Carotid Artery Intima-Media Thickness in Children With Type 1 Diabetes

We studied 50 children with type 1 diabetes (aged 11 ± 2 years) and 35 healthy control subjects. The groups were matched for age, sex, and body size. None of the children had hypercholesterolemia, as judged by the reference values for Finnish children, i.e., all children had total cholesterol and LDL cholesterol values less than the age- and sex-specific 90th percentile (26). The clinical characteristics of the study groups are shown in Table 1.

The patients with diabetes were recruited from the outpatient clinic of the Department of Pediatrics, Turku University Central Hospital. The inclusion criteria were age 7 to 14 years, diabetes duration >6 months, normotensive, nonsmoker, and no chronic diseases other than type 1 diabetes. We studied 50 (41%) consecutively seen consenting outpatient children of the total 123 eligible children with diabetes (meeting the inclusion criteria) who were treated at our clinic. The study subjects were representative of the total eligible children with diabetes, as age (11 ± 2 vs. 11 ± 2 years; P = 0.27), duration of diabetes (4.4 ± 3.0 vs. 5.2 ± 2.3 years), HbA_1c (8.9 ± 1.4 vs. 9.1 ± 1.5%), LDL cholesterol (2.3 ± 0.7 vs. 2.4 ± 0.5 mmol/l), and triglycerides (0.70 ± 0.50 vs. 0.76 ± 0.33 mmol/l) were similar in those who did not participate (all comparisons P > 0.10). The mean duration of diabetes was 4.4 ± 3.0 years (range, 6 months to 12 years). None of the children were taking regular medications other than daily insulin. The daily insulin dose was 0.95 ± 0.24 IU/kg (range, 0.62–1.53 IU/kg). None of the patients with diabetes had evidence of microvascular complications, such as diabetic retinopathy, neuropathy, or microalbuminuria. In the diabetic group, the mean HbA_1c level was 8.9 ± 1.4% (range, 6.2–12.7%; reference range, 4.2–6.0%). The healthy control children included in the study were friends and relatives of the children with diabetes studied or children of the staff members of the Turku University or Turku University Central Hospital. None of the control children had chronic diseases or were taking regular medications, and all were nonsmokers. Tanner staging was not systematically performed in all children. In unclear cases (not clearly prepubertal), puberty stage was evaluated by a pediatrician. Two girls and two boys, both in the diabetic group and in the control group, turned out to be pubertal (defined as menarche in girls and/or advanced pubarche and breaking of voice and/or advanced pubarche in boys). Written informed consent was acquired from the legal guardians of the children, and they were also encouraged to get approval from the child. The study was conducted according to the declaration of Helsinki, and the study protocol had been approved by the Joint Commission on Ethics of the Turku University and the Turku University Central Hospital.

All studies were performed using an Acuson Sequoia 512 mainframe (Acuson, Mountain View, CA) and a 13.0-MHz linear array transducer. All ultrasound scans were performed by an experienced vascular operator who was unaware of children’s clinical details. The studies were performed in the morning between 7:30 and 9:30 a.m. after the children had fasted overnight. Blood pressure was measured from the brachial artery three times during the ultrasound study using a standard mercury sphygmomanometer.

All studies were done following a predetermined, standardized scanning protocol for the right and left carotid arteries, using images of the far wall of the distal common carotid arteries, as described previously (27). The place of measurement was anatomically standardized in every study by identifying the proximal part of the carotid bulb and then scanning the common carotid artery (28). The bulb region was first scanned carefully in many interrogation angles to identify the beginning of the bulb. The scan was focused on the posterior (far) wall, and resolution box function was used to magnify the arterial far wall. Several images of common carotid far wall segment from 10 to 20 mm proximal to the carotid bulb (a far wall segment of 10 mm in width) were acquired. Images of the common carotid segment were acquired by using two interrogation angles in each case; anterior oblique (30° from midline) and lateral (100° from midline). To justify the use of these two interrogation angles, we performed IMT measurements in 10 children (5 control subjects and 5 diabetic children) using 15 different interrogation angles covering ∼120° of the carotid far wall circumference. The measurement of common carotid IMT, using either a mean of the two interrogation angles (anterior oblique and lateral) or a mean of 15 interrogation angles, yielded similar results (intraclass correlation coefficient, r = 0.98, coefficient of variation = 1.1%; mean difference, 0.006 ± 0.004 mm).

All scans were digitally stored on the ultrasound system internal hard disk for subsequent off-line analysis. Two end-diastolic frames were selected and analyzed for mean IMT and maximum IMT, and the average reading from these two frames was calculated for both right and left carotid arteries. Several measurements of IMT were taken covering the entire width of the common carotid segment in every subject by two independent and experienced readers who were blinded to the children’s clinical details. Average values of the two readers were used in the analyses.

