Association between epigenomic biomarkers and baseline clinical characteristics in patients with mCRPC treated with rucaparib in TRITON2.

5086 Background: TRITON2 is a phase 2 study evaluating rucaparib in 277 patients with metastatic castration resistant prostate cancer (mCRPC). PSA has been used to predict radiographic progression free survival (rPFS), overall survival (OS) and monitor response, however, new approaches are needed to improve performance. Recently, it has been demonstrated that hypermethylated tumor DNA in the peripheral blood can be used to detect and monitor response and as a prognostic marker. Here, we evaluate the association between baseline methylation-based tumor fraction (TF) and clinical outcomes in patients with mCRPC. Methods: Pre-treatment plasma samples from patients participating in TRITON2 were sequenced using Guardant Infinity, a next-generation sequencing platform that evaluates genomics and epigenomics with ~15Mb of coverage for methylation-profiling and quantification. TF is calculated from thousands of cancer-type specific differentially hypermethylated regions. Patient outcomes were evaluated using the median-split baseline values for TF and PSA vs rPFS and OS by Cox proportional hazard model. Results: Two hundred thirty of 277 patients were eligible for baseline analysis and were successfully sequenced. Methylation-based TF was detected in 229/230 patients (99.6%). TF ranged from 0.02% to 99%, with a median of 25.1%. Baseline methylation-based TF was associated with Gleason score (p=0.012) but was only very weakly correlated with PSA levels (R 2 =0.09). Patients with baseline TF ≤ median demonstrated superior rPFS and OS vs those > median (HR=0.54, p=0.013; HR=0.42, p= 3.72e-07), respectively. In contrast, baseline PSA was only associated with superior OS (HR=0.52, p=1.31e-04), but not rPFS (HR=0.71, p=0.186). Conclusions: In patients from TRITON2, methylation-based TF appeared superior to PSA for tracking disease activity, both in terms of rPFS and OS. This suggests that methylation-based TF is a candidate disease monitoring tool that should be further investigated as a potential replacement for both radiological and PSA-based disease monitoring. Clinical trial information: NCT02952534 .