Apolipoprotein E Phenotype Frequency in Type II Diabetic Patients with Different Forms of Hyperlipoproteinemia K. G. Parhofer, W. O. Richter and P. Schwandt II. Medizinische Klinik, Klinikum Grosshadern der Ludwig-Maximilians-Universitat, Munchen, Germany
Atherosclerosis is the main cause of death in diabetes mellitus. This may at least in part be due to lipoprotein abnormalities which have been described in these patients. Apolipoprotein-E is a component of most lipoprotein fractions and plays an important role in the catabolism of VLDL. The different apolipoprotein-E phenotypes determined genetically are associated with certain hyperlipoproteinemias in a various degree in nondiabetic patients. In most cases apolipoprotein-E phenotype E2/2 is characteristic for familial dysbetalipoproteinemia. Phenotype E3/2 was found to be more frequent in hypertriglyceridemia while phenotype E4/3 was associated with hypercholesterolemia as well as with type V hyperlipoproteinemia. We studied apolipoprotein-E phenotypes and serum lipids in 141 type II diabetic patients (36 normolipidemic, 41 type IIa hyperlipidemic, 32 type IIb hyperlipidemic, 24 type IV hyperlipidemic, 8 type V hyperlipidemic). The phenotype E3/3 was more common in normolipidemic diabetic (77.8%) than in hyperlipoproteinemic diabetic patients (42.9%) or in the control
Introduction The major cause of premature death in diabetes mellitus is atherosclerosis (Tunbridge 1981). The accelerated vascular disease may in part be due to lipoprotein abnormalities which have been described in these patients (Winocour, Durrington, Ishola and Anderson 1986; Reaven 1987; Falco, Parr, Simpson and Wynn 1987). Although the most common abnormality of lipid metabolism in diabetes mellitus is the altered metabolism of VLDL (Howard 1987) there are also abnormalities in HDL and to lesser extent in LDL metabolism [Reaven 1987). Apolipoprotein-E (apo-E) is a constituent of most lipoprotein fractions {Blum, Aron and Sciacca 1980; Schwandt 1982) and is one of the major protein components of
Horm. metab. Res. 22 (1990) 589-594 © Georg Thieme Verlag Stuttgart -New York
group (57.5%). On the other hand phenotype E3/2 was more frequent in hypertriglyceridemic (50%) than in normolipidemic (5.6%) or hypercholesterolemic (hyperlipoproteinemia IIa: 4.9%, IIb: 9.4%) diabetic patients. The phenotype E4/3 was more frequent in all hyperlipoproteinemic diabetic patients, especially in those having hypercholesterolemia (34.2%) or mixed hyperlipidemia (50%). In conclusion we found a strong association between apo-E2 and hypertriglyceridemia in diabetic patients. This association was stronger than the one found in the general population. The association between apo-E4 and hypercholesterolemia in diabetic patients was similar to the one described in non-diabetic patients. We therefore conclude that type II diabetes mellitus is a possible cofactor in the apolipoprotein-E2 associated hyperlipoproteinemia. Key words Type II Diabetes Mellitus — Apolipoprotein E Phenotype — Hypertriglyceridemia — Hypercholesterolemia — Secondary Hyperlipoproteinemia
VLDL, modulating the catabolism of this triglyceride rich lipoprotein (Davignon, Gregg and Sing 1988). The three major isoforms of apo-E (E2, E3, E4) are genetically determined and controlled by three different alleles (e2, e3, e4) at the same gene locus and inherited in a co-dominant fashion, resulting in six possible phenotypes (E2/2, E3/2, E3/3, E4/3, E4/4, E2/4) (Zannis and Breslow 1982). All of them are known to be associated with hyperlipoproteinemias to a different degree. E3/3 is the most common phenotype in Germany and in other populations investigated and is more frequent in normolipidemic subjects than in any form of hyperlipidemia (Utermann, Kindermann, Kaffarnik and Steinmetz 1984). The phenotypes containing Apo-E2 (E2/2, E3/2) have been reported to be associated with higher levels of total and VLDL-triglycerides and E2/2 homozygosity is found in type III hyperlipidemia (dysbetalipoproteinemia) (Utermann et al. 1984; Assmann,
Received: 7 Febr. 1990
Accepted: 3 May 1990
Downloaded by: Universite Laval. Copyrighted material.
Summary
Horm. metab. Res. 22 (1990)
K. G. Parhofer, W. O. Richter and P. Schwandt
Table 1 Clinical data of type II diabetic patients.
men/women age (a) BMI (kg/m2) diab-durat. (a) therapy: diet oral agents insulin HbAic(%)
all n = 141
normolipidemic n = 36
77/64 63.3 2 26.1+0.55 7.3 8 22 45 74 7.79 8
18/18 67.0 5 25.0 8 8.3 7 6 13 17 7.34 0
1
hyperlipidemic n = 105 59/46 62.0 9 26.5 5 7.0 9 16 32 57 7.92 2
2
lla3 n = 41
lib4 n = 32
22/19 4 25.4 8 7.2 4 9 14 18 7.82 4
20/12 61.9+1.8 26.9 9 6.1 5 4 7 21 7.85 5
IV5 n = 24 12/12 64.0 9 26.6 4 8.8 2 1 11 12 8.06 0
V6 n=8 5/3 56.5 29.5 4.9 2 0 6 8.63
4 8 9
3
1
LDL-cholesterol < 3.89 mmol/l; triglycerides < 2.28 mmol/l LDL-cholesterol > 3.89 mmol/l and/or triglycerides > 2.28 mmol/l 3 LDL-cholesterol > 3.89 mmol/l; triglycerides < 2.28 mmol/l 4 LDL-cholesterol > 3.89 mmol/l; triglycerides > 2.28 mmol/l 5 LDL-cholesterol < 3.89 mmol/l; triglycerides > 2.28 mmol/l 6triglycerides > 5.17 mmol/l; chylomicrons detectable 2
Schmitz, Menzel and Schulte 1984; Eto, Watanabe and Ishii 1986; Robertson and Cumming 1985). A higher frequency of phenotypes containing apo-E4 (E4/3, E4/4) have been described in hypercholesterolemia (Utermann et al. 1984; Assmann et al. 1984) and in type V hyperlipoproteinemia (Ghiselli, Schaefer, Zech, Gregg and Brewer 1982a; Kuusi, Taskinen, Solakivi and Kauppinen-Maekelin 1988), but the latter finding is still a matter of controversy (Stuyt, Stahlenhoef, Demacker and Laar 1982). However the presence of either apo-E2 or apoE4 does not necessarily result in manifest hyperlipoproteinemia even in the case of homozygosity. Thus it can be concluded that only in the presence of other factors unidentified until now hyperlipoproteinemia occurs. The diabetic state is possibly one such cofactor, thus explaining the high incidence of hyperlipoproteinemia in diabetes mellitus. If this is the case, it should be expected that the isoforms E2 and E4 are much more predictive for hyperlipoproteinemia in diabetic than in non-diabetic patients. We therefore studied serum lipoproteins, apoE phenotype and glucose control in 141 patients with type II diabetes mellitus.
Lipoproteins Blood samples were collected after an overnight fast. Serum was immediately separated by low speed centrifugation. Cholesterol and triglycerides were measured enzymatically. Lipid assay procedures were standardized throughout the trial against reference material supplied by a national center of quality control for clinical laboratories (INSTAND). Furthermore all lipid values were determined in duplicate differing no more than . The VLDL fraction was separated by ultracentrifugation (d = 1.006 g/ml; 50000 rpm; 20 h; 4 °C; Beckmann TI50 rotor). The content of apolipoprotein B (VLDL-apoB) was measured nephelometrically in the top fraction. In the bottom fraction HDL-cholesterol was measured after precipitation of LDL with phosphotungstic acid. LDL-cholesterol was calculated by subtracting HDL-cholesterol from total bottom-cholesterol.
Apolipoprotein E Phenotype The isolated VLDL were stored at — 20 °C. Detection of apolipoprotein E pattern was done according to Utermann, Hees and Steinmetz (1977) with certain modifications: VLDL was delipidated with ethanol/ether (3:1 v/v). After drying under a stream of nitrogen the proteins were dissolved in 30 ul 0.01 mol/1 Tris HC1, 8 mol/1 urea, 0.01 mol/1 dithiothreitol, pH 8.6 and subjected to isoelectrofocussing in a pH gradient from 5—7 in thinlayer polyacrylamide gels containing 8 mol/1 urea. Protein bands were stained with coomassie brilliant blue.
