Prediction of early prostate cancer recurrence using a liquid biopsy approach.

Abstract

386 Background: A critical clinical concern after radical prostatectomy for prostate cancer (PCa) is the timely identification of residual disease. Recurrent disease (biochemical recurrence: BCR) develops in approximately 30% of radical prostatectomies within 5 years of surgery. Currently, clinicopathological variables, including pathological tumor stage (pT-stage), Gleason score and PSA, or algorithmic combinatorial calculations (e.g., CAPRA-S) are used to predict BCR. Early and objective prediction of individuals at high risk of BCR would enable stratification of follow-up strategies and facilitate therapeutic therapy. To achieve these goals, we developed a liquid biopsy, the PROSTest, to identify PCa. This is a 27 multigene algorithmic signature with a high sensitivity and specificity (>90%) for PCa detection. We investigated if the PROSTest had utility as a predictive biomarker for BCR. Methods: Prospective recruitment of 60 PCa for radical prostatectomy with assessment of standard pathological, clinical and biomarker (PSA) data. D’Amico Risk scores and CAPRA-S were calculated. Blood was collected for PROSTest measurement pre-surgery. Target genes were amplified using qPCR and scored (0-100) using algorithmic analysis. Pre-surgical PROSTest scores were evaluated as predictors of BCR and compared with standard criteria as well as DR and CAPRA-S scores. Data was evaluated using Mann-Whitney U-test, multiple regression analyses, Kaplan-Meier survival analysis and Cox-proportional modeling. All data: median (range). Results: Consent was obtained in 48 (80%) patients. Median age (range) was 64 (50-82). Gleason was predominantly 7 (85%; 26: 7A, 15: 7B); TNM was primarily T2c (48%) and T3a (32%) with nodal disease evident in 8% and 0% cM1 disease. Resections were R0 (85%) and 7 R1. The median follow-up was 42 days (range: 14-782). Early BCR occurred in 8 (17%) patients. This included 3/7 (43%) of R1 and 5/41 (12%) R0 resections. PSMA imaging confirmed 3 LN recurrences and new visceral ( n=1) and bone ( n=1) disease. D’Amico Risk scores were mostly “high” (88% with risk score ≥50%) and were not associated with early BCR. CAPRA-S scores were higher in those who developed early BCR (5: 1-9) than in those who did not (2: 0-5). Pre-surgical PROSTest scores were elevated in all (median 59: 15-81). Multiple regression analysis identified only PROSTest score ≥60 and nodal status were associated with BCR. The median Recurrence Free Survival (mRFS) was 89 days compared to undefined in those with baseline PROSTest scores ≥60 (HR: 9.7; 95%CI: 2.16-43.7; p=0.003). No recurrences were identified in those with scores <60. Conclusions: Early biochemical recurrence (within 3 months of surgery) can be accurately predicted by elevated (≥60) pre-surgical PROSTest blood gene expression scores. This suggests the marker could be used as a stratification tool for neoadjuvant therapy, or to guide the frequency of monitoring during follow-up.