|Home | About | Journals | Submit | Contact Us | Français|
Intensive diabetes therapy of type 1 diabetes (T1DM) reduces diabetes complications but can be associated with excess weight gain, central obesity, and dyslipidemia.
The purpose of this study was to determine if excessive weight gain with diabetes therapy of T1DM is prospectively associated with atherosclerotic disease.
Subjects with T1DM (97% Caucasian, 45% female, mean age 35 years) randomly assigned to intensive (INT) or conventional (CONV) diabetes treatment during the Diabetes Control and Complications Trial (DCCT) underwent intima-media thickness (IMT) (n=1015) and coronary artery calcium (CAC) score (n=925) measurements during follow-up in the Epidemiology of Diabetes Interventions and Complications (EDIC) Study. INT subjects were classified by quartile of BMI change during the DCCT. Excess gainers (4th quartile, including CONV subjects meeting this threshold) maintained greater BMI and waist circumference (WC), needed more insulin, had greater IMT (+5%, P<0.001 EDIC year 1, P=0.003 EDIC year 6), and trended towards greater CAC scores (OR 1.55, CI 0.97 – 2.49, P=0.07) than minimal gainers. DCCT subjects meeting metabolic syndrome criteria for WC and blood pressure had greater IMT in both EDIC years (P =0.02 to <0.001); those meeting HDL criteria had greater CAC scores (OR 1.6 and CI 1.1 – 2.4, P=0.01) during follow-up. Increasing frequency of a family history of diabetes, hypertension, and hyperlipidemia was associated with greater IMT thickness with INT but not CONV.
Excess weight gain in DCCT is associated with sustained increases in central obesity, insulin resistance, dyslipidemia and blood pressure, as well as more extensive atherosclerosis during EDIC.
Intensive therapy (INT) of type 1 diabetes (T1DM) reduces the incidence and progression of microvascular complications,1 risk factors for macrovascular complications,2 coronary arteries calcium (CAC),3 and intima-media wall thickness (IMT).4 After 17 years of combined follow-up in the randomized, controlled Diabetes Control and Complications Trial (DCCT) and its observational follow-up study, the Epidemiology of Diabetes Interventions and Complications (EDIC), INT was also shown to reduce the incidence of major cardiovascular disease (CVD) events by 58% compared with conventional therapy (CONV).5
Intensive diabetes therapy is, however, associated with side effects, including increased frequency of severe hypoglycemia and weight gain. We have previously shown that the 25% of DCCT intensively treated subjects with the greatest weight gain increased their BMI from a mean of 24 kg/m2 at baseline to 31 kg/m2 or greater during treatment, meeting criteria for obesity.6 Compared to INT subjects who remained weight stable during DCCT, excess gainers with INT achieved similar glycemic levels but had greater mean waist circumference; increased levels of LDL cholesterol, triglyceride, and small-dense LDL particles; and lower HDL cholesterol levels.6 The greater central distribution of body weight and dyslipidemia associated with this weight gain are consistent with the hallmark characteristics of the metabolic syndrome (MetS), typically seen in subjects with insulin resistance and type 2 diabetes7-9 but also associated with increased risk of CVD events and mortality in observational studies of T1DM.10, 11
In the present analyses we examine relationships between excess weight gain, components of MetS, and other non-traditional CVD risk factors during the DCCT with carotid IMT and CAC during follow-up in EDIC. In addition, based on our finding of increased susceptibility to weight gain and dyslipidemia in INT treated DCCT subjects with a parental history of type 2 diabetes (T2DM),12 we hypothesized that family histories of T2DM, hypertension, and hyperlipidemia would be associated with worsening markers of atherosclerosis in the INT treated subjects during follow-up in EDIC.
