- Email: [email protected]
Relation of acanthosis nigricans to hyperinsulinemia and insulin sensitivity in overweight African American and white children
Relation of acanthosis nigricans to hyperinsulinemia and insulin sensitivity in overweight African American and white children Tuc T. Nguyen, MS, Margaret F. Keil, RN, NP, Deserea L. Russell, MS, Anuttara Pathomvanich, MD, Gabriel I. Uwaifo, MD, Nancy G. Sebring, MEd, RD, James C. Reynolds, MD, and Jack A. Yanovski, MD, PhD Objectives: Acanthosis nigricans (AN) has been proposed as a reliable marker of hyperinsulinemia, but its utility for predicting hyperinsulinism has not been systematically evaluated in overweight children. We examined the relationship of AN to hyperinsulinemia and body adiposity. Study design: One hundred thirty-nine children underwent physical examination for AN, body composition studies, an oral glucose tolerance test, and a hyperglycemic clamp. Results: Thirty-five children (25%) had AN. AN was more prevalent in African Americans (50.1%) than in white subjects (8.2%, P < .001). Independent of race, children with AN had greater body weight and body fat mass (P < .001); greater basal and glucose-stimulated insulin levels during oral glucose tolerance test (P < .001); greater first-phase, second-phase, and steady-state insulin levels (P < .001); and lower insulin sensitivity (P < .001) during the hyperglycemic clamp. After adjusting for body fat mass and age, none of these differences remained significant. When categorized by fasting insulin, 35% with fasting insulin levels >20 µU/mL and 50% with fasting insulin levels >15 µU/mL did not have AN. Eighty-eight percent of children with fasting insulin levels ≥15 µU/mL had a body mass index SD score ≥3.0. Conclusions: AN is not a reliable marker for hyperinsulinemia in overweight children. Children with a race-, sex-, and age-specific body mass index SD scores ≥3.0 should be screened for hyperinsulinemia, whether or not they have AN. (J Pediatr 2001;138:474-80)
Acanthosis nigricans is a skin lesion characterized by hyperpigmentation and velvety thickening that occurs on
the neck, axillae, and other skinfolds. Hyperinsulinemia appears to predispose individuals to AN; patients with
From Unit on Growth and Obesity, Developmental Endocrinology Branch, National Institute of Child Health and Human Development, National Institutes of Health (NIH); Nutrition Department, Warren Grant Magnuson Clinical Center, NIH; and Nuclear Medicine Department, Warren Grant Magnuson Clinical Center, NIH, Bethesda, Maryland. Dr Yanovski is supported by the Office of Research on Minority Health, NIH and NIH HD-00641. Dr Yanovski and N. Sebring are commissioned officers in the US Public Health Service. Submitted for publication July 20, 2000; revision received Sept 26, 2000; accepted Oct 26, 2000. Reprint requests: Jack A. Yanovski, MD, PhD, Unit on Growth and Obesity, National Institutes of Health, Building 10, Room 10N262, MSC 1862, 10 Center Dr, Bethesda, MD 20892-1862.
9/21/112657 doi:10.1067/mpd.2001.112657
474
inactivating insulin receptor mutations often have severe AN.1,2 Because the presence of AN is commonly associated with hyperinsulinemia3-6 and obesity,4-6 the use of AN as a marker of hyperinsulinemia and insulin resistance has been proposed for the early detection and prevention of type 2 diabetes.3,4,6,7
See related articles, p 453 and p 469. A few studies have suggested that adults with AN have higher fasting plasma insulin levels than non-AN cohorts,3,4,6 with one report showing that the width of the band of AN in the posterior neck is positively related to fasting insulin levels.3 The prevalence of AN is also disproportionately greater in groups with a high prevalence of type 2 diabetes, such as African Americans, Native Americans, and Hispanics compared with white and Asian subjects.4,7,8 AN BMI DXA IGF-I OGTT
Acanthosis nigrans Body mass index Dual-energy x-ray absorptiometry Insulin-like growth factor I Oral glucose tolerance test
Few studies have examined the association of AN with hyperinsulinism in children or adolescents.4,5,8 Stuart et al5 reported that 66% of 1412 adolescents with a body weight >200% of ideal had AN. In the subset with AN in whom insulin measurements were made, insulin levels were high.5 However, insulin measurements were ob-
NGUYEN ET AL
THE JOURNAL OF PEDIATRICS VOLUME 138, NUMBER 4 Table I. Clinical characteristics
African Americans
Sex Age (y) Bone age (y) Weight (kg) Height (cm) BMI (kg/m2) BMI SD score DXA body fat mass (kg) Hollingshead socioeconomic status score Boys’ testis size (cc) Girls’ breast stage (%) I II III Pubic hair stage (%) I II III IV
White subjects
AN+ (n = 28)
AN– (n = 27)
AN+ (n = 7)
AN– (n = 77)
8 M, 20 F 8.8 ± 0.2 10.9 ± 0.4† 61.7 ± 4.3‡ 142.2 ± 2.2* 29.6 ± 1.3‡ 5.4 ± 0.5‡ 28.6 ± 13.6‡ 3.8 ± 0.2 2.1 ± 0.2
11 M, 16 F 8.2 ± 0.3 9.2 ± 0.4 41.0 ± 2.8 133.9 ± 2.1 22.3 ± 1.0 2.7 ± 0.4 16.6 ± 10.8 3.7 ± 0.1 2.1 ± 0.4
4 M, 3 F 9.7 ± 0.3* 12.4 ± 0.3‡ 70.6 ± 2.1‡ 149.0 ± 3.0† 31.8 ± 0.5‡ 5.0 ± 0.5‡ 34.9 ± 3.0‡ 4.3 ± 0.2 2.3 ± 0.9
33 M, 34 F 8.4 ± 0.2 9.3 ± 0.2 41.0 ± 1.6 134.8 ± 1.2 22.1 ± 0.6 2.2 ± 0.2 16.4 ± 10.6 4.1 ± 0.1 2.3 ± 0.2
7 (35) 8 (40) 5 (25)
8 (50) 7 (44) 1 (6)
1 (33) 1 (33) 1 (33)
16 (47) 16 (47) 2 (6)
5 (18)† 16 (57) 5 (18) 2 (7)
18 (66) 5 (19) 4 (15) 0 (0)
5 (71) 1 (14) 0 (0) 1 (14)
65 (84) 7 (9) 4 (6) 1 (1)
*P < .05, †P < .005, ‡P < .001 for comparisons between African Americans with and without AN or between white subjects with and without AN. BMI-SD score, Body mass index expressed as the SD score for age, sex, and race.28
tained in only 12 subjects with AN, so that the strength of the relationship of AN to hyperinsulinemia in childhood remains unclear. Although children with AN are often overweight, there have been no large studies of the relationship of AN to hyperinsulinemia or insulin sensitivity that have attempted to separate the roles of body adiposity and AN as markers of insulin resistance. Therefore we studied body adiposity, fasting insulin levels, and insulin sensitivity in overweight African American and white children to delineate the nature of the relationship between AN and hyperinsulinemia.
METHODS Subjects We studied 139 African American and white children (aged 6-10 years, 73 girls and 66 boys) who were selected to have a body mass index greater than the 85th percentile for age, sex,
and race, as determined by the First National Health and Nutrition Examination Survey9 (Table I). Subjects were recruited through notices mailed to 6- to 10-year-old children in the Montgomery and Prince Georges County, Maryland, school districts, through referrals from local physicians, and through advertisements placed in local newspapers for a natural history study of body weight. Race was self-reported by each subject. African American and white subjects were selected to have all 4 grandparents of the same race. All subjects had normal findings on physical examination and normal hepatic, renal, and thyroid function. Subjects were medication-free for at least 2 weeks before the start of the study, and all were free of significant medical disease. The study was approved by the National Institutes of Health Intramural Clinical Research Subpanel. Each subject gave written assent and a parent gave
written consent for participation in the study.
