Predictive and Prognostic Value of Androgen Receptor Expression Across All Molecular Subtypes in Patients with Breast Cancer Who Received Neoadjuvant-Treatment: Single-Center Retrospective Analysis

Abstract

Purpose Androgen receptor (AR) expression is detectable in the majority of breast cancers across all molecular subtypes; however, its predictive value for pathological complete response (pCR) and its prognostic contribution to long-term survival in an all-comer neoadjuvant chemotherapy (NACT) cohort remain incompletely defined. Patients and Methods In this retrospective, single-center cohort study, 196 patients with early-stage breast cancer across all molecular subtypes received NACT at Istanbul Medipol University Hospital between 2013 and 2025. Pre-NACT AR expression was assessed by immunohistochemistry (SP107 clone) with a pre-specified threshold of 50% applied. The primary endpoint was event-free survival (EFS); 35 EFS events occurred. Multivariable Cox proportional hazards regression (three pre-specified covariates), Kaplan–Meier analyses, and competing risk modeling were performed. Results AR ≥50% was detected in 139 patients (70.9%). The overall pCR rate was 30.1% (59/196). AR ≥50% was independently associated with superior EFS in the multivariable model (hazard ratio [HR]=0.396, 95% CI: 0.199–0.786; p=0.008). The estimated 5-year EFS was 82.2% vs 69.8% for AR ≥50% versus AR <50% patients (log-rank p=0.031). A statistically significant AR × pCR status interaction was identified (p=0.032), with the prognostic gradient of AR most pronounced among patients with residual disease (5-year EFS 77.9% vs 52.9% for AR-high vs AR-low). In HR+/HER2− disease, AR-low status paradoxically predicted higher pCR rates (median AR 40% vs 90%; p=0.027) but inferior long-term EFS, consistent with a “luminal paradox”. Subgroup analyses are exploratory and hypothesis-generating. Conclusion In this retrospective single-center cohort, pre-NACT AR ≥50% was independently associated with superior EFS in the multivariable model, complementary to pCR status. A formally significant AR × pCR interaction supports that AR provides differential prognostic information in the residual-disease population. These hypothesis-generating findings support prospective evaluation of AR-stratified risk assessment after NACT.