Abstract
BACKGROUND:
Human scalp hair is a validated bio-substrate for monitoring various exposures in childhood including contextual stressors, environmental toxins, prescription or non-prescription drugs. Linear hair growth rates (HGR) are required to accurately interpret hair biomarker concentrations.
METHODS:
We measured HGR in a prospective cohort of preschool children (N = 266) aged 9–72 months and assessed demographic factors, anthropometrics, and hair protein content (HPC). We examined HGR differences by age, sex, race, height, hair pigment, and season, and used univariable and multivariable linear regression models to identify HGR-related factors.
RESULTS:
Infants below 1 year (288 ± 61 μm/day) had slower HGR than children aged 2–5 years (p = 0.0073). Dark-haired children (352 ± 52 μm/day) had higher HGR than light-haired children (325 ± 50 μm/day; p = 0.0019). Asian subjects had the highest HGR overall (p = 0.016). Younger children had higher HPC (p = 0.0014) and their HPC-adjusted HGRs were slower than older children (p = 0.0073). Age, height, hair pigmentation, and HPC were related to HGR in multivariable regression models.
CONCLUSIONS:
We identified age, height, hair pigment, and hair protein concentration as significant determinants of linear HGRs. These findings help explain the known hair biomarker differences between children and adults and aid accurate interpretation of hair biomarker results in preschool children.
INTRODUCTION
Discovery of hair biomarkers in the past few decades has transformed scientific disciplines like toxicology, pharmacology, epidemiology, forensics, healthcare, and developmental psychology.1–4 Moreover, the frequent use of hair in large population cohorts to measure environmental exposures in children has emerged from the discovery of hair biomarkers.5–9 Cortisol in human scalp hair is a validated biomarker for monitoring time-averaged acute/chronic responses to various familial, health, psychosocial, and economic conditions across the lifespan.10–14 Hair biomarkers capture these ecological exposures and can serve as organic record logs of various exposures.11,15–17 To identify the time-periods for these exposures, it is important to understand the determinants of hair growth and adjust for confounding factors.18–21
Scalp hair is characterized by growth cycles of active growth (anagen), transition (catagen), and resting (telogen) phases; hair density, diameter, or linear length; and even according to its texture (straight or curly), shape, pigmentation, or elasticity.22–25 These characteristics may reflect nutritional status, mitochondrial metabolism, puberty, pregnancy, or seasonal changes.18,25–29 Furthermore, a few hair studies demonstrate the diversity in hair quality and character across individual factors, such as race, ethnicity, age, sex, and disease.2,11,30–32
Studies of hair growth have been reported on exceptionally few children and focused on keratinization, medullation, or diameter/cross-sectional growth patterns.22,33–36 Human scalp hair, derived from the neuroectoderm and mesoderm,34,37 evolves via prenatal lanugo, postnatal vellus, intermediate medullary, and post-pubertal terminal hair stages.24,25,38 These studies also revealed that: (a) scalp hair medullation increases diameter with the greatest increases occurring at 1–3 years of age and no sex differences in hair diameter; (b) males have greater hair density (follicles/cm2) and faster linear hair growth rates at 3–9 years.22,24,36 Linear hair growth rates in adults are correlated with hair diameter, medullation, and loss of pigmentation; microscopically correlated with the hair follicle size, and interscale distances within the hair shaft.27,39,40
While these are important observations, linear hair growth rates (HGRs) are required to accurately interpret the time frame during which biomarkers are incorporated into the growing hair shaft, and linear HGRs remain undetermined in children. Current hair biomarker research practices involve the application of adult HGRs to the interpretation of biomarker levels in children.11,41 Rapid physical growth and development in childhood may alter HGRs; thus, child HGRs may differ from HGRs reported in adults.27 For example, differing biomarker (cortisol) levels in adults vs. children demonstrated by Neumann et al. and de Kruijff et al. may be due to age-dependent differences in HGRs.42,43
Race-dependent differences in hair growth linked to hair color, texture, biochemical composition, and hair growth cycles are determined in adults (specifically in African, Asian, European/Caucasian groups), but remain undetermined in children.44 Child hair biomarker studies have described analogous race-dependent differences linked to hair qualities.11,38 Furthermore, Barth and Pecoraro et al. reported higher HGRs in male vs. female children (5–15% faster hair growth; 340 vs. 302 μm/day, respectively), but both examined small sample sizes and only performed descriptive statistical analyses.38,45 On the whole, adult hair studies provide biomarker research with more sophisticated findings (more studies, larger sample sizes, rigorous methodologies, and statistical analyses) than child hair studies. For example, while the composition of adult hair is known (0.25–0.95% trace elements,1–9% lipids, 3–5% endogenous water, and 65–95% protein),46–50 the composition of children’s hair largely remains unknown.
We designed this study to understand hair growth physiology in early childhood, specifically investigating anthropometric measures and demographic factors that determine linear HGRs in preschool children. Results from this study will enable better interpretation of the timeframe in which biomarkers are incorporated into children’s growing hair so that appropriate adjustments can be made to hair biomarker levels associated with environmental exposures and the mind-brain-body connectome.
MATERIALS AND METHODS
Study cohort
Following approval from the Stanford University IRB, parents or guardians gave informed consent for participation in the Hair Biomarkers Study (https://childwellness.stanford.edu/hair-biomarkers-study). Healthy preschool children aged 9–72 months (N = 266) from Santa Clara County, CA were enrolled in a 3-week longitudinal observational study to evaluate HGRs during the period of November 2017 to February 2020. Subjects with tinea capitis, alopecia areata, eczema, or other scalp conditions, or those receiving any prescription or other drugs (except daily vitamins) were excluded. Subjects receiving systemic steroid therapy, or those with a history of chronic medical conditions such as cystic fibrosis, sickle cell disease, congenital heart disease, or other disorders, and subjects with chemical exposures (e.g., dying, bleaching, chemical straightening, perming) to their hair in the 3 months prior to study entry were also excluded. Qualitative assessment of social and demographic factors confirmed that our study sample only included healthy children from nuclear families free from adversity. The COVID-19 pandemic limited the enrollment and in-person contact with several participants for serial HGR measurements.
