Clinical evaluation of modifications to a human papillomavirus assay to optimise its utility for cervical cancer screening in low-resource settings: a diagnostic accuracy study

摘要: Summary Background HPV-based screen and treat is the recommended approach for cervical cancer screening in low-resource settings, but quite low specificity of human papillomavirus (HPV) testing, particularly women living with HIV, leads to overtreatment. We evaluated whether HPV type restriction more stringent cutoffs on Xpert optimise performance characteristics this assay treat. Methods recruited HIV-negative HIV-positive aged 30–65 years from a primary care facility referral colposcopy clinic Cape Town, South Africa. Women included had no history any anogenital or treatment dysplasia, hysterectomy, were not pregnancy at time recruitment. All samples collected (an that detects high-risk types five channels: 16; 18 45, both; 31, 33, 35, 52, 58, than one these types; 51 59, 39, 56, 66, 68, types) underwent histological sampling consensus pathology review. Logistic regression receiver operating characteristic curves used evaluate improvements attained by modifying cycle threshold define screen-positive results. Results 1121 years, 586 whom 535 HIV-positive. Sensitivity detecting intraepithelial neoplasia grade 2 greater using manufacturer-defined all channels was 88·7% (95% CI 83·1–94·3), 86·9% (83·4–90·4). 93·6% (90·0–97·3) 59·9% (54·1–65·7) women. Cycle values 16, 45 (or both), 58 informative predict greater. Shifting three allowing sensitivity decline 75–85%, led specificities 91·3–95·3% 77·0–85·8% Interpretation More selected improve only modest losses sensitivity, making an optimal choice settings high prevalence HIV. These modifications can be made standard output need new engineering. Decision about testing shifted programme implementers according resource availability community preferences. Funding Supported National Cancer Institute UH2/3 CA189908.