Evaluation of the aggressive variant prostate cancer molecular profile (AVPCm) in CLIA environments.

250 Background: 20-30% of prostate cancers respond poorly to androgen receptor (AR)-directed therapies and have an atypical, virulent course. The aggressive variant prostate cancers ( AVPC) encompass virulent androgen indifferent tumors and are characterized by combined defects in ≥ 2/3 of TP53, RB1, and PTEN (AVPC-m). Post-hoc analysis in a phase I/II trial indicated a benefit of carboplatin+cabazitaxel in men with AVPCm+ metastatic castrate resistant prostate cancer (mCRPC). We assess the feasibility of detecting the AVPCm in CLIA-certified environments. Methods: Men with progression of mCRPC undergoing biopsies for standard molecular profiling at MD Anderson Cancer Center (MDACC) were eligible. TP53, RB1, and PTEN expression was assessed by immunohistochemistry (IHC) and read by 2 pathologists independently. Next generation sequencing of solid tumor DNA (stDNA) and circulating tumor DNA (ctDNA) was performed on the MDACC Molecular Diagnostics Laboratory and the University of Washington OncoPlex ctDNA platforms respectively . Results: Of 64 eligible patients, 49 (77%), 37 (58%), and 54 (84%) had results reported for IHC, stDNA, and ctDNA respectively. Concordance of IHC reads between pathologists was variable: Kappa 0.78, 0.67, and 0.91 for TP53, RB1, and PTEN respectively. 36 (66.7%) ctDNA samples had low tumor content, such that false negatives could not be excluded. AVPC-m was detected in 14 (29%) of patients by IHC per the original reading pathologist, 2 (5%) by stDNA NGS, and 7 (39%) by ctDNA NGS. Median turnaround times for IHC, stDNA and ctDNA were 12 (2-35), 28 (12-56) and 16 (11-56) days respectively. Conclusions: We show the operational characteristics of the AVPCm detection using CLIA certified assays. IHC detected a greater proportion of patients with AVPCm+ tumors than stDNA, likely due, partly, to the challenges of copy number loss detection. ctDNA detected the greatest proportion of AVPCm+ tumors but was not evaluable in two thirds of patients. These data serve to inform the design of clinical trials using AVPCm as a criterion for patient selection or stratification. [Table: see text]