Test Accuracy of Human Papillomavirus in Urine for Detection of Cervical Intraepithelial Neoplasia

Study population.Between November 2016 and January 2019, we recruited a convenience sample of women, 25 to 65 years of age, who were attending colposcopy clinics at the University of North Carolina at Chapel Hill (UNC) Women’s Hospital or Duke University Hospital for one of the following indications: abnormal cytology results, infection with HPV-16/18, persistent infection with other hrHPV subtypes, or treatment for CIN2+. We also aimed to include a sample of HPV-positive women at lower risk for CIN2+, who would be more similar to a primary screening population than the patients undergoing clinically indicated colposcopy. Therefore, we invited women with normal cytology findings and positive hrHPV results for HPV types other than HPV-16/18 at the time of their routine screening. This group was referred to as “research only,” because current U.S. guidelines do not recommend immediate referral for colposcopy for such women (4).

Potentially eligible women were identified through review of electronic medical records and were contacted via telephone or during their clinic visits. Women were excluded if they were pregnant or had had their cervix removed; additionally, women in the research only group were excluded if they were taking anticoagulants or if the enrollment date was not within 3 months after their original hrHPV diagnosis. Written informed consent was obtained from all participating women. The study was approved by the institutional review boards at UNC and Duke University.

Sample collection.During the clinic visit, participating women received detailed verbal and written instructions, in English or Spanish, concerning the study procedures. The women then provided two urine samples, i.e., an initial-stream sample of ∼20 ml for hrHPV testing and a midstream sample of up to 100 ml for pregnancy testing, if clinically indicated. Next, the women self-collected a cervicovaginal sample by inserting a Viba brush (Rovers Medical Devices BV, Oss, The Netherlands) to the top of the vaginal canal, rotating it five times, and releasing the brush head into a vial prefilled with 6 ml of preservative liquid-based cytology medium (ThinPrep; Hologic Inc., Bedford, MA).

Next, a pelvic examination was performed by a clinical provider, during which a cervical scraping was collected with two 360° turns, in a clockwise fashion, of a brush-like cervical cell collector (Wallach Papette; Wallach Surgical Devices, Trumbull, CT). The provider-collected cervical sample was preserved in a 20-ml vial of ThinPrep medium for subsequent hrHPV testing. Colposcopy was performed for all women, following cervical treatment with 3 to 5% acetic acid (followed by Lugol’s iodine at the Duke University Hospital site), according to standard clinical procedures. Directed biopsy samples were taken from visible cervical lesions, and endocervical curettage (ECC) was performed if the transformation zone or the limits of a lesion near the cervical os could not be fully visualized. If no cervical lesions were observable, then one random biopsy sample was taken at the 12 o’clock position of the cervix and ECC was performed. At least two cervical biopsy samples were obtained during each procedure. The loop electrosurgical excision procedure was performed when clinically indicated. At the end of the visit, the women received a gift card ($40 for women who had undergone clinically indicated colposcopy and $120 for research only participants).

Sample processing and laboratory analyses.From the initial-stream specimen, 2 ml of urine was transferred (within 5 min after collection) into a Becton Dickinson (BD) molecular tube containing 0.2 ml of a proprietary preservative to ensure sample stability. The tubes are fitted with a pierceable cap to facilitate automated sample processing on the BD Viper LT System. All samples and the BD molecular tube were placed in a cooler with frozen gel packs within 10 min after sample collection and were kept cool until they could be further aliquoted the same day. The self- and provider-collected samples were vortex-mixed for 10 to 30 s, and 0.5 ml of each sample was transferred to separate BD molecular tubes containing 1.7 ml of an HPV diluent buffer. All three BD molecular tubes were stored at –20°C until they were shipped to BD for hrHPV testing using the Onclarity assay (BD, Sparks, MD). The staff at BD did not have access to any clinical information concerning the participants. The Onclarity assay uses PCR and nucleic acid hybridization to detect DNA of 14 hrHPV subtypes. Six hrHPV subtypes (subtypes 16, 18, 31, 45, 51, and 52) are individually genotyped, and the other 8 subtypes are identified in three groups (subtypes 33/58, 56/59/66, and 35/39/68). Provider- and self-collected Hologic ThinPrep specimens were processed using the standard liquid-based cytology workflow on the BD Viper LT System. The Onclarity assay has been approved by the FDA for use with provider-collected samples in BD SurePath preservative fluid, but it shows similar performance when used with ThinPrep PreservCyt transport medium (23). Urine samples were processed using a co-collection device workflow, without a prewarming step. The remaining provider-collected cervical sample was sent to the UNC cytopathology laboratory for cytological analysis, if clinically indicated, or storage. The Onclarity hrHPV test results were obtained for research purposes only and were not shared with participants.

