Main

Diabetes mellitus is an important risk factor for atherosclerosis and both the incidence and mortality of cardiovascular disease are increased in diabetic patients (1). Among the various pathophysiological mechanisms mediating the atherosclerotic process, both oxidative stress (OS) and endothelial dysfunction occur at an early stage in animal models of diabetes (2).

Oxidative stress is defined as the change in the pro-oxidant/antioxidant balance in favor of the former, potentially leading to biologic damage to macromolecules and cell dysfunction (3). As a result of hyperglycemia, excessive pro-oxidants (free radicals and reactive oxygen species) are formed via auto-oxidation of glucose, nonenzymatic glycation and formation of advanced glycation end products, increased flux through the polyol and hexosamine pathways, and activation of protein kinase C. These processes also lead to decreased antioxidant defenses. Brownlee (4) has linked all these abnormalities to the excessive production of superoxide by the mitochondria.

In children with type 1 diabetes mellitus (T1DM), increased OS has been reported to be present even shortly after diagnosis (5). Other reports showed the parallelism between OS and abnormal markers of endothelial cell function (such as E-Selectin and ICAM-1) in young T1DM patients, suggesting a link between these two abnormalities (6). Ultrasound testing of skin microcirculation and of brachial artery flow-mediated dilatation (FMD) have demonstrated early endothelial dysfunction in diabetic children and adolescents (7).

Since it has been shown that foam cell accumulation in the vascular wall is already present in 69% of adolescents in the general population (8), it can be postulated that the diabetes-induced endothelial abnormalities might be directly related to an increase in intima media thickness and thus be involved in the early pathogenesis of vascular dysfunction that underlies the increased atherosclerotic risk in diabetes. The mechanisms mediating or modulating these possible relationships have not been fully identified yet. In the present study, we investigated the relationship between endothelial function, carotid intima media thickness, and the oxidant-antioxidant balance. The ultimate aim is to investigate whether markers of OS can help to identify the diabetic children who are more susceptible to develop diabetic vascular disease.

MATERIALS AND METHODS

Study subjects and design.

Diabetic children and adolescents regularly attending the Diabetes Outpatient Clinic of the Antwerp University Hospital were recruited consecutively. All patients were treated with a basal-bolus insulin regimen with ≥4 subcutaneous injections daily. Exclusion criteria were presence of other diseases, regular medications other than insulin, urinary albumin excretion exceeding 15 μg/min in an overnight timed urine collection, neuropathy or proliferative retinopathy. A nondiabetic control group was recruited among the children or friends of the hospital staff members or of the diabetic children. In view of the influence of age, gender, height, and weight on the cardiovascular function parameters, great care was taken to carefully match all diabetic subjects and controls for these parameters. Pubertal stage classification was based on the determination of circulating levels of hormones, clinical examination, and age. All study subjects were on an average standard Flemish school-child normocaloric diet, which did not differ in the diabetic group. There were no vegetarians and they did not receive any supplements of vitamins or antioxidants. Similarly, physical activity was that of an average school-going population (ranging from 2 to 12 h of sport per week).

Out of 81 diabetic and nondiabetic children initially recruited for the study, 7 diabetic children were not included in the final statistical analysis because of celiac disease, obesity, hypercholesterolemia, maturity-onset diabetes of the young (MODY), and lack of a matched control. Two controls were not included because of bronchitis and use of steroid medication at the time of the study.

The study was approved by the Hospital Ethics Committee (Comité voor Medische Ethiek, Universitair Ziekenhuis Antwerpen, Wilrijkstraat 10, approval number 3/29/101) and all participants or their parents signed an informed consent form. On the day of the study, a fasting blood sample was obtained and FMD and carotid intima media thickness (IMT) were measured.

Routine blood tests including glucose, glycated Hb (HbA1c), lipid profile and routine blood count, and biochemistry were analyzed in the laboratory of the Antwerp University Hospital. Total plasma homocysteine was assayed by HPLC (Bio Rad kit 195-4075, Hercules, Ca) (coefficient of variation CV 3.4%).

