Influence of Insulin Resistance and Body Mass Index at Age 13 on Systolic Blood Pressure, Triglycerides, and High-Density Lipoprotein Cholesterol at Age 19

Methods

This study was approved by the Human Subjects Committee of the University of Minnesota. Consent was obtained from all of the children and their parents/guardians.

The cohort for this report consists of black and non-Hispanic white children participating in a longitudinal insulin/blood pressure study.¹⁴ The children were randomly recruited after blood pressure, height, and weight screening of 12 043 grade 5 to 8 Minneapolis Public School students (3819 black, 4216 white, 4008 other; 6035 boys and 6008 girls), representing 93% of all of the students in those grades. Euglycemic insulin clamps were conducted at mean age 13 (11 to 15 years) in 357 participants, and the clamps were repeated at mean age 15 (13 to 17 years) in 309 participants and at mean age 19 (18 to 21 years) in 224 participants.

At the initial clamp study, the children underwent a complete physical examination including Tanner staging and anthropometric measurements. Height was measured by a wall-mounted stadiometer. Weight was measured by a balance scale. Blood pressure was measured twice on the right arm using a random-zero sphygmomanometer with participants seated; the averages of the 2 measurements (systolic and 5th phase Korotkoff diastolic) were used in the analyses. Data from children in the 5 Tanner stages were combined, based on previously published analyses from this cohort.¹⁵

Euglycemic insulin clamp studies were conducted in the University of Minnesota Clinical Research Center as described previously.¹⁴ Blood samples for fasting serum insulin levels, triglycerides, and HDL-C were obtained at baseline (before starting the insulin infusion). Plasma glucose was measured at baseline and every 5 minutes during the clamp. The insulin infusion was started at time 0 and continued at 1 mU/kg per minute for 3 hours. An infusion of 20% glucose was started at time 0 and adjusted, based on plasma glucose levels, to maintain plasma glucose at 5.6 mmol/L (100 mg/dL). Insulin sensitivity, “M,” was determined from the amount of glucose administered over the final 40 minutes of the euglycemic clamp and was expressed as M_lbm (ie, glucose use/kg lean body mass [LBM] per minute). LBM, or fat-free mass, was calculated by the skinfold formula method of Slaughter et al¹⁶ at the first and second clamps and by dual emission x-ray absorptiometry at the third clamp. The LBM values from the first and second clamps were adjusted to the dual emission x-ray absorptiometry values according to equations derived from studies in siblings of the present cohort, and within the same age range, as published previously.¹⁷

Blood samples were analyzed for glucose immediately at the bedside with a Beckman Glucose Analyzer II (Beckman Instruments Inc). Insulin levels (radioimmunoassay) and serum lipids were determined in the laboratory of the Fairview-University Medical Center, as reported previously.¹⁴

The extent of clustering of components of the insulin resistance syndrome ([IRS] fasting insulin, SBP, triglycerides, and HDL-C) was based on z scores for the participants, as reported previously.¹⁸ The z score for each of the 4 variables was calculated by determining the difference between each participant’s value for the respective variable and the gender-specific mean value for that variable and then dividing the result by the corresponding SD. The average of the z scores for the 4 variables was computed (with reversed sign for HDL-C) and defined as the insulin resistance (metabolic) syndrome score (IRS score). Thus, a higher IRS score indicates that the 4 variables tend to cluster in the higher distributions (ie, higher risk). Measures of body size were not included in the cluster analysis, because BMI was one of the criteria under investigation, and all of the other measures of body size are highly correlated (r>0.9) with BMI; fasting glucose was not used as a component of the IRS score because of the narrow range of values found in children in contrast to the wide range found in adults and used as a component of the metabolic syndrome. To ensure that the cluster analysis was not unduly influenced by a single abnormal component, we reviewed the individual clusters and found the expected distribution, that is, a high number of high individual component z scores in the highest clusters, a high number of low individual component z scores in the lowest clusters, and a mix of individual z scores in the middle clusters.

