Annual Review of Cosmetic Ingredient Safety Assessments: 2007-2010

This Annual Review of Cosmetic Ingredient Safety Assessments updates and affirms the findings of the Cosmetic Ingredient Review (CIR) Expert Panel’s assessment of almost 30 compounds used in cosmetic ingredients. The review also summarizes new findings from epidemiology studies of hair dyes.

The CIR Expert Panel’s re-review process is intended to uncover any new data since the last safety assessment. In some cases, newly available data are largely redundant compared with the data available in the original safety assessment. In other cases, new data present new safety issues. If after considering the newly available information, the CIR Expert Panel decides to not reopen a safety assessment, this finding, along with any background material, is summarized and announced publicly. To assure that the scientific community is aware of any new information and the decision not to reopen, this Annual Review of Cosmetic Ingredient Safety Assessments is prepared.

A list of reference sources is provided after each ingredient re-review summary that updates the available published literature and includes any unpublished data made available since the original safety assessment. The re-review also captures information on the industry’s current practices of ingredient use, updating the data available in the earlier report. Although this material provides the opinion of the CIR Expert Panel regarding the new data described, it does not constitute a full safety review.

The CIR Expert Panel has assessed the safety of over 2100 cosmetic ingredients since its inception in 1976. These safety assessments were published in the Journal of Environmental Pathology and Toxicology in 1980, the Journal of the American College of Toxicology, from 1982 to 1996, and since then in the International Journal of Toxicology.

The ingredients the CIR Expert Panel reconsidered during the 2007-2010 period and did not reopen are:

• Acetamide MEA

• 5-Bromo-5-nitro-1,3-dioxane

• Butyl benzyl phthalate

• t-Butyl hydroquinone (TBHQ)

• Chlorhexidine, chlorhexidine dihydrochloride, chlorhexidine digluconate, chlorhexidine diacetate

• 2-Chloro-p-phenylenediamine and 2-chloro-p-phenylenediamine sulfate

• 2,4-Diaminophenol and 2,4-diaminophenol dihydrochloride

• Diisopropylamine

• Disperse blue 1

• Disperse violet 1

• Ethyl hexanediol

• HC Blue No. 2

• HC Red No. 3

• HC Yellow No. 2

• Hydroxybenzomorpholine

• Isopropyl isostearate

• Lauramine oxide and stearamine oxide

• Methenamine

• 1-Naphthol

• Phenoxyethanol

• Phenyl methyl pyrazolone

• N-Phenyl-p-phenylenediamine

• Polyoxymethylene urea

• Polyquaternium-7

• Quaternium-22

• Shellac

• Sodium and potassium bromate

• Stearpyrium chloride and lapyrium chloride

Among these are several cosmetic ingredients used in hair dye products (2-chloro-p-phenylenediamine and 2-chloro-p-phenylenediamine sulfate, 2,4-diaminophenol and 2,4-diaminophenol dihydrochloride, disperse blue 1, disperse violet 1, HC Blue No. 2, HC Red No. 3, HC Yellow No. 2, 1-naphthol, phenyl methyl pyrazolone, and N-phenyl-p-phenylenediamine).

Hair Dye Epidemiology

As part of its continuing assessment of cosmetic ingredient safety, the CIR Expert Panel reviews all epidemiology studies on hair dyes. While the safety of individual hair dye ingredients are not addressed in epidemiology studies that seek to determine links, if any, between hair dye use and disease, such studies do provide broad information. The CIR Expert Panel has developed an approach in which a given hair dye ingredient is categorized into groupings that are relevant to the epidemiology data; for example, permanent or oxidative versus semipermanent or nonoxidative.

The epidemiology studies reviewed by the Expert Panel are summarized on the CIR Web site (http://www.cir-safety.org/findings.shtml); only a summary of these data are provided in a safety assessment of a specific hair dye ingredient.

Summaries of these studies the Panel reviewed during 2007-2010 are presented by the outcome studied.

Bladder Cancer

Bolt and Golka ¹ reviewed the published literature on bladder cancer risk and personal use of hair dyes (17 publications) or occupation as a hairdresser and/or barber (23 publications) and concluded that, based on these studies, there seems to be no relevant bladder cancer risk from the use of oxidative hair dyes currently available. The authors could not rule out a bladder cancer risk in hairdressers who worked with oxidative hair dyes available decades ago.

Kelsh et al ² conducted meta-analyses of primary epidemiology studies of hair dye use and bladder cancer and performed their own meta-analysis that examined regular use of hair dye, including permanent hair dye, gender, duration of use, lifetime extent of use, and use of dark hair dye. No association was found between any use of hair dye and bladder cancer among women, men, or both sexes. No statistically significant differences were found for permanent hair dye; duration of any use; duration of permanent hair dye use or lifetime application of any hair dye, permanent hair dye, or dark color hair dye.

