Accuracy of genotyping for HPV16 and 18 to triage women with low-grade squamous intraepithelial lesions: a pooled analysis of VALGENT studies. (3rd June 2019)

• Record Type: Journal Article
• Authors: Xu, Lan; Benoy, Ina; Cuschieri, Kate; Poljak, Mario; Bonde, Jesper; Arbyn, Marc
• Is Part Of: Expert review of molecular diagnostics. Volume 19:Number 6(2019)
• Page Start: 543
• Page End: 551
• Publication Date: 2019-06-03
• Subjects: Cervical cancer screening -- HPV genotyping -- triage -- low-grade squamous intraepithelial lesions -- diagnostic test accuracy; Molecular diagnosis -- Periodicals; 616.0758205
• DOI: 10.1080/14737159.2019.1613890
• Languages: English
• ISSNs: 1473-7159

Abstract

ABSTRACT: Background : Genotyping for the most carcinogenic human papillomavirus (HPV) types (HPV16/HPV18) can identify high risk of underlying cervical precancer and guide further management. Research design and methods : A pooled analysis was performed of the clinical accuracy of high-risk HPV (hrHPV) testing and HPV16/18 genotyping in triage of women with low-grade squamous intraepithelial lesions (LSIL). Data regarding 24 assays evaluated in four VALGENT validation panels were used. Results : In women with LSIL, hrHPV had a pooled sensitivity for CIN2+ of 95.5% (95% CI: 91.0–97.8%) and a specificity of 25.3% (95% CI: 22.2–28.6%). HPV16/18 genotyping had a sensitivity and specificity for CIN2+ of 52.9% (95% CI: 48.4–57.4%) and 83.5% (95% CI: 79.9–86.5%), respectively. The average risk of CIN2+ was 46.1% when HPV16/18-positive, 15.5% in women who were HPV16/18-negative but positive for other hrHPV types and 4.3% for hrHPV-negative women. Conclusions : Triage of women with LSIL with HPV16/18 genotyping increases the positive predictive value compared to hrHPV testing but at the expense of lower sensitivity. Arguably, women testing positive for HPV16/18 need further clinical work-up. Whether colposcopy referral or further surveillance is recommended for women with other hrHPV types may depend on the post-test risk of precancer and the local risk-based decision thresholds.