Acanthosis Nigricans Is a Strong Predictor of Low Blood Calcidiol Levels in Children and Adolescents

Abstract

Background:

Clinical consensus differs as to when blood vitamin D (VD) levels should be measured in children. Obesity and metabolic syndrome are risk factors for low VD levels and are also associated with acanthosis nigricans (AN).

Objectives:

To test whether the clinical diagnosis of AN is a strong predictor for vitamin D deficiency (VDD) in children.

Methods:

Within the study period (2015–2020), we identified 677 consecutive individuals (age <18 years) with available calcidiol measurements and compared those with (n = 273) and without (n = 404) AN. Bivariate associations and the occurrence of AN were tested using the chi-squared test. Multivariate logistic regression was performed to control for confounding variables, and adjusted odds ratios with 95% confidence intervals (CI) were reported. Multiple regression analysis was performed, and unstandardized beta coefficients, standard errors, and standardized beta coefficients were reported.

Results:

Individuals with AN had 3.6 times higher odds of VDD than those without (95% CI: 1.38–9.51, P = 0.009). Males had 0.41 times lower odds of having AN than females (95% CI: 0.21–0.79, P = 0.008). Individuals with vitamin D sufficiency (VDS) were much less likely to be diagnosed with metabolic syndrome compared with those who were vitamin D deficient (P = 0.011), even after adjusting for body mass index z-scores.

Conclusion:

Children and adolescents with AN are at a higher risk of VDD and should likely be tested for low calcidiol levels.

Introduction

Acanthosis nigricans (AN) is a common pediatric diagnosis among African American and Latin American children and adolescents and a frequent referral to pediatric clinics from school nurses. Several types of AN have been described in the literature, including those associated with obesity (most common), acral AN in people with dark skin, syndromic AN (including metabolic syndrome and fibroblast growth factor receptor defects), nonsyndromic malignancy-associated AN, familial AN, and drug-induced AN, among others.¹ Although the causes of AN are numerous, it is most commonly associated with insulin resistance.² Excessive insulin production causes overstimulation of epidermal keratinocytes and dermal fibroblast growth receptors, leading to abnormalities in cell proliferation and AN.¹

Vitamin D deficiency (VDD) is also common in children, and it is associated with obesity, diabetes, and other components of metabolic syndrome. However, screening guidelines differ as to when to measure calcidiol (25-hydroxyvitamin D [25(OH)D]) levels in children. This knowledge gap is partly explained by the lack of a clear definition of “sufficiency” or acceptable 25(OH)D levels at different stages of puberty, during growth spurts, or among individuals with obesity, especially those with the metabolic syndrome phenotype. Due to the discrepancy between different clinical guidelines as to when to measure 25(OH)D levels, it is essential to determine if there are other strong predictors for VDD.

Interestingly, VDD and AN are both associated with obesity and metabolic syndrome. While studies have shown that AN and VDD occur in the same patient population,³ very few studies have been conducted to determine if AN can predict the presence of VDD. One clinical study proposed that children and adolescents with obesity and AN had lower VD levels than those who had obesity without AN.³ However, this study was conducted in a predominately non-Hispanic White population and only included individuals with obesity. Identifying if AN is predictive across minority groups, as well as if other clinical markers can predict VDD, would be an important tool for pediatricians as they decide when VD testing should be performed. Therefore, the present study aimed to test whether the clinical diagnosis of AN is a strong predictor for VDD in children.

Materials and Methods

Patients

A total of 12,959 children and adolescents (95% African American and Latin American) were cared for at the Kids ‘N Teens Clinics P.A., a community health center in Houston, Texas (latitude 29°N), over a 5-year period (April 2015–April 2020). Most high-risk individuals, especially those with AN and obesity [body mass index (BMI) ≥95th tile or ≥30 kg/m²], were routinely recommended to undergo testing for a fasting lipid profile, blood glucose level, insulin level, liver function, and calcidiol (25(OH)D; as measured by the QuestAssureD™ test) level. Individuals who presented with significant chronic idiopathic musculoskeletal complaints, recent long bone fractures, chronic idiopathic fatigue, mood swings, or depression were also routinely recommended for 25(OH)D testing. Written informed consent was not required considering that the study individual's identity was coded and deidentified by the principal investigator. This study was approved by the Baylor College of Medicine Institutional Review Board.

Definitions

VD status was defined by 25(OH)D levels as follows: VDD, <20 ng/mL; vitamin D insufficiency (VDI), 20–29 ng/mL; and vitamin D sufficiency (VDS), ≥30 ng/mL.⁴ Although other guidelines have suggested that a 25(OH)D level of <12 ng/mL should be used to define VDD,⁵ all agree that a level <20 ng/mL is insufficient.⁵ Thus, we have used a broader definition endorsed by the Endocrine Society and frequently used to define VD status in nonbone health outcomes.^6
–8

The definition for the metabolic syndrome was modified from the original definition to include BMI z-scores (BMI adjusted for child age and sex) instead of waist circumference as a measure of obesity.⁹ Metabolic syndrome was defined as meeting at least three of the following critical phenotypes: (1)

Systolic or diastolic Blood pressure (BP) >95th percentile, >130 mmHg (systolic), or >90 mmHg (diastolic).