Protocol All subjects were studied at the National Institutes of Health (NIH) Warren Grant Magnuson Clinical Center on 2 occasions. At the initial visit, each subject underwent anthropometric measurements and an oral glucose tolerance test, 1.75 g glucose/kg (maximum 75 g), after an overnight fast. Levels of glucose and insulin were measured at 0 hour and 2 hours after glucose ingestion. Glucose assays were performed with standard calorimetric methods on a Hitachi 736-30 analyzer (Boehringer Mannheim, Inc, Indianapolis, Ind), and insulin was measured by radioimmunoassay (Covance, Vienna, Va). Insulin resistance was estimated by the homeostasis model assessment for insulin resistance (HOMA-IR) index by using the following equation10: 475
NGUYEN ET AL
HOMA – IR = [Fasting insulin (µU/L) × Fasting glucose (mmol/L)/22.5] Measurements of weight, height, skinfolds, and waist, abdomen, and hip circumferences were obtained with the use of standardized techniques.11 Weight was measured to the nearest 0.1 kg by using a calibrated digital scale (Scale-Tronix, Wheaton, Ill). Height was measured 3 times to the nearest 1 mm by using a stadiometer calibrated before each height measurement (Holtain Ltd, Crymych, Wales). Waist, abdomen, and hip circumferences were measured by using a flexible non-elastic tape measure. Waist was measured at the narrowest part of the torso or just below the 12th rib, abdomen was measured at the maximum circumference or just above the iliac crest, and hips were measured around the buttocks at the maximum extension. Roentgenograms of the left hand and wrist were obtained to assess bone maturation.12 Each child was examined for the presence or absence of AN on the neck, axillae, and antecubital and intertriginous regions by a trained nurse practitioner or pediatric endocrinologist and assigned to one of two groups: those having acanthosis nigricans (AN+) and those without acanthosis nigricans (AN–). Concordance between raters for the presence or absence of AN was 100%. The second visit occurred 4 to 6 weeks later, at which time subjects underwent dual energy x-ray absorptiometry and a hyperglycemic clamp study. Body composition analysis by DXA was obtained by using the Hologic QDR2000 instrument in the pencil beam mode (software version 5.64), as previously described.13 For the clamp study, 114 subjects were studied after an overnight fast. The plasma glucose level was acutely increased during a 2-minute period to approximately 225 mg/dL by bolus infusion of 25% dextrose and then maintained by continuous infusion of 20% dextrose 476
THE JOURNAL OF PEDIATRICS APRIL 2001 for 120 minutes. Plasma glucose levels were measured every 2.5 minutes for the first 15 minutes and then every 5 minutes until the end of the study. Insulin levels were measured every 5 minutes for the first 15 minutes and then every 15 minutes from 15 to 120 minutes. Plasma glucose levels were measured by using a glucose analyzer (Yellow Springs Instrument, Yellow Springs, Ohio), calibrated to within 5% of multiple glucose standards (50, 100, 125, 150, 250, and 500 mg/dL) before each study. Plasma insulin levels were measured by radioimmunoassay (Covance). First-phase insulin was calculated as the mean of measurements obtained during the first 15 minutes, second-phase insulin as the mean of the measurements from 15 to 120 minutes, and steady-state insulin (I) as the mean of the measurements obtained from 60 to 120 minutes. Whole-body glucose uptake was estimated as the metabolic rate (M), defined as the infusion rate of exogenous glucose administered, corrected for urinary glucose losses and the glucose space correction.14 As a measure of insulin sensitivity, the ratio of metabolic rate to steady-state insulin (M/I) was calculated.
the log of DXA body fat mass and chronological age as covariates to determine whether the relation between the presence of AN and hyperinsulinemia (or other variables) was independent of body composition and age. Least-squares adjusted means are reported and were compared in order to determine the significance of results after adjusting for DXA body fat mass. The receiver operating characteristics15 of AN, BMI SD score, DXA total body fat, and DXA percentage body fat for the diagnosis of fasting hyperinsulinemia were also determined by using the RuleMaker program (Digital Medicine, Inc, Hanover, NH). Sensitivity, specificity, positive and negative predictive values, and diagnostic accuracy were calculated for each test variable.16 For each variable, the criteria with the highest specificity, given 100% sensitivity for an insulin level ≥15 µU/mL, were identified and used for between-test comparisons. For all analyses, significance was adjusted by using the Bonferroni correction for multiple comparisons as appropriate. Values are reported as means ± SD unless otherwise indicated.
Data Analysis
RESULTS
Parametric data were analyzed with SuperAnova 1.11 and StatView 4.5 software (Abacus Concepts, Inc, Berkeley, Calif). Results were first analyzed by using analysis of variance with race and AN as factors to detect differences between those with and those without AN. Although African Americans were more likely to have AN (see results), there were no race × AN status interactions for the measures studied once age was included as a covariate in the model. Including or excluding race as a factor in analyses did not alter the direction or the significance of differences between those with and those without AN in any case. Therefore race was not included in the presented analyses. Analysis of covariance was then performed with
Of the 139 study participants, 55 were African American and 84 were white (Table I). AN+ and AN– subjects were well-matched for socioeconomic status, as assessed by the Hollingshead Scale,17 and consisted of subjects with varying pubertal and adrenarchal development. AN+ African American and white subjects had significantly greater bone age and weight and correspondingly greater BMI and BMI SD scores than AN– subjects. Although neither breast Tanner stage nor testicular volume was significantly different in African American or white AN+ and AN– children and pubic hair stage was similar in AN+ and AN– white children, AN+ African American children had greater
NGUYEN ET AL
THE JOURNAL OF PEDIATRICS VOLUME 138, NUMBER 4 Table II. Body composition and 2-hour OGTT in all subjects and hyperglycemic clamp study in 114 subjects
Unadjusted
Body composition Waist (cm) Abdomen (cm) Hip (cm) Triceps skinfold (mm) Biceps skinfold (mm) Subscapular skinfold (mm) Supra-iliac skinfold (mm) Sum of 4 skinfolds (mm) OGTT Basal glucose (mg/dL) Two-hour glucose (mg/dL) Basal insulin (µU/mL) Two-hour insulin (µU/mL) Basal insulin/glucose ratio HOMA-IR index Hyperglycemic clamp First-phase insulin (µU/mL) Second-phase insulin (µU/mL) Steady-state insulin (µU/mL) Basal C peptide (ng/mL) Steady-state C peptide (ng/mL) Basal insulin/C peptide ratio Steady-state insulin/C peptide Metabolic rate (mg/kg/min) Insulin sensitivity (mg/kg/min/µU/mL)
Adjusted for DXA fat mass and age
AN+
AN–
AN+
AN–
86.2 ± 2.4* 92.2 ± 2.5* 96.8 ± 2.1* 31.0 ± 1.2* 16.3 ± 0.8* 32.0 ± 2.0* 34.5 ± 1.9* 108.7 ± 5.0*
69.3 ± 1.2 74.2 ± 1.4 80.0 ± 1.2 21.5 ± 0.9 11.0 ± 0.6 15.7 ± 1.0 19.2 ± 1.0 65.2 ± 3.3
70.7 ± 2.0 74.7 ± 2.0 82.8 ± 1.4 24.6 ± 1.7 13.0 ± 1.2 19.6 ± 2.4 26.7 ± 2.7 81.0 ± 6.3
72.6 ± 0.5 78.0 ± 0.5 83.0 ± 0.4 23.7 ± 0.5 12.4 ± 0.3 18.1 ± 0.6 21.2 ± 0.7 72.0 ± 1.5
89.8 ± 1.1 115.7 ± 3.8 17.7 ± 4.0* 98.2 ± 35.2* 0.24 ± 0.03* 4.75 ± 3.28*
88.7 ± 0.6 108.7 ± 1.7 8.1 ± 1.1 40.7 ± 6.8 0.11 ± 0.01 2.19 ± 1.57
93.1 ± 2.4 115.3 ± 6.9 11.3 ± 4.6 59.0 ± 36.7 0.12 ± 0.03 2.39 ± 3.02
88.8 ± 0.7 110.2 ± 1.9 9.2 ± 0.9 47.0 ± 6.8 0.13 ± 0.01 2.45 ± 1.80
142.6 ± 38.5* 144.9 ± 33.4* 168.7 ± 38.9* 2.9 ± 0.3* 13.3 ± 0.9* 6.8 ± 0.6* 14.5 ± 1.1* 10.0 ± 0.70* 5.5 ± 1.9*
52.4 ± 9.5 72.4 ± 10.9 86.3 ± 13 2.0 ± 0.1 10.1 ± 0.4 4.8 ± 0.2 9.6 ± 0.4 13.3 ± 0.60 14.7 ± 2.6
74.6 ± 57.8 79.6 ± 45.4 93.3 ± 56.6 2.5 ± 0.5 9.9 ± 1.7 4.9 ± 0.8 10.9 ± 1.5 14.1 ± 2.1 15.3 ± 10.5
60.5 ± 8.5 81.5 ± 9.7 95.9 ± 11.5 2.2 ± 0.1 10.8 ± 0.4 5.0 ± 0.3 10.1 ± 0.5 12.6 ± 0.5 12.6 ± 1.7
After adjustment for body fat mass determined by DXA, there were no significant differences in any body composition measurement between those with AN and those without AN. HOMA-IR, Homeostasis model assessment of insulin resistance. *P < .001 versus unadjusted AN–.