In an initial data analysis to understand the relative contributions of age 13 insulin resistance (M_lbm), BMI and SBP to SBP at age 19, we regressed age 19 SBP on age 13 M_lbm, BMI, and SBP, adjusting for age, race, and sex in the 208 participants who had both age 13 and age 19 measures. Next, data analysis was performed using a repeated-measures regression model to predict SBP, triglycerides, HDL-C, and the IRS score at mean age 19 from levels of M_lbm and BMI at mean age 13 and changes in M_lbm and BMI from age 13 to age 19. We assumed a compound symmetry variance structure, meaning that a constant correlation was assumed to exist between first clamp and second clamp, first clamp and third clamp, and second clamp and third clamp measures. The assumption of more complex variance structures had little effect on the estimates. The dependent variable was the current value of each of the variables in 4 separate regression analyses, 1 regression for SBP, 1 for triglycerides, and so forth. The fixed-effects variables included visit (categorical variable); current age; sex; ethnicity (black or white); M_lbm at visit 1; M_lbm with time; the change in M_lbm, namely, the value at each visit minus the value at visit 1 plus corresponding variables for BMI (visit 1, interaction with each visit, and change from visit 1); and the baseline value for the respective dependent variable. In this format, the regression coefficients of visit 1 M_lbm represent the slope of the dependent variable at visit 3 on M_lbm at visit 1, and the regression coefficient of the change in M_lbm represents the slope of the dependent variable at visit 3 on the change in M_lbm, visit 3 minus visit 1. The coefficients of BMI are similarly interpreted. Parallel analyses were then run with fasting insulin replacing the independent variable M_lbm. Goodness-of-fit of the regression was assessed by predicting the dependent variables from quartiles of each independent variable; adjusted means from these analyses are plotted on the same graph as the predicted lines in Figures 1 through 3⇓⇓.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1. Relation of M_lbm at age 13 to levels of SBP, triglycerides, and the IRS score (z score) at age 19. Goodness-of-fit of the regression was assessed by predicting the dependent variables from quartiles of each independent variable; the symbols (squares) plotted on the same graph as the predicted lines represent the adjusted means from these analyses.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2. Relation of change in M_lbm from age 13 to 19 to levels of triglycerides and the IRS score at age 19 (see legend for Figure 1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3. Relation of change in BMI from age 13 to 19 to SBP, triglycerides, HDL-C, and the IRS score at age 19 (see legend for Figure 1).

Results

The clinical and laboratory variables for the participants who had clamp studies at mean ages 13 and 19 years are listed in Table 1. There were no significant differences for any of the variables between these individuals and the individuals with <3 clamps. The mean values for all of the variables were not significantly different from the mean values of the entire cohort. There were more boys (58%) and whites (82%). Table 1 shows the developmental changes in the variables between 13 and 19 years. By mean age 13, boys were taller and slightly heavier with greater LBM and lower percentage of body fat than girls, and these differences were similar at mean age 19. BMI was not significantly different between the genders at either age. Percentage of body fat decreased in the boys and increased in the girls from age 13 to 19. SBP was greater in boys, and diastolic blood pressure was greater in girls at mean age 13, but there was no significant difference at mean age 19. Fasting insulin was not significantly different between genders at either age. M_lbm was significantly greater in boys at age 13 but was significantly greater in girls at age 19. Triglycerides became significantly higher and HDL-C significantly lower in boys from age 13 to 19.

The tracking correlations (r) for the anthropometric measures and risk factors between mean ages 13 and 19 were all statistically significant. The highest correlations were generally associated with the anthropometric measures (weight: 0.82; BMI: 0.85; waist: 0.74; triceps and subscapular skinfolds: 0.60 and 0.69, respectively; and percentage of body fat: 0.71, with the LBM lower at 0.51; all P<0.0001). Other correlations were SBP (0.48), diastolic blood pressure (0.37), triglycerides (0.41), and HDL-C (0.67; all P<0.0001). The correlation for M_lbm was 0.42 (P<0.0001), and fasting insulin had the lowest correlation (0.18; P=0.01). Adjustment for baseline BMI (mean age 13) had virtually no effect on the correlations for blood pressure, lipids, or M_lbm, but the correlation for fasting insulin fell to 0.11 (P>0.05).