Determinants of hair growth
Hair samples, demographic information, and anthropometric measures were collected at seven participating sites. Parents were asked to classify their child’s race into racial categories recognized by the Federal government (White, Asian, African American, Alaska Native, or Pacific Islander). Ethnicity was assigned in accordance with the Department of Homeland Security’s code of ethnicities (based on the geographic location of birth, namely, Africa, Asia, Caribbean, Central America, Europe, North America, Oceania, and South America). Subjects were designated as “Mixed” if maternal and paternal race or ethnicity differed, assigned to the “Other” race group if they were identified as African American, Native Alaskan, or Pacific Islander, and into the “Other” ethnicity group if they were identified as African, Caribbean, or Oceanian. To assess nutritional status, we measured the child’s height, head, and waist circumferences.
We classified children into three hair pigment groups (light, medium, and dark hair) using a modification of Loussouarn’s methodology, which involved matching hair color to a reference scale compromised of a 10-unit gradient ranging from black (1) to pale blonde (10).27 Dark pigmentation included black/brown, black and dark black hair colors (pigments 1–3). Medium pigmentation included brown/red, light brown, brown and dark brown hair colors (pigments 4–7). Light pigmentation included gradients from pale blonde, blonde/brown and blonde/red hair colors (pigments 8–10). Hair sampling seasons (Spring, Summer, Autumn, Winter) were identified by dates of the two solstices and two equinoxes to assess seasonal changes in hair growth.
Hair sampling
Researchers were trained to trim a 1 cm2 area near the posterior vertex using a Philips Norelco Multigroom 3000 trimmer®, which cuts hair at 0.1 mm from the scalp. Fine digital calipers (Fisher Scientific) were used to measure the length of new hair at weekly follow-up visits. The same researcher who obtained hair measured the hair growth at weekly appointments across 7–21 days. Hair growth was averaged based on serial measurements of 3–5 hairs. The difference in hair length from the previous measurement was divided by the number of days between measurements to calculate the averaged hair growth rate (HGR) (μm/day).
Hair protein extraction and measurement
The distal end of hair was taped and the proximal (scalp) end was indicated by an arrow on a folded paper containing the sample. The hair sample was placed into a zip-lock bag, labeled, and stored at 4 °C until processing.12 The proximal 0–3 cm length of hair was weighed (10–50 mg) in a glass vial, finely cut to a powdery consistency, and processed with a four-step extraction procedure, with alternating two cycles of hair extraction in 1 ml of methanol (15 h, at 52 °C, rotating 200 rpm) followed by 1 ml acetone (5 min, at room temperature, rotating 200 rpm). Samples were centrifuged, supernatants removed and air dried at 4 °C. Dried residue was reconstituted with phosphate buffered saline (70 μl/10 mg hair) with transfer to microcentrifuge tubes. Samples were cleaned by cold (4 °C) centrifugation, 12,000 rpm for 15 min, and the supernatant was transferred to fresh tubes. The soluble hair protein content (HPC) was measured by spectrophotometric absorption at 260/280 (Take-3, Epoch plate reader, Gen5.5 software, BioTek Instruments).
Statistical methods
Descriptive statistics were used to summarize demographic characteristics and anthropometric measures for children enrolled. These variables were summarized using means with standard deviations or medians with interquartile ranges for continuous variables, frequencies and percentages for categorical variables. HGRs for each follow-up visit were averaged to obtain an overall mean HGR and then summarized by age group, sex, race, height tertiles, age-adjusted height tertiles, hair pigmentation, and hair sampling season. ANOVA and post hoc two-stage linear step-up procedure of Benjamini, Krieger and Yekutieli was performed to assess whether HGRs differed by age, sex, race, height, age-adjusted height, hair pigment, or season.
To examine associations between HGR and factors of interest, we performed several simple and multiple linear regression models. Simple linear regression was used to model the bivariate relationship between HGR and each factor. Multiple linear regression was used to examine the association between HGR and all factors simultaneously. Data were fitted to a linear regression model that included age, sex, race, height, hair pigmentation, and hair sampling season (Model 1) and HPC was added in Model 2.
Multiple imputations using fully conditional specification were used to impute any missing independent covariates used in these models. We created ten imputed datasets, and estimates were combined using Rubin’s rules. The ‘type3_MI_glm’ macro developed by Wang et al. was used to generate p-values from type-III analyses.51 All analyses were conducted using SAS 9.4 (SAS Institute, Inc., Cary, NC) and R (version 4.1.2.; R Foundation for Statistical Computing, Vienna, Austria). All tests were two-sided and evaluated using an alpha-error of 0.05.
RESULTS
Table 1 shows child demographics and measurements for the study sample (N = 266). Participants had an average age 3.5 ± 1.3 years, height 101.2 ± 12.2 cm, head circumference 49.7 ± 3.0 cm, and waist circumference 51.2 ± 4.5 cm. Nearly half of the subjects (44.7%) were White, while the remaining subjects were Asian (25.2%), Mixed (16.5%), and Other (13.5%, including African American, Pacific Islander, and Alaska Native). Most subjects belonged to non-Hispanic (88.7%) ethnicity. We studied similar numbers of male (54.1%) and female (45.9%) subjects and subjects with light (27.8%), medium (39.5%), and dark (32.7%) hair. Table 2 shows the distribution of sex, race, hair pigment, and anthropometric measures for children within each age group (<1-year-old, 1-, 2-, 3-, 4-, and 5-year-old children).
Table 1.
