Clinical Performance of the RealTi m e High Risk HPV Assay on Self-Collected Vaginal Samples within the VALHUDES Framework

Abstract

The VALHUDES framework (NCT03064087) was established to evaluate the clinical accuracy of HPV testing on self-samples compared with HPV testing on matched clinician-taken cervical samples. Women referred to colposcopy due to previous cervical abnormalities were recruited at five Belgian colposcopy centers. A total of 486 pairs of matched cervical samples and vaginal self-samples were included in the analysis (228 collected with Evalyn Brush and 258 with Qvintip). The dry vaginal brushes were transferred into 20 mL ThinPrep PreservCyt solution. All specimens were tested with the Abbott RealTime High Risk HPV assay (Abbott RT). Testing on vaginal and cervical specimens was considered the index and comparator tests, respectively, and colposcopy and histology as the reference standard. The clinical sensitivity for CIN2+ of Abbott RT (cutoff ≤32 cycle number [CN]) on vaginal self-samples (Evalyn Brush and Qvintip combined) was 8% lower than on the cervical clinician-collected samples (ratio = 0.92 [95% CI, 0.87 to 0.98]), while the specificity was similar (ratio = 1.04 [95% CI, 0.97 to 1.12]). Sensitivity (ratio = 0.95 [95% CI, 0.89 to 1.02]) and specificity (ratio = 1.11 [95% CI, 0.995 to 1.23]) on Evalyn Brush samples was similar to cervical, while on Qvintip samples, the sensitivity was 12% lower than cervical samples (ratio = 0.88 [95% CI, 0.78 to 0.998]) with similar specificity (0.99 [95% CI, 0.90 to 1.10]). Exploratory cutoff optimization (cutoff ≤35 CN) resulted in an improvement of the relative sensitivity (self-sampling versus clinician sampling: ratio = 0.96 [95% CI, 0.91 to 1.02]) but yielded a loss in relative specificity (ratio = 0.92 [0.85 to 1.00]). The clinical accuracy of Abbott RT differed from the self-sampling device. However, after cutoff optimization, the sensitivity on self-samples taken with either of two vaginal brushes became similar to clinician-collected samples. IMPORTANCE Self-samples are becoming a crucial part of HPV-based cervical cancer screening programs to reach nonattendee women and increase screening coverage. Therefore, the VALHUDES framework was established to validate and evaluate HPV tests and devices on self-samples. Here, in the present manuscript, we evaluated the accuracy of the RealTime High Risk HPV assay (Abbott RT) on two different vaginal devices to detect cervical intraepithelial neoplasia grade two or higher (CIN2+). The study results demonstrated that the Abbott RT assay is similarly accurate on vaginal self-samples as on matched clinician-taken cervical samples after adjusting cutoff values. Moreover, we observed that some vaginal devices perform better than others in CIN2+ detection. We also underline the necessity of standardization and validation of general workflow and sample handling procedures for vaginal self-samples.

Keywords: Abbott RealTime High Risk HPV assay; HPV; VALHUDES; cervical cancer screening; diagnostic test accuracy; self-sampling.

FIG 1

Flow chart of samples included in the VALHUDES trial tested with the Abbott RealTime High Risk HPV (Abbott RT) assay. Gray boxes represent excluded samples. Detailed exclusions are reported in the Supplemental Material.

FIG 2

RealTime High Risk HPV cycle number (CN) values between cervical and vaginal samples were collected with Evalyn Brush or Qvintip. Boxplots indicate median CN values, interquartile ranges, and extreme values (whiskers) for HPV16, 18, other hrHPV, and β-globin. Median values are presented in Table S3 in Supplemental File 1.

FIG 3

Boxplot of the RealTime β-globin cycle number by age group in vaginal (top) and cervical samples (bottom), by age group. The boxplot plots show median CN values, interquartile ranges, and extreme values (whiskers).