Abstract Cervical, endometrial and vulvar cancer are three common types of gynecological tumor that threaten the health of females worldwide. Since their underlying mechanisms and associations remain unclear, a comprehensive and systematic bioinformatics analysis is required. The present study downloaded [33]GSE63678 from the GEO database and then performed functional enrichment analyses, including gene ontology and pathway analysis. To further investigate the molecular mechanisms underlying the three types of gynecological cancer, protein-protein interaction (PPI) analysis was performed. A biological network was generated with the guidance of the Kyoto Encyclopedia of Genes and Genomes database and was presented in Cytoscape. A total of 1,219 DEGs were identified for the three types of cancer, and 25 hub genes were revealed. Pathway analysis and the PPI network indicated that four main types of pathway participate in the mechanism of gynecological cancer, including viral infections and cancer formation, tumorigenesis and development, signal transduction, and endocrinology and metabolism. A preliminary gynecological cancer biological network was constructed. Notably, following all analysis, the phosphoinositide 3-kinase (PI3K)/Akt pathway was identified as a potential biomarker pathway. Seven pivotal hub genes (CCNA2, CDK1, CCND1, FGF2, IGF1, BCL2 and VEGFA) of the three gynecological cancer types were proposed. The seven hub genes may serve as targets in gynecological cancer for prevention and early intervention. The PI3K/Akt pathway was identified as a critical biomarker of the three types of gynecological cancer, which may serve a role in the pathogenesis. In summary, the present study provided evidence that could support the treatment of gynecologic tumors in the future. Keywords: cervical cancer, endometrial carcinoma, vulvar carcinoma, bioinformatics, phosphoinositide 3-kinase/Akt Introduction A gynecological tumor is a type of malignant tumor that occurs in the female reproductive system and seriously threatens the life of the patient. Among the types of gynecological tumor, cervical cancer (CC), endometrial carcinoma (EC) and vulvar carcinoma (VC) are the top three most common tumors of the female genital system, besides ovarian cancer ([34]1). Despite an overall decline in the incidence and mortality rates due to increased understanding of the disease, gynecological cancer remains a significant health care burden worldwide ([35]1). Early detection and treatment are essential for improving patient outcomes; however, these require improved understanding of the molecular pathology of the disease, in addition to identification of appropriate biomarkers and drug targets. Previous studies have demonstrated that the occurrence of CC is closely associated with human papillomavirus (HPV) infection ([36]2–[37]4). VC can be separated into two types, including one type that more frequently occurs in young females. This type involves the progression of a vulvar intraepithelial neoplasia caused by HPV infection, particularly HPV 16 and 18 ([38]5). Based on pathogenetic perspectives, EC is also classified into two groups according to estrogen dependence ([39]6). Although there have been a number of previous etiology studies, the exact pathogenesis of these three types of cancer remains unclear. There are certain pathological and etiological associations between CC and VC, as both are squamous cell cancers and both are associated with HPV infection ([40]2,[41]5). Unlike CC and VC, EC is associated with sex hormones, which is similar to common invasive tumors in females, including breast and ovarian cancer ([42]6). In addition, clinical diagnoses of these three cancer types rely predominantly on pathology ([43]7). Precise biomarkers in early stages of CC, EC and VC remain unknown. It is understood that cervical, endometrial and vulvar tissues all originate from the same embryological origin, the paramesonephric ducts, which give rise to the whole female reproductive tract and develop into different organs, following complex regulatory process ([44]8). For this reason, although there a number of differences between CC, EC and VC, it has been hypothesized that these three types of gynecological tumor share a similar mechanism and certain specific marker molecules may be common to their tumorigenesis and development. Therefore, a comprehensive analysis may improve understanding of these three types of tumor. Advancements in biotechnology have improved the availability of high-throughput data, including genomic, proteomic and metabolomics data, which supports in-depth scientific research. High-throughput data can assist with effective early diagnosis, prognosis prediction and investigations of molecular mechanisms for numerous types