Risk Factors Exploration Of High-Risk Human Papillomavirus Infection And Cervical Cancer And Development Of Risk Prediction Models

ObjectivesTo explore the risk factors and develop risk prediction models for high risk human papillomavirus(HR-HPV)infection and cervical cancer and precancer lesions in Chinese population,and then to provide some scientific evidence for identifying the high risk population and generating the screening strategy.Materials and Methods1.Risk factor evaluation and prevalence model building for HR-HPV infection,cervical cancer and precancer lesions:(1)Based on the pooled data of 18 population based cervical cancer screening studies conducted by Cancer Hospital Chinese Academy of Medical Sciences(CICAMS)between 1999-2008,we obtained the risk factor information by epidemiological questionnaire,HR-HPV infection and pathology results were provided by center laboratory and department of pathology of CICAMS.(2)Independent and interaction association between demographic characteristics,sexual related factors,parity,smoking,trichomonas infection and HR-HPV infection,cervical intraepithelial neoplasia grade 2 or worse(CIN2+)were evaluated using logistic regression by single variate and multivariate analysis and stratified analysis.(3)Prevalence risk prediction models including determined factors were built using logistic regression model,and accuracy and calibration of these models were evaluated by area under curve(AUC)and 95%confidence interval and Hosmer-Lemeshow test.2.Development of incidence risk prediction model for cervical cancer and precancer lesions:(1)Based on the prospective cohort which was followed up four times(in 1999,2005,2010 and 2014,respectively)data,the women with whole four time follow-up results were included for the association analysis between potential risk factors and HR-HPV infection,cervical cancer and precancer lesions using the generalized estimation equation.(2)Based on the later three-time visits excluding the baseline CIN2+ cases,we developed the incidence risk prediction model for CIN2+ using generalize estimation equation,and the accuracy and calibration of the model was evaluated by area under curve(AUC)and 95%confidence interval and calibration slope.Results1.Risk factor evaluation and prevalence model building for HR-HPV infection,cervical cancer and precancer lesions:(1)Age,area,education,number of female sexual parterner,husband's affair,condom use and meonopause status were independent factors for HR-HPV infection.While age,age of sexual debut,oral contraceptive were shown as the independent risk factors to the CIN2+ after adjusting for HR-HPV(2)After adjusting for confounding factors,active smokers had 1.45(95%CI:1.10-1.91)times increased risk for HR-HPV infection and had 1.89(95%:1.03-3.44)increased times for CIN2+ compared to the non-smokers.While passive smoker has slightly increased risk of infecting HR-HPV,adjusted OR and 95%CI was 1.11(1.00-1.24);There is no significant association was found between passive smoking and CIN2+.(3)Women with trichomonas infection has increased risk for HPV infection either by LBC diagnosis or by self-reporting,the ORs and 95%confidence interval were 1.33(1.12-1.58)and 1.00(0.91-1.10).After adjusting for HPV infection and other confounding factors,the trichomonas showed a decreased risk for CIN2+ by LBC diagnosis and by self-reporting with ORs and 95%CIs:0.56(0.35-0.91)and 0.70(0.56-0.88).(4)The sensitivity and specificity of prevalence risk model for HR-HPV infection were 61.9%and 59.9%;the CIN2+ model including VIA test had 57.6%sensitivity and 84.9%specificity,another CIN2+ model including the HPV test and VIA showed 96.3%sensitivity and 85.5%specificity.2.Development of the risk prediction model for cervical cancer and precancer lesions:(1)Both of the Age and extramarital sexual behavior were the independent risk factors for the multi-time HR-HPV infection with ORs and 95%CIs of 1.40(1.22-1.630)and 1.58(1.39-1.81),respectively.After adjusting for HR-HPV infection,postmenopausal women had 1.67(95%CI:1.26-2.23)higher risk of developing CIN2+ compared to pre-menopausal women;no significant association was found between other factors and CIN2+ development.(2)Incidence risk prediction model for cervical cancer and precancer lesions were built by including age,menopause status,follow-up time and their interaction,the AUC and 95%CI for the incidence model was 0.866(0.857-0.875),regression coefficient slope of observed response probability and predictive probability is 1.Conclusions1.Given the various effect of age,age of sexual debut,oral contraceptive,smoking and trichomonas infection on HR-HPV infection and CIN2+,we recocomend a comprehensive prevention and control stratigies:avoid early sexual debut,encourage women with age above 35 yrs and early sexual debut to receive regular cervical screening,quit active smokeing and launch anti-smoking activities,prevent low genital tract infection and encourage early treatment.Additionally,the models for HR-HPV infection and CIN2+ model including VIA test in our study have a potential value for HR-HPV infection and cervical cancer screening,while another CIN2+model including the HPV test and VIA is not recommended.2.According to our results based on longitudinal data,we recommend a close monitoring and follow-up for women with HPV infection,high-risk sexual behavior factors(early sexual behavior,women with more than two sexual partners)and postmenopausal women to prevent cervical cancer.In addition,the incidence predictive model of cervical cancer and precancerous lesions in our study can be used as a scientific basis for risk assessment of health management.