Association Between Natural Hair Color, Race, and Alopecia

Abstract

Introduction

Limited epidemiologic data has suggested direct associations between hair pigment, race, and incidence of alopecia areata (AA). Here, we examine the relationship between natural hair color, race, and the lifetime risk alopecia.

Methods

In this case–control study, we included UK Biobank patients of all races and self-reported hair color with diagnoses of AA, androgenetic alopecia (AGA), or scarring alopecia (SA). Multivariable logistic regression was used to detect differences in lifetime risk.

Results

Findings reveal a significantly increased risk of AA among individuals with black hair compared to dark brown hair (OR 1.71 [95% CI 1.22–2.38], p < 0.001). Those with red or blonde hair showed a decreased risk of AA (0.74 [0.56–0.97]; 0.62 [0.41–0.95], p < 0.05). No racial differences in AA prevalence were observed among individuals with black hair.

Conclusions

Darker hair colors may be associated with a higher risk of AA, lighter hair colors with a lower risk, and differences in hair color could contribute to previously noted racial variations in AA incidence, potentially influencing dermatologists’ perspectives on the disease’s epidemiology.

Supplementary Information

The online version contains supplementary material available at 10.1007/s13555-024-01218-9.

Key Summary Points

This study examined the relationship between natural hair color, race, and lifetime risk of alopecia.

Our findings show a significantly increased risk of alopecia areata (AA) among individuals with black hair compared to dark brown hair. No racial differences in AA prevalence were observed among individuals with black hair.

Darker hair colors may be associated with a higher risk of AA, while lighter hair colors may be associated with a lower risk. These differences in hair color could contribute to previously noted racial variations in AA incidence, potentially influencing dermatologists’ perspectives on the disease’s epidemiology.

Introduction

Hair color is a highly visible inherited trait determined by the type and amount of melanin within the hair shaft [1]. For example, in black or brown hair, dark eumelanin predominates, while in red or blonde hair, lighter pheomelanin predominates. In the field of dermatology, there has been increasing interest in how pigmentation may affect the natural course and risk of alopecia and other skin disorders [2].

Three common forms of alopecia are alopecia areata (AA), androgenetic alopecia (AGA), and scarring alopecia (SA). AA is an autoimmune disease most often affecting children and young adults in which CD8⁺ T cells attack hair follicles, leading to well-demarcated, nonscarring patches of hair loss [3]. AGA is a hereditary form of gradual hair loss driven by a heritable sensitivity of hair follicles to androgens, leading to gradual shaft miniaturization and eventual loss with age [4]. Furthermore, AGA is very common in women as well as men. Finally, SA, a group which includes frontal fibrosing alopecia, central centrifugal cicatricial alopecia, and lichen planopilaris, are auto-inflammatory disorders of varying etiologies that lead to replacement of hair follicles with scar tissue and consequent permanent hair loss [5].

While laboratory research has begun to investigate interactions between natural hair pigmentation and both inflammatory and non-inflammatory alopecia, there are limited large epidemiological studies examining this relationship. Furthermore, little is known about whether differences in natural hair color between races may be driving the higher incidence of AA observed among certain racial groups, including Asians and Blacks, and whether hair color can be understood as a risk factor for developing certain types of alopecia [6–8]. Prior research has also suggested that melanosomal proteins may serve antigenic targets, thus leading to preferential loss of pigmented hair in AA, which warrants further research into the association of hair color on alopecia [9].

In this study, we utilize the United Kingdom Biobank (UKBB), a prospective cohort of over 500,000 healthcare recipients [10], to compare the relative lifetime risk of alopecia among patients with different natural hair colors. Additionally, we explore and validate how race alone and in combination with natural hair color may influence this risk [6].

Methods

Participants and Study Design

The UKBB is a large-scale research and biomedical database that collects medical, genomic, and phenotypic health data, including electronic health records (EHR), biologic data, and lifestyle reports [10]. The aim of the UKBB is to improve disease diagnosis, prevention, and treatment. We performed a matched case–control study analyzing data using the latest version of the UKBB (March 2023). This study was deemed exempt from the MassGeneral Brigham Institutional Review Board as it is non-human subjects research.

Cases included adult (18+) participants of all self-reported races and both a documented self-reported hair color and a physician-verified diagnosis of AA, AGA, or SA. If a patient had gray hair, natural hair color was defined as their hair color before it went gray. Natural hair color options included black, dark brown, light brown, red, and blonde. Alopecia cases were identified through International Classification of Diseases (ICD-9 and ICD-10) coding [11] as well through review of primary care and hospital records using previously validated methods [12]. Controls were patients with self-reported race and hair color and no diagnosis of AA, AGA, or SA. In regression analyses, brown hair color was the control group as it was felt to be the most intermediate hair color in terms of melanin content.

