- Design
- systematic review and diagnostic meta-analysis across five databases, QUADAS-2 quality assessment
- Population
- eight studies, 7,494 participants undergoing cervical cancer screening
- Primary outcome
- pooled sensitivity and specificity of PAX1/JAM3 dual-gene methylation for high-grade cervical intraepithelial neoplasia
- Effect
- CIN2+ sensitivity 0.81 (95% CI 0.73–0.88), specificity 0.95 (0.94–0.96), AUC 0.96; CIN3+ sensitivity 0.85 (0.75–0.92), specificity 0.88 (0.86–0.89)
HPV-based cervical screening finds far more positives than there are high-grade lesions, and the triage step — deciding which HPV-positive women need colposcopy — is where the programme either works or drowns. Cytology triage is the current answer and is limited by needing a cytologist. This meta-analysis pooled eight studies in 7,494 participants evaluating PAX1/JAM3 dual-gene methylation testing.
For detecting CIN2 or worse, pooled sensitivity was 0.81 (95% CI 0.73–0.88) with specificity 0.95 (0.94–0.96) and an area under the curve of 0.96. For CIN3 or worse, sensitivity was 0.85 (0.75–0.92) and specificity 0.88 (0.86–0.89), area under the curve 0.90.
The specificity at the CIN2+ threshold is the important figure. Sending 5% of HPV-positive women without high-grade disease to colposcopy, rather than the far larger proportion that cytology triage or HPV genotyping alone produce, is what makes a screening programme affordable in a setting with limited colposcopy capacity. The test is objective, does not require a cytologist, and can run on the same self-collected or clinician-collected sample. A sensitivity of 0.81 means about one high-grade lesion in five is missed at that round, which is the cost and has to be set against screening interval.
- This is triage of HPV-positive women, not primary screening — it does not replace the HPV test.
- The specificity advantage matters most where colposcopy capacity is the bottleneck, which describes most of India.
- A negative methylation result does not clear the patient permanently; screening interval and follow-up still apply.
- Most of these studies come from Chinese cohorts; performance in populations with different HPV genotype distributions needs confirmation.
- Note the specificity falls from 0.95 at CIN2+ to 0.88 at CIN3+, which is the opposite of the usual pattern and reflects different study thresholds rather than better detection of milder disease.
Why it matters
The bottleneck in HPV-based screening is triage, and a test that needs no cytologist changes what a programme can do.
Don't overread it
Diagnostic accuracy pooled from eight studies — no screening programme has yet shown that using this triage improves outcomes.
The statistics, in plain English
Eight studies and 7,494 participants is a reasonable base for a diagnostic meta-analysis, but the confidence interval on sensitivity (0.73 to 0.88) is wide enough that the true miss rate could be one lesion in four. Area under the curve values of 0.96 and 0.90 describe performance across all possible thresholds; the sensitivity and specificity pairs quoted are at whatever cut-offs the individual studies chose, which is a common source of optimism in pooled diagnostic estimates. QUADAS-2 assessment was performed but the abstract does not report how many studies had high risk of bias.
Read the rest in the app
You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for oncology, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free