During 1983-89, 1441 subjects were randomly assigned to INT versus CONV diabetes therapy in the DCCT, and were treated until the final closeout visit in 1993, a mean of 6.5-years. Annual follow-up in EDIC was initiated in 1994 (EDIC year 1). Participants included in this analysis are a subset of the original 1441 DCCT subjects who were aged 18-years or older at baseline (n=1168),6 survived, and who then elected to continue in the EDIC study and underwent measurements of IMT at years 1 and 6 of EDIC (n=1015), and CAC at EDIC year 8 (n=925) (supplemental figure 1). The average ± SD age for the group at DCCT closeout visit was 35 ± 5.7 years, 45% were female, and 97% were Caucasian. Participants were categorized by quartile change in BMI during the DCCT within their respective treatment groups.6 The ranges of change in BMI for each quartile in the INT group were: from −6.27 to 0.95 kg/m2 in quartile 1; from 0.96 to 2.44 kg/m2 in quartile 2; from 2.45 to 4.37 kg/m2 in quartile 3; and from 4.39 to 17.7 kg/m2 in quartile 4. For reference, a change in 1 BMI unit for a male with a height of 1.76 m (5′ 9.5”) represents 3.1 kg and for a woman with a height of 1.62 m (5′ 4”) represents 2.6 kg. In contrast, the highest quartile of change in BMI in the CONV group was from 2.24 to 8.86 kg/m2.
Herein, excess gainers are defined as those whose BMI increased by at least 4.39 kg/m2 during the DCCT, the cutoff for the fourth quartile of BMI change in the INT group during the DCCT. For the present study, subjects in the CONV group that met this criterion were also classified as excess gainers.
In addition, because the amount of weight gain in the first through third quartiles in the INT therapy group were close in range and nearly half that of the fourth quartile (figure 1), data from the first three quartiles were combined (minimal gainers, n=394) for comparison with the excess gainers (n=122). In the CONV, 23 (4.6%) subjects met criteria for excess gainer and the remaining 476 subjects were classified as minimal gainers. Baseline ages, duration of diabetes, and BMI’s were the same in all BMI change quartiles in both the INT and CONV groups, as were hemoglobin A1c values at the final DCCT closeout visit with INT after a mean of 6.5-years of treatment.6 Family histories of T2DM as well as hypertension, and hyperlipidemia were obtained at the baseline visit in DCCT.
Measurements of lipid levels, blood pressure, and waist circumference obtained at the DCCT closeout visit were classified according to MetS criteria from the National Cholesterol Education Program.7
Carotid ultrasonography was performed approximately 1 and 6 years after the initiation of the EDIC study (approximately 8 and 13 years, respectively, after the beginning of the DCCT).4, 13 Carotid IMT (mm) was measured by a single longitudinal lateral view of the distal 10 mm of the right and left common carotid arteries and three longitudinal views in different imaging planes of each internal carotid artery.13 Studies were read in a central unit (Tufts University, Boston) by a single reader, who was masked to the subjects’ treatment assignments and the time of the studies (year 1 vs. year 6).
Computed tomography was performed once during approximately the 8th EDIC year, between 11–20 years after enrolment into the DCCT.3 At that visit, participants were scanned twice over calibration phantoms of known calcium concentration with scans read centrally at the Harbor-UCLA Research and Education Institute (Torrance, CA) by readers who were masked to subject identity and prior treatment assignment. The average coronary artery calcium (CAC) Agatston score from the two scans was used in the analysis. Presence of CAC was defined as an Agatston score > 6.25 mm3, which represents a value that is less than 1% likely to be due to interscan variability.14
Following an overnight fast of at least 8 hours, blood was drawn from the subjects. After the plasma was separated and stored briefly at −20°C (−4°F), it was placed on dry ice and sent immediately to the DCCT/EDIC Central Biochemistry Laboratory at the University of Minnesota, Minneapolis, where it was stored at −70°C. Total cholesterol, triglyceride, and HDL-C levels were determined by enzymatic methods.2 The LDL-C was calculated using the Friedewald equation.15 Hemoglobin A1c and urinary albumin excretion were also measured as previously described.16
Additional samples from the final closeout visit in the DCCT were sent on dry ice to the Northwest Lipid Research Laboratories in Seattle, Wash, where they were stored at −70°C prior to undergoing the following measurements as previously described6: Lp(a) mass, apolipoprotein B, and lipoprotein density distribution for determination of the LDL relative flotation (LDL Rf), a measure of peak LDL particle density.
