1. Introduction
Breast cancer (BC) is the most common cancer in women and the second-leading cause of cancer-related deaths among women worldwide. About one in eight US women will develop BC during their life, and the lifetime risk of dying from the malignant disease is 3.4%.[1] Early diagnosis of BC is important for timely targeted treatment and for a higher chance of long-term survival.[2] Accordingly, exploring more reliable prognostic biomarkers is a challenging task that could benefit greatly from the development of bioinformatics approaches.[3] In this context, cancer research has seen a growing concern in high-throughput technologies, whose integration has raised new hopes for improving cancer treatment.
In the last few decades, progress in targeted therapy has provided more treatment opportunities and prolonged survival times for BC patients.[4,5] Progress in microarray and RNA-sequencing technology has expanded our understanding of the complex and dynamic nature of the transcriptome.[6] Numerous studies have reported the importance of comparative transcriptome analysis in BC to assess the expression levels of functional genes, key pathways, and activity of cellular regulatory mechanisms,[7–9] and, subsequently, provide new insights into the prediction of disease prognosis and the discovery of targeted anticancer therapies.
The results of this study are expected to provide more insights into the identification of reliable molecular biomarkers and potential therapeutic targets for BC.
2. Material and methods
2.1. Ethics statement
Ethical approval was waived for the present study because all datasets were downloaded from a publicly available database (https://www.ncbi.nlm.nih.gov/geo/).
2.2. Microarray data source and information
The microarray data used in this study were downloaded from the public functional genomics data repository, Gene Expression Omnibus (GEO) database (http://www.ncbi.nlm.nih.gov/gds/). Three gene expression profiles (GSE55715, GSE124646, and GSE87049) of BC and normal breast tissue were chosen. The array data of GSE55715 consisted of five BC and three normal breast tissues, based on the GPL6947 platform. The array data of GSE124646 consisted of ten BC (100% BC) and 10 normal breast tissues, based on the GPL96 platform. The array data of GSE87049 included 124 BC and 35 normal adjacent tissues, based on the GPL6244 platform.
2.3. Data processing and identification of DEGs
The three mRNA expression microarray datasets obtained from the GEO database were analyzed using the GEO2R online tool (https://www.ncbi.nlm.nih.gov/geo/geo2r/).[10] The |log2-fold change (FC)|˃|1 and P value < .05 were conducted as the cutoff criteria for the differentially expressed genes (DEG). Then, the raw data in TXT format were selected using Venn diagram software (http://bioinformatics.psb.ugent.be/webtools/Venn/)[11] to obtain the common DEGs among the three datasets. DEGs with log FC > 0 were considered upregulated genes, while DEGs with log FC < 0 were considered downregulated genes (see Figure S1, Supplemental Digital Content, https://links.lww.com/MD/I681, which illustrates the flow chart of this study).
2.4. Gene ontology and pathway enrichment analysis
Gene ontology (GO) analysis is a commonly used approach for functional studies of high-throughput transcriptome or genome data. GO and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses were performed using the online bioinformatics tool Database for Annotation, Visualization, and Integrated Discovery (DAVID) software (version 6.8, https://david.ncifcrf.gov/).[12]P < .05 was set as the cutoff criterion.
2.5. PPI network and module analysis
To explore the interactions of DEGs, we used an online database (version 11.5, https://string-db.org/). A high confidence level (0.700) was chosen as the minimum required interaction score. After removing the disconnected nodes, protein-protein interaction (PPI) networks were visualized using Cytoscape 3.9.0 software.[12] In addition, the Molecular Complex Detection plug-in of Cytoscape software was used to screen modules of the PPI network (degree cutoff = 2, node score cutoff = 0.2, k-core = 2, and max. depth = 100).
2.6. Selection of hub genes
Hub genes were extracted using the Cytoscape plug-in CytoHubba software.[13] Then, the top ten essential hub genes were selected and ranked by Maximal Clique Centrality and Degree topological algorithms.
