Screening of Prognosis-Related Genes in Primary Breast Carcinoma Using Genomic Expression Data

This study aimed at exploring the genes that may be related to the prognosis of primary breast cancer (BC) patients. The gene expression microarray data, together with sample survival data were acquired from The Cancer Genome Atlas database. The top 20% genes according to expression value variance were subjected to hierarchical cluster analysis. Bootstrap methods were utilized to assess the stability of cluster. Cox regression was applied to screen genes related to the survival time of patients with BC, and the Beta-Uniform Mixture model was applied to adjust the significance of numerous tests. Further, ingenuity pathway analysis (IPA) was carried out to analyze the functions of the potential prognostic genes. Cluster analysis revealed that there were at least five stable BC subtypes, each with specific gene expression. Further, 42 survival time-associated genes were found (p-value = 0.0006, false discovery rate = 0.2) by Cox regression analysis. According to Gene Ontology (GO) functional annotation, genes in clusters A, B, C, D, and E separately were implicated in cell adhesion cooperation, cell stress response, cell cycle, the assembly of nucleosome and chromosome, and immune regulation. IPA results showed that prognosis-related genes mainly participated in the pathways of cell apoptosis, and cell communication and morphology. Genes such as JAK2, TBP, PTGES3, and RYBP may be promising prognostic biomarkers for BC patients.