Does natural hair pigmentation before graying modulate alopecia areata risk?
In this large matched case-control study of 8365 adults from the UK, black hair and dark brown hair had significantly increased risk of alopecia areata compared with light brown hair. Blond individuals were significantly less likely to receive a diagnosis of alopecia areata compared with those with light brown hair.
Hair pigment may be associated with a modulation of alopecia areata risk.
Alopecia areata (AA) is a complex immune-mediated disorder that causes nonscarring hair loss. Previous reports have documented preferential targeting of pigmented hair follicles with sparing of gray, nonpigmented hair follicles in alopecia lesions. Thus, immune targeting of melanogenesis-associated proteins in melanocytes and keratinocytes represents a potential mechanism for the inflammation that targets anagen hairs in alopecia areata.
To investigate the association of alopecia areata with hair color among White residents of the UK.
Design, Setting, and Participants
This matched, case-control study conducted in October 2020 used a large prospectively acquired cohort and included data that were collected from the UK Biobank, a large-scale prospective resource designed to study phenotypic and genotypic determinants in adults. A total of 502 510 UK Biobank participants were reviewed for inclusion. Among these individuals, 1673 cases of alopecia areata with reported hair color were captured and matched by age and sex to 6692 controls without alopecia areata using 1:4 matching.
Main Outcomes and Measures
Conditional logistic regression analysis was performed, in which the outcome variable was alopecia areata and the main predictor was natural hair color before graying. The variables considered included diabetes, hypothyroidism, hyperthyroidism, and vitiligo.
Of 464 353 participants, 254 505 (54.8%) were women, and the mean (SD) age for those with alopecia areata was 46.9 (16.5) years. Alopecia areata was significantly more common in individuals with black (adjusted odds ratio [aOR], 2.97; 95% CI, 2.38-3.71) and dark brown hair (aOR, 1.26; 95% CI, 1.11-1.42) compared with light brown hair. In contrast, blond individuals exhibited significantly decreased alopecia areata compared with those with light brown hair (aOR, 0.69; 95% CI, 0.56-0.85). Red hair color was not significantly different from light brown hair.
Conclusions and Relevance
The findings of this matched case-control study seem to indicate that alopecia areata is modulated by natural hair color, preferentially targeting darker hair. Our results support a previously proposed model of alopecia areata in which immunity is directed against melanogenesis-associated proteins in the anagen hair follicles. However, further study is needed to more precisely understand the immunopathogenic association between alopecia areata and hair color.
Alopecia areata (AA) is a common cause of nonscarring patchy hair loss. The involvement of melanogenesis in AA is suggested by the observation that hair follicles are only attacked during the melanogenically active stages of the hair cycle (anagen III-VI).1-4 Alopecia areata has been observed to selectively involve pigmented hairs while sparing adjacent white hairs. Following AA, regrowth of white hair is often seen in areas that were previously filled with darker hair.5,6 Alopecia areata is common in children and young adults and is seen less often in older age groups with graying hair. Thus, immune response in AA may target hair pigmentation, or the pigmentation pathway may play a mechanistic role in developing AA.
While treating patients with AA, we have noticed a relative paucity of blond-haired individuals with this disease. This observation prompted us to ask how varying darkness of natural hair color before graying is associated with AA. To address this question, we used the UK Biobank, a prospective cohort of about 500 000 individuals that includes deep medical, phenotypic, and genetic data. We compared the relative lifetime risk of AA in patients with blond, red, light brown, dark brown, and black hair.
The UK Biobank is a large-scale prospective resource with 502 510 participants.7 Baseline participant characteristics were recorded using touchscreen questionnaires and physical measurements.8 We analyzed the complete data set using the latest version as of October 2020. Centrally, the UK Biobank received ethical approvals from the North West Multicenter Research Ethics Committee, which covers the UK, the Community Health Index Advisory Group, covering Scotland, the Patient Information Advisory Group for gaining access to invite people to participate, and the National Research Ethics Service. All participants provided written informed consent. Locally, institutional review board approval was obtained through West Virginia University.
Case and Control Criteria
Cases were defined as self-reported White UK Biobank participants with physician-diagnosed AA and available self-reported hair color. Diagnoses of AA were verified by reviewing participant primary care and hospital records. Data from UK primary care electronic health record systems have been extensively used for research and previously shown to contain accurate diagnostic information.9,10 Furthermore, International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10) coding for alopecia areata, as used in this study, has been validated.11 Participants self-reported natural hair color at the baseline visit between 2006 and 2010. Controls were defined as White participants without physician-diagnosed AA and available self-reported hair color. Each case was subsequently matched to 4 controls by age and sex.
The main predictor analyzed was natural hair color. Participants were asked “What best describes your natural hair color? If your hair color is gray, what was the color before you went gray?” Participant responses included blond, red, light brown, dark brown, and black hair color. Matching variables included age and sex. Age was defined as the difference between the date of birth and date of AA diagnosis.
