Primary HPV testing verification: A retrospective ad-hoc analysis of screening algorithms on women doubly tested for cytology and HPV.

Abstract

To evaluate human papillomavirus (HPV) testing as a primary screening tool, we retrospectively analyzed data comparing (1) HPV testing to the algorithms of the ATHENA Study: (2) cytology alone, (3) cytology with ASCUS triage in women 25-29 and (4) cotesting ≥ 30 or (5) cotesting ≥ 25. We retrospectively analyzed data from women tested with both cytology and HPV testing from 2010 to 2013. Cumulative risk (CR) for CIN3+ was calculated. Crude and verification bias adjusted (VBA) sensitivity, specificity, predictive values, likelihood ratios, colposcopy rate, and screening test numbers were compared. About 15,173 women (25-95, 7.1% <30) had both HPV and cytological testing. Nearly 1,184 (8.4%) had biopsies. About 19.4% had positive cytology, 14.5% had positive HPV. HPV testing unassociated with ASCUS was requested in 40% of women <30, versus 84% ≥30, with similar HPV16/18 genotyping results (68% vs. 70%). 84 CIN3+ were detected with the following 3-year cumulative risk (CR) (95% confidence interval): HPV+/ASCUS+, 46% (32-66%), HPV+/NILM 30% (15-58%), HPV-/ASCUS+ 12% (6-23%), and HPV-/NILM 0.8% (0.2-3.6%). HPV had higher specificity 57% (54-60%) than cotesting ≥30 52% (49-55%). HPV sensitivity 78% (69-87%), positive 12.3% (9.8-15.3%), negative 97 (96-98%) predictive values, positive 1.8 (1.6-2.1) and negative likelihood ratios 0.6 (0.5-0.6), were not significantly different. Cotesting increased colposcopy rate and doubled testing per CIN3+ diagnosed. While HPV-/NILM cotesting results are associated with low CIN3+ risk, HPV testing had similar screening performance to cotesting and to cytology alone. Additionally, HPV testing and cytology incur false negatives in nonoverlapping subsets of patients. Diagn. Cytopathol. 2017;45:580-586. © 2017 Wiley Periodicals, Inc.