Background: We have previously reported that high expression of the erbB-2 gene (also known as HER-2/neu and ERBB2) in breast cancer is associated with patient response to dose-intensive treatment with cyclophosphamide, doxorubicin (Adriamycin), and 5-flurouracil (CAF) on the basis of short-term follow-up of 397 patients (set A) with axillary lymph node-positive tumors who were enrolled in Cancer and Leukemia Group B (CALGB) protocol 8541.
Methods: To validate those findings, we conducted immunohistochemical analyses of erbB-2 and p53 protein expression in an additional cohort of 595 patients (set B) from CALGB 8541, as well as a molecular analysis of erbB-2 gene amplification in tumors from all patients (sets A and B). Marker data were compared with clinical, histologic, treatment, and outcome data.
Results: Updated analyses of data from set A (median follow-up, 10.4 years) showed an even stronger interaction between erbB-2 expression and CAF dose, by use of either immunohistochemical or molecular data. A similar interaction between erbB-2 expression and CAF dose was observed in all 992 patients, analyzed as a single group. However, for set B alone (median follow-up, 8.2 years), results varied with the method of statistical analysis. By use of a proportional hazards model, the erbB-2 expression-CAF dose interaction was not significant for all patients. However, in the subgroups of patients randomly assigned to the high- or the moderate-dose arms, significance was achieved. When patient data were adjusted for differences by use of a prognostic index (to balance an apparent failure of randomization in the low-dose arm), the erbB-2 expression-CAF dose interaction was significant in all patients from the validation set B as well. An interaction was also observed between p53 immunopositivity and CAF dose.
Conclusions: The hypothesis that patients whose breast tumors exhibit high erbB-2 expression benefit from dose-intensive CAF should be further validated before clinical implementation. Interactions between erbB-2 expression, p53 expression, and CAF dose underscore the complexities of predictive markers where multiple interactions may confound the outcome.