Correlations among BMI, M_lbm, and fasting insulin at baseline (mean age 13) and BMI, M_lbm, fasting insulin, triglycerides, HDL-C, and SBP at mean age 19 were virtually all significant, as shown in Table 2. Baseline BMI was significantly related to all of the variables except M_lbm; M_lbm was significantly related to all of the variables, except BMI and triglycerides; and fasting insulin was significantly related to all of the variables, except triglycerides and SBP.

The results of the multiple regression analyses with SBP, triglycerides, HDL-C, and the IRS score as dependent variables and baseline BMI, baseline M_lbm, change in BMI (age 13 to 19), and change in M_lbm (age 13 to 19) as independent variables are shown in Table 3 and in Figures 1 to 3⇑⇑. There was no significant correlation between either M_lbm at age 13 or change in Mlbm from age 13 to 19 and BMI at age 19 (P=0.37 and 0.65, respectively), but both were significantly correlated with change in BMI from age 13 to 19 (r=0.16, P=0.02 and r=0.24, P=0.0006, respectively). However, in the multiple regression analysis, both baseline M_lbm and change in M_lbm significantly predicted most of the variables at age 19. There was a decrease of 0.41 mm Hg SBP, 0.02 mmol/L triglycerides, and 0.03 z score units per unit increase in baseline M_lbm (Figure 1) and a decrease of 0.02 mmol/L triglycerides and 0.02 z score units per unit increase in M_lbm from age 13 to 19 (Figure 2). HDL-C at 19 was not predicted by baseline M_lbm, and HDL-C and SBP at 19 were not predicted by change in M_lbm. Baseline BMI did not predict any of the variables in the multiple regression analysis, whereas increase in BMI from 13 to 19 significantly predicted all 4 of the variables (0.80 mm Hg increase in SBP, 0.03 mmol/L increase in triglycerides, 0.06 z score units increase in cluster score, and 0.007 mmol/L decrease in HDL-C per kg/m² increase in BMI; Figure 3). When fasting insulin was substituted for M_lbm in the multiple regression analyses (Table 3), fasting insulin and change in fasting insulin significantly predicted triglycerides and the cluster score, but neither predicted HDL-C nor SBP, whereas the estimated relation of SBP, triglycerides, HDL-C, and cluster score to baseline BMI and change in BMI from 13 to 19 was similar to the analyses in which the M_lbm and change in M_lbm from 13 to 19 were included in the model. When these relations were considered separately for boys and girls, the results were similar, except for baseline BMI, which had a significant relation with age 19 HDL-C, triglycerides, and IRS score in boys and IRS score in girls.

The relative contributions of age 13 SBP, M_lbm, and BMI on age 19 SBP were considered by putting the above coefficients on an SD scale. Age 19 SBP was estimated to increase by 4.3 mm Hg (P<0.0001) per SD of age 13 SBP and to decrease by a lower degree (1.8 mm Hg; P=0.03) per SD unit of age 13 M_lbm. Although age 19 SBP was also positively related to age 13 BMI (P=0.01), in analyses adjusted for age, race, and sex, this relation was attenuated almost completely by adjustment for age 13 SBP.

Discussion

A previous cross-sectional study in this cohort at mean age 15 showed that insulin resistance is related independent of BMI to triglycerides, HDL-C, and to clustering of the risk factors into the insulin resistance (metabolic) syndrome.¹⁸ The present study extends those findings by showing that the degree of insulin resistance at mean age 13 and the degree of change in insulin resistance from age 13 to 19 predict levels of the risk factors and the clustering of these factors with fasting insulin into the IRS in early young adulthood (mean age 19). Although it has been known that insulin resistance is significantly related to hypertension^12,18–20 and dyslipidemia^3,6,13,21 in adults and children, this study demonstrates for the first time not only that childhood insulin resistance significantly predicts future levels of the risk factors but also that it is independent of the effects of obesity. Moreover, the results suggest that higher degrees of insulin resistance predict a clustering of the risk factors (ie, the IRS score) into the insulin resistance (metabolic) syndrome.

