¹University of Virginia Health Sciences Center, Department of Pediatrics, Division of Endocrinology, Charlottesville, Virginia, USA

²Department of Human Services, Curry School of Education, University of Virginia, Charlottesville, Virginia, USA

³University of Virginia Health Sciences Center, Department of Medicine, Division of Endocrinology and Metabolism, Charlottesville, Virginia, USA

⁴University at Buffalo, School of Medicine and Biomedical Sciences, Department of Pediatrics, Division of Behavioral Medicine, Buffalo, New York, USA

⁵University of Virginia Health Sciences Center, Department of Pharmacology, Charlottesville, Virginia, USA

Correspondence to: J N Roemmich, Department of Pediatrics, Division of Behavioral Medicine, State University of New York at Buffalo, Farber Hall, Room G56, 3435 Main Street, Building No. 26, Buffalo, NY 14214-3000, USA. E-mail: roemmich@buffalo.edu

^a†JN Roemmich is currently with the Department of Pediatrics, Division of Behavioral Medicine, State University of New York at Buffalo; PA Clark is with the Department of Pediatrics, Division of Endocrinology, University of Louisville; and AD Rogol is with Insmed Pharmaceuticals Inc., Richmond, VA, USA.

Objective: To investigate the independent influence of alterations in fat mass, body fat distribution and hormone release on pubertal increases in fasting serum insulin concentrations and on insulin resistance assessed by the homeostasis model (HOMA).

Design and Subjects: Cross-sectional investigation of pre- (n=11, n=8), mid- (n=10, n=11), and late-pubertal (n=10, n=11) boys and girls with normal body weight and growth velocity.

Measurements: Body composition (by a four-compartment model), abdominal fat distribution and mid-thigh interfascicular plus intermuscle (extramyocellular) fat (by magnetic resonance imaging), total body subcutaneous fat (by skinfolds), mean nocturnal growth hormone (GH) release and 06:00 h samples of serum insulin, sex steroids, leptin and insulin-like growth factor-I (IGF-I).

Results: Pubertal insulin resistance was suggested by greater (P<0.001) fasting serum insulin concentrations in the late-pubertal than pre- and mid-pubertal groups while serum glucose concentrations were unchanged and greater (P<0.001) HOMA values in late-pubertal than pre- and mid-pubertal youth. From univariate correlation fat mass was most related to HOMA (r=0.59, P<0.001). Two hierarchical regression models were developed to predict HOMA. In one approach, subject differences in sex, pubertal maturation, height and weight were held constant by adding these variables as a block in the first step of the model (r²=0.36). Sequential addition of fat mass (FM) increased r² (r²_{(inc)remental}=0.08, r²=0.44, P<0.05) as did the subsequent addition of a block of fat distribution variables (extramyocellular fat, abdominal visceral fat, and sum of skinfolds; r²_inc=0.11, r²=0.55, P<0.05). Sequential addition of a block of hormone variables (serum IGF-I and log₍₁₀₎ leptin concentrations; r²_inc=0.04, P>0.05) did not reliably improve r² beyond the physical characteristic and adiposity variables. In a second model, differences in sex and pubertal maturation were again held constant (r²=0.25), but body size differences were accounted for using percentage fat data. Sequential addition of percentage body fat (r²_{(inc)remental}=0.11, r²=0.36, P<0.05), then a block of fat distribution variables (percentage extramyocellular fat, percentage abdominal visceral fat, and percentage abdominal subcutaneous fat; r²_inc=0.08, r²=0.44, P=0.058), and then a block of serum IGF-I and log₍₁₀₎ leptin concentrations (r²_inc=0.07, r²=0.51, P<0.05) increased r². Mean nocturnal GH release was not related to HOMA (r=-0.04, P=0.75) and therefore was not included in the hierarchical regression models.

Conclusion: Increases in insulin resistance at puberty were most related to FM. Accumulation of fat in the abdominal visceral, subcutaneous and muscular compartments may increase insulin resistance at puberty beyond that due to total body fat. Serum concentrations of leptin and IGF-I may further modulate HOMA beyond the effects of adiposity and fat distribution. However, the results are limited by the cross-sectional design and the use of HOMA rather than a criterion measure of insulin resistance.

International Journal of Obesity (2002) 26, 701-709. DOI:10.1038/sj/ijo/0801975

insulin resistance; myocellular triglyceride; abdominal visceral fat; growth hormone; leptin; insulin-like growth factor-I

Introduction

Puberty is associated with modest insulin resistance, compensated by increased insulin secretion and serum insulin concentrations.^1,2-³ Increased fasting serum insulin concentrations and insulin resistance occur in both lean and obese youth and is greater in girls than in boys.^1,2,3 Based on the universal nature of insulin resistance, it is probably caused by pubertal alterations in total adiposity, body fat-distribution and growth hormone (GH), insulin-like growth factor-I (IGF-I), and leptin release. Insulin resistance declines in early adulthood,^2,4 but it is important to understand the modulators of pubertal insulin resistance because puberty is a high-risk developmental period for obesity and type 2 diabetes mellitus,⁵ and insulin resistance may play a role in the development of these diseases. Pubertal insulin resistance is restricted to glucose metabolism while the anabolic effect on fat metabolism remains potent.⁶

An observable change during puberty is in the amount and regional distribution of body fat. The total fat mass (FM) and fat stored in the abdominal visceral (AVF^7,8,9,10), abdominal subcutaneous (ASF,^11,12), and the muscle depots^{11,13,14,15,16} increase with body size during puberty and are all associated with insulin resistance. Increased AVF is thought to lead to a cascade of metabolic derangements resulting in insulin resistance.^7,17 Increased muscular fat and subcutaneous fat may also reduce insulin sensitivity.¹¹ Inverse relationships between muscular fat and insulin sensitivity may be especially relevant during puberty because muscular fat increases during puberty and muscle is the primary site for insulin-stimulated glucose disposal under euglycemic conditions.^18,19 Whether the total FM or the fat distribution is most associated with increased serum insulin concentrations and reduced insulin sensitivity during puberty is unclear.