(2)

Obesity, BMI z-score ≥2.0, or BMI >97th percentile.

(3)

Diabetes or fasting glycaemia ≥100 mg/dL.

(4)

Oral glucose tolerance test 2-hr blood glucose of 140–200 mg/dL.

(5)

Dyslipidemia, which is, low serum levels of high-density lipoprotein (HDL) cholesterol and high serum levels of low-density lipoprotein (LDL) cholesterol.

(6)

Serum HDL cholesterol (mg/dL) <5th percentile, or <50 mg/dL (females) and <40 mg/dL (males).¹⁰

(7)

High triglycerides (mg/dL): >95th percentile or ≥150 mg/dL.¹⁰

Clinical measurement

We used a stadiometer to measure height in inches and a balanced standard mechanical scale for weight measurement of all subjects. BMI (kg/m²), its standard deviation z-scores, and percentile scores were calculated for everyone using the Center for Disease Control 2–20 years' weight and height for age data and clinical information as close to the 25(OH)D testing date as possible. The Children's BMI-percentile-for-age Calculator (https://www.bcm.edu/cnrc-apps/bodycomp/bmiz2.html) was utilized for these calculations. BPs were measured on the right arm with an appropriate cuff using the oscillometer method (Welch Allyn vital signs monitor; Welch Allyn, Inc.). BP percentiles for diastolic and systolic pressure were calculated using the Baylor College of Medicine's Age-based Pediatric BP calculator (https://www.bcm.edu/bodycomplab/BpappZjs/BPvAgeAPPz.html).

Laboratory methods

VD testing was ordered when the primary care provider judged it to be of clinical benefit (“targeted testing”) or when the subject's parents requested testing (“routine testing”). All laboratory assessments were performed using the Quest Diagnostics Laboratory in Houston, TX. We collected clinical information on the following: total blood levels of 25(OH)D (measured using the QuestAssureD vitamin D assay^11,12), lipids, insulin, glucose, alkaline phosphatase, calcium, and creatinine, as well as liver function tests. Assay methods and sensitivity in ng/mL have been previously published.^11,12 Homeostasis model assessment of insulin resistance (HOMA-IR) was calculated as follows: fasting plasma insulin level (μU/mL) x fasting plasma glucose (mM/L)/22.5.^13
–15 HOMA-IR status¹⁴ was defined as follows: normal, <2.6; elevated, 2.6–3.8; and significantly elevated, >3.8.

Statistical methods

We used chi-squared tests to compare VD status groups (deficient, insufficient, and sufficient) with categorical parameters. As a categorical variable, we categorized age into three groups to account for possible pubertal differences: <10 years, 10–14 years, and >14 years. We also compared VD status among several continuous variables, including age. For continuous variables, we compared groups using one-way analysis of variance and then, if a significant main effect was detected, we performed a post hoc test (Tukey's test).

We also used chi-squared tests to analyze hypothetical bivariate associations between age, race, sex, and VD status with the occurrence of AN. Separate multivariate logistic regression analyses were conducted to control for confounding variables in outcomes related to VD status (sufficient vs. insufficient/deficient), AN, metabolic syndrome, and low HDL. We also performed a multiple regression analysis to test for multivariate associations between AN, total VD level, age, binary season, binary race, and sex on HOMA-IR values. All analyses were performed using SPSS Version 26 (IBM Corp., Armonk, NY), and statistical significance was considered at an alpha value of 0.05.

Results

Of the 12,959 individuals cared for during the 5-year study period (2015–2020), 745 received 25(OH)D testing results. We excluded 68 (9.1%) individuals from the study whose medical records noted that they had been on VD supplements before testing and those with no BP or height measurement during the intake examination. Furthermore, individuals with disorders that are known to affect the blood VD concentration, such as those with chronic kidney disease (high creatinine), kidney transplant, liver transplant after hepatic failure, syndromes causing intestinal malabsorption, hyperparathyroidism, sarcoidosis, granulomatous diseases, and cancer, as well as those prescribed certain medications (antiepileptics, glucocorticoids, ketoconazole, or cholestyramine), were also excluded. Finally, a total of 677 (90.9%) individuals were included in the study.

Among these, VD levels were sufficient (≥30 ng/mL) in 6.9%, insufficient (20–29 ng/mL) in 30.4%, moderately deficient (10–19 ng/mL) in 52.6%, and severely deficient (0–9 ng/mL) in 10.1% individuals. Primary diagnosis and 25(OH)D test results of the study sample at an initial clinical encounter, including those of patients with and without AN diagnosis, are presented in the supporting Supplementary Table S1 for reference.

