Prevalence and Concomitants of Glucose Intolerance in European Obese Children and Adolescents

We studied a cohort of 710 obese children and adolescents of European origin by at least two generations. Subjects were referred by their general practitioner or their primary care pediatric consultant to the Istituto Auxologico Italiano, a specialized center for the study of obesity, between 1994 and 2001. They were admitted for metabolic evaluation and clinical management for 2 weeks. There were 345 male and 365 female subjects with ages ranging between 6 and 18 years. This article reports the baseline cross-sectional data of this population that is being followed-up for prospective evaluation of predictors of risk for metabolic and CVD. Patients with secondary obesity syndromes and acute illnesses were excluded from the study. Anthropometric measures were recorded. Duration of obesity, birth weight, and family history for obesity and diabetes were obtained by questionnaires filled in by the mother. Body weight was recorded to the nearest 0.1 kg using a standard beam balance scale, with the subjects wearing light indoor clothing and no shoes. Height was recorded to the nearest 0.5 cm using a standardized wall-mounted height board. The degree of obesity was quantified using Cole’s least mean square method, which normalizes the BMI skewed distribution and expresses BMI as an SD score (SDS). This measure gives ages and sex-specific estimates of the distribution median (M), coefficient of variation (S), and the degree of skewness (L) by a maximum-likelihood fitting technique (16). Visceral obesity was assessed by waist circumference measured at the level of the umbilicus and the superior iliac crest. The measurement was made at the end of a normal expiration while the subjects were in a standing position. Pubertal development was assessed by physical examination according to the criteria of Tanner (17). Diastolic and systolic blood pressure (BP) were measured to the nearest 2 mmHg in the supine position after 5 min rest, using a standard mercury sphygmomanometer with an appropriately sized cuff (18). The average of six measurements obtained on different days during admission was used in the analysis.

After a 12-h overnight fast, subjects underwent an oral glucose tolerance test (1.75 g/kg, up to a maximum of 75 g glucose in 250 ml of water). Plasma samples were drawn at baseline and after 30 and 120 min for determination of plasma glucose and insulin concentration, and categorization of glucose tolerance status was made using the World Health Organization criteria (19).

Insulin resistance was estimated using the homeostasis model assessment for insulin resistance (HOMA_IR; fasting insulin × fasting glucose/22.5) (20). β-Cell function was determined by the insulinogenic index (ΔI30/ΔG30), expressed as the ratio of the incremental (0–30 min) insulin and glucose response after commencement of oral glucose intake during the oral glucose tolerance test (21). Fasting blood samples were taken for lipids, uric acid, fibrinogen, and C-reactive protein (CRP) measurements.

Plasma glucose was measured by the glucose oxidase method, and total cholesterol, HDL cholesterol, triglycerides, and uric acid concentrations were measured by a colorimetric method on an automated analyzer (Roche Diagnostics, Mannheim, Germany). CRP concentrations were measured by an immunoturbidimetric assay (Roche Diagnostics), with a sensitivity of 0.2 mg/dl and intra- and interassay CVs of 1.5 and 7.2%, respectively. The normal range of CRP using this method in our laboratory is 0–1 mg/dl. Fibrinogen was measured in citrate plasma with a clot-rate assay using the ACL 200/IL instrument (Instrumentation Laboratory, Milan, Italy). The sensitivity of the assay was 7.5 mg/dl, and the intra- and interassay CVs were 4.8 and 5.2%, respectively, at a concentration of 275 mg/dl. Serum insulin was measured in duplicate by a chemiluminescent assay (DPC, Los Angeles, CA), with a sensitivity of 14.3 pmol/l and intra- and interassay CVs of 3.7 and 6.7%, respectively, at a concentration of 380 pmol/l.

Differences between groups and univariate regression analyses were calculated using both parametric and nonparametric tests depending on the distribution of the variables. Multivariate analysis was performed using variables statistically significant at the 5% level in univariate analysis. Closely interrelated variables were entered separately in these models. The 2-h postload glucose and SDS-BMI, as a measure of the degree of obesity, were the dependent variables in this analysis. To assess the effect of puberty on inflammatory and CVD markers, the population was divided into the five Tanner stages, and ANCOVA was performed using SDS-BMI and age as covariates. Between-group differences were assessed by the post hoc least significant difference test for multiple comparisons. A probability value <0.05 was considered significant. Data are given as the means ± SD unless otherwise stated. SPSS version 10.1 (SPSS, Chicago, IL) was used for analysis.

