The pathophysiology of insulin resistance and atherosclerosis may share a common inflammatory basis, maintaining endothelial dysfunction, suggesting why patients with T2DM (Type II diabetes mellitus) have an impaired prognosis after an MI (myocardial infarction), but it remains unclear how these parameters are inter-related. Forty patients with an MI (20 patients with and 20 patients without T2DM) took part in this cross-sectional study. Endothelium-dependent [FMD (flow-mediated dilation)] and -independent [NTG (nitroglycerine)] vasodilatation (determined by ultrasound), SI (insulin sensitivity index; determined by isoglycaemic–hyperinsulinaemic clamp) and serum levels of CRP (C-reactive protein), TNF-α (tumour necrosis factor-α), IL-6 (interleukin 6), resistin and adiponectin (determined by ELISA) were measured. Associations between FMD/NTG and SI, and CRP, TNF-α, IL-6, adiponectin, resistin, lipids, blood pressure, BMI (body mass index) and brachial artery diameter were then assessed. FMD (2.1 compared with 4.7%; P<0.05), NTG (14.9 compared with 21.2%; P<0.05) and SI [4.3 compared with 6.6 10−4 dl·kg−1 of body weight·min−1·(μ-units/ml)−1; P<0.05], and adiponectin levels (3.1 compared with 6.4 μg/ml; P<0.01) were all lower in patients with T2DM. TNF-α (6.9 compared with 1.8 pg/ml; P<0.01) and IL-6 (2.3 compared with 1.2 pg/ml; P<0.01) levels were higher in patients with T2DM, whereas differences in CRP and resistin levels did not attain statistical significance between the two groups. TNF-α concentrations and brachial artery diameter were negatively, whereas SI was positively, correlated with FMD. Adjustment for age weakened the association for SI, whereas TNF-α and brachial artery diameter remained significantly associated with FMD after adjustment for group, age and BMI. Endothelial dysfunction and low-grade inflammation co-exist in T2DM after MI. These results suggest that the endothelium is negatively impacted in multiple ways by the diabetic state after an MI.

INTRODUCTION

Patients with T2DM (Type II diabetes mellitus) have a worse prognosis following an MI (myocardial infarction) than patients without T2DM suffering from an MI, which could be partly explained by traditional risk factors [1]. In recent years, a number of studies have been published that point to chronic low-grade inflammation as an important player in the pathogenesis of insulin resistance and T2DM [2]. Also, it has been proposed that atherosclerosis is an inflammatory disease [3]. Thus the pathophysiology of T2DM and atherosclerosis may share a common inflammatory basis, maintaining endothelial dysfunction.

Endothelial function is a barometer for cardiovascular risk [4], i.e. endothelial dysfunction is a major factor in the atherosclerotic progression and prediction of CVD (cardiovascular disease) outcome in humans [5]. Recent studies have documented increased inflammation in human diabetic atherosclerosis [6]. In insulin-resistant states, endothelial dysfunction is common, and associations between dyslipidaemia, low-grade inflammation, impaired glucose homoeostasis, adipocyte-derived factors and hypertension have been suggested [7]. Certain adipocyte-derived factors, e.g. adiponectin and resistin, as well as IL-6 (interleukin-6) and TNF-α (tumour necrosis factor-α) may be involved in initiating a pro-inflammatory milieu in T2DM, factors that may even directly affect the endothelium.

We hypothesized that endothelial dysfunction would be more advanced, involving low-grade inflammation, in patients with compared with those without T2DM after an MI. Therefore we investigated the associations between endothelial function, inflammatory markers, i.e. CRP (C-reactive protein), IL-6, TNF-α, and other parameters related to coronary disease, e.g. adiponectin and resistin, insulin resistance, lipids and BP (blood pressure) in 40 patients (20 patients without and 20 patients with T2DM) suffering an MI.

MATERIALS AND METHODS

Subjects

The study population consisted of 40 male Caucasian individuals, matched for age and BMI (body mass index), of whom 20 had T2DM and 20 did not. All subjects had an established stable coronary artery disease, having suffered an MI, according to European Society of Cardiology criteria [8], 3–6 months prior to the onset of the study. The standard post-MI pharmacological treatments are shown in Table 1. The study was approved by the local Ethics Committee, and written informed consent was obtained from the patients.

Table 1
Clinical, biochemical and pharmacological treatment data of the study groups

Data are means±S.E.M.; ARB, angiotensin receptor blocker; CKMB, creatine kinase MB; STEMI, ST-segment elevation MI; NSTEMI, non-ST-segment elevation MI.

SubjectsPatients with T2DMPatients without T2DMP value
n 20 20  
Age (years) 59±2 60±2 NS 
BMI (kg/m229±1 28±1 NS 
Smokers (nNS 
Waist circumference (cm) 104±2 103±2 NS 
SBP (mmHg) 127±4 124±3 NS 
DBP (mmHg) 79±2 80±2 NS 
HbA1c (%) 6.4±0.3 4.7±0.1 <0.01 
Creatinine (μmol/l) 83±3 86±3 NS 
Total cholesterol (mmol/l) 4.2±0.1 4.7±0.2 <0.05 
HDL-cholesterol (mmol/l) 0.9±0.1 1.3±0.1 <0.01 
LDL-cholesterol (mmol/l) 2.6±0.1 2.8±0.2 NS 
Triacylglycerols (mmol/l) 1.8±0.2 1.5±0.2 NS 
Cardiovascular parameters    
 STEMI/NSTEMI (n9/11 7/13 NS 
 Peak CKMB (μg/l) 118±45 145±47 NS 
 Ejection fraction (%) 45±5 50±2 NS 
Patients treated with    
 ACE inhibitor/ARB (%) 86 85 NS 
 β-Blocker (%) 100 100 NS 
 Calcium flow inhibitor (%) 14 20 NS 
 Acetylsalicylic acid (%) 100 100 NS 
 Statin (%) 77 65 NS 
 Insulin (yes/no) 15/5 −  
 Insulin (units/day) 48±6 −  
 Sulphonylurea (yes/no) 7/13 −  
 Metformin (yes/no) 5/15 −  
SubjectsPatients with T2DMPatients without T2DMP value
n 20 20  
Age (years) 59±2 60±2 NS 
BMI (kg/m229±1 28±1 NS 
Smokers (nNS 
Waist circumference (cm) 104±2 103±2 NS 
SBP (mmHg) 127±4 124±3 NS 
DBP (mmHg) 79±2 80±2 NS 
HbA1c (%) 6.4±0.3 4.7±0.1 <0.01 
Creatinine (μmol/l) 83±3 86±3 NS 
Total cholesterol (mmol/l) 4.2±0.1 4.7±0.2 <0.05 
HDL-cholesterol (mmol/l) 0.9±0.1 1.3±0.1 <0.01 
LDL-cholesterol (mmol/l) 2.6±0.1 2.8±0.2 NS 
Triacylglycerols (mmol/l) 1.8±0.2 1.5±0.2 NS 
Cardiovascular parameters    
 STEMI/NSTEMI (n9/11 7/13 NS 
 Peak CKMB (μg/l) 118±45 145±47 NS 
 Ejection fraction (%) 45±5 50±2 NS 
Patients treated with    
 ACE inhibitor/ARB (%) 86 85 NS 
 β-Blocker (%) 100 100 NS 
 Calcium flow inhibitor (%) 14 20 NS 
 Acetylsalicylic acid (%) 100 100 NS 
 Statin (%) 77 65 NS 
 Insulin (yes/no) 15/5 −  
 Insulin (units/day) 48±6 −  
 Sulphonylurea (yes/no) 7/13 −  
 Metformin (yes/no) 5/15 −  

