Different Biological Processes Are Associated with Prognosis and Chemotherapy Sensitivity in the Different Molecular Subtypes of Breast Cancer.

Abstract

Abstract Background: Breast cancer includes multiple different molecular subtypes. We hypothesized that different biological processes and molecular markers may be associated with prognosis and chemotherapy sensitivity in the different subtypes.Methods: We performed gene set analysis in ER-positive/HER-2-normal (ER+) and ER-negative/HER-2-normal (ER-) cancers separately to identify gene sets that are associated with prognosis and chemotherapy sensitivity. We tested 5295 different, functionally annotated gene sets (GS) that collectively represent almost all known biological and metabolic pathways in human cells. Significance was estimated with permutation test. Three separate cohorts on untreated HER2-normal patients (n=234, n=175, n=170) and 3 cohorts of patients with neoadjuvant therapy (n=198, n=85, n=61) were included in this analysis. We performed analysis of each data set separately and also as pooled data.Results: When data sets were analyzed individually, 753 to 938 GS were associated with prognosis (P≤0.05) in ER+ cancers including 186 common to all, and 408 to 579 GS were associated with prognosis in ER- cancers including 1 common to all and 141 common to at least two data sets. Within each data set, the number of overlapping GS between the ER+ and ER- prognostic lists ranged from 33 to 112. In the neoadjuvant data sets, 555 to 674 GS were associated with response in ER+ cancers (195 common to at least two, 4 common to all) and 543 to 730 in ER- cancers (268 common to two, 22 common to all). When the predictive GS for ER+ cancers were compared to those for ER- cancers, there were only 10 GS in common. A meta analyses of the combined prognostic data sets yielded 384 GS with p≤0.0001 in ER+ cancers and 47 in ER- cancers, respectively, with 12 GS common to both. A meta analyses of the combined neoadjuvant data sets yielded 92 GS associated with pCR in ER+ cancers and 110 GS in the ER- cancers. Only 2 GS were common to both. Functional analysis indicated that DNA replication, mitotic spindle checkpoint and plasma cell function were the most common prognostic pathways in ER+ cancer. T cell differentiation, glycolipid metabolism and immune functions were the most commonly prognostic pathways in ER- cancer with a notable absence of proliferation related gene sets. DNA replication, spindle and microtubule activity and cell cycle regulation were associated with chemotherapy response in ER+ cancers; oxidative stress, blood vessel formation in ER- cancers.Conclusion: These data indicate that prognosis and chemotherapy response are associated with different gene sets in ER+ and ER- cancers. Citation Information: Cancer Res 2009;69(24 Suppl):Abstract nr 6124.