Table III. Receiver operating characteristic curve analysis for diagnosis of fasting insulin level ≥15 µU/mL
Area under the curve Criterion for 100% sensitivity Specificity Likelihood ratio Positive predictive value Negative predictive value
AN
BMI SD score
DXA percent body fat
DXA fat mass (kg)
0.664 ± 0.057 AN+ 0.00 1.00 0.25 0.00
0.833 ± 0.034* >1.57 0.46* 1.84* 0.37* 1.00*
0.845 ± 0.034* >34.6% 0.41* 1.69* 0.36* 1.00*
0.906 ± 0.025* >16.62 kg 0.60* 2.50* 0.45* 1.00*
Criteria were compared at 100% sensitivity. BMI SDS, Standard deviation score for body mass index with age-, sex-, and race-specific standards. *P < .02 versus AN.
pubic hair Tanner stage than AN– African American children (P = .0016). AN+ white children were older than those in the AN– white cohort. Once
age was used as a covariate in analysis of covariance models, breast Tanner stage, pubic hair Tanner stage, and testicular volume were not significantly
different in AN– and AN+ subjects and were not significantly correlated with measures of insulin resistance. Additional adjustments for Tanner 477
NGUYEN ET AL
THE JOURNAL OF PEDIATRICS APRIL 2001
A
B
C
D
Figure. Outcome variables from hyperglycemic clamp study. Subjects with (AN+) and without (AN–) acanthosis nigricans had significantly different (P < .001) first-phase insulin levels (A), steadystate insulin levels (I) (B), metabolic rate (M) (C), and insulin sensitivity (M/I) (D). After adjusting these variables for DXA fat mass and age, results were no longer different between those with and those without AN.
stage did not alter the significance of any described difference. Similarly, results were not different in male and female subjects. Of the 139 study subjects examined, 35 (25%) were AN+. More than 50% of all African Americans, but only 8.2% of overweight white children, were found to be AN+ (Table I, χ2, P < .0001). Because the significance and direction of all differences between AN+ and AN– African American and white subjects were similar, results for all subjects are presented in subsequent tables. Body composition measurements revealed that AN+ subjects had significantly greater waist, abdomen, and hip measurements (Table II, all P < .001) and greater triceps, biceps, subscapu478
lar, and supra-iliac skinfold thickness (all P < .001). After adjusting for DXA fat mass and age, there were no significant body circumference or skinfold differences between AN+ and AN– subjects (Table II). AN+ study subjects had fasting insulin levels that were twice as high as those of AN– subjects (Table II; P < .001). However, after adjusting for DXA fat mass, there were no differences in basal insulin levels between the 2 groups (Table II; P = .26). Similar results were found for glucose-stimulated insulin levels during a 2-hour oral glucose tolerance test. Although AN+ subjects had 2-hour insulin levels that were more than twice as great as those of AN– subjects (Table II, P < .001), these values were no longer sig-
nificant after adjusting for DXA fat mass (Table II; P = .37). Similarly, basal insulin/glucose ratio and homeostasis model assessment of insulin resistance, measures of insulin resistance, were significantly greater for AN+ subjects (P < .001) but were not significantly different after adjusting for DXA fat mass and age (Table II). There were no differences in basal glucose or 2-hour glucose levels during the OGTT between the 2 groups (Table II). During the hyperglycemic clamp study, first-phase insulin (Figure, A, Table II), second-phase insulin (Table II), and steady-state insulin levels (Figure, B, Table II) were significantly greater for AN+ subjects compared with AN– subjects (all P < .001) but were no longer different after adjusting for DXA fat mass (all P > .47). Total glucose disposal rate, estimated as the metabolic rate, M (Figure, C, Table II), was significantly lower for AN+ subjects (10.0 ± 0.70 vs 13.3 ± 0.60 mg/kg/min, P < .001) but was not different after adjustment for body fat and age (14.1 ± 2.1 vs 12.6 ± 0.5 mg/kg/min, P = .48). Because AN+ subjects had significantly lower metabolic rates and greater steady-state insulin, the M/I ratio (Figure, D, Table II) was correspondingly lower compared with that of AN– subjects (5.5 ± 1.9 vs 14.7 ± 2.6 mg/kg/min/µU/mL, P < .001). However, after adjusting for DXA fat mass, there was no difference in the M/I ratio between AN+ and AN– subjects (15.3 ± 10.5 vs 12.6 ± 1.7 mg/kg/min/µU/mL, P = .47). Basal and steady-state C peptide levels (Table II) were also significantly greater in AN+ subjects compared with AN– subjects (P < .001), but after adjusting for DXA fat mass, were no longer significantly different (P = .60). Among the 20 children with the highest insulin levels (fasting insulin >20 µU/mL), 7 (35%) were AN–. When the 34 children with a fasting insulin ≥15.0 µU/mL were examined, 17 (50%) were AN–. Six AN+ subjects
NGUYEN ET AL
THE JOURNAL OF PEDIATRICS VOLUME 138, NUMBER 4 had fasting insulin <10 µU/mL. When the presence of AN was examined as a diagnostic test for insulin ≥15 µU/mL by receiver operating characteristic curve analysis, the receiver operating characteristic area under the curve for AN (0.664 ± 0.057) was significantly less than that for BMI SD score (0.833 ± 0.034, P < .02), DXA percent fat (0.845 ± 0.034, P < .02), or DXA body fat mass (0.906 ± 0.025, P < .02). According to criteria chosen to detect 100% of subjects with hyperinsulinemia, BMI SD score, DXA percent body fat, and DXA body fat mass each had significantly greater specificity, likelihood ratio, and positive predictive value than AN (Table III). Of the 34 subjects with a fasting insulin level ≥15 µU/mL, 30 (88%) had a BMI SD score ≥3.0. Thus a BMI SD score ≥3.0 had 88% sensitivity, 67% specificity, and a likelihood ratio of 2.81 for the diagnosis of hyperinsulinemia.