The far wall of the common carotid artery and the carotid bulb region were also scanned for the presence of atherosclerotic plaques, defined as a distinct area of the vessel protruding >50% of the adjacent parts of the intima-media layer (10). For all measurements, the between-observer intraclass correlation coefficient for mean IMT was r = 0.86, with a mean between observer error of 0.018 ± 0.017 mm (range, 0–0.09 mm) and a coefficient of variation of 3%. The within-subject repeatability of mean carotid IMT measurements was studied in 22 children who were studied twice 5–8 days apart. The within-subject intraclass correlation coefficient was r = 0.94, with a mean absolute within-subject error of 0.041 ± 0.025 mm (range, 0.0–0.09 mm) and a coefficient of variation of 4%.

Venous blood samples were taken in the morning, after an overnight fast (10–12 h). Serum total cholesterol, HDL cholesterol, and triglyceride concentrations were measured using standard enzymatic methods with the use of Boehringer Mannheim reagents, with a fully automated analyzer (Hitachi 917; Hitachi, Tokyo, Japan). LDL cholesterol concentration was calculated using Friedewald’s equation (29). HbA_1c was measured with high-performance liquid chromatography (Variant Analyzer, Bio-Rad, CA).

The susceptibility of LDL to oxidation was determined in a subgroup of 42 children (21 children with diabetes and 21 matched control subjects) by monitoring the formation of conjugated dienes induced by copper. This subgroup was representative of all study children, as age, sex distribution, body size, LDL cholesterol (2.4 ± 0.8 vs. 2.2 ± 0.6 mol/l; P = 0.42), and the duration of diabetes (4.6 ± 2.9 vs. 4.2 ± 3.2 years; P = 0.60) and glycemic control (8.8 ± 1.5 vs. 8.9 ± 1.4%; P = 0.90) in children with diabetes were comparable to the entire study group.

The copper-induced LDL oxidation method used in this study has been described previously (30). In brief, LDL was isolated by single-step ultracentrifugation for 30 min at 338,000g (Beckman TL-100) with a TLV-100 rotor. A 1-ml sample of desalted and in-gel-filtered LDL solution was standardized to 0.05 mg protein/ml with PBS. The sample was oxidized at 37°C as described previously (31). The final concentration of CuSO₄ in the mixture was 1.67 μmol/l. Ultraviolet absorbance at 234 nm was monitored every minute for 300 min with the use of a Perkin Elmer Lambda Bio 10 spectrometer. Lag time to the start of the propagation phase of diene formation was defined as the intersection of the tangents of the initial phase (first 5 min) and maximal propagation.

Results are expressed as means ± SD. Data on serum triglycerides were skewed toward high values and were included as their logarithms in the analyses. Comparisons between the groups were conducted by Student’s t test. Univariate associations between the study variables were analyzed by calculating the Pearson’s correlation coefficients. Multivariate analyses were done using linear regression technique. All statistical analyses were performed using SAS (SAS Institute, Cary, NC).

The characteristics of the children are shown in Table 1. The groups were matched for age, sex, and BMI. The diabetic group had lower serum triglycerides and higher HDL cholesterol concentration compared with control subjects. There were no differences in serum total cholesterol, LDL cholesterol, blood pressure, or carotid artery baseline diameter between the study groups (Table 1).

The rate of diene formation was greater in the children with diabetes (0.49 ± 0.06 vs. 0.45 ± 0.07 μmol/min; P = 0.05). Also, the total amount of conjugated dienes produced tended to be greater in the diabetic group (521 ± 50 vs. 499 ± 45 μmol; P = 0.14). The lag time did not differ between the groups (74.1 ± 6.6 vs. 73.1 ± 5.5 min; P = 0.58).

The mean carotid IMT was significantly increased in children with diabetes compared with healthy control subjects (0.47 ± 0.04 vs. 0.42 ± 0.04 mm; P < 0.001) (Fig. 1). The maximal IMT was also significantly higher in children with diabetes (0.53 ± 0.05 vs. 0.48 ± 0.04 mm; P < 0.001), but no plaque formations (10) were observed in any of the children studied. Mean IMT was 0.47 ± 0.05 vs. 0.42 ± 0.04 mm (P = 0.02) in girls with diabetes and control girls, respectively, and 0.47 ± 0.04 vs. 0.42 ± 0.03 mm (P < 0.001) in boys with diabetes and control boys, respectively. Thus, the difference in IMT between children with diabetes and control subjects was similarly seen in boys and in girls. The difference in IMT remained highly significant when children who were considered to be pubertal (4 + 4) were excluded from the analysis (0.47 ± 0.04 vs. 0.42 ± 0.04 mm; P < 0.0001).