Demographics and measurements of preschool children (N = 266).
| Characteristics/Measures | ||
|---|---|---|
| Sex | N (%) | – |
| Female | 122 (45.9) | |
| Male | 144 (54.1) | |
| Age categories | N (%) | – |
| <1 year old | 9 (3.4) | |
| 1 year old | 30 (11.3) | |
| 2 years old | 47 (17.7) | |
| 3 years old | 61 (22.9) | |
| 4 years old | 85 (32.0) | |
| 5 years old | 34 (12.8) | |
| Age | Mean (SD) | Median (Q1, Q3) |
| Years | 3.5 (1.3) | 3.8 (2.6, 4.5) |
| Months | 42.4 (15.5) | 45.5 (31, 54) |
| Race | N (%) | – |
| Asian | 67 (25.2) | |
| White | 119 (44.7) | |
| Mixed | 44 (16.5) | |
| Other | 36 (13.5) | |
| Ethnicity (*N = 3) | N (%) | – |
| Hispanic | 30 (11.3) | |
| Not Hispanic | 236 (88.7) | |
| Hair Pigmentation | N (%) | – |
| Light | 74 (27.8) | |
| Medium | 105 (39.5) | |
| Dark | 87 (32.7) | |
| Anthropometric Measures | Mean (SD) | Median (Q1, Q3) |
| Height (cm) | 101.2 (12.2) | 103 (94, 111) |
| Head circumference (cm) (*N = 3) | 49.7 (3.0) | 50 (48.5, 51) |
| Waist circumference (cm) (*N = 2) | 51.2 (4.5) | 51 (48.5, 53.7) |
| Left arm circumference (cm) (*N = 6) | 16.9 (1.3) | 17 (16, 17.8) |
| Right arm circumference (cm) (*N = 7) | 16.9 (1.6) | 17 (16.3, 17.8) |
Number of missing values.
Table 2.
Child demographics and measurements by age groups.
| Characteristics/Measures | 0- < 1 year (N = 9) | 1 year (N = 30) | 2 years (N = 47) | 3 years (N = 61) | 4 years (N = 85) | 5 years (N = 34) |
|---|---|---|---|---|---|---|
| Age | Mean (SD) | Mean (SD) | Mean (SD) | Mean (SD) | Mean (SD) | Mean (SD) |
| Age in years | 0.8 (0.1) | 1.5 (0.2) | 2.5 (0.3) | 3.5 (0.3) | 4.4 (0.3) | 5.4 (0.2) |
| Age in months | 9.7 (0.7) | 17.7 (2.8) | 29.9 (3.5) | 42 (3.6) | 53.1 (3.1) | 64.2 (2.7) |
| Sex | N (%) | N (%) | N (%) | N (%) | N (%) | N (%) |
| Female | 7 (77.8) | 12 (40) | 25 (53.2) | 27 (44.3) | 36 (42.4) | 15 (44.1) |
| Male | 2 (22.2) | 18 (60) | 22 (46.8) | 34 (55.7) | 49 (57.6) | 19 (55.9) |
| Race | N (%) | N (%) | N (%) | N (%) | N (%) | N (%) |
| Asian | 2 (22.2) | 4 (13.3) | 8 (17) | 16 (26.2) | 25 (29.4) | 12 (35.3) |
| White | 6 (66.7) | 16 (53.3) | 21 (44.7) | 30 (49.2) | 34 (40) | 12 (35.3) |
| Mixed | 0 (0) | 6 (20) | 11 (23.4) | 7 (11.5) | 16 (18.8) | 4 (11.8) |
| Other | 1 (11.1) | 4 (13.3) | 7 (14.9) | 8 (13.1) | 10 (11.8) | 6 (17.6) |
| Hair Pigmentation | N (%) | N (%) | N (%) | N (%) | N (%) | N (%) |
| Light | 5 (55.6) | 11 (36.7) | 17 (36.2) | 17 (27.9) | 20 (23.5) | 4 (11.8) |
| Medium | 3 (33.3) | 14 (46.7) | 22 (46.8) | 22 (36.1) | 32 (37.6) | 12 (35.3) |
| Dark | 1 (11.1) | 5 (16.7) | 8 (17) | 22 (36.1) | 33 (38.8) | 18 (52.9) |
| Anthropometric Measures | Mean (SD) | Mean (SD) | Mean (SD) | Mean (SD) | Mean (SD) | Mean (SD) |
| Height (cm) | 74.6 (2.8) | 84 (7.7) | 92.6 (5.5) | 102 (6.1) | 109 (6.3) | 114.3 (6) |
| Head circumference (cm) (*N = 3) | 44.6 (1.1) | 47.6 (1.8) | 48.8 (4.9) | 49.7 (2.2) | 50.7 (1.5) | 51.2 (1.8) |
| Waist circumference (cm) (*N = 2) | 45.3 (3.2) | 48.5 (4.1) | 48.2 (5.5) | 51.5 (3.2) | 53.0 (3.3) | 53.6 (3.7) |
| Left arm circumference (cm) (*N = 6) | 15.3 (0.7) | 16.3 (1.1) | 16.4 (1.7) | 16.9 (1.1) | 17.3 (1.1) | 17.6 (1.0) |
| Right arm circumference (cm) (*N = 7) | 15.4 (0.8) | 16.3 (1.1) | 16.4 (1.8) | 17.0 (1.0) | 17.1 (2.1) | 17.7 (1.0) |
Number of missing values.
Hair growth rates
The average HGR ranged from 287.9 ± 60.7 μm/day in infants less than 1 year old to 345.1 ± 47.2 μm/day in 3-year-old children. HGR rates differed among age groups, with infants less than 1 year old showing slower HGRs than 2-, 3-, 4-, and 5-year-old children (all p < 0.05, Table 3). We found no sex differences in HGRs (female 338.4 ± 47.2 μm/day vs. male 340.7 ± 50.4 μm/day; p = 0.71). While HGR was not associated with waist or head or arm circumferences, subjects in the top tertiles for height and age-adjusted height had higher HGRs than those in the bottom tertile (p = 0.0025 and p = 0.04, respectively, Table 3).
