Identification and evaluation of network modules for the prognosis of basal-like breast cancer

Abstract

Purpose: Basal-like breast cancer (BLBC) is a molecular subtype of breast cancer associated with poor clinical outcome, although some patients with BLBC experience long-term survival. Apart from nodal status, current clinical/histopathological variables show little capacity to identify BLBC patients at either high- or low-risk of disease recurrence. Accordingly, we sought to develop a network based genomic predictor for predicting the outcome of patients with BLBC.

Experimental Design: We performed network analysis on global gene expression profiling data of BLBCs, and identified BLBC network modules associated with AP-1 transcription, G-protein coupled receptors, and T-, B-, and NK-cells that are significant predictors of BLBC patient survival.

Results: In gene expression and tissue microarray (TMA) validation cohorts of 210 and 102 BLBC patients, respectively, the identified network modules were robustly associated with patient outcome. In the gene expression validation cohort, the Kaplan-Meier estimate for 10-year survival in the low-risk group was 90%, whereas in the high-risk group it was a 56%. In the TMA cohort, the Kaplan-Meier estimate for 10-year survival in the low-risk group was 98%, whereas in the high-risk group it was 71%.

Conclusions: The capacity to distinguish between patients with BLBC at high- or low-risk of recurrence at the time of diagnosis could permit timely intervention with more aggressive therapeutic regimens in those patients predicted to be high-risk, and to avoid such therapy in low-risk patients.