of disease. The present study used [45]GSE63678 microarray data downloaded from the Gene Expression Omnibus (GEO) to determine the differentially expressed genes (DEGs, which were identified between cancerous samples and non-cancerous samples) of CC, EC and VC ([46]9). Subsequently, functional enrichment analyses were performed, including gene ontology (GO) and pathway analysis, and a protein-protein interaction (PPI) network was generated to identify the significant biological terms associated with the DEGs. The genes that were screened out by the PPI network were considered as the hub genes, which may serve important roles in the mechanism of CC, EC and VC. In addition, a gene-pathway network was constructed and further analysis was performed. The complete flowchart of the present study is presented in [47]Fig. 1. In summary, the current study may provide a new perspective for elucidating the biological significance of three types of gynecological cancer, and assist with the identification of potential candidate biomarkers for diagnosis, prognosis and therapy. Figure 1. [48]Figure 1. [49]Open in a new tab Flowchart of the study. CC, cervical cancer; EC, endometrial cancer; VC, vulvar cancer, PPI, protein-protein interaction. Materials and methods Microarray data The gene expression profile [50]GSE63678 on the platform of the [51]GPL571 Affymetrix Human Genome U133A 2.0 Array was downloaded from the GEO database ([52]http://www.ncbi.nlm.nih.gov/geo/). [53]GSE63678 is a dataset submitted by Pappa et al ([54]9), containing 18 cancer samples, including five cervical, seven endometrial and six vulvar samples, and 17 normal samples, including, five cervical, five endometrial and seven vulvar samples. Identification of DEGs GEO2R ([55]http://www.ncbi.nlm.nih.gov/geo/geo2r) is an interactive network analysis tool of the GEO database based on R, in which two sets of samples can be compared under the same experimental conditions ([56]10). Associated gene data were divided into CC, EC and VC groups. Subsequently, the GEO2R ([57]10) tool and limma package ([58]11) available through Bioconductor (version 3.8) of R Studio (version 3.5) were used to compare the gene expression of the CC, EC and VC groups. P<0.05 and a fold-change >2 were considered to indicate a DEG. The distribution of the DEGs in the three tumor types was presented as a Venn diagram using FunRich software (version 3.0) ([59]12). Construction of the PPI network and identification of hub genes Search Tool for the Retrieval of Interacting Genes ([60]http://string-db.org/; version 10.5) is a software system that is commonly used to search for known proteins and predict interactions ([61]13). The experimentally validated interactions with a combined score >0.7 were selected as significant and DEGs with a connection number <2 were removed. The PPI network was visualized using Cytoscape ([62]https://cytoscape.org/; version 3.6.0). The nodes with degree, closeness and betweenness scores higher than the mean, as calculated by the Cytoscape plugin Centiscape, were considered hub nodes. GO analysis of the hub genes WEB-based Gene SeT AnaLysis Toolkit ([63]http://www.webgestalt.org/; revision 2017) is a popular software tool for functional enrichment analysis, which covers seven biological contexts, including GO ([64]14). Therefore, this software was used in the present study for GO enrichment analysis. The false discovery rate (FDR) was set at <0.05 to conduct the GO analysis of the DEGs. Pathway enrichment analysis of the hub genes The hub genes were uploaded to ToppGene ([65]https://toppgene.cchmc.org/,) for pathway enrichment analysis. The two frequently used databases, Kyoto Encyclopedia of Gene and Genomes (KEGG; [66]www.genome.jp/kegg) and Biocarta ([67]www.biocarta.com), were used to perform this analysis ([68]15). The FDR was set at <0.05. Pathway crosstalk analysis The enriched pathways were recruited for further crosstalk analysis to investigate the associations between them. As described previously ([69]15), to measure the association between two pathways, Jaccard coefficient (JC)=A∩B/A∪B and overlap coefficient (OC)=A∩B/min(|A|,|B|) were adopted, where A and B are the gene items contained in the two pathways, min is the minimum, ∩ is the intersection of A and B, and ∪ is the union of A and B. Since limited biological information was available, pathways containing <3 genes were excluded. Similarly, the pathway pairs with <2 overlapping genes were removed. Subsequently, the pathway network was presented with Cytoscape according to the JC and OC value of each selected pair ([70]16), and the MCODE plug-in ([71]17) (version 1.4.2; [72]apps.cytoscape.org/apps/MCODE) for Cytoscape was used to find clusters and highly interconnected regions in any network was used to