Hair dye use and risk of lymphoid neoplasms and soft tissue sarcomas
We analyzed the relation between hair dye use and the risk of Hodgkin's disease (HD), non-Hodgkin's lymphoma (NHL), multiple myeloma (MM) and soft tissue sarcomas (STS) in a hospital-based case-control study conducted between 1985 and 1997 in northern Italy. Cases included 158 patients with histologically confirmed incident HD, 446 with NHL, 141 with MM, 221 with STS and controls included 1,295 patients with acute nonneoplastic conditions. Compared to never use of any type of hair dyes, the odds ratio (OR) for ever use was 0.68 (95% confidence interval, CI, 0.40–1.18) for HD, 1.03 (95% CI 0.73–1.44) for NHL, 1.17 (95% CI 0.70–1.97) for MM and 0.73 (95% CI 0.45–1.17) for STS. The OR were close to unity for permanent and semipermanent dyes analyzed separately or when the analysis was restricted to women. Our study indicates that there is no appreciable association between ever use of any type of hair dyes and the risk of HD, NHL, MM or STS.
Hair coloring products are used in Europe, North America and Japan by over 1/3 of women above the age of 18 and by over 10% of men above the age of 40, with permanent type of hair dyes accounting for about 3/4 of global use.1, 2 In the past, hair dyes contained some in vitro mutagenic substances and some animal carcinogens. These were evaluated for their carcinogenicity to humans by a Working Group of the International Agency for Research on Cancer of the World Health Organization, which, however, concluded that personal use of hair colorants could not be considered carcinogenic (Group 3).1, 3 It has also been suggested that some hair dye components are carcinogenic in rodents4, 5 and cause DNA damaging in human peripheral lymphocytes6 after oxidation, although the evidence remains inconclusive.7 Nonetheless, this has raised concerns that hair color products may increase the risk of cancer in humans, including lymphoid neoplasms.
The evidence on the relation between personal use of hair dyes and risk of any type of lymphoid neoplasms is inconclusive.2 Data are available from at least 2 cohort [the Nurses' Health Study8, 9 and the American Cancer Society (ACS) prospective study]10, 11 and 8 case-control studies.12, 13, 14, 15, 16, 17, 18, 19 The last follow-up of the Nurses' Health Study included 244 cases of hematolymphopoietic malignancies; the relative risks were 0.9 for Hodgkin's disease (HD), 1.1 for non-Hodgkin's lymphoma (NHL), 0.6 for chronic lymphoid leukemia, 0.8 for other leukaemias and 0.4 for multiple myeloma (MM).9 The ACS prospective study10 included 1,991 female cases of lymphoid neoplasms, and the relative risks were 1.1 for NHL, 1.1 for leukaemias and 1.0 for MM.11
Data from case-control studies are inconsistent.12, 13, 14, 15, 16, 17, 18, 19 Among the most recent ones, an Italian hospital-based investigation based on 252 cases of acute myeloid leukaemias, 100 of acute lymphocitic leukemia, 111 with refractory anemia, 156 with chronic myeloid leukemia, and 1,161 outpatient hematological controls found some nonsignificant excess risks of acute and chronic myeloid leukaemias in women but not in men, and no consistent pattern for acute lymphocitic leukemia or refractory anemia with dark hair dyes.17 Another Italian case-control study that provided information on personal and occupational exposure to hair dyes found no consistent association of personal use of hair dyes with any lymphoid neoplasm considered: the OR was 0.7 for HD, 1.0 for NHL, 0.9 for leukaemias and 0.8 for MM.18 In a case-control study of NHL in women, conducted in Connecticut and based on 601 cases and 717 controls, the OR was 1.1 for ever use of hair dyes and 1.2 for ever use of permanent hair dyes.19
There is no epidemiological information on the potential relation between hair dye use and the risk of soft tissue sarcomas (STS), and main reviews on hair dyes and cancer have not considered STS.
Given the widespread use of hair dyes, mainly among women, we analyzed data from a case-control study of HD, NHL, MM and STS in Italy.