The adipose tissue and pancreatic islets are functionally connected through the adipoinsular axis.²⁰ Leptin directly inhibits insulin secretion from isolated human islets and, after food intake, leptin feeds back to reduce insulin secretion.²⁰ Perhaps increases in serum leptin concentration at the initiation of puberty²¹ cause modest leptin resistance at the -cells resulting in increased insulin concentrations.

GH secretion also increases during puberty. GH has well-known insulin antagonistic effects²² and, along with alterations in adiposity and fat distribution, increased GH secretion is thought to be a primary mechanism of pubertal insulin resistance.^23,24,25

Measuring the likely modulators in the same subjects is necessary to determine their relative importance because they are all related to one another and to insulin sensitivity. Therefore, we examined the relationships among total adiposity, fat distribution, leptin concentrations, and the GH-IGF-I axis on fasting serum insulin concentrations and homeostasis model assessments (HOMA) of insulin resistance²⁶ in pre-, mid- and late-pubertal boys and girls. We hypothesized that after holding constant initial differences in body size and pubertal maturation, increases in insulin resistance would be primarily related to increases in total FM but, also directly and independently related to body fat distribution and hormone release.

Methods

Subjects

The sample included pre- (n=10, n=8), mid- (n=13, n=8), and late-pubertal (n=10, n=12) boys and girls with a normal weight, height and weight-for-height. After visual inspection of the development of the genitalia (G) of boys and breasts (B) of girls by a pediatric endocrinologist, subjects were placed into pubertal groups based on Tanner's criteria²⁷ as follows G1, B1, prepubertal; G2-4, B2-4, midpubertal; G5, B5, late-pubertal. Informed consent from a parent and assent from the child were obtained before entry into the study. All subjects were enrolled in an ongoing longitudinal study of growth and maturation and were growing at appropriate rates for their stage of sexual maturation. This paper includes a cross-sectional analysis of the fasting serum insulin concentrations of these youths.

Blood sampling and hormone assays

After admission to the General Clinical Research Center at 08:00 h, a catheter was inserted into a forearm vein at 16:00 h and kept patent with a heparin lock. To determine mean nocturnal GH release, serial blood sampling (every 10 min) was initiated at 18:00 h and continued until 06:00 h as previously described.²⁸ The Nichols Luma Tag hGH chemiluminescence assay (San Juan Capistrano, CA, USA) was used to measure the serum GH concentrations with a sensitivity of 0.002 µg/l. The intra-assay coefficients of variation was 4.9% at 0.2 µg/l, 6.7% at 2 µg/l, and 6.4% at 4.9 µg/l, while the inter-assay coefficients of variation was 7.2% at both 1.7 and 4.2 µg/l. GH pulse characteristics were assessed by the model-free Cluster algorithm, version 6.01.²⁹

Serum insulin, total testosterone, estradiol, IGF-I, leptin and glucose concentrations were measured from the 06:00 h blood sample. Insulin, testosterone and estradiol were measured by RIA kits from Diagnostic Products Corporation (Los Angeles, CA, USA). The sensitivity of the insulin assay was 1.3 µIU/ml with an intra-assay coefficient of variation (CV) of 8.3-6.4% within the range of 4.8-54.6 µIU/ml and interassay CV of 12.2-4.7% within the range of 4.9-52.9 µIU/ml. The sensitivity of the testosterone assay was 10.0 ng/dl with an intra-assay CV of 5-6% within the range of 100-800 ng/dl and inter-assay CV of 9.2-12.9% within the range of 70-840 ng/dl. The sensitivity of the estradiol assay was 10.0 pg/ml with an intra-assay CV of 4-7% within the range of 50-1100 pg/ml. The inter-assay CV ranged from 4.2 to 8.1% within the range of 50-1025 pg/ml. IGF-I concentrations were measured by RIA (Nichols Institute, San Juan Capistrano, CA, USA) after acid-ethanol extraction and had an intra-assay CV of 2.4 and 3.0% at 0.53 and 0.92 ng/ml and an inter-assay CV of 5.2 and 8.4% at 0.54 and 0.82 ng/ml. The sensitivity was 0.06 ng/ml. Serum leptin concentration was measured by RIA with a detection limit of 0.5 ng/ml and a within- and between-assay CV of 4.4 and 6.9% at 2.9 ng/ml and 5.7 and 9.0% at 14.1 ng/ml, respectively.²¹ Serum glucose concentrations were determined using an Olympus (Olympus Optical Company, Tokyo, Japan) automated chemistry analyzer.

