Breast cancer (BC) is highly prevalent and despite therapeutic advances, has the highest cancer-related mortality in women (Siegel et al., 2020). Although long-term all-round radiation therapy can enhance patient survival, there are undiscovered regional lymphatics within the radiation zone, and still, over 30% of BC patients develop distant metastasis, causing a high fatality rate (Moreno et al., 2017; Zhao et al., 2019). Therefore, exploring the molecular mechanisms associated with BC prognosis and treatment is of great significance. The Cancer Genome Atlas (TCGA, https://portal.gdc.cancer.gov/) and Gene Expression Omnibus (GEO, https://www.ncbi.nlm.nih.gov/geo) databases have high flux and efficiency and have been widely applied in various disease research gene chip platforms, including for lung (Cai et al., 2021), kidney (Xu et al., 2021), and colon cancers. In recent years, some bioinformatics assays on BC have been reported and different chip databases have been selected to investigate the potential mechanisms.

The Nottingham Prognosis Index (NIP) has been adopted for investigating gene features (Zhou et al., 2022), and machine learning modules have been utilized to select and confirm the prediction accuracy of biomarkers (Tabl et al., 2019). In this preliminary study, the integrated bioinformatics method is applied. We downloaded two gene chip datasets (GSE31448 and GSE42568) from the GEO database, consisting of 457 BC and 21 non-carcinoma mammary tissue samples. Then, differentially expressed genes (DEGs) were identified by adopting R statistical software from the aforementioned data, while the intersection of the aforementioned maps was obtained by Venn diagram. The survival of the DEGs was investigated using online KM-plotter. Online analysis using The Database for Annotation, Visualization and Integrated Discovery (DAVID), KOBAS, Gene Ontology (GO) analysis, and the Kyoto Encyclopedia of Gene and Genome (KEGG) pathway analysis was conducted to establish survival-associated DEGs. A survival-associated protein–protein interaction (PPI) network was constructed on the basis of String online tools and visualized by Cytoscape software and the cytoHubba plugin to detect hub genes. The top three degrees of hub genes (EZH2, CCNB1, and PPARG) were then selected and the TCGA_BRCA dataset was utilized to confirm them. Gene Set Enrichment Analysis (GSEA) was conducted to study the potential mechanisms of these hub genes. Having obtained credible findings, BC samples were used for verification of hub genes, with consistent results obtained. This work identified three BC survival-associated biomarkers, which help us to understand the molecular mechanisms of BC development.

The incidence of BC among female cancer patients is approximately 30%,—the highest for cancer—and its mortality rate ranks second. Similarly, the incidence of BC in China is up to 17.07%, which also ranks first but with a younger trend. BC has the highest fatality of all cancers in female patients worldwide, representing a major mortality factor for women (Cheng et al., 2020). Based on the newest cancer statistics, there will be 51,400 ductal carcinoma in situ cases diagnosed in BC women in the USA in 2022 (Siegel et al., 2022). Despite advances in diagnosis, BC is a huge threat to women due to the rapid progress of drug resistance and a lack of therapeutic target methods. Consequently, a sensitive and special biomarker for BC is urgently needed.

Our study involved two gene chips based on the GEO database. A total of 169 DEGs (30 up- and 139 downregulated) were selected. Due to the high BE mortality rate, early diagnostic markers play a vital role in favorable prognoses. Although many genes are recognized as specific diagnostic indicators for BC, clinical evidence is lacking. In this study, 105 (24 with upregulation and 81 with downregulation) survival-associated DEGs were identified by GEO database analysis. Based on GO and KEGG enrichment, upregulated survival-associated DEGs were the most enriched in “cell division” and “spindle,” “microtubule binding,” and “cell cycle” in, respectively, BP, CC, and MF segments and KEGG pathways. Downregulated survival-associated DEGs were most enriched in “cellular response to hormone stimulus” and “lipid droplet,” “norepinephrine binding,” and “Tyrosine metabolism” in, respectively, BP, CC, and MF terms and KEGG pathways. To determine which gene is the pivotal regulator for BC, a survival-associated PPI network was constructed, with three hub genes (EZH2, CCNB1, and PPARG) chosen using the cytoHubba plugin. These three hub genes were verified by the TCGA_BRCA dataset and tissue samples from BC patients. To investigate the underlying mechanism of hub genes, GSEA was conducted. VANTVEER_BREAST_CANCER_POOR_PROGNOSIS’ and ‘CELL_CYCLE’ gene sets were positively correlated with the three hub genes. We recognized these three genes to be the hub genes related to poor prognoses of BC. Multivariate and univariate analyses of genetic association studies and genome-wide association studies have received a remarkable attention as they improve analytical precision (Puddu et al., 1997; Dimou et al., 2018) From the multivariate and univariate analyses, the aforementioned genes are associated with patients’ clinical features, such as the TNM-based stage (Supplementary Figures S3, S4). The details of these genes are as follows.

The enhancer of zeste homolog 2 (EZH2) is a member of the polycomb gene (PcG) family. EZH2 is an important factor in regulating cell autophagy and apoptosis (Yao et al., 2016), which can promote DNA damage repair and the cell cycle (Nutt et al., 2020). This gene not only has a diversified role in cell function but is also related to many diseases, including cancer. An aberrant expression of EZH2 enhances cell proliferation, which may cause tumor development. Others have reported that EZH2 was correlated with poor clinical outcomes for prostate cancer (Varambally et al., 2002). More evidence shows that EZH2 is strongly denoted in numerous tumors, including esophageal cancer (Qiu et al., 2020), gastric cancer (Gan et al., 2018), and endometrial carcinoma (Krill et al., 2020), which also contains BC (Bachmann et al., 2006). An in vivo mice model experiment demonstrated that EZH2 is positively related to the lymph nodes and distant metastases (Zingg et al., 2015). EZH2 can thus act as a prognosis biomarker for BC. Cyclin B1 (CCNB1) is a cell cycle regulator which can form complexes with CCNB2 and CDK1 to regulate the G2/M phases of the cell cycle. The dysregulation of CCNB1 is associated with aberrant cell growth. Moreover, the high expression of CCNB1 shows positive association with unfavorable outcomes for non-small-cell lung, breast, gastric, and esophageal cancers (Roskoski 2019). Nevertheless, its precise functions in BC need elucidation. As the nuclear receptor, peroxisome proliferator-activated receptor gamma (PPARG) can antagonize the transcription of immune and inflammatory factors and is also involved in the energy metabolism (Medina-Gomez et al., 2007), adipocyte differentiation, and inflammation (Mirza et al., 2019; Villa et al., 2020). Studies have revealed that PPARG exerts a vital role in a number of tumors, including colorectal (Villa, Parra, Feitosa, Camargo, Machado, Tirapelli, Rocha and Feres 2020), renal (Fujita et al., 2011), and bladder cancers (Plissonnier et al., 2010). Some argue that PPARG expression inhibits the Wnt/beta-catenin pathway (Vallee et al., 2019) and is positively related to patient prognosis (Ogino et al., 2009).

To sum up, our study has identified three hub genes (EZH2, CCNB1, and PPARG) related to the prognosis and therapeutic target of BC based on comprehensive gene chip datasets combined with bioinformatic analyses. These findings provide us with a high perspicacity for gene therapy of BC. However, due to the limitations of the present study, more cell research and experiments are essential to study the roles of hub genes in BC and more experiments need to be conducted for validation.