Assessment of the ability of Decipher Prostate Genomic Classifier (DGC) >0.85 to identify patients who benefit from adding docetaxel (DOC) to androgen deprivation therapy (ADT) plus enzalutamide (ENZ): Level 1B evidence from the ENZAMET study.

5001 Background: DGC predicts overall survival (OS) benefit from DOC when added to ADT while the benefit of adding DOC to ADT and ENZ is unknown. We hypothesized in a locked prespecified statistical analysis plan that higher GC score (> 0.85, the locked threshold on clinical test report) will identify patients who benefit from addition of DOC to ADT plus ENZ independently of metastases volume and timing. Methods: DGC scores were generated from transcriptome profiling using a clinical test (Veracyte) on tumor index cores from participants (pts) randomised on ENZAMET study (N = 1,125) 1:1 ADT with NSAA or ENZ with clinical discretion to add DOC as part of standard of care. Primary tumor samples were available from 764 pts. Differential DOC OS benefit was assessed by testing the marker-by-treatment interaction term in propensity score-weighted Cox models, with weights based on factors associated with planned DOC use. Prognostic effects of DGC were estimated using Cox for age, WHO PS, Gleason, randomized arm +/- planned DOC, metastases volume and timing, with adjusted hazard ratios (aHR) and 95% CIs. Analyses were prespecified and conducted independently by two statisticians. Results: DGC scores were available for 634 (83%) unique pts (median follow-up 5.6 years) with clinical factors representative of the overall trial. Median age was 68 yrs with 50% HV, 62% synchronous presentation, and 44% with planned DOC. Planned DOC was associated with younger age, high volume and synchronous presentation (all p≤0.002). Median DGC score was 0.88 (IQR 0.75-0.96) and 55% had DGC > 0.85 (62% in pts with DOC planned, 49% in others). Overall, higher GC was significantly associated with poorer OS (aHR 1.37 [1.06-1.78], p = 0.02). DOC benefit differed by GC level [higher vs lower] in pts treated with ADT plus ENZ (p-interaction 0.043). Higher DGC was associated with worse OS in pts treated with ADT plus ENZ (aHR 2.31 [95% CI 1.26-4.21], p = 0.007) while pts treated with DOC added to ADT plus ENZ showed no significant difference by DGC (aHR 1.08 [0.63–1.86]). Effects were more pronounced in pts with high vs. low volume disease. Conclusions: DGC > 0.85 is predictive of benefit from adding DOC to ADT plus ENZ as it negated the poor prognostic outcome of DGC > 0.85 with ADT plus ENZ alone, whereas there was no evidence of benefit for adding DOC for pts with GC ≤0.85. 5yr OS N Lower Decipher Higher Decipher Higher vs. loweradj. HR (95% CI) Overall 634 71%(66-77%) 57%(51-62%) 1.37(1.1, 1.8) ADT+ENZ 178 86%(78-93%) 62%(52-72%) 2.29(1.3, 4.2) ADT+ENZ+DOC 142 66%(53-79%) 58%(48-68%) 1.08(0.6, 1.9) ADT+ENZ (LV) 123 88%(80-96%) 72%(61-84%) 1.96(0.9, 4.3) ADT+ENZ+DOC (LV) 44 86%(67-100%) 79%(64-94%) 1.70(0.3, 9.0) ADT+ENZ (HV) 55 81%(65-96%) 41%(23-60%) 2.83(1.1, 7.5) ADT+ENZ+DOC (HV) 98 58%(42-75%) 48%(36-61%) 1.07(0.6, 1.9)