Atherosclerosis, 87 (1991) 39-46 0 1991 Elsevier Scientific Publishers Ireland, Ltd. 0021-9150/91/$03.50 ADONIS 002191509100081U
39
ATHERO 04603
Intermediate density lipoprotein levels are strong predictors of the extent of aortic atherosclerosis in the St. Thomas’s Hospital rabbit strain Bsrge G. Nordestgaard
‘, * and Barry Lewis 2
’Department of Chemical Pathology and Metabolic Disorders, and ’Rayne Institute, St. Thomas’s Hospital, London (U.K.) (Received 26 June, 1990) (Revised, received 1 November, 1990) (Accepted 5 November, 1990)
sunlmaly
This study assessed nonfasting cholesterol and triglyceride in plasma and in lipoproteins as predictors of the extent of aortic atherosclerosis in 2 similar groups of rabbits from the St. Thomas’s Hospital strain; the lipoprotein classes studied in the 2 groups were very low (VLDL), intermediate (IDL), low (LDL), and high density lipoprotein(HDL), and Sf > 60 lipoprotein, Sf 12-60 lipoprotein, LDL and HDL, respectively. These rabbits exhibit elevated plasma levels of VLDL, IDL, and LDL, with plasma cholesterol and triglyceride of up to 23 mmol/l and 7 mmol/l, respectively, and with up to 100% of the aortic intima bearing atherosclerosis-like lesions. In group 1 rabbits (n = 25), univariate linear regression showed that cholesterol in plasma, LDL, IDL and in VLDL each were positively associated with the extent of aortic atherosclerosis. In group 2 rabbits (n = 20), cholesterol in plasma, LDL and Sf 12-60, but not in Sf > 60 lipoprotein, was consistently positively associated with the extent of aortic atherosclerosis. Neither plasma triglyceride, triglyceride in lipoprotein fractions nor HDL cholesterol was associated consistently with the extent of atherosclerosis. Using step-up multiple linear regression among lipoprotein lipids, IDL and Sf 12-60 lipoprotein cholesterol were the most powerful independent predictors of the extent of aortic atherosclerosis in the 2 groups of rabbits. LDL cholesterol was the only other independent predictor. The results suggest that remnant lipoproteins, whether defined as IDL or Sf 12-60 lipoprotein, play an important causal role in atherosclerosis under conditions where plasma levels of these lipoproteins are elevated.
* Present address Department of vej 9, DK-2100 33 98 ext. 4116;
and correspondence: Barge G. Nordestgaard, Cardiology 2142, Rigshospitalet, BlegdamsCopenhagen, Denmark. Phone: + 45 (35) 45 Telefax: +45 (35) 37 55 40
40
Key words: Atherosclerosis; Cholesterol; Familial combined hyperlipidemia; Hypercholesterolemia; Hypertriglyceridemia; Intermediate density lipoprotein; Low density lipoprotein; Remnant lipoproteins; St. Thomas’s Hospital rabbit strain; Very low density lipoprotein
Introduction
Evidence from many sources indicates that elevated plasma levels of cholesterol and of LDL are causally associated with atherosclerosis, and epidemiological data suggest that elevated plasma levels of HDL may have a protective effect. Much less consensus exists on the atherogenic potential of elevated plasma triglyceride and of elevated levels of the triglyceride-rich lipoproteins IDL and VLDL [1,2]. However, some evidence has accumulated that remnant lipoproteins may play a role in atherogenesis, in humans without major genetic forms of hyperlipidemia [3-61 as well as in humans with type III hyperlipoproteinemia [7]. Recently, a rabbit model, the St. Thomas’s Hospital strain, has been described [8]; it has many features in common with the human disorder familial combined hyperlipidemia. Affected rabbits from this strain, fed normal rabbit chow, exhibit hypercholesterolemia, hypertriglyceridemia, or combined hyperlipidemia. Hyperlipidemic rabbits have elevated plasma levels of VLDL, IDL, and LDL due to overproduction of these lipoproteins, and such rabbits develop intimal lesions resembling human atherosclerosis [9]. We used rabbits from this