Introduction

A rapidly rising epidemic of type 2 diabetes worldwide portends a gloomy increase of cardiovascular disease (CVD). In type 2 diabetes, dyslipidaemia is a powerful modifiable CVD risk factor [1, 2]. Diabetic dyslipidaemia is a cluster of atherogenic lipid alterations characterised by raised fasting triglycerides (TG), excessive postprandial lipaemia, small dense LDL and low HDL cholesterol [3]. Whether triglycerides are an independent CHD risk factor has been an ongoing topic of debate [4–6]. Recent data from the Hoorn study by Bos and co-workers clarified the issue showing that high triglycerides are a risk factor of CVD in subjects with abnormal glucose metabolism, but not in those with normal glucose metabolism [7]. VLDL particles are heterogeneous and can be separated into two major subspecies, large buoyant VLDL 1 particles (Sf 60–400) and smaller dense VLDL 2 particles (Sf 20–60) [8]. Substantial evidence demonstrates that VLDL 1 and VLDL 2 particles are metabolically diverse and regulated independently of each other [8, 9]. Although VLDL particles are the major carriers of triglycerides, there is little information on the distribution and composition of VLDL subspecies in people with type 2 diabetes.

Apolipoproteins (apo) CIII, CII and E are key functional proteins of VLDL metabolism. Apo CIII is synthesised in the liver and intestine and is a component of triglyceride-rich lipoproteins (TRLs), LDL and HDL [10]. Apo CIII is the most abundant apolipoprotein in VLDL particles and is correlated closely with concentrations of serum total and VLDL triglycerides [11, 12]. Apo CIII modulates metabolism of TRLs at two subsequent steps [13]. Firstly, apo CIII is an inhibitor of lipoprotein lipase activity, and secondly it interferes with apolipoprotein E-mediated receptor binding, delaying removal of remnant particles. So far little attention has been paid to the distribution of apo CIII between VLDL subspecies. Likewise only a few small studies have reported apo CIII levels in a diabetic population and the data are inconclusive [14–17]. In the context of CHD risk, current evidence suggests that the increase of apolipoprotein CIII in TRLs may be an independent risk factor for coronary heart disease [18–21].

In contrast to apo CIII, apo CII is a specific activator of lipoprotein lipase [10] and patients with mutations of the apo CII gene display variable elevation of serum triglycerides [22]. The balance between apo CII and apo CIII is considered to be critical for the efficacy of TRL hydrolysis by lipoprotein lipase [10]. Apo E acts as a primary ligand for LDL receptor and LDL receptor-related protein [23, 24] and consequently apo E plays a central role in the removal process of VLDL remnants. Emerging evidence suggests that apo E also has a critical role in intracellular lipid metabolism [25, 26]. These functions of apo E should be protective against atherosclerosis. So far the available data on apo CII and apo E in type 2 diabetic subjects are surprisingly limited.

The purpose of this study was to assess the composition and concentration of VLDL 1 and VLDL 2 subspecies, specifically apo CIII concentration in VLDL species in type 2 diabetic and non-diabetic subjects, to gain insight into the metabolism and atherogenicity of VLDL subspecies.

Subjects, materials and methods

Subjects

The FIELD (Fenofibrate Intervention and Event Lowering in Diabetes) study is a multinational, double-blind study started in 1998 in Australia, New Zealand and Finland [27]. In the Helsinki Centre 270 type 2 diabetic patients were recruited to the FIELD study. Of these patients, 217 (69 women) volunteered for this substudy. A group of 93 healthy subjects (44 women) was recruited by letters to the spouses of study patients, through advertisements in the hospital magazine and on the intranet, and by contacting retired pilots. The subjects were between 50- and 75-years-old, with no signs or history of clinical CVD except mild hypertension and no other major chronic diseases. We performed an oral 75-g glucose tolerance test to ensure normal glucose tolerance in accordance with WHO criteria [28]. Subjects with (1) impaired glucose tolerance or impaired fasting glucose, (2) cholesterol >6.0 mmol/l, (3) LDL cholesterol >4.0 mmol/l, (4) S-triglycerides >2.5 mmol/l, (5) transaminase levels over twice the upper limit of normal values, or (6) abnormal thyroid stimulating hormone concentrations, were not eligible.

