Impaired Glucose Tolerance and Bone Mineral Content in Overweight Latino Children With a Family History of Type 2 Diabetes

A cohort of overweight Latino children with a positive family history of type 2 diabetes was established with the intention to follow them longitudinally in the USC SOLAR diabetes project. The current investigation reports the cross-sectional findings from the first 184 children in this cohort.

A total of 184 children (106 boys, 78 girls) were recruited to the SOLAR project through clinics, health fairs, newspaper announcements, and word of mouth. The children were required to meet the following inclusion criteria: 1) age 8–13 years; 2) BMI ≥85th percentile for age and sex based on the Centers for Disease Control and Prevention (CDC) standards (8) and an initial telephone prescreening; 3) Latino ancestry (all four grandparents Latino by self-report); and 4) family history of type 2 diabetes in at least one parent, sibling, or grandparent. Children were of Mexican-American (71%), Central American (16%), or mixed Mexican–Central American (13%) heritage. Children were excluded if they had prior major illness including type 1 or type 2 diabetes, took medications, or had a condition known to influence body composition, bone mass, insulin action, or insulin secretion (e.g., pregnancy, glucocorticoid therapy, or hyper- or hypothyroidism). The SOLAR study was approved by the Institutional Review Board, Health Science Campus, USC. Informed consent and assent were obtained from all parents and children, respectively.

Children arrived at the USC General Clinical Research Center (GCRC) at ∼8:00 a.m. after an overnight fast. Weight and height were measured followed by a physical examination conducted by a board-certified pediatric endocrinologist (Marc J. Weigensberg). Physical examinations included Tanner staging in girls (breast stage) and boys (pubic hair stage) as well as testicular volume measurements in a subsample (n = 84) of boys. A medical history was conducted including parental interview detailing family history of diabetes. After these procedures, an oral glucose tolerance test was conducted. Children who met the following screening criteria were invited back for further testing during an inpatient GCRC visit: 1) BMI ≥85th percentile for age and sex according to the CDC standards (8) based on height and weight measures at the GCRC and 2) absence of type 1 or type 2 diabetes using the American Diabetes Association guidelines (9). Note that for the purposes of this study, we used the CDC definitions for weight status in children (i.e., at risk of overweight/obesity is a BMI age/sex percentile >85th). All children met the BMI criteria, and none of the 184 children screened positive for diabetes.

Of the 184 children completing the outpatient visit, all returned to the GCRC within 15 ± 10 (± SD) days. Children were admitted to the GCRC in the early afternoon and then completed tests for bone mass and body composition using dual-energy X-ray absorptiometry. Children were served dinner and an evening snack, with only water permitted after 8:00 p.m. The following morning, insulin sensitivity and AIR were determined from an intravenous glucose tolerance test.

Height (by a wall-mounted stadiometer) and weight (by a balance beam medical scale) were recorded at each visit to the nearest 0.1 cm and 0.1 kg, respectively, and the average of the two measurements was used for analysis. BMI and BMI percentiles for age and sex were determined based on established CDC normative curves using EpiInfo 2000, Version 1.1.

A whole-body dual-energy X-ray absorptiometry scan was performed to determine whole-body BMC (units: grams), BMD (units: grams/cm²), and whole-body composition using a Hologic QDR 4500W (Hologic, Bedford, MA). The whole-body scan requires the subject to be placed supine with the arms and legs positioned according to the manufacturer’s specifications. Quality control was performed daily using a phantom, and measurements were maintained within the manufacturers precision standards of ≤1%.

A subsample (n = 146) of participants completed an all-out progressive continuous treadmill protocol (10). After being familiarized with the exercise equipment, the children practiced walking on the motorized treadmill until they were able to walk without holding the railings. Children walked for 4 min at 0% grade and 4 km/h, after which the treadmill grade was raised to 10%. Each ensuing work level lasted 2 min, during which the grade was increased by 2.5%. The speed remained constant until a 22.5% grade was reached, at which time the speed was increased by 0.6 km/h until the subject reached exhaustion. Respiratory gases were collected and measured via open circuit spirometry and analyzed on a MedGraphics CardiO₂ combined Vo₂/ECG Exercise System. Heart rate was measured continuously using a Polar Vantage XL heart rate monitor (Port Washington, NY). Vo_2max was achieved if individuals satisfied two of the following three criteria: respiratory exchange ratio >1.0, heart rate ≥195, and a leveling or plateauing of Vo₂ defined as an increase of oxygen uptake <2 ml · kg⁻¹ · min⁻¹ with a concomitant increasing workload.