Body composition

Body composition was estimated using a four-compartment model described by Lohman.³⁰ We have described and validated the use of this model in our laboratory.³¹ Body density is measured by underwater weighing and corrected for residual lung volume by nitrogen washout. The body density is corrected for the total body water by deuterium oxide dilution and bone mineral content by dual-energy X-ray absorptiometry (Hologic QDR 2000, Waltham, MA, USA).

Magnetic resonance imaging (MRI)

Subcutaneous and visceral fat areas at the level of the L4-L5 intervertebral space and the intermuscle plus interfascicle fat (extramyocellular) area of the mid-thigh were measured with magnetic resonance imaging (MRI) using a Siemens Vision 1.5 T scanner. A T₁-weighted spin-echo sagittal scout scan with a repetition of 500 ms, echo time (TE) of 20 ms, 10 mm slice thickness with a 10 mm gap, 128´256 matrix and two signal averages was used to locate the L4-L5 disk space and the mid-thigh defined as one-half the distance between the greater trochanter of the femur and superior border of the patella. Adipose and lean tissue areas were assessed using a T₁-weighted spin-echo sequence. The fat and water based tissue areas were determined using MedX Software (Sensor Systems, Sterling, VA, USA). Percentage of the sagittal abdominal area consisting of visceral fat and subcutaneous fat were calculated by dividing both the abdominal visceral and abdominal subcutaneous cross-sectional areas by the total area of the abdomen in the image. A similar ratio was used to determine the percentage extramyocellular fat of the thigh.

Anthropometry

A trained anthropometrist (JNR) completed all measures. Height, waist girth, hip girth, trunk skinfolds (subscapular, chest, mid-axillary, suprailliac, abdominal) and peripheral skinfolds (triceps, biceps, thigh and medial calf) were measured as recommended.³²

Data analysis

Insulin resistance was estimated with the HOMA model as described by Matthews et al.²⁶ Analysis of variance (2 (gender)´3 (maturation)) and covariance was used to test for group differences in body composition, body fat distribution and serum hormone concentrations. Pearson correlations were used to examine the strength of the relationship between fasting serum insulin concentrations and body composition, body fat distribution and hormone variables. Hierarchical regression was used to determine whether addition of information regarding body fat distribution and then hormone release improved prediction of fasting serum insulin concentrations beyond that provided by FM while holding constant differences in physical characteristics. In order to determine the independent predictive ability of body fat distribution beyond FM, FM was added to the model before the body fat distribution variables because they are a subcomponent of the FM and highly interrelated with FM. Leptin concentrations were log transformed because the absolute values were not normally distributed.

Results

Table 1 shows the subjects' physical characteristics. The age (P=0.001), bone age (P<0.001), height (P<0.001), weight (P=0.001), and fat-free mass (P<0.001) differed for all three pubertal groups. The late-pubertal group had greater amounts of fat mass (P<0.001), extramyocellular fat (P<0.001), ASF (P=0.002), and sum of skinfolds (P=0.003) than the pre- and mid-pubertal groups. The late-pubertal group had a greater (P=0.02) amount of AVF than the prepubertal group. The boys had a greater chronological age (P=0.006) and bone age (P=0.001) than the girls due to the later puberty of boys. Height (P<0.001), weight (P=0.007), and fat-free mass (P<0.001) were also greater in the boys than the girls. Percentage body fat (P<0.001), FM (P=0.004), extramyocellular fat (P=0.01), percentage extramyocellular fat (P=0.01), percentage AVF (P=0.04), ASF (P=0.017), percentage ASF (P<0.001), and sum of skinfolds (P=0.013) were greater in the girls than the boys. The percentage body fat (gender-maturation interaction effect, P=0.03), FM (gender-maturation interaction effect, P=0.008), extramyocellular fat (gender-maturation interaction effect, P=0.01), percentage extramyocellular fat (gender-maturation interaction effect, P=0.05), and sum of skinfolds (gender-maturation interaction effect, P=0.045) of the late-pubertal girls was greater than that of the late-pubertal boys. The boys' fat-free mass was greater than the girls' in the mid- and late-pubertal groups but not the pre-pubertal group (interaction effect, P=0.01).

Serum hormone and glucose concentrations are shown in Table 2. Fasting serum insulin concentrations and HOMA values were greater (P<0.001) in the late-pubertal group than the pre- and mid-pubertal groups but did not differ between boys and girls. There were no group differences or interaction effect when serum insulin concentrations or HOMA was covaried for FM, but the late pubertal subjects had greater (P<0.001) serum insulin concentrations and HOMA when covaried for percentage body fat. Serum glucose concentrations did not differ between boys and girls (P=0.08) or maturational groups (P=0.77). Serum testosterone (interaction effect, P<0.001) and estradiol (interaction effect, P=0.04) concentrations of the boys and girls differed at mid- and late-puberty but not prepubertally. Mean nocturnal GH release was lower (P=0.009) in the prepubertal group than the mid- and late-pubertal groups. Serum IGF-I concentrations were different (P<0.001) in each pubertal group. Serum log₍₁₀₎ leptin concentrations were greater in the girls than the boys (P<0.001; interaction effect, P=0.08).