Table 3.
Factors affecting hair growth rates for children.
| N | Mean (SD) | ANOVA | P-value | |
|---|---|---|---|---|
| Agea | 0.0073 | |||
| <1 year | 9 | 287.9 (60.7) | ||
| 1 year old | 30 | 322.5 (57.1) | ||
| 2 years old | 47 | 343.6 (47.1) | ||
| 3 years old | 61 | 345.1 (47.2) | ||
| 4 years old | 85 | 343.5 (46.8) | ||
| 5 years old | 34 | 343.3 (39.7) | ||
| Sex | 0.71 | |||
| Female | 122 | 338.4 (47.2) | ||
| Male | 144 | 340.7 (50.4) | ||
| Raceb | 0.016 | |||
| Asian | 67 | 355.6 (50.3) | ||
| White | 119 | 332.1 (48.5) | ||
| Mixed | 44 | 338.4 (48.7) | ||
| Other | 36 | 336.0 (42.2) | ||
| Heightc | 0.0025 | |||
| Bottom tertile | 89 | 325.4 (54.3) | ||
| Middle tertile | 89 | 344.2 (43.8) | ||
| Top tertile | 88 | 349.3 (45.0) | ||
| Age-Adjusted Heightd | 0.040 | |||
| Bottom tertile | 85 | 329.2 (57.4) | ||
| Middle tertile | 94 | 341.5 (37.5) | ||
| Top tertile | 87 | 347.7 (49.3) | ||
| Hair Pigmentatione | 0.0019 | |||
| Light | 74 | 325.3 (50) | ||
| Medium | 105 | 339.1 (43) | ||
| Dark | 87 | 352.4 (51.6) | ||
| Season | 0.22 | |||
| Winter | 56 | 338.1 (51) | ||
| Spring | 40 | 349.1 (27.6) | ||
| Summer | 29 | 351.3 (36.9) | ||
| Fall | 141 | 335.1 (54.2) |
Age: Tukey post hoc tests: <1 year vs. 2-year olds, p = 0.019, <1 year vs. 3-year olds, p = 0.012, <1 year vs. 4-year olds, p = 0.013, <1 year vs. 5-year-olds, p = 0.027.
Race: Tukey post hoc tests Asian vs. White race, p = 0.0085.
Height: Tukey post hoc tests: bottom vs. middle tertile, p = 0.025, bottom vs. top tertile, p = 0.0030.
Age-adjusted height: post hoc tests bottom vs. top tertile, p = 0.034.
Hair Pigmentation: Tukey post hoc test light vs. dark pigment hair, p = 0.0012.
Subjects with dark hair pigment (352.4 ± 51.6 μm/day) had higher HGRs compared to subjects with light hair pigment (325.3 ± 50.0 μm/day; p = 0.0019, Table 3). Asian subjects had the highest HGRs among all races and significantly higher HGR compared to White subjects (355.6 ± 50.3 μm/day vs. 332.1 ± 48.5 μm/day, respectively, p = 0.016, Table 3). HGRs differed across the age groups for White children (p = 0.0026) but not Asian children (p = 0.19), and were inconclusive for “Mixed” and “Other” race groups. No HGR differences were observed between the four seasons.
Table 4 presents the results of linear regression analyses. Consistent with the findings from ANOVA in Table 3, simple linear regression showed significant relationships between HGR and age (p = 0.0073), height (p = 0.0025), race (p = 0.016) and hair pigmentation (p = 0.0019). Age (p = 0.0066) and height (p = 0.005) remained significant when other factors were adjusted in the linear regression model (Model 1). Model 1 explained 15% of the variation in children’s HGR. With addition of hair protein content (HPC) to the model (Model 2), significant differences in HGR were seen for age (p = 0.018), height (p = 0.0073) and HPC (p = 0.013), whereas hair pigmentation had marginal effects (p = 0.062). Model 2 explained 17% of the variation in HGR and fit the data better (p = 0.01) as compared to Model 1. An R-squared of 0.17 corresponds to Cohen’s effect size F-squared of 0.20, which based on Cohen’s 1988 criteria, indicates a medium effect size for HGR variability in children. We also tested the assumptions for both linear models. As shown in Fig. 1, the density plots and scatterplots of studentized residuals for hair growth rates followed a linear pattern for both Models 1 and 2, showing that the linearity and the equal variance assumptions were being met.52 In addition, the residuals for both models were not skewed, because all points in the Quantile-Quantile plots (Q-Q plots) were distributed along the 45-degree diagonal reference line, further substantiating that the normality assumption was satisfied (Fig. 1).52
Table 4.
Linear regression analyses for determinants of hair growth rates.