OS status was evaluated by measuring blood concentrations of individual antioxidants, global plasma antioxidant capacity, and products of lipid peroxidation as recently described in detail (9). In short, alpha-tocopherol, retinol, and ascorbate were measured by reversed phase HPLC. Glutathione (GSH) in whole blood and protein thiols in plasma were measured by a colorimetric method using Ellman's reagent. Total plasma antioxidant capacity was evaluated by inhibition of peroxyl-induced chemiluminescence (TAC-PI) and by radical scavenging capacity, expressed as Trolox equivalents (TAC-TE). Plasma d-ROM (determinable reactive oxygen metabolites) was measured using a commercial kit (Pharmalab d-ROMs, Parma, Italy) and expressed as tert-butyl hydroperoxide equivalents. Plasma malondialdehyde (MDA) was analyzed by reverse phase HPLC of the adduct formed by reaction with thiobarbituric acid. Copper/zinc superoxide dismutase (Cu/Zn SOD) and glutathione peroxidase (GPX) in erythrocyte hemolysate were assayed using commercial kits (RANSOD and RANSEL respectively, Randox, Crumlin, UK, CV 14.5% and 5.9%).

FMD was evaluated by ultrasonography of the right brachial artery at rest (baseline brachialis diameter, BBD) and during reactive hyperemia after inflating and deflating a forearm blood pressure cuff (200 mm Hg or at least 50 mm Hg above peak systolic blood pressure for 4 min). Continuous ECG registration was used to measure the correct end-diastolic diameter, coincident with the R-wave. Postocclusion measurements were taken every 30 s over the following 240 s. To calculate the maximal dilatation compared with baseline (peak FMD %), the mean of three measurements at baseline, and the maximal postocclusion value were used.

Carotid IMT was evaluated on patients in the supine position, with the head turned 45° away from the side being scanned. The reference point for measurement was the beginning of the dilatation of the carotid bulb. The two-dimensional B-mode image of the posterior wall of the right common carotid artery was gained 1 to 2 cm proximal to the carotid bifurcation. The radiofrequency signals originating from an M-line perpendicular to the longitudinal and transversal axes of the artery, R-wave triggered, were analyzed three times using three different interrogation angles: 0° from midline, 30–60° from midline (anterior oblique), 90–100° from midline (lateral). Mean IMT of the three measurements was calculated.

Ultrasound studies were performed using the AU5 Ultrasound system (Esaote, Biomedica, Genova, Italy), equipped with a 10 Mhz linear-array transducer. Data analysis was performed using the Wall-Tracker System (WTS, P-Medical, Maastricht, The Netherlands).

Statistical analysis.

Results were expressed as mean ± SD or as geometric mean (95% confidence intervals) for data not compatible with a Gaussian distribution. Data were analyzed using the statistical package SPSS Version 11.0 (Chicago, IL). Differences between groups were calculated using two-factor ANOVA to test for the independent effect of diabetes and of gender as well as the interaction between both factors. Non-Gaussian data were log-transformed before application of the parametric tests. Multiple linear regression (stepwise) and analysis of covariance were applied to identify the relationship between the various parameters. Two-tailed p values ≤0.05 (or adjusted for multiple comparisons according to Bonferroni) were considered as statistically significant. The authors had full access to the data and take responsibility for its integrity. All authors have read and agree to the manuscript as written.

RESULTS

With regard to clinical characteristics (Table 1), the healthy control (CO) and diabetic groups (DM) did not differ in age, gender distribution, pubertal stage, weight, and height. Serum cholesterol was higher in DM (4.71 ± 0.89 versus 4.07 ± 0.61 mM, p = 0.001 for the independent effect of diabetes). Likewise, serum triacylglycerols were higher in DM (1.09 ± 0.47 versus 0.79 ± 0.41 mM, p = 0.006).

Table 1 General clinical and metabolic characteristics of the study subjects
Full size table

