High Birth Weight and Early Postnatal Weight Gain Protect Obese Children and Adolescents From Truncal Adiposity and Insulin Resistance: Metabolically healthy but obese subjects? Free

Between February 2001 and September 2004, this cross-sectional study included 117 children and adolescents (69 girls and 48 boys), aged 6–15 years, who were referred to the Pediatric Endocrinology Unit for obesity (BMI over +2 SD score for age and sex [17], which corresponds to the cutoff for overweight from the International Obesity Task Force [18]). Exclusion criteria were maternal diabetes (either preexisting to the index pregnancy or developed during or after the index pregnancy), since the genetic factors associated with maternal diabetes may be inherited and influence later insulin sensitivity in the offspring. Other exclusion criteria were obesity of genetic or endocrine origin, diabetes, and the use of any medication. In addition, children whose retrospective data were not fully available were excluded. Gestational age was obtained from early ultrasound scan during pregnancy for all the children included in the study. Birth weight and length and weight and height at 2 and 4 years are mandatory information in the French health care system and were retrospectively obtained from the individual health book, a record of growth measurements, vaccinations, and significant health-related events for every child in France (19). Weight at birth was converted into percentiles and SD score for age and sex according to the French customized birth weight curves, Association of Users of Computerized Medical Records in Paediactrics, Obstetrics, and Gynaecology (AUDIPOG) (14), which accounted for maternal age, weight before pregnancy, height, parity, gestational age, and sex. Current weight and height and pubertal stage were assessed. The children were selected from an initial cohort of 380 children and adolescents seen in our clinic for obesity during this 3.6-year period; 295 children met the inclusion and exclusion criteria. The children were assigned to one of three groups according to their fetal growth category, with a goal of including at least 30 children into each category; the birth weight distribution in this population allowed for the inclusion of 33 subjects with customized LBW (LBWcust, birth weight ≤10th percentile), 32 with customized HBW (HBWcust, birth weight ≥90th percentile), and 52 eutrophic subjects (birth weight ≥10th percentile and ≤90th percentile; recruitment of eutrophic subjects was stopped thereafter). Mean age and BMI, as well as sex and pubertal stage distributions, were similar between the selected and unselected children. The study was approved by the ethics committee of the University of Angers. Written informed consent was obtained from all children and their parents.

After a 12-h overnight fast, an antecubital catheter was inserted in a peripheral vein for blood sampling. An OGTT was performed with the administration of 1.75 g glucose/kg body wt (maximal dose 75 g). Blood samples were drawn at −30, 0, 30, 60, 90, and 120 min for measurements of glucose, insulin, and FFA concentrations. Blood samples were also obtained for measurement of fasting levels of adiponectin, leptin, and visfatin.

Body composition was investigated by dual-energy X-ray absorptiometry using a QDR 4500A densitometer (Hologic, Waltham, MA) (20). Whole-body scans were performed, and body compartments were analyzed using software from Hologic (version V8.24a:3). To assess the effects of fat distribution, the body was divided into four areas: arms, legs, trunk, and head. The trunk region was bounded by a horizontal line below the chin, two vertical lines along the edges of the rib cage, and oblique lines through the hips; the leg region was all tissue below these oblique lines. Body fat distribution was defined as trunk-to-leg fat-mass ratio calculated as the ratio of the amount of fat tissue in the trunk region to the amount of fat tissue in the leg region.

The homeostasis model assessment of insulin resistance index was calculated (16). Three insulin sensitivity indexes, the composite whole-body insulin sensitivity index (WBISI), the hepatic insulin resistance index, and the skeletal muscle insulin sensitivity index, were calculated from OGTT as previously validated (15,16).

WBISI = 10000/√[(fasting glucose in mmol/l × 18.182 × fasting insulin in μUI/ml) × (mean glucose_OGTT × 18.182 × mean insulin _OGTT)]. This index showed good correlation with the insulin sensitivity measurement from the euglycemic-hyperinsulinemic clamp in obese children (r = 0.78, P < 0.0005) (21) and adults (r = 0.73, P < 0.0001) (16).

Hepatic insulin resistance index = (glucose_0–30OGTT [AUC][mmol/l × min⁻¹] × 18.182) × insulin _{0–30 OGTT} [AUC][μUI/ml × min⁻¹]/1,000,000. This index showed good correlation with the hepatic insulin resistance measurement from the euglycemic-hyperinsulinemic clamp using tritiated glucose in obese subjects (r = 0.65, P < 0.0001) (15).

Muscle insulin sensitivity index = 100 × (18.182 × dG/dt [mmol/l × min⁻¹]) /mean insulin _OGTT in μUI/ml, where dG/dt is the rate of decline in plasma glucose concentration and is calculated as the slope of the least square fit to the decline in plasma glucose concentration from peak to nadir. This index showed good correlation with the muscle insulin sensitivity measurement from the euglycemic-hyperinsulinemic clamp in obese subjects with normal glucose tolerance (r = 0.78, P < 0.0001) (15).