| Parameter | Unadjusted | Model 1 | Model 2 | |||
|---|---|---|---|---|---|---|
| Estimate (95% CI) | P-value | Estimate (95% CI) | P-value | Estimate (95% CI) | P-value | |
| Sex | 0.71 | 0.53 | 0.46 | |||
| Female | Ref | Ref | Ref | |||
| Male | 2.3 (−9.6, 14.1) | −3.8 (−15.5, 8) | −4.3 (−15.9, 7.2) | |||
| Age | 0.0073 | 0.0066 | 0.018 | |||
| < 1 year | Ref | – | – | Ref | – | |
| 1 year old | 34.6 (−1.2, 70.4) | 0.058 | 37.4 (1.8, 72.9) | 0.039 | 24.1 (−12.4, 60.6) | 0.20 |
| 2 years old | 55.7 (21.4, 90.0) | 0.0015 | 53.7 (19.5, 87.9) | 0.0022 | 41.6 (6.6, 76.6) | 0.020 |
| 3 years old | 57.2 (23.6, 90.9) | 0.0009 | 36.5 (−0.9, 73.8) | 0.056 | 24.8 (−13.1, 62.7) | 0.20 |
| 4 years old | 55.6 (22.5, 88.6) | 0.0011 | 21.2 (−17.9, 60.4) | 0.29 | 8.7 (−31.2, 48.5) | 0.67 |
| 5 years old | 55.4 (20.1, 90.8) | 0.0022 | 12.2 (−30.5, 54.9) | 0.57 | −1.1 (−44.4, 42.2) | 0.96 |
| Height | 0.0025 | 0.0050 | 0.0073 | |||
| Bottom | Ref | – | Ref | – | Ref | – |
| Middle | 18.8 (4.7, 33.0) | 0.0093 | 24.6 (3.8, 45.5) | 0.021 | 23.4 (2.8, 44) | 0.026 |
| Top | 23.9 (9.7, 38.1) | 0.0010 | 41.6 (16.7, 66.6) | 0.0012 | 39.7 (15.1, 64.3) | 0.0016 |
| Race | 0.016 | 0.40 | 0.52 | |||
| White | Ref | – | Ref | – | Ref | |
| Asian | 23.5 (9.0, 38.0) | 0.0016 | 10.1 (−11.3, 31.4) | 0.35 | 6.1 (−15.1, 27.4) | 0.57 |
| Mixed | 6.3 (−10.5, 23.0) | 0.46 | −5.6 (−23.3, 12.0) | 0.53 | −7.4 (−24.8, 10) | 0.41 |
| Other | 3.9 (−14.1, 22.0) | 0.67 | −5.6 (−25.5, 14.3) | 0.58 | −6.1 (−25.7, 13.5) | 0.54 |
| Hair Pigmentation | 0.0019 | 0.17 | 0.062 | |||
| Light | Ref | – | Ref | – | Ref | – |
| Medium | 13.7 (−0.6, 28.1) | 0.060 | 13.4 (−1.9, 28.6) | 0.085 | 16.6 (1.4, 31.8) | 0.033 |
| Dark | 27.1 (12.2, 42.0) | 0.0004 | 17.6 (−3.9, 39.1) | 0.11 | 22.7 (1.1, 44.2) | 0.039 |
| Season | 0.22 | 0.34 | 0.14 | |||
| Winter | Ref | – | Ref | – | Ref | – |
| Spring | 11.1 (−8.8, 30.9) | 0.27 | 1.3 (−19.4, 21.9) | 0.90 | −0.3 (−20.7, 20.1) | 0.98 |
| Summer | 13.2 (−8.8, 35.1) | 0.24 | 14.6 (−6.8, 36.0) | 0.18 | 19 (−2.3, 40.4) | 0.081 |
| Fall | −3.0 (−18.1, 12.2) | 0.70 | −3.2 (−18.1, 11.7) | 0.67 | −4.2 (−18.9, 10.5) | 0.57 |
| Hair Protein (ug/mg of hair) | −0.1 (−0.2, −0.04) | 0.0046 | – | – | −0.1 (−0.2, −0.03) | 0.013 |
| R-squared | – | 0.15 | – | 0.17 | ||
| Δ R-squared | – | – | – | 0.02 | 0.01 | |
Fig. 1. Statistical analyses to test the normality and equal variance assumptions for the two linear regression models.
Each row shows the density plots, scatterplots of studentized residuals, and the Q-Q plots, respectively, for hair growth rates in Models 1 and 2. Density plots show the distribution of HGR values, scatterplots of studentized residuals follow a linear pattern that meets the linearity and the equal variance assumptions, and all points in the Q-Q plot for each model were distributed along the 45-degree diagonal reference line, further satisfying the normality assumption.
HPC was negatively correlated to HGR (r = −0.18, p = 0.001) and significantly altered by age, showing higher values in infants below 1 year compared to children at older ages (p = 0.0014, Fig. 2a). There were no differences in HPC between children grouped by race (p = 0.2261). HGRs were lower in the youngest children and significantly higher at 2–5 years (p = 0.0073, Fig. 2b). Given these findings and improvement in the HGR regression model after adding HPC, we adjusted each child’s HGR by their measured hair protein content. The ratio of HPC to HGR revealed the highest levels of hair protein synthesis below age 1 year with steady decline through to age 5 (p = 0.0001, Fig. 2c).
Fig. 2. Whisker plots for HPC, HGR, and HPC:HGR by age.
a shows whisker plots for the Hair Protein Content (HPC, μg protein/mg of hair) at different ages showing higher protein content in the youngest children (p = 0.0014); b shows the HPC-adjusted Hair Growth Rate (HGR) by age suggesting faster HGRs at 2, 3, 4, and 5 years compared to the youngest children (p = 0.0073); c shows the HPC to HGR ratio for each child, suggesting higher rates of daily hair protein synthesis in the youngest children (p = 0.0002). ANOVA and post hoc two-stage linear step-up procedure of Benjamini, Krieger and Yekutili tests were performed to limit the false discovery rate to less than 5%.
DISCUSSION
This is the first study to determine potential covariates of linear hair growth in preschool children. Linear HGRs in preschool children are age-dependent with overall increases in the HGR as children age; HGRs are lowest in children under 1 year of age and highest in children aged 3–5 years. These findings are consistent with the findings of Barth and de Kruijff et al., who reported lower HGRs in children below 2 years of age.38,42 Furthermore, age explains a moderate degree of HGR variance when comparing adult vs. child HGRs from Asian, European, and North American groups (30, 39, and 55% variance, respectively; see Supplementary).