OS status was assessed by measuring antioxidant concentrations, antioxidant enzyme activities, body iron stores, and products of peroxidation (Table 2). Independently of gender, diabetic children had lower levels of plasma antioxidant capacity (TAC-TE), which was accompanied by lower levels of hydrophilic antioxidants (uric acid and ascorbate) and lower serum proteins and albumin in particular. Likewise, the thiol content in plasma proteins was lower in DM (4.22 ± 1.69 versus 5.25 ± 1.80 μmol/g protein, p = 0.025). Serum retinol did not differ and α-tocopherol tended to be lower in DM only when expressed relative to serum lipids (3.97 ± 1.14 versus 4.67 ± 1.32 μmol/mmol cholesterol + triacylglycerols in CO, p = 0.054). In contrast to no differences in glutathione peroxidase (GPX), superoxide dismutase (Cu/Zn SOD) was higher in DM (1008 ± 224 versus 845 ± 195 U/g Hb, p = 0.003). Plasma lipid peroxidation products in the form of d-ROM and MDA were higher in DM (0.622 ± 0.176 versus 0.478 ± 0.130 μM, p < 0.0005) but this difference disappeared when expressing MDA relative to cholesterol or to total lipids (cholesterol + triacylglycerols) in serum. In the DM group, glycemic control monitored as HbA1c was negatively related to the endogenous circulating antioxidants uric acid (r = –0.55, p = 0.007), glutathione, (r = –0.38, p = 0.038) and bilirubin (r = –0.49 p = 0.017) and positively to d-ROM (r = 0.39 p = 0.039). These relationships were not observed in the CO.

Table 2 Oxidative stress status of the study subjects
Full size table

Endothelial function as assessed by brachial artery FMD, was significantly impaired in DM (6.68 ± 1.98 versus 7.92 ± 1.60%, p = 0.004 for the independent effect of DM) and this impairment tended to be more pronounced in boys (p = 0.06 for the interaction DM and gender). Carotid intima media thickness (0.495 ± 0.063 in DM versus 0.481 ± 0.052 mm in CO, p = 0.14) and blood pressure were not significantly different in diabetes and were not affected by gender. It should be noted, however, that the power to detect differences in these parameters only amounted to <0.31 in this study.

Analysis of the factors influencing FMD revealed, respectively, a positive and negative relationship with serum HDL cholesterol and triacylglycerols in CO but these relationships were lost in DM (Fig. 1, A and B). In contrast, there was no significant correlation with LDL cholesterol (Fig. 1C) or with parameters of glycemic control (fasting plasma glucose or HbA1c) in either group. In the diabetic group but not in the CO, FMD was negatively related to carotid IMT (Fig. 1D). Moreover, IMT was strongly inter-related with daily insulin dose (units/kg) in the DM (r = 0.46, p = 0.007). A multiple regression model including these variables shows that the variance of FMD is explained by triacylglycerols (24%) in CO and by IMT (or insulin dose) in DM (18%). In the whole study population or in either group separately, neither FMD nor carotid IMT were significantly related to levels of circulating nonenzymatic antioxidants or peroxidation products. The sole marker of OS status, which was related to FMD, was Cu/Zn SOD but only in DM (Fig. 2A). Comparison of the groups with Cu/Zn SOD values higher or lower than 955 U SOD/g Hb (median of the whole population) revealed that the high Cu/Zn SOD group had better FMD in both girls and boys of the DM and the CO groups (p = 0.035 for the independent effect of high Cu/Zn SOD after correcting for the effect of diabetes and gender) (Fig. 2B). Diabetic girls had normal FMD values when their Cu/Zn SOD activity was above the median (8.08 ± 2.29 versus 6.38 ± 1.18% in the low Cu/Zn SOD subgroup). In diabetic boys, the FMD was also better in the high Cu/Zn SOD subgroup (6.44 ± 1.43% versus 4.94 ± 1.92% in the low Cu/Zn SOD subgroup) but did not reach the levels seen in control boys (Table 2).

Figure 1
figure 1

Scatter plots of the relationship between flow mediated vasodilatation and serum HDL cholesterol (A), triacylglycerol (B), LDL cholesterol (C), and carotid intima media thickness (D) in diabetic (• and continuous line) and nondiabetic controls (X and discontinuous line).

Full size image
Figure 2
figure 2

(A) Scatter plot of the relationship between flow mediated vasodilatation and erythrocyte Cu/Zn superoxide dismutase activity in diabetic (• and continuous line) and nondiabetic controls (X and discontinuous line). (B) Histogram showing flow mediated dilatation in the subgroups divided according to diabetes, gender, and level of erythrocyte Cu/Zn SOD. Shown are means + SEM of diabetic patients (□) and control subjects (▪). High and low Cu/Zn SOD refers to the subgroups having Cu/Zn SOD levels respectively ≥ and <955 units/g Hb, which corresponds to the median of the whole study population. Three-factor analysis of variance detects an independent effect of diabetes (p = 0.002) and SOD level (p = 0.035) but not of gender (p = 0.14). The interaction between the factors was: diabetes × gender (p = 0.08), diabetes × SOD level (p = 0.21), and gender × SOD level (p = 0.98).