analyze the clusters. Gene-pathway network analysis To further investigate the developmental mechanisms of CC, EC and VC, the hub genes were mapped into a crosstalk network. By analyzing the interactions between the genes and pathways with KEGG and Biocarta, the connected nodes were linked with arrows. The gene-pathway network was constructed and visualized in Cytoscape. The degree was calculated and nodes with a degree greater than the mean degree of all nodes were selected to constitute a sub network. Results Identification of DEGs Following screening with the criteria of P<0.05 and fold-change >2, a total of 1,219 DEGs were identified. In the CC group 138 DEGs were revealed, including 87 upregulated genes. In addition, 479 DEGs were identified in the EC group, including 272 upregulated genes. Finally, 734 DEGs, including 172 upregulated genes, were revealed in the VC group. As demonstrated in [73]Fig. 2A, 84, 378 and 630 DEGs were exclusively identified in CC, EC and VC groups, respectively. However, 23 DEGs were present in both the CC and EC group, 73 DEGs were identified in both the EC and VC groups, and 26 DEGs were revealed in both the CC and VC groups. Furthermore, five mutual genes, including signal sequence receptor subunit 1 (SSR1), flap structure-specific endonuclease 1 (FEN1), cyclin A2 (CCNA2), signal transducer and activator of transcription 1 (STAT1) and C-X-C motif chemokine ligand 12 (CXCL12), were identified in all three groups. Figure 2. [74]Figure 2. [75]Open in a new tab DEGs in CC, EC and VC. (A) Venn diagram of three gynecological tumor types. The yellow circles represent the DEGs in CC, the blue circles represent the DEGs in EC and the red circles represent the DEGs in VC. The overlapping areas indicate the shared genes of any two or three groups. (B) The protein-protein interaction network of the 25 hub genes. The circle nodes represent genes. Red indicates genes in CC, yellow represents genes in EC and green indicates genes in VC. The thickness of the line between any two nodes represents the strength of the connection. (C) Gene ontology analysis of 25 DEGs. The results are presented for the following three categories: Biological process, cellular component and molecular function DEG, differentially expressed gene; CC, cervical cancer; EC, endometrial cancer; VC, vulvar cancer. Hub genes and PPI network Following calculation by Centiscape, the mean values of degree, closeness and betweenness were 12.64080, 3.73×10-4 and 2081.81034, respectively. Additionally, 25 hub genes were identified, including six downregulated genes and 19 upregulated genes ([76]Table I). Three histone cluster family members were revealed as hub genes, including histone H2B type 1-H (HIST1H2BH), histone cluster 1 H2B family member D (HIST1H2BD) and histone cluster 1 H2B family member K (HIST1H2BK), and the five hub genes were cell cycle regulatory proteins, including CCNA2, cyclin B1 (CCNB1), cyclin D1 (CCND1), aurora kinase A (AURKA) and cell division cycle 20 (CDC20). Furthermore certain genes associated with tumor progression were identified, including vascular endothelial growth factor A (VEGFA), FYN proto-oncogene, Src family tyrosine kinase (FYN), baculoviral IAP repeat containing 5 (BIRC5) and the apoptosis regulator B-cell lymphoma 2 (BCL2). Table I. Topological parameters of the hub genes. Gene Degree Betweeness Closeness Group Regulation Mean 13 2081.81034 3.73×10^−4 – – CDK1 115 20448.4838 5.21×10^−4 CC/EC Up CCNB1 100 16335.9142 5.23×10^−4 CC/EC Up CDC20 92 8148.54809 4.76×10^−4 EC Up CCNA2 91 8126.90458 4.89×10^−4 CC/EC/VC Up AURKA 90 9181.94344 4.76×10^−4 EC/VC Up TOP2A 87 21409.6376 5.15×10^−4 EC/VC Up UBE2C 82 14916.6491 4.82×10^−4 EC/VC Up BIRC5 80 16279.0157 5.13×10^−4 EC/VC Up PCNA 62 26080.2036 5.25×10^−4 CC/EC Up VEGFA 56 32712.0335 5.43×10^−4 VC Up PIK3R1 46 24576.8505 4.91×10^−4 VC Down HIST1H2BK 43 8689.20717 4.70×10^−4 CC/VC Up HIST1H2BD 43 8689.20717 4.70×10^−4 VC Up HIST1H2BH 42 8291.98846 4.70×10^−4 VC Up ACACB 41 40853.547 4.98×10^−4 VC Down CXCL8 40 28927.396 5.16×10^−4 VC Up H2AFZ 39 8793.56814 4.67×10^−4 CC Up IGF1 38 16278.2011 5.07×10^−4 CC/VC Down ACLY 37 34157.4527 4.81×10^−4 EC Up CCND1 36 28548.907 5.39×10^−4 EC Up GAPDH 33 41598.1258 5.39×10^−4 EC Up PPP2R5C 32 10238.8057 4.75×10^−4 CC UP FYN 32 17254.3572 5.22×10^−4 CC/VC Down FGF2 32 19465.3169 5.20×10^−4 EC Down BCL2 32 17923.6004 4.86×10^−4 VC Down [77]Open in a new tab CC, cervical cancer; EC, endometrial cancer; VC, vulvar cancer. The PPI network of the 25 hub genes with 25 nodes and 114 edges is presented in [78]Fig. 2B. The top five genes with the highest degrees were CDK1, CCNB1, CDC20, CCNA2 and AURKA. All five of these genes are associated with cell cycle regulation, which indicates that cell cycle dysfunction serves an important