DISCUSSION The use of AN as a predictive marker of hyperinsulinemia has become a common practice. Previous studies have associated the presence of AN with high insulin levels, thus identifying a subgroup believed to be at greater risk for type 2 diabetes.3,7 Stuart et al7 found that African American adults with AN had significantly greater fasting insulin values than subjects without AN and concluded that AN is a marker for hyperinsulinemia among African Americans. Others studying adults have noted the same association in other racial groups, including white subjects, Native Americans, and Hispanics; they have found that among each race, those individuals with AN have the greatest insulin levels.3 Many studies have suggested an association between AN and obesity; however, none have clearly demonstrated the inter-relationship of AN, obesity, and hyperinsulinemia in overweight children.3-6
Consistent with results of prior studies, we found greater insulin levels and lower insulin sensitivity in overweight children with AN than in children without AN. We then studied whether the greater insulin levels of overweight children with AN could be accounted for by their greater body fat mass. After adjusting the measurements obtained during either an OGTT or a hyperglycemic clamp for DXA body fat mass and age, there were no significant differences between those with and those without AN. These findings led us to consider the possibility that for overweight children, body fat may be a better marker than AN for elevated fasting insulin levels. We found that all of the measures of body adiposity we studied were superior to AN for the diagnosis of an insulin level ≥15 µU/mL. The most easily determined value among those compared is the BMI SD score. We found that 88% of subjects with hyperinsulinemia had a BMI SD score ≥3.0 compared with 65% of subjects with AN. In our group of overweight children, 62 children had a BMI SD score ≥3.0. Thus measuring fasting insulin levels in such children would lead to the identification of significant hyperinsulinemia in almost 50% of such children and would identify 88% of all children with significant hyperinsulinemia. Consistent with previous findings,3,5-8 more than 50% of all African Americans and less than 10% of white subjects in this study had AN. We found no race-specific differences in the relationship of AN to insulin between African American and white subjects. The possible reasons underlying the greater prevalence of AN in African Americans are unclear. Many groups have hypothesized that AN is caused by insulin binding to insulin-like growth factor I receptors located on keratinocytes and dermal fibroblasts.18-23 In childhood24,25 and adulthood,26,27 African Americans have been noted to have greater IGF-I levels than white subjects. We hypothe-
size that the greater IGF-I levels of African Americans may possibly contribute to the greater prevalence of AN among African Americans, without necessarily predisposing them to hyperinsulinemia.24 We conclude that AN should not be used exclusively as the marker for predicting which overweight children have insulin levels in excess of 15 µU/mL. Use of AN as the sole indicator of hyperinsulinemia may lead practitioners to miss the diagnosis in half of all children with significant hyperinsulinemia. Based on the results of this study, we propose that children with a race-, sex-, and age-specific BMI SD score ≥3.0 should be screened for fasting hyperinsulinemia, whether or not they have AN.
REFERENCES 1. Flier JS, Moses AC. Characterization of monoclonal antibodies to the IGF-I receptor that inhibit IGF-I binding to human cells. Biochem Biophys Res Commun 1985;127:929-36. 2. Kuzuya H, Matsuura N, Sakamoto M, Makino H, Sakamoto Y, Kadowaki T, et al. Trial of insulin like growth factor I therapy for patients with extreme insulin resistance syndrome. Diabetes 1993;42:696-705. 3. Stuart CA, Gilkison CR, Keenan BS, Nagamani M. Hyperinsulinemia and acanthosis nigricans in African Americans. J Natl Med Assoc 1997;89:523-7. 4. Stuart CA, Smith MM, Gilkison CR, Shaheb S, Stahn RM. Acanthosis nigricans among Native Americans: an indicator of high diabetes risk. Am J Public Health 1994;84:1839-42. 5. Stuart CA, Pate CJ, Peters EJ. Prevalence of acanthosis nigricans in an unselected population. Am J Med 1989; 87:269-72. 6. Hud JA, Cohen JB, Wagner JM, Cruz PD. Prevalence and significance of acanthosis nigricans in an adult obese population. Arch Dermatol 1992;128:941-4. 7. Stuart CA, Driscoll MS, Lundquist KF, Gilkison CR, Shaheb S, Smith MM. Acanthosis nigricans. J Basic Clin Physiol Pharmacol 1998;9:407-18. 8. Bent KN, Shuster GF, Hurley JS, Frye D, Loflin P, Brubaker C. Acanthosis nigricans as an early clinical 479
NGUYEN ET AL
9.
10.
11.
12.
13.
14.
480
proxy marker of increased risk of type II diabetes. Public Health Nurs 1998;15:415-21. Must A, Dallal GE, Dietz WH. Reference data for obesity: 85th and 95th percentiles of body mass index (wt/ht2) and triceps skinfold thickness. Am J Clin Nutr 1991;53:839-46. Matthews DR, Hosker JP, Rudenski AS, Naylor BA, Treacher DF, Turner RC. Homeostasis model assessment: insulin resistance and beta-cell function from fasting plasma glucose and insulin concentrations in man. Diabetologia 1985,28:412-9. Lohman TG, Roche AF, Martorell R. Anthropometric standardization reference manual. Champaign (IL): Human Kinetics Publishers, Inc; 1988. Gruelich WW, Pyle SI. Radiographic atlas of skeletal development of hand and wrist. 2nd ed. Stanford (CA): Stanford University Press; 1959. Yanovski SZ, Reynolds JC, Boyle A, Yanovski JA. Resting metabolic rate in Caucasian and African American girls. Obes Res 1997;5:321-5. Matsuda M, DeFronzo RA. In vivo measurement of insulin sensitivity in humans. In: Draznin B, Rizza R, editors. Clinical research in diabetes and
THE JOURNAL OF PEDIATRICS APRIL 2001
15.
16.
17.
18.
19.
20. 21.
22.
obesity, part I: methods, assessment, and metabolic regulation. Totowa (NJ): Humana Press, Inc; 1997. p. 23-65. Swets JA. Measuring the accuracy of diagnostic systems. Science 1988;240: 1285-93. McNeil BJ, Keeler E, Adelstein SJ. Primer on certain elements of medical decision making. N Engl J Med 1975; 293:211-5. Hollingshead AB. Four factor index of social status. New Haven (CT): Yale University Press; 1975. Taylor S, Arioglu E. Syndromes associated with insulin resistance and acanthosis nigricans. J Basic Clin Physiol Pharmacol 1998;9:419-39. Schwartz R, Janniger C. Childhood acanthosis nigricans. Pediatr Dermatol 1995;55:337-41. Schwartz R. Acanthosis nigricans. J Am Acad Derm 1994;31:1-19. Matsuoka LY, Wortsman J, Goldman J. Acanthosis nigricans. Clin Dermatol 1993;11:21-5. Accili D, Barbetti F, Cama A, Kadowaki H, Kadowaki T, Imano E, et al. Mutations in the insulin receptor gene in patients with genetic syndromes of insulin resistance and acanthosis nigricans. J Invest Dermatol 1992;98:77S-81S.
