc.
CONFLICT OF INTEREST DISCLOSURES
The authors have no conflicts of interest to disclose.
ETHICS STATEMENT
The Emory IRB protocol Prostate Satellite Tissue Bank (IRB00045859) was used to facilitate storage and distribution of patient samples enrolled on EDRN U01 Harvard-Michigan-Cornell or training cohort. For validation, secondary deidentified data were used in collaboration with the EDRN PCA3 trial team.
DATA AVAILABILITY STATEMENT
The data sets generated during and/or analyzed during this study are available from the corresponding author on reasonable request.
AUTHOR CONTRIBUTIONS
Conceptualization: Zheng, Wei, Patil, Chinnaiyan, Tomlins, Lin, Sanda, Narayan. Formal analysis: Eyrich, Zheng, Wei, Narayan. Funding acquisition: Sanda, Narayan. Methodology: Eyrich, Chan, Patil, Wang, Narayan. Data curation: Huang, Zheng, Sokoll, Chan, Patil, Wang. Formal analysis: Huang, Sokoll, Chan, Patil. Investigation: Huang, Zheng, Scherr, Bidair, Sanda, Narayan. Supervision: Patil, Chinnaiyan, Tomlins, Lotan, Lin, Scherr, Kibel, Taneja, Thompson, Sanda, Narayan. Validation: Patil, Kibel, Taneja, Thompson, Sanda, Narayan. Writing—review & editing: Huang, Zheng, Wei, Sokoll, Chan, Patil, Chinnaiyan, Tomlins, Lotan, Lin, Scherr, Wang, Kibel, Taneja, Bidair, Thompson, Sanda, Narayan. Writing—original draft: Eyrich, Sanda, Narayan.
ACKNOWLEDGMENTS
The authors would like to acknowledge contributions from collaborators affiliated with Early Detection Research Network efforts represented in this work.
Footnotes
Presented in abstract form at the American Urological Association Annual Meeting, May 2024, San Antonio, TX.
Shared Senior Authorship.
Contributor Information
Nicholas W. Eyrich, Email: neyrich@emory.edu.
Yijian Huang, Email: yhuang5@emory.edu.
Yingye Zheng, Email: yzheng@fredhutch.org.
John Wei, Email: jtwei@umich.edu.
Lori Sokoll, Email: lsokoll@jhmi.edu.
Daniel W. Chan, Email: dchan@jhmi.edu.
Dattatraya Patil, Email: dattatraya.patil@emory.edu.
Arul Chinnaiyan, Email: arul@umich.edu.
Scott A. Tomlins, Email: tomlinss@med.umich.edu.
Yair Lotan, Email: yair.lotan@utsouthwestern.edu.
Daniel Lin, Email: dlin@uw.edu.
Douglas Scherr, Email: dss2001@med.cornell.edu.
Yanqing Wang, Email: ywang159@gsu.edu.
Adam Kibel, Email: akibel@bwh.harvard.edu.
Samir S. Taneja, Email: samir.taneja@nyumc.org.
Mohamed Bidair, Email: mobidair@sdclinicaltrials.com.
Ian M. Thompson, Email: ian.thompson2@christushealth.org.
Martin G. Sanda, Email: martinsanda@emory.edu.
REFERENCES
- 1.Hogenhout R Remmers S van Slooten-Midderigh ME, et al. ERSPC Rotterdam Study Group. From screening to mortality reduction: an overview of empirical data on the patient journey in European randomized study of screening for prostate cancer Rotterdam after 21 years of follow-up and a reflection on quality of life. Eur Urol Oncol. 2024;7(4):713-720. [DOI] [PubMed] [Google Scholar]
- 2.Frånlund M, Månsson M, Godtman RA, et al. Results from 22 years of followup in the Göteborg randomized population-based prostate cancer screening trial. J Urol. 2022;208(2):292-300. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Fenton JJ, Weyrich MS, Durbin S, Liu Y, Bang H, Melnikow J. Prostate-specific antigen-based screening for prostate cancer evidence report and systematic review for the US Preventive Services Task Force. JAMA. 2018;319(18):1914-1931. [DOI] [PubMed] [Google Scholar]
- 4.Eyrich NW, Morgan TM, Tosoian JJ. Biomarkers for detection of clinically significant prostate cancer: contemporary clinical data and future directions. Transl Androl Urol. 2021;10(7):3091-3103. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Loeb S, Catalona WJ. The prostate health index: a new test for the detection of prostate cancer. Ther Adv Urol. 2014;6(2):74-77. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.De La Calle C, Patil D, Wei JT, et al. Multicenter evaluation of the prostate health index to detect aggressive prostate cancer in biopsy Naïve men. J Urol. 2015;194(1):65-72. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Loeb S, Sanda MG, Broyles DL, et al. The prostate health index selectively identifies clinically significant prostate cancer. J Urol. 2015;193(4):1163-1169. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.White J, Shenoy BV, Tutrone RF, et al. Clinical utility of the prostate health index (Phi) for biopsy decision management in a large group urology practice setting. Prostate Cancer Prostatic Dis. 2018;21(1):78-84. