little evidence that this locus influences smoking behavior. Thus, these authors concluded that the association was primarily with lung cancer. Because the association of smoking with this disease is very strong, it is difficult to detect any residual independent association with a gene variant that primarily acts through increasing exposure to smoking. Here, we show with biomarkers that carriers of the risk variants in CHRNA3 and CHRNA5 smoke more intensively and are exposed to a higher internal dose of a carcinogenic tobacco-specific nitrosamine per cigarette dose than non-carriers. These data are consistent with the overall increased lung cancer risk associated with this locus and suggest that number of cigarettes per day, even if it could be accurately assessed, is not an adequate surrogate for smoking dose when examining the independent effect of the 15q locus on lung cancer.
