A subset of triple-negative breast cancer (TNBC) expresses the androgen receptor (AR), but thresholds for AR positivity and its clinical significance vary. We hypothesize that objective assessment outperforms subjective methods, and that high AR negatively impacts prognosis. In a population-based TNBC cohort (n = 198) with long follow-up (4–383 months), AR expression was evaluated via subjective scoring (AR-Manual) and automated digital image analysis (AR-DIA). A 10% cut-off value via AR-DIA was the strongest negative prognostic threshold for distant metastases (p = 0.008). High AR-DIA correlated with lower grade (p = 0.014), and lower proliferation (p = 0.004) but also with larger tumors (p = 0.047), distant metastasis (p = 0.052), and lymph node (LN) positivity (p < 0.001), highlighting its dual roles. Multivariate analysis revealed interaction between LN status and AR-DIA (p < 0.001) as the strongest prognostic factor, followed by fibrotic focus (FF; p = 0.009), mitotic activity index (MAI; p = 0.018), and stromal tumor-infiltrating lymphocytes (sTILs; p = 0.041). AR-DIA had no additional prognostic value in favorable subgroups but was significant in unfavorable subgroups. In high AR-DIA patients with unfavorable characteristics, ACT did not improve survival, and patients may benefit from AR-targeted therapy. Overall, the DIA method provides reproducibility, high AR-DIA (≥10%) shows opposing survival effects in different TNBC subgroups, and AR evaluation is crucial for prognosis and AR-targeted therapies.
Read full abstract