HOMA values and fasting serum insulin concentrations were directly and significantly related to all variables shown in Table 3 and all variables except percentage AVF shown in Table 4. Both HOMA and fasting serum insulin concentrations were most highly related to FM (Table 3). HOMA values and fasting serum insulin concentrations were also related to chronological age (r=0.45, P<0.001; r=0.43, P=0.001), bone age (HOMA, see Figure 1; fasting serum insulin, r=0.52, P<0.001), and ASF (r=0.56, P<0.001; r=0.67, P<0.001), but not related to mean GH concentration (r=-0.04, 0.75; r=-0.07, P=0.60), respectively.

Two sets of hierarchical regression analyses were performed and the incremental semipartial regressions (sr_i²), and the r, r², and adjusted r² after entry of all blocks of independent variables are shown in Tables 3 and 4. For the set of regressions shown in Table 3, subject differences in sex, pubertal maturation, height and weight were held constant by adding these variables as a block in the first step of the models. These variables were added first so that the predictive value of the remaining independent variables could be determined while holding constant initial differences in body size and pubertal maturation. In the HOMA prediction model these physical characteristics produced an r²=0.36. Sequential addition of FM increased r² (r²_{(inc)remental}=0.08, r²=0.44, P<0.05). For the remaining blocks, order of entry of variables was according to our hypothesis that increased insulin resistance during puberty is most related to increased FM with independent influences of body fat distribution and hormone release. The subsequent addition of a block of fat distribution variables (extramyocellular fat, AVF, and sum of skinfolds) further increased r² (r²_inc=0.11, r²=0.55, P<0.05). Sequential addition of a block of hormone variables (serum IGF-I and log₍₁₀₎ leptin concentrations; r²_inc=0.04, P>0.05) did not reliably improve r² beyond the physical characteristic and adiposity variables. For the fasting serum insulin model, the block of physical characteristics produced an r²=0.46 (P<0.001). After adding FM to the physical characteristics there was a significant increment in r² of 0.06 to a new total r² of 0.52 (P=0.01). Adding the block of body fat distribution variables to FM and the physical characteristics produced a significant 0.08 unit increment in r² to a new total r² of 0.60 (P=0.01). Adding the hormone variables (serum log₍₁₀₎ leptin and IGF-I concentrations) to the body composition, FM and physical characteristic variables increased r² by an increment of 0.04 units to a new total r² of 0.64 but, this was not a reliable improvement (P=0.08).

In the second set of models (Table 4), differences in sex and pubertal maturation were again held constant (r²=0.25), but body size differences were accounted for using percentage fat data. These models were developed due to concern that height (r=0.34, P=0.007) and weight (r=0.66, P<0.001) are directly related to FM and inclusion of height and weight in the initial block may decrease the variance that can be accounted for by FM in the second block. For the HOMA model, the block of sex and pubertal maturation group produced an r²=0.25 (P<0.001). Sequential addition of percentage body fat increased r² (r²_inc=0.11, r²=0.36, P<0.05), then a block of fat distribution variables (percentage extramyocellular fat, percentage AVF, and percentage ASF; r²_inc=0.08, r²=0.44, P=0.058), and then a block of serum IGF-I and log₍₁₀₎ leptin concentrations (r²_inc=0.07, r²=0.51, P<0.05) increased r². For the insulin model, the sex and pubertal maturation block produced an r²=0.31 (P<0.001). Sequential addition of percentage body fat (r²_inc=0.16, r²=0.47, P<0.001), the block of fat distribution variables (r²_inc=0.08, r²=0.55, P<0.05), and the block of serum IGF-I and log₍₁₀₎ leptin concentrations (r²_inc=0.05, r²=0.60, P=0.059) increased incremental r². Mean nocturnal GH release was not related to HOMA (r=-0.04, P=0.75) nor fasting serum insulin concentrations (r=-0.07, P=0.60) and therefore was not included in the hierarchical regression models.

Discussion

Pubertal insulin resistance is compensated by increased insulin secretion resulting in increased serum insulin concentrations.^1,2,3 The cause of pubertal insulin resistance is uncertain but, due to its universal nature, is likely due to pubertal alterations in adiposity, body fat-distribution and hormone release. Therefore, we determined the relationship of total adiposity, body fat distribution, and GH, IGF-I, and leptin release with fasting serum insulin concentrations in normal-weight boys and girls. This is the first attempt to measure all of these variables in the same subjects.

Determining whether maturational increases in fasting insulin are more related to alterations in body composition, body fat distribution or hormonal release is difficult because these variables are all related to one another and to fasting serum insulin or insulin resistance measures. We used hierarchical regression to test our hypothesis that pubertal increases in insulin resistance would be primarily related to increases in total adiposity and independently related to body fat distribution and hormone release, especially GH release. Two sets of hierarchical regression models were developed. The first set adjusted for differences in body size and biological maturity by adding a block of physical characteristic variables to the model in the first step. The second set of models adjusted for differences in maturation, but body size differences were accounted for using percentage adiposity data rather than absolute adiposity data. The results of both sets of models were very consistent (Tables 3 and 4).