The mechanisms linking insulin resistance to hypertension and dyslipidemia have not been defined. It has been suggested that hypertension may be related to the increase in circulating insulin associated with insulin resistance. Insulin has been shown to induce renal sodium absorption,²² to increase sympathetic nervous system activity,²³ and to stimulate vascular smooth muscle growth.²⁴ The dyslipidemic changes may also be related to hyperinsulinemia because of an adverse effect on hepatic lipid metabolism³ or to increased fatty acid secretion resulting from insulin resistance of adipose tissue.²¹ A role for hyperinsulinemia is further supported by the present study in which both fasting insulin at mean age 13 and increase in insulin from 13 to 19 years predicted triglyceride levels and the clustering of IRS factors. Studies in Pima Indians have shown that the level of fasting insulin has an effect, independent from insulin resistance, on the onset of diabetes over a mean period of observation of 6.4 years.²⁵

Previous studies have shown a relation of fasting insulin to future risk, and evidence linking insulin resistance to hypertension has been summarized recently.¹² In adults, an association between fasting insulin and future development of hypertension was identified over 10 years in a Swedish study,²⁶ over 7 years in the San Antonio Heart Study,²⁷ and over 6 years in the Atherosclerosis Risk In Communities (ARIC) study.²⁸ A significant relation between fasting insulin and future blood pressure level was also shown in Finnish children over 6 years of observation¹⁰ and over an 8-year period in white and black children in the Bogalusa Heart Study.⁹ Similarly, baseline fasting insulin levels have been related to future dyslipidemia in children.^9,29

The relation of insulin resistance to hypertension and dyslipidemia is confounded by the strong independent relations of these 2 factors to obesity. The adverse metabolic effect of obesity on blood pressure begins in early childhood³⁰ and tracks through adolescence into adulthood.⁷ Degree of fatness correlates with levels of triglycerides (positive) and HDL-C (negative) from childhood through adolescence,^13,31,32 and BMI predicts lipid levels in young adulthood.⁷ The significant correlation between obesity and insulin resistance also begins in childhood,^13,33,34 and increasing obesity is associated with higher levels of blood pressure and triglycerides and lower levels of HDL-C and insulin resistance, as measured by the homeostatic model assessment method.⁶ The present study uses the euglycemic insulin clamp to better define these interrelations by assessing the independent influence of insulin resistance and BMI in predicting future levels of blood pressure, triglycerides, and HDL-C. The results show slightly different effects. Both baseline insulin resistance and degree of increase in insulin resistance from age 13 to 19 predicted adverse levels of the risk factors. In contrast, although baseline BMI was correlated significantly with blood pressure and the lipid levels at age 19, when considered in a multiple regression analysis together with SBP at age 13 and degree of increase in BMI from age 13 to 19, the change in BMI, but not the baseline BMI, significantly predicted the risk factor levels. Thus, the effect of baseline SBP and change in BMI over time had a greater influence than the baseline BMI. Results from our earlier studies in another cohort⁷ were similar to the present study, showing that change in BMI during either childhood or adolescence had a more predominant effect on levels of SBP and lipids in young adulthood than the baseline BMI at age 7 years.

It is not clear why insulin resistance at age 13, in contrast to BMI, remains an independent predictor of SBP at age 19. Although the strong relation of obesity to insulin resistance is widely recognized and believed by many to be integral to the development of insulin resistance, the independent effects of BMI and insulin resistance have been shown previously. Studies in adults have reported a significant independent relation between insulin resistance and cardiovascular risk factors and an accentuation of the effect of obesity on cardiovascular risk in the presence of insulin resistance.³⁵ Previous studies in children have also shown the independent effects of insulin resistance and BMI and an interaction between the 2.¹⁸ Recent reviews have discussed the independent relations of insulin resistance and obesity to the metabolic syndrome.^36,37 Thus, the results from this study suggest that: (1) the effect of BMI on cardiovascular risk begins early in life, but its impact is less significant than changes in BMI that occur during the transition from childhood to young adulthood; and (2) insulin resistance has an independent effect on cardiovascular risk that also begins early in life and seems to continue during adolescent development.