Breast cancer has emerged as the leading cause of high incidence and mortality among women, surpassing lung cancer in recent years [25]. However, the pathogenesis of breast cancer remains unclear and may be influenced by factors such as the patient’s age, lifestyle habits, family history, genetic mutations, and hormonal levels. Therefore, investigating the molecular mechanisms underlying the physiology and pathology of breast cancer is essential for a comprehensive understanding of the disease’s occurrence, progression, and prognosis, as well as for informing clinical judgment and treatment strategies. High-throughput microarray and bioinformatics approaches have enabled researchers to explore genetic alterations in breast cancer, proving to be effective methods for identifying new biomarkers across various diseases. In this study, our short-term goal is to conduct follow-up experiments based on the results of bioinformatics analysis and preliminary experiments, while our long-term objective is to identify molecular markers that can aid in the prediction and treatment of breast cancer.

It is possible for the analysis of a single microarray dataset to overfit the data, resulting in low training error but high test error. For this reason, this study conducted a comprehensive analysis of 3 microarray datasets (GSE86374, GSE120129, and GSE29044), which included 226 breast cancer (BC) tissue samples and 175 normal breast tissue samples. Consequently, 323 differentially expressed genes (DEGs) were identified across the three microarray datasets, comprising 139 down-regulated genes and 184 up-regulated genes. Gene Ontology (GO) analyses indicated that the alterations in the gene modules primarily focused on cell division, extracellular regions, spindles, extracellular matrices, and extracellular matrix structural constituents. Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis of the DEGs revealed that they were mainly associated with signaling molecules and interactions, as well as cell growth and apoptosis. Additionally, 37 hub genes were identified as the most significant components within the protein–protein interaction (PPI) network. Our findings suggest that these enhanced modules and pathways have a genetic impact on breast cancer. Furthermore, the predictive genes within these datasets may be interconnected and jointly regulate breast cancer within the network.

We identified three key genes associated with very low survival rates in breast cancer (BC) across these three datasets: RACGAP1, SPAG5, and KIF20A. RACGAP1 (Rac GTPase-activating protein 1) is a component of the central spindle protein complex and encodes a GTPase-activating protein. This protein interacts with the active form of Rho GTPases and promotes GTP hydrolysis, which can lead to the negative regulation of Rho-mediated signal transduction. Numerous studies have indicated that RACGAP1 is a highly expressed gene linked to poor prognosis in several types of human cancer [26,27,28]. Bioinformatics analyses have demonstrated that RACGAP1 is significantly overexpressed in BC tissues associated with poor prognosis [29]. Additionally, analyses of online databases revealed a substantial increase in RACGAP1 mRNA expression across various tumor tissues [30, 31], which aligns with our findings.

SPAG5 (sperm-associated antigen 5) encodes a protein that may play a crucial role in the function and dynamic regulation of mitotic spindles. The SPAG5 gene is overexpressed in various types of tumor tissues, and elevated expression levels have been associated with poorer prognoses, as indicated by bioinformatics analyses. It has been observed that SPAG5 exhibits carcinogenic activity as a spindle-related protein during mitosis in both solid tumors and hematological malignancies [32,33,34]. In triple-negative breast cancer (TNBC), the expression levels of RAD51, BRCA1, and BRCA2 have been positively correlated with SPAG5 expression [35].

KIF20A, a member of the kinesin family of motor proteins, is localized in several cellular components, including the cleavage furrow, intercellular bridge, and midbody. It exhibits protein kinase binding activity and plays a crucial role in the formation of microtubule bundles, midbody abscission, and the regulation of cytokinesis. The expression of KIF20A has been associated with overall survival rates in various human cancers, including kidney cancer, prostate cancer, fibrosarcoma, and hepatocellular carcinoma [36,37,38,39]. Additionally, KIF20A can bind to microRNAs, which may influence the efficacy of chemotherapy in breast cancer cells [40].

Through a series of bioinformatics methods, we identified three genes: RACGAP1, SPAG5, and KIF20A. We conducted a preliminary validation of these genes using immunohistochemistry (IHC), and the results corroborated our bioinformatics analysis. Further literature searches revealed that the interactions between breast cancer and these key genes have not been extensively documented.

Despite the valuable insights provided by this study, it has several limitations. Firstly, the experiments conducted utilized immunohistochemistry, which revealed weak labeling of the three selected biomarkers. To achieve more effective detection of protein expression, we recommend employing methods such as Western blotting (WB) or immunofluorescence assays. While immunofluorescence is generally more effective due to the incorporation of fluorescent dye labeling, which produces stronger signals than immunohistochemistry, the research team in this study was not qualified to utilize this method. Secondly, the mechanisms by which these biomarkers influence breast cancer (BC) were not clarified, indicating that the relationships of these genes with BRCA1, BRCA2, and other BC-related genes warrant further investigation. Thirdly, although bioinformatics analysis is a powerful tool for understanding molecular mechanisms and identifying potential biomarkers, additional experimental validation is needed at the molecular, cellular, and biological levels. Fourthly, the roles these genes play in diagnosis were not adequately explored. Consequently, further investigations should focus on elucidating the underlying mechanisms connecting BC and these key genes.