Abstract
Cervical cancer (CC) is a growing health concern, emphasizing the need for reliable biomarkers in treatment selection and prognosis assessment. We analyzed gene expression profiles and clinicopathological data from The Cancer Genome Atlas (TCGA) for CC. Using Consensus Cluster Plus, we applied machine learning to cluster the CC cohort. Differential analysis was performed using the edge R package, while weighted correlation network analysis (WGCNA) was conducted using the WGCNA package. Single-sample gene set enrichment analysis (ssGSEA) evaluated immune cell abundance and computed the m6Ascore. Western blot and Q-PCR validated the m6A score in CC. Common copy number variation alterations were observed in the 23 m6A-related genes in CC, and their mutation frequency was summarized in a waterfall chart. Patients were grouped into two clusters, m6AclusterA and m6AclusterB. Improved clinical outcomes were observed in m6AclusterA, while m6AclusterB exhibited higher infiltration of 14 immune cell types. WGCNA analysis generated seven integrated modules, enriched in several biological processes. Prognostic differential genes were used to generate two gene clusters (gene Cluster I and gene Cluster II). Using ssGSEA, the m6Ascore was calculated for each patient. Lower m6Ascore correlated with better clinical outcomes, lower gene mutation frequency, and wild-type status. We investigated the sensitivity of high and low m6Ascore to immunotherapy, visualized through violin and UMAP diagrams showcasing crosstalk among single-cell clusters. The key gene PFKFB4 showed higher expression in CC cell lines and tumor tissues compared to normal cells and tissue. Our study elucidates the role of m6A molecules in predicting prognosis, biological features, and appropriate treatment for CC patients.