Meta-analysis of microarray and RNA-Seq gene expression datasets for carcinogenic risk: An assessment of Bisphenol A

Abstract

Bisphenol A (BPA) is an endocrine-disrupting chemical that is related to many diseases, including heart attacks and diabetes. Recently, several studies have reported the carcinogenic potential of BPA in rodents, yet carcinogenic effects of BPA in humans remains unclear. In this study, meta-analysis was applied to independent GEO datasets, based on 158 Affymetrix microarrays and 8 Illumina RNA-Seqs. Additionally, we performed functional enrichment analysis, disease similarity analysis based on Disease Ontology (DO) analysis, and network analysis. 1,993 (1,457 up-, 536 down-regulated) differentially expressed genes (DEGs) were identified from five GEO datasets by adjusting for batch effects. Using disease similarity analysis, we demonstrated that results of DO analysis of the top 20 diseases were highly related to breast cancer. Moreover, we showed that the DEGs were significantly enriched in gene expression datasets on human breast cancer tissue via gene set enrichment analysis. By performing network analysis, we finally identified 85 (68 up- and 17 down-regulated) DEGs, and some of their expression levels were validated by quantitative PCR. The identified DEGs were regarded as genetic markers for carcinogenic risks, indicating that BPA may be a potential carcinogenic chemical contributing to the cause of breast cancer in humans.