Sixteen control subjects (17%) and 100 diabetes patients (46%) had medication for arterial hypertension. The median duration of diabetes was 6 years. The mode of diabetes treatment in the study subjects was (1) diet only in 29 (13%), (2) oral antihyperglycaemic agents (OHA) in 135 (62%), (3) insulin only in ten (5%), and (4) a combination of OHA and insulin in 43 (20%) subjects. There were 31 current and 95 former smokers among the diabetes patients, 91 of the patients had never smoked. Among the control subjects there were seven current and 33 former smokers, 52 control subjects had never smoked.

All patients and control subjects signed informed consent forms. The Ethical Committee of the Helsinki University Central Hospital approved the protocol.

Laboratory analyses

All examinations were performed during the placebo run-in period of the FIELD study before any fenofibrate intervention. All lipid measurements were performed in the research laboratory of the Helsinki University Central Hospital, Division of Cardiology, Helsinki, Finland. Blood samples were obtained after an overnight fast. Serum and EDTA plasma were separated by centrifugation and stored at −80°C until analysed. Fasting serum lipoproteins were isolated from fresh serum by sequential ultracentrifugation as described previously [29]. TRL subfractions were isolated by density gradient ultracentrifugation as previously described [30]. Enzymatic colorimetric assays were used to measure cholesterol (Unimate 7 CHOL, Hoffman-La Roche, Basel, Switzerland), triglyceride (Unimate 7 TRIG, Hoffman-La Roche), free cholesterol (Boehringer Mannheim, Mannheim, Germany) and phospholipid (Wako Chemicals, Neuss, Germany) concentrations in whole sera and in lipoprotein fractions using Cobas Mira automatic analyser (Hoffman-La Roche). Protein concentrations in the lipoprotein subfractions were measured by modification of the method of Lowry [31] (DC protein assay, Bio-Rad, Hercules, CA, USA). Immunoturbidimetric kits were used to measure total apo B (Orion Diagnostica, Espoo, Finland). Serum apo CII and apo CIII were measured using commercially available kits using a single radial immunodiffusion method (Daiichi Pure Chemicals, Tokyo, Japan). Due to discontinued production of the apo CII and CIII kits, the levels of only 77 control subjects were determined. Apo E phenotype was determined by a method based on isoelectric focusing of delipidated serum followed by immunoblotting [32]. Plasma glucose concentrations were analysed by a glucose dehydrogenase method (Precision-G Blood Glucose Testing System; Medisense, Abbott, IL, USA). HbA1c was measured using a commercially available kit (DCA 2000 Analyzer; Bayer Diagnostics, NY, USA). Serum-free insulin concentrations were determined by radioimmunoassay using the Phadeseph Insulin RIA kit (Pharmacia, Uppsala, Sweden).

Plasma samples and fractions of VLDL 1, VLDL 2 and IDL (intermediate-density lipoprotein) were frozen to −80°C and sent to the laboratory of Dr Fruchart (Department of Atherosclerosis, Pasteur Institute of Lille). Plasma apo CIII and apo E, as well as apo CIII in TRL fractions were measured by nephelometry. The polyclonal antibodies against apo CIII and apo E were generated in rabbit, using total synthetic apo CIII and purified apo E. An ELISA system was used to screen the antisera for antibody activity. The pooled antisera were precipitated with PEG and then dialysed. The specificity of the antibodies was tested by Western blotting against VLDL and HDL. For each antibody, only one band in VLDL and HDL lanes was detected at the expected weight. To confirm the specificity of the assays, the reactivity of the antisera was tested by nephelometry against the purified apolipoproteins and no signal was detected except with the corresponding purified protein against which the antibody was generated. The quantitation assays of apo CIII and apo E were performed on the Beckman immunochemistry system (IMMAGE), which measures the rate of change in light scatter induced by the formation of the immunoprecipitation complex between the polyclonal antibody and the corresponding apolipoproteins. The control and the standard used for this study were purchased from Behring. Samples (15 and 40 μl) were used for the quantitation of apo CIII and apo E, respectively, 200 μl of commercialised potassium phosphate polymer-enhancing buffer (buffer 1, ref: 447650, Beckman Coulter) and 20 μl of antisera were mixed in a reaction cell. The maximum rate of change of light scatter was measured at an optimal gain setting (gain 4 for apo CIII and 3 for apo E quantitation). The inter-assays CV were 5.2 and 4.2%, respectively, for apo E and apo CIII measurements. The recovery of apo CIII in lipoprotein fractions (VLDL, IDL, LDL, HDL and the bottom) was measured in eight samples and averaged 82±17%. The amount of apo CIII found in the density fraction >1.21 g/ml averaged 1.7±1.3 μg/dl, (i.e 0.06±0.05%). Plasma apo CIII levels measured by the two assays were closely correlated (r=0.772, p<0.001) although the absolute values obtained by nephelometry were lower.