2.7. Survival analysis and expression validation of hub genes
The online database Kaplan–Meier Plotter (www.kmplot.com)[14] was utilized to assess the effect of ten core genes on the overall survival (OS) of BC patients. To estimate the OS rate of patients with BC, samples were split into high- and low-expression groups based on the median expression. The hazard ratios (HR), corresponding 95% confidence intervals, and log-rank P values were calculated and displayed on the plot. The GEPIA2 (http://gepia2.cancer-pku.cn/) database tool[15] was used to analyze the RNA sequencing expression data between BC and control samples (|Log2FC| Cutoff = 1, and P value cutoff = .05).
2.8. Transcription factor target gene
Transcription factors of hub genes derived from the ENCODE ChIP-seq data were determined using the NetworkAnalyst database (https://www.networkanalyst.ca/).[16] Then, the Genotype-Tissue Expression Project (GTEx) database and breast tissue were selected to filter genes by tissue specificity. Only peak intensity signals < 500 and predicted regulatory potential scores < 1 were used, using the BETA Minus algorithm. Finally, the transcription-regulated network was visualized using Cytoscape software.
2.9. BC immune infiltration cells
TIMER (https://cistrome.shinyapps.io/timer/)[17] was used to estimate the infiltration of the six types of immune cells in BC. The Somatic Copy Number Alteration (SCNA) module of TIMER provides a comparison of the abundance of immune cells among tumors with different somatic copy number variations (deep deletion, arm-level deletion, diploid/normal, arm-level gain, and high amplification) for a given gene.
3. Results
3.1. DEGs identification
DEGs were screened among each microarray dataset using GEO2R online tools with P < .05 and |logFC| ≥ 1.5. This allowed the extract of 4240, 723, and 1726 DEGs from GSE55715, GSE87049, and GSE124646, respectively. The Venn diagram software was used to identify common DEGs in the three GEO datasets. Overall, 146 common DEGs, including 60 upregulated genes (logFC ˃ 0) and 86 downregulated genes (logFC ˂ 0), were found in the three datasets (Fig. 1 and Table 1).
Table 1 -
List of DEGs detected from three profile datasets.
| DEGs |
Genes names |
| Up-regulated (60) |
TPX2 KPNA2 COL1A1 CDK1 RACGAP1 AURKA KIF14 MAD2L1 IFI27 KIF4A SQLE OLR1 TYMS MELK CCNA2 BGN STAT1 PTTG1 CKS2 ECT2 KIF23 ARMT1 CENPE LMNB1 CCNB2 CEP55 OAS2 MCM4 DLGAP5 CENPI KIAA0101 HSD17B6 NEK2 IFIT3 MMP13 PLK1 EZH2 KIF11 KIFC1 PCNA MATN3 KIF2C GPRC5A LRRC15 CDC20 BUB1 SHCBP1 ASPM BST2 ATAD2 UBE2C RRM2 TOP2A SPC25 KIF18A HMMR KIF20A CDKN3 NCAPG CENPF |
| Down-regulated (86) |
BBOX1 EDN1 CNN1 GHR MAOA TRIM29 MYBPC1 OXTR ZBTB20 HLF SPRY2 KRT15 PLSCR4 PDGFD PI15 KIT PTGS2 PIK3C2G PDK4 FAM13A NR3C2 LALBA MT1E NTRK2 APOD GABRP HBB TFPI ALDH1A3 TGFBR3 AK5 SOSTDC1 PIK3R1 SYNM CFD PTN MPPED2 ANXA1 CRYAB CD36 PIGR MT1M MID1 TMEM47 ZBTB16 EDNRB SLPI PAK3 PER2 MT1X SLC6A14 CSRNP3 RGS2 KRT14 CLCA4 PLCB1 SDPR SEPP1 NFIB ACKR1 TAT OPRPN LIFR CX3CL1 LDB2 LTF ITM2A FAM189A2 COL17A1 FMO2 ACTG2 TGFBR2 IL33 PLEKHS1 ELF5 FABP4 CLDN8 PDLIM3 GPM6B KRT5 SEMA5A RELN ADAMTS5 SFRP1 JCHAIN TP63 |
DEGs = differentially expressed genes.
;)
Figure 1.: Identification of common DEGs in the three datasets (GSE55715, GSE87049, and GSE124646) through Venn diagrams software. A, 60 DEGs were up-regulated in the three datasets (logFC > 0). B, 86 DEGs were downregulated in three datasets (logFC < 0). DEGs = differentially expressed genes.