strain to study the relative atherogenicity of VLDL, IDL and LDL under conditions where all 3 lipoprotein fractions are elevated in plasma, and where no other cardiovascular risk factor except male sex is present. Since plasma concentrations of IDL and VLDL after an overnight fast may not represent the average lipoprotein levels to which the arterial wall is exposed, the mean of 2 non-fasting plasma samples of the rabbits fed ad libitum were analysed. Using univariate and multiple linear regression, we examined the predictive value of cholesterol and triglyceride in VLDL, IDL, LDL, and HDL for the development of aortic atherosclerosis. In one group of rabbits, the conventional limit between VLDL and IDL (d = 1.006 g/ml or Sf = 20) was used. Since Sf 12-60 lipoprotein may
better represent the functional entity described as remnant lipoproteins, than does conventional IDL (= Sf 12-20 lipoprotein) [lo], in another, otherwise similar, group of rabbits the Sf = 60 upper cutpoint was chosen. Methods
Animals The 45 rabbits, 29 females and 16 males, were descendants of the founder male of the St. Thomas’s Hospital rabbit strain [8]; the parents of the rabbits were either 2 hyperlipidemic rabbits or one hyperlipidemic and one normolipidemic rabbit, the latter also a New Zealand White. The rabbits were divided into 2 groups; except for the difference in lipoprotein fractionation described below, these 2 groups were similar. SG 1 pellets (Grain Harvesters Ltd., Wigham, Kent, U.K.) were fed ad libitum. Experimental protocols were in accordance with UK guidelines for experiments with animals. Lipoproteins Two 4-ml non-fasting blood samples for lipoprotein fractionation were drawn at least 2 h apart between noon and 6 p.m. during the day before, or the day of killing. The blood containing the anticoagulant and antioxidant Na, EDTA (1.2 mg/ml) (Sigma, Poole, Dorset, U.K.), the antibiotics chloramphenicol (80 pg/ml) (Sigma) and gentamycin sulfate (80 pg/ml) (Sigma), and the protease inhibitors r-amino-n-caproic acid (2.6 mg/ml) (Sigma), benzamidine (10 pg/ml) (Sigma) and aprotinin (5-9 Kallikrein units/ml) (Sigma) was kept at 4” C until plasma was separated at 4” C. Lipoproteins were separated promptly at 4O C using salt solutions with the above mentioned concentrations of the 2 antibiotics and the 3 protease inhibitors and with 0.1 mg/ml of Na, EDTA. In the first group of rabbits (n = 25), 18 females and 7 males, VLDL (Sf > 20 lipoprotein: d < 1.006 g/nil), IDL (Sf 12-20 lipoprotein: 1.006
41 g/ml < d < 1.019 g/ml), LDL (Sf O-12 lipoprotein: 1.019 g/ml < d < 1.063 g/ml), and HDL (d > 1.063 g/ml) were isolated from plasma by sequential ultracentrifugation in a Kontron TFT 45.6 rotor (Watford, Hertfordshire, U.K.) for 2.4 x lo”, 2.6 x 106, and 2.9 x lo6 g,, x h in solvent densities of 1.006, 1.019, and 1.063 g/ml, respectively. In the second group of rabbits (n = 20) 11 females and 9 males, Sf > 60 lipoprotein (large VLDL) was first separated from plasma followed by sequential ultracentrifugation at solvent densities of 1.019 and 1.063 g/ml to isolate Sf 12-60 lipoprotein (small VLDL + IDL), LDL, and HDL. To isolate Sf > 60 lipoprotein, 3 ml plasma was overlayered with 2 ml of d = 1.006 g/ml salt solution in 13 x 64 mm polyallomer tubes (Kontron). After centrifugation for 1.1 X 10’ g,, X min [ll] in a Kontron TFT 45.6 rotor the top 1.5 ml was harvested to provide the Sf > 60 lipoprotein. Enzymatic methods were used to measure cholesterol (CHOD-PAP, Boehringer Mannheim, Mannheim, Germany) and triglyceride (GPO-PAP, Wako Chemicals GmbH, Neuss, Germany) in plasma and lipoprotein fractions. Average recoveries of cholesterol and triglyceride for the three consecutive ultracentrifugations together were 81% and 83%. The mean values, in the two plasma samples from each animal, of plasma lipids and of lipoprotein lipids, corrected for loss during