prospective study of NAT2 acetylation genotype, cigarette smoking

Hunter et al. have computed the relative risks (RRs) of smoking and breast cancer among women with the slow acetylator genotype using as reference group non-smoking rapid acetylators. This way of analysing their data tends to pull the relative risk towards unity and therefore to attenuate the apparent effect of smoking among slow acetylators. In Table III, for example, the multivariate RR of breast cancer among current smokers of 151 cigarettes daily who are slow acetylators is 1.5 (0.7–3.2) relative to never-smoking rapid acetylators. This same relative risk would be 1.9 if it were computed relative to never-smoking slow acetylators (1.5/0.851.9). For the same reason, the RRs shown in Tables III–V for slow acetylators should all be incremented by a factor 1.25 (1/0.8). The rationale for using never-smoking rapid acetylators instead of never-smoking slow acetylators needs to be explained by the authors since it is inconsistent with the tests for interaction reported in their paper. These tests compare the effect of smoking in slow acetylators versus the effect of smoking in rapid acetylators. Unfortunately, readers cannot recompute the RRs and their confidence limits on the basis of the data given in the tables since this is a matched study. We would be therefore very grateful to Hunter et al. if they could present all the matched and multivariate RRs among slow acetylators using never-smoking slow acetylators as reference