3.2. Gene ontology and KEGG pathway analysis of DEGs
To investigate the biological functional roles of the 146 DEGs, GO enrichment analysis was performed using DAVID software. Only 87 genes (45 upregulated and 42 downregulated) were found to be involved in the three major attributes associated with gene products. The biological processes of upregulated DEGs were mostly concentrated in microtubule-based movement, sister chromatid cohesion, mitotic metaphase plate congression, and anaphase-promoting complex-dependent catabolic process, while downregulated DEGs were in response to estradiol, estrous cycle, response to drugs, and negative regulation of growth (see Table S1, Supplemental Digital Content, https://links.lww.com/MD/I682, which shows the gene ontology analysis of DEGs in BC). The cellular components of the upregulated genes were mainly enriched in kinesin complex, condensed chromosome kinetochore, spindle microtubule, and cell-cell adherens junction, and downregulated DEGs in intermediate filament, proteinaceous extracellular matrix, and perinuclear region of the cytoplasm. The molecular functions of upregulated DEGs were concentrated in microtubule motor activity, ATP-dependent microtubule motor activity plus-end-directed, microtubule binding, and protein kinase activity; however, the downregulated DEGs were enriched in heparin binding, structural constituent of cytoskeleton, double-stranded DNA binding, and sequence-specific DNA binding.
Moreover, to gain more insight into the enriched pathways, KEGG pathway analysis was performed using the DAVID tool. Upregulated genes were enriched in six pathways, including cell cycle, oocyte meiosis, progesterone-mediated oocyte maturation, p53 signaling pathway, HTLV-I infection, and hepatitis B. Eleven pathways were identified in downregulated genes, including phenylalanine metabolism, tyrosine metabolism, melanogenesis, mineral absorption, TNF signaling pathways, and pathways in cancer (see Table S2, Supplemental Digital Content, https://links.lww.com/MD/I683, which shows the KEGG pathway analysis of DEGs).
3.3. PPI interaction network and module analysis
To investigate the associations between DEGs, the STRING database was used to establish a PPI network. The network was visualized using Cytoscape software, which showed 87 nodes and 512 edges, including 45 upregulated and 42 downregulated genes (Fig. 2A). Furthermore, the Cytoscape Molecular Complex Detection plugin analysis showed a significant module including 29 nodes and 392 edges within the PPI network of DEGs (Fig. 2B). Finally, the cytoHubba plug-in was used to filter out ten genes, including BUB1, HMMR, MAD2L1, CEP55, AURKA, CCNB2, TPX2, MELK, KIF20A, and CDK1 (Fig. 2C).
Figure 2.: PPI network constructed by STRING online database and visualized by Cytoscape. (A) PPI network without isolated or partially disconnected nodes. (B) Module analysis via Cytoscape software (degree cutoff = 2, node score cutoff = 0.2, k-core = 2, and max. Depth = 100). (C) Selection of hub genes using the plug-in cytoHubba of Cytoscape. All of them were upregulated genes. PPI = protein-protein interaction.
3.4. Survival analysis and expression validation
The overall survival of hub genes was then analyzed using the Kaplan–Meier Plotter database. As shown in Figure 3, the analyses suggested that nine genes, BUB1, HMMR, MAD2L1, CEP55, AURKA, CCNB2, TPX2, MELK, and KIF20A, had significantly worse survival, while only CDK1 was not significant (P < .05). Using the GEPIA2 dataset, we found that the mRNA expression of the nine hub genes was higher in BC tissues than in normal breast tissues (Fig. 4).
Figure 3.: Kaplan–Meier overall survival analysis for 10 hub genes expressed in BC patient samples. Nine of the ten genes had a significantly worse survival rate (P < .05). AURKA = aurora kinase, BC = breast cancer, BUB1 = BUB1 mitotic checkpoint serine/threonine kinase, CCNB2 = cyclin B2, CDK1 = cyclin-dependent kinase 1, CEP55 = centrosomal protein 55, HMMR = hyaluronan-mediated motility receptor, HR = hazard ratio, KIF20A = kinesin family member 20A, MAD2L1 = mitotic arrest deficient 2 like 1, MELK = maternal embryonic leucine zipper kinase, TPX2 = TPX2 microtubule nucleation factor.