role in the development of gynecological tumors. GO enrichment analysis A total of 25 DEGs were used to perform GO enrichment analysis ([79]Fig. 2C). For cellular component terms, 22 out of the 25 genes were revealed to be located in the ‘nucleus’ and approximately 80% were identified to participate in the ‘macromolecular complex’ (19 genes) and ‘membrane-enclosed lumen’ (18 genes). In the biological process category, the DEGs were associated with ‘biological regulation’ (22 genes), ‘metabolic process’ (22 genes), ‘cellular component organization’ (21 genes) and ‘response to stimulus’ (21 genes). In the molecular function category all 25 DEGs were associated with ‘protein binding’ (25 genes). Pathway enrichment analysis of the hub genes By uploading the 25 genes into ToppGene, 86 significant pathways were identified. The biological processes involved in these pathways can be divided into the following five main categories: i) viral infections and cancer formation, including ‘viral carcinogenesis’ and ‘hepatitis B’, ii) tumorigenesis and development, including ‘colorectal cancer’ and ‘proteoglycans in cancer’, iii) signal transduction, including ‘PI3K-Akt signaling pathway’ and ‘AMPK signaling pathway’, iv) endocrinology and metabolism, including ‘AGE-RAGE signaling pathway in diabetic complications’ and ‘endocrine resistance’, and v) others, including ‘genes encoding secreted soluble factors’ and ‘NFAT and hypertrophy of the heart (transcription in the broken heart)’. In addition, 19 pathways were identified to be downregulated and 17 pathways were revealed to be upregulated ([80]Table II). Table II. Pathways enriched in three types of gynecological cancer. Pathway Regulation P-value Genes in the pathway Viral carcinogenesis – 3.43×10^−9 HIST1H2BD, CCND1, CDK1, HIST1H2BH, CDC20, PIK3R1, HIST1H2BK, CCNA2 Hepatitis B – 9.66×10^−9 BIRC5, CCND1, BCL2, PIK3R1, PCNA, CXCL8, CCNA2 AMPK signaling pathway – 1.13×10^−7 CCND1, PPP2R5C, IGF1, ACACB, PIK3R1, CCNA2 Oocyte meiosis – 1.31×10^−7 AURKA, PPP2R5C, IGF1, CDK1, CDC20, CCNB1 Cell cycle Up 1.31×10^−7 CCND1, CDK1, CDC20, PCNA, CCNA2, CCNB1 EGFR tyrosine kinase inhibitor resistance – 4.35×10^−7 FGF2, BCL2, IGF1, PIK3R1, VEGFA Pathways in cancer – 6.44×10^−7 FGF2, BIRC5, CCND1, BCL2, IGF1, PIK3R1, CXCL8, VEGFA Progesterone-mediated oocyte maturation – 1.15×10^−6 IGF1, CDK1, PIK3R1, CCNA2, CCNB1 AGE-RAGE signaling pathway in diabetic complications – 1.35×10^−6 CCND1, BCL2, PIK3R1, CXCL8, VEGFA HIF-1 signaling pathway – 1.49×10^−6 BCL2, IGF1, GAPDH, PIK3R1, VEGFA Focal adhesion – 2.13×10^−6 CCND1, BCL2, IGF1, FYN, PIK3R1, VEGFA PI3K-Akt signaling pathway – 3.48×10^−6 FGF2, CCND1, BCL2, PPP2R5C, IGF1, PIK3R1, VEGFA p53 Signaling Pathway – 3.94×10^−6 CCND1, BCL2, PCNA IL-7 Signal Transduction Down 4.77×10^−6 BCL2, FYN, PIK3R1 Colorectal cancer – 5.72×10^−6 BIRC5, CCND1, BCL2, PIK3R1 p53 signaling pathway – 1.00×10^−5 CCND1, IGF1, CDK1, CCNB1 Melanoma – 1.00×10^−5 FGF2, CCND1, IGF1, PIK3R1 Cyclins and Cell Cycle Regulation Up 1.23×10^−5 CCND1, CDK1, CCNB1 Platinum drug resistance – 1.25×10^−5 BIRC5, BCL2, PIK3R1, TOP2A Regulation of BAD phosphorylation Down 1.80×10^−5 BCL2, IGF1, PIK3R1 Prostate cancer – 2.52×10^−5 CCND1, BCL2, IGF1, PIK3R1 Endocrine resistance – 3.71×10^−5 CCND1, BCL2, IGF1, PIK3R1 Proteoglycans in cancer – 4.47×10^−5 FGF2, CCND1, IGF1, PIK3R1, VEGFA Bladder cancer Up 7.25×10^−5 CCND1, CXCL8, VEGFA Sphingolipid signaling pathway – 8.32×10^−5 BCL2, PPP2R5C, FYN, PIK3R1 FoxO signaling pathway – 1.29×10^−4 CCND1, IGF1, PIK3R1, CCNB1 Systemic lupus erythematosus Up 1.32×10^−4 H2AFZ, HIST1H2BD, HIST1H2BH, HIST1H2BK NFAT and Hypertrophy of the heart (Transcription in the broken heart) Down 1.66×10^−4 FGF2, IGF1, PIK3R1 Breast cancer – 1.80×10^−4 FGF2, CCND1, IGF1, PIK3R1 Activation of Src by Protein-tyrosine phosphatase alpha Up 2.11×10^−4 CDK1, CCNB1 Sonic Hedgehog (SHH) Receptor Ptc1 Regulates cell cycle Up 2.11×10^−4 CDK1, CCNB1 AKAP95 role in mitosis and chromosome dynamics Up 2.52×10^−4 CDK1, CCNB1 Glioma – 2.75×10^−4 CCND1, IGF1, PIK3R1 Pancreatic cancer – 2.75×10^−4 CCND1, PIK3R1, VEGFA Expression of cyclins regulates progression through the cell cycle by activating cyclin-dependent kinases. Up 2.98×10^−4 CCND1, CCNA2 The IGF-1 Receptor and Longevity Down 4.00×10^−4 IGF1, PIK3R1 Alcoholism Up 4.22×10^−4 H2AFZ, HIST1H2BD, HIST1H2BH, HIST1H2BK B Cell Survival Pathway – 4.57×10^−4 BIRC5, PIK3R1 Small cell lung cancer – 6.12×10^−4 CCND1, BCL2, PIK3R1 Stathmin and breast cancer resistance to antimicrotubule agents Up 6.48×10^−4 CDK1, CCNB1 Epstein-Barr virus infection – 6.64×10^−4 BCL2, CDK1, PIK3R1, CCNA2 Skeletal muscle hypertrophy is regulated via AKT/mTOR pathway Down 7.19×10^−4 IGF1, PIK3R1 Rap1 signaling pathway – 7.54×10^−4 FGF2, IGF1, PIK3R1, VEGFA IGF-1 