Material and Methods Patients 141 patients with type II diabetes mellitus were recruited, their characteristics are shown in Table 1. The patients were subdivided intofivegroups according to their plasma lipid levels. Normolipidemia was defined as both LDL-cholesterol below 3.89 mmol/l and triglycerides below 2.28 mmol/l. The different forms of hyperlipidemia were defined as type Ha if LDL-cholesterol was above 3.89 mmol/l and triglycerides below 2.28 mmol/l, as type IIb if LDLcholesterol was below 2.28 mmol/l and triglycerides above 2.28 mmol/l and as type V if triglycerides were above 5.7 mmol/l and chylomicrons were present. For the purpose of this study no distinction was made between primary and secondary hyperlipidemia. All patients were in steady state conditions concerning their blood glucose control. Quality of blood glucose control was registered as HbAic. 22 of the diabetic patients were treated with diet alone, 45 took sulfonyl ureas and 74 were on insulin. Furthermore, the distribution of apolipoprotein E phenotypes (control group) was evaluated in 711 unselected patients (401 men, 310 women, age 44.0 1.5 years).
Statistics All values are expressed as mean values SEM and compared by Mann-Whitney-test; correlation was calculated by linear regression coefficient. Results The lipoprotein levels for all patients and the different groups of hyperlipoproteinemias are shown in Table 2. We found a significant inverse correlation between HDL-cholesterol and triglycerides as well as VLDLtriglycerides, VLDL-cholesterol and VLDL-apoB in the whole group as well as in all subgroups (except group IV and group V) (all: r = - 0 . 2 7 1 ; P < 0.001; without hyperlipoproteinemia: r = — 0.422; P < 0.005; with hyperlipoproteinemia:
Downloaded by: Universite Laval. Copyrighted material.
590
Apolipoprotein EPhenotype in Diabetic Patients
Horm. metab. Res. 22 (1990)
591
Table 2 Lipoproteins in type II diabetic patients with and without hyperlipoproteinemia. all n = 141 cholesterol triglycerides HDL-cholest. LDL-cholest. VLDL-cholest. VLDL-triglyc. VLDL-apoB (umol/l) VLDL-chol/trig VLDL-apoB/trig
normolipidemic n = 36
6.60 2.91 1.01 4.07 1.04
6 2 3 1 9 2 0.41 0.05 0.24 1 0.15 + 0.01
4.74 1.32 1.11 3.03 0.44 0.87 0.20 0.24
4 8 9 1 4 8 3 1 1
hyperlipidemic n = 105 7.25 1 3.45 7 0.98 3 4.43 2 1.22 2 2.62 7 0.52 6 0.24 1 0.14 + 0.01
lla n = 41
1 1
1 1 1 1
7 8 6 9 3 8 4 2 2
1.44 1.11 4.66 0.47 0.89 0.26 0.25 0.15
IV n = 24
lib n = 32 1
1
7.59 3.48 0.96 5.34
0 4 5 7 0 9 3 1 3
2.33 0.58 0.21 0.14
1 1
1 1 1 1
5.65 6 4.43 3 0.83 6 3.06 5 1.58 + 0.19 3.47 3 0.65 8 0.20 1 0.14 + 0.01
V n=6 2 1 2
1 1
14.7 25.2 0.67 3.08 5.41 6.35
0 1 8 1 7 3 2 3 1 2 1
0.37
3
1 3
3
All values in mmol/l except VLDL-chol/trig and VLDL-apoB/trig; VLDL-apoB was measured only in a subgroup of 46 patients (16 without hyperlipoproteinemia, 9 with type lla, 12 with type lib and 9 with type IV hyperlipoproteinemia); Significant vs. normolipidemic diabetics P < 0.001 Significant vs. normolipidemic diabetics P < 0.005 Significant vs. normolipidemic diabetics P < 0.05
Table 3 Frequencies of apolipoprotein E phenotypes in type II diabetic patients with and without hyperlipoproteinemia.
n = 141 E2/2 E3/2 E3/3 E4/3 E4/4 E2/4
normolipidemic n = 36
hyperlipidemic n = 105
lla
lib
IV
n = 41
n = 32
n = 24
V
n = 711
/
/
/
/
/
/
/
14.9% 51.8% 31.2% 0.7% 1.4%
5.6% 77.8% 13.9%
18.1% 42.9% 37.1% 1% 1%
4.9% 58.5% 34.2% 2.5%
9.4% 40.6% 50%
50% 16.7% 29.2%
25% 50% 25%
/ /
/ 4.2%
/ /
/ 2.8%
/
Controls
2.4% 13.9% 57.5% 24.2% 1.0% 1.0%
Table 4 Apolipoprotein E alleles frequencies in type II diabetic patients with and without hyperlipoproteinemia hyperlipidemic n = 105
lla
lib
IV
V
n = 141
normolipidemic n = 36
n = 41
n = 32
n = 24
n=8
Controlgroup n = 711
0.08 0.75 0.17
0.04 0.88 0.08
0.10 0.70 0.20
0.02 0.78 0.20
0.05 0.70 0.25
0.27 0.56 0.17
0.13 0.75 0.12
0.10 0.77 0.13
all
E2 E3 E4
r = - 0 . 2 7 4 ; P < 0.002; group lla: group lib: r = - 0.304; P < 0.001).
0.342; P < 0 . 0 1 ;
The apo-E phenotype frequency distribution (Table 3) of the whole group studied was similar to the one found in our control group and in the general population (Utermann et al. 1984; Assmann et al. 1984; Ghiselli, Gregg, Zech, Schaefer and Brewer 1982b). The phenotypes E3/2 and E4/3 were slightly more, the phenotype E3/3 slightly less frequent than in the normolipidemic non-diabetic population {Utermann et al. 1984; Assmann et al. 1984; Ghiselli et al. 1982b). Normolipidemic type II diabetic patients more frequently had the phenotype E3/3 (77.8%) than hyperlipidemic diabetic patients (42.9%; P < 0.001). The frequency of the E3/3 was different in the hyperlipidemic groups being highest and almost normal in the group lla (58.5 %) and lowest in group IV (16.7%). The frequency of the phenotype E3/2 was low in normolipidemic (5.6%) and hypercholesterolemic diabetic patients (group lla: 14.9%; group lib: 9.4%), but was as high as 50 % in group IV and 25 %> in group V. The frequency of the phenotype E4/3 was low in the normolipidemic group
(13.9%) and somewhat higher in group IV (29.2%) and group V (25%). But this phenotype was as frequent as 34.2 %> and 50%o in group lla respectively group lib. Statements concerning the rare phenotypes E2/2, E2/4 and E4/4 are not possible because of their low numbers. However these phenotypes are taken into account if the apo-E allele frequencies are compared (Table 4). While the frequency of the e3 allele is lowered in all hyperlipidemic groups (56 %—75 %) compared to the normolipidemic group (88%), the frequency of the e2 allele is elevated in group IV (27 %) and group V (13 %), whereas the e4 allele frequency is elevated in all hyperlipidemic diabetic patients being highest in group lla (20 %) and group lib (25 %). Comparing the average lipoprotein levels in the most common apo-E phenotypes (E3/3, E3/2, E4/3) (Table 5) we found that diabetic patients with the phenotype E3/2 had higher triglycerides, VLDL-triglycerides, VLDLcholesterol and VLDL-apoB whereas HDL-cholesterol and LDL-cholesterol were found to be lower compared to diabetic patients with the phenotype E3/3.
Downloaded by: Universite Laval. Copyrighted material.
all
Horm. metab. Res. 22 (1990)
K. G. Parhofer, W. O. RichterandP.
Table 5 Lipoproteins in type II diabetic patients depending on the apolipoprotein E phenotype. E3/3 n = 73
E3/2 n = 21 cholesterol triglycerides HDL-cholest. LDL-cholest. VLDL-cholest. VLDL-triglyc. VLDL-apoB (u.mol/1) VLDL-chol/trig. VLDL-apoB/trig.
7.10 4.26 1.01 3.86 1.53 3.24 0.71 0.22
8 4 1 8 8 6 1 2 1 1 1 1
6.42 2.44 1.06 4.04 0.80 1.71 0.38 0.25 0.16
E4/3 n = 44 4 7 5 3 1 6 7 2 2
6.73 4 2.93 5 0.96 5 4.30 1 1.09 + 0.16 55.0 + 8.76 0.35 5 0.22 1 2
3/5
3/5 3/6
5 4
All values in mmol/l except VLDL-chol/trig and VLDL-apoB/trig; VLDL-apoB was measured only in a subgroup of 46 patients (20 with phenotype E3/3, 11 with E3/2 and 15 with E4/3). Significant vs diabetic patients with phenotype E3/3; P < 0.001; Significant vs diabetic patients with phenotype E3/3; P < 0.005; Significant vs diabetic patients with phenotype E3/3; P < 0.01; Significant vs diabetic patients with phenotype E3/3; P < 0.05; Significant vs diabetic patients with phenotype E3/2; P < 0.001; Significant vs diabetic patients with phenotype E3/2; P < 0.005.
