International Collaborative Study on Human Papillomavirus Analytical Thresholds for Sensitivity and Specificity in Cervical Screening

ABSTRACT Human papillomavirus (HPV) assays vary regarding the minimum amount of virus they detect. We investigated analytical thresholds of HPV detection and cervical screening sensitivity and specificity. One hundred cervical intraepithelial neoplasia grade 2 or worse (CIN2+) cases and 200 matched population‐based controls were obtained at the Swedish National HPV Reference Laboratory and analyzed by 10 laboratories across 10 countries. Cumulative sensitivity (weighted according to the global HPV type distribution in invasive cervical cancer (ICC)) and specificity were estimated at varying analytical detection thresholds. Consensus results found HPV in 99/100 CIN2+ cases and 52/200 controls. HPV16 prevalence declined in HPV‐vaccinated birth cohorts, among both cases and controls. Line plots of 1‐specificity and ICC‐weighted sensitivity found optimal analytical detection thresholds as 3 International Units (IU)/µl for HPV16/18, 25 IU/µl for HPV31/33/35/45/52/58 and 100 genome equivalents (GE)/µl for HPV 39/51/56/59 resulting in 92.00% cumulative specificity and 90.08% ICC‐weighted sensitivity. Thresholds defined using virus amounts per 10 4 human cells gave similar results. Comparator assay testing using manufacturer‐defined thresholds achieved high ICC‐weighted sensitivity (96.61%) but low specificity (82.50%). This international collaborative study has identified HPV analytical detection thresholds optimizing the sensitivity and specificity of cervical screening.