Multiple parameters were associated with VD status in our study groups, potentially confounding the association of VD status with AN (Table 1). Among the VD status groups, we found statistically significant main effects of age, race, sex, season, AN, metabolic syndrome, low HDL, BMI, BMI z-score, systolic BP, diastolic BP, calcium, and alkaline phosphatase. No significant differences between the VD status groups for high LDL or high triglycerides were found. Participants in the VDD group were significantly older than those in the VDI (P = 0.01) and VDS (P = 0.002) groups. Patients with VDD had higher BMI z-score values than those in the VDI and VDS groups (P = 0.002) and significantly higher systolic and diastolic BP values than those in the VDI group (P < 0.001 and P = 0.019, respectively). Serum calcium levels were significantly lower in the VDD group than in the VDI (P = 0.002) and VDS (P = 0.001) groups. No differences were found among the VD status groups for LDL, HDL, triglycerides, creatinine, aspartate aminotransferase, alanine aminotransferase, and hemoglobin A1C levels (Table 1).

Table 1.

Comparison for Vitamin D Status Groups

Variables/VD	Deficient (0–19 ng/mL)	Insufficient (20–29 ng/mL)	Sufficient (≥30 ng/mL)	P–
Age^a				0.001
<10 years	102 (51.5%)	79 (39.9%)	17 (8.6%)
10–14 years	195 (69.9%)	69 (24.7%)	15 (5.4%)
15–18 years	132 (66.0%)	57 (28.5%)	11 (5.5%)
Race^a				0.023
Hispanic	335 (61.2%)	174 (31.8%)	38 (6.9%)
Black	86 (75.4%)	24 (21.1%)	4 (3.5%)
Asian	2 (100.0%)	0 (0.0%)	0 (0.0%)
White	3 (27.3%)	7 (63.6%)	1 (9.1%)
Middle Eastern	3 (100.0%)	0 (0.0%)	0 (0.0%)
Sex^a				<0.001
Male	159 (52.6%)	115 (38.1%)	28 (9.3%)
Female	270 (72.0%)	90 (24.0%)	15 (4.0%)
Season^a				0.003
Spring	88 (62.9%)	42 (30.0%)	10 (7.1%)
Summer	114 (54.3%)	79 (37.6%)	17 (8.1%)
Fall	112 (62.9%)	55 (30.9%)	11 (6.2%)
Winter	115 (77.2%)	29 (19.5%)	5 (3.4%)
Acanthosis nigricans diagnosis^a				<0.001
No	234 (57.9%)	133 (32.9%)	37 (9.2%)
Yes	195 (71.4%)	72 (26.4%)	6 (2.2%)
Metabolic syndrome diagnosis^a				<0.001
No	317 (60.0%)	169 (32.0%)	42 (8.0%)
Yes	112 (75.2%)	36 (24.2%)	1 (0.7%)
Low serum HDL level^a				0.004
No	147 (60.5%)	83 (34.2%)	13 (5.3%)
Yes	201 (74.2%)	59 (21.8%)	11 (4.1%)
High serum LDL level				0.46
No	319 (68.0%)	127 (27.1%)	23 (4.9%)
Yes	21 (70.0%)	9 (30.0%)	0 (0.0%)
High serum triglyceride level				0.30
No	239 (67.7%)	97 (27.5%)	17 (4.8%)
Yes	108 (71.5%)	40 (26.5%)	3 (2.0%)
Values below are represented as mean (SD)
Age (years)^a	12.7 (3.1)	11.8 (3.5)	11.2 (4.2)	0.001
BMI (kg/m²)^a	27.4 (7.6)	24.9 (6.9)	22.4 (5.7)	<0.001
BMI z-score^a	1.55 (1.08)	1.32 (1.14)	1.00 (1.44)	0.002
Systolic BP (mmHg)^a	115.0 (12.9)	109.6 (13.5)	112.2 (11.6)	<0.001
Diastolic BP (mmHg)^a	68.8 (9.4)	66.7 (9.2)	66.5 (9.9)	0.014
LDL (mg/dL) (n = 511)	88.3 (28.0)	86.3 (25.5)	77 (21.5)	0.13
HDL (mg/dL) (n = 536)	46.1 (12.1)	47.3 (13.0)	47.0 (13.0)	0.60
Triglycerides (mg/dL) (n = 511)	115.3 (65.2)	116.4 (68.7)	86.9 (59.2)	0.14
Serum creatinine (mg/dL) (n = 407)	0.6 (0.1)	0.6 (0.2)	0.6 (0.2)	0.96
Serum calcium (mg/dL) (n = 396)^a	9.7 (0.3	9.8 (0.3)	9.9 (0.4)	<0.001
Serum AST (U/L) (n = 431)	22.4 (16.9)	23.5 (17.3)	17.6 (11.4)	0.64
Serum ALT (U/L) (n = 423)	26.3 (37.6)	25.4 (42.0)	17.6 (11.4)	0.55
Hemoglobin A1C (%) (n = 230)	5.4 (0.6)	5.4 (0.4)	5.2 (0.2)	0.64
Serum alkaline phosphatase (U/L) (n = 294)^a	188.4 (93.2)	207.5 (92.4)	242.6 (107.0)	0.04

Statistically significant.

