Comparison of human papillomavirus testing strategies for triage of women referred with low-grade cytological abnormalities.

AIM To compare triage strategies using different human papillomavirus (HPV) consensus and genotyping tests and a p16(INK4a) test. METHODS 1228 women referred with a borderline or single mildly dyskaryotic smear. Samples were taken at colposcopy using PreservCyt. Tests included Hybrid Capture 2, Abbott RealTime PCR, BD HPV, Cobas 4800, PreTect HPV-Proofer, APTIMA and p16(INK4a). Results were based on the worst histology within 9 months. RESULTS 97/1228 (7.9%) women had CIN3+ (203/1228 (17%) CIN2+). HPV testing alone using Hybrid Capture 2, Abbott RealTime PCR, BD HPV, Cobas 4800 or APTIMA had a sensitivity for CIN3+ ranging from 99.0% to 100.0% and specificity for <CIN2 from 23.3% to 34.7%. p16(INK4a) had a sensitivity of 86.8% and specificity of 50.7%. PreTect HPV-Proofer had a sensitivity of 85.1% and specificity of 73.2%. Testing for HPV type 16 only had sensitivities ranging from 66.0% to 75.5% and specificities from 81.3% to 87.6%. Dual testing with HPV type 16 combined with p16(INK4a) gave a high sensitivity for CIN3+ (78.7% to 98.0%) and specificity for <CIN2 of 58.6% to 81.5%. CONCLUSIONS Triage with sensitive HPV testing assays can substantially reduce the number of unnecessary referrals in women with low grade cytology with virtually no loss of sensitivity. Even greater gains can be made if p16 and type 16 are used, but some cases of CIN2 will be missed. In both cases short term surveillance will be needed.