Identification of putative gene signatures associated with diagnosis and prognosis of breast cancer

Purpose: This study aimed to identify potential diagnostic and therapeutic biomakers for the development of breast cancer (BC). Methods: GSE86374 dataset containing 159 samples was acquired from the Gene Expression Omnibus (GEO) database followed by differentially expressed genes (DEGs) identification and cluster analysis. Corresponding functional enrichment and protein-protein interaction (PPI) network analyses were performed to identify hub genes. Prognostic evaluation using clinical information obtained from TCGA database and hub genes was conducted to screen for crucial indicators for BC progression. The risk model was established and validated. Results: In total, 186 DEGs were identified and grouped into four clusters: 96 in cluster 1; 69 in cluster 2; 16 in cluster 3; and 5 in cluster 4. Functional enrichment analysis showed that DEGs, including ADH1B in cluster 1, were dramatically enriched in the tyrosine and drug metabolism pathways, while genes in cluster 2, including SPP1 and RRM2, played crucial roles in PI3K-Akt and p53 signalling pathway. SPP1 and RRM2 served as hub genes in the PPI network, resulting in an support vector machine classifier with good accuracy and specificity. Additionally, the results of prognostic analysis suggest that age, metastasis stage, SPP1 and ADH1B were correlated with risk of BC, which was validated by using the established risk model analysis. Conclusion: SPP1, RRM2 and ADH1B appear to play vital roles in the development of BC. Age and TNM stage were also preferentially associated with risk of developing BC. Evaluation of the risk model based on larger sample size and further experimental validation are required.