Colposcopy referral and CIN3+ risk of human papillomavirus genotyping strategies in cervical cancer screening

Background: High-risk human papillomavirus (hrHPV)-based cervical cancer screening in the Netherlands led to a substantial increase in number of colposcopy referrals and low-grade lesions detected. Genotyping strategies may be employed to lower the screening-related burden. Methods: We evaluated 14 triage strategies with genotyping (HPV16/18 or HPV16/18/31/33/45/52/58) for hrHPV-positive borderlineormilddyskaryosis (BMD)ornormal cytology,usingdata from a population-based hrHPV-based screening trial with 5-year interval (POBASCAM). We considered colposcopy referral at baseline, after 6-month repeat cytology and after 5-year hrHPV testing. Performance was evaluated by one-round positive and negative predictive value (PPVandNPV) for CIN3+ and by two-roundcolposcopy referral rate. To identify efficient strategies, they were ordered by the one-round colposcopy referral rate. Adjacent strategies were compared by the marginal PPV for detecting one additional CIN3+ (mPPV). Results: The most conservative strategy (repeat cytology after BMD and HPV16/18/31/33/45/52/58-positive normal cytology, next round otherwise) yielded an mPPV of 28%, NPV of 98.2%, and two-round colposcopy referral rate of 47.2%. Adding direct referral after BMD or genotype-positive BMD yielded an mPPV ≤ 8.2%, NPV ≥ 98.5% and an increase in colposcopy referral rate of 1.9% to 6.5%. Adding direct referral after HPV16/18-positive normal cytology yielded an mPPV ≤ 3.5%, NPV ≥ 99.5% and an increase in colposcopy referral rate of 13.9%. Conclusions: Direct colposcopy referral of women with BMD or normal cytology is unlikely to be efficient, but genotype-guided direct referral after BMD may be considered because the increase in colposcopies is limited. Impact: hrHPV screening programs can become very efficient when immediate colposcopy referral is limited to women at highest CIN3+ risk.