Clinical accuracy of OncoPredict HPV Quantitative Typing (QT) assay on self-samples

Abstract

Background

The VALHUDES initiative was established to assess the clinical accuracy of HPV assays to detect cervical precancers using urine and vaginal self-samples compared to cervical clinician-collected samples. Here, the clinical performance of OncoPredict HPV Quantitative Typing (QT) assay (OncoPredict QT) was evaluated.

Methods

490 women referred to colposcopy self-collected a urine and a vaginal specimen using Colli-Pee and FLOQSwab, respectively. Subsequently, a colposcopy was performed, and a cervical sample was collected with Cervex-Brush, followed by biopsy if clinically indicated. Vaginal samples were transported dry and resuspended in 5 mL of eNAT medium, whilst cervical brushings were immediately transferred in 20 mL ThinPrep.

Results

The clinical sensitivity of OncoPredict HPV QT testing for CIN2+ in urine and vaginal self-samples was similar to cervical samples (ratios of 0.99 [95 % CI 0.94–1.05] and 1.00 [95 % CI 0.96–1.04]), respectively, when manufacturer's cut-offs were applied. The specificity for <CIN2 on both self-samples was lower than on cervical samples (urine/cervical ratio = 0.91 [95 % CI 0.84–0.98]; vaginal/cervical ratio = 0.90 [95 % CI 0.84–0.98]). Cut-off optimisation improved specificity without compromising sensitivity. Median viral load values adjusted for cellularity were significantly higher in cervical samples compared to urine or vaginal self-samples, in general for all 12 high-risk HPV and in particular for HPV16, 18, 31, 33, 35, 45, 51, 58 (p < 0.05). No difference was observed in median viral loads between urine and vaginal samples.

Conclusion

Following cut-off optimisation OncoPredict HPV QT assay demonstrated similar accuracy on self-collected versus cervical samples.