ALT, alanine aminotransferase; AST, aspartate aminotransferase; BMI, body mass index; BP, blood pressure; HDL, high-density lipoprotein cholesterol; LDL, low-density lipoprotein cholesterol; SD, standard deviation; VD, vitamin D.

We also determined parameters associated with AN (Table 2). The group with AN had significant difference in terms of age, VD status, and BMI z-scores. The highest rates of AN were found in individuals 14–18 years of age and those with VDD. There were no differences among the groups when considering race or sex.

Table 2.

Comparison of Acanthosis Nigricans Groups

Variable	AN (absent)	AN (present)	P–
Age^a			0.025
<10 years	128 (64.6%)	70 (35.4%)
10–14 years	172 (61.6%)	107 (38.4%)
14–18 years	104 (52.0%)	96 (48.0%)
Race
Hispanic	324 (59.2%)	223 (40.8%)	0.37
Black	67 (58.8%)	47 (41.2%)
Asian	2 (100.0%)	0 (0.0%)
White	8 (72.7%)	3 (27.3%)
Other	3 (100.0%)	0 (0.0%)
Sex			0.51
Male	176 (58.2%)	126 (41.7%)
Female	228 (60.8%)	147 (39.2%)
VD status^a			<0.001
Deficient (0–19 ng/mL)	234 (57.9%)	195 (71.4%)
Insufficient (20–29 ng/mL)	133 (32.9%)	72 (26.4%)
Sufficient (≥30 ng/mL)	37 (9.2%)	6 (2.2%)
BMI z-score (mean/SD)^a	1.17 (0.06)	2.20 (0.43)	<0.001

Statistically significant.

AN, acanthosis nigricans.

Given the multiple factors associated with AN, we determined if the presence of AN was a significant predictor of VDI or VDD. Indeed, after adjusting for age, race, season (winter/spring vs. summer/fall), sex, and BMI z-scores, we found that those with AN had a 3.62 times higher risk of having VDD/VDI (adjusted odds ratio (AOR): 3.62; 95% confidence intervals (CIs): 1.38–9.51, P = 0.009] (Table 3). Despite AN being associated with BMI z-scores, BMI and BMI z-scores were not independent predictors of VDS versus VDI/VDD.

Table 3.

Logistic Regression Model Parameters

Outcome	Predictors	Level	AOR (95% CI)	P–
VD	Age	—	1.076 (0.98–1.18)	0.111
Sufficient versus insufficient+deficient	Race	Non-BlackBlack	Referent1.82 (0.63 − 5.30)	0.271
	Season	Summer/FallWinter/Spring	Referent0.67 (0.34–1.32)	0.252
	Sex^a	Female	Referent
	Sex^a	Male	0.41 (0.21–0.79)	<0.001
	AN^a	No	Referent
	AN^a	Yes	3.62 (1.38 − 9.51)	0.009
	BMI	BMI z-score	1.14 (0.87–1.49)	0.335
AN	Age^a	—	1.17 (1.09–1.26)	<0.001
	Race	Non-Black	Referent
	Race	Black	0.75 (0.43–1.3)	0.305
	Season	Summer/Fall	Referent
	Season	Winter/Spring	1.10 (0.73–1.66)	0.643
	Sex	Female	Referent
	Sex	Male	0.68 (0.45–1.04)	0.078
	VD status^a	Deficient^a	Referent	0.036
		Insufficient	0.86 (0.55 − 1.36)	0.528
		Sufficient^a	0.24 (0.07–0.71)	0.010
	BMI^a	BMI z-score	11.02 (7.24 − 16.77)	<0.001
Metabolic syndrome	Age^a	—	1.09 (1.01–1.17)	0.018
	Race	Non-Black	Referent
	Race	Black	0.61 (0.34–1.10)	0.102
	Season	Summer/Fall	Referent
	Season	Winter/Spring	0.80 (0.50–1.19)	0.248
	Sex	Female	Referent
	Sex	Male	0.80 (0.51–1.26)	0.335
	VD status^a	Deficient^a	Referent	0.026
		Insufficient	0.76 (0.47–1.25)	0.285
		Sufficient^a	0.03 (0.02–0.46)	0.011
	BMI^a	BMI z-score^a	9.78 (5.84 − 16.34)	<0.001
Low HDL	Age	—	1.04 (0.97–1.02)	0.23
	Race^a	Non-Black	Referent
	Race^a	Black	0.44 (0.26–0.75)	0.002
	Season	Summer/Fall	Referent
	Season	Winter/Spring	1.05 (0.71–1.56)	0.805
	Sex^a	Female	Referent
	Sex^a	Male	0.28 (0.18–0.42)	<0.001
	VD status	Deficient	Referent	0.135
		Insufficient	0.66 (0.42–1.03)	0.065
		Sufficient	1.26 (0.49–3.24)	0.636
	BMI^a	BMI z-score	2.32 (1.82 − 2.96)	<0.001

Statistically significant.

