Personal Permanent Hair Dye Use Is Not Associated with Bladder Cancer Risk: Evidence from a Case-Control Study

Hair dyes are widely used in the United States, Europe, and East Asia. It is estimated that more than one third of women ages >18 years and >10% of men ages >40 years in the United States use some type of hair dye products, among which permanent hair dyes are the most widely used category and contribute to >80% market share in United States (1). The fact that almost any hair dye on the market contains small amounts of aromatic amines and that aromatic amines are carcinogenic in animals (2, 3) has generated great interest in studying the association between hair dye use and cancer risk, especially bladder cancer risk. Although occupational exposure to hair dyes by hairdressers and barbers has been shown to increase bladder cancer risk in a number of epidemiologic studies (4), whether or not personal hair dye use may increase bladder cancer risk remains controversial with fewer studies published and fewer cases and exposed individuals (4-6). In this study, we used one of the largest bladder cancer case-control studies to evaluate the association between personal hair dye use and bladder cancer risk.

Bladder cancer patients were recruited from The University of Texas M.D. Anderson Cancer Center and Baylor College of Medicine. Cases are enrolled from two ongoing studies: a bladder cancer case-control study and a case cohort study with the common goal of identifying genetic susceptibility to bladder carcinogenesis and clinical outcome. The case-control study started patient recruitment in 1999 and the case cohort study began in 2001. The procedures for subject recruitment and eligibility criteria were previously described (7). Briefly, cases are newly diagnosed and histologically confirmed urinary bladder cancer patients who had not received any previous chemotherapy or radiotherapy before enrollment. There were no recruitment restrictions on age, gender, ethnicity, or cancer stage. Cases enrolled in the case-control study are all Texas residents. Case recruitment is expanded to include all U.S. residents in the case cohort study. Among all cases recruited, the majority (∼70%) is from Texas. We compared education level and total household income between residents of Texas and those from other states and found no significant differences. The controls are recruited in collaboration with the Kelsey-Seybold clinics, the largest private multispecialty physician group consisted of >23 clinics and >300 physicians in the Houston metropolitan area. The majority of controls visit the clinics for annual health check-ups. Controls are frequency matched to cases by age (±5 years), gender, and ethnicity. Controls have no prior history of cancer (except nonmelanoma skin cancer). For both cases and controls, after obtaining written informed consent, trained M.D. Anderson staff interviewers gave risk factor questionnaires to study participants. Data were collected on demographic characteristics (age, gender, ethnicity, etc.), work history, tobacco use history, and personal hair dye use. For personal hair dye use data, each participant answered questions about ever use of hair dye products, age when first started using hair dyes, frequency of use, type of hair dye used (permanent or semipermanent), and color of the most frequently used hair dye. Regular use of hair dye is defined as using hair dye products for 1 year or more.

Differences between cases and controls in the distributions of age, gender, ethnicity, smoking, and hair dye use were tested using the χ² test, Student's t test, or Wilcoxon rank sum test when appropriate. Multivariate conditional logistic regression was done to calculate odds ratios (OR) while adjusting residual confounding effects of age and smoking status. All statistical tests were two sided with a type I error rate of 5%. Statistical analyses were done with the Stata software (version 8, College Station, TX).

The study included 712 bladder cancer cases and 712 healthy controls matched to cases by age, gender, and ethnicity. Gender and ethnicity were perfectly matched and there was no significant age difference between the cases (63.4 ± 11.2) and the controls (62.5 ± 10.9; P = 0.14). The cases had a significantly higher percentage of current (26.5%) and ever smokers (73.4%) than the controls (8.6% and 54.2%, respectively; P < 0.001). Among ever smokers, cases reported significantly higher levels of cigarettes consumption than controls (median pack-years, 37.5 versus 21.0; P < 0.001). The percentage of cases reporting ever use of hair dye was similar to that of the controls (26.5% versus 26.8%; P = 0.91). The use of different colors of hair dyes and the starting age of use were also similar in cases and controls (P = 0.75 and 0.12, respectively; data not shown).