23. Rogers D. Acanthosis nigricans. Semin Dermatol 1991;10:160-3. 24. Yanovski J, Sovik KN, Sebring NG. Insulin-like growth factors and bone mineral density in African American and white girls. J Pediatr 2000;137: 826-32. 25. Arslanian S, Suprasongsin C, Janosky JE. Insulin secretion and sensitivity in black versus white prepubertal healthy children. J Clin Endocrinol Metab 1997;82:1923-7. 26. Wright NM, Renault J, Willi S, Veldhuis JD, Pandey JP, Gordon L, et al. Greater secretion of growth hormone in black than in white men: possible factor in greater bone mineral density. J Clin Endocrinol Metab 1995;80: 2291-7. 27. Wright NM, Papadea N, Willi S, Veldhuis JD, Pandey JP, Key LL, et al. Demonstration of a lack of racial difference in secretion of growth hormone despite a racial difference in bone mineral density in premenopausal women. J Clin Endocrinol Metab 1996;81: 1023-6. 28. Frisancho AR. Anthropometric standards for the assessment of growth and nutritional status. Ann Arbor (MI): The University of Michigan Press; 1990.
Acanthosis nigricans is a skin lesion characterized by hyperpigmentation and velvety thickening that occurs on
the neck, axillae, and other skinfolds. Hyperinsulinemia appears to predispose individuals to AN; patients with
From Unit on Growth and Obesity, Developmental Endocrinology Branch, National Institute of Child Health and Human Development, National Institutes of Health (NIH); Nutrition Department, Warren Grant Magnuson Clinical Center, NIH; and Nuclear Medicine Department, Warren Grant Magnuson Clinical Center, NIH, Bethesda, Maryland. Dr Yanovski is supported by the Office of Research on Minority Health, NIH and NIH HD-00641. Dr Yanovski and N. Sebring are commissioned officers in the US Public Health Service. Submitted for publication July 20, 2000; revision received Sept 26, 2000; accepted Oct 26, 2000. Reprint requests: Jack A. Yanovski, MD, PhD, Unit on Growth and Obesity, National Institutes of Health, Building 10, Room 10N262, MSC 1862, 10 Center Dr, Bethesda, MD 20892-1862.
9/21/112657 doi:10.1067/mpd.2001.112657
474
inactivating insulin receptor mutations often have severe AN.1,2 Because the presence of AN is commonly associated with hyperinsulinemia3-6 and obesity,4-6 the use of AN as a marker of hyperinsulinemia and insulin resistance has been proposed for the early detection and prevention of type 2 diabetes.3,4,6,7
See related articles, p 453 and p 469. A few studies have suggested that adults with AN have higher fasting plasma insulin levels than non-AN cohorts,3,4,6 with one report showing that the width of the band of AN in the posterior neck is positively related to fasting insulin levels.3 The prevalence of AN is also disproportionately greater in groups with a high prevalence of type 2 diabetes, such as African Americans, Native Americans, and Hispanics compared with white and Asian subjects.4,7,8 AN BMI DXA IGF-I OGTT
Acanthosis nigrans Body mass index Dual-energy x-ray absorptiometry Insulin-like growth factor I Oral glucose tolerance test
Few studies have examined the association of AN with hyperinsulinism in children or adolescents.4,5,8 Stuart et al5 reported that 66% of 1412 adolescents with a body weight >200% of ideal had AN. In the subset with AN in whom insulin measurements were made, insulin levels were high.5 However, insulin measurements were ob-
NGUYEN ET AL
THE JOURNAL OF PEDIATRICS VOLUME 138, NUMBER 4 Table I. Clinical characteristics
African Americans
Sex Age (y) Bone age (y) Weight (kg) Height (cm) BMI (kg/m2) BMI SD score DXA body fat mass (kg) Hollingshead socioeconomic status score Boys’ testis size (cc) Girls’ breast stage (%) I II III Pubic hair stage (%) I II III IV
White subjects
AN+ (n = 28)
AN– (n = 27)
AN+ (n = 7)
AN– (n = 77)
8 M, 20 F 8.8 ± 0.2 10.9 ± 0.4† 61.7 ± 4.3‡ 142.2 ± 2.2* 29.6 ± 1.3‡ 5.4 ± 0.5‡ 28.6 ± 13.6‡ 3.8 ± 0.2 2.1 ± 0.2
11 M, 16 F 8.2 ± 0.3 9.2 ± 0.4 41.0 ± 2.8 133.9 ± 2.1 22.3 ± 1.0 2.7 ± 0.4 16.6 ± 10.8 3.7 ± 0.1 2.1 ± 0.4
4 M, 3 F 9.7 ± 0.3* 12.4 ± 0.3‡ 70.6 ± 2.1‡ 149.0 ± 3.0† 31.8 ± 0.5‡ 5.0 ± 0.5‡ 34.9 ± 3.0‡ 4.3 ± 0.2 2.3 ± 0.9
33 M, 34 F 8.4 ± 0.2 9.3 ± 0.2 41.0 ± 1.6 134.8 ± 1.2 22.1 ± 0.6 2.2 ± 0.2 16.4 ± 10.6 4.1 ± 0.1 2.3 ± 0.2
7 (35) 8 (40) 5 (25)
8 (50) 7 (44) 1 (6)
1 (33) 1 (33) 1 (33)
16 (47) 16 (47) 2 (6)
5 (18)† 16 (57) 5 (18) 2 (7)
18 (66) 5 (19) 4 (15) 0 (0)
5 (71) 1 (14) 0 (0) 1 (14)
65 (84) 7 (9) 4 (6) 1 (1)
*P < .05, †P < .005, ‡P < .001 for comparisons between African Americans with and without AN or between white subjects with and without AN. BMI-SD score, Body mass index expressed as the SD score for age, sex, and race.28
tained in only 12 subjects with AN, so that the strength of the relationship of AN to hyperinsulinemia in childhood remains unclear. Although children with AN are often overweight, there have been no large studies of the relationship of AN to hyperinsulinemia or insulin sensitivity that have attempted to separate the roles of body adiposity and AN as markers of insulin resistance. Therefore we studied body adiposity, fasting insulin levels, and insulin sensitivity in overweight African American and white children to delineate the nature of the relationship between AN and hyperinsulinemia.
METHODS Subjects We studied 139 African American and white children (aged 6-10 years, 73 girls and 66 boys) who were selected to have a body mass index greater than the 85th percentile for age, sex,
and race, as determined by the First National Health and Nutrition Examination Survey9 (Table I). Subjects were recruited through notices mailed to 6- to 10-year-old children in the Montgomery and Prince Georges County, Maryland, school districts, through referrals from local physicians, and through advertisements placed in local newspapers for a natural history study of body weight. Race was self-reported by each subject. African American and white subjects were selected to have all 4 grandparents of the same race. All subjects had normal findings on physical examination and normal hepatic, renal, and thyroid function. Subjects were medication-free for at least 2 weeks before the start of the study, and all were free of significant medical disease. The study was approved by the National Institutes of Health Intramural Clinical Research Subpanel. Each subject gave written assent and a parent gave
written consent for participation in the study.