Covariates were selected based on previously known risk factors that are associated with AA that may confound the outcome variable, including diabetes, hypothyroidism, hyperthyroidism, and vitiligo. Diabetes was defined as answering yes to the question “Has a doctor ever told you that you have diabetes?” or as regularly taking insulin. Hypothyroidism was defined as an ICD-10 diagnosis of E03.9, self-reported during verbal interview, or an existing primary care record with all cases confirmed by an existing date of diagnosis. Hyperthyroidism was defined as an ICD-10 diagnosis of E05.0-9, self-reported during verbal interview, or an existing primary care record with all cases confirmed by an existing date of diagnosis. Vitiligo was defined as an ICD-10 diagnosis of L80 or an existing primary care record with all cases confirmed by an existing date of diagnosis. All codes used for case-control and covariate selection are listed in the eTable in the Supplement.
For statistical analysis, SPSS (version 24.0; IBM) was used for data entry and management, and SAS, version 9.4 (SAS Institute) was used for statistical testing. Descriptive statistics were presented as frequencies with percentages for categorical variables and means (SDs) for continuous variables. Statistical significance was set at P < .05.
To determine significant differences between cases and control participants, χ2 tests were used. In addressing the association of natural hair color with AA, adjusted odds ratios (aORs) and associated 95% CIs were estimated by conditional logistic regression analysis, in which the outcome variable was AA, the main predictor natural hair color, and variables accounted for diabetes, hypothyroidism, hyperthyroidism, and vitiligo. Each case was matched to 4 controls by age and sex. A sensitivity analysis was performed on the full cohort to confirm the association of hair color and AA using unconditional multivariate logistic regression analysis without age and sex matching and instead using both variables as additional covariates.
The full cohort included 464 353 participants after participant identification, screening, and eligibility assessment, of whom 8365 (1.8%) were included in the main analysis (Figure 1). A total of 1673 AA cases were matched to 6692 non-AA controls, matching each case to 4 controls by age and sex. The AA cases were predominately in women (1077 [64.4%]), and the mean (SD) age of diagnosis was 46.9 (16.5) years (Table 1). The AA cases were more likely than controls to receive a diagnosis of hypothyroidism (199 [11.9%] vs 535 [8.0%]; P < .001), but not hyperthyroidism (1.3% each; P = .96). Vitiligo was more frequently diagnosed in AA cases compared with matched controls (23 [1.4%] vs 19 [0.3%]; P < .001).
Among controls, the most frequent natural hair color reported was light brown (2722 [40.7%]), followed by dark brown (2575 [38.5%]), blond (842 [12.6%]), red (297 [4.4%]), and black (256 [3.8%]). Therefore, light brown was selected as the reference natural hair color in the main analysis.
Conditional logistic regression analysis was performed to test the association between AA and natural hair color. After accounting for diabetes, hypothyroidism, hyperthyroidism, and vitiligo, dark brown and black hair phenotypes were significantly more likely than light brown hair to be diagnosed with AA (dark brown hair: aOR, 1.26; 95% CI, 1.11-1.42; P < .001; black hair: aOR, 2.97; 95% CI, 2.38-3.71; P < .001) (Table 2). Those with blond hair were significantly less likely than those with light brown hair to receive a diagnosis of AA (aOR, 0.69; 95% CI, 0.56-0.85; P < .001). In fact, black hair increased the risk of developing AA 4-fold compared with blond hair. Individuals with red hair were not significantly different than those with light brown hair regarding AA diagnosis (aOR, 0.94; 95% CI, 0.70-1.26; P = .68). The effect sizes of hair color on AA are shown in Figure 2.
A sensitivity analysis was performed on the full data set that met inclusion criteria to confirm the observed association of hair color and AA. A second multivariate logistic regression analysis was conducted without age and sex matching, instead using both variables as covariates. After accounting for age, sex, diabetes, hypothyroidism, hyperthyroidism, and vitiligo, similar trends were observed for each hair color (Table 3).
Hair color is a highly visible trait in the European population, and twin studies suggest that as much as 97% of this variation in hair color is inherited.12 This study’s data indicate that, in White individuals, risk for AA is significantly associated with hair pigmentation, with incremental darkening of hair posing increased risk. The largely hereditary determination of hair color (before graying) indicates that the association of AA with hair color likely represents a common genetic pathway driving AA and hair color or, alternatively, an immune response that is directed at melanin or melanogenesis-associated proteins.