Signaling Pathway Down 7.94×10^−4 IGF1, PIK3R1 Ras signaling pathway – 1.01×10^−3 FGF2, IGF1, PIK3R1, VEGFA Erk and PI-3 Kinase Are Necessary for Collagen Binding in Corneal Epithelia Down 1.04×10^−3 FYN, PIK3R1 Cell Cycle: G2/M Checkpoint Up 1.04×10^−3 CDK1, CCNB1 Influence of Ras and Rho proteins on G1 to S Transition – 1.22×10^−3 CCND1, PIK3R1 Genes related to IL4 rceptor signaling in B lymphocytes Down 1.32×10^−3 BCL2, PIK3R1 Inactivation of Gsk3 by AKT causes accumulation of b-catenin in Alveolar Macrophages Up 1.32×10^−3 CCND1, PIK3R1 Cholinergic synapse Down 1.41×10^−3 BCL2, FYN, PIK3R1 Cell Cycle: G1/S Check Point Up 1.42×10^−3 CCND1, CDK1 VEGF, Hypoxia, and Angiogenesis – 1.52×10^−3 PIK3R1, VEGFA HTLV–I infection – 1.57×10^−3 CCND1, CDC20, PIK3R1, PCNA Control of skeletal myogenesis by HDAC and calcium/calmodulin-dependent kinase (CaMK) Down 1.63×10^−3 IGF1, PIK3R1 Apoptosis-multiple species – 1.97×10^−3 BIRC5, BCL2 How Progesterone Initiates Oocyte Membrane Up 2.09×10^−3 CDK1, CCNB1 Measles – 2.36×10^−3 CCND1, FYN, PIK3R1 Aldosterone-regulated sodium reabsorption Down 2.47×10^−3 IGF1, PIK3R1 Apoptosis – 2.56×10^−3 BIRC5, BCL2, PIK3R1 IL-2 Receptor Beta Chain in T cell Activation Down 2.60×10^−3 BCL2, PIK3R1 Signaling pathways regulating pluripotency of stem cells Down 2.62×10^−3 FGF2, IGF1, PIK3R1 Fluid shear stress and atherosclerosis – 2.78×10^−3 BCL2, PIK3R1, VEGFA Phospholipase D signaling pathway – 3.01×10^−3 FYN, PIK3R1, CXCL8 Jak-STAT signaling pathway – 3.63×10^−3 CCND1, BCL2, PIK3R1 Members of the BCR signaling pathway Down 3.80×10^−3 BCL2, PIK3R1 Hedgehog signaling pathway – 3.96×10^−3 CCND1, BCL2 T Cell Receptor Signaling Pathway Down 3.96×10^−3 FYN, PIK3R1 Endometrial cancer – 4.47×10^−3 CCND1, PIK3R1 Genes encoding secreted soluble factors – 4.58×10^−3 FGF2, IGF1, CXCL8, VEGFA Acute myeloid leukemia – 5.39×10^−3 CCND1, PIK3R1 Non-small cell lung cancer – 5.97×10^−3 CCND1, PIK3R1 VEGF signaling pathway – 6.18×10^−3 PIK3R1, VEGFA Viral myocarditis – 6.18×10^−3 CCND1, FYN Longevity regulating pathway-multiple species Down 6.80×10^−3 IGF1, PIK3R1 Renal cell carcinoma – 7.45×10^−3 PIK3R1, VEGFA Fc epsilon RI signaling pathway – 8.13×10^−3 FYN, PIK3R1 Prolactin signaling pathway – 8.60×10^−3 CCND1, PIK3R1 Chronic myeloid leukemia – 8.84×10^−3 CCND1, PIK3R1 Longevity regulating pathway – 1.36×10^−2 IGF1, PIK3R1 Genes related to Wnt-mediated signal transduction Up 1.36×10^−2 CCND1, GAPDH Rheumatoid arthritis Up 1.39×10^−2 CXCL8, VEGFA NF-kappa B signaling pathway – 1.54×10^−2 BCL2, CXCL8 Amoebiasis – 1.57×10^−2 PIK3R1, CXCL8 Cdc25 activates the cdc2/cyclin B complex to induce the G2/M transition. Up 1.60×10^−2 CDK1 Inflammatory mediator regulation of TRP channels Down 1.60×10^−2 IGF1, PIK3R1 [81]Open in a new tab Pathway crosstalk analysis To further investigate how the identified pathways interact with each other, a pathway crosstalk analysis was conducted among the pathways that met the criteria. The approach was based on the assumption that two pathways can be considered to be associated if they share a proportion of genes ([82]18). A total of 45 pathways contained more than two hub genes, of which 41 pathways met the criterion for crosstalk analysis. The network of crosstalk, which includes these 41 pathways, is presented in [83]Fig. 3A. The thickness of edge connecting two nodes represents the strength of the association between them, which was measured by the mean value of OC and JC. Using MCODE, two major clusters were identified from the whole network. The simple cluster involves three pathways associated with cell cycle, including ‘Cell cycle’, ‘Oocyte meiosis’ and ‘Cyclins and Cell Cycle Regulation’. The complicated cluster containing a total of 32 nodes and 376 edges is presented in [84]Fig. 3B. The five aforementioned types of pathways were interconnected to form the complex network, which indicates the complexity of the pathogenesis of CC, EC and VC. Figure 3. [85]Figure 3. [86]Open in a new tab Pathway networks. (A) Model of the pathway network. The red circle nodes indicate a pathway and the lines represent the interactions between two pathways. The strength of the interaction is represented by the thickness of the line, according to the mean value of JC and OC. (B) Model of the core pathway network. Red circle nodes indicate a pathway and the lines represent the interactions between two pathways The strength of the interaction is represented by the thickness of the line according to the mean value of JC and OC. JC, Jaccard coefficient; OC, overlap coefficient. Gene-pathway network construction of DEGs By mapping the hub genes into the complicated sub-network according to the KEGG and Biocarta databases, a potential gene-pathway network was constructed to verify the associations between the