We adjusted for comorbid conditions that have been associated with risk of alopecia in the literature. These included vitamin D deficiency, psoriasis, type 1 and 2 diabetes mellitus, hypo- and hyperthyroidism, ulcerative colitis, vitiligo, and systemic lupus erythematosus (SLE). Each condition was identified using validated ICD-9 and 10 coding, confirmation through primary care and hospital records, or through self-report during recorded interviews detailed in the UKBB (Supplemental Material) [11, 13–18].

Statistical Analysis

Descriptive statistics were calculated as means and standard deviations for continuous variables and frequencies with percentages for categorical variables. We performed multiple multivariable logistic regressions, controlling for the covariates mentioned above, to study the association between natural hair color, race, and AA, AGA, and SA. For the first regression, the outcome variable was the type of alopecia and the main predictor was natural hair color, with dark brown as the reference color. Each case was matched to five controls by age and sex via the nearest neighbor method. Odds ratios were presented with associated 95% confidence intervals (CI). We also performed a sensitivity analysis to confirm the association between hair color and alopecia with unconditional multivariate logistic regression without age/sex matching, instead using those variables as covariates.

Next, we performed a matched multivariate logistic regression to assess the association between alopecia and race among individuals with all hair colors. Finally, we performed matched multivariable logistic regression to assess the association between AA, race, and black natural hair color only, given the predominance of black hair in non-White racial groups [19]. Statistical significance was set at P < 0.05. All statistical analysis was conducted with R version 4.0.3.

Results

Baseline Characteristics

Out of the 472,574 White participants in the UKBB dataset, we identified 918 with AA, 161 with AGA, and 236 with SA. AA, AGA, and SA cases were predominantly female (64.7%, 77.0%, 76.7%) with a mean (SD) age of 55.6 (8.0), 55.5 (8.3), and 57.5 (7.1) years, respectively (Table 1).

Table 1.

Demographic characteristics of patients with alopecia areata, androgenetic alopecia, and scarring alopecia and their controls in the United Kingdom Biobank

	Alopecia areata		Androgenetic alopecia		Scarring alopecia
	Control (N = 4590)	Case (N = 918)	Control (N = 805)	Case (N = 161)	Control (N = 1180)	Case (N = 236)
Sex
Female	2970 (64.7%)	594 (64.7%)	620 (77.0%)	124 (77.0%)	905 (76.7%)	181 (76.7%)
Male	1620 (35.3%)	324 (35.3%)	185 (23.0%)	37 (23.0%)	275 (23.3%)	55 (23.3%)
Age (years)
Mean (SD)	55.6 (8.04)	55.6 (8.05)	55.5 (8.28)	55.5 (8.30)	57.5 (7.11)	57.5 (7.12)
Median [Min, Max]	56.0 [40.0, 70.0]	56.0 [40.0, 70.0]	56.0 [40.0, 70.0]	56.0 [40.0, 70.0]	58.0 [41.0, 70.0]	58.0 [41.0, 70.0]
Race
White	4317 (94.1%)	804 (87.6%)	747 (92.8%)	130 (80.7%)	1117 (94.7%)	205 (86.9%)
Asian	102 (2.2%)	67 (7.3%)	19 (2.4%)	16 (9.9%)	24 (2.0%)	9 (3.8%)
Black	102 (2.2%)	23 (2.5%)	17 (2.1%)	6 (3.7%)	17 (1.4%)	17 (7.2%)
Multiracial	28 (0.6%)	13 (1.4%)	9 (1.1%)	4 (2.5%)	8 (0.7%)	1 (0.4%)
Other	41 (0.9%)	11 (1.2%)	13 (1.6%)	5 (3.1%)	14 (1.2%)	4 (1.7%)
Natural hair color
Dark brown	1725 (37.6%)	370 (40.3%)	285 (35.4%)	57 (35.4%)	407 (34.5%)	82 (34.7%)
Black	345 (7.5%)	127 (13.8%)	69 (8.6%)	22 (13.7%)	79 (6.7%)	28 (11.9%)
Blonde	484 (10.5%)	71 (7.7%)	85 (10.6%)	10 (6.2%)	130 (11.0%)	25 (10.6%)
Light brown	1822 (39.7%)	323 (35.2%)	323 (40.1%)	65 (40.4%)	504 (42.7%)	86 (36.4%)
Red	214 (4.7%)	27 (2.9%)	43 (5.3%)	7 (4.3%)	60 (5.1%)	15 (6.4%)