A key finding was that insulin resistance as measured by the HOMA model was greater in the late-pubertal group than in the pre- and mid-pubertal groups (Table 2). Fasting serum insulin concentrations were greater in the late-pubertal group, while fasting serum glucose concentrations were not different between pubertal maturation groups, which also suggests pubertal insulin resistance (Table 2). Although insulin resistance is best measured by the euglycemic clamp technique, the HOMA model^26,33 and fasting insulin concentrations^2,17,34,35 adequately parallel insulin resistance measures. However, HOMA values and fasting serum insulin concentrations were highly related (r=0.92, P<0.001) and HOMA may provide little information beyond fasting serum insulin concentrations. Furthermore, we cannot be certain that similar results would have been found if we had measured insulin resistance by the euglycemic clamp because serum insulin concentrations, insulin secretion and insulin resistance may all be influenced by different physiological factors.

Consistent with our hypothesis, insulin resistance was most related to FM (Table 3) and percentage body fat (Table 4). Furthermore, there were no maturational differences in insulin resistance after statistically adjusting HOMA values or fasting serum insulin concentrations for FM. This hypothesis was based on strong evidence that insulin resistance and fasting serum insulin concentrations increase in obese children and adolescents.^17,36,37,38 A potential argument against the body fat hypothesis is that insulin resistance may peak at mid-puberty and then decrease although FM remains constant or continues to increase. As shown in Figure 1, there was a direct relationship between bone age and insulin resistance, suggesting that in the present subjects insulin resistance was increasing through puberty. Others have also reported that insulin resistance remains elevated throughout puberty.^2,24,39 However, insulin resistance does eventually decrease during adulthood, while the amount of FM and percentage body fat increases, so factors other than FM are probably also involved.

Our results that insulin resistance was most related to total adiposity are contrary to adult data that have shown AVF is most strongly linked to insulin resistance in adults.⁴⁰ Few empirical data support the AVF supposition in youth. Arslanian and Suprasongsin³⁶ found that the percentage body fat was the primary variable related to insulin resistance in prepubertal and pubertal subjects but AVF was not measured. Others^7,8,9,10 have reported direct relationships between basal or stimulated insulin secretion and AVF in children, but total body composition data were not reported. Gower and colleagues⁴¹ reported that, in children, fasting insulin was more related to the total fat mass than AVF. In the present investigation, HOMA values and fasting serum insulin concentrations were also less related to AVF or percentage AVF than FM or percentage body fat (Tables 3 and 4). The inclusion of mostly normal-weight youth in our study may have prevented the observation of a stronger direct relationship between HOMA values or fasting serum insulin concentrations and AVF. Perhaps a threshold amount of AVF is necessary before adverse events are observed. A threshold of 130 cm² has been suggested in adults⁴⁰ and none of our subjects had an AVF area greater than 120 cm². When presented as percentage AVF data, the girls (14.8%) had a slightly yet significantly greater percentage AVF (Table 1) than the boys (12.3%). In contrast, adult women have about half the percentage AVF of men, but males are at their leanest during puberty and young adulthood and then start to experience gains in AVF, probably due to declining testosterone and GH concentrations.

Based on the studies discussed above and on our previous investigations^21,28 that demonstrated smaller relationships of AVF than total adiposity with hormone release, we included a block of body fat distribution variables (AVF, extramyocellular fat, subcutaneous fat) as secondary predictors in the hierarchical model. In agreement with our hypothesis, after holding differences in total adiposity constant, body fat distribution added an independent amount of predictive value to HOMA values and fasting serum insulin concentrations. To the best of our knowledge these are the first data demonstrating an independent relationship of extramyocellular fat and other body fat distribution variables with fasting insulin in children and adolescents. Inverse relationships between muscle lipid and insulin sensitivity have been reported for obese nondiabetic adults^11,15 and are corroborated by muscle biopsy studies, which demonstrated inverse relationships between muscle triglyceride and insulin sensitivity.^13,14,16 The extramyocellular fat measured in the present study is not the same depot as intramyocellular triglyceride. The intramyocellular triglyceride is probably the contributing factor to insulin resistance,^42,43 but in most individuals amounts of intra- and extramyocellular triglyceride are probably highly related and therefore indicative of insulin resistance. Potential reductions in insulin sensitivity due to muscular lipids are a serious concern because muscle is the primary site for insulin-stimulated glucose disposal under euglycemic conditions.^18,19 The inverse relationship between extramyocellular fat and fasting serum insulin concentrations in healthy normal weight youth provides evidence that the accumulation of muscular triglyceride may be a precursor of type 2 diabetes mellitus and not simply a metabolic complication of the disease.¹⁴

In contrast to our hypothesis, mean nocturnal GH release was not related to HOMA values or fasting serum insulin concentrations (Tables 3 and 4). We hypothesized that GH would increase insulin resistance through GH's antagonistic effects on insulin action.^25,44,45,46 Similar to the present study (Table 3), Cook et al²⁴ found that serum IGF-I concentrations were directly related to insulin resistance while GH release was not. Furthermore, others have reported direct relationships between increasing IGF-I and insulin resistance measures^4,23,47 during puberty. Although the insulin antagonistic effects of GH may be mediated through its downstream stimulation of IGF-I, this is unlikely because exogenous IGF-I reduces insulin requirements in type 1 diabetics.⁴⁸

Adolescence is a critical time for the development of obesity⁵ and pubertal insulin resistance may add to the risk of developing obesity and type 2 diabetes mellitus but it may also benefit the growth process by amplifying the effects of GH and IGF-I. Pubertal insulin resistance occurs for carbohydrate, but not for protein or fat metabolism resulting in an enhanced insulin effect on protein anabolism during the rapid growth of adolescence.⁶ Furthermore, hyperinsulinemia reduces IGF binding protein-I resulting in greater amounts of biologically active free-IGF-I.²⁵ Insulin feeds back to the hypothalamus to regulate food intake.⁴⁹ Insulin resistance, then, may also help promote growth by ensuring sufficient energy and nutrient intake due to a reduction in the satiating signals of insulin while the peripheral anabolic effects for protein and fat are maintained.

