Differential treatment outcomes in BRCA1/2-, CDK12-, and ATM-mutated metastatic castration-resistant prostate cancer.

Background: DNA damage repair mutations (DDRm) are common in patients with metastatic castration-resistant prostate cancer (mCRPC). The optimal standard therapy for this population is not well described.

Methods: A multi-institutional, retrospective study of patients with mCRPC and DDRm was conducted. Patient data, including systemic therapies and responses, were collected. The decline in prostate-specific antigen ≥ 50% from baseline (PSA50) and overall survival (OS) from the treatment start were compared by mutation and treatment type. A multivariable Cox proportional hazards model for OS was created that controlled for DDRm, first-line treatment received for mCRPC, and clinical factors.

Results: The most common DDRm observed among 149 men with mCRPC were BRCA1/2 (44%), CDK12 (32%), and ATM (15%). The majority received first-line abiraterone (40%) or enzalutamide (30%). The PSA50 rate with first-line abiraterone was lower for CDK12 (52%) than BRCA1/2 (89%; P = .02). After first-line abiraterone or enzalutamide, the median OS was longest with second-line carboplatin-chemotherapy (38 months) in comparison with abiraterone or enzalutamide (33 months), docetaxel (17 months), or cabazitaxel (11 months; P = .02). PSA50 responses to carboplatin-based chemotherapy were higher for BRCA1/2 (79%) than ATM (14%; P = .02) or CDK12 (38%; P = .08). In a multivariable analysis, neither the specific DDRm type nor the first-line treatment was associated with improved OS.

Conclusions: Responses to standard therapies were generally superior in patients with BRCA1/2 mutations and inferior in patients with ATM or CDK12 mutations. The DDRm type did not independently predict OS. After progression on first-line abiraterone or enzalutamide, carboplatin-based chemotherapy was associated with the longest OS. These findings may inform treatment discussions and clinical trial design and require prospective validation.