Background: Recent evidence implicates smoking as a risk factor for cervical cancer (CC), but the confounding from high-risk human papillomavirus (HPV) infections is not clear.
Objectives: To analyse the role of smoking as an independent predictor of CIN2+ and HR-HPV infections in a population-based prospective (NIS, New Independent States of former Soviet Union) cohort study.
Study design and methods: A cohort of 3,187 women was stratified into three groups according to their smoking status: (i) women who never smoked; (ii) those smoking in the past; and (iii) women who are current smokers. These groups were analysed for predictors of (a) HR-HPV; (b) high-grade CIN, and (c) outcome of HR-HPV infections and cytological abnormalities during prospective follow-up (n = 854).
Results: The three groups were significantly different in all major indicators or risk sexual behaviour (or history) implicating strong confounding. There was no increase in HSIL/LSIL/ASC-US cytology or CIN1+/CIN2+/CIN3+ among current smokers. Only few predictors of HR-HPV and CIN2+ were common to all three groups, indicating strong interference of the smoking status. There was no difference in outcomes of cervical disease or HR-HPV infections between the three groups. In multivariate model, being current smoker was one of the five independent predictors of HR-HPV (P = 0.014), with adjusted OR = 1.52 (95%CI 1.09-2.14). In addition to age, HR-HPV was the only independent predictor of CIN2+ in multivariate model (OR = 14.8; 95%CI 1.72-127.31).
Conclusions: These data indicate that cigarette smoking is not an independent risk factor of CIN2+, but the increased risk ascribed to smoking is mediated by acquisition of HR-HPV, of which current smoking was an independent predictor in multivariate model.