Using subjects who had never used any hair dye products as the reference group, the ORs for regular use of permanent hair dyes and semipermanent hair dyes were 0.81 [95% confidence interval (95% CI), 0.50-1.30] and 0.82 (0.52-1.31), respectively (Table 1). The lack of association was seen in both women and men (Table 1). We then assessed the association between duration of use, frequency of use, lifetime use, age at first use, and bladder cancer risk (Table 2). The results were consistent in overall analysis and in either sex alone. However, due to the small sample size of male users, we only reported results for all subjects and for women only. In all variables, we chose cutoff points using quartiles and tertiles in controls, as well as cutoff points reported in the literature. As results were consistent regardless of the type of cutoff points chosen, we only reported tertile analysis and results based on literature cut off points. There were no significant associations overall and in women between bladder cancer risk and duration of use, frequency of use, lifetime use, and age at first use (Table 2). In addition, we also stratified the analysis by hair dye color because dark-colored dye contains higher concentration of dye loads than light-colored dyes (8). Again, no significant associations were found in any strata although the ORs tended to increase in users of black dyes (Table 2). Further, major formulation changes in hair dyes occurred around 1980, when many manufacturers replaced carcinogenic compounds in hair dye products (9); therefore, we did stratified analyses according to whether the subject started using hair dye products before or at 1980 or after 1980. No association was found in either stratum. The above analyses were also stratified by smoking status and there was no evidence of joint effects between hair dye use and smoking in bladder cancer risk (data not shown).

There are a limited number of studies evaluating the association between personal hair use and bladder cancer risk and the results have been inconsistent. Two recent meta-analyses yielded contradictory conclusions, one found a pooled relative risk of 1.01 (95% CI, 0.89-1.14; ref. 5) and the other concluded that there was a 22% to 50% increased bladder cancer risk for permanent hair dye users (6). However, the latter report did not include the most recent negative case-control study (10) and the pooled significant ORs were obtained by excluding selected studies that the authors deemed questionable in study design. The largest cohort study with ∼600,000 women (including 336 bladder cancer deaths) produced a RR of 1.08 (95% CI, 0.84-1.38) between permanent hair dye use and bladder cancer mortality after 16 years of follow-up (11). The caveat is that this report used cancer mortality as the end point, which would leave out the predominant bladder cancer subtype—superficial bladder cancer. It is known that superficial and invasive bladder cancer have different etiology (12). The largest multicenter case-control study (2,249 male cases, 733 female cases, and 5,792 controls) reported no significant association between hair dye use and bladder cancer risk (OR, 0.9; 95% CI, 0.8-1.1 for women; OR, 1.1; 95% CI, 0.8-1.4 for men; ref. 8). The lone positive study with 897 cases and matched controls found women who used permanent hair dyes at least once a month exhibited a 2.1-fold increased risk of bladder cancer (P_trend = 0.04). The risk increased to 3.3 (95% CI, 1.3-8.4) among regular monthly users of ≥15 years (13). This current study is the third largest case-control study and we did not find any significant associations. Neither did all the other earlier smaller studies find overall associations (14-18). Given the proportion of exposed subjects in controls, total number of cases, and the case/control ratio, our study has 90% power to detect an increased OR of 1.46 and 80% power to detect an increased OR of 1.39.

Based on these results and the very modest increase of bladder cancer risk—slightly above unity—in hair care professionals, it seems that personal hair dye use does not cause an appreciable increase in bladder cancer risk. However, a possibility of increased risk in certain subgroups cannot be rued out given the nonsignificant increased risk in users of dark-colored dyes. Further, the overall lack of association does not exclude possible increased risk in certain genetic susceptible subpopulations (19).

As this is a hospital-based case-control study, selection bias is a potential limitation. To control for confounding from variables that could possibly correlate with hair dye use habits, we further adjusted education level and total household income in the multivariate model. The similar nonsignificant results were obtained. As bladder cancer cases were recruited from a large referral center, invasive bladder cancer subtype is overrepresented in the study population (∼45.0% of cases have invasive disease). It should also be noted that the small sample size in stratified analyses results in unstable risk estimates with rather wide confidence intervals. Further, although this is a large study, only a small percentage of subjects are regular hair dye users. For example, the percentages of regular users are 26.1% in cases and 26.5% in controls.

Our study has the strength of collecting data on hair dye type, duration of use, frequency of use, and color of use. Very few previous studies collected similarly detailed information (13). As one of the largest studies examining personal hair dye use and bladder cancer risk, our study provides valuable data for future meta-analyses to further address the putative associations.