Comparison of urine specimen collection times and testing fractions for the detection of high-risk human papillomavirus and high-grade cervical precancer.

Background: Urine testing for high-risk human papillomavirus (HR-HPV) detection could provide a non-invasive, simple method for cervical cancer screening.

Objectives: We examined whether HR-HPV detection is affected by urine collection time, portion of urine stream, or urine fraction tested, and assessed the performance of HR-HPV testing in urine for detection of cervical intraepithelial neoplasia grade II or worse (CIN2+).

Study Design: A total of 37 female colposcopy clinic attendees, ≥ 30 years, provided three urine samples: "first void" urine collected at home, and "initial stream" and "mid-stream" urine samples collected at the clinic later in the day. Self- and physician-collected brush specimens were obtained at the same clinic visit. Colposcopy was performed and directed biopsies obtained if clinically indicated. For each urine sample, HR-HPV DNA testing was conducted for unfractionated, pellet, and supernatant fractions using the Trovagene test. HR-HPV mRNA testing was performed on brush specimens using the Aptima HPV assay.

Results: HR-HPV prevalence was similar in unfractionated and pellet fractions of all urine samples. For supernatant urine fractions, HR-HPV prevalence appeared lower in mid-stream urine (56.8%[40.8-72.7%]) than in initial stream urine (75.7%[61.9-89.5%]). Sensitivity of CIN2+ detection was identical for initial stream urine and physician-collected cervical specimen (89.9%[95%CI=62.7-99.6%]), and similar to self-collected vaginal specimen (79.1%[48.1-96.6%]).

Conclusion: This is among the first studies to compare methodologies for collection and processing of urine for HR-HPV detection. HR-HPV prevalence was similar in first void and initial stream urine, and was highly sensitive for CIN2+ detection. Additional research in a larger and general screening population is needed.