Background
High-risk human papillomavirus (hrHPV) DNA testing has been the major primary method for cervical cancer screening. Despite the high sensitivity of hrHPV testing, its limited specificity may lead to unnecessary referrals for colposcopy. Emerging evidence suggests that host DNA methylation markers may help identify women at high risk for cervical cancer and its precursors; however, their performance for triaging hrHPV-positive women yet to be comprehensively evaluated.
 
Objective
This umbrella review aimed to systematically synthesise evidence from existing systematic reviews and meta-analyses to evaluate the accuracy of DNA methylation testing as a triage tool for hrHPV-positive women in cervical cancer screening.
 
Methods
We systematically searched five databases (PubMed, Web of Science, Cochrane Library, China National Knowledge Infrastructure, and Wanfang) from inception to September 1, 2025 for eligible systematic reviews and meta-analyses evaluating host DNA methylation markers for triage among hrHPV-positive women. Search terms included: (“Systematic Review” OR “Meta-Analysis”) AND (“Human Papillomavirus Viruses” OR “HPV” OR “Human Papillomavirus”) AND (“Uterine Cervical Neoplasms” OR “Cervical Cancer” OR “Cervix Neoplasm” OR “Cervix Cancer” OR “Uterine Cervical Dysplasia” OR “Cervical Dysplasia” OR “Squamous Intraepithelial Lesions”) AND (“DNA Methylation”). Triage performance, including pooled sensitivity, specificity, and area under the receiver operating characteristic curve (AUC), were systhesised using meta-analytic approaches in population-based studies. This study was registered with PROSPERO, CRD420251130315.
 
Results
A total of 25 systematic reviews and meta-analyses were included. Across reviews, the sensitivity of DNA methylation markers ranged from 0.55-0.86 for CIN2+, 0.70-0.86 for CIN3+, and 0.87-0.99 for cervical cancer, with corresponding specificity ranging from 0.69 to 0.95, from 0.69-0.87 and from 0.36-0.91. According to AMSTAR-2, the methodological quality of the included reviews was generally low (moderate: 10; low: 7; critically low: 8).
19 population-based cervical cancer screening studies cited in systematic reviews met the inclusion criteria. Overall, DNA methylation testing demonstrated moderate accuracy for detecting CIN2+ (pooled sensitivity 0.647, specificity 0.768, AUC 0.755) and CIN3+ (sensitivity 0.724, specificity 0.743, AUC 0.795), and high accuracy for cervical cancer (sensitivity 0.885, specificity 0.736, AUC 0.892) among hrHPV positive women. For biomarkers reported with multiple cut-offs, only the threshold with the highest Youden index was included in pooled analyses. Subgroup analyses indicated comparable performance between physician-collected and self-collected samples, although heterogeneity was higher in self-collected settings. Several methylation panels exhibited consistent triage performance across outcomes. Publication bias was detected for CIN2+ (P=0.0447) and cervical cancer (P<0.001), but not for CIN3+ (P=0.3551).
 
Conclusions
DNA methylation testing demonstrates promising potential as a triage strategy for hrHPV-positive women. Its application may help optimise cervical cancer screening pathways and reduce unnecessary referrals for colposcopy. However, the limited number of population-based screening studies, biomarker heterogeneity, and potential publication bias warrant cautious interpretation. Further large-scale prospective studies are required to support implementation in population-based screening programs.