Background: Tailor-made therapies are currently gaining prominence, and the clarification of predictive markers for anticancer agents represents an important research issue. From our institutional neoadjuvant experience, apocrine carcinoma showed a high correlation with therapeutic effect against breast cancer. We thus considered that apocrine metaplasia (AM) might represent a predictive marker for breast cancer.
Methods: A total of 210 primary invasive breast cancers from Japanese patients were scored according to the size of cytoplasmic granules and abundance of cytoplasm, then classified into three categories: non-AM, incomplete AM and complete AM. Clinicopathological features were evaluated based on these classifications.
Results: Distribution according to the classification of AM was: non-AM, 61%; incomplete AM, 36%; and complete AM, 3%. No significant differences with regard to estrogen receptor, progesterone receptor, human epidermal growth factor receptor type 2, androgen receptor or bcl-2 were observed among the three groups. Gross cystic fluid protein-15 showed a high positive rate (83%) for complete AM. Cases of incomplete AM and complete AM were combined to form the AM group. Among lymph node-positive patients without chemotherapy, the 10-year recurrence-free survival (RFS) rate was 85% for non-AM and 44% for AM (p < 0.05). Conversely, among the lymph node-positive group with chemotherapy, the 10-year RFS rate was 45% for non-AM and 75% for AM (p < 0.05). Prognoses for non-AM and AM were turned around by chemotherapy. Lymph node metastasis was related to prognosis in multivariate analysis, although AM did not remain an independent prognosticator.
Conclusions: Apocrine metaplasia appears to offer an effective predictive marker for anticancer therapy.