A topical anesthetic (EMLA cream; AztraZeneca, Wilmington, DE) was applied to the antecubital area of one arm, and a flexible intravenous catheter was placed in an antecubital vein. Subjects ingested 1.75 g oral glucose solution per kilogram body weight (to a maximum of 75 g) at “time 0.” Blood was sampled and assayed for glucose and insulin at times −5 min (“fasting”) and 120 min (“2 h”) relative to glucose ingestion. IGT was defined by “2-h glucose ≥140 mg/dl,” as recommended by the American Diabetes Association Expert Committee on the Diagnosis and Classification of Diabetes Mellitus (11). AUC was determined with the trapezoidal method (12). Incremental insulin AUC was calculated by dividing the AUC by 180 min and subtracting the fasting insulin value.

At 6:30 a.m., the EMLA cream was applied, followed ∼1 h later by flexible intravenous catheter placement in bilateral antecubital veins. Two fasting blood samples were drawn at −15 and −5 min for determination of basal glucose and insulin. At time 0, glucose (25% dextrose, 0.3 g/kg body wt) was administered intravenously. Blood samples were collected at time points 2, 4, 8, 19, 22, 30, 40, 50, 70, 100, and 180 min. Insulin (0.02 units/kg body wt, Humulin R [regular insulin for human injection]; Eli Lilly, Indianapolis, IN) was injected intravenously at 20 min. Values for glucose and insulin were entered into the Minmod Millenium 2002 computer program (Version 5.16, Richard N. Bergman) for determination of insulin sensitivity and the acute insulin response to glucose (13).

Blood samples during the oral glucose tolerance test were centrifuged, and plasma was placed on ice and analyzed within 1 h at the Los Angeles County USC Medical Center Core Laboratory with a Dimension Clinical Chemistry system using the hexokinase method (Dade Behring, Deerfield, IL). Blood samples from the intravenous glucose tolerance test were centrifuged, and plasma was stored on ice before storage at −80°C. Aliquots were assayed in duplicate for glucose using the glucose oxidase method and a Yellow Springs Instruments 2700 Analyzer (Yellow Springs Instruments, Yellow Springs, OH). Insulin was assayed in duplicate using an enzyme-linked immunosorbent assay kit from Linco (St. Charles, MO).

All analyses were performed using SPSS version 11.0 (SPSS, Chicago, IL), with a type I error set at P < 0.05. Partial correlations (controlling for lean mass, age, and Tanner stage) were performed to determine the associations between BMC and BMD and the independent variables. Children with NGT versus IGT were compared using ANOVA or general linear models when covariates were included. As determined by the Kolmogorov-Smirnov test of normality, BMC and BMD were not normally distributed and were log transformed; multiple linear regression analysis was repeated with log-transformed dependent variables.

A summary of the adolescents’ characteristics based on glycemic control (i.e., IGT vs. NGT) is shown in Table 1. Mean BMC was 1,501.4 ± 410.9 g in IGT and 1,553.5 ± 449.2 g in NGT children; BMC had a range of 748–3,199 g. BMD values ranged from 0.71 to 1.34 g/cm² and were identical (mean of 0.92 g/cm²) in IGT and NGT children. Bone area was 1,600.1 ± 309.1 cm² in IGT and 1,658.1 ± 296.3 cm² in NGT children. There were no significant differences in BMC, BMD, or bone area in IGT versus NGT children when the analysis was repeated separately by sex (data not shown). Mean 2-h glucose and 2-h insulin were significantly greater in IGT than NGT children. There were no other significant differences in groups.

Table 2 reports partial correlations (controlling for lean mass, age, and Tanner stage) between the glucose/insulin variables and BMC and BMD. Fasting and 2-h glucose levels were not significantly correlated with BMC or BMD. Fasting and 2-h insulin levels were both inversely correlated with BMC (r = −0.16) and BMD (r = −0.17). AIR and AUC were also inversely correlated with BMC and BMD. The inverse relationships between insulin AUC and BMC and BMD were stronger than the inverse correlations between incremental insulin AUC and BMC and BMD.