There are several plausible explanations for the age-dependent differences in HGRs. First, newborn hair has much less medulla than infant hair, and the degree of medullation, which is associated with the anagen phase of hair growth, increases throughout childhood.22 Furthermore, it is hypothesized that the four types of infant hair (i.e., lanugo in newborns, vellus hair, an intermediate form of hair, followed by terminal hair) grow at different rates.53 Second, around the first year of life the hair growth cycle transitions from a state of synchronous to asynchronous activity in the hair follicles.38,54 This transition results in variable hair loss and significant differences in the percentage of terminal hair follicles; subsequently, there is a transient decrease in HGR around 1 year of age. Once hair growth has transitioned from synchronous to asynchronous cycles, hairs are uniformly distributed to all stages of the growth cycle, which results in a relative net increase in HGR (when compared to the transition period) in older children—this is maintained unless disrupted by ageing, hormonal changes, or illness.19,20,38,55–57
When comparing within ethnicities, child HGRs observed in our study differ from published adult HGR ranges.27 The discrepancy between adult and child HGRs may promote inaccurate interpretations of biomarker levels; thus, we caution investigators applying adult linear HGRs for defining the time-period of biomarker incorporation into children’s hair. For example, variations in linear HGRs (child vs. adult) may partially account for the higher hair cortisol levels observed in young children as compared to adults.11,42,43
Our study also determined that race, hair pigmentation, height, and age-adjusted height are covariates of HGR in preschool children. Specifically, HGRs were higher in Asian vs. White children, those with dark vs. light pigmented hair, and in the taller children. Examination of the age-dependent inverse relationship between HGR and HPC (i.e., younger children have lower HGR and higher HPC) via the HPC to HGR ratio revealed that hair protein synthesis is highest in children below 1 year of age and decreases as children get older. One plausible explanation for these findings may be that the protein utilization required for increased height growth velocity may lead to lower HPC in older/taller children, although further research directly comparing the rates of total body protein synthesis with hair protein synthesis will be required to assess this relationship. In adults, decreases in hair pigmentation are associated with decreases in telogen (resting phase) density, with likely increases in the proportion of anagen hair follicles in adults with lighter color hair.27 Longitudinal analyses indicate that cycling of hair follicles is independent of linear HGRs in adults.56 Therefore, even if children with light-colored hair had a higher proportion of anagen hairs, that does not contradict our finding of lower HGRs in these children.
There is limited data on the effect of malnutrition on hair growth rates in preschool children.58 Our findings did not indicate an association between waist circumference and HGR. However, our study protocol included a healthy population at low risk of malnutrition. No other published data on anthropometric measures and HGRs was identified, therefore, future studies enrolling at-risk populations may provide insight on the undetermined association of HGR and other anthropometric measures in preschool children.
Strengths of this study include a larger and more diverse sample than all prior studies on hair growth. We also studied a broad range of potential covariates and identified several HGR-related factors in preschool children. Thus, our study provides determinants of linear HGRs for previously understudied preschool children. Weaknesses of this study include the uneven distribution of participants from different racial/ethnic groups. Very few subjects identified as Pacific Islander, Alaska Native, or African American; these subjects were collectively assigned to the “Other” group. Uneven distribution of boys and girls also occurred in the <1-year age group, though HGRs did not differ across sex in this or other age groups. Lack of hair quality assessment (given the established racial and ethnic differences in hair qualities, such as texture, form, and thickness) was another weakness of our study.27,35,38,59,60 We advise that future reference studies must include diverse populations and incorporate objective metrics for evaluating hair qualities, so that meaningful associations between hair growth and race/ethnicity can be investigated to facilitate the interpretation of hair biomarkers in children.
CONCLUSIONS
Compared to the existing literature on linear hair growth in children, our data provides potential covariates of hair growth in a racially and ethnically diverse population of preschool children. Determinants of linear HGR in preschool children will be essential to implement biomarker testing and to accurately interpret their results in pediatric research and clinical practice. Our study reveals age, race, hair pigment (color), and height as covariates of HGRs in preschool children; studies measuring hair biomarkers should account for these determinants. Moreover, our initial observations on hair protein content and HGRs may provide physiological explanations for differences in HGRs and biomarkers in preschool children as compared to adults.
Supplementary Material
IMPACT:
Discovery of hair biomarkers in the past few decades has transformed scientific disciplines like toxicology, pharmacology, epidemiology, forensics, healthcare, and developmental psychology.
Identifying determinants of hair growth in children is essential for accurate interpretation of hair biomarker results in pediatric clinical studies.
Childhood hair growth rates define the time-periods of biomarker incorporation into growing hair, essential for interpreting the biomarkers associated with environmental exposures and the mind-brain-body connectome.
Our study describes age-, sex-, and height-based distributions of linear hair growth rates and provides determinants of linear hair growth rates in a large population of children.
Age, height, hair pigmentation, and hair protein content are determinants of hair growth rates and should be accounted for in child hair biomarkers studies.
Our findings on hair protein content and linear hair growth rates may provide physiological explanations for differences in hair growth rates and biomarkers in preschool children as compared to adults.
ACKNOWLEDGEMENTS
We thank Grace K-Y. Tam, Clinical Research Coordinator, Pain/Stress Neurobiology Laboratory, and Dr. Sukyung Chung, Quantitative Sciences Unit, Stanford University School of Medicine, Stanford, CA for their contributions to this study and to earlier versions of this manuscript.
FUNDING
Grants from the Eunice Kennedy Shriver National Institute for Child Health & Human Development (R01 HD099296) and the Maternal & Child Health Research Institute to KJSA supported this study. Study sponsors had no role in the design and conduct of the study; the collection, management, analysis, or interpretation of the data; the preparation, review, approval, or decision to publish this manuscript.
Footnotes
COMPETING INTERESTS
The authors declare no competing interests.
ETHICS APPROVAL AND CONSENT TO PARTICIPATE
Parents/Guardians of child participants gave written consent to trained research coordinators at the IRB-approved recruitment sites.
ADDITIONAL INFORMATION
Supplementary information The online version contains supplementary material available at https://doi.org/10.1038/s41390-023-02791-z.
DATA AVAILABILITY
The datasets generated during and/or analyzed during the current study are available from the corresponding author on request.
