LETTER TO THE EDITOR Permanent Hair Dyes and Bladder Cancer Risk Manuela GAGO-DOMINGUEZ 1 , Kenneth K. CHAN 2 , Ronald K. ROSS 1 and Mimi C. YU 1 * 1 USC/Norris Comprehensive Cancer Center, University of Southern California Keck School of Medicine, Los Angeles, CA, USA 2 The Ohio State University College of Pharmacy, Columbus, OH, USA Dear Sir, Dr. Henley and Dr. Thun presented the latest results from the ACS/CSP-II study, a cohort study begun by the American Cancer Society in 1982. 1 Consistent with their earlier null findings, no association was observed between use of perma- nent hair dyes and death from bladder cancer, irrespective of subjects’ smoking status. Our case-control results, 2 which are based on incident data, are clearly incompatible with the ACS cohort findings, which are based on mortality data. As Dr. Henley and Dr. Thun pointed out in their letter, 1 it is conceiv- able but unlikely that use of permanent hair dyes may affect disease incidence but not disease mortality. We have no fully satisfying explanation for the discrepancy between these data sets, but we provide additional evidence below that we think further supports a causal association between hair dye use and bladder cancer risk. Among published studies on hair dye use and bladder cancer risk, the National Bladder Cancer Study 3 is by far the largest, involving close to 3,000 population-based incident cases and roughly 5,800 population controls in 10 geographic areas of the United States. The study comprised 733 female cases and 1,500 female controls. Types of hair dyes were not investigated in that study. We examined whether our results are consistent with those of Hartge et al. 3 by mimicking their Tables 2 and 3, which show distributions of frequencies and duration of use of hair dyes, respectively, in female cases and controls separately. Among cases, both the National Bladder Study and our study observed a 61% rate of hair dye use (Table I). Likewise, the rates of use in the two sets of controls were very similar (58% in the National Bladder Study versus 55% in Los Angeles). Distributions by frequency of use were similar between the two data sets, among both case patients and control subjects (Table I). On the other hand, the subjects (both cases and controls) in Los Angeles generally had longer duration of use than their counterparts in the National Bladder Cancer Study (Table I). This discrepancy could reflect a cohort effect in prevalence of hair dye use among US women: the National Bladder Cancer Study was conducted in the late 1970s, while the Los Angeles Study was conducted in the 1990s. Table II shows the simultaneous effects of cigarette smok- ing and permanent hair dye use on female bladder cancer, based on the Los Angeles Study data. Highest risk was observed in smokers who were users of permanent hair dyes. The smoking  hair dye interaction data are suggestive of an additive, rather than multiplicative, effect of the two exposures on bladder cancer risk. In our earlier paper, 2 we had presented separate relative risks for lifelong nonsmokers and ever smokers (last paragraph on page 576). These latter relative risks are not as easy to interpret as those presented in Table II, since the baseline risk in smokers differs from that in nonsmokers. Certain aromatic amines, such as 4-aminobiphenyl and b- naphthylamine, are recognized human bladder carcinogens. We and others have hypothesized that the aromatic amines in hair dyes may be the putative carcinogens responsible for increased bladder cancer risk in users. 2 Aromatic amines require meta- bolic activation to realize their full carcinogenic potential, and hepatic N-acetylation, catalyzed by the N-acetyltransferase 2 enzyme, is a recognized detoxification pathway. 4 Consistent with this biological understanding, NAT2 slow acetylators have been shown in multiple studies to exhibit increased blad- der cancer risk, especially among those with documented ex- posures to carcinogenic aromatic amines. 5 The NAT2 pheno- type was assessed using a caffeine-based urinary assay 6 on all consenting participants of our Los Angeles study, including 124 (61%) of the 203 female case patients and 122 (60%) of the 203 female control subjects. Table III presents the salient hair dye-bladder cancer associations stratified by subjects’ NAT2 acetylation status. Irrespective of the exposure index under *Correspondence to: USC/Norris Comprehensive Cancer Center, Uni- versity of Southern California Keck School of Medicine, Los Angeles, CA 90033-0800, USA. Received 4 June 2001; Accepted 29 June 2001 TABLE I – DISTRIBUTIONS OF HAIR DYE USE AMONG FEMALE CASES AND FEMALE CONTROLS IN THE NATIONAL BLADDER CANCER STUDY AND THE LOS ANGELES STUDY Use of hair dyes National Bladder Cancer Study 1 % Cases/ % Controls Los Angeles Study 2 % Cases/ % Controls Non-users 39/42 39/45 Users 61/58 61/55 Number of times per year 3 12/13 7/4 3–9 26/23 30/31 10–12 11/12 20/14 13+ 12/10 4/6 Number of years of use 5 15/13 5/7 5–10 10/10 8/8 10–19 21/18 15/15 20+ 15/17 33/25 1 733 cases and 1,500 controls.– 2 203 cases and 203 controls. Int. J. Cancer: 94, 905–906 (2001) © 2001 Wiley-Liss, Inc. Publication of the International Union Against Cancer