All hair types contain a mixture of eumelanins and pheomelanin. Color is determined by the mix of and amount of melanins in the hair shaft.13 Black or brown eumelanins cause black or brown hair, whereas pheomelanin imparts a red color. Blond hair results from small amounts of brown eumelanin.14 We have shown that AA risk is highest with black hair, followed by dark brown, light brown, red, and blond, respectively. Although the mechanism is unknown, our findings show that the greater content of eumelanin appears to be associated with greater risk for AA. Melanins are best known for the ability to protect higher organisms from ultraviolet radiation.15,16 However, melanins also function as antioxidants and immunomodulatory molecules.17,18 Moreover, melanocytes themselves have immune functions.19 Thus, various potential mechanisms could provide an association between melanogenesis and inflammatory disease of the hair follicle.20
Genome-wide association studies of AA have identified numerous loci that are associated with immunity, as well as a few genes expressed in the hair follicle.21,22 A variant near STX17, a gene that is expressed in hair follicles and involved in autophagy, increases risk for AA.23 This finding provides a possible mechanistic association between hair color and AA, as STX17 is well established to cause white hair in horses.24 Moreover, human melanosome biosynthesis is mechanistically regulated by autophagocytosis.25 Thus, autophagy could represent a common pathway that controls hair pigmentation and causes AA.
The anagen hair bulb is a privileged immunological site, and AA is associated with a collapse of immune privilege.26 Alopecia areata is associated with melanocyte degeneration, leading to melanin deposition into the follicular epithelium, dermal papilla, and follicular stella, potentially exposing melanosomes to the immune system.27,28 Interestingly, antibodies directed against melanocytes have been observed in AA.29,30
A limitation of our study is that natural hair color was self-reported rather than measured by trained observers or using optical instrumentation. Additionally, this study was unable to account for heterochromia. However, hair color varies over a person's lifetime, eventually turning gray. Hair color measurement at any point may not provide an accurate assessment of lifetime natural hair color, and self-reported natural hair color has been used for most studies of hair color genetics.31-33 The diagnosis of AA was determined from codes from the inpatient and outpatient records rather than a direct review of patient health records by the authors. However, AA is common and typically straightforward to diagnose clinically and histologically. The female predominance of AA and the associations with vitiligo and autoimmune thyroid disease in this data set are consistent with previous studies.34-37
Darker hair indicates greater risk for AA in White residents of the UK. Our findings support and expand on the previously hypothesized association between AA and hair pigmentation. The mechanism through which darker pigmentation is associated with AA remains unknown. However, our findings suggest that investigations into the commonalities between mechanisms of hair pigmentation and autoimmunity or immunity that targets melanogenesis-associated proteins may illuminate this very common but poorly understood disease.
Accepted for Publication: January 22, 2021.
Published Online: March 10, 2021. doi:10.1001/jamadermatol.2021.0144
Retraction: This article was retracted on June 2, 2021.
Corresponding Author: Michael S. Kolodney, MD, PhD, West Virginia University, One Medical Center Dr, PO Box 9158, Morgantown, WV 26506 (firstname.lastname@example.org).
Author Contributions: Drs Yousaf and Kolodney had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Yousaf, Fang, Kolodney.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Yousaf, Lee, Kolodney.
Critical revision of the manuscript for important intellectual content: Fang, Kolodney.
Statistical analysis: Yousaf, Lee, Fang.
Obtained funding: Kolodney.
Administrative, technical, or material support: Kolodney.
Conflicts of Interest Disclosures: None reported.
Funding/Support: This work was supported through the William Welton Endowment Fund.
Role of the Funder/Sponsor: The William Welton Endowment Fund had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Additional Contributions: The study was conducted using the UK Biobank resource (approved UK Biobank application: 66911). We thank the UK Biobank participants for their involvement, as well as Stephen Davis, PhD, West Virginia University, and Spiros Denaxas, PhD, University College London, for advice. No individuals received compensation for their contributions.
UK Biobank. UK Biobank: Rationale, Design and Development of a Large-Scale Prospective Resource. UK Biobank Coordinating Centre, 2007.
et al. UK biobank: an open access resource for identifying the causes of a wide range of complex diseases of middle and old age. PLoS Med
. 2015;12(3):e1001779. doi:10.1371/journal.pmed.1001779
X-H. Melanocytes and Skin Immunity. Elsevier; 2015:37-39.
et al. Mechanisms underlying the dysfunction of melanocytes in vitiligo epidermis: role of SCF/KIT protein interactions and the downstream effector, MITF-M. J Pathol
. 2004;202(4):463-475. doi:10.1002/path.1538PubMedGoogle ScholarCrossref
et al. Genome wide analysis of copy number variants in alopecia areata in a Central European cohort reveals association with MCHR2. Exp Dermatol
. 2017;26(6):536-541. doi:10.1111/exd.13123
et al. A cis-acting regulatory mutation causes premature hair graying and susceptibility to melanoma in the horse. Nat Genet
. 2008;40(8):1004-1009. doi:10.1038/ng.185
T. The Hair Follicle and Immune Privilege. Elsevier; 2003:188-194.
et al; International Visible Trait Genetics Consortium. Genome-wide association meta-analysis of individuals of European ancestry identifies new loci explaining a substantial fraction of hair color variation and heritability. Nat Genet
. 2018;50(5):652-656. doi:10.1038/s41588-018-0100-5
FG, Erayman Demirbaş
A. The frequency of autoimmune thyroid disease in alopecia areata and vitiligo patients. Biomed Res Int
. 2015;2015:435947. doi:10.1155/2015/435947PubMedGoogle Scholar