Comparisons of quantitative variables between excess gainers versus minimal gainers were made using the Mann-Whitney rank sum test. Proportions of medication use and MetS criteria between groups were tested by Chi-square analysis. Prospective associations of IMT and CAC scores using lipid and clinical outcome variables determined at the closeout visit of the DCCT, including MetS criteria, were tested using multiple linear regression for IMT and multiple logistic regression for CAC scores. Co-variables included in the regression analyses represent updated, weighted (adjusted for differences in sampling schedule between DCCT and EDIC study visits) mean values at the time the imaging measure was obtained—EDIC years 1 and 6 for IMT and EDIC year 8 for the CAC. Bonferoni correction was used for analyses of repeated measures. Statistical analyses were performed using SAS version 9.1.
From the DCCT closeout visit until EDIC year 6, INT and CONV subjects originally categorized as minimal gainers at the DCCT closeout visit continued to gain weight (P<0.001 for both treatment groups) (figure 1 and supplemental figure 2). INT and CONV subjects originally categorized as excessive weight gainers experienced slight, but nonsignificant, weight gain during this same time (figure 1 and supplemental figure 2). All weight gain groups, however, experienced significant increases in waist circumferences (P<0.001) (figure 1, supplemental figure 2). In addition, the separations in BMI and waist circumference between the excess gainers and the minimal gainers were maintained during EDIC follow-up (Table 1 and supplemental Table 1).
With INT, although both the excess gainers and minimal gainers had similar glycemic control at the closeout visit in the DCCT and experienced rises in hemoglobin A1c levels during EDIC follow-up (Table 1), the rise in hemoglobin A1c was significantly greater in the excess gainers compared to the minimal gainers despite using greater insulin doses (Table 1). With CONV, insulin dose was higher and hemoglobin A1c level trended lower at the DCCT closeout visit in the excess than the minimal gainers (supplemental Table 1); and both weight gain groups experienced improvement in glycemic control during EDIC follow-up with similar A1c and insulin doses by EDIC year 6 (supplemental Table 1).
At the DCCT closeout visit, levels of total cholesterol, LDL, non-HDL cholesterol, and percent of subjects with LDL > 2.59 mmol/L (100 mg/dL) were higher in excess gainers than minimal gainers with INT, and remained higher at both EDIC follow-up visits (Table1). HDL cholesterol was lower among excess gainers than minimal gainers at the DCCT closeout visit and EDIC year 1 visits (Table 1). By EDIC year 6, however, HDL cholesterol had increased in both groups such that levels were no longer statistically different between them, which occurred despite the weight gain and significant increases in TG levels experienced by both groups (Table 1). During EDIC follow-up, use of lipid lowering medications was greater in the excess gainers than minimal gainers (7% at EDIC year 1, 26% at EDIC year 6, and 48% by EDIC year 8-9 for excess gainers vs. 2%, 11%, and 32% for corresponding time points in minimal gainers, respectively, P =0.003, P <0.001, and P =0.04, respectively). Of the lipid values measured only at the DCCT closeout visit, mean (± SD) levels of apolipoprotein B (0.89 ± 0.18 vs. 0.81 ± 0.22 g/L, P <0.001) and LDL peak particle density (Rf, 0.30 ± 0.024 vs. 0.31 ± 0.023, P =0.004) were higher among excess gainers than minimal gainers, but Lp(a) levels were not different (0.69 ± 0.70 vs. 0.66 ± 0.69 μmol/L, P =0.28). Systolic and diastolic blood pressures were higher among excess gainers than minimal gainers at each study visit (P <0.001), despite greater use of medications for hypertension in the excess gainer compared to the minimal gainer group (17% at EDIC year 1, 44% at EDIC year 6, and 88% by EDIC year 8-9 for excess gainers vs. 9%, 27%, and 48% for corresponding time points in minimal gainers, respectively. All comparisons P <0.001). For the CONV group, there were no significant differences in lipid levels or blood pressure at any time point (Supplemental Table 1).