Lymphoma and Leukemia

Chiu et al ³ evaluated non-Hodgkin lymphoma subtypes defined according to the presence or absence of t(14:18) translocation from samples taken from a case–control study conducted in Nebraska during 1983-1986. Exposures in 65 t(14:18)-positive cases and 107 t(14:18)-negative cases were compared with those among 1432 controls. Among women, hair dye use was not associated with either t(14;18)-positive or t(14:18)-negative subtypes. (There were too few cases for meaningful analysis of hair dye use among men.) The use of permanent hair dye was associated with a 40% higher risk of the t(14:18)-negative subtype (odds ratio [OR] of 1.4). Hair dye use was not associated with follicular lymphoma or diffuse large B-cell lymphoma in either sex.

Morton et al ⁴ conducted a US population-based case–control study of non-Hodgkin lymphoma. In 1321 cases and 1057 controls, hair dye use included when hair dye use first occurred, use of permanent versus semipermanent dyes, frequency of use, color (black, brown, red, blonde) and intensity of dye used (light vs dark), and total lifetime use. Blood samples obtained from 773 cases and 668 controls and buccal cell samples from 399 cases and 314 controls were used to extract DNA for analyzing NAT1 and NAT2 genotypes and NAT2 acetylation phenotype.

The authors reported no evidence of increased non-Hodgkin lymphoma risk among women, who began hair dye use in or after 1980, or in men. For women whose year of first use was prior to 1980, the odds ratio (OR) for any permanent hair dye use was 1.3 (95% confidence interval [CI] 0.9-1.9); for permanent dark hair dye use was 1.3 (95% CI 0.9-2.0); and for intense tone (ie, black, dark brown, and dark blonde) permanent hair dye use was 1.6 (95% CI 0.9-2.7). For women whose year of first use was 1980 or later, the ORs, for each of the 3 groups listed above, were 0.9 (95% CI 0.6-1.4), 1.2 (95% CI 0.7-1.7), and 0.6 (95% CI 0.4-1.1), respectively.

The risk did not increase consistently with frequency of use, duration of use, or total lifetime use. Women with the NAT2 slow acetylator phenotype or who had no copies of the NAT1*10 allele and used intense tone permanent hair dyes before 1980 did not have an increased risk of non-Hodgkin lymphoma (OR 1.5; 95% CI 0.6-3.6 and OR 1.5; 95% CI 0.7-3.3, respectively), but women with the NAT2 rapid/intermediate acetylator phenotype or those carrying 1 or 2 copies of the NAT1*10 allele did have an increased risk (OR 3.3; 95% CI 1.3-8.6 and OR 2.5; 95% CI 0.9-7.6, respectively).

Zhang et al ⁵ concluded that there was an increased risk of non-Hodgkin lymphoma in women who started using hair dyes before 1980, but not in women who started use after 1980. They performed an evaluation of pooled data from 4 previous studies in which information on sex, duration of use, number of applications, dates of use, and type and color of the hair dye used was available and in which non-Hodgkin lymphoma was classified by histologic type. A total of 4461 cases and 5799 controls were included. In the analysis of non-Hodgkin lymphoma subtypes, there was an increased risk of follicular lymphoma and chronic lymphocytic leukemia/small lymphocytic lymphoma mainly in users who began use before 1980, but no increased risk of diffuse large B-cell lymphoma, marginal-zone lymphoma, or T cell lymphoma. For women who began use after 1980, an increased risk of follicular lymphoma only was suggested for users of dark hair dyes. Other comparisons failed to show the differences between risks of dark versus light hair dyes and permanent versus semipermanent hair dyes. In the analysis of subtypes, similar failures to show differences were present in the data.

Koutros et al ⁶ found no association between hair dye use and myeloma risk, semipermanent hair dye use, permanent dye use, or dark permanent hair dye use. The authors also reported no significant association of myeloma with use of hair dyes before age 30 years, use begun before 1980, >240 lifetime uses, or use of dark permanent dyes for ≥28 years. They conducted a population-based case–control study of 175 cases of multiple myeloma with 679 matched controls. Information on type and color of hair-coloring product, age at first use, age when the use was stopped, duration of use, and frequency of use per year were obtained from interviews.

Reproductive and Developmental Outcomes

Axmon and Rylander ⁷ conducted a cohort study of Swedish women, who had attended vocational schools for hairdressers (3137), and their sisters (3952) based on data obtained from Swedish national registries. Low-birth-weight (less than 2500 g), small for gestational age, and large for gestational age data were gathered from Swedish birth registries of infants born to the hairdressers (6223 infants) and their sisters (8388 infants). The authors reported no association between occupation as a hairdresser and increased risk of low-birth-weight or small for gestational age parameters. Among the infants born to the hairdressers’ sisters, the distribution of birth weights were wider than that among the infants born to the hairdressers. For “large for gestational age” data, there was a reduced risk of women who had actually worked as hairdressers during at least 1 pregnancy. The infants born to these women also had a significantly lower mean birth weight (3387 g vs 3419 g; P = .033).

