Abstract
Background and objective
NRG/RTOG 0521 randomized men with high-risk localized prostate cancer (PC) to androgen suppression (AS) and definitive radiotherapy (RT) ± docetaxel-based chemotherapy (CT). The overall survival (OS) benefit with CT initially reported was lost on longer follow-up. The Decipher genomic classifier (GC) measures multiple transcripts relevant to docetaxel action. Basal/luminal differentiation portends differential response to AS and CT for high-risk localized and metastatic hormone-sensitive PC. We validated the Decipher GC in pretreatment biopsy samples for risk stratification and examined basal-luminal subtyping to predict docetaxel response.
Methods
Decipher GC scores and basal-luminal cellular subtypes were generated for specimens from NRG/RTOG 0521. The primary objective was to validate the independent prognostic ability of GC for metastasis-free survival (MFS). Treatment effects in luminal proliferating (LP) and non-LP cell subtypes were examined in relation to MFS, OS, and distant metastasis (DM).
Key findings and limitations
Samples were obtained from 283 patients and yielded 183 GC scores. Over median follow-up of 9.9 yr, 67 metastasis events were observed, including 34 DM events. Multivariable analysis revealed that GC was independently associated with DM (subdistribution hazard ratio 1.45) and MFS (hazard ratio 1.20). No biomarker-by-treatment interaction with GC and docetaxel was detected. The 10-yr restricted mean survival time difference in OS with CT was 13.7 mo for LP (p = 0.053) and 2.5 mo for non-LP (p = 0.63) tumors.
Conclusions and clinical implications
The Decipher GC score was independently associated with DM and MFS, and LP tumors may benefit from addition of CT. Validation of these findings may allow more effective use of CT in men with localized PC. The original NRG/RTOG 0521 trial is registered on ClinicalTrials.gov as NCT00288080.