Diagnostic accuracy of urine-based HPV testing for cervical neoplasia and cancer: A landmark systematic review and meta-analysis.

Abstract

e17506 Background: Urine-based human papillomavirus (HPV) testing offers a non-invasive screening alternative that could improve accessibility. We quantified diagnostic accuracy of urine HPV tests for detecting cervical intraepithelial neoplasia grade 2 or worse (CIN2+), CIN3+, and cervical cancer. Methods: We searched PubMed, Embase, Scopus, and Web of Science through January 2026. Studies evaluating urine HPV testing against histologically confirmed disease were included. Bivariate random-effects meta-analyses using the Reitsma model pooled sensitivity and specificity. We analyzed different biomarker types (high-risk HPV DNA, mRNA, methylation), population and test characteristics, and examined disease threshold effects through meta-regression. Results: Fifty-five studies (72 arms, n=38,009) were included. For CIN2+, pooled sensitivity was 81.4% (95% CI: 77.5-84.7%) and specificity 59.3% (95% CI: 50.2-67.8%). hrHPV DNA tests (k=45) showed sensitivity 82.7% and specificity 58.2% (AUC 0.812). hrHPV mRNA (k=2) had lower sensitivity (45.1%, p=0.001) but higher specificity (68.4%). Methylation markers (k=2) achieved balanced performance: sensitivity 78.0%, specificity 75.2% (AUC 0.832). Performance improved with disease severity. For cancer (k=6), sensitivity was 91.8% (95% CI: 85.4-95.5%) and specificity 84.1% (95% CI: 74.2-90.6%; AUC 0.943). Meta-regression confirmed significant disease threshold differences (p<0.022). Conclusions: Urine-based HPV testing, particularly hrHPV DNA assays, demonstrates promising accuracy for CIN2+ and excellent performance for cancer detection, supporting expanded screening access. Standardization and larger studies of methylation markers are needed. Diagnostic performance of urine HPV testing by disease threshold and biomarker type. Analysis Subgroup Studies Participants Sensitivity % (95% CI) Specificity % (95% CI) Accuracy % DOR AUC Disease Threshold Cancer 6 894 91.8 (85.4-95.5) 84.1 (74.2-90.6) 86.8 59.05 0.943 CIN3+ 17 11,141 83.7 (78.0-88.1) 53.9 (36.3-70.6) 64.8 5.98 0.814 CIN2+ 49 25,974 81.4 (77.5-84.7) 59.3 (50.2-67.8) 69 6.36 0.800 Biomarker (CIN2+) hrHPV DNA 45 25,159 82.7 (79.0-85.8) 58.2 (48.3-67.4) 68.9 6.63 0.812 hrHPV mRNA 2 277 45.1 (35.6-54.9) 68.4 (54.3-79.8) 59.6 1.78 0.455 Methylation 2 538 78.0 (67.4-86.0) 75.2 (62.3-84.8) 77.1 10.79 0.832 Overall All studies 55 (72 arms) 38,009 82.8 (79.8-85.4) 60.7 (52.9-68.0) 68.2 7.42 0.819 CI, confidence interval; DOR, diagnostic odds ratio; AUC, area under curve; hrHPV, high-risk HPV; CIN, cervical intraepithelial neoplasia. All estimates from bivariate random-effects meta-analysis (Reitsma model) against histology reference standard. Meta-regression p-values: disease threshold p=0.019 (sensitivity), p=0.022 (specificity); biomarker p=0.001 (sensitivity). Heterogeneity I²: 34.7-50.8%.