Statistical analysis

Statistical analysis was performed using SPSS 11.0 for Windows (SPSS, Chicago, IL, USA). Results are shown as means and standard deviation or medians and interquartile range (IQR) in the case of skewed distribution of the variable. Factors that were not normally distributed were log transformed. Samples were compared using the general linear model, univariate analysis of variance, adjusted for age, BMI and sex. In the matched-pairs study, samples were compared using the paired samples t-test. The frequency distribution of apo E phenotype between the groups was compared with the Fisher’s exact test and the means of plasma apo E concentration in the groups of apo E phenotype were compared with the F-test and further with Scheffe’s method. Significant correlations were studied using two-tailed Pearson correlation for normally distributed variables and two-tailed Spearman correlation for non-normally distributed variables. Linear regression slopes for patients and controls were compared using GraphPad Prism version 3.00 for Windows (GraphPad Software, San Diego, CA, USA). A p value of <0.05 was considered significant in all analyses.

Results

Subject characteristics

As expected, control subjects were leaner with lower BMI and waist circumference (Table 1). Diabetic patients were in good glycaemic control with median HbA1c of 7.1%. The lipid profile of the diabetic patients was unfavourable, with higher triglyceride and lower HDL cholesterol concentrations. Expectedly, plasma total apo B was higher in the patients. The concentrations of plasma total apo CIII and apo CII were significantly elevated in diabetic patients (Table 1). Notably, the ratio of serum apo CII:apo CIII was elevated in type 2 diabetic patients (0.50±0.33 vs 0.40±0.10, p<0.001). Plasma apo E levels were reduced, resulting in markedly higher ratio of apo CIII:apo E in diabetic patients compared to control subjects (1.35±0.42 vs 0.73±0.21, p<0.001). The distribution of apo E phenotype did not differ between the groups. Plasma apo E concentrations were comparable between the groups of different apo E phenotypes in type 2 diabetic patients (data not shown). Likewise apo E phenotype did not have significant effects on triglyceride and cholesterol contents in VLDL subspecies (data not shown).

Table 1 Basic characteristics, metabolic variables, lipoprotein concentration and apolipoprotein concentration in type 2 diabetic patients and control subjects
Full size table

There was a close correlation between plasma apo CIII and triglycerides in non-diabetic subjects but the correlation was much weaker in type 2 diabetic subjects (r=0.762, p<0.001 vs r=0.473, p<0.001, p=0.001 for the difference between the slopes). In contrast the correlation between serum apo CII and triglycerides was stronger in type 2 diabetic subjects than in control subjects (r=0.641, p<0.001 vs r=0.431, p<0.001). Plasma apo E correlated positively with triglycerides in non-diabetic subjects but negatively in diabetic patients (r=0.513, p<0.001 vs r=−0.278, p<0.001, p<0.001 for the difference between the slopes). The levels of serum apo CII and apo CIII were closely correlated in control subjects (r=0.727, p<0.001) as well as in type 2 diabetic patients (r=0.601, p<0.001). Plasma apo CIII and apo E were strongly related in non-diabetic subjects (r=0.502, p<0.001) but this relationship was only marginal in the diabetic subjects (r=0.197, p<0.005).

Lipid and apolipoprotein composition of TRL subspecies

The percentage compositions of VLDL 1, VLDL 2 and IDL particles are presented in Table 2. In type 2 diabetic subjects VLDL 1 contained relatively more cholesterylester and phospholipids but less TG and free cholesterol than that of non-diabetic subjects. Consistently, VLDL 2 in type 2 diabetic patients was relatively enriched with cholesterylester and poor in free cholesterol and protein compared with control subjects. The composition of IDL particles was comparable in the two groups. Overall, the concentrations of all lipid components in both VLDL 1 and VLDL 2 were markedly higher in type 2 diabetic subjects than in non-diabetic subjects (data not shown). Consequently the mass concentrations of VLDL 1 and VLDL 2 were increased in type 2 diabetic compared with non-diabetic subjects by 125 and 82% (p<0.001). Apo CIII in VLDL1 particles was slightly higher in diabetic patients than in control subjects (0.52±0.33 vs 0.47±0.39 mg/dl). In contrast type 2 diabetic subjects had significantly less apo CIII in VLDL 2 (0.21±0.13 vs 0.31±0.22 mg/dl, p<0.001) and IDL particles (0.07±0.05 vs 0.11±0.07 mg/dl, p<0.001) than non-diabetic subjects.

