Association between genomic classifier scores and initial management of localized prostate cancer in a population-based cohort in the United States.

350 Background: Tissue-based gene expression assays provide prognostic information in prostate cancer, but their independent clinical value, particularly in guiding the use of initial treatment versus active surveillance (AS), remains uncertain. We evaluated the association between results of a 22-gene genomic classifier (GC) and initial management in a novel, population-based sample linking genomic and clinical data. Methods: We leveraged the Surveillance, Epidemiology, and End Results – Decipher linked database, which integrates incident prostate cancer diagnoses during 2010-2018 with Decipher Prostate GC test orders and results. Patients included in our analysis had low-risk or favorable intermediate-risk prostate cancer, underwent biopsy-based GC testing, and had complete demographic and clinical data (PSA, Gleason grade group, stage, NCCN risk category). First-course management was classified as immediate therapy (surgical or radiation) versus AS. GC scores were categorized based on commonly used thresholds: low (<0.45), intermediate (0.45–0.60), or high (>0.60). We used multivariable logistic regression to assess the association between the receipt of AS versus immediate treatment and GC scores, adjusting for demographic, clinical, and pathological covariates, with stratification by clinical risk group (low- or intermediate-risk). Results: Among 2,547 men with low- or favorable intermediate-risk prostate cancer, 744 (72.2%) with low-risk disease and 404 (26.6%) with intermediate-risk disease were managed initially with AS. In those with low NCCN risk disease, the proportion managed with initial AS was higher in those with low GC scores (419/527 [79.5%]), compared to 218/304 (71.7%) with intermediate GC scores and 107/199 (53.8%) with high GC scores (p<0.001). Similarly, among those with intermediate-risk disease, AS was more common in those with low GC scores (239/562 [42.5%]) compared to 103/425 (24.2%) with intermediate GC scores and 62/530 (11.7%) with high GC scores, p<0.001 (Figure). In the low-risk group, higher GC scores were independently associated with lower odds of AS versus treatment (compared to low GC, intermediate GC: odds ratio [OR] = 0.69, 95% confidence interval [CI]: 0.49–0.97, p=0.033; high GC: OR=0.29, 95% CI: 0.20–0.43, p<0.001). In the intermediate-risk group, the associations were of greater magnitude (intermediate GC: OR=0.42, 95% CI: 0.31–0.57, p<0.001; high GC: OR=0.19, 95% CI: 0.14–0.27, p<0.001). Conclusions: In this analysis of the first publicly available population-based linkage of genomic and clinical data, higher GC scores were independently associated with lower odds of AS among men with low- or intermediate-risk prostate cancer in the United States.