Estimating lung cancer risk from e-cigarettes and heated tobacco products: applications of a tool based on biomarkers of exposure and of potential harm.

Abstract

Background: Reliable epidemiological data are limited on the lung cancer risk of groups using e-cigarettes (ECIGs) and groups using heated tobacco products (HTPs).

Aim: We describe a methodology to estimate the lung cancer risk of these groups according to their levels of biomarkers of exposure (BOEs) and of potential harm (BOPHs).

Methods: Using 28 search terms for BOEs and 82 for BOPHs we sought publications reporting biomarker-specific data from North America and Europe comparing individuals who smoke cigarettes and individuals who use other established products (ETPs; cigars, pipes, smokeless tobacco (ST) and/or snuff/snus). Publications were selected using defined inclusion/exclusion criteria. Additionally using lung cancer relative risk (RR) estimates for users of specific ETPs derived from recent meta-analyses of epidemiological studies in these regions, we derived a regression model predicting the lung cancer RR by level of each specific biomarker. Separately for groups using ECIGs and using HTPs the lung cancer risk was then estimated by combining RR estimates for selected biomarkers. Our primary estimates only considered biomarkers statistically significantly (p < 0.01) related to lung cancer risk which showed no significant (p < 0.01) misfit to the RR of 1.0 for non-users-those with no use of ETPs, ECIGs or HTPs.

Results: Based on 38 available publications, we extracted biomarker-specific data for ETPs for 56 BOEs within 21 of the 28 search terms considered and for 54 BOPHs within 29 of the 82. The regression slope fitted to the lung cancer risk was significant (p < 0.01) for 22 BOEs and six BOPHs. However, the predicted RR for non-users significantly (p < 0.01) differed from 1.0 for 16 of these biomarkers. We estimated the lung cancer RR for using ECIGs, derived from 30 estimates for 10 biomarkers, as 1.88 (95% CI 1.60-2.22), the excess risk (ER = RR - 1) being 6.8% of that for smokers of cigarettes. The RR estimate varied little in most sensitivity analyses conducted, but increased markedly after removing the restriction to significant model fit. We estimated the lung cancer RR for using HTPs, combining estimates for four BOEs, as 1.44 (0.41-5.08), the ER being 3.4% of that for smokers of cigarettes.

Conclusions: Despite some methodological limitations, our approach estimates risk when reliable epidemiological data are unavailable. Using the biomarkers considered here, the model indicates that the lung cancer risk for individuals using ECIGs is much lower than for smokers of cigarettes, and suggests that the risk for those using HTPs is also low. Research using additional data could add precision to these findings.