Accuracy of human papillomavirus testing using self-collected urine samples for detecting high-grade squamous intra-epithelial lesion or worse: a diagnostic meta-analysis.

Abstract

Objective: Self-collection methods are regarded as a strategy to promote human papillomavirus (HPV) testing for cervical cancer screening. HPV testing using urine specimens is non-invasive and could eliminate barriers such as embarrassment and discomfort. However, concerns regarding the accuracy of this method persist. Hence, the current meta-analysis aims to verify the accuracy of urine HPV nucleic acid amplification testing.

Methods: A comprehensive search was conducted in the PubMed, Embase, and Cochrane Library databases to identify studies that evaluated the performance of urine HPV nucleic acid amplification testing. The inclusion criteria were as follows: studies that evaluated the diagnostic accuracy of HPV testing for high-grade squamous intra-epithelial lesion or worse (HSIL+) with self-collected urine specimens. In addition, studies that provided sufficient data for conducting a meta-analysis were assessed. To ensure the rigor of the study, those using histopathology or colposcopy as the reference standard were deemed sufficiently rigorous for inclusion. The meta-analysis was conducted using the bi-variate random-effects model.

Results: A total of 21 studies with 6603 samples were identified. The meta-analysis yielded a pooled sensitivity of 83.0% (95% CI 77.5% to 87.3%) and a pooled specificity of 51.3% (95% CI 39.2% to 63.3%) of HPV nucleic acid amplification testing with self-collected urine specimens for HSIL+. In terms of detection technology, a pooled sensitivity of 85.2% and a pooled specificity of 49.4% were obtained from the sub-group analysis of manuscripts that used DNA-based testing.

Conclusions: Our meta-analysis indicates that urine HPV nucleic acid amplification testing achieves high sensitivity for detecting HSIL+. A notable benefit of DNA-based urine HPV nucleic acid amplification testing is its heightened sensitivity compared with RNA-based methods.