Subtype-Specific Metagene-Based Prediction of Outcome after Neoadjuvant and Adjuvant Treatment in Breast Cancer

Abstract Purpose: In spite of improvements of average benefit from adjuvant/neoadjuvant treatments, there are still individual patients with early breast cancer at high risk of relapse. We explored the association with outcome of robust gene cluster–based metagenes linked to proliferation, ER-related genes, and immune response to identify those high-risk patients. Experimental Design: A total of 3,847 publicly available gene-expression profiles were analyzed (untreated, N = 826; tamoxifen-treated, N = 685; chemotherapy-treated, N = 1,150). Genes poorly performing in formalin-fixed samples were removed. Outcomes of interest were pathologic-complete response (pCR) and distant metastasis-free survival (DMFS). In ER+HER2−, the proliferation and ER-related metagenes were combined to define three risk groups. In HER2+ and ER−HER2− risk groups were defined by tertiles of an immune-related metagene. Results: The high-proliferation/low-ER group of ER+HER2− breast cancer had significantly higher pCR rate [OR, 5.01 (1.76–17.99), P = 0.005], but poorer outcome [HR = 3.73 (1.63–8.51), P = 0.0018] than the low-proliferation/high-ER. A similar association with outcome applied to patients with residual disease (RD) after neoadjuvant chemotherapy (P = 0.01). In ER−HER2− and HER2+ breast cancer, immune metagene in the high tertile was linked to higher pCR [33.7% vs. 11.6% in high and low tertile, respectively; OR, 3.87 (1.79–8.95); P = 0.0009]. In ER−HER2−, after adjuvant/neoadjuvant chemotherapy, 5-year DMFS was 85.4% for high-tertile immune metagene, and 43.9% for low tertile. The outcome association was similar in patients with RD (P = 0.0055). In HER2+ breast cancer treated with chemotherapy the association with risk of relapse was not significant. Conclusions: We developed metagene-based predictors able to define low and high risk of relapse after adjuvant/neoadjuvant therapy. High-risk patients so defined should be preferably considered for trials with investigational agents. Clin Cancer Res; 22(2); 337–45. ©2015 AACR.