At the DCCT closeout visit, among minimal gainers in INT, MetS criteria for HDL cholesterol and blood pressure were most commonly met whereas excess gainers most commonly met HDL, blood pressure, and waist circumference criteria (Table 1). Similarly, the most frequent MetS criteria met by minimal gainers in CONV at the DCCT closeout visit were HDL cholesterol and blood pressure. On the other hand, excess gainers in CONV had roughly equal proportions meeting waist circumference, lipids, and blood pressure criteria (Supplemental Table 1). Proportionally more excess weight gainers than minimal gainers met one or more MetS criteria at each DCCT and EDIC study visit in both INT (Table 1, P<0.001 for each time point) and CONV groups (Supplemental Table 1, P=0.05, <0.001, and < 0.05 for DCCT closeout, EDIC year 1, and EDIC year 6 respectively).
IMT at EDIC year 1 was greater among excess gainers than minimal gainers (P < 0.001), remaining significant with adjustments for treatment group (INT vs. CONV), age, sex, smoking status, type of scanning machine used, and updated, weighted mean values for urinary albumin excretion rate, hemoglobin A1c, and diastolic and systolic blood pressure levels (Table 2). Both excess and minimal gainers experienced increases in IMT thickness between years 1 and 6 of follow-up (P < 0.001) (figure 2, data shown for INT group only); however, there was an incrementally higher mean IMT among excess gainers vs. minimal gainers at year 6 (P=0.006, Table 3).
When traditional lipid risk factors (e.g. LDL-C, HDL-C, and triglyceride levels) measured at the DCCT closeout were added to the multivariable models presented in Tables 2 and and3,3, excess gainers continued to show significantly greater IMT compared to minimal-gainers at year 1 (P =0.003) and at year 6 (P =0.02). “Non-traditional” lipid risk factors measured at the DCCT closeout visit were also tested for association with IMT at EDIC years 1 and 6. These included apolipoprotein B, LDL peak particle density (relative flotation rate or Rf), Lp(a), and non-HDL cholesterol. When each was added individually to a model that included LDL, triglyceride, HDL, as well as adjustments for age, sex, and machine used, none independently reached significance with IMT at either time point or improved the model fit.
The likelihood of having an increased CAC score (> 6.25) trended higher among excess gainers compared to minimal gainers (OR 1.55 and CI: 0.97 – 2.49, P=0.07, Table 4; figure 3 showing data for INT group only) after accounting for updated, weighted mean values of covariates and lost even borderline significance with further adjustment for traditional lipid risk factors (LDL-C, HDL-C, and triglycerides) (OR 1.48, 95% CI 0.91 – 2.40, P=0.12). None of the “non-traditional” lipid variables individually added to a model that included LDL, triglyceride, and HDL levels, as well as adjustments for age, sex, and machine used, independently reached significance or improved the model fit.
The associations of meeting individual MetS criteria, irrespective of weight gain, using variables measured at the DCCT closeout visit with IMT thickness and CAC accumulation were tested (Table 5). Including covariates for age, sex, treatment group assignment, study performance site, smoking status, and updated, weighted mean values for hemoglobin A1c, albumin excretion rate, and total and LDL cholesterol, meeting MetS criteria for blood pressure and waist circumference were associated with greater IMT at EDIC years 1 (P=0.02 and <0.001, respectively) and 6 (P<0.001 for both). Meeting triglyceride criteria became significantly associated with increased IMT at EDIC year 6 (P=0.04). HDL criteria was not associated with IMT at either EDIC year but was the only MetS criteria associated with CAC accumulation (P=0.01).