The comparison of hypertriglyceridemic diabetic patients (group IV) with apo-E phenotype E3/2 to those with apo-E phenotype E3/3 showed some interesting differences: The apo-E2 containing subgroup had higher levels of VLDL-apoB (0.69 umol/1 vs. 0.58 u.mol/1) and a higher ratio of VLDL-cholesterol to VLDL-triglycerides (0.22 vs. 0.19), thus indicating that the VLDL of the E2/3 subgroup of hypertriglyceridemic diabetic patients are richer in apo-B and in cholesterol than the VLDL of the E3/3 subgroup presenting with the same form of hyperlipoproteinemia. We could not find any differences in lipoprotein levels or ratios when we subdivided group Ha or group lib according to their apo-E phenotype (data not shown). The form of therapy (diet, sulfonyl ureas, insulin) was not associated with certain apo-E phenotypes in normolipidemic or hyperlipidemic diabetic patients, although we found a higher proportion of phenotype apo-E4/3 in the insulin treated patients (36% vs. 23 % in the group treated with diet alone or with sulfonyl ureas). However, this may be the consequence of the high proportion of patients with hyperlipoproteinemias type Ha or lib in the insulin treated group. Discussion In type II diabetic patients (n = 141) we found a similar frequency distribution of the apo-E phenotypes as described in the non-diabetic population (Davignon, Gregg and Sing 1988; Utermann et al. 1984; Assmann et al. 1984; Ghiselli etal. 1982b). Subdividing the whole group into those without hyperlipoproteinemia and into those with different forms of hyperlipoproteinemia we found normolipidemic patients to have predominantly the apo-E phenotype E3/3, while phenotype E4/3 was more frequent in diabetic patients with hypercholesterolemia (group Ha) or mixed hyperlipidemia (group lib). In hypertriglyceridemic diabetic patients we found an increase in frequency of phenotype E3/2 and to a
Schwandt
lesser extent also of phenotype E4/3, while the frequency of phenotype E3/3 was reduced. Comparing these results to the data on apo-E phenotype frequency in diabetic patients already present {Eto, Watanabe, Iwashima, Morikawa, Oshima, Sekiguchi and Ishii 1986; Eto, Watanabe, Iwashima, Morikawa, Chonan, Oshima, Sekiguchi and Ishii 1987; James, Voliotis, Grab and Pometta 1987; Vogelberg and Maucy 1988; Winocour, Tetlow, Durrington, Ishola, Hillier and Anderson 1989) we found a very similar distribution as Eto et al. 1987), who presented the following data for Japanese normolipidemic diabetic patients: E3/2 3.7%; E3/3 79.9%; E4/3 14.9%. Also for diabetic patients with hypercholesterolemia (group Ha or lib) we found the same tendency as Eto et al. (1987), but in our study, we found the e4 allele more frequent in group lib than in group Ha, while Eto et al. (1987) described it the other way around. Here on the other hand we found the frequency of the e2 allele unchanged in group Ha or lib compared to normolipidemic diabetic patients, while hypercholesterolemic patients had a diminished frequency of the e2 allele in Eto's {Eto et al. 1987) study. Concerning hypertriglyceridemic diabetic patients there are no data directly comparable to the results presented here. However it has been shown {Eto et al. 1986) that phenotype E3/2 diabetic patients had higher levels of triglycerides, VLDL-triglycerides, VLDL-cholesterol, apolipoprotein E and VLDL-cholesterol/VLDL-triglyceride ratios than phenotypic E3/3 diabetic patients. Vogelberg and Maucy (1988) found that hyperlipoproteinemia in insulin dependent diabetic patients is associated with apo-E2 heterozygosity, while it is induced by apo-E2 homozygosity in non-insulin dependent diabetic patients. Inconsistent with this finding Winocour et al. (1989) described an unexpected increase in apo-E2 homozygosity in insulin treated diabetic patients. Dividing our patients according to the form of therapy (diet, oral agents, insulin) we could not find any differences in apo-E phenotype distribution. Ghiselli etal. (1982a) andXwutt et al. (1988) described that in non-diabetic patients with type V hyperlipoproteinemia the e4 allele is more frequent, whereas Stuyt et al. (1982) could not find this association. 8 of our diabetic patients had type V hyperlipoproteinemia. The apo-E phenotype frequencies in this small group did not show any gross abnormalities compared to other diabetic patients, but this form of hyperlipoproteinemia in diabetic patients may not be comparable to type V hyperlipoproteinemia of other origin. These patients had a higher HbAi c value (7.92% vs 7.34% in normolipidemic diabetic patients) and chylomicronemia may develop because of very poor glucose control {Greenfield, Koltermann, Olejsky andReaven 1980). The increased triglyceride levels associated with apo-E2 in diabetic and non-diabetic patients can be explained by the abnormal receptor binding activity {Weisgraber, Innerarity and Mahley 1982). In our study we found in the E3/2 subgroup of hypertriglyceridemic diabetic patients the VLDL to be richer in apolipoprotein B and in cholesterol. This could be the effect of the abnormal receptor binding activity of apo-E2 leading to an increase in IDL levels,
Downloaded by: Universite Laval. Copyrighted material.
592
Apolipoprotein EPhenotype in Diabetic Patients
Horm. metab. Res. 22 (1990)
593
Table 6 Apolipoprotein E phenotype in diabetic and non-diabetic subjects with different forms of hyperlipoproteinemia. normolipidemic
E3/2 E3/3 E4/3 other
1
diabetic
non-diabetic
5.6% 77.8% 13.9% 2.7%
11.7%—18.9% 55.4%-62.2% 19.9%-24.4% 4.1%-6.2%
hypercholesterolemic
(lla)
diabetic
non-diabetic
1
4.9% 58.8% 34.2% 2.4%
2.5%-8.0% 50.0%-65.6% 18.0%-37.5% 6.9%-14%
mixed lipidemic (lib)
hypertriglyceridemic (IV) 1
diabetic
non-diabetic1
diabetic
non-diabetic
9.4% 40.6% 50.0% /
14.2%-27.0% 40.8%-54.1% 13.5%-23.2% 5.4%-18.4%
50.0% 16.7% 29.2% 4.1%
14.8%-28.9% 46.6%-56.5% 17.8%-24.4% 6.4%-9%
According to Utermann et al. (1984); Assmann et al. (1984), Ghiselli e\ al. (1982b) and the own control group.
The receptor binding activity of apo-E4 has been reported to be normal (Weisgraber, Innerarity and Mahley 1982). As apo-E4 is associated with increased dietary fat clearance (Weintraub, Eisenberg and Breslow 1987) it would not be surprising if VLDL clearance is also enhanced by this isoform, thus leading to increased LDL-cholesterol levels (Kesdniemi, Ehnholm and Mietinen 1987). This mechanism could be especially important when VLDL production is increased because of poor glucose control (Taskinen, Beltz, Harper, Fields, Schonfeld, Grundy and Howard 1986). On the other hand it is well known that diabetes mellitus is associated with a defect in VLDL clearance (Taskinen et al. 1986) which should cancel out the clearance enhancing effect of apo-E4. Another possible mechanism could be that the apo-E gene locus is in linkage disequilibrium with the LDL receptor gene locus as discussed by Eto et al. (1987). Comparing our apo-E phenotype frequency distribution to the one described in non-diabetic subjects {Utermann et al. 1984; Assmann et al. 1984; Ghiselli et al. 1982b) (Table 6) the mean difference is a stronger association between apo-E2 and hypertriglyceridemia in diabetic than in non-diabetic patients, while the association between apo-E4 and hypercholesterolemia was similar in both groups. In hypertriglyceridemic non-diabetic subjects phenotype E3/2 frequency ranges between 14.8% and 28.9% (Utermann et al. 1984; Assmann et al. 1984; Ghiselli et al. 1982b), while it was 50% in the hypertriglyceridemic diabetic patients studied here. Phenotype E3/3 frequency is reduced in all hypertriglyceridemic patients (Utermann et al. 1984; Assmann et al. 1984; Ghiselli et al. 1982), but especially in those also having diabetes. The mechanism leading to the stronger association between apo-E2 and hypertriglyceridemia in diabetic patients may be non-enzymatic posttranslational glucosylation, which has been described in diabetic patients (Curtiss and Witztum 1985). The substitution of an arginine by a cysteine residue in apo-E2 (Weisgraber, Innerarity andMahley 1982) may result in enhanced glucosylation, thereby diminishing the already abnormal receptor binding activity of apo-E2 and resulting in delayed VLDL clearance. Still another possibility is that the apo-E receptor is glucosylated, which may inhibit binding of apo-E2 but may not interfere with the binding of apo-E3 or apo-E4. It is also possible that the increase in hepatic VLDLtriglyceride synthesis and secretion which can be observed in diabetic patients (Davignon, Gregg and Sing 1988) only leads to
manifest hyperlipoproteinamia, when also the catabolic rate is decreased, i. e. secondary to an abnormal receptor binding activity. The lack of a stronger association between apoE4 and hypercholesterolemia in diabetic than in non-diabetic patients supports the theory concerning the pathogenesis of apo-E4 associated hyperlipoproteinemia. If the apo-E4 associated hyperlipoproteinemia is caused by an increased clearance of IDL and chylomicron-remnants (Weintraub, Eisenberg and Breslow 1987) glucosylation (as a consequence of hyperglycemia) should not lead to a further increase of the clearance rate. Another interesting finding of our study is the increased frequency of phenotype E4/3 in diabetic patients with hyperlipoproteinemia type lib compared to non-diabetic patients suffering from the same form of hyperlipoproteinemia. The pathogenetic background of this finding remains unclear. It is known that there is a negative correlation between HDL-cholesterol and triglycerides (Assmann 1982). We could confirm this finding also for diabetic patients. Furthermore we found a significant inverse correlation for HDLcholesterol and VLDL-triglycerides as well as VLDLcholesterol and VLDL-apoB. Until now dyslipoproteinemia in type II diabetes mellitus was considered to be mainly a secondary phenomenon (Reaven 1987; Howard 1987; Greenfield et al. 1980; Haffner, Foster, Kushwaha and Hazzard 1984; Kostner and Karddi 1988). In the light of the findings presented an important part must however be conceded to genetic factors. In summary we found a stronger association between apo-E2 and hypertriglyceridemia in diabetic patients than it was described in non-diabetic patients. The association between apo-E4 and hypercholesterolemia was similar to the one found in non-diabetic patients. We therefore conclude that type II diabetes mellitus is a possible cofactor in the apoE2 associated hyperlipoproteinemia. Further studies must show by which mechanism this increased association is mediated. Acknowledgement The authors would like to acknowledge the expert technical assistance of Mrs. Margit Judenberg.
