Cigarette smoking and risk of lymphoma in adults: a comprehensive meta-analysis on Hodgkin and non-Hodgkin disease.

The aim of the present meta-analysis was to examine comprehensively the association between smoking and lymphoma [Hodgkin lymphoma (HL) and non-Hodgkin lymphoma (NHL)] in adults. Eligible studies were identified, and pooled-effect estimates (odds ratios and relative risks) were calculated for ever, current and former smoking, separately by lymphoma subtype and gender. Metaregression analysis with percentage of male patients, mean age, duration (years of smoking), intensity (pack-years and cigarettes per day) and years since quitting was carried out. Out of the 50 eligible articles, 41 used a case-control design (20 143 NHL cases, 4340 HL cases and 61 517 controls), whereas nine used a cohort design (5748 incident NHL cases, 334 HL cases, total cohort size comprising 1 530 833 smokers). Ever smoking was associated with increased risk for NHL [pooled-effect estimate=1.05, 95% confidence interval (CI): 1.00-1.09] mainly because of the association with T-NHL (pooled-effect estimate=1.23, 95% CI: 1.09-1.38). Ever smoking was also associated with increased risk for HL (pooled-effect estimate=1.15, 95% CI: 1.02-1.30); sizeable associations were observed regarding both nodular sclerosis and mixed cellularity subtypes. Although male study arms pointed to predominantly increased risk for HL, metaregression did not confirm the male preponderance. Dose-response patterns were particularly evident for HL. Cigarette smoking seems to be associated with increased lymphoma risk, especially HL and T-NHL. Further well-designed studies seem to be needed so as to investigate the risk thoroughly, especially for T-NHL subentities, and the extent to which confounding may interfere with gender-related disparities.