Abstract
BACKGROUND: Breast cancer (BC) is the most common cancer in women and accounts for approximately 15% of all cancer deaths among women globally. The underlying mechanism of BC patients with small tumor size and developing distant metastasis (DM) remains elusive in clinical practices. METHODS: We integrated the gene expression of BCs from ten RNAseq datasets from Gene Expression Omnibus (GEO) database to create a genetic prediction model for distant metastasis-free survival (DMFS) in BC patients with small tumor sizes (≤ 2 cm) using weighted gene co-expression network (WGCNA) analysis and LASSO cox regression. RESULTS: ABHD11, DDX39A, G3BP2, GOLM1, IL1R1, MMP11, PIK3R1, SNRPB2, and VAV3 were hub metastatic genes identified by WGCNA and used to create a risk score using multivariable Cox regression. At the cut-point value of the median risk score, the high-risk score (≥ median risk score) group had a higher risk of DM than the low-risk score group in the training cohort [hazard ratio (HR) 4.51, p < 0.0001] and in the validation cohort (HR 5.48, p = 0.003). The nomogram prediction model of 3-, 5-, and 7-year DMFS shows good prediction results with C-indices of 0.72-0.76. The enriched pathways were immune regulation and cell-cell signaling. EGFR serves as the hub gene for the protein-protein interaction network of PIK3R1, IL1R1, MMP11, GOLM1, and VAV3. CONCLUSION: Prognostic gene signature was predictive of DMFS for BCs with small tumor sizes. The protein-protein interaction network of PIK3R1, IL1R1, MMP11, GOLM1, and VAV3 connected by EGFR merits further experiments for elucidating the underlying mechanisms.