The demographic, clinical, and biochemical characteristics of the 710 obese children and adolescents are summarized in Table 1. The duration of obesity was apparently 7 years. Of the obese children, 61% had a positive family history for obesity and 54% for diabetes, with no differences between sexes. Compared with girls, boys were more obese but had similar birth weights. Lipid profiles and CRP were similar between the sexes, whereas uric acid was significantly higher in boys and fibrinogen in girls.

A substantial proportion of these obese children had abnormal parameters for their age (Table 2). Systolic BP was significantly higher in boys than in girls and was above the age-related 95th percentile adjusted for height in 29% of girls and 25% of boys. Diastolic BP exceeded the 95th percentile for height in 13% of boys and 17% of girls (Fig. 1).

When cutoff values for the adult populations were used, as published by recent Adult Treatment Panel III clinical guidelines (22), the percentage of abnormality was 10% for triglycerides, 52% for HDL cholesterol, and 15% for BP.

Of the obese children, 1 had diabetes, 30 had IGT, and 3 had impaired fasting glucose, 2 of whom also had IGT, to give an overall prevalence of glucose intolerance of 4.5%. Compared with children with normal glucose tolerance, children with IGT had significantly higher fasting blood glucose (BG; 4.8 ± 0.67 vs. 4.5 ± 0.43 mmol/l, P < 0.0001) and 2-h postload insulin levels (median [interquartile range] 711.4 pmol/l [546–906] vs. 414.0 pmol/l [270–576], P < 0.0001). Family history for diabetes and obesity were also more frequent in glucose-intolerant children, but not significantly so (62 vs. 53% and 71 vs. 61%, respectively; NS). The male-to-female ratio, age, fasting insulin levels, HOMA_IR, lipids, fibrinogen, CRP, BP, SDS-BMI, and birth weight were similar between the two groups. The proportion of children at Tanner II-IV stages was also comparable between glucose-intolerant and glucose-tolerant children (22 vs. 37%, NS). When 2-h BG was considered as a continuous variable in the whole population of obese children, it was found to be positively correlated, in univariate analysis, with HOMA_IR (r = 0.141, P < 0.0001) and negatively with ΔI30/ΔG30 (r = −0.159, P < 0.0001). A weak relationship was also found with diastolic BP (r = 0.097, P < 0.01) and triglycerides (r = 0.081, P < 0.05) (Table 3), but there was no association with SDS-BMI and birth weight. In a multivariate regression analysis with 2-h postload glucose values as the dependent variable, only ΔI30/ΔG30, HOMA_IR, and diastolic BP remained significantly and independently related (Table 4). Of the obese children, 64% had an HOMA_IR >2.5, which is the cutoff level to distinguish normal from impaired insulin sensitivity in adults (20). HOMA_IR was univariately correlated with lipids, BP, 2-h BG and ΔI30/ΔG30, SDS-BMI, and uric acid (Table 3).

The degree of obesity expressed as SDS-BMI was related, in univariate analysis, with inflammatory markers, lipids, BP, HOMA_IR, ΔI30/ΔG30, birth weight, and uric acid (Table 3). There was no correlation between SDS-BMI and 2-h BG.

In a multivariate regression analysis with SDS-BMI as the dependent variable, triglycerides, HOMA_IR, and diastolic BP were excluded. (Table 4). The variables independently correlated with SDS-BMI together explained 19% of the variance.

Of the obese children, 20, 29, and 51% were at Tanner stages I, II-IV, and V, respectively. After adjustment for SDS-BMI and age, the degree of insulin resistance progressively increased from Tanner I to Tanner IV pubertal stage and subsequently decreased at Tanner stage V. HDL cholesterol, triglycerides, and systolic BP showed a pattern of change consistent with that of the degree of insulin sensitivity (Fig. 2), whereas diastolic BP and uric acid tended to progressively increase and total cholesterol decrease throughout puberty (Tanner stage I to V: 73.9 ± 1.0 to 78.5 ± 0.80 mmHg, P < 0.01; 340 ± 15.60 to 370.2 ± 9.0 μmol/l, NS; 4.7 ± 0.10 to 4.1 ± 0.06 mmol/l, P < 0.0001; respectively). Fibrinogen and CRP remained stable throughout (data not shown).