Table 3 reports partial correlations of the independent variables with BMC and BMD based on glucose tolerance. Of the 184 subjects, 138 (75%) had NGT and 46 (25%) had IGT. All significant relationships were negative. In NGT children, the relationships between AIR and BMC and BMD were not significant. In IGT children, however, there were strong inverse correlations between AIR and BMC (r = −0.58, P = 0.000) and BMD (r = −0.51, P = 0.001). Although the relationships between insulin AUC and incremental insulin AUC and BMC and BMD were significant and negative in both NGT and IGT children, these inverse relationships were magnified in IGT children.

Figure 1 compares the adjusted means and SEs of BMC in IGT versus NGT children. Although IGT children had lower adjusted BMC (1,527 ± 22.3 g) than NGT children (1,545 ± 12.8 g), these differences were not significant.

Stepwise multiple linear regression analyses were performed to examine the independent influence of age, Tanner stage, lean mass, fat mass, Vo_2max, fasting/2-h glucose, fasting/2-h insulin, insulin sensitivity, AIR, and AUC on log-transformed whole-body BMC and BMD. Multiple regression coefficients and partial/total R values are presented in Table 4. Lean mass (87%), age (1%), and AIR (1%) explained a total of 89% of the variance in BMC with no contribution by the other variables. BMD was explained by lean mass (69%), Tanner stage (3%), and AUC (2%) for a total of 74% of the variance. Fat mass, Vo_2max, fasting/2-h glucose, fasting/2-h insulin, and insulin sensitivity did not make significant contributions to the variances in BMC or BMD. When the analysis was repeated without log transformations, the results did not materially change (data not shown).

Our findings with respect to lack of significant differences in BMC and BMD in IGT versus NGT subjects, although the first in a group of pre-diabetic (type 2) children, are in agreement with those of previous researchers who have examined the same hypothesis in different populations. Wakasugi et al. (14) found no relationship between the level of HbA_1c and deficits in bone mass of type 2 diabetic adults. Pascual et al. (15) found no correlation between BMD and HbA_1c in children with type 1 diabetes but acknowledged the fact that HbA_1c may not be an ideal marker for poor glycemic control because it represents only the patients’ glycemic status of the last 3 months; it is noteworthy that changes in bone metabolism take place over a longer period. Although IGT (i.e., 2-h glucose ≥140 mg/dl) is an important precursor of diabetes (11), it is difficult to speculate the duration of this condition in our IGT children; nevertheless, we believe that inadequate duration of the pre-diabetic state may have contributed to the lack of significant differences in BMC and BMD of the two groups. Other research has shown that insulin-dependent diabetic children have a normal spine BMD during the first year of disease (16,17); a decreased BMD has been observed in those with long-standing diabetes (17). We are aware that there are also individual differences in susceptibility to disease; in other words, not all subjects with IGT will worsen to diabetes, and another possible explanation for the lack of significant differences in BMC and BMD between IGT and NGT may be because only IGT children with the potential to progress to diabetes may be at risk for osteopenia.

Although differences in BMC and BMD in IGT versus NGT children were not significant, Table 1 and Fig. 1 highlight the fact that IGT children had lower unadjusted (Table 1) and adjusted (Fig. 1) means of BMC, which in growing children is a better measure than BMD (18). Mean 2-h insulin was significantly greater in IGT than in NGT children (210.5 vs. 128.8 μU/ml). Furthermore, partial correlations (Table 2) reveal that 2-h insulin was inversely correlated with both BMC and BMD. Therefore, in light of the greater levels of 2-h insulin in IGT children and the inverse relationship between 2-h insulin and BMC and BMD, it makes sense that IGT children had lower BMC than NGT children. We therefore believe that this adjusted trend toward lower bone mass may continue if our IGT children remain hyperglycemic and hyperinsulinemic.

