Abstract
Purpose
Recent data and National Comprehensive Cancer Network (NCCN) guidelines suggest that high-risk prostate cancer (cT3-4, Gleason score ≥8, or prostate-specific antigen [PSA] >20 ng/mL) is a heterogenous group in terms of long-term patient outcomes. We sought to determine whether subclassification of high-risk prostate cancer based on clinical factors correlates with genomic markers of risk.
Methods and materials
We identified 3220 patients with NCCN unfavorable intermediate-risk (n = 2000) or high-risk (n = 1220) prostate cancer from a prospective multi-institutional registry cohort. We defined the following subclassification of high-risk prostate cancer based on previously published data: favorable high risk (cT1c, Gleason 6, and PSA >20 ng/mL or cT1c, Gleason 4 + 4 = 8, PSA <10 ng/mL); very high risk (cT3b-T4 or primary Gleason pattern 5); and standard high risk (all others with cT3a, Gleason score ≥8, or PSA >20 ng/mL). We used a set of 33 previously developed genomic classifiers, including Decipher, to determine whether high-risk genomic features correlate with clinical subclasses of high-risk prostate cancer.
Results
Among those with favorable high-risk, standard high-risk, and very high-risk prostate cancer, 50.4%, 64.2%, and 81.6% had a high-risk Decipher score, respectively (P < .001). Among 32 other genomic signatures, 29 had a similar trend of increasing risk scores across the 3 subclasses of high-risk disease (P < .05 after correction for multiple hypothesis testing). Patients in the 3 subclasses of high-risk disease had a median of 4, 6, and 13 high-risk signatures, respectively. In comparison, among those with unfavorable intermediate-risk prostate cancer, 38.2% had a high-risk Decipher score, and the median number of high-risk signatures was 3.
Conclusions
Although NCCN guidelines currently use a 2-tiered system for high-risk prostate cancer, genomic markers of risk correlate with the clinically validated subclassification of high-risk prostate cancer into favorable high-risk, standard high-risk, and very high-risk disease, further confirming the prognostic utility of this 3-tiered stratification.