Diagnostic Accuracy of DNA Methylation Markers for Detecting Cervical Precancer and Cervical Cancer: A Systematic Review and Meta-analysis.

BACKGROUND: DNA methylation markers have been proposed as molecular triage tools for detecting cervical precancer and cancer, particularly among HPV-positive women. However, diagnostic performance varies across markers, assays, specimens, and clinical settings. This study aimed to evaluate the marker-specific diagnostic accuracy of DNA methylation markers and methylation panels for detecting CIN2+ and CIN3+ in cervical cancer screening, triage, and related diagnostic contexts. METHODS: This systematic review and diagnostic test accuracy meta-analysis was reported according to PRISMA-DTA 10. Marker-specific analyses were performed separately by endpoint to avoid double counting. Pooled sensitivity, specificity, diagnostic odds ratio, clinical utility estimates, risk of bias, publication bias, and certainty of evidence were assessed using diagnostic test accuracy methods, QUADAS-2 11, Deeks' test 14, and GRADE-DTA 12,13. RESULTS: The broad main diagnostic meta-analysis included 74 eligible studies 15-88, contributing 260 diagnostic test accuracy comparisons across 130 marker-endpoint-sample analyses. A strict sensitivity dataset included 164 comparisons from 51 studies. Seven marker analyses were available for CIN2+ and seven for CIN3+. Diagnostic performance varied by marker and endpoint. For CIN2+, S5 EPB41L3 plus HPV16/18/31/33 methylation showed the highest sensitivity but lower specificity, whereas PAX1, JAM3, SOX1, PAX1/JAM3, and Six-gene/GynTect showed higher specificity. For CIN3+, S5 showed the highest sensitivity, while PAX1/JAM3, JAM3, SOX1, PAX1, and Six-gene/GynTect had more balanced performance. Deeks' test did not show strong evidence of publication bias for most analyses, although borderline findings were observed in selected marker-endpoint analyses. GRADE-DTA certainty was low or very low across marker-endpoint analyses. CONCLUSIONS: DNA methylation markers showed heterogeneous diagnostic accuracy for detecting CIN2+ and CIN3+. Some markers demonstrated potentially useful sensitivity- or specificity-oriented profiles, but certainty of evidence was low or very low. Further standardized prospective validation is required before routine clinical implementation.