Abstract 4947: Branched chain RNA in situ hybridization for the detection of androgen receptor splice variants within archival prostate cancer tissue

Abstract

Abstract Introduction: Androgen receptor (AR) mRNA splice variants have emerged as predictive biomarkers for response to AR targeted therapies. There are no commercially available assays to detect AR splice variants. The branched chain RNA in situ hybridization (ISH) platform enables the highly sensitive detection of RNA transcripts in formalin-fixed, paraffin-embedded (FFPE) tissues. In this pilot study, we developed an RNA ISH assay to detect the constitutively active AR-V7 splice variant, and we correlated AR-V7 expression with clinical outcomes after first-line androgen deprivation therapy (ADT) for metastatic prostate cancer. Methods: We designed a branched chain RNA ISH probe to target the unique cryptic exon CE3 of AR-V7 using multiple tiling probes (Affymetrix). Automated ISH assays were performed using ViewRNA eZl Detection Kit on the BOND RX IHC and ISH Staining System. A semiquantitative scoring method was used to score the RNA ISH signal in FFPE tissue. We analyzed AR-V7 and full length AR within two prostate cancer cohorts that we hypothesized to represent “low likelihood” and “high likelihood” to have detectable AR-V7. “Low likelihood” men had undergone prostatectomy for Gleason 6 adenocarcinoma, and “high likelihood” men had metastatic castration resistant prostate cancer with tumor tissue obtained after disease progression despite at least one subsequent line of hormonal therapy (abiraterone, enzalutamide, or bicalutamide). We then analyzed two additional cohorts who experienced a “sustained response” (>2.5 years; n = 13) or “brief response” (<9 months; n = 9) to first-line ADT for metastatic prostate cancer and for whom baseline tumor tissue was available. Results: “High likelihood” cohort samples had detectable AR-V7 with a score of 1+ to 3+ in all samples (n = 12). “Low likelihood” cohort samples had no detectable AR-V7 (n = 10). Given the apparent binary distinction of AR-V7 signal among the above groups, we analyzed pre-treatment AR-V7 status as a predictive and prognostic biomarker in men with treatment-naive metastatic disease. Detectable AR-V7 was more common among “brief response” samples (6 of 9) than among “sustained response” samples (4 of 13) (not significant; P = 0.19). Patients without detectable AR-V7 RNA had significantly longer overall survival (OS, logrank P = 0.044), with a non-significant trend toward superior progression-free survival (PFS, logrank P = 0.055). Conclusions: Within an institutional cohort, the RNA ISH assay identified AR-V7 in all tested samples of high clinical likelihood for the splice variant RNA and no tested samples of low clinical likelihood. AR-V7 RNA was detectable in some pretreatment samples, and its presence was associated with significantly shorter overall survival. The potential prognostic and predictive utility of AR-V7 status as determined by RNA ISH from conventionally prepared FFPE tissue warrants further study in larger cohorts. Citation Format: Philip J. Saylor, Richard J. Lee, Kshitij S. Arora, David T. Ting, Vikram Deshpande, Miguel N. Rivera, Rong Hu, Chin-Lee Wu, David T. Miyamoto. Branched chain RNA in situ hybridization for the detection of androgen receptor splice variants within archival prostate cancer tissue. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr 4947.