REFERENCES
- 1.Niebel A, Pragst F, Krumbiegel F & Hartwig S Prevalence of cathinones and other new psychoactive substances in hair of parents and children of families with known or suspected parental abuse of conventional illegal drugs. Forensic Sci. Int 331, 111148 (2022). [DOI] [PubMed] [Google Scholar]
- 2.Bhattarai D et al. Hair microscopy: an easy adjunct to diagnosis of systemic diseases in children. Appl. Microsc 51, 18 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Plott TJ et al. Age-related changes in hair shaft protein profiling and genetically variant peptides. Forensic Sci. Int. Genet 47, 102309 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Leonhard MJ, Chang ET, Loccisano AE & Garry MR A systematic literature review of epidemiologic studies of developmental manganese exposure and neurodevelopmental outcomes. Toxicology 420, 46–65 (2019). [DOI] [PubMed] [Google Scholar]
- 5.Fabelova L et al. Hair concentration of trace elements and growth in homeless children aged <6years: results from the enfams study. Environ. Int 114, 318–325 (2018). [DOI] [PubMed] [Google Scholar]
- 6.Pineiro XF et al. Heavy metal contamination in Peru: implications on children’s health. Sci. Rep 11, 22729 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Takeuchi H et al. Lead exposure is associated with functional and microstructural changes in the healthy human brain. Commun. Biol 4, 912 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Lyon-Caen S et al. Deciphering the impact of early-life exposures to highly variable environmental factors on foetal and child health: design of sepages couple-child cohort. Int. J. Environ. Res. Public Health 16, 1–29 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Perng W et al. Early life exposure in Mexico to environmental toxicants (element) project. BMJ Open 9, e030427 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.DePasquale CE et al. Lifetime stressors, hair cortisol, and executive function: age-related associations in childhood. Dev. Psychobiol 63, 1043–1052 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Gray NA et al. Determinants of hair cortisol concentration in children: a systematic review. Psychoneuroendocrinology 87, 204–214 (2018). [DOI] [PubMed] [Google Scholar]
- 12.Slominski R, Rovnaghi CR & Anand KJS Methodological considerations for hair cortisol measurements in children. Ther. Drug Monit 37, 812–820 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Anand KJS et al. Measuring socioeconomic adversity in early life. Acta Paediatr 108, 1267–1277 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Anand KJS et al. Demographic and psychosocial factors associated with hair cortisol concentrations in preschool children. Pediatr. Res 87, 1119–1127 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Lakshmi Priya MD & Geetha A A biochemical study on the level of proteins and their percentage of nitration in the hair and nail of autistic children. Clin. Chim. Acta 412, 1036–1042 (2011). [DOI] [PubMed] [Google Scholar]
- 16.Jursa T, Stein CR & Smith DR Determinants of hair manganese, lead, cadmium and arsenic levels in environmentally exposed children. Toxics 6, 1–15 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Gareri J & Koren G Prenatal hair development: implications for drug exposure determination. Forensic Sci. Int 196, 27–31 (2010). [DOI] [PubMed] [Google Scholar]
- 18.Courtois M, Loussouarn G, Hourseau S & Grollier JF Periodicity in the growth and shedding of hair. Br. J. Dermatol 134, 47–54 (1996). [PubMed] [Google Scholar]
- 19.Grymowicz M et al. Hormonal effects on hair follicles. Int. J. Mol. Sci 21, 5342 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Pierard-Franchimont C, Petit L, Loussouarn G, Saint-Leger D & Pierard GE The hair eclipse phenomenon: sharpening the focus on the hair cycle chronobiology. Int J. Cosmet. Sci 25, 295–299 (2003). [DOI] [PubMed] [Google Scholar]
- 21.Van Neste DJ, de Brouwer B & De Coster W The phototrichogram: analysis of some technical factors of variation. Ski. Pharm 7, 67–72 (1994). [DOI] [PubMed] [Google Scholar]
- 22.Duggins OH & Trotter M Age changes in head hair from birth to maturity. Ii. Medullation in hair of children. Am. J. Phys. Anthropol 8, 399–415 (1950). [DOI] [PubMed] [Google Scholar]
- 23.Duggins OH & Trotter M Changes in morphology of hair during childhood. Ann. N. Y Acad. Sci 53, 569–575 (1951). [DOI] [PubMed] [Google Scholar]
- 24.Trotter M & Duggins OH Age changes in head hair from birth to maturity; index and size of hair of children. Am. J. Phys. Anthropol 6, 489–506 (1948). [DOI] [PubMed] [Google Scholar]
- 25.Trotter M & Duggins OH Age changes in head hair from birth to maturity. Am. J. Phys. Anthropol 8, 467–484 (1950). [DOI] [PubMed] [Google Scholar]
- 26.Liu C et al. Changes in Chinese hair growth along a full year. Int. J. Cosmet. Sci 36, 531–536 (2014). [DOI] [PubMed] [Google Scholar]
- 27.Loussouarn G et al. Diversity in human hair growth, diameter, colour and shape. an in vivo study on young adults from 24 different ethnic groups observed in the five continents. Eur. J. Dermatol 26, 144–154 (2016). [DOI] [PubMed] [Google Scholar]
- 28.Randall VA & Ebling FJ Seasonal changes in human hair growth. Br. J. Dermatol 124, 146–151 (1991). [DOI] [PubMed] [Google Scholar]
- 29.Eerkens JW, Ryder A, Evoy A & Hull B Hydrogen isotopes in serial hair samples record season of death in a mummified child from 19th century San Francisco, Ca. Am. J. Phys. Anthropol 173, 606–614 (2020). [DOI] [PubMed] [Google Scholar]
