Eur J Cancer · 2026 · Systematic Review & Meta-Analysis · IF 7.9

DNA Methylation Tests for High-Grade Cervical Lesions & Cancer: Systematic Review & Meta-Analysis

DNA 甲基化检测对高级别宫颈病变与宫颈癌的诊断效能:系统综述与 Meta 分析
125 studies · 49,242 women · JAM3, C13ORF18, PAX1 most accurate for CIN3+ in hrHPV+ women · EPB41L3 lowest NLR → 1.35% post-test probability
Ellis LB et al. · PROSPERO CRD42022299760
Open access · doi: 10.1016/j.ejca.2026.116823
2nd · DOR
C13ORF18
DOR 19.50 · Sp 93%
Se 61% (49–72)
1st · DOR
JAM3
DOR 20.82 · Se 74% · Sp 88%
CIN3+ · hrHPV+ women
3rd · DOR
PAX1
DOR 17.13 · Sp 87%
Se 71% (64–77)

1Background & Objective

  • Problem: cytology triage has modest sensitivity (53–75%) and inter-observer variability; many countries now use hrHPV primary screening, needing reliable triage.
  • Methylation: measurable epigenetic change correlated with malignant transformation; many markers proposed but rarely head-to-head compared.
  • Objective: collate all published human/viral DNA methylation tests for CIN3+/ICC and estimate pooled diagnostic accuracy in hrHPV+ women.

2Methods & Evidence Base

2,097
records screened
125
studies included
49,242
women
2,097 screened 784 full-text 125 included
13
markers · primary
32
markers · all women
GRADE
certainty
Pre-test riskPost-test (−) EPB41L310/13 markers
Low · 5%1.35%<2%
Moderate · 9%2.51%<2%
High · 17%≈5%<3%
EPB41L3
1.35%
JAM3
1.8%
PAX1
2.0%
FAM19A4/miR124-2
2.66%
  • Search: MEDLINE, Embase, CENTRAL to 31 Mar 2025; PROSPERO CRD42022299760; PRISMA-DTA.
  • Analysis: bivariate random-effects meta-analysis per marker; HSROC; QUADAS-2 risk of bias; GRADE.
  • Primary endpoint: CIN3+ in hrHPV+ women; secondary CIN2+/ICC; each marker analysed separately.
  • Pre/post-test: low (5%) / moderate (9%) / high (17%) risk settings; referral threshold ≥10% post-test probability.

3Top Markers — CIN3+ in hrHPV+ Women

MarkerDORSe %Sp %NLR
JAM320.8274880.30
C13ORF1819.5061930.42
PAX117.1371870.33
S5 classifier4.4684470.34
Highest sensitivity: S5 (84%); highest specificity: C13ORF18 (93%); lowest NLR: EPB41L3 0.26.

4Results — Post-Test Probability & Forest

Post-test probability plot by pre-test risk setting
Fig. Negative post-test probability — EPB41L3 1.35% in low-risk setting; 10/13 markers <2%.
Forest plot of marker accuracy
Fig. Marker-level forest plots of pooled accuracy (per marker meta-analysis).
  • Rule-out power: all 13/13 markers reach ≤3% negative post-test probability; none reach <1%.
  • Certainty: JAM3/C13ORF18/PAX1 best by DOR — consistent with CISCER® dual-gene test.
Clinical Significance