Identifying prognostically relevant transcription factor activity in breast cancer

Abstract

Breast cancer is the leading malignancy in women worldwide. It has five primary groups based on transcriptional characterization: luminal A, luminal B, HER2+, basal-like and normal-like. Previous studies have used transcriptomic profiling to identify genes of prognostic relevance within and across these subtypes. However, this analysis fails to provide direct insight into biological processes contributing to differential survival. Therefore, we estimate transcription factor activity from transcriptomic data and explore molecular characteristics of breast cancer based on it. The TCGA data contains 1090 breast primary tumor samples split into 567 luminal A, 207 luminal B, 82 HER2+, 194 basal-like, and 40 normal-like patients. The activity of 1333 transcription factors is estimated for each sample. There are 145 transcription factors with significantly different patient survival (log-rank test, p <0.01) identified when they are stratified into high and low activity groups. STAT3 is the leading candidate (p < 5e-4) whose activity is correlated to prognosis, but it is confounded by the subtype-specific survival differences. To eliminate this variable, further analysis was limited to patients within primary subtypes. Among luminal B samples, for instance, low FOSL1 and FOSL2 activity indicated better survival. FOS protein family has been implicated in cell proliferation and drug resistance. In contrast, among the HER2+ group, TP53 high activity indicated better survival. Mutation of the TP53 gene is common in breast cancer and it encodes proteins that cannot suppress tumors. All noted survival results are significant (p < 0.01) to impact of prognosis and consistent with known impact on cancer. Furthermore, the target genes regulated by prognostically relevant transcription factors are analyzed to assist the explanation of transcription factors on prognosis. For instance, lack of FAS leads to poor survival and promotes skeletal metastasis. With the strong activity of TP53, expression of FAS is high and hence supports its performance on prognosis. Therefore, the targets contribute to explaining the activity of transcription factors on prognosis.