Comparison of Human Papillomavirus Genotyping by Research vs. Clinical Assay for Two Self-Collection Devices

Abstract

Background: Human papillomavirus (HPV) assays and self-collection devices for HPV detection have evolved. We aim to compare two self-sampling devices against speculum-based testing for HPV genotype agreement and their accuracy for cervical intraepithelial neoplasia grade 2 (CIN2+) disease. Secondarily, we aim to compare two HPV assays for different HPV genotype detection agreement and their accuracy for CIN2+ disease. Methods: Women from colposcopy (N ¼ 97) and primary care (N ¼ 96) were block-randomized to two different self-sampling device groups. Self-sampling and speculum-collected pairs of HPV specimens were analyzed with the research assay. A second speculum-collected specimen provided clinical results using the clinical HPV assay. Agreement (prevalence-based κ) and accuracy (sensitivity/specificity ratios) provided the statistical comparison. Results: The two devices did not differ in their κ agreement scores for overall HPV detection compared with the speculum-collected sample [κ ¼ 0.83 (0.72–0.94) and κ ¼ 0.90 (0.81–0.98), respectively, nonsignificant exact McNemar test results]. The two devices did not differ in accuracy as measured by the relative sensitivity/specificity for overall HPV at the CIN2+ disease threshold [1.0 (0.15–6.77) and 1.19 (0.56–2.54), respectively]. The two assays did not differ in HPV agreement nor assay accuracy for CIN2+ (n ¼ 10). Conclusions: HPV self-sampling devices robustly detected high-risk HPV types for cervical cancer screening when using the research assay to compare them. Both research and clinical HPV assays provide equivalent HPV detection for specific and aggregated HPV types.