Comparative genomic analyses between Asian and Caucasian prostate cancers in an 80,829 patient cohort.

Abstract

273 Background: The Decipher 22-gene genomic classifier (GC) has been shown to predict for metastasis and survival in predominantly Caucasian and African-American men from Western cohorts. There is however little data on the clinical utility of GC in Asian prostate cancer (PCa). We investigated if GC prognosticates for metastasis-free survival (MFS) in an Asian cohort of localized PCa. Additionally, we performed comparative genomic analyses between our Asian patients and non-Asian cases from a large Decipher GRID database (Veracyte, San Francisco, CA). Methods: We used a cohort of PCa patients who were treated at a single institution from East Asia between 2006 to 2021. Patients underwent active surveillance or radical prostatectomy (RadP) and/or radiotherapy (RT) +/- hormonal therapy (HT). The GC assay (Decipher Biosciences, CA) was performed on diagnostic biopsies, following central review of the Gleason’s grade (GG) and tumor cellularity by an expert GU pathologist. MFS was the primary endpoint for survival analysis. Comparative analyses of 273 gene-signatures were performed against the full and a propensity-score matched (PSM) cohort identified from the Decipher GRID database. Adjusted p-values from Wilcoxon tests were used to select from these signatures. Results: We profiled 126 unique patient tumors, comprising of 19 (15.1%) NCCN-defined low-/favorable intermediate-, 22 (17.5%) unfavorable intermediate-, 34 (27.0%) high-, and 51 (40.5%) very high-risk groups. 24 (19.0%) patients had RadP as primary treatment, 98 (77.8%) had RT +/- HT, while 4 (3.2%) underwent active surveillance or HT alone. Using the GC, 70 (55.6%), and 56 (44.4%) were stratified as low/intermediate- (≤0.6) and high-risk ( > 0.6), respectively. GC high-risk was significantly associated with an inferior MFS than GC low/intermediate-risk ( HR 5.22 [95% CI:1.08–25.3], P = 0.04). Comparison between our Asian and the full unmatched GRID cohort (N = 80,703) revealed a lower proportion of ERG+ PCa (14% vs 41%, P < 0.001) and a higher proportion of PAM50 basal subtypes (41% vs 30%, P < 0.001), as well as a lower T-cell exclusion (median 0.080 vs 0.097; P < 0.001) and angiogenesis (-0.37 vs -0.08; P < 0.001) signature scores. These trends were also observed when comparing with the PSM-subset (N = 630; 5:1 ratio for NCCN, GG, age at diagnosis, and assay quality score). Interestingly, both high angiogenesis and T-cell exclusion signatures were associated with a worse MFS ( HR not available due to no events for low angiogenesis, P = 0.0015 by log-rank test; HR 5.12 [1.03–26.5], P = 0.046, respectively). Conclusions: We validated the Decipher 22-gene GC for prognostication of MFS in a predominantly high-risk PCa cohort. We also identified several gene expression signatures that were significantly different between Asian PCa and other cohorts. These observations may have implications for customizing treatment recommendations to an Asian population.