T2DM subjects

None of the patients showed signs of microvascular diabetic complications, such as overt nephropathy, neuropathy or retinopathy. The duration of T2DM was 5.0±1.1 years. Patients were treated with hypoglycaemic agents as follows: metformin, sulphonylurea or both and insulin alone, or insulin plus metformin.

Non-diabetic subjects

All subjects were tested twice before the start of the experiment, with at least a 1-week interval, and had fasting blood glucose levels of 4.7±0.3 and 4.9±0.2 mmol/l respectively.

Study design

This was a cross-sectional study. After a 12 h overnight fast, subjects arrived at our laboratory for the experiment at 08.00 hours. Subjects were not allowed to eat or drink anything but water, and they refrained from taking their medication on the morning of the test day. For those subjects taking insulin, this was not given after 18.00 hours on the day before the test. After resting for at least 30 min in a quiet and dark room, endothelial function was measured by ultrasonography. Thereafter blood samples were taken for fasting biochemical analyses. Finally, the sensitivity to insulin-mediated whole-body glucose uptake was measured with the isoglycaemic–hyperinsulinaemic clamp technique. Subjects were clamped with regard to their fasting blood glucose levels in order to avoid artefacts caused by acute lowering of glucose levels with insulin.

FMD (flow-mediated dilation) and NTG (nitroglycerine)-induced vasodilatation

The diameter of the target artery was measured from two-dimensional ultrasound images, using a 7.0 MHz linear array transducer and a standard 128XP/10 system (Accuson), as described by Celermajer et al. [9]. After a resting time of 60 min, the subject's right arm was immobilized and the transducer was fixed in the same position throughout the study with the assistance of a mechanical arm. The brachial artery was scanned longitudinally and the transmit (focus) zone was set to optimize images of the lumen arterial wall interface. B-mode images were magnified by a resolution box and obtained with gating from the R wave of the ECG as trigging mode. The condition of reactive hyperaemia was induced by inflation of a pneumatic tourniquet placed around the forearm to a pressure of 300 mmHg for 4.5 min, followed by release [endothelium-dependent vasodilatation (FMD)]. Measurements were made at baseline (after 30 min of supine rest), and between 45 and 60 s after cuff release. Then, after 10 min rest, 0.4 mg of NTG spray was applied and new images were obtained 4 min later (endothelium-independent vasodilatation). Changes in the dBA (brachial artery diameter), the means of two images, were measured with an automated computerized analysing system. Briefly, this analysing program is a PC/Windows-based software with digitized ultrasound image. The starting point of the measurement area is set by the operator, and a 10 mm box is automatically drawn. The different echo interfaces are automatically outlined. If obvious errors are detected, it is possible to modify the measurement by marking a correct echo in the ultrasound image. In this case, only one or two manually marked points are needed to guide the automatic system to the correct interface.

Arterial flow velocity rates were obtained using a pulsed Doppler signal at 70° angles to the vessel in the centre of the artery. Ultrasound images were made 15–30 s after cuff release with the freeze mode. The volume flow was calculated by multiplying the velocity time integral of the Doppler flow signal for the mean of three pulse waves by the heart rate and vessel cross-sectional area. Calculations of blood flow and changes in arterial vasodilatation were done unaware of the subjects or the procedure.

Repeatability coefficients (RC) [calculated using the formula: RC=2√(ΣDi2/n), where Di is the absolute difference between two measurements, and n is the number of measurements] based on 32 individuals 1 week apart was 7.2 for FMD and 10.7 for NTG [10].

Isoglycaemic–hyperinsulinaemic clamp

The hyperinsulinaemic clamp was performed as described by DeFronzo et al. [11], and has been reported in detail previously [10]. The glucose clamp-derived index of SI [insulin sensitivity; units are 10−4 dl·kg−1 of body weight·min−1·(μ-units/ml)−1] was calculated from the GIR (glucose infusion rate), corrected for body weight, during the final 30 min as follows:

 
formula

where GIRSS is the steady-state GIR (mg/min), GSS is the steady-state blood glucose concentration (in mg/dl), and ΔISS is the difference between basal and steady-state plasma insulin concentrations (in μ-unit/ml).

Blood chemistry

All blood samples were fasting samples and were collected after the ultrasound procedure. Aliquots were placed on ice, centrifuged within 30 min, and separated plasma stored at −20 °C pending analysis. Plasma levels of adiponectin, TNF-α, IL-6 (R&D Systems), high-sensitivity CRP (Immunodiagnostik) and resistin (BioVendor) were determined by using ELISAs according to the manufacturers' instructions. Routine laboratory methods were used for the determination of plasma glucose, serum cholesterol and triacylglycerols.