Serum concentrations of leptin, another satiety hormone, were positively related to HOMA values and fasting serum insulin concentrations but did not reliably increase r² after adjusting for total and regional adiposity and physical characteristics (Tables 3 and 4). We hypothesized that pubertal hyperinsulinemia was, in part, a function of leptin resistance. The adipose tissue and the pancreas are functionally connected through the adipoinsular axis.²⁰ Food intake stimulates insulin secretion and within the adipocyte, lipogenesis and leptin production increase. Leptin then feeds back to the pancreatic islets to reduce insulin secretion. A direct inhibitory effect of leptin on insulin secretion has been found in isolated human²⁰ and rodent⁵⁰ islets. During puberty, leptin resistance may occur at the islets²⁰ so that insulin secretion remains elevated. Although we did not find strong support for this hypothesis in the modest insulin resistance that occurs in normal weight youth at puberty, leptin resistance may modulate insulin sensitivity in overt obesity.

In conclusion, in boys and girls with normal amounts of adiposity, HOMA values and fasting serum insulin concentrations were greater in the late-pubertal group than in the pre- and mid-pubertal groups, suggesting pubertal insulin resistance. Increased total body fat was the primary and body fat distribution (extramyocellular fat, AVF, subcutaneous fat) an independent secondary adiposity variable related to increased fasting insulin during puberty. Normal pubertal increases in FM, extramyocellular fat and AVF may be responsible for the reduction in insulin sensitivity during puberty. Future studies should longitudinally investigate the alterations in body composition, body fat distribution, and hormone release at puberty as such studies would more easily separate the effects of pubertal maturation and body fat on insulin resistance.

Acknowledgements

The authors are indebted to Ms Sandra Jackson and the nursing staff at the University of Virginia General Clinical Research Center who provided patient care and Katy Nash, MS and Catherine DeGood for study coordination. We thank Bret Goodpaster PhD for his insightful suggestions on an earlier version of this paper. We acknowledge the subjects for their enthusiasm for the research program. This work was supported in part by grants from the National Institutes of Health HD 32631 (ADR), General Clinical Research Center grants MO1 RR00847 (University of Virginia), the Genentech Foundation for Growth and Development (PAC), and the University of Virginia Children's Medical Center (JNR).

1 Bloch CA, Clemons P, Sperling MA. Puberty decreases insulin sensitivity. J Pediatr 1987; 110: 481-487. MEDLINE

2 Moran A, Jacobs DR, Steinberger J, Hong C-P, Prineas R, Luepker R, Sinaiko AR. Insulin resistance during puberty. Results from clamp studies in 357 children. Diabetes 1999; 48: 2039-2044. MEDLINE

3 Potau N, Ibanez L, Rique S, Carrascosa A. Pubertal changes in insulin secretion and peripheral insulin sensitivity. Horm Res 1997; 48: 219-226. MEDLINE

4 Caprio S, Plewe G, Diamond MP, Simonson DC, Boulware SD, Sherwin RS, Tamborlane WV. Increased insulin secretion in puberty: a compensatory response to reductions in insulin sensitivity. J Pediatr 1989; 114: 963-967. MEDLINE

5 Dietz WH. Critical periods in childhood for the development of obesity. Am J Clin Nutr 1994; 59: 955-959. MEDLINE

6 Amiel SA, Caprio S, Sherwin RS, Plewe G, Haymond MW, Tamborlane WV. Insulin resistance of puberty: a defect restricted to peripheral glucose metabolism. J Clin Endocrinol Metab 1991; 72: 277-282. MEDLINE

7 Caprio S, Hyman LD, Limb C, McCarthy S, Lange R, Sherwin RS, Shulman G, Tamborlane WV. Central adiposity and its metabolic correlates in obese adolescent girls. Am J Physiol 1995; 269: E118-E126. MEDLINE

8 Caprio S, Hyman LD, McCarthy S, Lange R, Bronson M, Tamborlane WV. Fat distribution and cardiovascular risk factors in obese adolescent girls: importance of the intrabdominal fat depot. Am J Clin Nutr 1996; 64: 12-17. MEDLINE

9 Caprio S, Tamborlane WV, Silver D, Robinson C, Leibel R, McCarthy S, Grozman A, Belous A, Maggs D, Sherwin RS. Hyperleptinemia: an early sign of juvenile obesity. Relations to body fat depots and insulin concentrations. Am J Physiol 1996; 271: E626-E630. MEDLINE

10 Brambilla P, Manzoni P, Sironi S, Simone P, Del Maschio A, di Natale B, Chiumello G. Peripheral and abdominal adiposity in childhood obesity. Int J Obes Relat Metab Disord 1994; 18: 795-800. MEDLINE