Histological diagnoses (<CIN2 versus CIN2+) served as the reference standard for this test accuracy study. Cervical biopsy samples from women who underwent colposcopy as part of their scheduled clinical appointments were sent to the UNC or Duke University Hospital surgical pathology laboratory for histological evaluation, according to standard procedures. Pathologists had access to clinical information captured in the electronic medical records but were unaware of the study hrHPV results. The colposcopy patients were informed of the histological results by the clinical team responsible for their care. Biopsy samples taken from women in the research only group were analyzed by a gynecological pathologist at the UNC translational pathology laboratory, who did not have access to clinical information. Women in the research only group were contacted with histological results by the study team after review by a practicing gynecologist. Women whose samples were inadequate for pathology review were invited to return for additional sampling. Women with CIN2+ were referred for further treatment, according to the standard of care.

Sample size considerations, definitions, and statistical analyses.Study recruitment was guided by an expected CIN2+ prevalence of 31% among the colposcopy clinic patients (13) and 6% among the research only participants (24). Our sample size calculations were focused on estimating the sensitivity of urine-based hrHPV testing with adequate precision. We expected 90% sensitivity and aimed to achieve sampling error margins of 10 percentage points. Thus, we aimed to recruit 160 colposcopy patients (49 CIN2+ cases expected) and 250 research only participants (15 CIN2+ cases expected). However, recruitment of research only participants turned out to be logistically challenging, with 79% of eligible women declining participation, cancelling clinic visits, or not being reachable. Therefore, we recruited more colposcopy patients and fewer research only participants than planned.

We included women with valid triple-matched results (urine, self-collected cervicovaginal, and provider-collected cervical samples available) in the test accuracy analysis; participants with missing samples or invalid test results were excluded. Participants without a valid cervical biopsy sample result (e.g., due to insufficient tissue) were also excluded. There is no established PCR cycle threshold (C_T) for the Onclarity assay in urine; therefore, we used the same cutoff value as for provider-collected cervical samples (≤34.2 for all hrHPV subtypes except HPV-16, with a C_T of ≤38.3). In an exploratory analysis, we evaluated the effects of different C_T cutoff values (≤36.0, ≤38.3, and ≤40.0) on the accuracy of urine-based hrHPV testing for CIN2+. Samples that were reactive for any hrHPV subtypes detected by the Onclarity assay were regarded as positive.

We used descriptive statistics to assess the sociodemographic characteristics of included women. We calculated the sensitivity and specificity of hrHPV testing for CIN2+ detection for different sample types. McNemar’s test was used to assess whether the sensitivity of hrHPV testing with urine samples was different from that with self-collected cervicovaginal samples and provider-collected cervical samples. The level of statistical significance was set at <0.05. We computed the overall agreement and the specific positive agreement and negative agreement between hrHPV test results for urine samples and those for self-collected cervicovaginal samples or provider-collected cervical samples (25). We calculated the unweighted Cohen’s kappa and its 95% confidence interval (CI) to assess the degree of agreement beyond chance between the different sample types. The following interpretation was used: ≤0, no agreement; 0.01 to 0.20, slight agreement; 0.21 to 0.40, fair agreement; 0.41 to 0.60, moderate agreement; 0.61 to 0.80, substantial agreement; 0.81 to 1.00, almost perfect agreement (26).

In sensitivity analyses, we examined whether the test accuracy and agreement results changed if we restricted the study population to patients with clinically indicated colposcopy or lower-risk research only participants. We also computed test accuracy with all valid results for a given sample type (403 results for urine samples, 380 results for cervicovaginal samples, and 343 results for provider-collected cervical samples), and we estimated hrHPV test accuracy for CIN3+ detection. Analyses were performed using SAS/STAT software (SAS Institute Inc., Cary, NC).

Data availability.Study data will be shared upon request, in accordance with the National Institutes of Health (NIH) guidelines on data sharing, as proposed in our original grant submission.