Statistical analysis

Results are means±S.E.M., or percentages. As CRP, adiponectin, IL-6, TNF-α and triacylglycerols were skewed, these data were logarithmically transformed before entering the statistical models. Statistical comparisons between patients without and with T2DM were made using Student's unpaired t test and χ2 test. Pearson's correlation analysis was undertaken regarding every variable conceivably influencing endothelial function (FMD and NTG). We used multiple regression analysis for evaluating associations between each of the two dependent variables (FMD and NTG) and the independent variables. The independent variables were divided into five blocks: (i) lipid block [total cholesterol, LDL (low-density lipoprotein)-cholesterol, HDL (high-density lipoprotein)-cholesterol and triacylglycerols]; (ii) glucose homoeostasis block (fasting blood glucose, fasting insulin, fasting C-peptide and SI); (iii) inflammatory markers block (CRP, IL-6 and TNF-α); (iv) adipokine block (adiponectin and resistin), and (v) vascular variables block [dBA, SBP (systolic blood pressure) and DBP (diastolic blood pressure)]. Finally, a combined block model was created using the most significant variables from each block. The rationale for this block setting was the small sample size in combination with many independent variables. For adjustment of group (β1), a dummy variable was created (0, patients with T2DM; and 1, patients with out T2DM). Every block was then successively introduced for adjustment of the covariates and age (β2), and age and BMI (β3). P<0.05 was considered statistically significant. All statistical analyses were performed using the Statistica 6.0 software package (Statsoft).

RESULTS

Routine data

All clinical, biochemical and pharmacological treatment data of the study groups are shown in Table 1. Significant differences were seen in levels of HbA1c (glycated haemoglobin), total cholesterol, HDL-cholesterol and fasting blood glucose, which is consistent with the metabolic syndrome in T2DM [12].

Glucose homoeostasis data

All clamp data are given in Table 2. As we chose to clamp subjects under isoglycaemic conditions, i.e. at their fasting blood glucose levels, the steady-state blood glucose was higher in patients with T2DM compared with the patients without T2DM. GIR and SI were significantly higher in patients without T2DM, indicating a less insulin-resistant state in this group. However, patients without T2DM also had a far from a normal insulin sensitivity response, suggesting insulin resistance was also present in this group.

Table 2
Glucose homoeostasis, brachial artery flow, inflammatory and adipokine data in the study groups

Data are means±S.E.M. Steady-state clamp, mean of 90–120 min in the clamp procedure.

VariablesPatients with T2DMPatients without T2DMP value
n 20 20  
Glucose homoeostasis data    
 Glucose (mmol/l) 6.5±0.5 5.0±0.2 <0.01 
 Insulin (pmol/l) 116±51 76±21 NS 
 C-peptide (pmol/l) 698±93 883±100 NS 
 Steady-state clamp    
  Glucose (mmol/l) 6.5±0.2 4.9±0.1 <0.01 
  Insulin (pmol/l) 582±53 528±20 NS 
  C-peptide (pmol/l) 420±26 620±54 <0.05 
 GIR (mg·kg−1 of body weight·min−13.3±0.3 5.1±0.4 <0.01 
SI [10−4 dl·kg−1 of body weight·min−1·(μ-units/ml)−14.3±0.5 6.6±0.8 <0.05 
Brachial artery flow data    
 FMD (%) 2.1±0.6 4.7±0.6 <0.05 
 NTG (%) 14.9±1.5 21.2±1.2 <0.05 
Inflammatory markers    
 CRP (mg/l) 2.7±0.6 1.9±0.4 NS 
 IL-6 (pg/ml) 2.3±0.2 1.2±0.1 <0.01 
 TNF-α (pg/ml) 6.9±1.8 1.8±0.4 <0.01 
Adipokines    
 Adiponectin (μg/ml) 3.1±0.3 6.4±0.8 <0.01 
 Resistin (ng/ml) 3.7±0.2 4.4±0.6 NS 
VariablesPatients with T2DMPatients without T2DMP value
n 20 20  
Glucose homoeostasis data    
 Glucose (mmol/l) 6.5±0.5 5.0±0.2 <0.01 
 Insulin (pmol/l) 116±51 76±21 NS 
 C-peptide (pmol/l) 698±93 883±100 NS 
 Steady-state clamp    
  Glucose (mmol/l) 6.5±0.2 4.9±0.1 <0.01 
  Insulin (pmol/l) 582±53 528±20 NS 
  C-peptide (pmol/l) 420±26 620±54 <0.05 
 GIR (mg·kg−1 of body weight·min−13.3±0.3 5.1±0.4 <0.01 
SI [10−4 dl·kg−1 of body weight·min−1·(μ-units/ml)−14.3±0.5 6.6±0.8 <0.05 
Brachial artery flow data    
 FMD (%) 2.1±0.6 4.7±0.6 <0.05 
 NTG (%) 14.9±1.5 21.2±1.2 <0.05 
Inflammatory markers    
 CRP (mg/l) 2.7±0.6 1.9±0.4 NS 
 IL-6 (pg/ml) 2.3±0.2 1.2±0.1 <0.01 
 TNF-α (pg/ml) 6.9±1.8 1.8±0.4 <0.01 
Adipokines    
 Adiponectin (μg/ml) 3.1±0.3 6.4±0.8 <0.01 
 Resistin (ng/ml) 3.7±0.2 4.4±0.6 NS 

Inflammatory markers and adipokine data

All the data for inflammatory markers and adipokines are shown in Table 3. Interestingly, although the groups did not significantly differ with respect to age, BMI or waist circumference, there were significant differences in IL-6 and TNF-α, but not CRP, concentrations. Plasma levels of adiponectin were significantly higher in patients without T2DM compared with those with T2DM, whereas resistin levels did not differ between the groups.

Table 3
Pearson correlation coefficients between endothelial function (FMD and NTG), lipids, glucose homoeostasis, inflammatory markers and vascular variables in all subjects (n=40)

*P<0.05, †P<0.01, ‡P<0.001.

