| ObjectivesThis study was aimed to explore the clinical performance of liquid-based cytology(LBC),viral-load,HPV genotyping,E6/E7 oncoprotein,and p16 protein to triage HPV positive women,so as to provide scientific basis for optimizing HPV-based cervical cancer screening program.Materials and MethodsBased on the Shanxi Province Cervical Cancer Screening Study]and 2(SPOCCS1 and SPOCCS2),we targeted the population from the last screening of both studies(SPOCCS1:n=1416;SPOCCS2:n=2112)and performed different triage tests including LBC,HPV DNA genotyping,E6/E7 oncoprotein testing and p16 testing on the residual samples or fresh samples of HPV positive women.Triage performance,including sensitivity,specificity,positive predictive values(PPV)and negative predictive value(NPV),and their respective 95%confidence interval(95%CI)for Cervical Intraepithelial Neoplasia grade 2/3 or worse(CIN2+/CIN3+)detection were estimated.To evaluate the colposcopy efficiency,colposcopy referral rates and number needed to be referred to colposcopy to detect per case(NNR)were estimated.Besides,Pre-test and Post-test Probability(PPP)was estimated to reveal the utility of triage tests.ResultsFor detection of CIN2+among HPV positive women,sensitivity and specificity of LBC(ASC-US+)were 85.0%and 61.2%.By increasing the cytology abnormal cut-off to LS1L+ or ASC-H+,the decreased sensitivity and increased specificity were observed,and the colposcopy referral rates decreased as well.When using HPV viral-load as the triage method,the same trend was observed.Viral-load triage at 40 RLU/CO cut-off achieved the similar triage performance with LBC(ASC-US+)with the sensitivity and specificity at 83.8%and 61.7%.Increasing the number of targeted genotypes resulted the increased sensitivity and decreased specificity.Triage strategy targeting 5 hrHPV(HPV16/18/31/33/45)achieved the sensitivity of 71-73%,which was significantly higher than HPV 16/1 8 triaging(51-55%).While targeting 7 hrHPV(HPV16/18/31/33/45/5 2/58),the sensitivity increased to-90%,while the specificity was only at-45%.HPV E6/E7 oncoprotein test targeting 8 hrHPV types showed the sensitivity of 67.7%,which was about 15%higher than HPV 16/18 E6 test,but only-4%less specific(89.5%vs 94.3%).p16 triaging showed slightly superior sensitivity(87.8%)and specificity(68.8%)than LBC(ASC-US+),but not statistically significant.Compared to one-step triage,two-step triage algorithm using HPV2 6/18 with reflex LBC achieved the highest sensitivity(98.0%)and slightly less specificity(-57%)than LBC alone,but not statistically significant.While HPV 16/1 8 with reflex p16 also achieved the similar highest sensitivity,but not statistically significant less specificity(-61%).PPP plots showed that if two-step triage both negative,the risk of CIN2+was<2%and the risk of CIN3+ was<1%,which fall into the "safe"zone.ConclusionsOur study suggested that extended genotyping and viral-load with increased cut-point could achieve the similar performance to cytology triage.E6/E7 oncoprotein showed the superiority in specificity.p16 testing combined the relatively high sensitivity and specificity better than other triage algorithms.As the molecular-based triage algorithms and thereby the objectivity in result interpretation,mentioned methods could be the promising alternatives to cytology triage for triaging HPV positive women.Combined triage algorithms of HPV16/18 with reflex p16 or cytology,could achieve the better risk stratification of HPV positive women,which is of great importance to detect the most cases,meanwhile reduce the unnecessary colposcopy referrals or repeat testing.Different settings need to consider the local risk threshold,resource availability and balance the benefits and harms of different screening algorithms when incorporating them into practice. |