Table 2 VLDL 1, VLDL 2 and IDL percentage composition in type 2 diabetic patients and control subjects
Full size table

We calculated the ratio of apo CIII:phospholipids and free cholesterol, the surface components of TRL particles. Notably, the ratio of apo CIII:phospholipids and free cholesterol in both VLDL 1 (0.037 vs 0.070, p<0.001) and in VLDL 2 (0.017 vs 0.040, p<0.001) as well as in IDL (0.005 vs 0.009, p<0.001) were lower in type 2 diabetic subjects than in control subjects. We also calculated the ratio of apo CIII:cholesterylester and triglyceride. Diabetic patients had markedly reduced ratios of apo CIII:cholesterylester and triglyceride in VLDL 1 (0.008 vs 0.014, p<0.001), VLDL 2 (0.006 vs 0.015, p<0.001) and IDL (0.003 vs 0.006, p<0.001). This implies that apo CIII does not increase in concert with core lipids in type 2 diabetes.

The relationship of VLDL triglycerides and serum triglycerides

Strong positive correlations existed between triglycerides in VLDL 1 and VLDL 2 and serum total triglycerides in both diabetic subjects and in control subjects (Fig. 1a,b). However, the slope for VLDL 1 was much steeper than that for VLDL 2 in both groups (p<0.001). The data indicate that for each increment of serum triglycerides there is a higher increase of triglycerides in VLDL 1 than in VLDL 2 particles. The regression lines of VLDL 1 TG and total TG were comparable between diabetic and the control group (Fig. 1c). Similarly, the regression lines of VLDL 2 TG and total TG were comparable between the two groups (Fig. 1d).

Fig. 1
figure 1

The relationship of VLDL 1 and VLDL 2 triglyceride and serum triglyceride concentrations in type 2 diabetic patients and control subjects. a Closed symbol: VLDL 1 TG; open symbol: VLDL 2 TG. r=0.879, p<0.001 for VLDL 1 TG; r=0.870, p<0.001 for VLDL 2 TG in the control group. Difference between the slopes p<0.001. b Closed symbol: VLDL 1 TG; open symbol: VLDL 2 TG. r=0.899, p<0.001 for VLDL 1 TG; r=0.771, p<0.001 for VLDL 2 TG in the patient group. Difference between the slopes p<0.001. c, d Closed symbol: type 2 diabetic patients; open symbol: control subjects. Difference between the slopes c p=0.27, d p=0.85

Full size image

Relationship between triglycerides and apolipoproteins in VLDL subspecies

We observed strong correlations between apo CIII and triglycerides in both VLDL 1 and in VLDL 2 in control subjects and in type 2 diabetic patients. Notably the slope for apo CIII in VLDL 1 in non-diabetic subjects was considerably steeper than in type 2 diabetic subjects (Fig. 2a, p<0.001). Likewise the slope for apo CIII in VLDL 2 was considerably steeper in normal subjects than in diabetic patients (Fig. 2b, p<0.001). Thus, for each increment of TG in VLDL 1 and VLDL 2 there was a smaller increase of apo CIII in VLDL 1 and VLDL 2 of diabetic subjects than in control subjects. The association between TG and apo CIII in IDL was significant in diabetic subjects (r=0.616, p<0.001) but not in control subjects (r=0.204, p=0.051).

Fig. 2
figure 2

The relationship of apo CIII and triglycerides in a VLDL 1 and b VLDL 2. Closed symbols: type 2 diabetic patients; open symbols: control subjects. a r=0.900, p<0.001 for control group; r=0.756, p<0.001 in type 2 diabetic patients. b r=0.803, p<0.001 for control group; r=0.575, p<0.001 for type 2 diabetic patients. Difference between the slopes: p<0.001 in both panels