candidate pathways and genes ([87]Fig. 4). This network included 37 important pathways and 18 hub genes, including CCND1 presented in the middle with direct or indirect associations with all other genes. As the only overlapping gene of all three groups, CCNA2 possessed complicated connections with ‘viral carcinogenesis’, ‘hepatitis B’, ‘cell cycle’ and six other pathways. In addition, insulin-like growth factor-1 (IGF1), fibroblast growth factor 2 (FGF2) and CCND1 were located close to the middle of the gene-pathway network. Figure 4. [88]Figure 4. [89]Open in a new tab Gene-Pathway network. Orange diamonds indicate a pathway and circles represent a gene. Red indicates genes in cervical cancer, yellow represents genes in endometrial cancer and green indicates genes in vulvar cancer. Sub gene-pathway network of DEGs To screen the key factors, including genes and pathways, in the gene-pathway network, the degrees of all of nodes were calculated and nodes with a degree greater than the mean degree of all nodes were selected ([90]Fig. 5). Seven genes (CCNA2, CDK1, CCND1, BCL2, IGF1, FGF2 and VEGFA) and six pathways (‘Viral carcinogenesis’, ‘Hepatitis B’, ‘Focal adhesion’, ‘Pathways in cancer’, ‘PI3K-Akt signaling pathway’ and ‘Proteoglycans in cancer’) were selected. Since these gene and pathways had more connections with other nodes, they were considered to more likely serve a role in CC, EC and VS. Figure 5. [91]Figure 5. [92]Open in a new tab Sub-molecular network of CC, EC and VC. Orange diamonds indicate a pathway and circles represent a gene. Red indicates genes in CC, yellow represent genes in EC and green indicates genes in VC. CC, cervical cancer; EC, endometrial cancer; VC, vulvar cancer. Discussion In the past few decades, gynecological cancer research has developed rapidly, particularly regarding the recognition of etiological factors. Although a number of studies have investigated CC, EC and VC separately, few studies have focused on these three types of cancer in combination. The aim of the present study was to perform a systematic and comprehensive analysis to investigate the pathogenesis of CC, EC and VC and make a preliminary assessment of the associations between these three cancer types. By performing an analysis of microarray data, 1,219 DEGs were identified, including 138 in CC, 479 in EC and 734 in VC. Five DEGs were revealed in all three cancer types, which suggests these genes may participate in a number of important biological processes in gynecological cancer and may serve as crucial biomarkers following further research. Together with the 25 hub genes identified, these data may provide a direction for future research on gynecological cancer and assist with the identification of clinical biological targets. Pathway enrichment analysis indicated that 86 pathways are closely associated with the 25 hub genes. In particular, it was identified that viral infection and carcinogenesis pathways were significantly enriched, including ‘viral carcinogenesis’, ‘HTLV-1 infection’ and ‘hepatitis B’, which supports the association of virus with the three gynecological cancer types, particularly CC. Furthermore, cancer-association pathways, including ‘pathways in cancer’ and ‘proteoglycans in cancer’ were revealed to be associated with the biological process of the three malignant tumor types. Notably, multiple different types of human cancer, including melanoma, prostate cancer, bladder cancer, breast cancer and glioma, were also identified to be associated with the 25 hub genes. This indicates that gynecological cancer types may exhibit homologous mechanisms with tumor types of other systems. With the gene-pathway sub-network model, seven critical hub genes and six important pathways of the three gynecological cancer types were identified. The hub genes with the highest degrees included CDK1, which was enriched in CC and VC. As reported, CDK1 is a member of the Ser/Thr protein kinase family and is encoded by cell division cycle gene 2 (cdc2) ([93]19). In addition, CDK1 has been revealed to serve a role in numerous types of cancer, including EC ([94]20), breast cancer ([95]21) and ovarian cancer ([96]22). Consistent with the present bioinformatics results, CDK1 has been demonstrated to serve a comprehensive role in mediating genetic networks involved in the progression of CC; therefore, it may be an important therapeutic target for improving prognosis ([97]23). A study regarding ovarian cancer identified that CDK1 is associated with proliferation and can serve as a prognostic factor in epithelial ovarian cancer ([98]22). In EC, the overexpression of CDK1 in endometrial carcinoma cells is closely associated with the occurrence of tumors, indicating a role in tumor prognosis ([99]24). The CDK1 gene can contribute to the