This study describes the clinical, metabolic, and biochemical features of a large, recent cohort of grossly obese children, all of European origin. Boys were more obese than girls, in line with the observation that the prevalence of obesity among U.S. adolescents has increased more among boys than among girls (23). A positive family history for obesity and diabetes was present in more than half of the children, as was also reported in a smaller group of obese children (24), and the prevalence of insulin resistance was high. IGT was present in 4.5% of these children, a prevalence that is lower than that previously reported by other authors in obese children of Caucasian origin (11–15). The reasons for these differences are not entirely clear, but methodological and calendar effects could be important. A number of previous reports used a different definition of IGT, which would favor detection of higher prevalences (25), and some observations date back several decades (11–13). The prevalence of IGT in our study was also significantly lower than the >20% recently reported in American obese children and adolescents from a multiethnic background (6). The frequency of diabetes in that population was 4% among the adolescents, which contrasted with 0.2% in our subgroup of 609 adolescents (11–18 years of age). Racial differences, lifestyle, and dietary habits may account for the disparity of results between our present cohort and the obese American children. In our cohort of grossly obese children, postload BG levels were independently correlated with a decrease in both insulin sensitivity and insulin secretion. This means that even before the development of IGT, both insulin resistance and insulin secretory dysfunction contribute to raised glucose levels. Two longitudinal studies, in Mexican-American and Pima Indians, have found that primary defects in both insulin action and secretion are responsible for the transition from normal glucose tolerance to IGT and from IGT to overt diabetes (7,8). We observed that similar defects are present in obese children of an ethnic group at significantly lower risk for type 2 diabetes. The recent report by Sinha et al. (6) in the obese American children and adolescents found a slightly smaller contribution of β-cell dysfunction. Furthermore, the children in our study showed a positive and independent correlation between postchallenge glucose and some CVD risk factors, such as diastolic BP, confirming in obese children the increasing cardiovascular risk with increasing glycemia observed in adults (26,27).

A substantial proportion of children had abnormal values for lipids, uric acid, CRP, fibrinogen, and BP, factors that are often interrelated. The true significance of these abnormal frequencies is limited by the use of literature-derived normal ranges (28). A significant percentage of abnormalities, however, were found even when cutoff values for adults were used. These results are in accord with previous smaller observations (29–35) and, for the first time in a large cohort with a wide age range, indicate that many obese children have, like obese adults, elevated levels of CRP and uric acid, factors that, in adulthood, are related to an increased risk of CVD and diabetes (36–41).

In the present study, 25% of boys and 29% of girls had elevated values of systolic BP, and ∼15% showed elevated levels of diastolic pressure, a prevalence that is lower than that of 40% previously reported in a Hungarian group of obese children (15). The independent relationship of systolic BP with the degree of obesity and of diastolic BP with 2-h glycemia suggests that the level of BP in the obese child is affected by factors that may have differential effects on either systolic or diastolic BP.

In univariate analysis, HOMA_IR and obesity were interrelated and were in turn associated with a variety of other cardiovascular risk factors. However, the relationship of the degree of obesity with cardiovascular risk factors was independent of and not affected by HOMA_IR (or 2-h BG), strongly suggesting that obesity per se, separately from insulin sensitivity or glucose intolerance, impacts on CVD risk.

A low birth weight, particularly if followed by a rapid weight gain, has been associated with an increased risk of insulin resistance, type 2 diabetes, hypertension, and atherosclerosis in adult life (42). This does not seem to be the case in grossly obese children, because those who developed glucose intolerance had similar birth weights compared with those who did not. Furthermore, birth weights were positively correlated with the degree of obesity. This finding agrees with the results of recent epidemiological studies that have shown that the incidence of obesity rises with increasing birth weight (43) and that adolescent BMI is predicted by birth weight (44).

The changes in insulin sensitivity during progression through puberty were independent of the degree of obesity and associated with parallel and consistent changes in triglycerides, HDL cholesterol, and systolic BP, whereas uric acid and diastolic BP showed a physiological age-dependent increase. The pattern of change in insulin sensitivity and lipids was similar to that reported in normal-weight children (45,46), and the changes in systolic BP are first described here. It was interesting to note that at Tanner stage IV, 12.2% of subjects had triglyceride levels above the 95th percentile of distribution. This percentage fell to 6.9% at Tanner stage V. The prognostic significance of this observation remains to be established.

From a cross-sectional study, we cannot tell whether the abnormalities seen in these obese children will impact on their long-term health; however, many of the variables measured in this study track with age and are likely to increase the risk for metabolic and CVDs in adulthood. The few studies that have examined the long-term effects of childhood or adolescent obesity on adult morbidity and mortality have observed an increased occurrence of CVDs and digestive and metabolic diseases among those subjects who were obese during adolescence (47–49).

In conclusion, we have found, in a large cohort of grossly obese children of European origin, that the prevalence of glucose intolerance is low compared with that observed in the U.S., and CVD risk factors were independently related to the level of postload glycemia and the degree of obesity, but they were not associated with low birth weight.

This work was supported by Grant ICS030.11/RS00138 from the Italian Ministry of Health.

We thank Dr. Stefano Omboni for statistical advice.