We (19,20) and many others (21–25) have shown that body size is the strongest single determinant of BMD, a relationship that is attributed to weight bearing and to the conversion of androstenedione to estrogen in adipose tissue (26,27). Consistent with our previous findings in Asian adolescents (19), lean body mass was the best covariate of BMC and BMD in this group of overweight Latino children. When lean mass was replaced with fat mass in the regression models, age (50%) and Tanner stage (41%) became the primary covariates of BMC and BMD, respectively; fat mass became a secondary covariate of BMC (17% of the total variance) and BMD (10% of the total variance). These findings are similar to the conclusions of others who have shown that lean mass and not fat mass independently predicts whole-body (28) and lumbar spine (29) BMD. Our results regarding the strong independent contribution of lean mass to bone mass provides additional evidence on the importance of mechanical impact and muscular forces for the growing skeleton. Further, it indirectly substantiates that skeletal muscle is also capable of aromatizing androstenedione to estrogen (30).

Because previous studies (31,32) of type 1 diabetic children with poor metabolic control have reported reductions in linear growth velocity and a diminished pubertal growth spurt, we thought it is necessary to also examine whether there are differences in height and Tanner stage in our two groups. As such, we found no significant differences in height or Tanner stage between IGT and NGT children (Table 1). Furthermore, it is noteworthy that testicular volume was highly correlated (r = 0.90, P < 0.0001) with Tanner stage (based on pubic hair) in boys. Because of this strong correlation and the lack of testicular volume data on the entire sample of boys, we chose to stage pubertal development based on pubic hair, and we believe that this was an accurate measure of puberty in boys. When we examined testicular volume values in IGT (n = 18) versus NGT (n = 66) children, we found no significant differences.

Other subtle contributors to the total variances in BMC and BMD of this population were the acute and cumulative insulin responses to oral glucose. Although some research (6,33) has suggested a link between hyperinsulinemia and better bones, reduced levels of IGF-I, a polypeptide synthesized by bone cells, reported in hyperglycemic, hyperinsulinemic, type 1 diabetic, and type 2 diabetic conditions (34–37) could alternatively cause osteopenia (34,37–39). In in vitro studies, receptors for insulin and IGF-I are present on osteoblastic cells (40,41), and both substances have a direct stimulatory effect on the recruitment of osteoblastic cells (42). We are in agreement with others (4,34,38) who have concluded that type 2 diabetes and the hyperinsulinemic state typically associated with obesity and insulin resistance in these patients could cause low bone mass (independent of body size) and that this relationship is possibly mediated by deficiency in insulin-like growth factors. Studies have found low (43), normal (44,45), and high (46) IGF-I levels in obese subjects. Whether or not overweight children with IGT have abnormal IGF-I levels should be investigated in longitudinal studies of this population.

Studies have shown that trabecular bone has high turnover and responds to factors such as endocrine changes to a greater extent than does cortical bone (24,32,47). Previous research on the association between diabetes and osteopenia has also shown site specificity. Roe et al. (48) found that in diabetic children, cortical but not trabecular BMD was lower than that in control subjects. In contrast, Lettgen et al. (49) found lower trabecular BMD but no differences in cortical or total BMD in diabetic versus nondiabetic children. Therefore, although it would seem more logical that trabecular bone would be more sensitive to metabolic changes, measurement of both cortical (femoral neck) and trabecular (lumbar spine) bone is necessary in our future work.

We conclude that IGT does not seem to predispose young pre-diabetic children to osteopenia. However, the lower unadjusted and adjusted means for BMC in children with IGT as well as the independent inverse relationship of acute and cumulative insulin responses to oral glucose on BMC and BMD suggest that, independent of body size, the skeleton of adolescents at risk for type 2 diabetes may not be protected, possibly because of the yet unproven role of IGF deficiency on bone accretion during growth. Because diabetic status may be altered in this population, examination of changes in the skeleton of these children over time may provide some answers into whether possible deficits in peak bone mass is a consequence of type 2 diabetes.

This study was supported by the National Institutes of Health (R01 DK 59211) and in part by the GCRC, National Center for Research Resources (grant M01 RR 00043).

We are grateful to the nursing and bionutrition staff at the GCRC at USC and the laboratory staff of Dr. Bergman for conducting insulin and glucose assays. We thank Dr. Marc Weigensberg for conducting the physical examinations. We are indebted to the adolescents and their families who participated in this study.

This work was presented at the International Osteoporosis Foundation World Congress on Osteoporosis held in Rio de Janeiro, Brazil, in May 2004.