FMDNTGCholesterolHDLLDLTriacylglycerolGlucoseInsulinC-peptideSICRPIL-6TNF-αAdiponectinResistinBMISBPDBP
FMD                   
NTG 0.29                  
Cholesterol 0.21 0.32                 
HDL 0.38† 0.40† 0.31                
LDL −0.01 0.16 0.80‡ 0.03               
Triacylglycerol −0.05 0.01 0.28 −0.39† 0.01              
Glucose −0.43† −0.32* 0.07 −0.34* 0.03 0.43†             
Insulin 0.01 −0.12 0.02 −0.22 0.11 0.10 0.43†            
C-peptide 0.21 −0.06 −0.05 0.02 −0.10 0.13 −0.07 0.08           
SI 0.32* −0.01 −0.23 0.12 −0.27 −0.30 −0.32* −0.39* −0.16          
CRP −0.13 0.01 0.08 −0.05 0.18 −0.05 0.06 0.11 0.32* −0.17         
IL-6 0.02 0.12 0.21 0.17 0.24 −0.13 0.40* −0.25 0.27 −0.08 0.43†        
TNF-α −0.47† −0.40* −0.31 −0.46† −0.18 0.11 0.44† 0.43† −0.23 −0.02 −0.05 −0.33*       
Adiponectin 0.26 0.24 0.10 0.58‡ 0.07 −0.36* −0.49† −0.42† −0.03 0.49† −0.12 0.14 −0.47†      
Resistin −0.26 −0.17 −0.20 −0.20 0.10 −0.27 −0.04 −0.07 0.13 0.17 0.26 0.19 0.37* 0.02     
BMI −0.18 −0.29 −0.03 −0.28 0.10 0.26 0.21 0.14 0.14 −0.37* 0.15 0.16 0.34* −0.38† 0.06    
SBP −0.14 −0.16 −0.01 −0.05 0.15 −0.09 0.02 0.24 −0.09 −0.19 −0.01 0.02 0.17 0.17 0.02 0.31   
DBP −0.20 0.10 0.08 −0.05 0.22 −0.11 −0.04 0.07 −0.13 −0.02 0.05 0.06 0.13 0.13 0.01 0.14 0.72‡  
dBA −0.36* −0.14 −0.21 −0.02 −0.34* 0.12 0.26 −0.14 −0.07 0.12 −0.01 0.05 0.12 0.11 −0.14 0.04 −0.26 −0.19 
FMDNTGCholesterolHDLLDLTriacylglycerolGlucoseInsulinC-peptideSICRPIL-6TNF-αAdiponectinResistinBMISBPDBP
FMD                   
NTG 0.29                  
Cholesterol 0.21 0.32                 
HDL 0.38† 0.40† 0.31                
LDL −0.01 0.16 0.80‡ 0.03               
Triacylglycerol −0.05 0.01 0.28 −0.39† 0.01              
Glucose −0.43† −0.32* 0.07 −0.34* 0.03 0.43†             
Insulin 0.01 −0.12 0.02 −0.22 0.11 0.10 0.43†            
C-peptide 0.21 −0.06 −0.05 0.02 −0.10 0.13 −0.07 0.08           
SI 0.32* −0.01 −0.23 0.12 −0.27 −0.30 −0.32* −0.39* −0.16          
CRP −0.13 0.01 0.08 −0.05 0.18 −0.05 0.06 0.11 0.32* −0.17         
IL-6 0.02 0.12 0.21 0.17 0.24 −0.13 0.40* −0.25 0.27 −0.08 0.43†        
TNF-α −0.47† −0.40* −0.31 −0.46† −0.18 0.11 0.44† 0.43† −0.23 −0.02 −0.05 −0.33*       
Adiponectin 0.26 0.24 0.10 0.58‡ 0.07 −0.36* −0.49† −0.42† −0.03 0.49† −0.12 0.14 −0.47†      
Resistin −0.26 −0.17 −0.20 −0.20 0.10 −0.27 −0.04 −0.07 0.13 0.17 0.26 0.19 0.37* 0.02     
BMI −0.18 −0.29 −0.03 −0.28 0.10 0.26 0.21 0.14 0.14 −0.37* 0.15 0.16 0.34* −0.38† 0.06    
SBP −0.14 −0.16 −0.01 −0.05 0.15 −0.09 0.02 0.24 −0.09 −0.19 −0.01 0.02 0.17 0.17 0.02 0.31   
DBP −0.20 0.10 0.08 −0.05 0.22 −0.11 −0.04 0.07 −0.13 −0.02 0.05 0.06 0.13 0.13 0.01 0.14 0.72‡  
dBA −0.36* −0.14 −0.21 −0.02 −0.34* 0.12 0.26 −0.14 −0.07 0.12 −0.01 0.05 0.12 0.11 −0.14 0.04 −0.26 −0.19 

Endothelium-dependent (FMD) and -independent (NTG) vasodilatation

Baseline dBA FMD [4.2±0.1 compared with 4.1±0.1 mm; P=NS (not significant)] or NTG (4.3±0.1 compared with 4.1±0.1; P=NS) did not differ between patients with or without T2DM respectively. The maximal shear stress stimuli, FMD maximal flow (199±20 compared with 145±18 ml/min; P=NS), did not differ between patients with or without T2DM. However, patients with T2DM had impaired FMD and NTG responses (Table 2).

Correlation data

Pearson correlation data are outlined in Table 3. Between FMD and the glucose homoeostasis markers, inflammatory markers, adipokines and vascular variables, there were positive correlations with HDL-cholesterol and SI, whereas negative correlations were seen with fasting blood glucose, TNF-α (Figure 1) and dBA. Between NTG and the above-described variables, there was a positive correlation with HDL-cholesterol, whereas negative correlations were seen with fasting glucose and TNF-α. Additionally, between SI and the above-described variables there was a positive correlation with adiponectin, whereas negative correlations were seen with fasting blood glucose, fasting insulin and BMI. Surprisingly, no correlation between SI and inflammatory markers was observed.

Pearson correlation data between FMD (dependent variable) and TNF-α

Figure 1
Pearson correlation data between FMD (dependent variable) and TNF-α

The scatter diagram, involving all subjects (n=40), shows a significant correlation between FMD and TNF-α.

Figure 1
Pearson correlation data between FMD (dependent variable) and TNF-α

The scatter diagram, involving all subjects (n=40), shows a significant correlation between FMD and TNF-α.

Multiple regression data

FMD (Table 4)

For each separate block, the glucose homoeostasis and inflammatory blocks were significantly associated with FMD. In contrast, a poor association was seen between FMD and the lipid, adipokine and vascular block models. To adjust for potential covariates, we successively introduced covariates: group, then group and age, and finally group, age and BMI. After adjustment for these covariates, TNF-α and dBA remained significantly associated with FMD in each separate block. In contrast, the SI association with FMD disappeared after introducing age as a covariate, revealing an age-confounding effect. We then created a combined block model with the three most significant factors: TNF-α, dBA and SI, which resulted in an increased association with FMD and contributed 27% of the FMD variance. Again, we successively adjusted this combined block for the potential covariates age and BMI. After adjustment for age, the association between SI and FMD disappeared again, revealing an age-dependent effect responsible for this association. However, TNF-α and dBA remained significantly associated with FMD after adjustment. Finally, we wanted to adjust for other potential covariates in the last combined block model (SI, TNF-α and dBA). To this end, we tested, one by one, the following putative covariates: fasting glucose, fasting insulin, HDL-cholesterol, adiponectin, resistin, IL-6, CRP and HbA1c, revealing no changes in the significant association between TNF-α or dBA and FMD, thus ruling out these above factors as confounding factors (results not shown).

