Background: Identifying biomarkers that can predict the prognosis and treatment response is helpful for individualizing breast cancer (BC) therapy. A neoadjuvant treatment setting is ideal for testing biomarkers capable of predicting the treatment response. This study analyzed the value of immunohistochemical biomarkers for predicting pathological complete response (pCR) and prognosis in a group of BC patients receiving standardized treatment.
Patients and methods: A total of 100 BC patients were treated with neoadjuvant chemotherapy (four cycles of epirubicin and cyclophosphamide) between 2000 and 2005. Formalin-fixed and paraffin-embedded core biopsies were taken before chemotherapy for immunohistochemical staining of ER, PgR, HER2, Bcl-2, p53, cyclin D1, CK5/6, CK8, CK18, and TOP2A. Patient and tumor characteristics and biomarker scores were used to predict pCR and prognosis, using logistic regression and Cox proportional hazard models.
Results: pCR was achieved in 11 patients and was predicted by the established marker Ki-67. In addition, CK5/6 and CK18 improved the prediction model and were associated with lower pCR rates. For the prognosis, only the established markers nodal status, Ki-67, and PgR predicted overall survival and nodal status; Ki-67 and PgR predicted distant disease-free survival.
Conclusions: In this small retrospective study, CK5/6 and CK18 appeared to improve prediction of pCR in addition to the established markers. CK5/6 may indicate a tumor type resembling a basal phenotype that is more resistant to anthracycline-based therapy, and CK18 may indicate a luminal subtype that is more resistant to chemotherapy. However, these results need to be replicated in larger studies.