11 Goodpaster BH, Thaete FL, Simoneau JA, Kelley DE. Subcutaneous abdominal fat and thigh muscle composition predict insulin sensitivity independently of visceral fat. Diabetes 1997; 46: 1579-1585. MEDLINE

12 Abate N, Garg A, Peshock RM, Stray-Gundersen J, Grundy S. Relationships of generalized and regional adiposity to insulin sensitivity in men. J Clin Invest 1995; 96: 88-98. MEDLINE

13 Pan DA, Lillioja S, Kriketos AD, Milner MR, Baur LA, Bogardus C, Jenkins AB, Storlein LH. Skeletal muscle triglyceride levels are inversely related to insulin action. Diabetes 1997; 46: 983-988. MEDLINE

14 Phillips DIW, Caddy S, Ilic V, Fielding BA, Frayn KN, Borthwick AC, Taylor R. Muscle triglyceride and muscle insulin sensitivity: evidence for a relationship in nondiabetic subjects. Metabolism 1996; 45: 947-950. MEDLINE

15 Simoneau JA, Colberg SR, Thaete FL, Kelley DE. Skeletal muscle glycolytic and oxidative enzyme capacities are determinants of insulin sensitivity and muscle composition in obese women. FASEB J 1995; 9: 273-278. MEDLINE

16 Storlien LH, Jenkins AB, Chrisholm DJ, Pascoe WS, Khouri S, Kraegen EW. Influence of dietary fat composition on development of insulin resistance in rats. Relationship to muscle triglyceride and -3 fatty acids in muscle phospholipid. Diabetes 1991; 40: 280-289. MEDLINE

17 Caprio S, Bronson M, Sherwin RS, Rife F, Tamborlane WV. Co-existence of severe insulin resistance and hyperinsulinaemia in pre-adolescent obese children. Diabtologia 1996; 39: 1489-1497.

18 DeFronzo RA, Tobin JD, Andres R. Glucose clamp technique: a method for quantifying insulin secretion and resistance. Am J Physiol 1979; 237: E214-E233. MEDLINE

19 Shulman GI, Rothman DL, Jue T, Stein P, DeFronzo RA, Shulman RG. Quantitation of muscle glycogen synthesis in normal subjects with non-insulin-dependent diabetes by ¹³C nuclear magnetic resonance spectroscopy. New Engl J Med 1990; 322: 223-228. MEDLINE

20 Fehmann H-C, Berghofer P, Brandhorst D, Brandhorst H, Hering B, Bretzel RG, Goke B. Leptin inhibition of insulin secretion from isolated human islets. Acta Diabetol 1997; 34: 249-252. Article MEDLINE

21 Roemmich JN, Clark PA, Berr S, Mai V, Mantzoros C, Weltman A, Flier JS, Rogol AD. Gender differences in leptin levels during puberty are related to the subcutaneous fat depot and sex steroids. Am J Physiol 1998; 275: E543-E551. MEDLINE

22 Bareille P, Azcona C, Matthews DR, Conway GS, Stanhope R. Lipid profile, glucose tolerance, and insulin sensitivity after more than four years of growth hormone therapy in non-growth hormone deficient adolescents. Clin Endocrinol 1999; 51: 347-353.

23 Amiel SA, Sherwin RS, Simonson DC, Lauritano AA, Tamborlane WV. Impaired insulin action in puberty. New Engl J Med 1986; 315: 215-219. MEDLINE

24 Cook JS, Hoffman RP, Stene MA, Hansen JR. Effects of maturational stage on insulin sensitivity during puberty. J Clin Endocrinol Metab 1993; 77: 725-730. MEDLINE

25 Caprio S. Insulin: the other anabolic hormone of puberty. Acta Paediatr 1999; 433: ((Suppl)) 84-87.

26 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: 412-419. MEDLINE

27 Tanner JM. Growth at adolescence Blackwell Scientific: Oxford, 1962.

28 Roemmich JN, Clark PA, Mai V, Berr S, Weltman A, Veldhuis JD, Rogol AD. Alterations in growth and body composition during puberty: III. Influence of maturation, gender, body composition, fat distribution, aerobic fitness, and energy expenditure on nocturnal growth hormone release during puberty. J Clin Endocrinol Metab 1998; 83: 1440-1447. MEDLINE

29 Veldhuis JD, Johnson ML. Cluster analysis: a simple, versatile, and robust algorithm for endocrine detection. Am J Physiol 1986; 250: E486-E493. MEDLINE

30 Lohman TG. Advances in body composition assessment Human Kinetics: Champaign, IL, 1992.

31 Roemmich JN, Clark PA, Weltman A, Rogol AD. Alterations in growth and body composition during puberty: I. Comparison among 2-, 3- and 4-compartment models of body composition. J Appl Physiol 1997; 83: 927-935. MEDLINE

32 Harrison GG, Buskirk ER, Lindsay Carter JE, Johnston FE, Lohman TG, Pollack ML, Roche AF, Wilmore JH. 1988 Skinfold thicknesses and measurement technique. In: Lohman TG, Roche AF, Martorell R (eds). Anthropometric standardization reference manual Human Kinetics: Champaign, IL, 1988, 55-70.

33 Radziuk J. Insulin sensitivity and its measurement: structural commonalities among the methods. J Clin Endocrinol Metab 2000; 85: 4426-4433. MEDLINE

34 Petrie JR, Morris AD, Dorrian CA, Small M, Connell JM. Specific insulin assays, insulin sensitivity and blood pressure. Q J Med 1997; 90: 465-475.