Table 4
Multiple regression block analysis with FMD or NTG as the dependent variable

β, Standardized coefficient; β1, adjusted for group; β2, adjusted for group and age; β3, adjusted for group, age and BMI. †P<0.05 and ‡P<0.01.

FMDNTG
BlockModel-adjusted βββ1β2β3Model-adjusted βββ1β2β3
Lipids R2=0.148, P=NS     R2=0.138, P=NS     
 Total cholesterol  −0.34 0.20 0.24 0.20  −5.34 −4.93 −4.61 −4.64 
 LDL-cholesterol  0.33 −0.27 −0.31 −0.27  5.1 4.64 4.38 4.41 
 HDL-cholesterol  0.65 0.17 0.16 0.15  2.8 2.47 2.37 2.36 
 Triacylglycerol  0.32 0.09 0.07 0.08  2.7 2.50 2.31 2.32 
Glucose homoeostasis R2=0.198, P<0.05     R2=0.034, P=NS     
 Glucose  −0.49‡ −0.11 −0.12 −0.18  −0.35† 0.20 0.19 0.16 
 Insulin  0.12 0.25 0.33 0.28  −0.09 0.10 0.26 0.23 
 C-peptide  0.25 0.13 0.11 0.15  −0.13 −0.31 −0.28 −0.23 
SI  0.28† 0.29† 0.20 0.22  0.21 0.18 0.21 0.25 
Inflammatory markers R2=0.152, P<0.05     R2=0.051, P=NS     
 CRP  −0.09 −0.16 −0.18 0.17  −0.06 −0.01 −0.01 −0.01 
 IL-6  −0.13 −0.15 −0.20 −0.19  0.01 −0.26 −0.20 −0.13 
 TNF-α  −0.48‡ −0.42† −0.40† −0.41†  −0.35† −0.05 0.02 0.06 
Adipokines R2=0.050, P=NS     R2=0.042, P=NS     
 Adiponectin  0.21 0.12 0.12 0.19  0.19 −0.20 −0.15 −0.23 
 Resistin  −0.07 −0.16 −0.16 −0.10  −0.12 −0.21 −0.23 −0.17 
Vascular variables R2=0.122, P=NS     R2=0.028, P=NS     
 dBA  −0.42† −0.33† −0.33† −0.30†  −0.21 −0.12 −0.06 −0.02 
 SBP  −0.18 −0.01 −0.02 0.09  −0.30 −0.12 0.04 0.17 
 DBP  −0.13 −0.24 −0.23 −0.28  0.34 0.24 0.14 0.08 
Combined analysis R2=0.271, P<0.01          
SI  0.27† 0.26† 0.20 0.14      
 TNF-α  −0.30† −0.28† −0.28† −0.26†      
 dBA  −0.34† −0.30† −0.31† −0.32†      
FMDNTG
BlockModel-adjusted βββ1β2β3Model-adjusted βββ1β2β3
Lipids R2=0.148, P=NS     R2=0.138, P=NS     
 Total cholesterol  −0.34 0.20 0.24 0.20  −5.34 −4.93 −4.61 −4.64 
 LDL-cholesterol  0.33 −0.27 −0.31 −0.27  5.1 4.64 4.38 4.41 
 HDL-cholesterol  0.65 0.17 0.16 0.15  2.8 2.47 2.37 2.36 
 Triacylglycerol  0.32 0.09 0.07 0.08  2.7 2.50 2.31 2.32 
Glucose homoeostasis R2=0.198, P<0.05     R2=0.034, P=NS     
 Glucose  −0.49‡ −0.11 −0.12 −0.18  −0.35† 0.20 0.19 0.16 
 Insulin  0.12 0.25 0.33 0.28  −0.09 0.10 0.26 0.23 
 C-peptide  0.25 0.13 0.11 0.15  −0.13 −0.31 −0.28 −0.23 
SI  0.28† 0.29† 0.20 0.22  0.21 0.18 0.21 0.25 
Inflammatory markers R2=0.152, P<0.05     R2=0.051, P=NS     
 CRP  −0.09 −0.16 −0.18 0.17  −0.06 −0.01 −0.01 −0.01 
 IL-6  −0.13 −0.15 −0.20 −0.19  0.01 −0.26 −0.20 −0.13 
 TNF-α  −0.48‡ −0.42† −0.40† −0.41†  −0.35† −0.05 0.02 0.06 
Adipokines R2=0.050, P=NS     R2=0.042, P=NS     
 Adiponectin  0.21 0.12 0.12 0.19  0.19 −0.20 −0.15 −0.23 
 Resistin  −0.07 −0.16 −0.16 −0.10  −0.12 −0.21 −0.23 −0.17 
Vascular variables R2=0.122, P=NS     R2=0.028, P=NS     
 dBA  −0.42† −0.33† −0.33† −0.30†  −0.21 −0.12 −0.06 −0.02 
 SBP  −0.18 −0.01 −0.02 0.09  −0.30 −0.12 0.04 0.17 
 DBP  −0.13 −0.24 −0.23 −0.28  0.34 0.24 0.14 0.08 
Combined analysis R2=0.271, P<0.01          
SI  0.27† 0.26† 0.20 0.14      
 TNF-α  −0.30† −0.28† −0.28† −0.26†      
 dBA  −0.34† −0.30† −0.31† −0.32†      

NTG (Table 4)

After adjustment for group, all significant associations with NTG disappeared, consistent with a group effect explaining the association of HDL-cholesterol, fasting glucose and TNF-α with NTG seen in the Pearson correlation analysis. For the separate blocks in this multiple regression model, none of the block models were significantly associated with NTG.

DISCUSSION

Endothelial dysfunction has been proposed to be a barometer for cardiovascular risk [4], which is widespread in subjects with T2DM in whom several independent risk factors, e.g. low-grade inflammation [7,13], may prevail. In the present study, patients with T2DM had impaired endothelial function with concomitantly increased plasma levels of TNF-α and IL-6, indicative of pro-inflammatory activity. It cannot be excluded that putative subtle differences in drug therapy between groups may have influenced FMD and the pro-inflammatory activity. The effects of statins, ACE (angiotensin-converting enzyme) inhibitors and insulin have been described in the literature to be mediated directly or indirectly by nitric oxide signalling pathways. Moreover, the anti-diabetic treatment in the T2DM group was not optimal, reflected by HbA1c of 6.4%, together with a low usage of metformin, both of which may have affected the results. TNF-α was the strongest variable associated with FMD, albeit only 15% of the change in FMD can be explained by changes in TNF-α, whereas the association of SI with FMD was age dependent. dBA was also associated with FMD, although greater FMD seen in small arteries does not necessarily reflect better conduit artery endothelial function [14]. No significant association of any of the tested variables was seen with NTG.

