Identification of markers associated with brain metastasis from breast cancer through bioinformatics analysis and verification in clinical samples

Background Brain metastasis from breast cancer (BC) is an important cause of BC-related death. The present study aimed to identify markers of brain metastasis from BC. Methods Datasets were downloaded from the public databases Gene Expression Omnibus (GEO) and The Cancer Genome Atlas (TCGA). Weighted gene co-expression network analysis (WGCNA) was performed to identify metastasis-associated genes (MAGs). Least absolute shrinkage and selection operator (LASSO) Cox proportional hazards regression models were constructed for screening key MAGs. Survival analysis and receiver operating characteristic (ROC) curves were used for evaluating the prognostic value. The factors associated with tumor metastasis were integrated to create a nomogram of TCGA data using R software. Gene Set Enrichment Analyses (GSEA) was performed for detecting the potential mechanisms of identified MAGs. Immunohistochemistry (IHC) was used to verify the expression of the key genes in clinical samples. Results The genes in 2 modules were identified to be significantly associated with metastasis through WGCNA. LASSO Cox proportional hazards regression models were constructed successfully. Subsequently, a clinical prediction model was constructed, and a nomogram was mapped, which had better sensitivity and specificity for BC metastasis. Two key genes, discs large homolog 3 (DLG3) and growth factor independence 1 (GFI1), were highly expressed in clinical samples, and the expression of these 2 genes was associated with patients' survival time. Conclusions We successfully constructed a clinical prediction model for brain metastasis from BC, and identified that the expression of DLG3 and GFI1 were strongly associated with brain metastasis from BC.