Plasma glucose was measured with a Hitachi 917 analyzer (Roche-Diagnostics, Meylan, France). Plasma FFAs were measured using enzymatic methods (Wako Chemical, Richmond) on a Cobas Mira analyzer (Roche-Diagnostics). Plasma insulin was measured by radioimmunoassay (RIA) (Schering AG CIS-Bio International, Gif sur Yvette, France). The intra- and interassay coefficients of variation (CVs) were 1.9 and 2.7%, respectively. Serum leptin was measured by RIA (Mediagnost, Tubingen, Germany). Sensitivity was 0.04 μg/l, and the intra- and interassay CVs were 5 and 7.6%, respectively. Serum adiponectin was determined by RIA (Linco Research, St. Charles, MO). The intra- and interassay CVs were 3.9 and 8.5%, respectively, and assay sensitivity was 1 μg/ml. Plasma visfatin was determined by enzymatic immunoassay (Phoenix Pharmaceuticals, Belmont, CA). The intra- and interassay CVs were <5 and <10%, respectively, and assay sensitivity was 2 ng/ml.

The variables following a non-Gaussian distribution, as assessed by the Komolgorov-Smirnov test, are presented as median (25th–75th percentiles); other variables are presented as means ± SD. Comparisons between subjects were performed using ANOVA and Fisher's least significant difference post hoc tests for normally distributed variables and nonparametric Kruskal-Wallis and Mann-Whitney U tests for non-normally distributed variables. Comparisons between subjects’ plasma glucose, insulin, and FFA concentrations during OGTT were performed using ANOVA with repeated measures. To analyze the combined contributions of birth weight and postnatal weight gain during three time periods (infant weight gain SD score between 0 and 2 years, preschool weight gain SD score between 2 and 4 years, or later child weight gain SD score after 4 years) to WBISI independently of the current fat mass, several multiple regression analyses were performed. In each regression analysis, the WBISI was the dependent variable, and age, sex, pubertal stage, height SD score, and percent fat were entered as adjusting variables. In the first analysis, independent variables were birth weight category (LBWcust, eutrophic, HBWcust), infant weight gain, and their interactions. In the second analysis, independent variables were birth weight category, preschool weight gain, and their interactions. In the last analysis, independent variables were birth weight category, annualized later child weight gain, and their interactions. Later child weight gain was annualized to account for the different time periods from 4 years to the evaluation between children. Any nonsignificant interaction was removed from the final model. Since plasma glucose, insulin, and FFA concentrations and WBISI were not normally distributed, they were log₁₀ transformed before their use in the parametric tests and their distribution again checked by the Komolgorov-Smirnov test. Significance was defined as P < 0.05. All analyses were performed with the SPSS 14.0 package.

The clinical and biological characteristics of the 117 obese subjects are presented in Table 1 according to the fetal growth category. Mean age, percentage of boys and girls, and percentage of prepubertal and pubertal subjects were similar between groups. Whereas the groups were comparable for the percent body fat, the subjects born with HBWcust had a lower trunk fat percentage than the eutrophic (P < 0.05) and LBWcust subjects (P < 0.05). After adjustment for percent body fat, trunk fat percentage was still lower in HBWcust compared with eutrophic subjects (P < 0.05), whereas it was higher in LBWcust (P < 0.05 vs. eutrophic subjects). Accordingly, the trunk-to-leg fat-mass ratio was significantly lower in HBWcust than eutrophic subjects (P < 0.05), whereas it was higher in LBWcust subjects (P < 0.05 vs. eutrophic subjects), indicating a preferential peripheral and central fat distribution in HBWcust and LBWcust subjects, respectively. Adiponectin concentration was significantly higher in HBWcust than in eutrophic subjects (P < 0.05), whereas no difference was found for leptin and visfatin concentrations between the three groups.

With regard to insulin sensitivity, the WBISI and the hepatic insulin resistance index were respectively higher and lower in HBWcust compared with eutrophic (P < 0.05) and LBWcust subjects (P < 0.05). Muscle insulin sensitivity was similar in HBWcust and eutrophic subjects and lower in LBWcust than in eutrophic and HBWcust subjects (P < 0.05), whereas LBWcust and eutrophic subjects had similar WBISI and hepatic insulin resistance index. During the OGTT, plasma insulin and FFAs were significantly lower in HBWcust than eutrophic and LBWcust subjects (P < 0.05, ANOVA with repeated measures), whereas blood glucose was similar (not shown), thus suggesting a higher sensitivity of adipose tissue to the antilipolytic effect of insulin.

