HPV-related methylation-based reclassification and risk stratification of cervical cancer

Human papillomavirus (HPV) is a clear etiology of cervical cancer (CC). However, the associations between HPV infection and DNA methylation have not been thoroughly investigated. Additionally, it remains unknown whether HPV-related methylation signatures can identify subtypes of CC and stratify the prognosis of CC patients. DNA methylation profiles were obtained from The Cancer Genome Atlas to identify HPV-related methylation sites. Unsupervised clustering analysis of HPV-related methylation sites was performed to determine the different CC subtypes. CC patients were categorized into cluster 1 (Methylation-H), cluster 2 (Methylation-M), and cluster 3 (Methylation-L). Compared to Methylation-M and Methylation-L, Methylation-H exhibited a significantly improved overall survival (OS). Gene set enrichment analysis (GSEA) was conducted to investigate the functions that correlated with different CC subtypes. GSEA indicated that the hallmarks of tumors, including KRAS signaling, TNF alpha signaling via NF-kappa B, inflammatory response, epithelial-mesenchymal transition, and interferon-gamma response, were enriched in Methylation-M and Methylation-L. Based on mutation and copy number variation analyses, we found that aberrant mutations, amplifications, and deletions among the MYC, Notch, PI3K-AKT, and RTK-RAS pathways were most frequently detected in Methylation-H. Additionally, mutations, amplifications, and deletions within the Hippo, PI3K-AKT, and TGF-beta pathways were presented in Methylation-M. Genes within the cell cycle, Notch, and Hippo pathways possessed aberrant mutations, amplifications, and deletions in Methylation-L. Moreover, the analysis of tumor microenvironments revealed that Methylation-H was characterized by a relatively low degree of immune cell infiltration. Finally, a prognostic signature based on six HPV-related methylation sites was developed and validated. Our study revealed that CC patients could be classified into three heterogeneous clusters based on HPV-related methylation signatures. Additionally, we derived a prognostic signature using six HPV-related methylation sites that stratified the OS of patients with CC into high- and low-risk groups.