TNF-α is produced by a broad variety of tissues, including adipocytes, but also from macrophages recruited into adipose tissue, sustaining inflammation and impaired adipocyte function [15,16]. TNF-α also exerts vascular effects, i.e. contact-mediated activation of monocytes in the vessels [17]. TNF-α may cause endothelial dysfunction in several different ways; it diminishes the ability of arterial rings to relax in response to the endothelium-dependent vasodilator acetylcholine [18] and induces a transient reversible endothelial dysfunction in humans [19]. In cultured endothelial cells, TNF-α profoundly inhibits NO production, both after insulin and fluid shear stress, consistent with alterations in insulin function in the endothelium [20]. Also, TNF-α impairs intracellular insulin signalling, which improves after neutralization of TNF-α in a rodent model [21]. In humans, TNF-α inhibits insulin-sensitive glucose uptake and endothelium-dependent vasodilatation, which has been suggested to be a mediator between insulin resistance and endothelial dysfunction [22]. Interestingly, anti-(TNF-α) antibody treatment improves endothelial dysfunction in rheumatoid arthritis patients [23]. Although not all studies show benefits on endothelial function after lowering plasma TNF-α levels [24], there is a considerable amount of data supporting this idea. Surprisingly, we found no correlation between TNF-α and SI. A previous report [16] indicates that TNF-α mRNA expression and secretion in adipose tissue are inversely associated with insulin sensitivity, whereas a poor correlation exists with circulating TNF-α plasma levels. Therefore we cannot entirely rule out that TNF-α may still be linked to insulin resistance and endothelial dysfunction in our patients. An alternative explanation is that low-grade inflammation and insulin resistance contribute to endothelial dysfunction in parallel.

Plasma IL-6 levels were higher in patients with T2DM, but without any correlation to FMD. In addition, CRP did not correlate with FMD, which is contradictory with our recent findings [25], where changes in CRP followed changes in FMD in patients suffering an MI. However, considering that the present report was a cross-sectional study, differences in study design may explain these apparent discrepancies [25]. Very recently, two large cross-sectional studies, addressing whether inflammatory markers are related to endothelial function, have emerged. First, Verma et al. [26] studied the correlation between CRP and FMD in healthy subjects and found no correlation at all. Interestingly, in the same study, a weak correlation between CRP and FMD was observed in a subgroup of patients with severe endothelial dysfunction and risk factors for CVD [26]. Secondly, Vita et al. [27] observed a modest unadjusted correlation between CRP and IL-6 and FMD, which lost significance after adjustment for traditional CVD risk factors. Even if CRP has emerged as one of the most important predictors of CVD, it seems that the association with endothelial function is weak, at least in healthy subjects [26,27]. However, searching for new markers of inflammation linked to endothelial dysfunction is still a field in its infancy [28,29].

The NTG response was impaired in patients with T2DM. It has been suggested that smooth muscle dysfunction occurs independently of impaired endothelium-dependent vasodilatation in adults at risk of atherosclerosis [30]. However, we cannot rule out that a decreased sensitivity of the smooth muscle to nitric oxide and/or a structural change limiting vasodilatation could represent possible explanations for decreased FMD beyond a decrease in endothelium-derived nitric oxide release to shear stress stimulus. However, in the Pearson correlation analysis, TNF-α, fasting glucose and HDL-cholesterol concentration appeared to correlate with NTG, but, after group adjustment, this association disappeared.

Insulin has vasorelaxant effects, activating eNOS (endothelial nitric oxide synthase) and releasing nitric oxide [31], an effect that is blunted in insulin-resistance states [32,33]. SI was positively correlated with endothelial function, albeit age dependently. The association between insulin resistance and endothelial dysfunction has been demonstrated by some [34], but not all [35], groups. Nonetheless, it is believed that the principal effects of insulin on vascular beds are anti-atherogenic and antithrombotic and likely to be impaired under conditions of insulin resistance [34].

Patients with T2DM are over represented with the metabolic syndrome [12], consistent with the present study. Also, typical features of diabetic dyslipidaemia in T2DM were seen, i.e. decreased levels of HDL-cholesterol and to some extent increased levels of triacylglycerols. Lowered HDL-cholesterol may be involved in pro-inflammatory processes [36]. In the present study, there was a positive correlation between HDL-cholesterol and FMD, a relationship that disappeared after multivariate analysis. BMI was positively correlated with TNF-α, but negatively with SI. However, adjustment for BMI did not reveal any confounding effects of TNF-α association with FMD. Although BMI and waist circumference did not differ between groups, it is conceivable that macrophages or even endothelial cells may contribute to the differences in TNF-α levels between groups. Also, we cannot rule out that difference in visceral fat and its secretion products may be a reason for the differences seen in plasma TNF-α between groups, e.g. the adipocyte-specific plasma protein adiponectin. Plasma levels of adiponectin were lower in patients with T2DM compared with those without T2DM, whereas no difference was seen for resistin. Consistent with recent studies, no correlation between either adiponectin or resistin and FMD were seen [37,38]. Interestingly, adiponectin was positively correlated with SI, but negatively correlated with TNF-α and BMI. However, adjustment for adiponectin in the combined model did not reveal any changes in TNF-α association with FMD. Adiponectin has insulin-sensitizing and anti-inflammatory properties [39,40], and also significantly inhibits phagocytic activity and suppresses lipopolysaccharide-induced production of TNF-α [41].

It has been suggested that T2DM is the final result of an acute-phase reaction, during which cytokines are released in large amounts from adipose tissue as well as from activated macrophages recruited into adipose tissue [36]. We now suggest that patients with T2DM with established CAD have pronounced endothelial dysfunction concomitant with low-grade inflammatory activity. It seems that the endothelium is adversely impacted in multiple ways, where inflammation and insulin resistance are two important factors. The pro-inflammatory signalling pathways involved need to be explored in greater detail and may form the basis of potential drug targets against the global epidemic of insulin resistance and atherosclerosis.