Subjects and methods
Data were derived from a case-control study of lymphoid neoplasms and STS, conducted between 1985 and 1997 in the greater Milan area and the province of Pordenone, northern Italy.20, 21
The updated dataset included 158 cases of HD (ICD-9 201), including 91 men and 67 women, whose median age was 41 years (range 14–77); 446 of NHL (ICD-9 200 and 202), including 256 men and 190 women, whose median age was 58 years (range 17–79); 141 cases of MM (ICD-9 203), including 70 men and 71 women, whose median age was 65 years (range 38–79) and 221 cases of STS (ICD-9 171), including 117 men and 104 women, whose median age was 53 years (range 16–79). All cases were incident (interviewed during the admission to hospital for diagnosis establishment) and histologically confirmed. Controls were 1,295 subjects (791 men and 504 women, median age 56 years, range 17–79), admitted to hospital for a wide range of acute, nonneoplastic diseases (15% for nonalcohol related traumas, 27% for other orthopedic disorders, 29% for surgical conditions and 28% for other miscellaneous illnesses, such as ear, nose and throat, eye, dental or skin disorders). All admission diagnoses known to be related to long-term modifications in diet, cigarette smoking or alcohol drinking were excluded from the control group.
Trained interviewers used a structured questionnaire to collect data on sociodemographic characteristics, anthropometric measures, a problem-oriented medical history, selected occupational exposures, smoking status, alcohol and coffee drinking and intake of 14 indicator foods. Interviews were conducted during hospital stay for both cases and controls, and less than 3% of the eligible cases and controls refused the interview. To obtain information on hair dye use, subjects were asked whether they had ever used permanent and/or semipermanent hair dyes.
Odds ratios (OR), and the corresponding 95% confidence intervals (CI), for ever use of each type of hair dyes compared to never use of any type were derived by multiple logistic regression equations, including terms for quinquennium of age, sex, area of residence, education and smoking.22 For the analysis of MM, controls aged less than 38 years were excluded, since there were no cases below this age.
The distribution of cases of HD, NHL, MM and STS and 1,295 controls according to ever use of different types of hair dyes in men and women is shown in Table I. Among control women, 29.6% reported use of permanent hair dyes, 23.4% of semipermanent hair dyes only and 46.2% of both. Compared to women, very few men (about 1%) reported ever use of any type of hair dyes.
Table I. Distribution of 158 Cases of Hodgkin's Disease, 446 Cases of Non-Hodgkin's Lymphoma, 141 Cases of Multiple Myeloma, 221 Cases of Soft Tissue Sarcomas and 1,295 Controls According to Use of Different Types of Hair Dyes in Men and Women
|Hodgkin's disease|| || || || || || |
|Non-Hodgkin's lymphoma|| || || || || || |
|Multiple myeloma|| || || || || || |
|Soft tissue sarcoma|| || || || || || |
|Controls|| || || || || || |
Table II reports the OR according to ever use of different types of hair dyes. All the risk estimates for hair dye use were close to unity. Compared to never use of hair dyes, the OR for ever use of permanent dyes was 1.14 for HD, 1.25 for NHL, 1.28 for MM and 1.23 for STS; the OR for ever use of semipermanent dyes was 0.58 for HD, 1.14 for NHL, 1.78 for MM and 0.81 for STS, and the OR for any use of hair dyes was 0.68 for HD, 1.03 for NHL, 1.17 for MM and 0.73 for STS. None of these estimates was significant, except that for ever use of semipermanent dyes and MM. The OR were also close to or, if anything, below unity when the analysis was restricted to women, the OR for any type of hair dye use being 0.56 for HD, 0.96 for NHL, 0.95 for MM and 0.61 for STS.
Table II. Odds Ratios and 95% Confidence Intervals of 158 Cases of Hodgkin's Disease, 446 Cases of Non-Hodgkin's Lymphoma, 141 Cases of Multiple Myeloma, 221 Cases of Soft Tissue Sarcomas and 1,295 Controls According to Ever Use of Different Types of Hair Dyes
|Hodgkin's disease|| || || |
| All subjects||1.14 (0.63–2.08)||0.58 (0.30–1.13)||0.68 (0.40–1.18)|
| Women||1.01 (0.52–1.98)||0.40 (0.18–0.88)||0.56 (0.30–1.02)|
|Non-Hodgkin's lymphoma|| || || |
| All subjects||1.25 (0.87–1.79)||1.14 (0.77–1.67)||1.03 (0.73–1.44)|
| Women||1.22 (0.83–1.78)||1.09 (0.72–1.64)||0.96 (0.67–1.37)|
|Multiple myeloma|| || || |
| All subjects||1.28 (0.75–2.19)||1.78 (1.02–3.12)||1.17 (0.70–1.97)|
| Women||1.14 (0.65–1.99)||1.49 (0.81–2.74)||0.95 (0.55–1.65)|
|Soft tissue sarcoma|| || || |
| All subjects||1.23 (0.75–2.00)||0.81 (0.46–1.44)||0.73 (0.45–1.17)|
| Women||1.05 (0.63–1.75)||0.72 (0.38–1.36)||0.61 (0.37–1.01)|
Table III considers ever use of any hair dyes in strata of age at diagnosis (median for each neoplasm) in women only, given the low number of users among men. No heterogeneity was found for any of the neoplasm considered and all the CI included unity. For elderly women, the OR was 0.51 for HD, 1.00 for NHL, 1.09 for MM and 0.79 for STS.