Downloaded by: Universite Laval. Copyrighted material.
which are known to be rich in cholesterol and apo-B (Sata, Havel and Jones 1972).
Horm. metab. Res. 22 (1990) References Assmann, G., G. Schmitz, H. J. Menzel, H. Schulte: Apolipoprotein E polymorphism and hyperlipidemia. Clin. Chem. 30: 641 — 643 (1984) Assmann, G.: Lipidstoffwechsel und Atherosklerose. Schattauer-Verlag, Stuttgart-New York (1982) Blum, C. B., C. Awn, R. Sciacca: Radioimmunoassay studies of human apolipoprotein. Europ. Clin. Invest. 66:1240-1250 (1980) Curtiss, L. K., J. L. Witztum: Plasma apolipoproteins AI, All, B, CI and E are glucosylated in hyperglycemic diabetic subjects. Diabetes 34:452-461 (1985) Davignon, J., R. E. Gregg, C. F. Sing: Apolipoprotein E polymorphism and atherosclerosis. Arteriosclerosis 8:1—21 (1988) Eto, M., K. Watanabe, K. Ishii: Reciprocal effects of apolipoprotein E alleles e2 and e4 on plasma lipid levels in normolipidemic subjects. Clin. Genet. 29:477-484 (1986) Eto, M., K. Watanabe, Y. Iwashima, A. Morikawa, E. Oshima, M. Sekiguchi, K. Ishii: Apolipoprotein E polymorphism in type II diabetics. Diabetes 35:1374-1382(1986) Eto, M., K. Watanabe, Y. Iwashima, A. Morikawa, N. Chonan, E. Oshima, M. Sekiguchi, K. Ishii: Increased frequency of apolipoprotein e4 allele in type II diabetes with hypercholesterolemia. Diabetes 36:1301-1306(1987) Falco, J. M., J. M. Parr, R. N. Simpson, V. Wynn: Lipoprotein analyses in varying degrees of glucose tolerance. Amer. J. Med. 83:641—647 (1987) Ghiselli, G., E. J. Schaefer, L. A. Zech, R. E. Gregg, H. B. J. Brewer: Increased prevalence of apolipoprotein E4 in type V hyperlipoproteinemia. J. Clin. Invest. 70:474-477 (1982a) Ghiselli, G., R. E. Gregg, L. A. Zech, E. J. Schaefer, H. B. J. Brewer: Phenotype study of apolipoprotein E isoforms in hyperlipoproteinemic patients. Lancet II: 405-407 (1982b) Greenfield, M., O. Koltermann, J. Olefsky, G. M. Reaven: Mechanism of hypertriglyceridemia in diabetic patients with fasting hyperglycemia. Diabetologia 18:441 - 4 4 6 (1980) Haffner, S. M., D. M. Foster, R. S. Kushwaha, W. R. Hazzard: Retarded chylomicron apolipoprotein B catabolism in type II non-insulin-dependent diabetic subjects with lipaemia. Diabetologia 26: 349-354(1984) Howard, B. V.: Lipoprotein metabolism in diabetes mellitus. J. Lipid Res. 28:613-628(1987) James, W., C. Voliotis, B. Grab, D. Pometta: Phenotypes de l'apoproteine E. apoE et lipides seriques des diabetiques. Schweiz. Med. Wschr. 117:2021 - 2 0 2 3 (1987) Kesdniemi, Y. A., C. Ehnholm, T. A. Mietinen: Intestinal cholesterol absorption efficiency in man is related to apolipoprotein E phenotype. J. Clin. Invest. 80: 578 (1987) Kostner, G. M., I. Karddi: Lipoprotein alterations in diabetes mellitus. Diabetologia 31:717-722(1988) Kuusi, T., M. R. Taskinen, T. Solakivi, R. Kauppinen-Maekelin: Role of apolipoprotein E and C in type V hyperlipoproteinemia. J. Lipid. Res. 29:293-298 (1988)
K. G. Parhofer,
W. O. Richter and P.
Schwandt
Reaven, G. M.: Abnormal lipoprotein metabolism in non-insulin-dependent diabetes mellitus. Amer. J. Med. 83 (Suppl. 3A): 3 1 - 4 0 (1987) Robertson, F. W., A. M. Cumming: Effects of apoprotein E polymorphism on serum lipoprotein concentration. Arteriosclerosis 5: 283-292(1985) Sata, T., R. J. Havel, A. L. Jones: Characterization of subfractions of triglyceride-rich lipoproteins separated by gel chromatography from blood plasma of normolipidemic and hyperlipidemic humans. J.LipidRes. 13:757-762(1972) Schwandt, P.:Die Apolipoproteine. Klin. Wschr. 60: 637-649 (1982) Stuyt, R., A. F. H. Stalenhoef P. N. M. Demacker, A. V. Laar: Hyperlipoproteinemia type V and apolipoprotein E4. Lancet II: 934 (1982) Taskinen, M. R., W. F. Beltz, J. Harper, R. M. Fields, G. Schonfeld, S. M. Grundy, B. W. Howard: The effects of noninsulin-dependent diabetes mellitus on VLDL-triglyceride and VLDL-apoB metabolism: studies before and after sulfonylurea therapy. Diabetes 35: 1268-1277(1986) Tunbridge, W. M.: Factors contributing to death of diabetics under fifty years ofage. Lancet 1:569-572(1981) Utermann, S., I. Kindermann, H. Kaffarnik, A. Steinmetz: Apolipoprotein E phenotypes and hyperlipidemia. Hum. Genet. 65: 232— 236(1984) Utermann, G., J. Hees, A. Steinmetz: Polymorphism of apolipoprotein E and occurrence of dysbetalipoproteinemia in man. Nature 269:604-607 (1977) Vogelberg, K. H., E. Maucy: Apo E2 phenotypes in type II diabetics with and without insulin therapy. Klin. Wschr. 66:690-693 (1988) Weintraub, M. S., S. Eisenberg, J. L. Breslow: Dietary fat clearance in normal subjects is regulated by genetic variation in apolipoprotein. Europ. Clin. Invest. 80:1571 - 1 5 7 7 (1987) Weisgraber, K. H., T. L. Innerarity, R. W. Mahley: Abnormal lipoprotein receptor binding activity of the human E apoprotein due to cysteine-arginine interchange at a single site. J. Biol. Chem. 257: 2518-2521(1982) Winocour, P. H., P. N. Durrington, M. Ishola, D. C. Anderson: Lipoprotein abnormalities in insulin-dependent diabetes mellitus. Lancetl: 1176-1178(1986) Winocour, P. H., L. Tetlow, P. N. Durrington, M. Ishola, V. Hillier, D. C. Anderson: Apolipoprotein E polymorphism and lipoproteins in insulin-treated diabetes mellitus. Atherosclerosis 75: 167—173 (1989) Zannis, V. I, J. L. Breslow: Apolipoprotein E. Molec. Cell Biochem. 42:3-20(1982)
Requests for reprints should be addressed to: Priv.-Doz. Dr. W. O. Richter II. Medizinische Klinik Klinikum Grosshadern Marchioninistr. 15 D-8000 Mtinchen 70 (Germany)
Downloaded by: Universite Laval. Copyrighted material.