carcinogenesis of HPV ([100]25), and CC and VC are associated with HPV infection; therefore, CDK1 may be an important molecule in the pathogenesis of gynecological tumors. Another cell cycle regulatory gene, CCNA2, was revealed as a shared DEG of CC, EC and VC, and complicated connections were identified between it and other nodes. According to recent studies, CCNA2 belongs to the highly conversed cyclin family and is expressed in multiple tissues in the human body, including numerous types of cancer, which indicates it may serve a role in cancer transformation and progression ([101]26,[102]27). Gao et al ([103]28) revealed that CCNA2 is a prognostic biomarker for estrogen receptor-positive breast cancer and is associated with Tamoxifen resistance. Combined with another biological analysis of EC that demonstrated CCNA2 is one of the top two upregulated nodes ([104]29), the present study hypothesizes that CCNA2 serves a role as a biomarker in gynecological tumors ([105]29). In addition, a study associated with ovarian cancer revealed a similar result, in which CCNA2 was upregulated in the chemo-resistant epithelial ovarian cancer ([106]30). Therefore, it can be suggested that CCNA2 is a potential biomarker in gynecological cancer; however, this requires in vivo or in vitro experimental verification. CCND1 is an important positive regulator of the G1/S phase of the cell cycle and has been identified as a co-factor of HPV in the initiation of cervical carcinogenesis ([107]31). Similar studies regarding EC and VC have also widely been reported ([108]32–[109]34). BCL2 and IGF1 were revealed as the only two downregulated genes in the sub-network. BCL2 is an intracellular membrane protein that prevents apoptotic cell death and overexpression of BCL2 can block p53-mediated G1 arrest ([110]35). Kamaraddi et al ([111]36) demonstrated that BCL2 expression is higher in malignant lesions compared with premalignant lesions, which differs from the current findings. It has been suggested that alterations of BCL2 expression are associated with early events in cervical tumorigenesis and a lower BCL2 expression level has also been demonstrated to be associated with an improved 5-year survival rate and prognosis ([112]37). The significance of BCL2 in gynecological tumors requires further investigation. Furthermore, IGF1 is closely associated with the occurrence of numerous tumor types; however, its exact mechanism remains unclear. Iyer et al ([113]38) identified that IGF-1 expression levels in advanced CC increase with chemo-radiotherapy and decline during follow-up ([114]38). With a limited specificity in gynecologic tumors, IGF1 is of limited value in the early prediction of gynecological tumors; however it may serve a role in targeted treatment strategies, and the assessment and improvement of prognosis ([115]39,[116]40). Angiogenesis serves an important role in tumor growth, development, progression and metastasis ([117]41). As a pro-angiogenesis factor, VEGFA is involved in the proliferation, differentiation and migration of endothelial cells, and participates in the invasion and metastasis of numerous types of cancer ([118]42). Chen et al ([119]43) demonstrated that VEGFA may be a target for inhibiting angiogenesis in EC 42). Similarly, Hua and Tian ([120]44) revealed that CCL4 can promote cell proliferation, invasion and migration of EC by targeting the VEGFA signal pathway ([121]44). Combined with the present results, this indicates that VEGFA serves an important role in gynecological tumor invasion and metastasis. FGF2 is a typical fibroblast growth factor that stimulates the growth of various cell types, from fibroblasts to tumor cells ([122]45). In addition, FGF2 is a fundamental signaling molecule in tumor-induced angiogenesis ([123]46). It has been demonstrated that FGF2 is mitogenic in various cell types and is associated with the regulation of tumor angiogenesis and metastasis ([124]47). Certain studies regarding the receptor family of FGF2 have revealed that it is associated with the occurrence and development of CC, in addition to HPV16 infection ([125]48,[126]49). Aberrant FGF/FGF receptor signaling has been demonstrated in multiple types of tumor ([127]50,[128]51). The expression level of FGF2 has been revealed to be higher in EC compared with normal tissues, and the highest expression level was observed in tumors with dedifferentiation, myometrial invasion and advanced staging ([129]52). Therefore, angiogenesis has an important impact in the pathogenesis of gynecological cancers. ‘PI3K-Akt signaling pathway’, ‘hepatitis B’, ‘pathways in cancer’, ‘focal adhesion’, ‘viral carcinogenesis’ and ‘proteoglycans in cancer’ were located in the sub-network, which indicates that these processes serve an important role in the