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

Abstract

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 in women living with HIV, leads to overtreatment. We evaluated whether HPV type restriction and more stringent cutoffs on Xpert HPV optimise performance characteristics of this assay for screen and treat.

Methods: We recruited HIV-negative and HIV-positive women aged 30-65 years from a primary care facility and a referral colposcopy clinic in Cape Town, South Africa. Women included had no history of any anogenital cancer or treatment for cervical dysplasia, had no hysterectomy, and were not pregnancy at the time of recruitment. All women had cervical samples collected for Xpert HPV (an assay that detects high-risk HPV types in five channels: HPV type 16; HPV types 18 or 45, or both; HPV types 31, 33, 35, 52, or 58, or more than one of these types; HPV types 51 or 59, or both; and HPV types 39, 56, 66, or 68, or more than one of these types) and underwent colposcopy and histological sampling with consensus pathology review. Logistic regression and receiver operating characteristic curves were used to evaluate improvements in specificity attained by modifying cycle threshold cutoffs to define screen-positive results.

Results: We recruited 1121 women aged 30-65 years, 586 of whom were HIV-negative and 535 HIV-positive. Sensitivity of detecting cervical intraepithelial neoplasia grade 2 or greater in HIV-negative women using manufacturer-defined cycle threshold cutoffs for all channels was 88·7% (95% CI 83·1-94·3), and specificity was 86·9% (83·4-90·4). Sensitivity was 93·6% (90·0-97·3) and specificity 59·9% (54·1-65·7) in HIV-positive women. Cycle threshold values from channels detecting HPV type 16, HPV types 18 or 45 (or both), and HPV types 31, 33, 35, 52, or 58 (or more than one of these types) were informative to predict cervical intraepithelial neoplasia grade 2 or greater. Shifting cycle threshold cutoffs on these three channels allowing sensitivity to decline to 75-85%, led to specificities of 91·3-95·3% in HIV-negative women and 77·0-85·8% in HIV-positive women.

Interpretation: More stringent cycle threshold cutoffs on selected channels in Xpert HPV improve specificity with only modest losses in sensitivity, making this assay an optimal choice for HPV-based screen and treat in settings with a high prevalence of HIV. These modifications can be made from standard output with no need for new engineering. Decision making about performance characteristics of HPV testing can be shifted to programme implementers and cutoffs selected according to resource availability and community preferences.

Funding: Supported by the National Cancer Institute UH2/3 CA189908.