Gallicchio et al ⁸ conducted a study of cosmetologists to determine whether they are at increased risk of poor pregnancy outcomes compared with women of the same age who are not cosmetologists. Participants were self-selected through mass mailing of questionnaires. Respondents to the survey had to be between 21 and 55 years of age and not have had a hysterectomy or oophorectomy. A cohort of 350 cosmetologists and 397 women in other occupations who self-reported that they met the inclusion criteria and had 5 or fewer singleton pregnancies were included. Outcome measures were miscarriage, stillbirth, and occurrence of maternal health conditions during pregnancy (preeclampsia, high blood pressure, and diabetes), hospitalization or physician-ordered bed rest during pregnancy, preterm labor, and premature delivery (before 37 weeks at delivery). The authors reported no statistically significant associations between occupation as a cosmetologist and any pregnancy outcomes after adjustment for age, race, education, and smoking and alcohol use at the time of pregnancy.

Halliday-Bell et al ⁹ identified all singleton births to hairdressers (10    662) and cosmetologists (2490) from the 1990-2004 Finnish Medical Birth Registry. Singleton births to teachers (18    594) were used as the control population. Data were obtained from the confidential registry on the parameters of sex distribution, low-birth-weight, small for gestational age, preterm delivery, and perinatal death. In comparing occupation as a hairdresser to that of a teacher, the authors reported a higher incidence of low-birth-weight, small for gestational age, preterm delivery, and perinatal death, but no difference in sex distribution. In comparing occupation as a cosmetologist to that of a teacher, the authors reported a higher incidence of small for gestational age and perinatal death but no difference in sex distribution, incidence of low birth weight, or preterm delivery. The authors concluded that occupation as a hairdresser or cosmetologist may reduce fetal growth and that occupation as a hairdresser may also increase the incidence of preterm delivery and perinatal death.

Herdt-Losavio et al ¹⁰ conducted a retrospective cohort study of cosmetologists (15 003) compared with realtors (4246) or compared with the general population (12 171) using birth records in New York State from 1997 to 2003. From the birth records, the authors obtained information on low-birth-weight, small for gestational age, and preterm birth. The incidence of low–birth-weight (1500-2499 g) was higher for cosmetologists compared with realtors but not when compared with the general population. For nonwhite cosmetologists, the incidence of low-birth-weight was higher compared with realtors and the general population. No statistically significant differences were reported for any comparison of small for gestational age or preterm births.

Baste et al ¹¹ conducted a cross-sectional study among women in Hordaland county in Norway. Among 10 512 women providing information by self-report, 221 were hairdressers. An association between occupation as a hairdresser and increased risk of infertility (OR 1.3; 95% CI 1.08-1.55) or spontaneous abortion (OR 1.31; 95% CI 1.07-1.60) were found. These associations were found primarily among those individuals who reported that they were never smokers.

Other Outcomes

Mendelsohn et al ¹² examined personal hair dye use and cancer risk in a prospective cohort of 73    366 Chinese women (29 076 hair dye users and 44    290 nonusers). Cancer diagnoses were ascertained through the Shanghai Cancer Registry in 1536 nonusers and 901 hair dye users. Self-administered questionnaires followed by interviews were used to gather data on use of hair dyes over the past 3 years, with follow-up questions about frequency of use over the past year and years of use for those individuals who answered “yes” to the initial question about recent hair dye use; the authors noted that hair dye color was not determined, but that most hair dye use would be expected to be dark hair dyes. The authors reported no significant association with hair dye use and overall cancer or for several common cancers, including breast cancer, lung cancer, stomach cancer, bladder cancer, hematopoietic cancer, including subtypes of non-Hodgkin lymphoma, multiple myeloma, and leukemia. The authors cautioned that the study is limited by small numbers of certain cancer types.

Exposure Assessment

Ambrosone et al ¹³ evaluated DNA adducts in breast ductal epithelial cells isolated from breast milk obtained from 64 women. A questionnaire was used to determine prior hair dye use, meat intake, and tobacco exposure. The following were reported:

• 2-Amino-6-phenylimidazo[4,5-b]pyridine (PhIP)-DNA adducts (30 participants, mean detectable level of 4.7 ± 1.7 adducts/107 nucleotides),

• Benzo[a]pyrene-7,8-diol-9,10-epoxide (BPDE)-adducts (13 participants, mean detectable level of 1.7 ± 0.7 adducts/10 nucleotides), and

• 4-Amino biphenyl (4-ABP)-DNA adducts (18 participants, mean detectable level of 4.7 ± 2.2 adducts/107 nucleotides).