AOR, adjusted odds ratio; CI, confidence interval; HDL, high-density lipoprotein.

For AN, VDS status, BMI z-score, and age were statistically significant predictors, whereas race, season, and sex were not (Table 3). Individuals in the VDS group had 0.20 times lower odds of having AN when compared with those in the VDD group (P = 0.010) after adjusting for potential confounders. VDI status (P = 0.528) was not a significant predictor of AN when compared with VDD (Table 3).

Our logistic regression model without adjusting for BMI z-scores suggested that VDI individuals were 62% less likely than VDD individuals to have lower HDL. However, after adjusting for BMI z-scores in the model, we found that VDI was a weak predictor (P = 0.06), and VDS was not a significant predictor. This suggests that obesity (high BMI z-scores) is the strongest risk factor (AOR: 2.3; 95% CI: 1.82 − 2.96, P < 0.001) for low HDL levels (Table 3).

In our study, metabolic syndrome was significantly associated with VD status, age, and BMI z-scores (Table 3). However, race, season, sex, and VDI status were not significant predictors. Importantly, individuals in the VDS group were less likely to have metabolic syndrome when compared with those in the VDD group (P = 0.011), even after adjusting for BMI z-scores.

Since AN has been linked to insulin resistance, we tested whether AN and VD levels were significant predictors for HOMA-IR and insulin levels. Only 187 individuals had appropriate laboratory results for HOMA-IR calculation. Thus, AN was a significant predictor of HOMA-IR as well as insulin levels (Supplementary Table S2). This association between AN and HOMA-IR was lost in a further sensitivity analysis when BMI-z scores were added (resulting in neither AN nor BMI z-scores being significant independent predictors of HOMA-IR). These data indicate that the effects of AN on HOMA-IR are likely influenced, partly, by BMI-z score status. Black race was a significant predictor for HOMA-IR but not for insulin levels. Lower VD levels did not predict higher HOMA-IR scores or insulin levels (Supplementary Table S2).

Discussion

VDD, as determined through the measurement of 25(OH)D levels, is highly prevalent in our society. In a study of 15,787 adults in the USA, 12.7% were found to have VDD (<20 ng/mL).⁸ The 2006 National Health and Nutrition Examination Survey determined that 18% of children 1–11 years of age had VDD (<20 ng/mL).¹⁶ In a recent study conducted in Dallas, Texas (USA-latitude 34°N), where adequate sun exposure is not a problem, ∼50% of the participants were found to have VDD. Additionally, this study noted that VDD was more clinically significant among African American and Latin American participants.¹⁷ In our study, individuals without AN had a slightly higher prevalence of VDD than the Dallas cohort (57.7% vs. 50%), but those with AN had a much higher prevalence (71.4%). However, global testing for VD is not recommended.¹⁸ Thus, determining possible clinical markers to guide testing would be useful.

There are a few clinical studies that consider AN as a phenotypic predictor of VDD. This is clinically important considering that there is a much higher prevalence of VDD among people with obesity; hence, determining the target population for VD testing would be useful. A previous study showed that AN was associated with lower VD levels but did not determine if AN truly predicted VDI or VDD.³ Our results showed that the clinical diagnosis of AN significantly predicts VDD and thus suggests that the presence of AN may be a proxy for VDD. Only 6 (2.2%) individuals diagnosed with AN had a sufficient VD level (≥30 ng/mL). Even with using a 25(OH)D level of 20 ng/mL to define sufficiency, almost three-quarters of individuals with AN had VDI in our study. This is not secondary to the presence of obesity or race because after adjusting for potential confounders, including BMI-z scores, those with AN were 3.6 times more likely to have VDD or insufficiency than those without AN.

As expected, individuals with darker skin color were also more likely to be VDD or insufficient. A small study of 150 individuals with obesity showed that AN was present in 59.3%, and of those 89 individuals with obesity and AN, 39.3% also had VDD.¹⁹ Our study has a much higher prevalence of VDD with AN likely due to the different racial composition. While our study consisted of mostly individuals of darker skin complexity, the former study consisted of mostly white adolescent individuals.

