Androgen Receptor Immunohistochemistry as a Companion Diagnostic Approach to Predict Clinical Response to Enzalutamide in Triple-Negative Breast Cancer.

Purpose: The androgen receptor (AR) is increasingly recognized as a potential biomarker for identifying a subset of patients with possible hormonally driven triple-negative breast cancer (TNBC). However, its performance as a companion diagnostic remains elusive. Thus, we evaluated AR expression by immunohistochemistry in patients with advanced TNBC before treatment with the AR inhibitor enzalutamide.

Methods: We optimized and validated immunohistochemistry assays in breast and prostate cancer cell lines and tissues using two commercial AR monoclonal antibodies (SP107 and AR441). AR expression was then examined in patients with advanced TNBC enrolled in a phase II study of enzalutamide (ClinicalTrials.gov identifier: NCT01889238) on archived or fresh tissue before treatment. Association with clinical response was assessed by sensitivity, specificity, positive predictive value (PPV), drop-out rate, and survival.

Results: AR expression was detected in 80% and 63% of breast cancer tissue using SP107 and AR441, respectively. SP107 was selected for additional analyses because of its higher sensitivity and robustness. Total AR nuclear staining demonstrated the best accuracy in predicting clinical response (area under receiver operating characteristic curve, 0.72; = .0001). At a threshold of 10%, 74.6% of patients were AR positive, leading to 30% PPV, 90% sensitivity, and 30% specificity. These patients showed a significantly higher median progression-free survival (hazard ratio, 0.56; 95% CI, 0.36 to 0.88; = .011) and overall survival (hazard ratio, 0.54; 95% CI, 0.32 to 0.91; = .019) compared with those with AR-negative (< 10%) TNBC.

Conclusion: At a threshold of ≥ 10% nuclear expression, the AR was associated with TNBC response to enzalutamide. However, the modest PPV may restrict its clinical application, and additional diagnostic tools may be helpful for improved patient selection.