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Tomlins SA, Day JR, Lonigro RJ, et al. Urine TMPRSS2:ERG plus PCA3 for individualized prostate cancer risk assessment. Eur Urol. 2016;70(1):45-53. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Sanda MG, Feng Z, Howard DH, et al. ; and the EDRN-PCA3 Study Group. Association between combined TMPRSS2:ERG and PCA3 RNA urinary testing and detection of aggressive prostate cancer. JAMA Oncol. 2017;3(8):1085-1093. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Van Neste L, Hendriks RJ, Dijkstra S, et al. Detection of high-grade prostate cancer using a urinary molecular biomarker-based risk score. Eur Urol. 2016;70(5):740-748. [DOI] [PubMed] [Google Scholar]
- 12.Leyten GH, Hessels D, Jannink SA, et al. Prospective multicentre evaluation of PCA3 and TMPRSS2-ERG gene fusions as diagnostic and prognostic urinary biomarkers for prostate cancer. Eur Urol. 2014;65(3):534-542. [DOI] [PubMed] [Google Scholar]
- 13.McKiernan J, Donovan MJ, O’Neill V, et al. A novel urine exosome gene expression assay to predict high-grade prostate cancer at initial biopsy. JAMA Oncol. 2016;2(7):882-889. [DOI] [PubMed] [Google Scholar]
- 14.Ferro M, Bruzzese D, Perdona S, et al. Prostate health index (Phi) and prostate cancer antigen 3 (PCA3) significantly improve prostate cancer detection at initial biopsy in a total PSA range of 2-10 ng/ml. PLoS One. 2013;8(7):e67687. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Wei JT, Feng Z, Partin AW, et al. Can urinary PCA3 supplement PSA in the early detection of prostate cancer?. J Clin Oncol. 2014;32(36):4066-4072. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Tosoian JJ, Trock BJ, Morgan TM, et al. Use of the MyProstateScore Test to rule out clinically significant cancer: validation of a straightforward clinical testing approach. J Urol. 2021;205(3):732-739. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Tuck MK, Chan DW, Chia D, et al. Standard operating procedures for serum and plasma collection: early detection research network consensus statement standard operating procedure integration working group. J Proteome Res. 2009;8(1):113-117. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Chiu PK, Ng CF, Semjonow A, et al. A multicentre evaluation of the role of the prostate health index (PHI) in regions with differing prevalence of prostate cancer: adjustment of PHI reference ranges is needed for European and Asian settings. Eur Urol. 2019;75(4):558-561. [DOI] [PubMed] [Google Scholar]
- 19.Loeb S, Bjurlin MA, Nicholson J, et al. Overdiagnosis and overtreatment of prostate cancer. Eur Urol. 2014;65(6):1046-1055. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Paschen U, Sturtz S, Fleer D, Lampert U, Skoetz N, Dahm P. Assessment of prostate-specific antigen screening: an evidence-based report by the German Institute for Quality and Efficiency in Health Care. BJU Int. 2022;129(3):280-289. [DOI] [PubMed] [Google Scholar]
- 21.Catalona WJ, Partin AW, Sanda MG, et al. A multicenter study of [-2]pro-prostate specific antigen combined with prostate specific antigen and free prostate specific antigen for prostate cancer detection in the 2.0 to 10.0 ng/ml prostate specific antigen range. J Urol. 2011;185(5):1650-1655. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Walz J, Graefen M, Chun FK-H, et al. High incidence of prostate cancer detected by saturation biopsy after previous negative biopsy series. Eur Urol. 2006;50(3):498-505. [DOI] [PubMed] [Google Scholar]
- 23.Ahdoot M, Wilbur AR, Reese SE, et al. MRI-targeted, systematic, and combined biopsy for prostate cancer diagnosis. N Engl J Med. 2020;382(10):917-928. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Zhu M, Fu Q, Zang Y, et al. Different diagnostic strategies combining prostate health index and magnetic resonance imaging for predicting prostate cancer: a multicentre study. Urol Oncol. 2024;42(5):159.e17-159.e23. [DOI] [PubMed] [Google Scholar]
- 25.Tosoian JJ, Singhal U, Davenport MS, et al. Urinary MyProstateScore (MPS) to rule out clinically-significant cancer in men with equivocal (PI-RADS 3) multiparametric MRI: addressing an unmet clinical need. Urology. 2022;164:184-190. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The data sets generated during and/or analyzed during this study are available from the corresponding author on reasonable request.