The association of VDD with obesity^17,19 and insulin resistance^3,19 has been reported in the literature. Williams et al¹⁹ found that “hyperinsulinism” was a statistically significant covariate predictor of VDD (<20 ng/mL). However, high insulin levels were a much weaker predictor of VDD (OR = 1.04, P = 0.01) when compared with race (OR = 4.96, P = < 0.01), season of the year (OR = 3.69, P = 0.01), and location (OR = 5.91, P = 0.05).¹⁸ AN has also been reported in the literature to be an important phenotypic expression of hyperinsulinism,^20
–22 although this relationship appears to be complex and likely mediated by the presence of obesity in children and adolescents.²¹ Interestingly, a study found that HOMA-IR could only predict variations in 25(OH)D levels when AN was included in the model.³ Similar to Soliman et al,²⁰ we found that the presence of AN was a strong predictor of higher HOMA-IR scores and higher insulin levels. While this significance partly depends on BMI z-scores,²¹ it is likely that we were underpowered to detect independent associations of both AN and BMI z-scores in the same model, given the smaller sample size for HOMA-IR measurements.

Evidence exists to suggest a causal relationship between AN and VDD. VD receptors (VDR) are found in the skin as well as other multiple organs, including the liver, kidney, and pancreas. VD is known to have anti-inflammatory effects by modulating the function of T and B lymphocytes,²³ which can augment insulin action in the cells and regulate brain neuronal excitability through the insulin-sensitive phosphoinositide 3-kinase pathway.²⁴ Moreover, it has been suggested that 1,25-dihydroxyvitamin D (the most potent endogenous VDR ligand) and the VDR promote gene signals that inhibit keratinocyte proliferation and stimulate the differentiation (keratinization) of basal keratinocytes.²⁵ Taken together, the literature suggests that low VD levels may promote excessive keratinocyte proliferation, causing the appearance of AN, which could be worsened in the presence of excess insulin signaling.

The cross-sectional design of our study did not permit us to look at AN severity as a predictor of VDD and/or lower VD levels. A previous study suggests that the severity of AN predicts lower 25(OH)D levels,³ but since they did not categorize by VD status, it is not clear which level of AN severity would help to predict VDD. A large prospective study utilizing validated severity scores is needed to further address this question.

Our clinical practice of targeting VD testing to high-risk subjects uncovered a high prevalence of VDD (62.6%). Interestingly, there were very few subjects with VDD in those tested for routine reasons, validating our clinical approach. Demographic variables, age (as a proxy for pubertal status), obesity, and time of the year when the VD D test was conducted,³ are also important variables that were adjusted for in the final analysis. We found that the prevalence rate of VDD was much higher in winter than in summer. Additionally, children 10–14 years of age were also at higher risk of VDD/VDI. It is conceivable that pubertal changes in these individuals may be contributing to their pathology; however, since we did not have accurate sexual maturation ratings (Tanner staging) at each visit, we could not test this hypothesis. Furthermore, we detected high levels of VDD in our younger age group, which likely reflects their increased risk due to obesity and darker skin.

There is controversy regarding the clinical diagnosis of metabolic syndrome among children and adolescents. We modified existing criteria using a more practical clinical parameter (BMI z-scores ≥2) as a “proxy” for abdominal obesity, since waist circumference is not a routine vital sign in our clinic. This methodology, although not validated for the diagnosis of metabolic syndrome, is practical, simple, and may be useful for future pediatric obesity research. Our adjusted logistic regression model suggests that children and adolescents with VDD were more likely to have metabolic syndrome than those with VDS. Since this study was observational, we could not test whether VDD causes the development of metabolic syndrome (or vice versa). Obesity is an important phenotype of metabolic syndrome ²⁶ and has also been inversely associated with blood levels of bioavailable VD.²⁷

Similarly, in our study, individuals in the VDD group had higher BMI z-scores than those in the VDS and VDI groups. Although there are many theories as to why VDD is more likely to occur as BMI values increase, there is no definitive explanation. Despite the association of low VD levels with increasing BMI, we did not observe associations of low VD with abnormal lipid, liver, or glucose parameters. We found an association (independent of BMI) between low VD levels and higher BP readings, but this analysis cannot explain if this relationship is causal or not. Several studies have suggested an important association between VDD and metabolic syndrome in adults,^28
–30 and an inverse association between VD levels and metabolic syndrome.^29,30 Thus, further study is required to understand the relationship between these disorders.

Overall, our results suggest that AN is a reliable proxy marker for VDD, especially in minority young individuals. However, there are many interesting unresolved clinical questions pertaining to our research. Further study is required to explore whether abnormalities in lipoprotein metabolism (low HDL phenotype) contribute to low VD levels, possibly through sequestration in the adipose tissue. Additionally, causal links and possible therapeutic uses of VD for other abnormal parameters, such as BP, obesity, AN, and lipids in children and adolescents with metabolic syndrome, still need to be elucidated.