594
Atherosclerosis is the main cause of death in diabetes mellitus. This may at least in part be due to lipoprotein abnormalities which have been described in these patients. Apolipoprotein-E is a component of most lipoprotein fractions and plays an important role in the catabolism of VLDL. The different apolipoprotein-E phenotypes determined genetically are associated with certain hyperlipoproteinemias in a various degree in nondiabetic patients. In most cases apolipoprotein-E phenotype E2/2 is characteristic for familial dysbetalipoproteinemia. Phenotype E3/2 was found to be more frequent in hypertriglyceridemia while phenotype E4/3 was associated with hypercholesterolemia as well as with type V hyperlipoproteinemia. We studied apolipoprotein-E phenotypes and serum lipids in 141 type II diabetic patients (36 normolipidemic, 41 type IIa hyperlipidemic, 32 type IIb hyperlipidemic, 24 type IV hyperlipidemic, 8 type V hyperlipidemic). The phenotype E3/3 was more common in normolipidemic diabetic (77.8%) than in hyperlipoproteinemic diabetic patients (42.9%) or in the control
Introduction The major cause of premature death in diabetes mellitus is atherosclerosis (Tunbridge 1981). The accelerated vascular disease may in part be due to lipoprotein abnormalities which have been described in these patients (Winocour, Durrington, Ishola and Anderson 1986; Reaven 1987; Falco, Parr, Simpson and Wynn 1987). Although the most common abnormality of lipid metabolism in diabetes mellitus is the altered metabolism of VLDL (Howard 1987) there are also abnormalities in HDL and to lesser extent in LDL metabolism [Reaven 1987). Apolipoprotein-E (apo-E) is a constituent of most lipoprotein fractions {Blum, Aron and Sciacca 1980; Schwandt 1982) and is one of the major protein components of
Horm. metab. Res. 22 (1990) 589-594 © Georg Thieme Verlag Stuttgart -New York
group (57.5%). On the other hand phenotype E3/2 was more frequent in hypertriglyceridemic (50%) than in normolipidemic (5.6%) or hypercholesterolemic (hyperlipoproteinemia IIa: 4.9%, IIb: 9.4%) diabetic patients. The phenotype E4/3 was more frequent in all hyperlipoproteinemic diabetic patients, especially in those having hypercholesterolemia (34.2%) or mixed hyperlipidemia (50%). In conclusion we found a strong association between apo-E2 and hypertriglyceridemia in diabetic patients. This association was stronger than the one found in the general population. The association between apo-E4 and hypercholesterolemia in diabetic patients was similar to the one described in non-diabetic patients. We therefore conclude that type II diabetes mellitus is a possible cofactor in the apolipoprotein-E2 associated hyperlipoproteinemia. Key words Type II Diabetes Mellitus — Apolipoprotein E Phenotype — Hypertriglyceridemia — Hypercholesterolemia — Secondary Hyperlipoproteinemia
VLDL, modulating the catabolism of this triglyceride rich lipoprotein (Davignon, Gregg and Sing 1988). The three major isoforms of apo-E (E2, E3, E4) are genetically determined and controlled by three different alleles (e2, e3, e4) at the same gene locus and inherited in a co-dominant fashion, resulting in six possible phenotypes (E2/2, E3/2, E3/3, E4/3, E4/4, E2/4) (Zannis and Breslow 1982). All of them are known to be associated with hyperlipoproteinemias to a different degree. E3/3 is the most common phenotype in Germany and in other populations investigated and is more frequent in normolipidemic subjects than in any form of hyperlipidemia (Utermann, Kindermann, Kaffarnik and Steinmetz 1984). The phenotypes containing Apo-E2 (E2/2, E3/2) have been reported to be associated with higher levels of total and VLDL-triglycerides and E2/2 homozygosity is found in type III hyperlipidemia (dysbetalipoproteinemia) (Utermann et al. 1984; Assmann,
Received: 7 Febr. 1990
Accepted: 3 May 1990
Downloaded by: Universite Laval. Copyrighted material.
Summary
Horm. metab. Res. 22 (1990)
K. G. Parhofer, W. O. Richter and P. Schwandt
Table 1 Clinical data of type II diabetic patients.
men/women age (a) BMI (kg/m2) diab-durat. (a) therapy: diet oral agents insulin HbAic(%)
all n = 141
normolipidemic n = 36
77/64 63.3 2 26.1+0.55 7.3 8 22 45 74 7.79 8
18/18 67.0 5 25.0 8 8.3 7 6 13 17 7.34 0
1
hyperlipidemic n = 105 59/46 62.0 9 26.5 5 7.0 9 16 32 57 7.92 2
2
lla3 n = 41
lib4 n = 32
22/19 4 25.4 8 7.2 4 9 14 18 7.82 4
20/12 61.9+1.8 26.9 9 6.1 5 4 7 21 7.85 5
IV5 n = 24 12/12 64.0 9 26.6 4 8.8 2 1 11 12 8.06 0
V6 n=8 5/3 56.5 29.5 4.9 2 0 6 8.63
4 8 9
3
1
LDL-cholesterol < 3.89 mmol/l; triglycerides < 2.28 mmol/l LDL-cholesterol > 3.89 mmol/l and/or triglycerides > 2.28 mmol/l 3 LDL-cholesterol > 3.89 mmol/l; triglycerides < 2.28 mmol/l 4 LDL-cholesterol > 3.89 mmol/l; triglycerides > 2.28 mmol/l 5 LDL-cholesterol < 3.89 mmol/l; triglycerides > 2.28 mmol/l 6triglycerides > 5.17 mmol/l; chylomicrons detectable 2
Schmitz, Menzel and Schulte 1984; Eto, Watanabe and Ishii 1986; Robertson and Cumming 1985). A higher frequency of phenotypes containing apo-E4 (E4/3, E4/4) have been described in hypercholesterolemia (Utermann et al. 1984; Assmann et al. 1984) and in type V hyperlipoproteinemia (Ghiselli, Schaefer, Zech, Gregg and Brewer 1982a; Kuusi, Taskinen, Solakivi and Kauppinen-Maekelin 1988), but the latter finding is still a matter of controversy (Stuyt, Stahlenhoef, Demacker and Laar 1982). However the presence of either apo-E2 or apoE4 does not necessarily result in manifest hyperlipoproteinemia even in the case of homozygosity. Thus it can be concluded that only in the presence of other factors unidentified until now hyperlipoproteinemia occurs. The diabetic state is possibly one such cofactor, thus explaining the high incidence of hyperlipoproteinemia in diabetes mellitus. If this is the case, it should be expected that the isoforms E2 and E4 are much more predictive for hyperlipoproteinemia in diabetic than in non-diabetic patients. We therefore studied serum lipoproteins, apoE phenotype and glucose control in 141 patients with type II diabetes mellitus.
Lipoproteins Blood samples were collected after an overnight fast. Serum was immediately separated by low speed centrifugation. Cholesterol and triglycerides were measured enzymatically. Lipid assay procedures were standardized throughout the trial against reference material supplied by a national center of quality control for clinical laboratories (INSTAND). Furthermore all lipid values were determined in duplicate differing no more than . The VLDL fraction was separated by ultracentrifugation (d = 1.006 g/ml; 50000 rpm; 20 h; 4 °C; Beckmann TI50 rotor). The content of apolipoprotein B (VLDL-apoB) was measured nephelometrically in the top fraction. In the bottom fraction HDL-cholesterol was measured after precipitation of LDL with phosphotungstic acid. LDL-cholesterol was calculated by subtracting HDL-cholesterol from total bottom-cholesterol.
Apolipoprotein E Phenotype The isolated VLDL were stored at — 20 °C. Detection of apolipoprotein E pattern was done according to Utermann, Hees and Steinmetz (1977) with certain modifications: VLDL was delipidated with ethanol/ether (3:1 v/v). After drying under a stream of nitrogen the proteins were dissolved in 30 ul 0.01 mol/1 Tris HC1, 8 mol/1 urea, 0.01 mol/1 dithiothreitol, pH 8.6 and subjected to isoelectrofocussing in a pH gradient from 5—7 in thinlayer polyacrylamide gels containing 8 mol/1 urea. Protein bands were stained with coomassie brilliant blue.