Protocol All subjects were studied at the National Institutes of Health (NIH) Warren Grant Magnuson Clinical Center on 2 occasions. At the initial visit, each subject underwent anthropometric measurements and an oral glucose tolerance test, 1.75 g glucose/kg (maximum 75 g), after an overnight fast. Levels of glucose and insulin were measured at 0 hour and 2 hours after glucose ingestion. Glucose assays were performed with standard calorimetric methods on a Hitachi 736-30 analyzer (Boehringer Mannheim, Inc, Indianapolis, Ind), and insulin was measured by radioimmunoassay (Covance, Vienna, Va). Insulin resistance was estimated by the homeostasis model assessment for insulin resistance (HOMA-IR) index by using the following equation10: 475
NGUYEN ET AL
HOMA – IR = [Fasting insulin (µU/L) × Fasting glucose (mmol/L)/22.5] Measurements of weight, height, skinfolds, and waist, abdomen, and hip circumferences were obtained with the use of standardized techniques.11 Weight was measured to the nearest 0.1 kg by using a calibrated digital scale (Scale-Tronix, Wheaton, Ill). Height was measured 3 times to the nearest 1 mm by using a stadiometer calibrated before each height measurement (Holtain Ltd, Crymych, Wales). Waist, abdomen, and hip circumferences were measured by using a flexible non-elastic tape measure. Waist was measured at the narrowest part of the torso or just below the 12th rib, abdomen was measured at the maximum circumference or just above the iliac crest, and hips were measured around the buttocks at the maximum extension. Roentgenograms of the left hand and wrist were obtained to assess bone maturation.12 Each child was examined for the presence or absence of AN on the neck, axillae, and antecubital and intertriginous regions by a trained nurse practitioner or pediatric endocrinologist and assigned to one of two groups: those having acanthosis nigricans (AN+) and those without acanthosis nigricans (AN–). Concordance between raters for the presence or absence of AN was 100%. The second visit occurred 4 to 6 weeks later, at which time subjects underwent dual energy x-ray absorptiometry and a hyperglycemic clamp study. Body composition analysis by DXA was obtained by using the Hologic QDR2000 instrument in the pencil beam mode (software version 5.64), as previously described.13 For the clamp study, 114 subjects were studied after an overnight fast. The plasma glucose level was acutely increased during a 2-minute period to approximately 225 mg/dL by bolus infusion of 25% dextrose and then maintained by continuous infusion of 20% dextrose 476
THE JOURNAL OF PEDIATRICS APRIL 2001 for 120 minutes. Plasma glucose levels were measured every 2.5 minutes for the first 15 minutes and then every 5 minutes until the end of the study. Insulin levels were measured every 5 minutes for the first 15 minutes and then every 15 minutes from 15 to 120 minutes. Plasma glucose levels were measured by using a glucose analyzer (Yellow Springs Instrument, Yellow Springs, Ohio), calibrated to within 5% of multiple glucose standards (50, 100, 125, 150, 250, and 500 mg/dL) before each study. Plasma insulin levels were measured by radioimmunoassay (Covance). First-phase insulin was calculated as the mean of measurements obtained during the first 15 minutes, second-phase insulin as the mean of the measurements from 15 to 120 minutes, and steady-state insulin (I) as the mean of the measurements obtained from 60 to 120 minutes. Whole-body glucose uptake was estimated as the metabolic rate (M), defined as the infusion rate of exogenous glucose administered, corrected for urinary glucose losses and the glucose space correction.14 As a measure of insulin sensitivity, the ratio of metabolic rate to steady-state insulin (M/I) was calculated.
the log of DXA body fat mass and chronological age as covariates to determine whether the relation between the presence of AN and hyperinsulinemia (or other variables) was independent of body composition and age. Least-squares adjusted means are reported and were compared in order to determine the significance of results after adjusting for DXA body fat mass. The receiver operating characteristics15 of AN, BMI SD score, DXA total body fat, and DXA percentage body fat for the diagnosis of fasting hyperinsulinemia were also determined by using the RuleMaker program (Digital Medicine, Inc, Hanover, NH). Sensitivity, specificity, positive and negative predictive values, and diagnostic accuracy were calculated for each test variable.16 For each variable, the criteria with the highest specificity, given 100% sensitivity for an insulin level ≥15 µU/mL, were identified and used for between-test comparisons. For all analyses, significance was adjusted by using the Bonferroni correction for multiple comparisons as appropriate. Values are reported as means ± SD unless otherwise indicated.
Data Analysis
RESULTS
Parametric data were analyzed with SuperAnova 1.11 and StatView 4.5 software (Abacus Concepts, Inc, Berkeley, Calif). Results were first analyzed by using analysis of variance with race and AN as factors to detect differences between those with and those without AN. Although African Americans were more likely to have AN (see results), there were no race × AN status interactions for the measures studied once age was included as a covariate in the model. Including or excluding race as a factor in analyses did not alter the direction or the significance of differences between those with and those without AN in any case. Therefore race was not included in the presented analyses. Analysis of covariance was then performed with
Of the 139 study participants, 55 were African American and 84 were white (Table I). AN+ and AN– subjects were well-matched for socioeconomic status, as assessed by the Hollingshead Scale,17 and consisted of subjects with varying pubertal and adrenarchal development. AN+ African American and white subjects had significantly greater bone age and weight and correspondingly greater BMI and BMI SD scores than AN– subjects. Although neither breast Tanner stage nor testicular volume was significantly different in African American or white AN+ and AN– children and pubic hair stage was similar in AN+ and AN– white children, AN+ African American children had greater
NGUYEN ET AL
THE JOURNAL OF PEDIATRICS VOLUME 138, NUMBER 4 Table II. Body composition and 2-hour OGTT in all subjects and hyperglycemic clamp study in 114 subjects
Unadjusted
Body composition Waist (cm) Abdomen (cm) Hip (cm) Triceps skinfold (mm) Biceps skinfold (mm) Subscapular skinfold (mm) Supra-iliac skinfold (mm) Sum of 4 skinfolds (mm) OGTT Basal glucose (mg/dL) Two-hour glucose (mg/dL) Basal insulin (µU/mL) Two-hour insulin (µU/mL) Basal insulin/glucose ratio HOMA-IR index Hyperglycemic clamp First-phase insulin (µU/mL) Second-phase insulin (µU/mL) Steady-state insulin (µU/mL) Basal C peptide (ng/mL) Steady-state C peptide (ng/mL) Basal insulin/C peptide ratio Steady-state insulin/C peptide Metabolic rate (mg/kg/min) Insulin sensitivity (mg/kg/min/µU/mL)
Adjusted for DXA fat mass and age
AN+
AN–
AN+
AN–
86.2 ± 2.4* 92.2 ± 2.5* 96.8 ± 2.1* 31.0 ± 1.2* 16.3 ± 0.8* 32.0 ± 2.0* 34.5 ± 1.9* 108.7 ± 5.0*
69.3 ± 1.2 74.2 ± 1.4 80.0 ± 1.2 21.5 ± 0.9 11.0 ± 0.6 15.7 ± 1.0 19.2 ± 1.0 65.2 ± 3.3
70.7 ± 2.0 74.7 ± 2.0 82.8 ± 1.4 24.6 ± 1.7 13.0 ± 1.2 19.6 ± 2.4 26.7 ± 2.7 81.0 ± 6.3
72.6 ± 0.5 78.0 ± 0.5 83.0 ± 0.4 23.7 ± 0.5 12.4 ± 0.3 18.1 ± 0.6 21.2 ± 0.7 72.0 ± 1.5
89.8 ± 1.1 115.7 ± 3.8 17.7 ± 4.0* 98.2 ± 35.2* 0.24 ± 0.03* 4.75 ± 3.28*
88.7 ± 0.6 108.7 ± 1.7 8.1 ± 1.1 40.7 ± 6.8 0.11 ± 0.01 2.19 ± 1.57
93.1 ± 2.4 115.3 ± 6.9 11.3 ± 4.6 59.0 ± 36.7 0.12 ± 0.03 2.39 ± 3.02
88.8 ± 0.7 110.2 ± 1.9 9.2 ± 0.9 47.0 ± 6.8 0.13 ± 0.01 2.45 ± 1.80
142.6 ± 38.5* 144.9 ± 33.4* 168.7 ± 38.9* 2.9 ± 0.3* 13.3 ± 0.9* 6.8 ± 0.6* 14.5 ± 1.1* 10.0 ± 0.70* 5.5 ± 1.9*
52.4 ± 9.5 72.4 ± 10.9 86.3 ± 13 2.0 ± 0.1 10.1 ± 0.4 4.8 ± 0.2 9.6 ± 0.4 13.3 ± 0.60 14.7 ± 2.6
74.6 ± 57.8 79.6 ± 45.4 93.3 ± 56.6 2.5 ± 0.5 9.9 ± 1.7 4.9 ± 0.8 10.9 ± 1.5 14.1 ± 2.1 15.3 ± 10.5
60.5 ± 8.5 81.5 ± 9.7 95.9 ± 11.5 2.2 ± 0.1 10.8 ± 0.4 5.0 ± 0.3 10.1 ± 0.5 12.6 ± 0.5 12.6 ± 1.7
After adjustment for body fat mass determined by DXA, there were no significant differences in any body composition measurement between those with AN and those without AN. HOMA-IR, Homeostasis model assessment of insulin resistance. *P < .001 versus unadjusted AN–.
