Abstract
Metastasis, the fatal hallmark of breast cancer (BC), is a serious hurdle for therapy. Current prognostic approaches are not sufficient to predict the metastasis risk for BC patients. Therefore, in the present study, we analyzed gene expression data from GSE139038 and TCGA database to develop predictive markers for BC metastasis. Initially, the data from GSE139038 which contained 65 samples consisting of 41 breast tumor tissues, 18 paired morphologically normal tissues and 6 from non-malignant breast tissues were analyzed for differentially expressed genes (DEGs). DEGs were obtained from three different comparisons: paired morphologically normal (MN) versus tumor samples (C), apparently normal (AN) versus tumor samples (C), and paired morphologically normal (MN) versus apparently normal samples (AN). Multiple bioinformatic methods were employed to evaluate metastasis, EMT and triple negative breast cancer (TNBC) specific genes. Further, regulation of gene expression, clinicopathological factors and DNA methylation patterns of DEGs in BC were validated with TCGA datasets. Our bioinformatic analysis showed that 40 genes were upregulated and 294 were found to be downregulated between AN vs C; 124 were upregulated and 760 genes were downregulated between MN vs C; 4 were upregulated and 13 were downregulated between MN vs AN. Analysis using TCGA dataset revealed 18 genes were significantly altered in nodal positive BC patients compared to nodal negative BC patients. Our study showed novel candidate genes as predictive markers for BC metastasis which can also be used for therapeutic targets for BC treatment.