Figure 4.: Expression profiles of hub genes in BC patients (GEPIA2). The red box represents BC samples, the gray box represents normal samples. AURKA = aurora kinase, BC = breast cancer, BUB1 = BUB1 mitotic checkpoint serine/threonine kinase, CCNB2 = cyclin B2, CEP55 = centrosomal protein 55, HMMR = hyaluronan-mediated motility receptor, KIF20A = kinesin family member 20A, MAD2L1 = mitotic arrest deficient 2 like 1, MELK = maternal embryonic leucine zipper kinase, N = normal, TPX2 = TPX2 microtubule nucleation factor.
3.5. Transcription factor and gene target data
The transcription-regulated network of hub genes included 153 nodes and 214 edges (Fig. 5). Results showed that TPX2 had 70 connections, followed by HMMR and KIF201 with 39 and 28 connections, respectively. Several transcription factors (TFs) implicated in the regulation of hub genes are mainly involved in multiple molecular functions, including histone demethylation, DNA-binding activity, and transcription regulation (Table 2).
Table 2 -
The
transcription factors of hub genes and their main function. Only TFs having more than 2 connections were represented.
| TFs |
Main function |
Regulated hub genes |
| KDM5A |
Histone demethylation |
TPX2, HMMR, CEP55, AURKA, MELK, KIF20A |
| ETV4 |
Transcription activation |
TPX2, HMMR, CEP55, MAD2L1 |
| GTF2E2 |
Initiation of RNA transcription |
HMMR, AURKA, MELK, KIF20A |
| RFX1 |
Activates transcription from class II MHC promoters |
AURKA, KIF20A, MAD2L1, CCNB2 |
| ZNF2 |
Transcription regulation |
HMR, CEP55, BUB1, MAD2L1 |
| ZBTB11 |
Transcription regulation |
AURKA, BUB1, MAD2L1, CCNB2 |
| CHD1 |
Control the cellular machinery’s access to DNA |
TPX2, HMMR, KIF20A |
| NFRKB |
Chromatin remodeling molecule |
TPX2, HMMR, CCNB2 |
| SMAD5 |
Transcription modulation |
TPX2, AURKA, HMMR |
| RFXANK |
Activates transcription from class II MHC promoters |
TPX2, AURKA, MAD2L1 |
| ZNF382 |
Sequence-specific transcription repression |
KIF20A, CCNB2, MAD2L1. |
| E2F4 |
Transcription activation, cell cycle regulation |
BUB1, MAD2L1, CEP55 |
CHD1 = chromodomain helicase DNA binding protein 1, E2F4 = E2F transcription factor 4, ETV4 = ETS variant transcription factor 4, GTF2E2 = general transcription factor IIE subunit 2, KDM5A = lysine demethylase 5A, NFRKB = nuclear factor related to kappaB binding protein, RFX1 = regulatory factor X1, RFXANK = regulatory factor X associated ankyrin containing protein, SMAD5 = SMAD family member 5, TFs = transcription factors, ZBTB11 = zinc finger and BTB domain containing 11, ZNF2 = zinc finger protein 2, ZNF382 = zinc finger protein 382.
Figure 5.: Network of transcription factors and hub genes. The red nodes represent hub genes, the yellow nodes represent transcription factors, and the green ones represent TFs that are associated with more than 2 hub genes. AURKA = aurora kinase, BUB1 = BUB1 mitotic checkpoint serine/threonine kinase, CCNB2 = cyclin B2, CEP55 = centrosomal protein 55, HMMR = hyaluronan-mediated motility receptor, KIF20A = kinesin family member 20A, MAD2L1 = mitotic arrest deficient 2 like 1, MELK = maternal embryonic leucine zipper kinase, TF = transcription factors, TPX2 = TPX2 microtubule nucleation factor.
3.6. BC immune infiltration cells
The abundance of six types of infiltrating immune cells, including B Cells, CD8+ T Cells, CD4+ T Cells, macrophages, neutrophils, and dendritic cells, was estimated to predict the influence of immune cell infiltration on the progression and clinical outcomes of BC. The results revealed that the expression of candidate hub genes was significantly correlated with the infiltration levels of CD8+ T cells, B cells, neutrophils, dendritic cells, macrophages, and CD4+ T cells (P < .05) (Fig. 6).