Most subjects (753 out of 1015, or 74%) reported a family history of type 2 diabetes, hyperlipidemia, or hypertension. After adjusting for age, sex, machine type, smoking status, urinary albumin excretion rate, hemoglobin A1c, increasing frequency of one or more of these familial diagnoses in the INT group was associated with increasing IMT thickness at both EDIC year 1 (regression coefficient = 0.011, standard error = 0.0038, P=0.004) and EDIC year 6 (regression coefficient = 0.014, standard error = 0.0050, P=0.007). On the other hand, CAC scores were not significantly associated with these family histories (OR 1.05, 95% CI 0.81 - 1.37, P=0.70) in the INT group. Corresponding analyses for CONV were all non-significant: IMT EDIC year 1, P=0.58; IMT EDIC year 6, P=0.56; CAC, P=0.56.
The DCCT study previously demonstrated that weight gain and obesity were associated with INT of T1DM, but little is known about the effect of this weight gain on CVD risk factors and outcomes. With longer follow-up of the DCCT subjects during EDIC, we have shown that the original DCCT groups continued to gain weight accompanied by increases in waist circumference and insulin dose (units/kg), and that those in the highest quartile of weight gain (excess weight gainers) maintained their average BMI in the obese range (≥ 30 kg/m2).
Along with the increase in central obesity and presumed insulin resistance (higher insulin needs for similar or worse hemoglobin A1c levels), lipid and blood pressure levels worsened during EDIC follow-up among excess gainers versus minimal gainers with INT. Of note is the paradoxical rise in HDL cholesterol during EDIC follow-up in both excess gainers and minimal gainers with INT. This occurred despite increases in central adiposity (waist circumference) and triglyceride levels in both groups, conditions usually associated with lower, more atherogenic HDL levels in association with increased activities of the enzymes involved in depleting cholesterol from HDL, including cholesteryl ester transfer protein and hepatic lipase.17, 18 Based on our previous findings, it is unlikely that changes in phospholipid transfer protein activity mediated this HDL effect18 but other candidate mechanisms include increased lecithin:cholesterol acyltransferase enzyme activity19, changes in endothelial lipase20, or other alterations leading to increased cholesterol efflux or reduced HDL clearance. In addition, the percentage of subjects meeting one or more criteria of metabolic syndrome (not including impaired fasting glucose) was greater in excessive gainers than minimal gainers at the DCCT closeout visit and remained so during EDIC years 1 and 6.
Increased IMT is a measure of subclinical atherosclerosis and a predictor of cardiovascular events in the general population21 and has been shown to be greater in patients with T1DM compared with controls without diabetes.13, 22, 23 Excess weight gain in the DCCT was associated with increased IMT at both years 1 and 6 of follow-up in EDIC compared to the minimal gainers, differences that may have been attenuated by the increased use of blood pressure and lipid lowering medications in the excess gainers. CAC is another subclinical marker associated with increased risk for cardiovascular disease,24 and like IMT, CAC scores were higher in the excess gainer group, though this relationship only reached borderline significant after adjusting for covariates. With only one cross-sectional CAC measurement in EDIC, we were unable to test whether a higher insulin dose and greater BMI predicted greater CAC score progression, as was recently reported in an observational study of patients with T1DM.25 We also explored relationships between individual criteria for MetS, previously described as an independent CVD risk in non-diabetic populations26 and in T1DM,10, 11 with both IMT and CAC measures. We found that meeting MetS criteria for blood pressure and waist circumference was associated with increased IMT at both EDIC time points, triglyceride criteria became associated with IMT by EDIC year 6, but only HDL criteria was associated with increased CAC score. Taken together, these data demonstrate the potentially adverse effect of excess weight gain (and increasing frequency of meeting MetS criteria that accompany this weight gain) with INT on IMT and CAC and are consistent with recent long-term observational studies of patients with T1DM that have found obesity27 to be predictive of future cardiovascular events and mortality.
