Estimating lung cancer risk from e-cigarettes and heated tobacco products based on biomarkers of exposure and of potential harm (b) Results

Abstract

Background: Epidemiological data on the lung cancer relative risk (RR) from using e-cigarettes or heated tobacco products (HTPs) is scarce. Our first paper described methodology to estimate this using data on biomarkers of exposure (BOE) and potential harm (BOPH). This used RR estimates for cigarette smokers and users of other tobacco products (e.g. cigars) and data on biomarker levels for smokers, other product users and non-users. It was then applied, to illustrate the methodology, to evidence from three studies reporting BOE data for smokers and users of e-cigarettes and/or HTPs. Aim: To estimate the lung cancer RR for e-cigarette and HTP use from studies providing relevant biomarker data for smokers and users of e-cigarettes and/or HTPs. Methods: Searches were carried out for terms where relevant biomarker data were identified previously, and papers identified providing data for smoking and for either e-cigarettes or HTPs. Our main RR estimates for e-cigarettes and HTPs were derived from results for biomarkers showing significant (p<0.01) fit to the regression model used, and no significant (p<0.01) misfit to the non-user RR of 1.0. Sensitivity analyses were carried out, as previously described. Results: Our estimated RR for e-cigarettes, based on 10 different biomarkers, is 1.83 (95% CI 1.53-2.19), represents 6.5% of the excess risk (ER = RR−1) for cigarette smokers. The RR generally varied little in sensitivity analyses, but increased markedly where the restriction to significant model fit was removed, the results combined then including some very large and implausible RRs. Our estimated HTP RR of 1.44 (95% CI 0.41-5.08) represents 3.4% of the ER for smokers. This RR estimate is as given previously, no additional biomarker studies being found satisfying the requirements for inclusion in the combined estimates. As noted previously, sensitivity analyses little affected this estimate. Conclusions: Our methodology has limitations, but is useful for estimating disease risk where epidemiological data is absent. Applying it to biomarkers with satisfactory model-fit suggests the lung cancer risk from e-cigarettes is much less than from cigarettes. Limited data indicates the risk from HTPs is also low. Future research using additional biomarker data can extend these findings.