Abstract 1378P
Background
Subtype differentiation in prostate cancer (PCa) has been shown to be prognostic and predictive of androgen deprivation therapy (ADT) response in the postoperative setting. We sought to determine whether basal-luminal subtypes based on gene expression profiling of pretreatment biopsy samples would be predictive for the benefit of long- (LT; 24-28 months) vs short-term (ST; 4 months) ADT, hypothesizing that patients with basal tumors benefit more from longer term ADT.
Methods
Whole-transcriptome arrays of biopsy samples from patients enrolled in the NRG/RTOG 9202, 9413, and 9902 randomized phase III trials were analyzed. The prognostic and predictive ability of a 215 gene basal-luminal prostate subtyping classifier (PSC) was evaluated for biochemical failure (BF), distant metastasis (DM), metastasis-free survival (MFS), prostate cancer-specific mortality (PCSM), and overall survival (OS) with multivariable Cox models (MVA). Event rates were estimated by the cumulative incidence method and compared with Gray’s or logrank tests.
Results
265 samples were analyzed (40% PSC basal, 60% luminal). In MVA, the PSC basal subtype had a worse prognosis compared to PSC luminal for MFS (HR [95% CI] 1.8 [1.3-2.5], p<0.001), PCSM (HR 2.8 [1.5-5.0], p<0.001), and OS (HR 1.8 [1.3-2.6], p<0.001). PCSM rates at 10 years were 26% (95% CI 17-35%) for basal vs 11% (6-15%) for luminal subtypes. A significant interaction between PSC and ADT duration was observed for BF (p=0.02) and PCSM (p=0.007). Similar but non-significant trends were observed for DM, MFS, and OS. Basal subtypes significantly benefitted from LT- vs STADT (10-year PCSM 5% [95% CI 0-11%] vs 42% [29-56%], p<0.001), while luminal outcomes did not differ by ADT duration (p=0.72).
Conclusions
Gene expression analysis of pretreatment biopsy samples from three NRG phase III trials of high-risk PCa suggests that basal-luminal subtyping is predictive for the benefit of LT- vs STADT. Patients with basal tumors benefitted from longer term ADT while those with luminal did not, supporting the hypothesis that basal-luminal subtyping from biopsy samples can help personalize ADT duration in high-risk PCa.
Clinical trial identification
Editorial acknowledgement
Legal entity responsible for the study
NRG Oncology.
Funding
Veracyte, Inc., National Cancer Institute, Pfizer, Bristol Myers Squibb & Takeda Pharmaceutical.
Disclosure
V. Liu, J.A. Proudfoot, E. Davicioni, Y. Liu: Financial Interests, Personal, Full or part-time Employment: Veracyte. All other authors have declared no conflicts of interest.