Material and Methods Patients 141 patients with type II diabetes mellitus were recruited, their characteristics are shown in Table 1. The patients were subdivided intofivegroups according to their plasma lipid levels. Normolipidemia was defined as both LDL-cholesterol below 3.89 mmol/l and triglycerides below 2.28 mmol/l. The different forms of hyperlipidemia were defined as type Ha if LDL-cholesterol was above 3.89 mmol/l and triglycerides below 2.28 mmol/l, as type IIb if LDLcholesterol was below 2.28 mmol/l and triglycerides above 2.28 mmol/l and as type V if triglycerides were above 5.7 mmol/l and chylomicrons were present. For the purpose of this study no distinction was made between primary and secondary hyperlipidemia. All patients were in steady state conditions concerning their blood glucose control. Quality of blood glucose control was registered as HbAic. 22 of the diabetic patients were treated with diet alone, 45 took sulfonyl ureas and 74 were on insulin. Furthermore, the distribution of apolipoprotein E phenotypes (control group) was evaluated in 711 unselected patients (401 men, 310 women, age 44.0 1.5 years).
Statistics All values are expressed as mean values SEM and compared by Mann-Whitney-test; correlation was calculated by linear regression coefficient. Results The lipoprotein levels for all patients and the different groups of hyperlipoproteinemias are shown in Table 2. We found a significant inverse correlation between HDL-cholesterol and triglycerides as well as VLDLtriglycerides, VLDL-cholesterol and VLDL-apoB in the whole group as well as in all subgroups (except group IV and group V) (all: r = - 0 . 2 7 1 ; P < 0.001; without hyperlipoproteinemia: r = — 0.422; P < 0.005; with hyperlipoproteinemia:
Downloaded by: Universite Laval. Copyrighted material.
590
Apolipoprotein EPhenotype in Diabetic Patients
Horm. metab. Res. 22 (1990)
591
Table 2 Lipoproteins in type II diabetic patients with and without hyperlipoproteinemia. all n = 141 cholesterol triglycerides HDL-cholest. LDL-cholest. VLDL-cholest. VLDL-triglyc. VLDL-apoB (umol/l) VLDL-chol/trig VLDL-apoB/trig
normolipidemic n = 36
6.60 2.91 1.01 4.07 1.04
6 2 3 1 9 2 0.41 0.05 0.24 1 0.15 + 0.01
4.74 1.32 1.11 3.03 0.44 0.87 0.20 0.24
4 8 9 1 4 8 3 1 1
hyperlipidemic n = 105 7.25 1 3.45 7 0.98 3 4.43 2 1.22 2 2.62 7 0.52 6 0.24 1 0.14 + 0.01
lla n = 41
1 1
1 1 1 1
7 8 6 9 3 8 4 2 2
1.44 1.11 4.66 0.47 0.89 0.26 0.25 0.15
IV n = 24
lib n = 32 1
1
7.59 3.48 0.96 5.34
0 4 5 7 0 9 3 1 3
2.33 0.58 0.21 0.14
1 1
1 1 1 1
5.65 6 4.43 3 0.83 6 3.06 5 1.58 + 0.19 3.47 3 0.65 8 0.20 1 0.14 + 0.01
V n=6 2 1 2
1 1
14.7 25.2 0.67 3.08 5.41 6.35
0 1 8 1 7 3 2 3 1 2 1
0.37
3
1 3
3
All values in mmol/l except VLDL-chol/trig and VLDL-apoB/trig; VLDL-apoB was measured only in a subgroup of 46 patients (16 without hyperlipoproteinemia, 9 with type lla, 12 with type lib and 9 with type IV hyperlipoproteinemia); Significant vs. normolipidemic diabetics P < 0.001 Significant vs. normolipidemic diabetics P < 0.005 Significant vs. normolipidemic diabetics P < 0.05
Table 3 Frequencies of apolipoprotein E phenotypes in type II diabetic patients with and without hyperlipoproteinemia.
n = 141 E2/2 E3/2 E3/3 E4/3 E4/4 E2/4
normolipidemic n = 36
hyperlipidemic n = 105
lla
lib
IV
n = 41
n = 32
n = 24
V
n = 711
/
/
/
/
/
/
/
14.9% 51.8% 31.2% 0.7% 1.4%
5.6% 77.8% 13.9%
18.1% 42.9% 37.1% 1% 1%
4.9% 58.5% 34.2% 2.5%
9.4% 40.6% 50%
50% 16.7% 29.2%
25% 50% 25%
/ /
/ 4.2%
/ /
/ 2.8%
/
Controls
2.4% 13.9% 57.5% 24.2% 1.0% 1.0%
Table 4 Apolipoprotein E alleles frequencies in type II diabetic patients with and without hyperlipoproteinemia hyperlipidemic n = 105
lla
lib
IV
V
n = 141
normolipidemic n = 36
n = 41
n = 32
n = 24
n=8
Controlgroup n = 711
0.08 0.75 0.17
0.04 0.88 0.08
0.10 0.70 0.20
0.02 0.78 0.20
0.05 0.70 0.25
0.27 0.56 0.17
0.13 0.75 0.12
0.10 0.77 0.13
all
E2 E3 E4
r = - 0 . 2 7 4 ; P < 0.002; group lla: group lib: r = - 0.304; P < 0.001).
0.342; P < 0 . 0 1 ;
The apo-E phenotype frequency distribution (Table 3) of the whole group studied was similar to the one found in our control group and in the general population (Utermann et al. 1984; Assmann et al. 1984; Ghiselli, Gregg, Zech, Schaefer and Brewer 1982b). The phenotypes E3/2 and E4/3 were slightly more, the phenotype E3/3 slightly less frequent than in the normolipidemic non-diabetic population {Utermann et al. 1984; Assmann et al. 1984; Ghiselli et al. 1982b). Normolipidemic type II diabetic patients more frequently had the phenotype E3/3 (77.8%) than hyperlipidemic diabetic patients (42.9%; P < 0.001). The frequency of the E3/3 was different in the hyperlipidemic groups being highest and almost normal in the group lla (58.5 %) and lowest in group IV (16.7%). The frequency of the phenotype E3/2 was low in normolipidemic (5.6%) and hypercholesterolemic diabetic patients (group lla: 14.9%; group lib: 9.4%), but was as high as 50 % in group IV and 25 %> in group V. The frequency of the phenotype E4/3 was low in the normolipidemic group
(13.9%) and somewhat higher in group IV (29.2%) and group V (25%). But this phenotype was as frequent as 34.2 %> and 50%o in group lla respectively group lib. Statements concerning the rare phenotypes E2/2, E2/4 and E4/4 are not possible because of their low numbers. However these phenotypes are taken into account if the apo-E allele frequencies are compared (Table 4). While the frequency of the e3 allele is lowered in all hyperlipidemic groups (56 %—75 %) compared to the normolipidemic group (88%), the frequency of the e2 allele is elevated in group IV (27 %) and group V (13 %), whereas the e4 allele frequency is elevated in all hyperlipidemic diabetic patients being highest in group lla (20 %) and group lib (25 %). Comparing the average lipoprotein levels in the most common apo-E phenotypes (E3/3, E3/2, E4/3) (Table 5) we found that diabetic patients with the phenotype E3/2 had higher triglycerides, VLDL-triglycerides, VLDLcholesterol and VLDL-apoB whereas HDL-cholesterol and LDL-cholesterol were found to be lower compared to diabetic patients with the phenotype E3/3.
Downloaded by: Universite Laval. Copyrighted material.
all
Horm. metab. Res. 22 (1990)
K. G. Parhofer, W. O. RichterandP.
Table 5 Lipoproteins in type II diabetic patients depending on the apolipoprotein E phenotype. E3/3 n = 73
E3/2 n = 21 cholesterol triglycerides HDL-cholest. LDL-cholest. VLDL-cholest. VLDL-triglyc. VLDL-apoB (u.mol/1) VLDL-chol/trig. VLDL-apoB/trig.
7.10 4.26 1.01 3.86 1.53 3.24 0.71 0.22
8 4 1 8 8 6 1 2 1 1 1 1
6.42 2.44 1.06 4.04 0.80 1.71 0.38 0.25 0.16
E4/3 n = 44 4 7 5 3 1 6 7 2 2
6.73 4 2.93 5 0.96 5 4.30 1 1.09 + 0.16 55.0 + 8.76 0.35 5 0.22 1 2
3/5
3/5 3/6
5 4
All values in mmol/l except VLDL-chol/trig and VLDL-apoB/trig; VLDL-apoB was measured only in a subgroup of 46 patients (20 with phenotype E3/3, 11 with E3/2 and 15 with E4/3). Significant vs diabetic patients with phenotype E3/3; P < 0.001; Significant vs diabetic patients with phenotype E3/3; P < 0.005; Significant vs diabetic patients with phenotype E3/3; P < 0.01; Significant vs diabetic patients with phenotype E3/3; P < 0.05; Significant vs diabetic patients with phenotype E3/2; P < 0.001; Significant vs diabetic patients with phenotype E3/2; P < 0.005.