Table III. Distribution1 of 67 Cases of Hodgkin's Disease, 190 Cases of Non-Hodgkin's Lymphoma, 71 Cases of Multiple Myeloma, 104 Cases of Soft Tissue Sarcomas and 504 Controls in Women, and Corresponding Odds Ratios (or)2 and 95% Confidence Intervals (CI), According to Ever Use of Any Type of Hair Dyes at Ages Stratified by Median Years at Diagnosis for Each Neoplasm
|Hodgkin's disease|| || || |
| < 41 years||24:54||13:51||0.61 (0.24–1.56)|
| ≥ 41 years||17:192||10:182||0.51 (0.22–1.20)|
|Non-Hodgkin's lymphoma|| || || |
| < 58 years||42:119||45:122||0.91 (0.53–1.55)|
| ≥ 58 years||50:127||44:111||1.00 (0.60–1.64)|
|Multiple myeloma|| || || |
| < 65 years||14:115||14:125||0.84 (0.36–1.95)|
| ≥ 65 years||21:86||19:75||1.09 (0.53–2.24)|
|Soft tissue sarcoma|| || || |
| < 53 years||24:56||16:93||0.65 (0.32–1.35)|
| ≥ 53 years||32:150||21:140||0.79 (0.34–1.80)|
Our study confirms that there is no appreciable association between ever use of hair dyes and the risk of HD, NHL, MM and STS. The present data are consistent with most evidence on the absence of any relevant or consistent relation between hair dyes and lymphoid neoplasms,2 and particularly with the results of the Nurses' Health Study9 and the ACS11 prospective studies and of the most recent case-control studies.17, 18, 19 No information is available for the relation of personal use of hair dyes and the risk of STS; thus the present study provides the first evidence on this issue.
A major limitation of the present study is the restricted information on hair dye use and the absence of its validation, which however is a limitation common to most studies of hair dyes and cancer risk.12, 13, 14, 15, 16, 17, 18, 19, 23 Moreover, we collected information only on ever use and type of hair dyes but not on duration of use, shade, age at first use or any other time factor of potential interest. Larger studies with more detailed information on exposure timing and type of hair dyes are needed to explore these issues. However, our findings for ever use should be relatively reliable, and in the absence of overall association, it is unlikely that consistent excess risks are found in specific time-related subgroups. A case-control study of NHL from Connecticut found no overall association with hair coloring product use but some excess risk for long-term users of dark permanent color.19 Although we had no information on duration of use and on calendar year of use, elderly women had OR similar to younger ones, suggesting that there was no appreciable excess risk for formulations used in the distant past.
Potential limitations of our study are those of the hospital-based design; however, only patients admitted to hospital for acute conditions were included in the control group and patients with admission diagnosis related to chronic conditions, particularly those related to major changes in lifestyle factors were excluded. Among the strengths of our study there are its consistent results across a wide range of lymphoid neoplasms and STS. Furthermore, interviews were conducted in the same setting for cases and controls, information on several lifestyle habits was satisfactorily reproducible,24 cases and controls were drawn from the same catchment areas and participation was almost complete. With reference to confounding, adjustment for a number of potentially relevant covariates did not appreciably modify any risk estimates. Furthermore, the results were consistent in separate analysis by sex, age and types of hair dyes.
In conclusion these results provide further evidence that personal ever use of hair dyes is not related to the risk of lymphoid neoplasms and STS.
We thank Ms. M.P. Bonifacino for editorial assistance.