In an attempt to understand the risk factors accompanying weight gain with INT in the DCCT that led to the increased IMT and CAC measures, we performed adjusted analyses with “traditional” (triglyceride, LDL, and HDL) and “non-traditional” (apolipoprotein B, LDL particle density, Lp(a), and non-HDL cholesterol levels) lipid CVD risk factors. The non-traditional risk factors were not associated with increased IMT or CAC in EDIC independent of excess weight gain. Although the traditional risk factors associated with CVD were associated with CAC and IMT, excess weight gain remained significantly associated with IMT after their inclusion. However, the same was not true for the association between weight gain and CAC. These data suggest that the effect of excess weight gain on carotid artery disease is not fully explained by accompanying deterioration in lipid levels (or blood pressure levels), whereas traditional lipid risk factors are likely to be mediators of the relationship between weight gain and coronary artery disease.
Obesity, central obesity, dyslipidemia, and hypertension within middle-age populations are known to be strongly influenced by heritability.28-31 We have previously demonstrated that patients with T1DM with at least one parent with T2DM are more likely to exhibit central obesity and dyslipidemia. Specifically, we showed that subjects in the DCCT with parents with T2DM gained more weight with INT, had higher waist circumferences and insulin doses, and had lipid levels consistent with MetS.12 These findings were confirmed in an independent cohort of patients with T1DM32 and shown to be predictive of CAD in another observational cohort.33 In the present analysis, increasing prevalence of individual family histories of diabetes, hypertension, and hyperlipidemia in DCCT were associated with increased IMT with INT during EDIC. This relationship was not observed in the CONV group and supports our hypothesis that near normalization of glucose control with INT in T1DM may allow the expression of heritable traits, including excess weight gain,34 that are otherwise obviated by poor metabolic control. The expression of these traits may promote atherosclerosis and reduce the salutary effects of INT on CVD previously demonstrated.4, 5, 12
In conclusion, we have shown that excessive weight gain with INT is sustained during 6 years of EDIC follow-up and remains associated with central obesity, insulin resistance, a progressive rise in blood pressure, and dyslipidemia. Regardless of treatment group assignment, however, subjects with the greatest weight gain had the greatest IMT and highest CAC scores. Whether the atherogenic changes associated with weight gain will reduce the long-term benefit of INT on major cardiovascular events needs to be examined with future follow-up. However, based on the results of this study, efforts should be made to limit excess weight gain that accompanies intensive glucose treatment of T1DM.
We acknowledge the ongoing dedication of the DCCT/EDIC study researchers, staff, participants and their families. A complete list of participants in the DCCT/EDIC research group can be found in Archives of Ophthalmology, 2008; 126(12):1713.
Funding Sources: This study was supported by National Institutes of Health grant no. DK02456 (JDB) and UW Diabetes Endocrinology Research Center National Institutes of Health grant no. DK 17047. The DCCT/EDIC project is supported by contracts with the Division of Diabetes, Endocrinology and Metabolic Diseases of the National Institute of Diabetes and Digestive and Kidney Diseases, National Eye Institute, National Institute of Neurological Disorders and Stroke, the General Clinical Research Centers Program and the Clinical and Translational Science Awards Program, National Center for Research Resources, and by Genentech through a Cooperative Research and Development Agreement with the National Institute of Diabetes and Digestive and Kidney Diseases.
Contributors of free or discounted supplies and/or equipment: Lifescan, Roche, Aventis, Eli Lilly, OmniPod, Can-Am, B-D, Animas, Medtronic, Medtronic Minimed, Bayer (donation one time in 2008), Omron.
Disclosures: The authors have no disclosures to report.