The comparison of hypertriglyceridemic diabetic patients (group IV) with apo-E phenotype E3/2 to those with apo-E phenotype E3/3 showed some interesting differences: The apo-E2 containing subgroup had higher levels of VLDL-apoB (0.69 umol/1 vs. 0.58 u.mol/1) and a higher ratio of VLDL-cholesterol to VLDL-triglycerides (0.22 vs. 0.19), thus indicating that the VLDL of the E2/3 subgroup of hypertriglyceridemic diabetic patients are richer in apo-B and in cholesterol than the VLDL of the E3/3 subgroup presenting with the same form of hyperlipoproteinemia. We could not find any differences in lipoprotein levels or ratios when we subdivided group Ha or group lib according to their apo-E phenotype (data not shown). The form of therapy (diet, sulfonyl ureas, insulin) was not associated with certain apo-E phenotypes in normolipidemic or hyperlipidemic diabetic patients, although we found a higher proportion of phenotype apo-E4/3 in the insulin treated patients (36% vs. 23 % in the group treated with diet alone or with sulfonyl ureas). However, this may be the consequence of the high proportion of patients with hyperlipoproteinemias type Ha or lib in the insulin treated group. Discussion In type II diabetic patients (n = 141) we found a similar frequency distribution of the apo-E phenotypes as described in the non-diabetic population (Davignon, Gregg and Sing 1988; Utermann et al. 1984; Assmann et al. 1984; Ghiselli etal. 1982b). Subdividing the whole group into those without hyperlipoproteinemia and into those with different forms of hyperlipoproteinemia we found normolipidemic patients to have predominantly the apo-E phenotype E3/3, while phenotype E4/3 was more frequent in diabetic patients with hypercholesterolemia (group Ha) or mixed hyperlipidemia (group lib). In hypertriglyceridemic diabetic patients we found an increase in frequency of phenotype E3/2 and to a
Schwandt
lesser extent also of phenotype E4/3, while the frequency of phenotype E3/3 was reduced. Comparing these results to the data on apo-E phenotype frequency in diabetic patients already present {Eto, Watanabe, Iwashima, Morikawa, Oshima, Sekiguchi and Ishii 1986; Eto, Watanabe, Iwashima, Morikawa, Chonan, Oshima, Sekiguchi and Ishii 1987; James, Voliotis, Grab and Pometta 1987; Vogelberg and Maucy 1988; Winocour, Tetlow, Durrington, Ishola, Hillier and Anderson 1989) we found a very similar distribution as Eto et al. 1987), who presented the following data for Japanese normolipidemic diabetic patients: E3/2 3.7%; E3/3 79.9%; E4/3 14.9%. Also for diabetic patients with hypercholesterolemia (group Ha or lib) we found the same tendency as Eto et al. (1987), but in our study, we found the e4 allele more frequent in group lib than in group Ha, while Eto et al. (1987) described it the other way around. Here on the other hand we found the frequency of the e2 allele unchanged in group Ha or lib compared to normolipidemic diabetic patients, while hypercholesterolemic patients had a diminished frequency of the e2 allele in Eto's {Eto et al. 1987) study. Concerning hypertriglyceridemic diabetic patients there are no data directly comparable to the results presented here. However it has been shown {Eto et al. 1986) that phenotype E3/2 diabetic patients had higher levels of triglycerides, VLDL-triglycerides, VLDL-cholesterol, apolipoprotein E and VLDL-cholesterol/VLDL-triglyceride ratios than phenotypic E3/3 diabetic patients. Vogelberg and Maucy (1988) found that hyperlipoproteinemia in insulin dependent diabetic patients is associated with apo-E2 heterozygosity, while it is induced by apo-E2 homozygosity in non-insulin dependent diabetic patients. Inconsistent with this finding Winocour et al. (1989) described an unexpected increase in apo-E2 homozygosity in insulin treated diabetic patients. Dividing our patients according to the form of therapy (diet, oral agents, insulin) we could not find any differences in apo-E phenotype distribution. Ghiselli etal. (1982a) andXwutt et al. (1988) described that in non-diabetic patients with type V hyperlipoproteinemia the e4 allele is more frequent, whereas Stuyt et al. (1982) could not find this association. 8 of our diabetic patients had type V hyperlipoproteinemia. The apo-E phenotype frequencies in this small group did not show any gross abnormalities compared to other diabetic patients, but this form of hyperlipoproteinemia in diabetic patients may not be comparable to type V hyperlipoproteinemia of other origin. These patients had a higher HbAi c value (7.92% vs 7.34% in normolipidemic diabetic patients) and chylomicronemia may develop because of very poor glucose control {Greenfield, Koltermann, Olejsky andReaven 1980). The increased triglyceride levels associated with apo-E2 in diabetic and non-diabetic patients can be explained by the abnormal receptor binding activity {Weisgraber, Innerarity and Mahley 1982). In our study we found in the E3/2 subgroup of hypertriglyceridemic diabetic patients the VLDL to be richer in apolipoprotein B and in cholesterol. This could be the effect of the abnormal receptor binding activity of apo-E2 leading to an increase in IDL levels,
Downloaded by: Universite Laval. Copyrighted material.
592
Apolipoprotein EPhenotype in Diabetic Patients
Horm. metab. Res. 22 (1990)
593
Table 6 Apolipoprotein E phenotype in diabetic and non-diabetic subjects with different forms of hyperlipoproteinemia. normolipidemic
E3/2 E3/3 E4/3 other
1
diabetic
non-diabetic
5.6% 77.8% 13.9% 2.7%
11.7%—18.9% 55.4%-62.2% 19.9%-24.4% 4.1%-6.2%
hypercholesterolemic
(lla)
diabetic
non-diabetic
1
4.9% 58.8% 34.2% 2.4%
2.5%-8.0% 50.0%-65.6% 18.0%-37.5% 6.9%-14%
mixed lipidemic (lib)
hypertriglyceridemic (IV) 1
diabetic
non-diabetic1
diabetic
non-diabetic
9.4% 40.6% 50.0% /
14.2%-27.0% 40.8%-54.1% 13.5%-23.2% 5.4%-18.4%
50.0% 16.7% 29.2% 4.1%
14.8%-28.9% 46.6%-56.5% 17.8%-24.4% 6.4%-9%
According to Utermann et al. (1984); Assmann et al. (1984), Ghiselli e\ al. (1982b) and the own control group.
The receptor binding activity of apo-E4 has been reported to be normal (Weisgraber, Innerarity and Mahley 1982). As apo-E4 is associated with increased dietary fat clearance (Weintraub, Eisenberg and Breslow 1987) it would not be surprising if VLDL clearance is also enhanced by this isoform, thus leading to increased LDL-cholesterol levels (Kesdniemi, Ehnholm and Mietinen 1987). This mechanism could be especially important when VLDL production is increased because of poor glucose control (Taskinen, Beltz, Harper, Fields, Schonfeld, Grundy and Howard 1986). On the other hand it is well known that diabetes mellitus is associated with a defect in VLDL clearance (Taskinen et al. 1986) which should cancel out the clearance enhancing effect of apo-E4. Another possible mechanism could be that the apo-E gene locus is in linkage disequilibrium with the LDL receptor gene locus as discussed by Eto et al. (1987). Comparing our apo-E phenotype frequency distribution to the one described in non-diabetic subjects {Utermann et al. 1984; Assmann et al. 1984; Ghiselli et al. 1982b) (Table 6) the mean difference is a stronger association between apo-E2 and hypertriglyceridemia in diabetic than in non-diabetic patients, while the association between apo-E4 and hypercholesterolemia was similar in both groups. In hypertriglyceridemic non-diabetic subjects phenotype E3/2 frequency ranges between 14.8% and 28.9% (Utermann et al. 1984; Assmann et al. 1984; Ghiselli et al. 1982b), while it was 50% in the hypertriglyceridemic diabetic patients studied here. Phenotype E3/3 frequency is reduced in all hypertriglyceridemic patients (Utermann et al. 1984; Assmann et al. 1984; Ghiselli et al. 1982), but especially in those also having diabetes. The mechanism leading to the stronger association between apo-E2 and hypertriglyceridemia in diabetic patients may be non-enzymatic posttranslational glucosylation, which has been described in diabetic patients (Curtiss and Witztum 1985). The substitution of an arginine by a cysteine residue in apo-E2 (Weisgraber, Innerarity andMahley 1982) may result in enhanced glucosylation, thereby diminishing the already abnormal receptor binding activity of apo-E2 and resulting in delayed VLDL clearance. Still another possibility is that the apo-E receptor is glucosylated, which may inhibit binding of apo-E2 but may not interfere with the binding of apo-E3 or apo-E4. It is also possible that the increase in hepatic VLDLtriglyceride synthesis and secretion which can be observed in diabetic patients (Davignon, Gregg and Sing 1988) only leads to
manifest hyperlipoproteinamia, when also the catabolic rate is decreased, i. e. secondary to an abnormal receptor binding activity. The lack of a stronger association between apoE4 and hypercholesterolemia in diabetic than in non-diabetic patients supports the theory concerning the pathogenesis of apo-E4 associated hyperlipoproteinemia. If the apo-E4 associated hyperlipoproteinemia is caused by an increased clearance of IDL and chylomicron-remnants (Weintraub, Eisenberg and Breslow 1987) glucosylation (as a consequence of hyperglycemia) should not lead to a further increase of the clearance rate. Another interesting finding of our study is the increased frequency of phenotype E4/3 in diabetic patients with hyperlipoproteinemia type lib compared to non-diabetic patients suffering from the same form of hyperlipoproteinemia. The pathogenetic background of this finding remains unclear. It is known that there is a negative correlation between HDL-cholesterol and triglycerides (Assmann 1982). We could confirm this finding also for diabetic patients. Furthermore we found a significant inverse correlation for HDLcholesterol and VLDL-triglycerides as well as VLDLcholesterol and VLDL-apoB. Until now dyslipoproteinemia in type II diabetes mellitus was considered to be mainly a secondary phenomenon (Reaven 1987; Howard 1987; Greenfield et al. 1980; Haffner, Foster, Kushwaha and Hazzard 1984; Kostner and Karddi 1988). In the light of the findings presented an important part must however be conceded to genetic factors. In summary we found a stronger association between apo-E2 and hypertriglyceridemia in diabetic patients than it was described in non-diabetic patients. The association between apo-E4 and hypercholesterolemia was similar to the one found in non-diabetic patients. We therefore conclude that type II diabetes mellitus is a possible cofactor in the apoE2 associated hyperlipoproteinemia. Further studies must show by which mechanism this increased association is mediated. Acknowledgement The authors would like to acknowledge the expert technical assistance of Mrs. Margit Judenberg.