ultracentrifugation, were used in the statistical analysis. A therosclerosis The percentage of the aortic intimal surface that was covered with atherosclerosis-like lesions, was used as a measure of atherosclerosis. The rabbits were killed by i.v. pentobarbitone sodium (50-100 mg/kg) (May & Baker Ltd., Dagenham, UK), and the aorta was excised, opened longitudinally, and rinsed thoroughly with saline. The vessel was fixed with pins on a corkboard, the area was outlined on graph paper, and the aorta was divided into the aortic arch (from the heart to the first intercostal arteries), the thoracic aorta (to the diaphragm), and the abdominal aorta (to the bifurcation). In each of these aortic regions the fraction of the total aortic intimal surface that was covered with lesions was graded visually as 0, 2, 5, 10, 20, 30,40, 50,60,70,80, 90, or 100%. Based on the areas and on the % of lesions in each of the 3
aortic sites, the average % lesion for the whole aorta was calculated. The rabbits were always graded by the same person (BGN) without knowledge of the rabbits’ lipoprotein levels. Most of the rabbits used in the present study were also used for studying lipoprotein-arterial wall interactions (to be reported elsewhere), which necessitated immediate processing of arterial tissue, and thus excluded more time-consuming measurements of aortic lesions. To validate the visual grading of aortic lesions, the arches, thoracic and abdominal aortas from another set of rabbits from the St. Thomas’s Hospital strain [12], were graded visually by BGN and compared with % lesions determined independently by planimetry after staining with Oil red 0. The linear regression of one grading method versus the other had an R2 of 94% (P < 0.001, n = 78) and the line had an intercept not significantly different from zero. As a further validation, cholesterol contents in intimas of arches, thoracic and abdominal aortas (pg/cm2) m cholesterol-fed rabbits [13,14] were compared with visual grading as described above and also performed by BGN. Linear regression of log (intimal cholesterol content) versus percentage of the total aortic surface covered with lesions had an R2 of 80% (P < 0.001, n = 83). Statistics Univariate and multiple linear regression analysis were performed with the Minitap programme [15]. For the linear regressions shown in Tables 2, 3, and 5, age was entered before plasma or lipoprotein lipids as predictors of the extent of aortic atherosclerosis. To identify the best subset of predictors of the extent of aortic lesions, step-up multiple linear regression as described by Snedecar and Co&ran [16] was performed. The statistical analysis were performed both with unadjusted and with logarithmically transformed plasma triglyceride values. Results The characteristics of the 45 rabbits are in Table 1. Plasma cholesterol ranged from mmol/l, plasma triglyceride from 0.5 mmol/l, and the fraction of the aortic
shown 1 to 23 to 6.5 intimal
42 TABLE 1 AGE, NONFASTING PLASMA AND LIPOPROTEIN LIPIDS, AND EXTENT GROUPS OF RABBITS FROM THE ST. THOMAS’S HOSPITAL STRAIN Group 1 (n = 25)
Age (months)
ATHEROSCLEROSIS
IN 2
Group 2 (n = 20)
Mean k SD
Range
16
11
rt 4
OF AORTIC
-
31
Mean f SD
Range
17
9
+ 9
-
35
Cholesterol (mmol/l) Plasma VLDL IDL Sf > 60 Sf 12-60 LDL HDL
9.4+ 6.2 2.3& 2.7 2.0+ 1.9
1.9o.oo.o-
23.1 8.5 7.0
4.1+ 1.1*
2.4 0.5
0.60.4-
8.0 2.3
1.9+ 0.9* 0.3+
1.4 0.9 0.2
0.50.2o.o-
6.5 4.0 1.0
5.9*
4.4
1.2_ _
16.6
0.7* 0.5 1.9+2.3 2.2+ 2.0 1.1* 0.5
o.o1.9 0.1-7.7 0.27.3 0.41.9
2.1+ _
1.3
0.5-
0.8+ 0.5* 0.4+ 0.3*
0.8 0.4 0.3 0.2
Triglyceride (mmol/l) Plasma VLDL IDL Sf > 60 Sf 12-60 LDL HDL Lesions (W of area) Arch Thoracic Abdominal Whole aorta
_ 0.5* 0.2*
55 33 30 38
0.3 0.2
+37 +31 +28 *30
0.2o.o-
0 0 0 0
-100 -100 -100 -100
1.3 0.8
4.5
_
25 14 13 16
+29 +24 *20 k23
O.Oo.oO.lo.o-
0 0 0 0
2.9 1.7 1.3 0.7
-100 -100 - 90 - 96
SD = standard deviation.