The correlations between risk factors and carotid IMT are shown in Table 2, separately for children with diabetes and control subjects. In children with diabetes, carotid IMT correlated significantly with BMI, diabetes duration, total cholesterol, LDL cholesterol, and blood pressure. In control subjects, mean carotid IMT correlated with total cholesterol and LDL cholesterol, and maximum IMT also correlated with systolic and diastolic blood pressure and HDL cholesterol/total cholesterol ratio (Table 2). The scatter plots between mean carotid IMT and LDL cholesterol are shown in Fig. 2, separately for children with diabetes and control subjects. Susceptibility of LDL to oxidation, measured as the rate of diene formation, was significantly correlated with mean carotid IMT in children with diabetes (r = 0.47, P = 0.04) (Fig. 3) but not in control children (r = −0.29, NS).

In a multivariate regression model for all study subjects, including age, sex, BMI, study group, HbA_1c, LDL cholesterol, log-transformed triglycerides, and systolic blood pressure as independent variables, the significant correlates for mean carotid IMT were the study group (β = 0.054, P < 0.001), LDL cholesterol (β = 0.018, P < 0.001), and systolic blood pressure (β = 0.0010, P < 0.01), the total variance explained being 47%.

In a multivariate regression model restricted to those children with data on LDL oxidation (n = 42), including study group, LDL cholesterol, the rate of diene formation, and systolic blood pressure as independent variables, the significant correlates for mean carotid IMT were the study group (β = 0.060, P < 0.001) and LDL cholesterol (β = 0.024, P < 0.001), the total variance explained being 56%. To study whether other risk factors could account for the effect of diabetic status, the regression analysis was performed without the group variable in the model. In this model, the correlates of carotid IMT were LDL cholesterol (P < 0.05) and the rate of diene formation (P = 0.25), with the total variance explained being only 17%, indicating that the effect of diabetic status was not fully explained by other risk factors.

To examine whether the correlation between diene formation rate and carotid IMT in children with diabetes remains significant after controlling for the effect of LDL cholesterol, we included diene formation rate and LDL cholesterol as independent variables in the model for carotid IMT. Both of these variables emerged as independent correlates for mean IMT in children with diabetes: LDL cholesterol (β = 0.033, P < 0.01) and the rate of diene formation (β = 0.28, P < 0.05), the total variance explained being 48%.

The present study shows that young children with type 1 diabetes have significantly increased carotid artery IMT compared with healthy control children. These findings extend to observations of postmortem studies that have indicated a relation between early atherosclerotic vascular lesions and diabetic state (32), by demonstrating in vivo that diabetes predisposes to increased subclinical atherosclerosis at a very early age. Several previous studies demonstrated that carotid IMT is increased in adults with type 1 diabetes (13–17). The results of the recent Epidemiology of Diabetes Interventions and Complications study (18), however, were somewhat contradictory, showing increased IMT only in male subjects in the internal carotid arteries but not in the common carotid. Differences in methodology and study population may offer an explanation for the discrepancy. In our experience, the image quality of carotid scans in children is superior to the scans of adult subjects. Furthermore, in the present study, we used the latest digital ultrasound technology and 13-MHz scanning frequency, which yielded very high-resolution images.

The earliest histologic atherosclerotic vascular changes, i.e., fatty streaks, are commonly found in the arteries of adolescents, whereas the development of raised lesions mainly occurs after the age of 20 years (33). According to these postmortem findings and consistent correlations between lipid risk factors and IMT seen in the present study, it may be suggested that the diffusely increased carotid artery wall thickness in the children with diabetes reflects intimal changes related to early atherogenesis. Consistent with previous studies, blood pressure was an independent predictor of IMT in the present study. The relationship between increased IMT and blood pressure suggests that smooth muscle proliferation also plays a role in the early diffuse thickening of the arterial wall. It is not clear, however, whether increased carotid IMT without focal plaques reflects subclinical atherosclerosis or merely represents a preatherosclerotic change that predisposes to atherosclerosis. To address this issue, the relationship between the histologic prevalence of carotid atherosclerosis and ultrasonographically measured wall thickness would need to be studied in pediatric subjects.

