Structural classification of FOXA1 alterations and their association with clinical outcomes in prostate cancer patients: A multi-institutional retrospective analysis.

Abstract

256 Background: FOXA1 is a transcription factor with gene alterations in 10-15% of prostate cancers. Due to its complex dynamics with the androgen receptor, the effects of mutations in the FOXA1 gene are variable, and clinical outcomes associated with FOXA1 alterations have been challenging to decipher. A recent classification system for FOXA1 alterations defined by mutational mechanism and orientation with respect to FOXA1 protein domains has been developed to inform clinical management for prostate cancer (Hwang J. CCR 2025). Here, we report consolidated multi-institutional patient outcomes associated with FOXA1 alteration classes in prostate cancer patients. Methods: We retrospectively analyzed 189 patients across 5 institutions by FOXA1 alteration class and calculated per-patient duration of exposure to androgen deprivation therapy (ADT), androgen receptor pathway inhibitors (ARPIs), and taxane chemotherapy. Associations of FOXA1 class and overall survival (OS) was examined by univariable Kaplan-Meier plot with log-rank test. Multivariable Cox proportional hazards models were fitted for OS related to FOXA1 alteration class and systemic treatment duration as independent predictors, adjusted for age, metastatic disease at diagnosis, and Gleason grade group as covariates, stratified by treating institution. Additional models were fitted to investigate for interaction effects between treatment exposures and FOXA1 mutation class. Baseline risk was defined by patients with FOXA1 Class 1B mutations. Results: Our final cohort included 9 patients with Class 1A, 94 with Class 1B, 19 with Class 1C, 3 with Class 2, 18 with Class 3A, 27 with Class 3B, and 19 with Class 4 alterations. FOXA1 Class 2 alterations correlated with decreased OS (HR 4.4, p=0.04). On multivariable analysis, improved OS was independently associated with longer treatment duration on ADT (HR 0.97 per month, p=0.03) and ARPIs (HR 0.89 per month, p<0.001). No significant association was found with duration of taxane. Interaction models showed treatment effect modification with ADT duration in Class 4 (β [Class4:ADT ] +0.07, p=0.003) and ARPI duration in Class 3B and 4 alterations (β [Class3B:ARPI] +0.10, p=0.05; β [Class4:ARPI] +0.09, p=0.03), suggesting modulation of the per-month survival benefit of treatment duration. Conclusions: Prior studies show that FOXA1 alterations produce distinct prostate cancer phenotypes. Here, we found that FOXA1 Class 2 alterations were significantly correlated with worse OS. Multivariable models suggest more OS benefit of ADT and/or ARPI duration in FOXA1 Class 1B, compared to 3B and 4 alterations. These findings emphasize the importance of structural subclassification of FOXA1 variants, rather than just their presence or absence, for prognostication and therapeutic selection for patients with prostate cancer.