- 30.Commo S et al. Age-dependent changes in eumelanin composition in hairs of various ethnic origins. Int J. Cosmet. Sci 34, 102–107 (2012). [DOI] [PubMed] [Google Scholar]
- 31.Sarifakioglu E, Yilmaz AE, Gorpelioglu C & Orun E Prevalence of scalp disorders and hair loss in children. Cutis 90, 225–229 (2012). [PubMed] [Google Scholar]
- 32.Khumalo NP & Gumedze F African hair length in a school population: a clue to disease pathogenesis? J. Cosmet. Dermatol 6, 144–151 (2007). [DOI] [PubMed] [Google Scholar]
- 33.Alves R & Grimalt R Hair loss in children. Curr. Probl. Dermatol 47, 55–66 (2015). [DOI] [PubMed] [Google Scholar]
- 34.Dawber RP The embryology and development of human scalp hair. Clin. Dermatol 6, 1–6 (1988). [DOI] [PubMed] [Google Scholar]
- 35.Das-Chaudhuri AB & Chopra VP Variation in hair histological variables: medulla and diameter. Hum. Hered 34, 217–221 (1984). [DOI] [PubMed] [Google Scholar]
- 36.Wynkoop EM A study of the age correlations of the cuticular scales, medullas, and shaft diameters of human head hair. Am. J. Phys. Anthropol 13, 177–188 (1929). [Google Scholar]
- 37.Smith DW & Gong BT Scalp hair patterning as a clue to early fetal brain development. J. Pediatr 83, 374–380 (1973). [DOI] [PubMed] [Google Scholar]
- 38.Barth JH Normal hair growth in children. Pediatr. Dermatol 4, 173–184 (1987). [DOI] [PubMed] [Google Scholar]
- 39.Baque CS et al. Relationships between hair growth rate and morphological parameters of human straight hair: a same law above ethnical origins? Int J. Cosmet. Sci 34, 111–116 (2012). [DOI] [PubMed] [Google Scholar]
- 40.Van Neste D Thickness, medullation and growth rate of female scalp hair are subject to significant variation according to pigmentation and scalp location during ageing. Eur. J. Dermatol 14, 28–32 (2004). [PubMed] [Google Scholar]
- 41.Kirschbaum C, Tietze A, Skoluda N & Dettenborn L Hair as a retrospective calendar of cortisol production-increased cortisol incorporation into hair in the third trimester of pregnancy. Psychoneuroendocrinology 34, 32–37 (2009). [DOI] [PubMed] [Google Scholar]
- 42.de Kruijff I et al. Lc-Ms/Ms-based reference intervals for hair cortisol in healthy children. Psychoneuroendocrinology 112, 104539 (2020). [DOI] [PubMed] [Google Scholar]
- 43.Neumann A et al. Predicting hair cortisol levels with hair pigmentation genes: a possible hair pigmentation bias. Sci. Rep 7, 8529 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Buffoli B et al. The human hair: from anatomy to physiology. Int J. Dermatol 53, 331–341 (2014). [DOI] [PubMed] [Google Scholar]
- 45.Pecoraro V, Astore I, Barman J & Ignacioaraujo C The normal trichogram in the child before the age of puberty. J. Invest Dermatol 42, 427–430 (1964). [DOI] [PubMed] [Google Scholar]
- 46.Nicolaides N & Rothman S Studies on the chemical composition of human hair fat. I. The squalene-cholesterol relationship in children and adults. J. Invest Dermatol 19, 389–391 (1952). [DOI] [PubMed] [Google Scholar]
- 47.Adav SS et al. Studies on the proteome of human hair—identification of histones and deamidated keratins. Sci. Rep 8, 1599 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48.Cruz CF et al. Keratins and lipids in ethnic hair. Int J. Cosmet. Sci 35, 244–249 (2013). [DOI] [PubMed] [Google Scholar]
- 49.Nicolaides N & Rothman S Studies on the chemical composition of human hair fat. II. The overall composition with regard to age, sex and race. J. Invest Dermatol 21, 9–14 (1953). [DOI] [PubMed] [Google Scholar]
- 50.Robbins CR Chemical and Physical Behavior of Human Hair: Chapter 2. Chemical Composition of Different Hair Types 4th edn, 105–176 (Springer, 2002). [Google Scholar]
- 51.Wang B, Fang Y & Jin M Combining type-III analyses from multiple imputations. SAS J 14, 1543 https://Support.Sas.Com/Resources/Papers/Proceedings14/1543-2014.Pdf (2015). [Google Scholar]
- 52.Mishra P et al. Descriptive statistics and normality tests for statistical data. Ann. Card. Anaesth 22, 67–72 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 53.Janniger CK & Bryngil JM Hair in infancy and childhood. Cutis 51, 336–338 (1993). [PubMed] [Google Scholar]
- 54.Furdon SA & Clark DA Scalp hair characteristics in the newborn infant. Adv. Neonatal Care 3, 286–296 (2003). [DOI] [PubMed] [Google Scholar]
- 55.Courtois M, Loussouarn G, Hourseau C & Grollier JF Ageing and hair cycles. Br. J. Dermatol 132, 86–93 (1995). [DOI] [PubMed] [Google Scholar]
- 56.Halloy J, Bernard BA, Loussouarn G & Goldbeter A Modeling the dynamics of human hair cycles by a follicular automaton. Proc. Natl Acad. Sci. USA 97, 8328–8333 (2000). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 57.Thom E Stress and the hair growth cycle: cortisol-induced hair growth disruption. J. Drugs Dermatol 15, 1001–1004 (2016). [PubMed] [Google Scholar]
- 58.Neuberger FM, Jopp E, Graw M, Puschel K & Grupe G Signs of malnutrition and starvation–reconstruction of nutritional life histories by serial isotopic analyses of hair. Forensic Sci. Int 226, 22–32 (2013). [DOI] [PubMed] [Google Scholar]
- 59.Morgan MD et al. Genome-wide study of hair colour in Uk Biobank explains most of the SNP heritability. Nat. Commun 9, 5271 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 60.De la Mettrie R et al. Shape variability and classification of human hair: a worldwide approach. Hum. Biol 79, 265–281 (2007). [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The datasets generated during and/or analyzed during the current study are available from the corresponding author on request.