Clinical evaluation of primary human papillomavirus (HPV) testing with extended <scp>HPV</scp> genotyping triage for cervical cancer screening: A pooled analysis of individual patient data from nine population‐based cervical cancer screening studies from China

Abstract

OBJECTIVE: To assess the clinical values of extended human papillomavirus (HPV) genotyping in triage of high-risk HPV-positive women, focusing on the trade-off between cervical precancer detections and colposcopy referrals. METHODS: A bivariate random-effects model was used to estimate the diagnostic accuracy of primary HPV screening with following triage strategies to detect cervical precancers: (i) partial genotyping for HPV16/18 combined with cytological testing at atypical squamous cells of undetermined significance threshold (used as the comparator), (ii) genotyping for HPV16/18/58/52, (iii) genotyping for HPV16/18/58/52/33, (iv) genotyping for HPV16/18/58/33/31, (v) genotyping for HPV16/18/58/52/33/31, and (vi) genotyping for HPV16/18/58/52/33/31/39/51. Internal risk benchmarks for clinical management were used to evaluate the risk stratification of each triage strategy. RESULTS: A total of 16,982 women (mean age 46.1 years, range 17-69) were included in this analysis. For CIN3+ detection, triage with HPV16/18/58/33/31 genotyping achieved lower positivity (6.85% vs. 7.35%, p = 0.001), while maintaining similar sensitivity (91.35% vs. 96.42%, p = 0.32) and specificity (94.09% vs. 93.67%, p = 0.56) compared with the comparator strategy. Similar patterns were observed for CIN2+ detection. Women with a positive HPV16/18/58/33/31 genotyping test had high enough risk for CIN3+ for colposcopy referral, while the risk for women with a negative test was below the 1-year return decision threshold according to internal benchmarks. CONCLUSIONS: Our findings suggested extended HPV genotyping is of potential to be used as a triage technique integrated into HPV-based cervical cancer screening, leading to reduced need for colposcopy referral while maintaining similar disease detection and efficient risk stratification.