Performance and diagnostic accuracy of human papillomavirus testing on urine and self-collected vaginal samples in a referral population

초록

Background Human papillomavirus (HPV) is well established as the main cause of cervical cancer. Non-invasive self-collected urine and vaginal sampling have the potential advantage of increasing patient compliance with cervical cancer screening. Methods Women referred for colposcopy at Korea University Guro Hospital, following abnormal cytology, were included this study. A total of 314 paired urine, vaginal and cervical samples were collected. Primary endpoints were sensitivity for CIN2+/CIN3+ and specificity for <CIN2; secondary endpoints were the relative accuracy of hrHPV test results in vaginal and urine samples versus cervical samples. Results For clinician-collected cervical samples, Sejong Realtime HR-S HPV test sensitivity for detecting and specificity from were similar to well-established test (Anyplex™ II HPV 28) [sensitivity for CIN3+ (n=109) 93.27% (95% confidence interval (CI), 86.62–97.25); CIN2+ (n=130) 92.74% (95% CI, 86.67–96.63); specificity for<CIN2 31.82% (95% CI, 25.01–39.25)]. All the paired differences (cervical versus urine sampling, cervix versus vaginal sampling) in sensitivity were statistically significant. However, among women with ASCUS/LSIL cytology, hrHPV sensitivity on vaginal samples was comparable to that of cervical samples for detection of CIN2+ and CIN3+ lesions. In addition, hrHPV sensitivity of Anyplex II HPV 28 assay on urine was comparable to that of cervical samples for detection of CIN3+ lesions (p=0.07) in women with ASCUS/LSIL cytology. Conclusions HPV tests using urine and vaginal samples were still inferior to clinician-collected cervical samples in terms of detecting CIN2/3. However, these results indicate that combination of cytology with reflex hrHPV test using vaginal and urine samples may offer a reliable strategy for discriminating women at greater risk of precancerous lesion, increasing compliance of patients.