Figure 6.: The correlations between the copy number of hub genes expression and tumor immune cell infiltration levels in BC tissues. CD8+, CD8+ T cells, CD4+, CD4+ T cells. P value significant codes: ***<.001, **<.01, *<.05, ns < 0.1. AURKA = aurora kinase, BUB1 = BUB1 mitotic checkpoint serine/threonine kinase, CCNB2 = cyclin B2, CEP55 = centrosomal protein 55, DC = dendritic cells, HMMR = hyaluronan-mediated motility receptor, KIF20A = kinesin family member 20A, MAD2L1 = mitotic arrest deficient 2 like 1, Macr = macrophages, MELK = maternal embryonic leucine zipper kinase, Neut = neutrophils, TPX2 = TPX2 microtubule nucleation factor.
4. Discussion
Breast cancer is a malignant disease that differs greatly among patients, and its therapeutic approaches should be based on the molecular characteristics of BC. Therefore, identifying molecular biomarkers that can predict tumor behavior is particularly important to provide prognostic information and predict specific therapies.
In the present study, integrated bioinformatics analysis was performed to identify common DEGs expressed in BC. To investigate the biological significance of these DEGs, GO and pathway analyses were performed using the DAVID tool. The results showed that 60 upregulated genes were mainly enriched in microtubule-based movement, sister chromatid cohesion, mitotic metaphase plate congression, and G2/M transition of mitotic cell cycle, and the most common downregulated genes were significantly enriched in response to estradiol, estrous cycle, and response to drugs. These findings are in line with previous reports showing that microtubule-based movement, mitotic metaphase plate congression, establishment of sister chromatid cohesion during the S phase, and the G2/M transition are important processes that lead to cell division.[18–21] KEGG pathway analysis and PPI interaction network indicated that nine upregulated genes, including BUB1, HMMR, MAD2L1, CEP55, AURKA, CCNB2, TPX2, MELK, and KIF20A, were highly connected and especially enriched in cell cycle regulation, oocyte meiosis, progesterone-mediated oocyte maturation, p53 signaling pathway, HTLV-I infection, and hepatitis B. These results are consistent with the evidence that an uncontrolled cell cycle is closely associated with cancer and always depends on the functional disruption of at least one of its checkpoint pathways.[22] Several recent studies have shown a link between estrogens and the initiation and progression of BC. Moreover, most human BCs begin to be estrogen-dependent and express estrogen receptors.[23,24] Furthermore, the relationship between HTLV-1 infection and cancer progression is still controversial, but no reports have shown a significant difference in BC incidence between infected and uninfected HTLV-1-patients.[25] Additional in-depth analyses are required to investigate the roles of these biological processes in BC.
Over the last decade, multiple studies have shown the overexpression of hub genes in multiple human cancers, including BC, and their involvement in BC progression and metastasis, which constitute the main event related to poor prognosis and survival.[26–32] The overexpression of BUB1 and MAD2L1 in BC is associated with early metastasis and leads to chromosomal instability, which could help for the prognosis and diagnosis of BC,[33] and they considered as a target for developing new anti-breast cancer therapies.[34] HMMR was previously identified as a member of BRCA1-protein complexes, including AURKA and MAD1L1, which are involved in mitotic centrosome function.[31] Most importantly, AURKA-related proteins, including TPX2,[35] are associated with key molecules in cancer-related pathways, such as p53, PI3K-Akt, and Wnt pathways.[36] Wang et al[37] indicated that CCNB2 is upregulated in human cancers, and its inhibition leads to cell cycle arrest. In addition, CCNB2 and p53 function antagonistically in regulating AURKA-mediated centrosome separation and correcting chromosome segregation in the late G2 phase by inducing AURKA phosphorylation.[38] Additionally, CEP55 has been shown to regulate the proliferation and invasion of various types of cancer via the PI3K/Akt signaling pathway; thus, it can be used as a prognostic marker and a putative therapeutic target.[39] The interactions of hub genes in regulating cancer progression may explain the protein-protein interactions between hub genes involved in cell cycle regulation and BC progression, as shown in Figure 2. Giuliano et al[32] suggested that MELK, an oncogene, is an important regulator of tumorigenesis in BC, and if silenced, leads to apoptosis in basal-like breast cancer cells, indicating that BC-associated changes in MELK expression are of functional importance[40] thus suggesting it as a novel targeted cancer therapy. The overexpression of KIF20A has been reported to be associated with poor prognosis and tumor development in patients with BC, as well as promoting the proliferation of colorectal cancer via the JAK/STAT3 signaling pathway.[41]