Table III. Receiver operating characteristic curve analysis for diagnosis of fasting insulin level ≥15 µU/mL
Area under the curve Criterion for 100% sensitivity Specificity Likelihood ratio Positive predictive value Negative predictive value
AN
BMI SD score
DXA percent body fat
DXA fat mass (kg)
0.664 ± 0.057 AN+ 0.00 1.00 0.25 0.00
0.833 ± 0.034* >1.57 0.46* 1.84* 0.37* 1.00*
0.845 ± 0.034* >34.6% 0.41* 1.69* 0.36* 1.00*
0.906 ± 0.025* >16.62 kg 0.60* 2.50* 0.45* 1.00*
Criteria were compared at 100% sensitivity. BMI SDS, Standard deviation score for body mass index with age-, sex-, and race-specific standards. *P < .02 versus AN.
pubic hair Tanner stage than AN– African American children (P = .0016). AN+ white children were older than those in the AN– white cohort. Once
age was used as a covariate in analysis of covariance models, breast Tanner stage, pubic hair Tanner stage, and testicular volume were not significantly
different in AN– and AN+ subjects and were not significantly correlated with measures of insulin resistance. Additional adjustments for Tanner 477
NGUYEN ET AL
THE JOURNAL OF PEDIATRICS APRIL 2001
A
B
C
D
Figure. Outcome variables from hyperglycemic clamp study. Subjects with (AN+) and without (AN–) acanthosis nigricans had significantly different (P < .001) first-phase insulin levels (A), steadystate insulin levels (I) (B), metabolic rate (M) (C), and insulin sensitivity (M/I) (D). After adjusting these variables for DXA fat mass and age, results were no longer different between those with and those without AN.
stage did not alter the significance of any described difference. Similarly, results were not different in male and female subjects. Of the 139 study subjects examined, 35 (25%) were AN+. More than 50% of all African Americans, but only 8.2% of overweight white children, were found to be AN+ (Table I, χ2, P < .0001). Because the significance and direction of all differences between AN+ and AN– African American and white subjects were similar, results for all subjects are presented in subsequent tables. Body composition measurements revealed that AN+ subjects had significantly greater waist, abdomen, and hip measurements (Table II, all P < .001) and greater triceps, biceps, subscapu478
lar, and supra-iliac skinfold thickness (all P < .001). After adjusting for DXA fat mass and age, there were no significant body circumference or skinfold differences between AN+ and AN– subjects (Table II). AN+ study subjects had fasting insulin levels that were twice as high as those of AN– subjects (Table II; P < .001). However, after adjusting for DXA fat mass, there were no differences in basal insulin levels between the 2 groups (Table II; P = .26). Similar results were found for glucose-stimulated insulin levels during a 2-hour oral glucose tolerance test. Although AN+ subjects had 2-hour insulin levels that were more than twice as great as those of AN– subjects (Table II, P < .001), these values were no longer sig-
nificant after adjusting for DXA fat mass (Table II; P = .37). Similarly, basal insulin/glucose ratio and homeostasis model assessment of insulin resistance, measures of insulin resistance, were significantly greater for AN+ subjects (P < .001) but were not significantly different after adjusting for DXA fat mass and age (Table II). There were no differences in basal glucose or 2-hour glucose levels during the OGTT between the 2 groups (Table II). During the hyperglycemic clamp study, first-phase insulin (Figure, A, Table II), second-phase insulin (Table II), and steady-state insulin levels (Figure, B, Table II) were significantly greater for AN+ subjects compared with AN– subjects (all P < .001) but were no longer different after adjusting for DXA fat mass (all P > .47). Total glucose disposal rate, estimated as the metabolic rate, M (Figure, C, Table II), was significantly lower for AN+ subjects (10.0 ± 0.70 vs 13.3 ± 0.60 mg/kg/min, P < .001) but was not different after adjustment for body fat and age (14.1 ± 2.1 vs 12.6 ± 0.5 mg/kg/min, P = .48). Because AN+ subjects had significantly lower metabolic rates and greater steady-state insulin, the M/I ratio (Figure, D, Table II) was correspondingly lower compared with that of AN– subjects (5.5 ± 1.9 vs 14.7 ± 2.6 mg/kg/min/µU/mL, P < .001). However, after adjusting for DXA fat mass, there was no difference in the M/I ratio between AN+ and AN– subjects (15.3 ± 10.5 vs 12.6 ± 1.7 mg/kg/min/µU/mL, P = .47). Basal and steady-state C peptide levels (Table II) were also significantly greater in AN+ subjects compared with AN– subjects (P < .001), but after adjusting for DXA fat mass, were no longer significantly different (P = .60). Among the 20 children with the highest insulin levels (fasting insulin >20 µU/mL), 7 (35%) were AN–. When the 34 children with a fasting insulin ≥15.0 µU/mL were examined, 17 (50%) were AN–. Six AN+ subjects
NGUYEN ET AL
THE JOURNAL OF PEDIATRICS VOLUME 138, NUMBER 4 had fasting insulin <10 µU/mL. When the presence of AN was examined as a diagnostic test for insulin ≥15 µU/mL by receiver operating characteristic curve analysis, the receiver operating characteristic area under the curve for AN (0.664 ± 0.057) was significantly less than that for BMI SD score (0.833 ± 0.034, P < .02), DXA percent fat (0.845 ± 0.034, P < .02), or DXA body fat mass (0.906 ± 0.025, P < .02). According to criteria chosen to detect 100% of subjects with hyperinsulinemia, BMI SD score, DXA percent body fat, and DXA body fat mass each had significantly greater specificity, likelihood ratio, and positive predictive value than AN (Table III). Of the 34 subjects with a fasting insulin level ≥15 µU/mL, 30 (88%) had a BMI SD score ≥3.0. Thus a BMI SD score ≥3.0 had 88% sensitivity, 67% specificity, and a likelihood ratio of 2.81 for the diagnosis of hyperinsulinemia.