Full size image

Pubertal stage was not associated with differences in FMD, IMT, Cu/Zn SOD, insulin dose, or OS parameters. Indeed, the lower FMD in diabetic children (p = 0.004 for the independent effect of diabetes) persisted after correcting for the confounding effect of pubertal stage with and without that of gender (using three- and two-factor ANOVA). However, the power to detect an independent or interactive effect of pubertal stage only amounted to 0.25 in this study.

DISCUSSION

In this study, we describe OS status in young and adolescent type 1 diabetic patients and its relationship with endothelial function and carotid intima media thickness. Enhanced OS in this group of diabetic patients was illustrated by the increase in plasma peroxidation products (d-ROM and MDA), the decrease in antioxidant molecules (ascorbic acid, urate, and tocopherol relative to lipids) and the decrease in the antioxidant capacity of proteins (thiol content) and of plasma globally (TAC). The negative relationship between endogenous scavenger antioxidants (uric acid, glutathione, and bilirubin) and levels of HbA1c as well as the positive one with peroxidation products points out to the important role of glycemic control in antioxidant/oxidant balance in these young DM patients. Some antioxidants are probably lost due to renal hyperfiltration in the early phase of T1DM (10). In addition, hyperglycemia leads to redox imbalances and glycation of enzymes and can thus cause impairment in the recycling of molecules such as ascorbate and thiols (11). These findings, together with the increase in Cu/Zn superoxide dismutase, are in accordance with other reports on children (5,6) and on young and adult T1DM patients (12). Enhanced lipid peroxidation in vivo has recently been confirmed by detecting increases in urinary F2 isoprostanes which were more marked at onset of the disease and improved in parallel to glycemic control (13).

Diabetes was associated with impairment of FMD that was more outspoken in the diabetic boys. We did not find a correlation between FMD and LDL cholesterol levels in either group, as also seen in older T1DM patients with and without microalbuminuria (14) but in discordance with other studies on younger diabetic children (median 11 y) with a slightly shorter duration of diabetes (4 y) (7). The deleterious effect of triacylglycerols and the protection afforded by HDL cholesterol in the healthy controls were lost in the diabetic group. Taken together, these observations suggest that in this diabetic group of children the impairment of FMD was related to factors other than their worse lipid control.

Despite the fact that our study was not sufficiently powered to detect the diabetes-related differences in IMT which have been reported by Jarvisalo (7), we observed a thicker carotid intima media in the children receiving higher insulin doses (corrected for weight) as also observed in a survey of adult type 1 diabetic patients (15). Both these factors were negatively related to FMD. In contrast, the only factor positively related to FMD in diabetic children was Cu/Zn SOD. We observed that the sub-group of diabetic children with higher circulating erythrocyte Cu/Zn SOD levels had less FMD impairment. In girls, this subgroup had normal FMD function but in boys, the higher Cu/Zn SOD status did not fully counterbalance their more outspoken FMD impairment. These observations suggest that the level of increase in SOD activity seen in diabetic children may be relevant in the defence against the impairment of FMD caused by diabetes.

Since SOD is the enzyme responsible for the neutralization of superoxide and because we did not observe any relationship between FMD and glutathione peroxidase, our data suggest the dominant role of superoxide rather than of peroxides in this diabetic endothelial dysfunction. This idea is supported by the lack of correlations with other markers of OS. The batteries available to monitor OS are still limited and fragmentary and do not allow direct monitoring of superoxide production in vivo in human subjects. Nevertheless, multiple observations in animals models have demonstrated an increased production of superoxide in diabetes, especially in the vessel wall (2). Potential mechanisms are the hyperglycemia induced activation of protein kinase C which in turn stimulates NADPH-oxidases to generate superoxide and reactive oxygen species (16). Superoxide interferes with the generation of NO production in several ways such as a decrease in the endothelial nitric oxide synthase (eNOS) expression mediated by activator protein, AP-1 (17), a change in the electrophysiological state of the endothelial cell (18) and the availability of tetrahydrobiopterin, an essential cofactor of eNOS (19). The reaction of superoxide with NO in the vascular wall also contributes to the decreased bioavailability of vascular NO, thus impairing vasodilatory response, as observed in this and other studies.

