An investigation of methylation biomarkers for abnormal cervical cancer screening tests in the United States

Early detection and treatment of precancerous cervical lesions through population-based screening has significantly reduced cervical cancer morbidity and mortality. In the United States (US), the addition of high-risk human papillomavirus co-testing to traditional Papanicolaou testing has improved the identification of cervical abnormalities and refined management guidelines. However, millions of abnormalities still necessitate follow-up to confirm the presence of high-grade cervical intraepithelial neoplasia (CIN2+) versus low-grade (≤CIN1) histology, which determines subsequent management. DNA methylation patterns at cytosine-phosphate-guanine (CpG) sites in cervical samples are promising biomarkers to improve detection and risk stratification of screening-detected abnormalities. Epigenome-wide methylation studies are needed in the US—particularly prospective studies of low-grade lesions in diverse populations—to optimally inform national guidelines. We utilized data from the Cervical Intraepithelial Neoplasia Cohort Study (CINCS), a prospective clinical cohort of patients presenting to colposcopy following an abnormal screening test in North Carolina, to examine 1) CpG differential methylation (DM) and differential variability (DV) in baseline CIN2+ vs. ≤CIN1 (cross-sectional), and 2) associations between methylation and five-year progression to incident CIN2+ among those with ≤CIN1 (prospective). We conducted targeted analyses of pre-selected genes with known roles in cervical cancer and exploratory epigenome-wide association studies (EWAS) to identify novel methylation biomarkers. We confirmed previously observed cross-sectional patterns of DM at CCNA1, CDH1, and RARB CpGs in CIN2+, and we detected DV among CDH1 and FHIT CpGs. The EWAS detected 3,534 CpGs exhibiting DM in CIN2+ and 270 CpGs with DV; 202 CpGs exhibited both DM and DV. We identified 44 gene ontology terms enriched among DM CpGs and 5 among DV CpGs, including terms related to DNA transcription, neuronal structures, cell structure, and developmental processes. Prospectively, we confirmed associations between CADM1 and RARB CpGs and time-to-progression to CIN2+, with trends supporting associations among DAPK1 and PAX1 CpGs. We detected 336 novel CpGs in the time-to-progression EWAS. Methylation biomarkers may help distinguish lesions with differential risks of CIN2+ and improve risk-based stratification of screening-detected cervical abnormalities. This study contributes to the growing literature assessing the utility of methylation biomarkers to inform management of millions of cervical precancerous lesions detected annually in the US.