Downloaded by: Universite Laval. Copyrighted material.
which are known to be rich in cholesterol and apo-B (Sata, Havel and Jones 1972).
Horm. metab. Res. 22 (1990) References Assmann, G., G. Schmitz, H. J. Menzel, H. Schulte: Apolipoprotein E polymorphism and hyperlipidemia. Clin. Chem. 30: 641 — 643 (1984) Assmann, G.: Lipidstoffwechsel und Atherosklerose. Schattauer-Verlag, Stuttgart-New York (1982) Blum, C. B., C. Awn, R. Sciacca: Radioimmunoassay studies of human apolipoprotein. Europ. Clin. Invest. 66:1240-1250 (1980) Curtiss, L. K., J. L. Witztum: Plasma apolipoproteins AI, All, B, CI and E are glucosylated in hyperglycemic diabetic subjects. Diabetes 34:452-461 (1985) Davignon, J., R. E. Gregg, C. F. Sing: Apolipoprotein E polymorphism and atherosclerosis. Arteriosclerosis 8:1—21 (1988) Eto, M., K. Watanabe, K. Ishii: Reciprocal effects of apolipoprotein E alleles e2 and e4 on plasma lipid levels in normolipidemic subjects. Clin. Genet. 29:477-484 (1986) Eto, M., K. Watanabe, Y. Iwashima, A. Morikawa, E. Oshima, M. Sekiguchi, K. Ishii: Apolipoprotein E polymorphism in type II diabetics. Diabetes 35:1374-1382(1986) Eto, M., K. Watanabe, Y. Iwashima, A. Morikawa, N. Chonan, E. Oshima, M. Sekiguchi, K. Ishii: Increased frequency of apolipoprotein e4 allele in type II diabetes with hypercholesterolemia. Diabetes 36:1301-1306(1987) Falco, J. M., J. M. Parr, R. N. Simpson, V. Wynn: Lipoprotein analyses in varying degrees of glucose tolerance. Amer. J. Med. 83:641—647 (1987) Ghiselli, G., E. J. Schaefer, L. A. Zech, R. E. Gregg, H. B. J. Brewer: Increased prevalence of apolipoprotein E4 in type V hyperlipoproteinemia. J. Clin. Invest. 70:474-477 (1982a) Ghiselli, G., R. E. Gregg, L. A. Zech, E. J. Schaefer, H. B. J. Brewer: Phenotype study of apolipoprotein E isoforms in hyperlipoproteinemic patients. Lancet II: 405-407 (1982b) Greenfield, M., O. Koltermann, J. Olefsky, G. M. Reaven: Mechanism of hypertriglyceridemia in diabetic patients with fasting hyperglycemia. Diabetologia 18:441 - 4 4 6 (1980) Haffner, S. M., D. M. Foster, R. S. Kushwaha, W. R. Hazzard: Retarded chylomicron apolipoprotein B catabolism in type II non-insulin-dependent diabetic subjects with lipaemia. Diabetologia 26: 349-354(1984) Howard, B. V.: Lipoprotein metabolism in diabetes mellitus. J. Lipid Res. 28:613-628(1987) James, W., C. Voliotis, B. Grab, D. Pometta: Phenotypes de l'apoproteine E. apoE et lipides seriques des diabetiques. Schweiz. Med. Wschr. 117:2021 - 2 0 2 3 (1987) Kesdniemi, Y. A., C. Ehnholm, T. A. Mietinen: Intestinal cholesterol absorption efficiency in man is related to apolipoprotein E phenotype. J. Clin. Invest. 80: 578 (1987) Kostner, G. M., I. Karddi: Lipoprotein alterations in diabetes mellitus. Diabetologia 31:717-722(1988) Kuusi, T., M. R. Taskinen, T. Solakivi, R. Kauppinen-Maekelin: Role of apolipoprotein E and C in type V hyperlipoproteinemia. J. Lipid. Res. 29:293-298 (1988)
K. G. Parhofer,
W. O. Richter and P.
Schwandt
Reaven, G. M.: Abnormal lipoprotein metabolism in non-insulin-dependent diabetes mellitus. Amer. J. Med. 83 (Suppl. 3A): 3 1 - 4 0 (1987) Robertson, F. W., A. M. Cumming: Effects of apoprotein E polymorphism on serum lipoprotein concentration. Arteriosclerosis 5: 283-292(1985) Sata, T., R. J. Havel, A. L. Jones: Characterization of subfractions of triglyceride-rich lipoproteins separated by gel chromatography from blood plasma of normolipidemic and hyperlipidemic humans. J.LipidRes. 13:757-762(1972) Schwandt, P.:Die Apolipoproteine. Klin. Wschr. 60: 637-649 (1982) Stuyt, R., A. F. H. Stalenhoef P. N. M. Demacker, A. V. Laar: Hyperlipoproteinemia type V and apolipoprotein E4. Lancet II: 934 (1982) Taskinen, M. R., W. F. Beltz, J. Harper, R. M. Fields, G. Schonfeld, S. M. Grundy, B. W. Howard: The effects of noninsulin-dependent diabetes mellitus on VLDL-triglyceride and VLDL-apoB metabolism: studies before and after sulfonylurea therapy. Diabetes 35: 1268-1277(1986) Tunbridge, W. M.: Factors contributing to death of diabetics under fifty years ofage. Lancet 1:569-572(1981) Utermann, S., I. Kindermann, H. Kaffarnik, A. Steinmetz: Apolipoprotein E phenotypes and hyperlipidemia. Hum. Genet. 65: 232— 236(1984) Utermann, G., J. Hees, A. Steinmetz: Polymorphism of apolipoprotein E and occurrence of dysbetalipoproteinemia in man. Nature 269:604-607 (1977) Vogelberg, K. H., E. Maucy: Apo E2 phenotypes in type II diabetics with and without insulin therapy. Klin. Wschr. 66:690-693 (1988) Weintraub, M. S., S. Eisenberg, J. L. Breslow: Dietary fat clearance in normal subjects is regulated by genetic variation in apolipoprotein. Europ. Clin. Invest. 80:1571 - 1 5 7 7 (1987) Weisgraber, K. H., T. L. Innerarity, R. W. Mahley: Abnormal lipoprotein receptor binding activity of the human E apoprotein due to cysteine-arginine interchange at a single site. J. Biol. Chem. 257: 2518-2521(1982) Winocour, P. H., P. N. Durrington, M. Ishola, D. C. Anderson: Lipoprotein abnormalities in insulin-dependent diabetes mellitus. Lancetl: 1176-1178(1986) Winocour, P. H., L. Tetlow, P. N. Durrington, M. Ishola, V. Hillier, D. C. Anderson: Apolipoprotein E polymorphism and lipoproteins in insulin-treated diabetes mellitus. Atherosclerosis 75: 167—173 (1989) Zannis, V. I, J. L. Breslow: Apolipoprotein E. Molec. Cell Biochem. 42:3-20(1982)
Requests for reprints should be addressed to: Priv.-Doz. Dr. W. O. Richter II. Medizinische Klinik Klinikum Grosshadern Marchioninistr. 15 D-8000 Mtinchen 70 (Germany)
Downloaded by: Universite Laval. Copyrighted material.
594