To study the independent effects of birth category and the pattern of postnatal weight gain on WBISI, several regression analyses were performed with age, sex, pubertal stage, height SD score, and percent fat (to account for the current fat mass) as the adjusting variables. Birth category and either weight gain between 0 and 2 years, weight gain between 2 and 4 years, or annualized weight gain after 4 years were used as independent variables, as well as their interactions (Table 2). Birth category was a significant positive predictor (P < 0.05) of WBISI in all analyses, in association with either weight gain between 0 and 2 years, which was a significant positive predictor (P < 0.05) indicating that high early weight gain was associated with higher insulin sensitivity (multiple R = 0.57, P < 0.01), or in association with weight gain after 4 years, which was a significant negative predictor (P < 0.05) indicating that high weight gain after 4 years was associated with lower insulin sensitivity (multiple R = 0.54, P < 0.001) (Table 2 and Fig. 1). Weight gain between 2 and 4 years was not a significant predictor of WBISI.

This study showed that obese children and adolescents born after increased fetal growth exhibited lower central and higher peripheral fat distribution than those born eutrophic, despite similar percentages of body fat mass. Whole-body and hepatic insulin sensitivities, assessed from OGTT (15,16), were significantly higher in HBWcust subjects. Plasma FFA and insulin levels during OGTT were lower in HBWcust children, suggesting a higher sensitivity of adipose tissue to the antilipolytic effect of insulin. Adiponectin levels were significantly higher in HBWcust than eutrophic subjects, further indicating a different fat cell metabolism in these subjects. Birth weight was an independent positive determinant of the WBISI, as was weight gain between 0 and 2 years, whereas weight gain after 4 years was an independent negative determinant of insulin sensitivity, even after adjustment for the current fat percentage. Hence, these findings showed that increased fetal growth and infant weight gain (0–2 years) were associated with relative protection from the development of central obesity and insulin resistance among obese children and adolescents, possibly indicating metabolically healthy obesity.

In children and adolescents, several studies have found a U-shaped relation between birth weight and insulin resistance or type 2 diabetes (5,24). Conversely, we found here that obese children and adolescents born with HBWcust displayed ∼60% higher insulin sensitivity than those born eutrophic. Our findings could be linked with several adult studies that found standardized mortality rates from coronary artery disease ∼50% lower in subjects with HBW compared with those with average birth weights (2,7). Several factors could explain these discrepancies. First, we excluded HBW due to maternal diabetes, since it may promote larger birth size as well as confer a risk of diabetes to offspring. Second, we used customized birth weight standards (which are adjusted for maternal characteristics and gestation number) instead of population-based standards (adjusted only for gestational age and sex) to identify subjects with true altered fetal growth and distinguish them from constitutionally large infants (14). Finally, the pattern of postnatal weight gain may modulate insulin resistance in obese subjects born with HBW, as we found that high weight gain between 0 and 2 years protected from later insulin resistance whereas high weight gain after 4 years increased insulin resistance.

The relationship between birth weight and adult disease is thought to reflect the influence of fetal nutrition on the prenatal programming of processes that affect long-term metabolism (23). Adipose tissue may play a critical role in this process, as shown by several studies in LBW individuals: dramatically reduced at birth, adiposity undergoes a catch-up growth process during infancy and childhood leading to a disproportionately high fat mass in relation to muscle mass (24), a central repartition of fat that contributes to insulin resistance, and a particular adipocytokine profile, namely decreased adiponectin levels, which also favors insulin resistance (25). Accordingly, we found that obese subjects born after true restricted fetal growth had a higher trunk fat percentage and a lower muscle insulin sensitivity index than those born eutrophic. As a mirror image, our obese children born with HBWcust exhibited lower central fat distribution than those born eutrophic despite similar percent fat mass, in agreement with other studies (26). Fasting adiponectin levels and insulin sensitivity indexes were significantly higher in the obese subjects born with HBWcust. In line with these findings, obese insulin-sensitive adolescents have been characterized by greater levels of adiponectin than obese insulin-resistant adolescents and by lower lipid deposition in the intramyocellular lipid and visceral compartments despite a similar degree of adiposity (27). We also found that plasma FFAs and insulin during OGTT were significantly lower in HBWcust than eutrophic subjects, suggesting higher sensitivity of their adipose tissue to the antilipolytic action of insulin. Our results indicate that increased fetal growth and early weight gain induced a peculiar adipose tissue metabolism with increased sensitivity to insulin and increased production of adiponectin, all of which likely contributed to enhanced liver and WBISIs. Conversely, when the weight gain occurred predominantly after 4 years, it favored insulin resistance. This suggests a specific window encompassing antenatal and early postnatal life up to 2 years in which fat accumulation programs adipose tissue to favor insulin sensitivity. In contrast, further fat accumulation at a later age led to insulin resistance.

Not all obese adults display a clustering of metabolic and cardiovascular risk factors, just as not all insulin-resistant adults are obese (28,29). Some metabolically healthy but obese individuals have large quantities of fat but normal to high indexes of insulin sensitivity and favorable cardiovascular risk profiles (28). Although it was suggested that these subjects could be characterized by childhood onset of obesity (30), other studies found an increased risk of cardiovascular disease in subjects who were obese in adolescence (31). These apparent discrepancies may be explained by our findings, since only obese children and adolescents born with HBWcust showed high insulin sensitivity, especially those displaying very early weight gain.