Study limitations and strengths

Our study was observational and thus could not prove causality. Additionally, our population was skewed toward high-risk groups, especially those with obesity (more likely to have insulin resistance) and darker skin color. Thus, our study may not be generalizable to the entire US population. However, our study targeted the most at-risk populations for VDD. Due to the retrospective and cross-sectional nature of our research, we were unable to document sunlight exposure, skin phototypes, severity of AN, accurate diet history, or adherence to VD supplementation. Additionally, as noted above, we were unable to obtain other laboratory values, such as parathyroid hormone levels. Although we used standard measurement practices for vital signs, no interobserver variability was measured, and thus minor vital sign observer bias cannot be excluded. On the other hand, subject selection bias was reduced by including all subjects who underwent a 25(OH)D test during the 5-year study period.

Conclusions

Our results suggest that obesity-associated AN appears to be a reliable proxy for VDD. Future prospective studies to confirm these findings and determine whether VD treatment in patients with AN would improve the AN itself and/or its related comorbidities are required.

Footnotes

Acknowledgments

Kids ‘N Teens Clinics PA (www.kidsandteens.org) provided financial and material support. E.R.H., PhD, PStat Accredited Professional Statistician, owner, and Operator, Scalë, LLC performed the statistical analysis and guided interpretation (https://www.scalestatistics.com). A senior Science editor and a managing editor of Editage (www.editage.com) edited the final article and completed an extensive plagiarism check (DOISA_1_Plagiarism_Check).

Authors' Contributions

F.A.I. conceived and designed the study, acquired funding, administered the project, and drafted the article. F.J.I.-I. participated in the conception and design, data acquisition, and managed resources. E.R.H. performed the analysis and interpretation of data. S.S. participated in the conception and design of the study, revised the article for key critically important intellectual content, and mentored coauthors. All authors participated in the article drafting, editing, and approval processes.

Author Disclosure Statement

S.S. receives speaking and consulting fees from Rhythm Pharmaceuticals. F.A.I., F.J.I.-I., and E.R.H. report no conflicts of interest.

Funding Information

S.S. receives salary support in part by agreement no. 58-3092-0-001 of the United States Department of Agriculture (USDA), Agricultural Research Service, and from the American Diabetes Association grant no. 1-17-JDF-037.

Supplementary Material

Supplementary Table S1

Supplementary Table S2

References

Miller

JH.

Acanthosis nigricans. Medscape; 2020. Available from: https://emedicine.medscape.com/article/1102488-overview#a6 [Last accessed: June 29, 2022].

Higgins

, Freemark

, Prose

. Acanthosis nigricans: A practical approach to evaluation and management. Dermatol Online J, 2008; 14(9):2; doi:10.5070/D37mf6g290.

Slyper

, Kashmer

, Huang

, et al. Acanthosis nigricans, vitamin D, and insulin resistance in obese children and adolescents. J Pediatr Endocrinol Metab, 2014; 27(11–12):1107–1111; doi: 10.1515/jpem-2013-0465.

Ross

, Manson

, Abrams

, et al. The 2011 report on dietary reference intakes for calcium and vitamin D from the Institute of Medicine: What clinicians need to know. J Clin Endocrinol Metab, 2011; 96(1):53–58; doi: 10.1210/jc.2010-2704.

Munns

, Shaw

, Kiely

, et al. Global consensus recommendations on prevention and management of nutritional rickets. J Clin Endocrinol Metab, 2016; 101(2):394–415; doi:10.1210/jc.2015-2175.

Zhang

, Tan

, Tang

, et al. Supplementation on prevention of type 2 diabetes in patients with prediabetes: A systematic review and meta-analysis. Diabetes Care, 2020; 43(7):1650–1658; doi:10.2337/dc19-1708.

Manson

, Cook

, Lee

, et al. Vitamin D supplements and prevention of cancer and cardiovascular disease. N Engl J Med, 2019; 380(1):33–44; doi: 10.1056/NEJMoa1809944.

Ross

, Taylor

, Yaktine

, et al. Dietary Reference Intakes for Calcium and Vitamin D. Institute of Medicine (US) Committee to Review Dietary Reference Intakes for Vitamin D and Calcium. National Academies Press (US): Washington (DC), 2011. PMID: 21796828. doi:10.17226/13050.

Weiss

, Bremer

, Lustig

. What is metabolic syndrome, and why are children getting it?. Ann NY Acad Sci, 2013; 1281(1):123–140; doi: 10.1111/nyas.12030.

10.

Skinner

, Steiner

, Chung

, et al. Cholesterol curves to identify population norms by age and sex in healthy weight children. Clin Pediatr (Phila), 2012; 51(3):233–237; doi: 10.1177/0009922811430344.

11.

Clarke

NJ.

Vitamin D testing. Increasing Accuracy, Improving Patient Management. Clinical Education Center, Quest-Nichols Institute: San Juan Capistrano, CA. Available from: www.questdiagnostics.com [Last accessed: June 29, 2022].

12.

Chen

, McCoy

, Schleicher

, et al. Measurement of 25-hydroxyvitamin D3 (25OHD3) and 25-hydroxyvitamin D2 (25OHD2) in human serum using liquid chromatography-tandem mass spectrometry and its comparison to a radioimmunoassay method. Clin Chem Acta, 2008; 391(1–2):6–12; doi: 10.1016/j.cca.2008.01.017.