DISCUSSION The use of AN as a predictive marker of hyperinsulinemia has become a common practice. Previous studies have associated the presence of AN with high insulin levels, thus identifying a subgroup believed to be at greater risk for type 2 diabetes.3,7 Stuart et al7 found that African American adults with AN had significantly greater fasting insulin values than subjects without AN and concluded that AN is a marker for hyperinsulinemia among African Americans. Others studying adults have noted the same association in other racial groups, including white subjects, Native Americans, and Hispanics; they have found that among each race, those individuals with AN have the greatest insulin levels.3 Many studies have suggested an association between AN and obesity; however, none have clearly demonstrated the inter-relationship of AN, obesity, and hyperinsulinemia in overweight children.3-6
Consistent with results of prior studies, we found greater insulin levels and lower insulin sensitivity in overweight children with AN than in children without AN. We then studied whether the greater insulin levels of overweight children with AN could be accounted for by their greater body fat mass. After adjusting the measurements obtained during either an OGTT or a hyperglycemic clamp for DXA body fat mass and age, there were no significant differences between those with and those without AN. These findings led us to consider the possibility that for overweight children, body fat may be a better marker than AN for elevated fasting insulin levels. We found that all of the measures of body adiposity we studied were superior to AN for the diagnosis of an insulin level ≥15 µU/mL. The most easily determined value among those compared is the BMI SD score. We found that 88% of subjects with hyperinsulinemia had a BMI SD score ≥3.0 compared with 65% of subjects with AN. In our group of overweight children, 62 children had a BMI SD score ≥3.0. Thus measuring fasting insulin levels in such children would lead to the identification of significant hyperinsulinemia in almost 50% of such children and would identify 88% of all children with significant hyperinsulinemia. Consistent with previous findings,3,5-8 more than 50% of all African Americans and less than 10% of white subjects in this study had AN. We found no race-specific differences in the relationship of AN to insulin between African American and white subjects. The possible reasons underlying the greater prevalence of AN in African Americans are unclear. Many groups have hypothesized that AN is caused by insulin binding to insulin-like growth factor I receptors located on keratinocytes and dermal fibroblasts.18-23 In childhood24,25 and adulthood,26,27 African Americans have been noted to have greater IGF-I levels than white subjects. We hypothe-
size that the greater IGF-I levels of African Americans may possibly contribute to the greater prevalence of AN among African Americans, without necessarily predisposing them to hyperinsulinemia.24 We conclude that AN should not be used exclusively as the marker for predicting which overweight children have insulin levels in excess of 15 µU/mL. Use of AN as the sole indicator of hyperinsulinemia may lead practitioners to miss the diagnosis in half of all children with significant hyperinsulinemia. Based on the results of this study, we propose that children with a race-, sex-, and age-specific BMI SD score ≥3.0 should be screened for fasting hyperinsulinemia, whether or not they have AN.
REFERENCES 1. Flier JS, Moses AC. Characterization of monoclonal antibodies to the IGF-I receptor that inhibit IGF-I binding to human cells. Biochem Biophys Res Commun 1985;127:929-36. 2. Kuzuya H, Matsuura N, Sakamoto M, Makino H, Sakamoto Y, Kadowaki T, et al. Trial of insulin like growth factor I therapy for patients with extreme insulin resistance syndrome. Diabetes 1993;42:696-705. 3. Stuart CA, Gilkison CR, Keenan BS, Nagamani M. Hyperinsulinemia and acanthosis nigricans in African Americans. J Natl Med Assoc 1997;89:523-7. 4. Stuart CA, Smith MM, Gilkison CR, Shaheb S, Stahn RM. Acanthosis nigricans among Native Americans: an indicator of high diabetes risk. Am J Public Health 1994;84:1839-42. 5. Stuart CA, Pate CJ, Peters EJ. Prevalence of acanthosis nigricans in an unselected population. Am J Med 1989; 87:269-72. 6. Hud JA, Cohen JB, Wagner JM, Cruz PD. Prevalence and significance of acanthosis nigricans in an adult obese population. Arch Dermatol 1992;128:941-4. 7. Stuart CA, Driscoll MS, Lundquist KF, Gilkison CR, Shaheb S, Smith MM. Acanthosis nigricans. J Basic Clin Physiol Pharmacol 1998;9:407-18. 8. Bent KN, Shuster GF, Hurley JS, Frye D, Loflin P, Brubaker C. Acanthosis nigricans as an early clinical 479
NGUYEN ET AL
9.
10.
11.
12.
13.
14.
480
proxy marker of increased risk of type II diabetes. Public Health Nurs 1998;15:415-21. Must A, Dallal GE, Dietz WH. Reference data for obesity: 85th and 95th percentiles of body mass index (wt/ht2) and triceps skinfold thickness. Am J Clin Nutr 1991;53:839-46. Matthews DR, Hosker JP, Rudenski AS, Naylor BA, Treacher DF, Turner RC. Homeostasis model assessment: insulin resistance and beta-cell function from fasting plasma glucose and insulin concentrations in man. Diabetologia 1985,28:412-9. Lohman TG, Roche AF, Martorell R. Anthropometric standardization reference manual. Champaign (IL): Human Kinetics Publishers, Inc; 1988. Gruelich WW, Pyle SI. Radiographic atlas of skeletal development of hand and wrist. 2nd ed. Stanford (CA): Stanford University Press; 1959. Yanovski SZ, Reynolds JC, Boyle A, Yanovski JA. Resting metabolic rate in Caucasian and African American girls. Obes Res 1997;5:321-5. Matsuda M, DeFronzo RA. In vivo measurement of insulin sensitivity in humans. In: Draznin B, Rizza R, editors. Clinical research in diabetes and
THE JOURNAL OF PEDIATRICS APRIL 2001
15.
16.
17.
18.
19.
20. 21.
22.
obesity, part I: methods, assessment, and metabolic regulation. Totowa (NJ): Humana Press, Inc; 1997. p. 23-65. Swets JA. Measuring the accuracy of diagnostic systems. Science 1988;240: 1285-93. McNeil BJ, Keeler E, Adelstein SJ. Primer on certain elements of medical decision making. N Engl J Med 1975; 293:211-5. Hollingshead AB. Four factor index of social status. New Haven (CT): Yale University Press; 1975. Taylor S, Arioglu E. Syndromes associated with insulin resistance and acanthosis nigricans. J Basic Clin Physiol Pharmacol 1998;9:419-39. Schwartz R, Janniger C. Childhood acanthosis nigricans. Pediatr Dermatol 1995;55:337-41. Schwartz R. Acanthosis nigricans. J Am Acad Derm 1994;31:1-19. Matsuoka LY, Wortsman J, Goldman J. Acanthosis nigricans. Clin Dermatol 1993;11:21-5. Accili D, Barbetti F, Cama A, Kadowaki H, Kadowaki T, Imano E, et al. Mutations in the insulin receptor gene in patients with genetic syndromes of insulin resistance and acanthosis nigricans. J Invest Dermatol 1992;98:77S-81S.
23. Rogers D. Acanthosis nigricans. Semin Dermatol 1991;10:160-3. 24. Yanovski J, Sovik KN, Sebring NG. Insulin-like growth factors and bone mineral density in African American and white girls. J Pediatr 2000;137: 826-32. 25. Arslanian S, Suprasongsin C, Janosky JE. Insulin secretion and sensitivity in black versus white prepubertal healthy children. J Clin Endocrinol Metab 1997;82:1923-7. 26. Wright NM, Renault J, Willi S, Veldhuis JD, Pandey JP, Gordon L, et al. Greater secretion of growth hormone in black than in white men: possible factor in greater bone mineral density. J Clin Endocrinol Metab 1995;80: 2291-7. 27. Wright NM, Papadea N, Willi S, Veldhuis JD, Pandey JP, Key LL, et al. Demonstration of a lack of racial difference in secretion of growth hormone despite a racial difference in bone mineral density in premenopausal women. J Clin Endocrinol Metab 1996;81: 1023-6. 28. Frisancho AR. Anthropometric standards for the assessment of growth and nutritional status. Ann Arbor (MI): The University of Michigan Press; 1990.