Increasing evidence suggests that the immune microenvironment plays an important role in BC development and progression.[42] The composition and intensity of immune cell subsets vary among various cancer types.[43] Ostrand-Rosenberg[44] reported that both macrophages and CD4+ cells contribute to tumor destruction or promote tumor growth. Furthermore, BC progression is characterized by increased immune cell infiltrates in the tumor, including CD4+, CD8+, cytotoxic T cells, B cells, macrophages, and dendritic cells.[45] These indications explain the results of the current study that show a positive correlation between hub gene expression and the infiltration levels of immune cells, especially macrophages and CD4+ T cells. Additional studies are required to elucidate the mechanisms controlling the distribution and behavior of immune cells in BC subtypes.
A TFs network of hub genes was constructed to explore the molecular networks associated with BC. These TFs are mainly involved in transcriptional regulation and chromatin remodeling. These findings are consistent with those of previous studies demonstrating that transcription and chromatin remodeling factors are overexpressed in multiple human cancers and are associated with poor prognosis.[46,47] However, more investigations are needed to understand TFs and hub genes interactions in regulating carcinogenesis.
The present study has several limitations. First, experimental verification of the selected genes is lacking. The datasets were analyzed using bioinformatic tools and were not validated by reverse-transcription quantitative PCR. Second, although the data format and labels are consistent across datasets, inter-study differences may still exist because of the variation in microarray platforms, study design, and measurement errors. Third, only three datasets were chosen for the study, which may constrain the generalizability of the results. It may be necessary to compare a wider range of datasets to draw general conclusions. Therefore, future studies with experimental verification and a larger number of datasets are required.
In summary, the present study identified a set of nine core genes, including BUB1, HMMR, MAD2L1, CEP55, AURKA, CCNB2, TPX2, MELK, and KIF20A, which were partially different from those identified in previous BC biomarker studies. All of these genes were significantly associated with immune infiltration in BC, and their pathways were mainly enriched in BC development. These findings offer key insights into BC progression and the development of new anticancer therapies.
Author contributions
Conceptualization: Abdelkader Oumeddour.
Data curation: Abdelkader Oumeddour.
Formal analysis: Abdelkader Oumeddour.
Investigation: Abdelkader Oumeddour.
Methodology: Abdelkader Oumeddour.
Project administration: Abdelkader Oumeddour.
Resources: Abdelkader Oumeddour.
Software: Abdelkader Oumeddour.
Supervision: Abdelkader Oumeddour.
Validation: Abdelkader Oumeddour.
Visualization: Abdelkader Oumeddour.
Writing – original draft: Abdelkader Oumeddour.
Writing – review & editing: Abdelkader Oumeddour.
References
[1]. Libson S, Lippman M. A review of clinical aspects of
breast cancer. Int Rev Psychiatry. 2014;26:4–15.
[2]. Goldhirsch A, Winer EP, Coates AS, et al.; Panel members. Personalizing the treatment of women with early
breast cancer: highlights of the St Gallen International Expert Consensus on the primary therapy of early
breast cancer 2013. Ann Oncol. 2013;24:2206–23.
[3]. Alhammad R. Bioinformatics identification of TUBB as potential prognostic biomarker for worse prognosis in ERα-positive and better prognosis in ERα-negative
breast cancer. Diagnostics (Basel). 2022;12:2067.
[4]. Lev S. Targeted therapy and drug resistance in triple-negative
breast cancer: the EGFR axis. Biochem Soc Trans. 2020;48:657–65.
[5]. Mehmood S, Faheem M, Ismail H, et al.
