139 - Genomic and Transcriptomic Profiling of Radiation Resistant, Locally Recurrent Prostate Cancer
Presenter(s)
B. K. Neilsen1, R. R. Huang2, L. Valle1, J. A. Proudfoot3, E. Davicioni3, U. Ryg4, M. Schulz-Jaavall4, J. B. Weidhaas1, M. Santoso1, J. Calais5, R. E. Reiter6, M. Rettig7, M. L. Steinberg1, A. Sisk2, W. Brisbane2, L. S. Marks2, P. C. Boutros2, W. Lilleby4, and A. U. Kishan1; 1Department of Radiation Oncology, University of California, Los Angeles, Los Angeles, CA, 2University of California, Los Angeles, Los Angeles, CA, 3Veracyte Inc., San Diego, CA, 4Oslo University Hospital, Oslo, Norway, 5Ahmanson Translational Theranostics Division, University of California, Los Angeles, Los Angeles, CA, 6Department of Urology, University of California, Los Angeles, Los Angeles, CA, 7Department of Medical Oncology, University of California, Los Angeles, Los Angeles, CA
Purpose/Objective(s): Very little is known about the biological underpinnings of intraprostatic recurrences after definitive radiotherapy for prostate cancer (PC). An understanding of their biology may have implications for the management of both recurrent and newly diagnosed prostate cancer. We hypothesize there are conserved molecular profiles associated with local radioresistance.
Materials/Methods: Forty-one locally radiorecurrent (LRR) prostate cancer tumors from 36 unique patients treated at two large academic centers were profiled using a targeted pan-cancer DNA sequencing panel as well as RNA expression (Veracyte, San Diego, CA). Genomic alteration frequencies within LRR PC were compared to alteration frequencies in treatment naïve PC from The Cancer Genome Atlas (TCGA). Significance was assessed using Chi-squared analysis. Transcriptomic data were compared to de-identified data from the Decipher Genomics Resource Information Database (GRID; NCT02609269, n = 146,865), which was accessed to create a subset (n=22,320) of patients identified by matching Gleason grade (GG) group at the time of initial diagnosis with the LRR cohort. Standardized mean differences were used to assess for small (<0.4), moderate (0.4-0.7) and large (>0.7) differences between the cohorts.
Results: The LRR cohort included patients with relatively aggressive disease including 32% with GG5, 24% with GG4, 12% GG3 and 2% unassigned. The LRR PC cohort demonstrated enrichment in PI3K pathway alterations compared to treatment naïve PC within TCGA with alteration frequencies in PTEN of 12% (5/41) vs 3%, [p = 0.005] and PIK3CA of 12% (5/41) vs 2% [p < 0.001]. Other genetic alterations enriched in LRR PC include BRCA1 of 10% (4/41), PTCH1 of 22% (9/41), MSH3 of 12% (5/41), and STAG2 of 34% (14/41). However, no common genetic alteration was conserved across all samples. The LRR cohort had proportionately greater patients with Decipher high or very high risk signatures (71% LRR vs. 56% for GG-matched GRID [SMD 0.47] and vs. 34% for the overall GRID cohort [SMD 1.0]). The LRR cohort demonstrated significantly higher basal phenotype (compared to luminal) based on multiple previously validated scores including PAM50 [SMD 0.89] and PSC [SMD 0.76]. Lower AR-activity was enriched in LRR (61% for LRR vs 9% for GG-matched GRID cohort [SMD 1.29]).
Conclusion: Genetic profiling and transcriptomic analysis of LRR PC revealed that LRR disease had an enrichment in mutations associated with high grade disease as well as higher Decipher scores, lower AR activity, loss of tumor suppressors and basal subtypes, even after matching for GG.