Smoking and risk of glioma: a meta-analysis

Although causal relationships between smoking and cancer risk have been established for many sites, most studies of brain cancer have not supported an association. However, two recent cohort studies showed increased risks of glioma among smokers. We quantified the association between smoking and glioma through a meta-analysis of the literature. Of 20 eligible studies, 17 (6 cohort and 11 case-control) were included in an analysis of ever versus never smoking. Multivariate-adjusted risk estimates in the papers were pooled to calculate cumulative risk. The cumulative estimated risk associated with ever smoking was 1.06 (95% CI: 0.97-1.15), for all, 1.10 (95% CI: 1.01-1.20) for cohort, and 1.00 (95% CI: 0.88-1.15) for case-control studies. A significantly increased risk associated with past smoking was noted for cohort studies, OR = 1.16 (p = 0.007), while an increased risk of borderline significance was seen for all studies, OR = 1.10 (p = 0.08). In general, dose-response analysis did not support an association and was limited because very few studies included these variables and could be pooled. Overall, results of pooling of all studies suggested that smoking is not associated with risk of glioma. However, the small but significant increased risk seen for cohort studies remains to be clarified.