Hair Color Analysis in White Patients

Among White patients with all three types of alopecia, the most prevalent natural hair colors in descending order were light brown, dark brown, blonde, red, and black. In a matched multivariable regression adjusting for age, sex, and related comorbidities, patients with black natural hair color had a higher odds ratio for AA, but not AGA or SA, compared to the reference population of patients with dark brown hair (1.71 (1.22–2.38, p < 0.001) (Table 2). Patients with blonde and red hair both had significantly lower odds for AA but no significantly different odds for AGA or SA compared to dark brown-haired controls (0.74 (0.56–0.97) and 0.62 (0.41–0.95), respectively) (p = 0.03).

Table 2.

Matched multivariable regression of prevalence of alopecia areata, androgenetic alopecia, and scarring alopecia by natural hair color in White patients

Natural hair color	Alopecia areata		Androgenetic alopecia		Scarring alopecia
	Odds ratio (95% CI)	P value	Odds ratio (95% CI)	P value	Odds ratio (95% CI)	P value
Dark brown	1 (Reference)		1 (Reference)		1 (Reference)
Black	1.71 (1.22–2.38)	< 0.001	0.83 (0.25–2.79)	0.76	1.40 (0.64–3.06)	0.40
Blonde	0.74 (0.56–0.97)	0.03	0.74 (0.34–1.60)	0.44	1.10 (0.66–1.82)	0.72
Light brown	0.89 (0.75–1.05)	0.18	1.33 (0.87–2.03)	0.18	1.01 (0.71–1.43)	0.96
Red	0.62 (0.41–0.95)	0.03	1.26 (0.53–3.01)	0.60	1.34 (0.70–2.56)	0.37

Matched multivariable regression was performed using nearest-neighbor propensity score 1-to-5 matching and adjusted for age, sex, diagnosis of ulcerative colitis, type 1 diabetes mellitus, type 2 diabetes mellitus, vitamin D deficiency, psoriasis, systemic lupus erythematosus, hypothyroidism, hyperthyroidism, vitiligo

Racial Differences

Compared to White individuals of all hair natural hair colors, there was a significantly higher lifetime prevalence of AA in Asian (aOR 3.58 [2.58–4.96], p < 0.001), Black (aOR 1.25 [0.79–1.98, p = 0.03], and multiracial patients (aOR 2.58 [1.34–4.98], p = 0.005). Compared to White patients with AGA, there was a significantly higher odds of disease in Asian patients (aOR 4.81 [2.34–9.90], p < 0.001), but no differences between other racial groups. Finally, compared to White patients with SA, there was a significantly higher odds of disease in Black patients (aOR 5.75 [2.82–11.73], p < 0.001), but no other significant differences between racial groups (Table 3).

Table 3.

Matched multivariable regression of prevalence of alopecia areata, androgenetic alopecia, and scarring alopecia by race

Race	Alopecia areata		Androgenetic alopecia		Scarring alopecia
	Odds ratio (95% CI)	P value	Odds ratio (95% CI)	P value	Odds ratio (95% CI)	P value
White	1 (Reference)		1 (Reference)		1 (Reference)
Asian	3.58 (2.58–4.96)	< 0.001	4.81 (2.34–9.90)	< 0.001	1.98 (0.89–4.40)	0.09
Black	1.25 (0.79–1.98)	0.03	2.17 (0.82–5.78)	0.12	5.75 (2.82–11.73)	< 0.001
Multiracial	2.58 (1.34–4.98)	0.005	2.30 (0.74–7.14)	0.15	0.74 (0.09–6.08)	0.78
Other	1.56 (0.80–3.07)	0.20	1.95 (0.73–5.19)	0.18	1.74 (0.54–5.58)	0.35

Matched multivariable regression was performed using nearest-neighbor propensity score 1-to-5 matching and adjusted for age, sex, diagnosis of ulcerative colitis, type 1 diabetes mellitus, type 2 diabetes mellitus, vitamin D deficiency, psoriasis, systemic lupus erythematosus, hypothyroidism, hyperthyroidism, vitiligo

When focusing on individuals with AA and black natural hair color only, we found a male predominance (55.1%) and a mean (SD) age of 53 (8.8) years. Most participants were White (43.3%) followed by Asian (33.9%), and Black (13.4%). Compared to White participants with black natural hair color, Black patients with black natural hair color had a significantly lower odds of AA (0.50 (0.27–0.93), P = 0.03). Otherwise, there was no significant difference in AA prevalence between White participants with black hair and Asian, multiracial, or other race participants with black hair (Table 4).

