Oestrogen receptor phosphorylation profiles and in silico PAM50 subtyping reflect sexual dimorphism in breast cancer

Abstract

Breast cancer (BC) is most prevalent in females but also accounts for <1% of male cancer cases and 0.2% of male cancer-related deaths. Distribution of histological subtypes, receptor status, and age of diagnosis varies based on sex, and a growing body of evidence supports sex-specific molecular differences in BC. However, this is limited by the smaller number of male cases available for study compared to the thousands of cases of female BC. We combined publicly available male BC gene expression datasets for 195 patients from 4 studies and split randomly into discovery and validation sets. Clustering and gene expression analysis were performed. Two stable clusters were identified initially, confirmed in the validation set. Cluster C1 was enriched for genes associated with MAPK signalling and arylesterase activity. Cluster C2 showed enrichment of genes associated with proliferation, invasion, and metastasis, along with enrichment of gene ontology and pathway terms related to ECM regulation, particularly collagen-containing ECM. Of note, when stratified by ERα and PR status, no enrichment was observed with the predicted PAM50 classification. ERα and MAPK signalling were enriched in both clusters, albeit through different gene sets. Since these pathways were enriched, we investigated the signalling regulation of ERα based on immunohistochemical expression of phosphorylated ERα (S104, S118, S167, S294) and their prognostic roles. This analysis also revealed distinctions from female BC, showing a lack of prognostic outcome for any of these biomarkers. We show that male BC does not align with female BC in the same way that intrinsic subtypes of female BC are not identical. As BC heterogeneity is well recognised, we propose that male BC should be considered as a potentially unique clinical subtype of BC.