Breast cancer resistance likelihood and personalized treatment through integrated multiomics. Front Mol Biosci. 2022;9:783494.
[6]. Wang J, Dean DC, Hornicek FJ, et al. RNA sequencing (RNA-Seq) and its application in ovarian cancer. Gynecol Oncol. 2019;152:194–201.
[7]. Cheng YH, Chen YC, Lin E, et al. Hydro-Seq enables contamination-free high-throughput single-cell RNA-sequencing for circulating tumor cells. Nat Commun. 2019;10:2163.
[8]. Yang X, Kui L, Tang M, et al. High-throughput transcriptome profiling in drug and biomarker discovery. Front Genet. 2020;11:19.
[9]. Kui L, Kong Q, Yang X, et al. High-throughput in vitro gene expression profile to screen of natural herbals for
breast cancer treatment. Front Oncol. 2021;11:684351.
[10]. Barrett T, Wilhite SE, Ledouxet P, et al.
NCBI GEO: archive for functional genomics data sets-update. Nucleic Acids Res. 2013;41:D991–5.
[11]. Feng H, Gu ZY, Li Q, et al. Identification of significant genes with poor prognosis in ovarian cancer via bioinformatical analysis. J Ovarian Res. 2019;12:35.
[12]. Wang Z, Meng L, Shan J, et al. To study the mechanism of Scutellariae Radix and Astragaloside in the treatment of lung cancer based on network pharmacology. Medicine (Baltimore). 2022;101:e29729.
[13]. Chin CH, Chen SH, Wu HH, et al. cytoHubba: identifying hub objects and sub-networks from complex interactome. BMC Syst Biol. 2014;8(Suppl 4):S11.
[14]. Tian L, Liao Y. Identification of G6PC as a potential prognostic biomarker in hepatocellular carcinoma based on bioinformatics analysis. Medicine (Baltimore). 2022;101:e29548.
[15]. Tang Z, Kang B, Li C, et al. GEPIA2: an enhanced web server for large-scale expression profiling and interactive analysis. Nucleic Acids Res. 2019;47:W556–60.
[16]. Zhou G, Soufan O, Ewald J, et al. NetworkAnalyst 3.0: a visual analytics platform for comprehensive gene expression profiling and meta-analysis. Nucleic Acids Res. 2019;47:W234–41.
[17]. Yang X, Yang F, Lan L, et al. Potential value of PRKDC as a therapeutic target and prognostic biomarker in pan-cancer. Medicine (Baltimore). 2022;101:e29628.
[18]. Jin G, Wang W, Cheng P, et al. DNA replication and sister chromatid cohesion 1 promotes breast carcinoma progression by modulating the Wnt/β-catenin signaling and p53 protein. J Biosci. 2020;45:127.
[19]. Jamasbi E, Hamelian M, Hossain MA, et al. The cell cycle, cancer development and therapy. Mol Biol Rep. 2022;49:10875–83.
[20]. Maiato H, Gomes AM, Sousa F, et al. Mechanisms of chromosome congression during mitosis. Biology (Basel). 2017;6:13.
[21]. Wang I, Thomas HR, Chen Y, et al. Reduced sister chromatid cohesion acts as a tumor penetrance modifier. PLoS Genet. 2022;18:e1010341.
[22]. Matthews HK, Bertoli C, de Bruin RAM. Cell cycle control in cancer. Nat Rev Mol Cell Biol. 2022;23:74–88.
[23]. Parida S, Sharma D. The microbiome-estrogen connection and
breast cancer risk. Cells. 2019;8:1642.
[24]. Mohanty SS, Mohanty PK. Obesity as potential
breast cancer risk factor for postmenopausal women. Genes Dis. 2019;8:117–23.
[25]. Hirata M, Shinden Y, Nagataet A, et al. Clinical features of
breast cancer patients with human T-cell lymphotropic virus type-1 infection. Asian Pac J Cancer Prev. 2019;20:1909–12.
[26]. Lin X, Zhou M, Xu Z, et al. Bioinformatics study on genes related to a high-risk postoperative recurrence of lung adenocarcinoma. Sci Prog. 2021;104:368504211018053.
