Validation of the decipher genomic classifier (GC) in SAKK 09/10: A phase III randomized trial of dose-escalated salvage radiotherapy (SRT) after radical prostatectomy (RP).

Abstract

5010 Background: GC has been shown to independently prognosticate outcomes in prostate cancer. Herein, we validate the GC in a European randomized phase III trial of dose escalated SRT after RP. Methods: SAKK 09/10 (NCT01272050) randomized 350 patients with biochemical recurrence after RP to 64Gy vs 70Gy. No patients received androgen deprivation therapy (ADT) or pelvic nodal radiotherapy. A pre-specified statistical plan was developed to assess the impact of the GC on clinical outcomes. RP samples were centrally reviewed for the highest-grade tumor and those passing quality control (QC) were run on a clinical-grade whole-transcriptome assay to obtain the GC score (0 to 1; < 0.45, 0.45-0.6, > 0.6 for low-, intermediate-, and high, respectively). The primary aim of this study was to validate the GC for the prediction of freedom from biochemical progression (FFBP) using Cox multivariable analysis (MVA) adjusting for age, T-category, Gleason score, persistent PSA after RP, PSA at randomization, and randomization arm. The secondary aims were to evaluate the association of GC with clinical progression-free survival (CPFS) and use of salvage ADT. Results: Of 233 patients with tissue available, 226 passed QC and were included for analysis. The final GC cohort was a representative sample of the overall cohort, with a median follow-up of 6.3 years (IQR 6.0-7.2). GC score (continuous per 0.1 unit, score 0-1) was independently associated with FFBP (HR 1.14 [95% CI 1.03-1.25], p = 0.009). Higher GC scores were independently associated with CPFS, use of salvage ADT, and rapid biochemical failure ( < 18 months after SRT). High- vs. low/intermediate-GC showed a HR of 2.22 ([95% CI 1.37-3.58], p = 0.001) for FFBP, 2.29 ([95% CI 1.32-3.98], p = 0.003) for CPFS, and 2.99 ([95% CI 1.50-5.95], p = 0.002) for use of salvage ADT. Patients with high-GC had 5-year FFBP of 45% [95% CI 32-59] vs 71% [95% CI 64-78] in low-intermediate GC. Similar estimates for GC risk groups were observed in the 64Gy vs 70Gy in GC high (5-year FFBP of 51% [95% CI 32-70] vs 39% [95% CI 20-59]) and in low-intermediate GC (75% [95% CI 65-84] vs 69% [95% CI 59-78]). Conclusions: This study represents the first contemporary randomized controlled trial in patients with recurrent prostate cancer treated with early SRT without ADT that has validated the prognostic utility of the GC. Independent of standard clinicopathologic variables and radiotherapy dose, patients with a high-GC were more than twice as likely than a lower GC score to experience biochemical and clinical progression and receive salvage ADT. This data confirms the clinical value of Decipher GC for tailoring treatment in the postoperative salvage setting.