pathogenesis of CC, EC and VC. It is understood that the PI3K/Akt signaling pathway serves a central role in cell growth and proliferation, and it has also been suggested that its deregulation is associated with cancer ([130]53). Yung et al ([131]54) demonstrated that the activation of AMPK could significantly inhibit CC cell growth. Similar studies regarding the PI3K/Akt pathway in EC have also been reported ([132]55,[133]56), and it has been considered as a therapeutic target ([134]57). According to previous studies, the PI3K/Akt signaling pathway can serve as a therapeutic target in EC ([135]57) and ovarian cancer ([136]58,[137]59), and can be mediated by molecules, including VEGFA ([138]40). FGF2 has also been reported to serve an angiogenic role by the PI3K/Akt pathway ([139]60). Furthermore, BCL2 is a major downstream mediator of the PI3K/Akt pathway and serves a pivotal role in tumor response ([140]61,[141]62). It has been reported that CCNA2 expression promotes the migration, invasion and metastasis of hepatocellular carcinoma and ovarian cancer cells via the PI3K/Akt signaling pathway ([142]63). Therefore, this crucial pathway in cancer cells may be involved in the early developmental stages of formation and invasion. As indicated by the present results, the molecular mechanisms underlying CC, EC and VC are complicated, and further studies are required to fully understand their pathological mechanisms. Similarly, Suman and Mishra ([143]64) identified that the aurora kinase pathway has a crucial function in the pathogenesis of five gynecological cancer types, including breast cancer, EC, CC, ovarian cancer and VC, by analyzing the common core genes from the [144]GSE63678, [145]GSE57297 and [146]GSE26712 datasets. Furthermore, the present study identified seven genes (CCNA2, CDK1, CCND1, BCL2, IGF1, FGF2 and VEGFA) and six pathways (‘viral carcinogenesis’, ‘hepatitis B’, ‘focal adhesion’, ‘pathways in cancer’, ‘PI3K-Akt signaling pathway’ and ‘proteoglycans in cancer’) that may serve an important role in CC, EC and VC. A number of factors are involved in the progression of cancer; the present study focused on the factors associated with the female reproductive system. The results may provide comprehensive evidence that promotes the understanding of cancers of the female genital tract. While previous studies have focused on co-expressed DEGs ([147]23,[148]64), the present study aimed to establish a gene-pathway network based on the analysis of DEGs. Additionally, previous studies investigated genes co-expressed by the five cancer types (breast cancer, EC, CC, ovarian cancer and VC). However, the current study not only investigated the co-expressed DEGs of CC, EC and VC, but also examined the DEGs co-expressed by any combination of two of the cancer types. In summary, the current study focused of the associations between three types of tumor, which may make it more comprehensive compared with previous studies. Notably, there are certain limitations to the present study. The sample number was relatively small, which to a certain extent reduces the credibility of gene enrichment. Subject to conditions, long-term assessments of the patients' clinical conditions were not available. In addition, the literature regarding the pathways associated with CC, EC and EC, except for the PI3K/Akt pathway is limited; therefore, the present study lacked a solid foundation to adequately discuss the current results. Finally, certain genes that are associated with the pathogenesis of gynecological types of cancer may not have been statistically analyzed, possibly due to the exclusion criteria that was applied. In conclusion, the pathogenesis of CC, EC and VC is complicated. By performing a comprehensive analysis, the present study revealed a library of DEGs in CC, EC and VC, and identified 25 hub genes. Subsequently, viral infection, tumorigenesis, inflammation and the endocrine system were revealed to be involved in the development of these three types of cancer. Finally, a molecular network of CC, EC and VC was constructed. Most notably, it was identified that the PI3K/Akt pathway serves an important role in the three types of gynecological cancer and seven hub genes (CCNA2, CDK1, CCND1, FGF2, IGF1, BCL2 and VEGFA) present in the sub-network may act as therapeutic targets, and assist with early diagnosis and prevention. The present study may support the elucidation of the underlying mechanisms in CC, EC and VC, which would promote early detection and the development of targeted therapy. Further investigations that aim to improve understanding of the mechanisms of these three cancer types will be vital for developing highly sensitive and multifactorial strategies for the prevention, diagnosis and treatment of CC, VC and EC. Acknowledgements