Presurgical Biomarker Performance in the Detection of Gleason Upgrading in Prostate Cancer

Current practice during active surveillance (AS) is to monitor low-risk patients by repeating biopsies periodically. Patients with an initial biopsy Gleason Score (GS) of 6 are offered definitive treatment if a repeat biopsy during surveillance yields a GS ≥ 7. This suggests that the cancer may have progressed during the interval from the previous biopsy. Even so, a sizable proportion of GS 7 patients may have been misclassified GS 6 due to sampling error on low-volume tumors (1, 2). For this reason, a noninvasive test that could detect this relatively subtle change in Gleason sum would be of value.

Results with post-DRE urine (3, 4), have consistently mirrored those with expressed prostatic secretion (EPS; refs. 5, 6), suggesting that either specimen can be used in the detection and classification of prostate cancer. In previous work, improvements in AUC values of ≥0.1 (3–5) have been obtained for the prediction that cancer will be found at biopsy using these specimens. In this study, we sought to assess the performance of biomarkers in predicting Gleason upgrading in cohorts of men who underwent radical prostatectomy (RP). Our sample sizes were chosen for a statistical power of greater than 90% for an AUC improvement over baseline of ≥0.1.

EPS specimens were obtained by milking the urethra after an attentive DRE before the patients underwent a Robot-Assisted Radical Prostatectomy (RARP). Of 646 men who consented under an Institutional Review Board–approved protocol for the collection and evaluation of biomarkers in EPS, 642 had complete clinical data, 224 met the 2012 National Comprehensive Cancer Network (NCCN) criteria, and 206 met the 2014 NCCN criteria for active surveillance based on their preoperative characteristics.

EPS collection and processing were as described previously (5, 6). Standard clinicopathologic data (age, Gleason sum, ethnicity, presurgical PSA, T stage, DRE, and race), direct measurements of EPS volume, total recovered RNA, and quantitative reverse transcription PCR values for TMPRSS2:ERG, PCA3, and PSA RNAs were obtained.

ROC analysis of biomarker performance was carried out in an attempt to predict upgrading. Statistical power for AUC differences attributable to biomarkers was estimated with PASS software. Minimal models for each cohort emerged from stepwise logistic regression analysis on six clinical and ten laboratory variables (Table 1).

For the 2012 NCCN AS cohort EPS volume, total RNA and RNA expression biomarkers did not significantly increase the AUC value beyond the 0.6969 AUC achieved by clinical variables for ethnicity and serum PSA in the prediction of upgrading postsurgery (Fig. 1A). For the 2014 NCCN AS cohort, these same biomarkers failed to significantly enhance the AUC value beyond the 0.7222 value achieved with age, ethnicity, and serum PSA (Fig. 1B). For the 2012 cohort with complete clinical and laboratory data and sample sizes (125 not upgraded and 99 upgraded), the power to detect an increase in the AUC of 0.05 and 0.10 was approximately 40% and 90%, respectively. For the 2014 cohort with 206 patients with complete clinical and laboratory data, the sample sizes (112 not upgraded and 94 upgraded), the power to detect an increase in the AUC of 0.05 and 0.10 is also approximately 40% and 90%, respectively.

For NCCN AS cohorts, the AUC values achieved with standard clinical markers remained unchanged by the inclusion of biomarkers that have been successful in the prediction of biopsy outcome and upstaging. Although an increase in AUC value of 0.05 due to these biomarkers cannot be excluded, it is unlikely that such an increase would warrant the clinical use of these biomarkers in counseling patients in this setting in spite of their success in other areas of treatment. Moreover, the general concordance between results obtained with EPS specimens and those obtained with urine samples obtained after an attentive DRE argues that the results obtained with EPS will mirror those obtained with post-DRE urine. We conclude that in this cohort of men, models for predicting upgrading are dominated by the standard clinicopathologic parameters with no significant contribution from current biomarkers of RNA expression or prostate function tested in EPS.

No potential conflicts of interest were disclosed.

Conception and design: K. Wittig, S.S. Smith

Development of methodology: K. Wittig, S.S. Smith

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): K. Wittig

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): K. Wittig, J. Yamzon, D.D. Smith, D.R. Jeske, S.S. Smith

Writing, review, and/or revision of the manuscript: K. Wittig, J. Yamzon, D.D. Smith, D.R. Jeske, S.S. Smith

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): K. Wittig, D.D. Smith, S.S. Smith

Study supervision: K. Wittig, S.S. Smith

This work was supported by NIH NCI grants CA 102521 and CA 136055 (to S.S. Smith).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.