Alcohol drinking and non-Hodgkin lymphoma risk: a systematic review and a meta-analysis.
PURPOSE Limited data suggest that alcohol drinking may have an inverse relation to risk of non-Hodgkin's lymphoma (NHL). Prospective data about alcohol, NHL, and other hematologic malignancies (HM) are sparse. METHODS We carried out a cohort study in a multiethnic population of 126,293 adults who supplied baseline information at health examinations. There were subsequent HM diagnoses in 1244 persons. We used Cox proportional hazards models with seven covariates. The role of beverage types was studied by comparing groups with preponderant choices and by studying the role of frequency of drinking beverage types. RESULTS Using lifelong abstainers plus infrequent drinkers as referent, adjusted relative risks (95% confidence intervals) for HM follow: less than one drink per day=1.0 (0.9-1.2), one to two drinks per day=0.9 (0.7-1.0), greater than three drinks per day=0.7 (0.6-0.9, p=0.008). For 673 NHL these were 1.2 (1.0-1.5), 0.9 (0.7-1.2), and 0.9 (0.6-1.2). Persons reporting greater than three drinks/day had inverse relations to lymphocytic (n= 146) and myelocytic (n= 169) leukemias, with relative risk of 0.5 (0.2-1.0, p<0.05) for each. No major independent relation was seen for choice of wine, liquor, or beer. CONCLUSIONS Alcohol drinking is associated with slightly lower risk of HM, due largely to inverse relations to lymphocytic and myelocytic leukemia.