In several models of diabetes, SOD or SOD-mimetic treatments reverse the impairment of endothelium-dependent relaxation (20–22). It should, however, be stressed that care should be taken when extrapolating conclusions from diabetic animal models to the human situation since, for example, diabetogenic agents such as streptozotocin can influence the gene expression of SOD directly with as a result decreases in SOD activity (23).

Data on the protective role of SOD in human diabetic patients are scarce and may differ depending on the type and duration of diabetes as well as race. Brownlee's unifying theory of complications in diabetes focuses on superoxide release by the mitochondria (4), which would be cleared up by the mitochondrial SOD (the Mn isoform). Our study, however, identified a potential protective role of the cytoplasmic form (Cu/Zn SOD), which is present in the red blood cells. This suggests that the superoxide responsible for the endothelial dysfunction is largely extracellular and present in the lumen of the vessel where it can be neutralized by the red cell Cu/Zn SOD. The relevance of the extracellular form of SOD (ecSOD) in the protection against several phases in the atherosclerotic process is also highlighted by the association of a mutation in the ecSOD gene with prognosis in diabetic hemodialysis patients (24). Indeed, the lower coronary ecSOD activity in adult coronary artery disease patients and its negative correlation to FMD (25) indicate that it plays a key determinant role in the bioavailability of vascular NO (26).

In addition to genetically predetermined levels, up-regulation of SOD synthesis in response to the oxidative injury or glycation caused by hyperglycemia has also been described, again in cell cultures and animal models (27). Interestingly, in T1DM patients with nephropathy, SOD is less up-regulated than in patients without nephropathy, again suggesting that SOD plays an important role in the susceptibility to develop diabetic complications (28).

It is not clear why high Cu/Zn SOD levels are associated with a normal FMD in diabetic girls but not in boys in our study. Apart from the more pronounced impairment in the boys, which would require more intensive counter-regulatory measures, other (hormonal) factors might play a significant role. Apart from the well-established direct receptor-mediated beneficial effects of estrogens on the vascular wall (29), several observations suggest that they can also modulate the vascular oxidant-antioxidant balance (30,31). Our study was not sufficiently powered to study the independent effect of puberty and its influence on the diabetes-induced impairment of FMD. This question needs further investigations aimed at analyzing the relationship between status of specific hormones (for example, also including growth hormone) and endothelial function.

In view of the increasing perception of the serious cardiovascular risk already present in young T1DM patients (32) and of its rapid progression in young adults (33), early intervention should target the modifiable risk factors in genetically predisposed individuals (with a positive family history of atherosclerosis). Although metabolic control remains the primordial therapeutic target in these patients, it does not explain the total coronary artery disease risk in T1DM (34). In addition to current antismoking campaigns, stimulation of regular physical exercise from an early age, treatment of dyslipidemia and hypertension, our results suggest that targeting OS could possibly help to counteract the early disruption of the delicate redox balance within the vasculature and the endothelial dysfunction caused by diabetes. For example, both vitamin C and E mimic SOD in neutralizing the superoxide production induced by high glucose and have been shown to reduce endothelial dysfunction in DM (35). It has also been shown that in healthy individuals, ascorbate can reverse the acute impairment of vasodilatation caused by hyperglycemia (36). Since our results show that diabetic children have lower plasma ascorbate levels, the promotion of a healthier nutrition based on more fruit intake may be a feasible approach to target this vascular dysfunction in childhood diabetes.

In conclusion, diabetic children have an impaired endothelial function and elevated markers of OS. Those children with a higher erythrocyte Cu/Zn SOD activity have a better endothelium mediated vascular function. Our results suggest that SOD could play a relevant protective role but we acknowledge that an observed association does not demonstrate causality and that the important influence of insulin treatment and possibly resistance requires further investigation. Sufficiently powered prospective and intervention studies are therefore needed to investigate whether levels of erythrocyte Cu/Zn SOD are related to, and can even predict susceptibility to, early endothelial dysfunction and atherosclerosis in young diabetic patients.