Full size image

The matched-pairs study

The diabetic patients and control subjects were paired for sex, BMI (mean intra-pair difference 1.14 kg/m2) and serum TG levels (mean intra-pair difference 0.09 mmol/l). Altogether we found 52 matched pairs (31 male and 21 female pairs). Despite similar serum TG levels (1.2 mmol/l in both groups) diabetic subjects had a lower concentration of HDL cholesterol (1.5±0.4 vs 1.2±0.5 mmol/l, p=0.004). Notably the concentrations of both plasma apo B (96.0±13 vs 90.5±15 mg/dl, p=0.005) and apo CIII (4.37±0.8 vs 3.12±0.8 mg/dl, p<0.001) were higher in diabetic subjects than in control group. The concentrations of plasma apo CII and apo E did not differ between the groups. The concentration of apo CIII in VLDL 1 (0.34±0.18 vs 0.57±0.42 mg/dl, p=0.001), VLDL 2 (0.17±0.08 vs 0.35±0.25 mg/dl, p<0.001) and IDL (0.05±0.02 vs 0.12±0.08 mg/dl, p<0.001) was significantly lower in diabetic patients than in normal subjects.

Discussion

In this study we demonstrate that the elevation of VLDL 1 triglycerides is the major determinant of serum triglyceride concentration in healthy normolipidaemic subjects as well as in patients with type 2 diabetes. Both plasma apo CII and apo CIII levels were elevated but apo E levels were decreased in type 2 diabetes. Paradoxically we detected a relative deficiency of apo CIII in all TRL species in type 2 diabetic subjects, providing evidence that the concentration of apo CIII does not increase in concert with core lipids. Since apo CIII has a critical role of in the hepatic clearance of VLDL remnants, lack of apo CIII may result in an impaired removal of these particles and a prolonged residence time in circulation. Thus, the elevation of VLDL particles in type 2 diabetes will have adverse consequences on other lipoprotein subspecies like LDL and HDL subclasses, and together these changes will comprise a highly atherogenic cluster [3].

A potential limitation of this study is that we separated TRL fractions using density-gradient ultracentrifugation and may have lost some apo CIII into the density >1.20 g/ml lipid-free fraction during ultracentrifugation. However, we were able to detect only trivial amounts of apo CIII in the lipid free fraction. Our results are in line with the finding that ultracentrifugation causes less than 5% of total apo CIII to be stripped off in the density >1.21 g/ml fraction [33]. In contrast to apo CIII, the detachment of apo E seems to be more robust and about 20–40% of apo E is lost into the non-lipid fraction [34]. Clearly the lack of apo E measurements in VLDL subspecies prohibits any concise conclusion on the metabolic relevance of plasma apo E reduction in type 2 diabetes. However, in type 2 diabetes the inverse correlation between apo E and triglycerides opposed to the strong positive correlation observed in non-diabetic subjects hints at the derangement of apo E metabolism.

Lipoprotein kinetic studies have shown that VLDL particles are metabolically heterogeneous [8]. Substantial evidence demonstrates that both the production and catabolism of large TG-rich VLDL 1 (Sf 60–400) and smaller cholesterol-rich VLDL 2 (Sf 20–60) are independently regulated [8, 9]. The metabolic processing of VLDL 1 and VLDL 2 in the VLDL-IDL-LDL cascade has parallel but diverse pathways that produce IDL and LDL. So far very little attention has been paid to VLDL subspecies in diabetes although the elevation of serum triglycerides is a distinct feature of diabetic dyslipidaemia. Previous studies have reported a proportionally higher increase of VLDL 1 particles than that of VLDL 2 particles in small groups of type 2 diabetic patients [30, 35]. Our results expand these earlier observations demonstrating clearly that the elevation of VLDL 1 triglycerides is the major determinant of serum triglyceride concentration both in normal subjects and in type 2 diabetic individuals. In agreement, the NMR profile of VLDL subspecies indicated that the increase of VLDL in type 2 diabetes is primarily due to an increased number of large VLDL particles [36]. The slope for VLDL 1 TG over a substantial range of serum triglycerides was much steeper than that for VLDL 2. Thus, the actual contribution of VLDL 1 triglyceride to serum triglyceride concentration becomes more pronounced when the triglyceride level rises. The fact that both diabetic and non-diabetic subjects fell on the same regression line suggest that the relationship was probably not due to diabetes itself but linked to factors driving the elevation of VLDL 1 triglycerides.