No association was reported between PhIP-DNA adducts or BPDE-DNA adducts and either meat consumption or tobacco exposure. The authors did not present PhIP-DNA or BPDE-DNA adduct data as a function of hair dye exposure. The authors reported that the presence of 4-ABP-DNA adducts were associated with use of hair dyes in the previous year but the association was not statistically significant. Among the hair dye users, the OR for use 6 to 12 months prior to sample collection was 5.42 (95% CI 0.49-59.74) and for 0 to 6 months, the OR was 11.17 (95% CI 1.14-109.19).

For temporary hair dye users, the OR was 9.47 (95% CI 0.46-195.04) and for permanent hair dye users the OR was 8.20 (95% CI 0.94-71.83). Use of light hair dye colors had an OR of 18.12 (95% CI 1.45-226.83), medium colors had an OR of 5.61 (95% CI 0.35-90.35), and dark colors had an OR of 2.57 (95% CI 0.10-64.31).

The authors reported that rapid NAT1 and NAT2 genotypes were associated with an increase in PhIP-DNA adducts and 4-ABP-DNA adducts, compared with slow acetylator genotypes, and that the level of 4-ABP-DNA adducts was higher for rapid NAT2 genotypes compared to rapid NAT1 genotypes.

The authors did not provide data on BPDE-DNA adducts as a function of NAT genotype. The authors also did not present data on 4-APB-DNA adducts as a function of meat consumption or tobacco exposure. The authors acknowledged that the small sample size resulted in wide CIs and limited the statistical power of the results.

Hueber-Becker et al ¹⁴ monitored the exposure of hairdressers to oxidative hair dyes under controlled conditions to determine whether the process of oxidative hair dye use produces measurable external and/or systemic exposure, and, if so, which parts of the process are responsible. [14C]-p-Phenylenediamine dihydrochloride was combined with a commercial product containing p-phenylenediamine dihydrochloride (4%), resorcinol (2%), and m-aminophenol (2%) in the dye component and a cream formulation developer with hydrogen peroxide (6%). In addition to the task of hair dye preparation and dyeing, the study monitored exposure during shampooing, rinsing, and conditioning and during hair cutting and drying. Six artificial human training heads with implanted human hair were used in a given day. One hairdresser performed the hair dye preparation and dyeing, a second hairdresser did the shampooing, and so on, and a third hairdresser did the cutting and drying—on all 6 heads. This process was repeated 6 times, involving a total of 18 hairdressers and 36 artificial human heads. Other than the mixing bowl, hair wash, and hair, recoveries were all less than 0.5% and most <0.1%.

Urinary excretion of radioactivity in the 18 hairdressers determined over 3 intervals (0-12 hours, 12-24 hours, and 24-48 hours) resulted in excretion values (all values below the limit of detection were given the value of the limit of detection) of <8.8 ± 4.4; <6.8 ± 4.6, and <9.7 ± 6.0 mg [14C] p-phenylenediamine dihydrochloride equivalent. Excretion values per working day (including 6 complete procedures) was <25.3 ± 5.2 mg [14C] p-phenylenediamine dihydrochloride equivalent. For a 71-kg individual, the exposure was calculated to be <0.36 mg/kg per day. The authors expressed uncertainty as to the source of this low level of measured systemic exposure. The authors noted that detectable air levels of radioactivity were found during the preparation and dyeing phase and the hair-cutting phase but not in the washing phase. In those individuals exposed to detectable air levels, urinary excretion was not elevated compared to individuals exposed to air samples with no detectable radioactivity.

International Agency for Research on Cancer

In 2008, International Agency for Research on Cancer (IARC) issued new working group observations on bladder cancer and hematological cancers. ¹⁵ The working group considered 82 studies involving personal use of hair dyes and 85 studies of occupational exposures. They concluded that the data are insufficient (quality, consistency, or statistical power) to conclude the presence or absence of a causal link between personal use of hair dyes and cancer. They also reviewed animal studies and cited major limitations (qualitative and quantitative) that preclude concluding that hair dyes have carcinogenic effects in animals. Occupational exposure as a hairdresser, barber, or beautician was also assessed. The working group noted that such occupations may involve exposure to hair dyes, but that any risk cannot be attributed to hair dye exposure, but is linked to all exposures that individuals may experience in these occupations. With that caveat, the working group did find that exposures experienced in these occupations are probably carcinogenic.

Discussion

In considering these data on many different end points, the CIR Expert Panel concluded that the available epidemiology studies are insufficient to conclude there is a causal relationship between hair dye use and cancer and other end points based on lack of strength of the associations and inconsistency of findings. The Panel stated that use of direct hair dyes, while not the focus in all investigations, appears to have little evidence of an association with adverse events as reported in epidemiology studies. However, direct hair dyes are a diverse group of chemicals and the determination of safety may hinge on other safety test data.