Cancer metabolism is essential for tumour growth, and altered metabolic activities are a hallmark of cancer (Hoang et al., 2019), leading to the modulation of tumour progression, metastasis, and prognosis (Danhier et al., 2017; DeBerardinis et al., 2016). In the current study, we focused on analysing the characteristics of metabolism-related genes in cervical cancer. We also assessed the roles of these genes in carcinogenesis and further explored their genetic and epigenetic alterations. Our results provide potential directions for studying the overall metabolism of cervical cancer.
Gene function is affected by genomic variation. Therefore, we analysed the changes in metabolism-related genes in the genome. Genomic analysis identified 2493 genes with mutations that might be involved in modulating metabolic activities. Among these genes, TTN, PIK3CA, and KMT2C showed the highest mutation frequency, and C > T was the most common nucleotide mutation in CESC. The PIK3CA gene has been reported to be frequently mutated in patients with cervical cancer in the United States of America (Femi et al., 2018). Mutations in PIK3CA affect glucose metabolism and proliferation in cervical cancer by activating the AKT/glycogen synthase kinase 3β/β-catenin pathway signalling pathway (Jiang et al., 2018), thereby reversing radio-resistance (Jiang et al., 2020). Notably, Vasilios et al. showed that patients with a mutated PIK3CA genotype tend to have poor survival outcomes compared with patients harbouring wild-type PIK3CA, particularly in patients with CESC (Pergialiotis et al., 2020). Similarly, KMT2C mutations have been shown to be dominant suppressor gene alterations in cervical cancer (Scholl et al., 2020). In the current study, we identified mutations in TTN in CESC for the first time and showed that these mutations were relevant for metabolism in this disease.
To determine whether LOF mutations affect prognosis in patients with CESC, we analysed 616 genes mutated in more than five samples and identified 15 significant genes. In the subsequent gene mutual exclusion/co-occurrence analysis, we found that CUBN-KALRN-HERC1 has a co-occurrence relationship and affects the prognosis of CESC. Although there is no research on these three genes and CESC, in Gronhoj’s research, it was found that HERC1 can be mutated with human papilloma virus infection (Gronhoj et al., 2018). Therefore, it is possible that HPV infection related to cervical cancer might affect the prognosis of CESC by causing mutations in these three genes.
In this study, we further evaluated the effects of metabolism-related gene mutations on clinical parameters using all available gene mutation information to calculate the TMB for each sample. TMB is the total number of mutations present in a tumour specimen and is an emerging biomarker for the response to immunotherapy (Chalmers et al., 2017). We found no significant correlations between TMB and clinical parameters (including overall survival), which is consistent with Huang’s research (Huang et al., 2021). However, we found that clinical staging could be associated with metabolism-related TMB in CESC. Except in stage IV disease, metabolism-related TMB was increased with the progression of endocervical adenocarcinoma. Cervical squamous cell carcinoma and endocervical adenocarcinoma belong to two different histological subtypes in cervical cancer. However, the pathogenesis of the two is not the same. Due to different tissue types, TMB is different in different clinical stages. Therefore, for the treatment of two different tissue types, personalised therapeutics can be provided based on different TMB and clinical types.
We also observed changes in copy numbers (Cejas et al., 2020). Importantly, an evaluation of changes in copy numbers of metabolism-related genes showed that FGA, FGG, and FGL were related to recurrence and overall survival in patients with CESC. Similarly, Peng et al. found that EBP50 plays important roles in cell proliferation and that copy number variations in EBP50 could affect the prognosis of patients with cervical cancer (Peng et al., 2016). In addition to genomic changes in metabolism-related genes, we also assessed the expression of the aforementioned genes in CESC samples based on TCGA database. In total, 494 metabolism-related genes were aberrantly expressed in CESC compared with levels in normal tissues, 374 of which were related to prognosis.
WGCNA corresponds to an unsupervised classification method, establishing connections among co-expressed genes and grouping them into modules (Langfelder et al., 2008; Niemira et al., 2019). For example, Tian et al. identified two co-expressed functional modules involved in the process of metastasis of lung cancer through WGCNA (Tian et al., 2017). Using WGCNA, eight candidate biomarkers were proposed for the progression and metastasis of breast cancer (Wang et al., 2019). In this study, we identified four metabolism-related modules for CESC according to WGCNA and pathway enrichment analysis, including arachidonic acid metabolism, carbon metabolism, glycolysis/gluconeogenesis, and purine metabolism modules, and each of these modules was verified to be significantly related to prognosis. The arachidonic acid pathway is a metabolic process involved in cervical carcinogenesis that functions by targeting multiple pathways (Yarla et al., 2016), including the cyclooxygenase (COX)-2 pathway; frequent COX-2 expression is closely related to poor prognosis in patients with cervical cancer (Mandic et al., 2014). Moreover, the metabolic process of glycolysis was reported to be correlated with worse prognosis in patients with cervical cancer (Cai et al., 2020; Li et al., 2019).WGCNA is also known to be particularly suitable for identifying co-expressed modules significant to clinical variables and biological tumour behaviour (Choi et al., 2018; Ohue et al., 2016). Hence, we evaluated the relationship between immune cells and metabolism-related modules. Previous studies have shown that fibre cells are related to glycolysis/gluconeogenesis (Funes et al., 2007). In our research, we also found a positive correlation between the two (R2 = 0.4). In addition, as a metabolic pathway closely related to the occurrence of tumours, arachidonic acid metabolism is positively related to most tumour immune cells. Among them, it has the strongest correlation with T cells. This article by Kato also confirmed the relationship between arachidonic acid and T cells (Kato et al., 1983). Moreover, after further exploring the correlations between these metabolism-related modules and clinical parameters, we speculated that the modules of carbon metabolism, purine metabolism, and arachidonic acid metabolism might be related to different histological classifications and could affect the overall survival patients with CESC. Glycolysis/gluconeogenesis can influence tumour recurrence and has been proven in previous studies to be an independent prognostic factor for tumour recurrence in advanced CESC patients treated with definitive chemoradiotherapy (Hong et al., 2016). Furthermore, we also discovered that the arachidonic acid metabolism module was related to cell proliferation using the GSVA algorithm and that this might attenuate cell proliferation via a mechanism independent of calcium entry, thereby contributing to poor prognosis (Cantonero et al., 2020).
Genetic and epigenetic anomalies, responsible for genetic dysregulation, are the most common aberrations that determine the underlying heterogeneity of tumour cells. Therefore, we also explored potential epigenetic regulatory factors affecting the identified differentially expressed genes. We found that 11 genes were negatively regulated by miRNAs and that the expression levels of six genes were affected by their mutations. Additionally, the expression of 46 genes might be affected by promoter methylation. Interestingly, we identified 58 genes that were regulated by the aforementioned three mechanisms. NME5 and SLC40A1 were affected by methylation and miRNA regulation, whereas HK3 and JAK3 were regulated by both mutation and methylation. Promoter methylation in the SLC40A1 gene has been reported to affect differential expression (Ju et al., 2020), and the expression of HK3 has been shown to be upregulated in CpG island methylator phenotype-high tumours (Fedorova et al., 2019), supporting our findings. Our findings thus provide evidence to further study the association between genetic variation and the expression of differentially expressed genes in CESC.