Abbreviations

     
  • ACE

    angiotensin-converting enzyme

  •  
  • BMI

    body mass index

  •  
  • CRP

    C-reactive protein

  •  
  • CVD

    cardiovascular disease

  •  
  • dBA

    brachial artery diameter

  •  
  • DBP

    diastolic blood pressure

  •  
  • FMD

    flow-mediated dilation

  •  
  • GIR

    glucose infusion rate

  •  
  • HbA1c

    glycated haemoglobin

  •  
  • HDL

    high-density lipoprotein

  •  
  • IL-6

    interleukin-6

  •  
  • LDL

    low-density lipoprotein

  •  
  • MI

    myocardial infarction

  •  
  • NS

    not significant

  •  
  • NTG

    nitroglycerine

  •  
  • SBP

    sytolic blood pressure

  •  
  • SI

    insulin sensitivity index

  •  
  • T2DM

    Type II diabetes mellitus

  •  
  • TNF-α

    tumour necrosis factor-α

We thank Ms Lotta Larsson and Ms Christina Häll for excellent technical assistance, Mr Hans Pettersson and Ms Elisabeth Berg for excellent statistical advice, and Professor Mårten Rosenqvist for excellent and fruitful advice. Financial support was provided through the regional agreement on medical training and clinical research (ALF) between Stockholm County Council and the Karolinska Institute, and also was financially supported by Åke Wiberg's Foundation, the Nutricia Research Foundation, the European Foundation for the Study of Diabetes, the Sigurd and Elsa Golje Memorial Foundation, Svenska Diabetesstiftelsen, Berth von Kantzow's Foundation, and Stiftelsen Serafimerlasarettet.

References

References
1
Haffner
 
S. M.
Lehto
 
S.
Ronnemaa
 
T.
, et al 
Mortality from coronary heart disease in subjects with type 2 diabetes and in nondiabetic subjects with and without prior myocardial infarction
N. Engl. J. Med.
1998
, vol. 
339
 (pg. 
229
-
234
)
2
Sjöholm
 
A.
Nyström
 
T.
 
Endothelial inflammation in insulin resistance
Lancet
2005
, vol. 
365
 (pg. 
610
-
612
)
3
Ross
 
R.
 
Atherosclerosis: an inflammatory disease
N. Engl. J. Med.
1999
, vol. 
340
 (pg. 
115
-
126
)
4
Vita
 
J. A.
Keaney
 
J. F.
 
Endothelial function: a barometer for cardiovascular risk?
Circulation
2002
, vol. 
106
 (pg. 
640
-
642
)
5
Suwaidi
 
J. A.
Hamasaki
 
S.
Higano
 
S. T.
, et al 
Long-term follow-up of patients with mild coronary artery disease and endothelial dysfunction
Circulation
2000
, vol. 
101
 (pg. 
948
-
954
)
6
Moreno
 
P. R.
Fuster
 
V.
 
New aspects in the pathogenesis of diabetic atherothrombosis
J. Am. Coll. Cardiol.
2004
, vol. 
44
 (pg. 
2293
-
2300
)
7
Caballero
 
A. E.
 
Endothelial dysfunction, inflammation, and insulin resistance: a focus on subjects at risk for type 2 diabetes
Curr. Diabetes Rep.
2004
, vol. 
4
 (pg. 
237
-
246
)
8
The Joint European Society of Cardiology/American College of Cardiology Committee
Myocardial infarction redefined: a consensus document of The Joint European Society of Cardiology/American College of Cardiology Committee for the Redefinition of Myocardial Infarction
Eur. Heart. J.
2000
, vol. 
21
 (pg. 
1502
-
1513
)
9
Celermajer
 
D. S.
Sorensen
 
K. E.
Gooch
 
V. M.
, et al 
Non-invasive detection of endothelial dysfunction in children and adults at risk of atherosclerosis
Lancet
1992
, vol. 
340
 (pg. 
1111
-
1115
)
10
Nyström
 
T.
Nygren
 
A.
Sjöholm
 
A.
 
Tetrahydrobiopterin increases insulin sensitivity in patients with type 2 diabetes and coronary heart disease
Am. J. Physiol. Endocrinol. Metab.
2004
, vol. 
287
 (pg. 
E919
-
E925
)
11
DeFronzo
 
R. A.
Tobin
 
J. D.
Andres
 
R.
 
Glucose clamp technique: a method for quantifying insulin secretion and resistance
Am. J. Physiol.
1979
, vol. 
237
 (pg. 
E214
-
E223
)
12
Grundy
 
S. M.
Brewer
 
H. B.
Cleeman
 
J. I.
, et al 
Definition of metabolic syndrome: Report of the National Heart, Lung, and Blood Institute/American Heart Association conference on scientific issues related to definition
Circulation
2004
, vol. 
109
 (pg. 
433
-
438
)
13
Calles-Escandon
 
J.
Cipolla
 
M.
 
Diabetes and endothelial dysfunction: a clinical perspective
Endocr. Rev.
2001
, vol. 
22
 (pg. 
36
-
52
)
14
Silber
 
H. A.
Ouyang
 
P.
Bluemke
 
D. A.
, et al 
Why is flow mediated dilation dependent on arterial size? assessment of the shear stimulus using phase contrast magnetic resonance imaging
Am. J. Physiol. Heart. Circ. Physiol.
2004
, vol. 
288
 (pg. 
H822
-
H828
)
15
Xu
 
H.
Barnes
 
G. T.
Yang
 
Q.
, et al 
Chronic inflammation in fat plays a crucial role in the development of obesity-related insulin resistance
J. Clin. Invest.
2003
, vol. 
112
 (pg. 
1821
-
1830
)
16
Kern
 
P. A.
Saghizadeh
 
M.
Ong
 
J. M.
, et al 
The expression of tumor necrosis factor in human adipose tissue. Regulation by obesity, weight loss, and relationship to lipoprotein lipase
J. Clin. Invest.
1995
, vol. 
95
 (pg. 
2111
-
2119
)
17
Burger
 
D.
Dayer
 
J. M.
 
Cytokines, acute-phase proteins, and hormones: IL-1 and TNF-α production in contact-mediated activation of monocytes by T lymphocytes
Ann. N.Y. Acad. Sci.
2002
, vol. 
966
 (pg. 
464
-
473
)
18
Aoki
 
N.
Siegfried
 
M.
Lefer
 
A. M.
 
Anti-EDRF effect of tumor necrosis factor in isolated, perfused cat carotid arteries
Am. J. Physiol.
1989
, vol. 
256
 (pg. 
H1509
-
H1512
)
19
Bhagat
 
K.
Vallance
 
P.
 