Increased carotid IMT has previously been demonstrated in children with familial hypercholesterolemia (10–12). These studies, however, have not been able to show a significant relationship between carotid IMT and serum LDL cholesterol concentration within the normocholesterolemic range. In the present study, all children had normal LDL cholesterol levels (26). Despite this, LDL cholesterol concentration was significantly related to increased IMT, both in children with diabetes and in control subjects. Moreover, in multivariate analysis including all children, LDL cholesterol emerged as an independent correlate for IMT, together with the diabetic state and systolic blood pressure. Our observations thus suggest that serum LDL cholesterol concentration, even within a normal range, is an important determinant of structural arterial changes already in childhood. In line with this, autopsy studies in children have also shown a significant relationship between serum cholesterol concentration and early atherosclerotic lesions (1,2).

Increased IMT in children with diabetes was not explained by the levels of conventional risk factors, as these were similar between the study groups. Instead, the children with diabetes showed an increased rate of LDL diene formation in response to copper-induced oxidation, indicating increased in vitro susceptibility of LDL to oxidation. Postsecretory modifications of LDL, such as oxidation, have been suggested to increase its atherogenicity. LDL particles with increased susceptibility to oxidation may become more easily oxidatively modified within the arterial wall and then recognized by the scavenger receptor, leading to increased foam cell formation and accelerated atherogenesis (34). Studies in adults have shown that LDL oxidation is increased in diabetes (35–38) and may explain some of the enhanced cardiovascular risk in type 1 diabetes (39). In the present study, differences in LDL oxidizability between the study groups did not fully explain the effect of diabetic state on IMT, as the regression model without the group variable as an independent variable resulted in a low explanatory R² value for carotid IMT. This suggests that other unmeasured factors may also account for the higher IMT in individuals with diabetes. In the multivariate analysis restricted to children with diabetes, however, LDL oxidizability remained significantly associated with IMT, independent of LDL cholesterol level. Therefore, these data support the idea that oxidative modification of LDL may have a role in the development of early structural atherosclerotic vascular changes in children with diabetes.

Epidemiologic and clinical evidence has emphasized the role of hyperglycemia in explaining the increased cardiovascular morbidity and mortality in diabetes (40,41). Chronic state of hyperglycemia may induce atherogenesis by increasing oxidative stress (42), leading to increased LDL oxidation (43,44) and decreased nitric oxide bioavailability, including endothelial dysfunction (45,46). In the present study, the HbA_1c levels in the children with diabetes were comparable to those reported previously in a population-based sample of children and adolescents (47). We were unable to show a relationship between HbA_1c and carotid IMT in children with diabetes. These data suggest that LDL cholesterol concentration and/or postsecretory LDL modification may be more important correlates of carotid IMT in children with diabetes than measures of chronic hyperglycemia. Alternatively, hyperglycemia may exert its deleterious effects by leading to glycosylation of LDL, which may increase its atherogenicity (48,49).

The present study examined the relationships between type 1 diabetes and arterial IMT using a cross-sectional setting. An ideal approach would be a longitudinal study of diabetic subjects to investigate the progression/regression of atherosclerotic vascular changes in response to therapy. The study included a relatively small number of participants, especially regarding subjects with data on LDL oxidizability. This may have increased the risk of selection bias. However, children with diabetes who participated in the present study were representative of the total number of eligible children with diabetes treated in our clinic with regard to age, disease duration, glycemic control, and lipid levels. Moreover, the subgroup of children with LDL oxidation measures were representative of all children studied. LDL oxidizability was assessed by using an assay that is based on measuring the formation of conjugated dienes in LDL lipids in response to copper-induced oxidation in vitro (31). Increased susceptibility of LDL to in vitro oxidation, however, reflects only one aspect of LDL oxidizability and has still-uncertain physiologic significance in vivo.

These data may have implications in the treatment of pediatric patients with diabetes. Our results emphasize the importance of early detection and control of vascular risk factors in young children with diabetes. As diabetes is a chronic disease and cardiovascular morbidity is very high among individuals with diabetes, noninvasive methods for monitoring vascular changes might be useful in clinical practice. Prospective studies are needed to evaluate the usefulness of IMT measurement as an index of atherosclerosis in the treatment of children with diabetes.

This study was financially supported by the Finnish Medical Foundation, the Medical Society Duodecim in Turku, the Research Foundation of Orion Corporation, the Turku University Foundation, the Lydia Maria Julin Foundation, the Wäinö Edward Miettinen Foundation, the Academy of Finland, the Juho Vainio Foundation, the Finnish Foundation of Cardiovascular Research, and the Medical Research Fund of the Tampere University Hospital.

We thank Mia Laakkonen, Tuula Laukkanen, Nina Peltonen, and Marita Koli for skillful technical assistance.