The LEUKEMIAS constitute less than 5% of malignancies in most countries, but they have received extensive epidemiologic attention because of the high case-fatality rate and the heterogeneity of a growing number of recognized subtypes.1,2 Leukemia was the first neoplasm to occur in excess among the Japanese atomic bomb survivors, to develop with increased incidence among patients treated with alkylating agents, and to be conclusively linked with a human retrovirus. Because these and other recognized etiologic factors account for only a small fraction of all leukemia cases, more recent epidemiologic investigations have attempted to quantify leukemia risks that may be associated with cigarette smoking and other more common exposures, including diagnostic roentgenograms, non-ionizing electromagnetic field exposures, solvents, and other chemicals. In this issue of the Archives, Brownson and colleagues3 present a meta-analysis of epidemiologic studies examining the cigarette smoking—leukemia relationship.
Austin and Cole4 were the first to suggest,