[27]. Carvalho AM, Soares da Costa D, Reis RL, et al. RHAMM expression tunes the response of
breast cancer cell lines to hyaluronan. Acta Biomater. 2022;146:187–96.
[28]. Jeffery J, Sinha D, Srihari S, et al. Beyond cytokinesis: the emerging roles of CEP55 in tumorigenesis. Oncogene. 2016;35:683–90.
[29]. Huang Y, Guo W, Kan H. TPX2 is a prognostic marker and contributes to growth and metastasis of human hepatocellular carcinoma. Int J Mol Sci. 2014;15:18148–61.
[30]. Yang Y, Li DP, Shen N, et al. TPX2 promotes migration and invasion of human
breast cancer cells. Asian Pac J Trop Med. 2015;8:1064–70.
[31]. Mateo F, He Z, Mei L, et al. Modification of BRCA1-associated
breast cancer risk by HMMR overexpression. Nat Commun. 2022;13:1895.
[32]. Giuliano CJ, Lin A, Smith JC, et al. MELK expression correlates with tumor mitotic activity but is not required for cancer growth. Elife. 2018;7:e32838.
[33]. Wang Z, Katsaros D, Shen Y, et al. Biological and clinical significance of MAD2L1 and BUB1, genes frequently appearing in expression signatures for
breast cancer prognosis. PLoS One. 2015;10:e0136246.
[34]. Zhu XF, Yi M, He J, et al. Pathological significance of MAD2L1 in
breast cancer: an immunohistochemical study and meta analysis. Int J Clin Exp Pathol. 2017;10:9190–201.
[35]. Kahl I, Mense J, Finke C, et al. The cell cycle-related genes RHAMM, AURKA, TPX2, PLK1, and PLK4 are associated with the poor prognosis of
breast cancer patients. J Cell Biochem. 2022;123:581–600.
[36]. Du R, Huang C, Liu K, et al. Targeting AURKA in cancer: molecular mechanisms and opportunities for cancer therapy. Mol Cancer. 2021;20:15.
[37]. Wang X, Xiao H, Wu D, et al. miR-335-5p regulates cell cycle and metastasis in lung adenocarcinoma by targeting CCNB2. Onco Targets Ther. 2020;13:6255–63.
[38]. Moradpoor R, Zali H, Gharebaghian A, et al. Identification of CCNB2 as a potential non-invasive
breast cancer biomarker in peripheral blood mononuclear cells using the systems biology approach. Cell J. 2021;23:406–13.
[39]. Xu L, Xia C, Sheng F, et al. CEP55 promotes the proliferation and invasion of tumour cells via the AKT signalling pathway in osteosarcoma. Carcinogenesis. 2018;39:623–31.
[40]. Wang Y, Lee YM, Baitsch L, et al. MELK is an oncogenic kinase essential for mitotic progression in basal-like
breast cancer cells. Elife. 2014;3:e01763.
[41]. Xiong M, Zhuang K, Luo Y, et al. KIF20A promotes cellular malignant behavior and enhances resistance to chemotherapy in colorectal cancer through regulation of the JAK/STAT3 signaling pathway. Aging (Milano). 2019;11:11905–21.
[42]. Steven A, Seliger B. The role of immune escape and immune cell infiltration in
breast cancer. Breast Care. 2018;13:16–21.
[43]. Burugu S, Asleh-Aburaya K, Nielsen TO. Immune infiltrates in the
breast cancer microenvironment: detection, characterization and clinical implication.
Breast Cancer. 2017;24:3–15.
[44]. Ostrand-Rosenberg S. Immune surveillance: a balance between protumor and antitumor immunity. Curr Opin Genet Dev. 2008;18:11–8.
[45]. Goff SL, Danforth DN. The role of immune cells in breast tissue and immunotherapy for the treatment of
breast cancer. Clin
Breast Cancer. 2021;21:e63–73.
[46]. Yang GJ, Zhu MH, Lu XJ, et al. The emerging role of KDM5A in human cancer. J Hematol Oncol. 2021;14:30.
[47]. Boysen G, Rodrigues DN, Rescigno P, et al. SPOP-mutated/CHD1-deleted lethal prostate cancer and abiraterone sensitivity. Clin Cancer Res. 2018;24:5585–93.