Inflammatory cytokines impair endothelium-dependent dilatation in human veins in vivo
Circulation
1997
, vol. 
96
 (pg. 
3042
-
3047
)
20
Kim
 
F.
Gallis
 
B.
Corson
 
M. A.
 
TNF-α inhibits flow and insulin signaling leading to NO production in aortic endothelial cells
Am. J. Physiol. Cell. Physiol.
2001
, vol. 
280
 (pg. 
C1057
-
C1065
)
21
Hotamisligil
 
G. S.
Budavari
 
A.
Murray
 
D.
Spiegelman
 
B. M.
 
Reduced tyrosine kinase activity of the insulin receptor in obesity-diabetes. Central role of tumor necrosis factor-α
J. Clin. Invest.
1994
, vol. 
94
 (pg. 
1543
-
1549
)
22
Rask-Madsen
 
C.
Dominguez
 
H.
Ihlemann
 
N.
, et al 
Tumor necrosis factor-α inhibits insulin's stimulating effect on glucose uptake and endothelium-dependent vasodilation in humans
Circulation
2003
, vol. 
108
 (pg. 
1815
-
1821
)
23
Hurlimann
 
D.
Forster
 
A.
Noll
 
G.
, et al 
Anti-tumor necrosis factor-α treatment improves endothelial function in patients with rheumatoid arthritis
Circulation
2002
, vol. 
106
 (pg. 
2184
-
2187
)
24
Bilsborough
 
W.
O'Driscoll
 
G.
Stanton
 
K.
, et al 
Effect of lowering tumour necrosis factor-alpha on vascular endothelial function in Type II diabetes
Clin. Sci.
2002
, vol. 
103
 (pg. 
163
-
169
)
25
Nyström
 
T.
Nygren
 
A.
Sjöholm
 
A.
 
Persistent endothelial dysfunction is related to elevated CRP levels in Type II diabetic patients after acute myocardial infarction
Clin. Sci.
2005
, vol. 
108
 (pg. 
121
-
128
)
26
Verma
 
S.
Wang
 
C. H.
Lonn
 
E.
, et al 
Cross-sectional evaluation of brachial artery flow-mediated vasodilation and C-reactive protein in healthy individuals
Eur. Heart J.
2004
, vol. 
25
 (pg. 
1754
-
1760
)
27
Vita
 
J. A.
Keaney
 
J. F.
Larson
 
M. G.
, et al 
Brachial artery vasodilator function and systemic inflammation in the Framingham Offspring Study
Circulation
2004
, vol. 
110
 (pg. 
3604
-
3609
)
28
Szmitko
 
P. E.
Wang
 
C. H.
Weisel
 
R. D.
, et al 
New markers of inflammation and endothelial cell activation: Part I
Circulation
2003
, vol. 
108
 (pg. 
1917
-
1923
)
29
Szmitko
 
P. E.
Wang
 
C. H.
Weisel
 
R. D.
, et al 
Biomarkers of vascular disease linking inflammation to endothelial activation: Part II
Circulation
2003
, vol. 
108
 (pg. 
2041
-
2048
)
30
Adams
 
M. R.
Robinson
 
J.
McCredie
 
R.
, et al 
Smooth muscle dysfunction occurs independently of impaired endothelium-dependent dilation in adults at risk of atherosclerosis
J. Am. Coll. Cardiol.
1998
, vol. 
32
 (pg. 
123
-
127
)
31
Zeng
 
G.
Quon
 
M. J.
 
Insulin-stimulated production of nitric oxide is inhibited by wortmannin. Direct measurement in vascular endothelial cells
J. Clin. Invest.
1996
, vol. 
98
 (pg. 
894
-
898
)
32
Steinberg
 
H. O.
Chaker
 
H.
Leaming
 
R.
, et al 
Obesity/insulin resistance is associated with endothelial dysfunction. Implications for the syndrome of insulin resistance
J. Clin. Invest.
1996
, vol. 
97
 (pg. 
2601
-
2610
)
33
Scherrer
 
U.
Randin
 
D.
Vollenweider
 
P.
, et al 
Nitric oxide release accounts for insulin's vascular effects in humans
J. Clin. Invest.
1994
, vol. 
94
 (pg. 
2511
-
2515
)
34
Ritchie
 
S. A.
Ewart
 
M. A.
Perry
 
C. G.
, et al 
The role of insulin and the adipocytokines in regulation of vascular endothelial function
Clin. Sci.
2004
, vol. 
107
 (pg. 
519
-
532
)
35
Wendelhag
 
I.
Fagerberg
 
B.
Hulthe
 
J.
, et al 
Endothelium-dependent flow-mediated vasodilatation, insulin resistance and the metabolic syndrome in 60-year-old men
J. Intern. Med.
2002
, vol. 
252
 (pg. 
305
-
313
)
36
Pickup
 
J. C.
 
Inflammation and activated innate immunity in the pathogenesis of type 2 diabetes
Diabetes Care
2004
, vol. 
27
 (pg. 
813
-
823
)
37
Fernandez-Real
 
J. M.
Castro
 
A.
Vazquez
 
G.
, et al 
Adiponectin is associated with vascular function independent of insulin sensitivity
Diabetes Care
2004
, vol. 
27
 (pg. 
739
-
745
)
38
Shetty
 
G. K.
Economides
 
P. A.
Horton
 
E. S.
, et al 
Circulating adiponectin and resistin levels in relation to metabolic factors, inflammatory markers, and vascular reactivity in diabetic patients and subjects at risk for diabetes
Diabetes Care
2004
, vol. 
27
 (pg. 
2450
-
2457
)
39
Chandran
 
M.
Phillips
 
S. A.
Ciaraldi
 
T.
Henry
 
R. R.
 
Adiponectin: more than just another fat cell hormone?
Diabetes Care
2003
, vol. 
26
 (pg. 
2442
-
2450
)
40
Krakoff
 
J.
Funahashi
 
T.
Stehouwer
 
C. D.
, et al 
Inflammatory markers, adiponectin, and risk of type 2 diabetes in the Pima Indian
Diabetes Care
2003
, vol. 
26
 (pg. 
1745
-
1751
)
41
Yokota
 
T.
Oritani
 
K.
Takahashi
 
I.
, et al 
Adiponectin, a new member of the family of soluble defense collagens, negatively regulates the growth of myelomonocytic progenitors and the functions of macrophages
Blood
2000
, vol. 
96
 (pg. 
1723
-
1732
)