35 Gower BA, Nagy TR, Goran MI. Visceral fat, insulin sensitivity, and lipids in prepubertal children. Diabetes 1999; 48: 1515-1521. MEDLINE

36 Arslanian S, Suprasongsin C. Insulin sensitivity, lipids, and body composition in childhood: is 'syndrome X' present? J Clin Endocrinol Metab 1996; 81: 1058-1062. MEDLINE

37 Jiang X, Srinivasan SR, Berenson GS. Relation of obesity to insulin secretion and clearance in adolescents: the Bogalusa Heart Study. Int J Obes Relat Metab Disord 1996; 20: 951-956. MEDLINE

38 Radetti G, Ghizzoni L, Paganini C, Iughetti L, Caselli G, Bernasconi S. Insulin pulsatility in obese and normal prepubertal children. Horm Res 1998; 50: 78-82. MEDLINE

39 Smith CP, Dunger DB, Williams AJK, Taylor AM, Perry LA, Gale EAM, Preece MA, Savage MO. Relationship between insulin, insulin-like growth factor I, and dehydroepiandrosterone sulfate concentrations during childhood, puberty, and adult life. J Clin Endocrinol Metab 1989; 68: 932-937. MEDLINE

40 Pouliot M-C, Despres J-P, Lemieux S, Moorjani S, Bouchard C, Tremblay A, Nadeau A, Lupien PJ. Waist circumference and abdominal sagittal diameter: best simple anthropometric indexes of abdominal visceral adipose tissue accumulation and related cardiovascular risk in men and women. Am J Cardiol 1994; 73: 460-468. MEDLINE

41 Gower BA, Nagy TR, Trowbridge CA, Dezenberg C, Goran MI. Fat distribution and insulin response in prepubertal African American and white children. Am J Clin Nutr 1998; 67: 821-827. MEDLINE

42 Boden G, Lebed B, Schatz M, Homko C, Lemieu S. Effects of acute changes of plasma free fatty acids on intramyocellular fat content and insulin resistance of healthy subjects. Diabetes 2001; 50: 1612-1617. MEDLINE

43 Perseghin G, Scifo P, De Cobelli F, Pagliato E, Battezzati A, Arcelloni C, Vanzulli A, Testolin G, Pozza G, Del Maschio A, Luzi L. Intramyocellular triglyceride content is a determinant of in vivo insulin resistance in humans: a ¹H-¹³C nuclear magnetic resonance spectroscopy assessment in offspring of type 2 diabetic parents. Diabetes 2001; 48: 1600-1606.

44 Arslanian SA, Heil BV, Becker DJ, Drash AL. Sexual dimorphism in insulin sensitivity in adolescents with insulin-dependent diabetes mellitus. J Clin Endocrinol Metab 1991; 72: 920-926. MEDLINE

45 Heptulla RA, Boulware SD, Caprio S, Silver D, Sherwin RS, Tamborlane WV. Decreased insulin sensitivity and compensatory hyperinsulinemia after hormone treatment in children with short stature. J Clin Endocrinol Metab 1997; 82: 3234-3238. MEDLINE

46 Seminara S, Merello G, Masi S, Filpo A, La Cauza F, D'Onghia G, Martelli E, Loche S. Effect of long-term growth hormone treatment on carbohydrate metabolism in children with growth hormone deficiency. Clin Endocrinol 1998; 49: 125-130.

47 Travers SH, Jeffers BW, Blocj CA, Hill JO, Eckel RH. Gender and Tanner stage differences in body composition and insulin sensitivity in early pubertal children. J Clin Endocrinol Metab 1995; 80: 172-178. MEDLINE

48 Clemmons DR, Moses AC, McKay MJ, Sommer A, Rosen DM, Ruckle J. The combination of insulin-like growth factor I and insulin-like growth factor binding protein-3 reduces insulin requirements in insulin-dependent type 1 diabetes: evidence of in vivo biological activity. J Clin Endocrinol Metab 2000; 85: 1518-1524. MEDLINE

49 Schwartz MW, Baskin DG, Kaiyala KJ, Woods SC. Model for the regulation of energy balance and adiposity by the central nervous system. Am J Clin Nutr 1999; 69: 584-596. MEDLINE

50 Emilsson V, Liu YL, Cawthorne MA, Morton NM, Davenport M. Expression of the functional leptin receptor mRNA in pancreatic islets and direct inhibitory action of leptin on insulin secretion. Diabetes 1997; 46: 313-316. MEDLINE

Figure 1 Relationship between homeostasis model assessment of insulin resistance and bone age of the boys and girls.

Table 1 Physical charcteristics of the subject groups

Table 2 Serum hormone and glucose concentrations of the subject groups

Table 3 Univariate correlation coefficients of HOMA values and fasting serum insulin concentrations with predictor variables and hierarchical regression of physical characteristic, body composition, body fat distribution, and hormone variables on HOMA values and fasting serum insulin concentration

Table 4 Univariate correlation coefficients of HOMA values and fasting serum insulin concentrations with predictor variables and hierarchical regression models of percentage adiposity, percentage body fat distribution, and hormone variables on HOMA values and fasting serum insulin concentration