Abstract 3910: AR copy number amplification and AR/KLK3 expression patterns reveal mechanisms of AR signaling inhibitor (ARSI) resistance and highlight the need for AR-directed therapeutic innovation in metastatic castration resistance prostate cancer (mCRPC).

Abstract Metastatic CRPC is predominantly driven by androgen receptor (AR) signaling. Despite second-generation ARSI, resistance remains a challenge and is associated with AR ligand-binding domain (LBD) mutations and AR copy number amplification (CNA). Prostate-specific antigen (PSA), encoded by kallikrein-related peptidase 3 (KLK3), serves as a key marker of AR transcriptional activity. We hypothesized that AR genomic alterations underpin variation in AR expression and activity and that real-world profiling would identify distinct AR/KLK3 expression-defined subgroups. From the Caris Prostate Cancer Database (N=17,429), 511 mCRPC patients treated with ARSI (Abiraterone and/or Enzalutamide) were identified. Whole exome and transcriptome sequencing (Caris Life Sciences) were performed on pre- and post-treatment tumor biopsies to assess AR alterations and RNA expression of AR and KLK3. Patients were stratified by AR CNA (CN2) and pathogenic AR LBD mutations and categorized into four RNA subgroups: ARhi/KLK3hi, ARhi/KLK3lo, ARlo/KLK3hi, ARlo/KLK3lo. Associations with time on treatment (TOT) and overall survival (OS) were evaluated with Kaplan-Meier and log-rank tests. ARSI reduced AR wild-type (WT; mut–/CNA–) tumors and enriched for AR alterations, with AR CNA in 42–51% and AR mut in 3–10% of post-treatment samples. AR CNA+ and AR mut+ tumors showed significantly higher AR RNA and KLK3 RNA versus WT (all p0.05). AR CN correlated with AR RNA (ρ=0.77, p0.0001) and was associated with elevated KLK3 RNA (p0.0001); AR RNA correlated with KLK3 RNA (ρ=0.4, p0.0001). The dominant RNA subgroups were ARhi/KLK3hi and ARlo/KLK3lo (each 31%), followed by ARlo/KLK3hi and ARhi/KLK3lo (each 19%). AR CNA+ tumors were enriched in ARhi/KLK3hi (55%), AR mut+ in ARlo/KLK3hi (48%), and AR WT in KLK3lo (59%) subgroups. ARlo/KLK3hi patients had the best ARSI response; ARhi/KLK3hi and ARlo/KLK3lo showed intermediate TOT, and ARhi/KLK3lo the shortest. ARhi/KLK3hi patients had worse OS versus ARlo/KLK3hi (HR=2.79, 95% CI 1.05–7.42, p=0.04). AR CNA predicted shorter TOT versus AR WT (HR=1.94, 95% CI 1.17–3.22, p=0.01). AR CNA is the main driver of elevated AR signaling and reduced ARSI response, while AR mut is associated with lower AR expression and better outcomes. KLK3lo tumors, largely AR WT, show reduced AR dependence with limited benefit from AR-targeted therapies. These findings highlight a high-risk AR-driven subgroup defined by high AR expression and AR CNA. FX-111, a selective degrader of transcriptionally active, hormone-bound AR (ARON) in IND-enabling studies by Flare Therapeutics, offers a promising strategy to address this resistance. Ongoing analyses of AR/KLK3 subgroups aim to elucidate biological drivers of differential ARSI response. Citation Format: Celine Robert-Tissot, Phuong A. Nguyen, Elisabeth A. Murphy, Yawei J. Yang, Evisa Gjini, Michaela Bowden. AR copy number amplification and AR/KLK3 expression patterns reveal mechanisms of AR signaling inhibitor (ARSI) resistance and highlight the need for AR-directed therapeutic innovation in metastatic castration resistance prostate cancer (mCRPC) abstract. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 3910.