Table 4.

Matched multivariable regression of prevalence of alopecia areata by race in patients with black natural hair color only

Race	Alopecia areata
	Odds ratio (95% CI)	P value
White	1 (Reference)
Asian	1.27 (0.79–2.06)	0.32
Black/Caribbean	0.50 (0.27–0.93)	0.03
Multiracial	1.14 (0.47–2.82)	0.77
Other	1.02 (0.22–4.76)	0.98

Matched multivariable regression was performed using nearest-neighbor propensity score 1-to-5 matching and adjusted for age, sex, diagnosis of ulcerative colitis, type 1 diabetes mellitus, type 2 diabetes mellitus, vitamin D deficiency, psoriasis, systemic lupus erythematosus, hypothyroidism, hyperthyroidism, vitiligo

Discussion

In this study, we found that natural hair color in White patients was associated with a differential risk of AA, but not AGA and SA. Specifically, compared to participants with dark brown hair, those with black hair had a higher risk of AA, whereas those with red and blonde hair had a lower risk. We did not find any statistical difference in lifetime prevalence of AGA or SA by natural hair color (Table 2). These findings are consistent with the hypothesis that pigment-related proteins in hair follicles may serve as antigenic targets, and that increased expression might therefore constitute a potential risk factor for T cell-dependent autoimmune alopecia, but not for other forms of hair loss [20, 21]

When performing racial analyses, we found that Asian, Black, and multiracial patients had higher odds of having AA compared to White patients. These findings build on previous US data on AA from the Nurses’ Health Study, Explorys Database, All of Us Research Program, and the National Alopecia Areata Registry, which suggest that patients of color, primarily Black, Asian, multiracial, and Hispanic patients, have a higher standardized prevalence of AA compared to White patients [6–8]. Also consistent with previous literature, we found higher odds of AGA in Asian compared to White patients [22, 23], as well as a higher odds of SA in Black patients compared to White patients (Table 3) [6].

When controlling for natural hair color, we found that racial differences in AA risk attenuated. Although Black and Asian patients overall had higher odds of AA compared to White patients, Asian patients with black hair had similar odds when compared to White patients with black hair. These findings suggest that natural hair color, rather than skin color or race, may be underpinning a patients’ risk of autoimmune alopecia. Interestingly, we also found that Black patients with black hair had significantly lower odds of AA compared to White patients with black hair. This finding was based on a relatively low sample size, however, and should be interpreted with caution.

Our finding that dark natural hair color, rather than race, appears to be the potential main driver of AA in people of color is consistent with previous mouse studies that have suggested that hair follicles are only attacked during the melanogenically active stages of the hair cycle (anagen III–VI) [24]. This leads to the rapid, preferential loss of pigmented hairs with sparing of gray hairs, resulting in the so-called observed Marie Antoinette syndrome, in which patients appear to have their hair turn white rapidly [25]. Similarly, as hair begins to regrow, white hairs often grow back before re-pigmenting [26]. While the mechanism of preferential loss of pigmented hairs in AA is unknown, a leading hypothesis is that pigment proteins, such as premelanosome protein (PMEL), which helps facilitate the process of melanin polymerization, may act as antigenic targets [27]. It is unclear, however, whether these antigenic targets are modified or unmodified self-antigens [21].

Since the pigmentation pathway resides downstream of the MITF transcription factor, it is anticipated that numerous melanocyte lineage restricted antigens would be simultaneously over- or underexpressed in melanocytes of dark (eumelanotic) vs light (pheomelanotic) hair follicle melanocytes [28]. On the other hand, immune targeting of melanocytes should not necessarily prevent hair growth, since unpigmented follicles typically retain viability. Therefore there may be an epitope spreading phenomenon or inflammatory process within the hair follicle microenvironment, which is targeting hair follicle keratinocyte populations. Our finding that AA is more prevalent among people with more pigmented hair, and thus more of these associated proteins, supports this hypothesis.

There is strong evidence to suggest that genetics play a large part in the development of AA [29], ranging from observations of increased incidence in certain family kindreds [30] to genome-wide association studies (GWAS) [31]. Genomic studies have identified numerous immune-related genetic variants associated with AA, including a variant near the STX17 gene, expressed in hair follicles [32], the MCHR2 gene, involving in melanin production and pigmentation [33], PMEL [20], and certain human leukocyte antigen (HLA) genotypes [34]. This constellation of risk-conferring genes strongly suggests an interaction between immune activity and pigmentation phenotypes, which are both known to be highly heritable [30, 34].