The available data on lipid composition of VLDL particles in diabetic dyslipidaemia are controversial. The studies looking at VLDL particles as a single entity have reported VLDL particles to be triglyceride enriched [37, 38]. However, a relatively larger increase of VLDL 1 mass compared with VLDL 2 mass in type 2 diabetes will be reflected as a relative increase of TG when studying only the whole VLDL particles. In this study the relative content of triglycerides in VLDL 1 was slightly, albeit significantly less, in diabetic subjects than in control subjects. This is consistent with a previous study reporting a relative depletion of triglycerides in VLDL1 particles of diabetic subjects [39]. Overall our results agree with previous data demonstrating that the elevation of serum triglyceride concentration primarily reflects an increase in the number, but not the size, of particles in both normolipidaemic and hypertriglyceridaemic subjects [40].

So far the data on plasma apo CII and apo CIII in type 2 diabetes are limited and controversial. Apo CIII levels have been reported to be elevated in the diabetic population but the limitations of these studies are small cohorts and heterogeneity of the subjects [14–17]. Most studies have reported normal apo CII levels and a normal or slightly decreased apo CII:apo CIII ratio in type 2 diabetic subjects [17, 41, 42]. In this study we observed a significant increase of both apo CII and apo CIII concentrations. The fact that the ratio of apo CII:apo CIII in total plasma was elevated in type 2 diabetic subjects is noteworthy and argues against impaired lipolysis of VLDL particles. As expected, a positive correlation existed between apo CIII and triglycerides in both VLDL 1 and in VLDL 2 in normal subjects. Unexpectedly the regression lines for apo CIII in both VLDL 1 and VLDL 2 across a wide range of triglycerides were less steep in diabetic subjects than in non-diabetic subjects. Thus for each increment of triglycerides in VLDL 1 and VLDL 2 there was less apo CIII in diabetic subjects than in non-diabetic subjects. To explore further this unexpected finding we calculated the ratio of apo CIII over the surface lipids and core lipids. Both ratios confirmed the relative depletion of apo CIII in diabetic subjects as compared with non-diabetic subjects. Likewise, the relative deficiency of apo CIII in both VLDL 1 and VLDL 2 fraction was demonstrated in type 2 diabetic subjects matched for fasting triglyceride level and BMI with non-diabetic subjects. Interestingly a recent study [43] reported that diabetes status was associated with lower concentrations of apo C III containing large and small VLDL particles whereas hypertriglyceridaemia in itself was linked to an increase of apo CIII containing VLDL particles. Likewise in diabetic patients, a lower ratio of apo CIII:triglycerides in VLDL has been reported [16]. Thus, despite elevation of plasma apo CIII, diabetes status is linked with a relative deficiency of apo CIII in all TRL species.

Both apo CIII and apo CII have a broad distribution being components of VLDL and HDL particles and are readily exchangeable between VLDL and HDL depending on the dynamics of lipoprotein metabolism [11, 12, 44]. Further studies are requested to establish if there are abnormal shifts between apo CIII, CII and E between VLDL and HDL particles in type 2 diabetes. A large body of data provides evidence that the plasma apo CIII concentration is increased in hypertriglyceridaemic subjects and there is a close positive correlation between plasma apo CIII and TG concentrations [11, 12, 44]. Likewise, a positive correlation exists between levels of plasma apo CII and TG [33, 45]. This was true also in our type 2 diabetic and non-diabetic subjects, arguing against a major derangement of apo CII metabolism.

The key question is: how to explain the relative deficiency of apo CIII in TRL species in the context of increased VLDL production in type 2 diabetes? Direct input of apo CIII into VLDL seems to account for about 50% or more of apo CIII production in both normolipidaemic subjects and in patients with elevated serum triglycerides [44]. In agreement with this, the VLDL apo CIII production rate is increased in subjects with hypertriglyceridaemia and there is a close correlation between plasma apo CIII levels and apo B production [46, 47]. Obviously assumptions drawn from these studies in non-diabetic subjects do not necessarily apply to type 2 diabetes. Metabolic kinetic studies are required to establish the mechanism behind the relative deficiency of apo CIII in VLDL species in type 2 diabetes.

In summary, the elevation of VLDL 1 species is a key feature of diabetic dyslipidaemia. Our data also suggest that the associations of apo CIII and apo E with TRL metabolism are profoundly disturbed in type 2 diabetes. Our results emphasise the need to study the distribution and composition of VLDL subclasses separately to capture the atherogenicity of TRL subpopulations in diabetic dyslipidaemia.

Duality of interest M.-R. Taskinen has received lecture fees from Laboratoires Fournier.