Performance and Diagnostic Accuracy of a Urine-Based Human Papillomavirus Assay in a Referral Population.

Background: Human papillomavirus (HPV) testing from clinician-collected cervical and self-collected cervico-vaginal samples is more sensitive for detecting CIN2+/CIN3+ than cytology-based screening, stimulating interest in HPV testing from urine. The objective was to determine the performance of the Trovagene HPV test for the detection of CIN2+ from urine and PreservCyt cervical samples.Methods: Women referred for colposcopy at St Mary's Hospital (London, United Kingdom), following abnormal cytology, were recruited to this diagnostic accuracy study by convenience sampling (September 2011 to April 2013). A total of 501 paired urine and cervical samples were collected. Primary outcomes were sensitivity for CIN2+/CIN3+ and specificity for <CIN2; secondary outcomes were comparisons with other HPV tests and agreement/kappa values between urine and cervical samples.Results: Trovagene HPV test sensitivity and specificity from PreservCyt were similar to well-established tests [sensitivity for CIN3+ (n = 145) 96.3% (95% confidence interval (CI), 89.6-99.2); CIN2+ (n = 81) 94.5% (95% CI, 89.4-97.6); specificity for <CIN2 25.3% (95% CI, 20.8-30.1)]. Sensitivity from urine was slightly, but not significantly, lower [CIN3+ 91.4% (95% CI, 83.0-96.5), P = 0.3; CIN2+ 88.3% (95% CI, 81.9-93.0), P = 0.06]. Specificity for <CIN2 was similar: 24.7% (95% CI, 20.3-29.5), P = 0.9. A total of 403 Trovagene cervical and 396 urine HPV tests were positive. Overall agreement between paired samples was 82.6% (95% CI, 79.3-86.0).Conclusions: Trovagene HPV test's performance on PreservCyt cervical samples was comparable with established HPV tests. Sensitivity in urine, although slightly lower, may nevertheless be adequate for self-sampling. This referral population's higher HPV positivity rate affects specificity, warranting further studies in a screening population.Impact: This may prove useful for women not attending for cervical screening. Cancer Epidemiol Biomarkers Prev; 26(7); 1053-9. ©2017 AACR.