13.

, Li

, Rentfro

, et al. The definition of insulin resistance using HOMA-IR for Americans of Mexican descent using machine learning. PLoS One, 2011; 6(6):e21041; doi: 10.1137/journal.pone.0021041.

14.

Matthews

, Hosker

, Rudenski

, et al. Homeostasis model assessment: Insulin resistance and β-cell function from fasting plasma glucose and insulin concentrations in man. Diabetologia, 1985; 28(7);412–419; doi: 10.1007/bf00280883.

15.

Hill

, Levy

, Matthews

. Expansion of the homeostasis model assessment of the β-cell function and insulin resistance to enable clinical trial outcome modeling through the interactive adjustment of physiology and treatment effects: iHOMA2. Diabetes Care, 2013; 36(8):2324–2330; doi: 10.2337/dc12-0607.

16.

Mansbach

, Ginde

, Camargo

. Serum 25-hydroxyvitamin D levels among US children aged 1 to 11 years: Do children need more vitamin D?. Pediatrics, 2009; 124(5):1404–1410; doi:10.1542/peds.2008-2041.

17.

Olson

, Naim

, Oden

, et al. Vitamin D deficiency in obese children and its relationship to glucose homeostasis. J Clin Endocrinol Metab, 2012; 97(1):279–285; doi: 10.1210/jc.2011-1507.

18.

Vitamin

Testing in the General Population: A Review of the Clinical and Cost-Effectiveness and Guidelines [Internet]. Canadian Agency for Drugs and Technologies in Health: Ottawa (ON); 2015. Summary of Evidence. Available from: https://www.ncbi.nlm.nih.gov/books/NBK274110/ [Last accessed: June 29, 2022].

19.

Williams

, Novick

, Lehman

. Prevalence of hypovitaminosis D and its association with comorbidities of childhood obesity. Perm J Fall, 2014; 18(4):32–39; doi:10.7812/TPP/14-016.

20.

Soliman

, Al-suwaid

, Gamil

, et al. Prevalence and significance of Acanthosis nigricans in children and adolescents. Ann Saudi Med, 1996; 16(4):424–428; doi: 10.5144/0256-4947.1996.424.

21.

Lopez-Alvarenga

, Chittoor

, Paul

SFD

, et al. Acanthosis nigricans as a composite marker of cardiometabolic risk and its complex association with obesity and insulin resistance in Mexican American children. PLoS One, 2020; 15(10):e0240467; doi: 10.1371/1/journal.pone.0240467.

22.

Koh

, Lee

, Kim

, et al. Acanthosis nigricans as a clinical predictor of insulin resistance in obese children. Pediatr Gastroenterol Hepatol Nutr, 2016; 19(4):251–258; doi:10.5223/pghn.1016.19.4.251.

23.

Misra

, Pacaud

, Petryk

, et al. Vitamin D deficiency in children and its management: Review of current knowledge and recommendations. Pediatrics, 2008; 122(2):398–417; doi: 10.1542/peds.2007-1894.

24.

Da Silva Teixeira

, Harrison

, Uzodike

, et al. Vitamin D actions in neurons require the PI3K pathway for both enhancing insulin signaling and rapid depolarizing effect. J Steroid Biochem Mol Biol, 2020; 200:105690; doi: 10.1016/j.jsbmb.2020.105690.

25.

Bikle

. Vitamin D metabolism and function in the skin. Mol Cell Endocrinol, 2011; 347(1–2):80–89; doi: 10.1016/j.mec.2011.05.017.

26.

Weiss

, Dziura

, Burget

, et al. Obesity and the metabolic syndrome in children and adolescent. N Engl J Med, 2004; 350:2362–2374; doi: 10.1056/NEJMoa031049.

27.

Cediel

, Corvalán

, de Romaña

, et al. Prepubertal adiposity, vitamin D status and insulin resistance. Pediatrics, 2016; 138(1):e20160076; doi: 10.1542/peds.2016-0076.

28.

Theik

NWY

, Raji

, Shenwai

, et al. Relationship and effects of vitamin D on metabolic syndrome: A systematic review. Cureus, 2021; 13(8):e17419; doi: 10.7759/cureous.17419.

29.

Hajhashemy

, Shahdadian

, Moslemi

, et al. Serum vitamin D levels in relation to metabolic syndrome: A systematic review and dose-response meta-analysis of epidemiologic studies. Obes Rev, 2021; 22(7):e13223; doi: 10.1111/obr.13223.

30.

Fite

, Wang

, Seyoum

. Vitamin D deficiency is associated with metabolic syndrome. J Diabetes Complicat, 2020; 1:1–8; doi: 10.3148/7/j.JDMC.2020.01.03.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.01 MB