Molecular subtyping of prostate cancer into three categories—luminal A, luminal B, and basal—identifies differences in prognosis and response to treatment, according to a new study combining a retrospective and a prospective cohort.
“Prostate cancer was first thought to derive from glandular luminal cells; however, there is mounting evidence that basal cells may also play a role in prostate carcinogenesis,” wrote study author Felix Y. Feng, MD, of the Helen Diller Family Comprehensive Cancer Center at the University of California, San Francisco, and colleagues. Other cancers including breast and bladder cancers can be molecularly defined based on these cell types, and the authors hypothesized that using the PAM50 classifier could divide prostate cancer as well.
The study included a retrospective cohort of 1,567 prostate cancer samples, with a median follow-up of 10 years, and 2,215 prospectively collected samples. The PAM50 classifier was able to consistently separate the samples into three categories: luminal A disease (retrospective cohort, 34.3%; prospective cohort, 33.3%), luminal B disease (retrospective, 28.5%; prospective, 32.6%), and basal disease (retrospective, 37.1%; prospective, 34.1%). Results of the analysis of these cohorts were published in JAMA Oncology.
“Patients with luminal B tumors consistently have poorer outcomes for all end points compared with those with luminal A and basal tumors,” the authors wrote. The 10-year actuarial rate of biochemical relapse–free survival was 29% for luminal B, compared with 41% and 39% for luminal A and basal, respectively. For distant metastasis–free survival (DMFS), these rates were 53% for luminal B and 73% for both luminal A and basal subtypes.
This was similar for prostate cancer–specific survival (PCSS), with a 10-year actuarial rate for luminal B of 78%, compared with 89% for luminal A and 86% for basal. For overall survival, the rates were 69% for luminal B, 82% for luminal A, and 80% for basal.
On a multivariate analysis, luminal A had a hazard ratio (HR) for DMFS compared with luminal B of 0.55 (95% CI, 0.43–0.69; P < .001). The basal subtype had an HR compared with luminal B of 0.66 (95% CI, 0.53–0.82; P < .001). For PCSS, the HR for luminal A compared with luminal B was 0.50 (95% CI, 0.35–0.71; P < .001); basal subtype was not significantly better than luminal B on that analysis.
The classification also highlighted differences in response to treatment. In a subset of the retrospective cohort, luminal B tumors treated with androgen deprivation therapy (ADT) had a 10-year metastasis rate of 33% compared with 55% in untreated luminal B tumors. For non–luminal B tumors, treated tumors had a rate of 37% and untreated tumors had a rate of 21% (P = .006 for interaction).
“We believe this work not only represents a significant step forward in our understanding of prostate cancer heterogeneity but also is a potential classifier that may identify patients who benefit from postoperative ADT on a clinical-grade platform and provide guidance in personalizing patient care,” the authors concluded.
In an accompanying editorial, Wassim Abida, MD, PhD, and Howard I. Scher, MD, of Memorial Sloan Kettering Cancer Center in New York, wrote that the study “serves as a model for biomarker development,” and that it raises questions regarding potential drivers of aggressive early prostate cancer. “With the wealth of integrative genomic and clinical outcomes data now becoming available, we expect to see increasing application of molecular classification in the investigation of treatment response, leading to biomarker-driven treatment selection for patients with prostate cancer.”