Limitations

This study has some limitations. First, our study only included individuals from the UK, which may not be representative of other populations. Second, the UKBB aggregates the Asian racial category, which is 81.1% South Asian and 18.9% “any other Asian background.” As such, it is unclear which Asian subgroup is driving our AGA findings. Third, the number of non-White AA cases was relatively comparatively low, limiting our power to detect racial differences. Fourth, the AGA cases identified in the Biobank data highlight only those that were severe or distressing enough for patients to bring to the attention of their medical providers. Since hair loss may be more distressing to women, that may be why there are more female patients represented in the UKBB. AA may also be more prevalent because it is more clearly recognized as pathology, whereas AGA is often ignored as normal aging. Finally, our study included a handful of covariates. Other potential risk factors, such as smoking, environmental exposure, and family history, were not included in this analysis.

Future Directions

To address the above limitations, future studies should utilize more accurate measures of hair color. Additionally, researchers should replicate this study in a more diverse global population to examine potential differences in the prevalence of alopecia across different ethnic groups. Finally, future mechanistic studies are needed to better characterize the molecular pathophysiology connecting autoimmune alopecia and natural hair color.

Conclusions

The results of our study provide new insights into how the natural color of hair may impact the likelihood of developing different types of alopecia. Specifically, we found that people with darker hair colors, such as black, may have a higher risk of developing alopecia areata, while those with lighter hair colors, like red and blonde, may have a lower risk. We did not find any significant relationship between hair color and non-autoimmune alopecia such as AGA or SA. Additionally, our study suggests that differences in hair color may explain the majority of previously described racial differences in AA incidence, thereby potentially altering the way dermatologists view the epidemiology of this disease. Overall, this study adds to the growing body of literature on the role of pigmentary genotypes and phenotypes in dermatologic disease and may have important implications for patient counseling and management.

Supplementary Information

Below is the link to the electronic supplementary material.

Author Contributions

Kanika Kamal and David Xiang had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Concept and design: Kanika Kamal, David Xiang, Katherine Young, David E Fisher, Arash Mostaghimi, Nicholas Theodosakis. Acquisition, analysis, or interpretation of data: Kanika Kamal, David Xiang, Katherine Young, David E Fisher, Arash Mostaghimi, Nicholas Theodosakis. Drafting of the manuscript: Kanika Kamal, David Xiang. Critical revision of the manuscript for important intellectual content: Kanika Kamal, David Xiang, Katherine Young, David E Fisher, Arash Mostaghimi, Nicholas Theodosakis. Statistical analysis: David Xiang, Arash Mostaghimi, Nicholas Theodosakis. Administrative, technical, or material support: David E Fisher, Arash Mostaghimi, Nicholas Theodosakis. Supervision: David E Fisher, Arash Mostaghimi, Nicholas Theodosakis.

Funding

D.E.F. gratefully acknowledges grant support from the Dr. Miriam and Sheldon G. Adelson Medical Research Foundation and grants from the NIH (R01AR072304; R01AR043369; P01CA163222). The Rapid service fee was waived for the authors.

Data Availability

The data that support the findings of this study are openly available in UK Biobank at www.ukbiobank.ac.uk.

Declarations

Conflicts of Interest

The authors of this study have several conflicts of interest. Arash Mostaghimi receives consulting fees from Pfizer, hims, Digital Diagnostics, Concert, Lilly, Abbvie, Equillium, and Boehringer Ingelheim. Arash Mostaghimi also owns equity in hims, Figure 1, Acom, Seebe. He receives licensing fees and royalties from Pfizer, Concert, and Lilly. He serves on the medical advisory board for hims, Figure 1, and Digital Diagnostics. Arash Mostaghimi oversees clinical trials for Lilly and Concert. David E Fisher has a financial interest in Soltego, a company developing salt inducible kinase inhibitors for topical skin-darkening treatments that might be used for a broad set of human applications. The interests of David E Fisher were reviewed and are managed by Massachusetts General Hospital and Partners HealthCare in accordance with their conflict-of-interest policies. He also has consulting and equity in Tasca, and consulting relationships with Biocoz, and Torqur. Kanika Kamal, David Xiang, Katherine Young and Nicholas Theodosakis have nothing to disclose.

Ethical Approval

This study was deemed exempt from the MassGeneral Brigham Institutional Review Board as it is non-human subjects research. Patient consent acquired by the United Kingdom Biobank.