surface that was covered with atherosclerosis-like lesions ranged from 0 to 100%. In group 1 rabbits (Table 2) univariate linear regression adjusted for differences in age, consistently showed that cholesterol in plasma, VLDL, IDL, and in LDL were positively associated with the extent of lesions in aortic intima. There were no consistent associations between plasma or lipoprotein triglyceride, or between HDL cholesterol, and extent of aortic atherosclerosis. In group 2 rabbits (Table 3), univariate analysis allowing for age, showed consistent positive associations between extent of aortic lesions and cholesterol in plasma, Sf 12-60 lipoprotein, and in LDL, whereas less consistent associations were found for cholesterol in Sf > 60 lipoprotein and for triglyceride in Sf 12-60 lipoprotein and in LDL. Plasma triglyceride, and HDL cholesterol
and triglyceride, were not associated with extent of atherosclerosis. Table 4 shows intercorrelations among plasma and lipoprotein lipids in the 2 groups of rabbits. VLDL, IDL, and LDL cholesterol, and cholesterol in Sf > 60, Sf 12-60 lipoprotein and in LDL, were correlated in the 2 groups of rabbits, respectively. Using step-up multiple linear regression adjusted for differences in age (only including lipoprotein lipids, not plasma lipids), LDL and Sf 12-60 lipoprotein cholesterol were the most powerful independent predictors of the extent of aortic atherosclerosis in the 2 groups of rabbits (Table 5). To achieve more statistical power, all 45 rabbits were considered together: cholesterol and triglyceride in triglyceride-rich lipoproteins (d -c 1.019 g/ml = VLDL + IDL = Sf > 60 + Sf 12-60 lipoprotein), in LDL and in HDL as predictors of
43 TABLE
2
UNIVARIATE CORRELATION COEFFICIENTS OF EXTENT OF AORTIC ATHEROSCLEROSIS AS A FUNCTION OF NONFASTING LIPIDS IN GROUP 1 RABBITS (n = 25) Independent variables
Cholesterol Plasma VLDL IDL LDL HDL Triglyceride Plasma a VLDL IDL LDL HDL
Dependent
variable:
Arch
Thoracic
Abdominal
Whole aorta
0.81
0.76 0.64 0.76 0.63 NS
0.76 0.64 0.73 0.67 NS
0.82 0.69 0.79 0.71 NS
NS
NS NS NS NS NS
NS NS NS NS NS
0.67 0.75 0.73 NS
NS NS NS NS - 0.46
NS
NS NS NS
58 intima with lesions
NS = no statistically significant correlation (P > 0.05). Partial correlation coefficients adjusted for age are shown. a No significant correlation whether unadjusted or log-transformed plasma triglyceride values were used.
TABLE
3
UNIVARIATE CORRELATION COEFFICIENTS OF EXTENT OF AORTIC ATHEROSCLEROSIS AS A FUNCTION OF NONFASTING LIPIDS IN GROUP 2 RABBITS (n=20) Independent variables
Dependent
variable:
8 intima with lesions
Arch
Thoracic
Abdominal
Whole aorta
Cholesterol Plasma Sf > 60 Sf 12-60 LDL HDL
0.90 0.51 0.91 0.86 NS
0.84 NS 0.85 0.83 NS
0.75 NS 0.75 0.76 NS
0.86 NS 0.87 0.84 NS
Triglyceride Plasma ” Sf > 60 Sf 12-60 LDL HDL
NS
NS 0.60 0.66 NS
NS NS NS 0.48 NS
NS NS NS NS NS
NS NS NS 0.52 NS
NS = no statistically significant correlation (P > 0.05). Partial correlation coefficients adjusted for age are shown. a No significant correlation whether unadjusted or log-transformed plasma triglyceride values were used.
extent of atherosclerosis was examined. LDL cholesterol and cholesterol in the triglyceride-rich lipoproteins were both independent predictors. To exclude that differences in sex distribution might affect the above results, gender was entered in the various multiple linear regression analyses after age and the independent lipoprotein lipid predictors. Sex did not influence any of the independent predictors shown in Table 5. Discussion The present result that cholesterol in remnant lipoproteins, whether defined as IDL or Sf 12-60 lipoprotein, was the strongest independent predictor of extent of aortic atherosclerosis, suggests that these particles are at least as important as LDL particles in causing atherosclerosis in the St. Thomas’s Hospital rabbit strain. The groups of rabbits used in the present paper were well suited to study the relative atherogenicity of VLDL, IDL, and LDL: within the groups of rabbits large variation in concentrations was present for all 3 lipoprotein fractions as well as for the extent of atherosclerosis. The lipoprotein levels were similar to those commonly seen in hyperlipidemic humans. In humans without major genetic forms of hyperlipidemia, there is some evidence from within patient comparison and a case control study, that mildly elevated plasma levels of remnant lipoproteins, defined as IDL or Sf 12-60 lipoprotein, are a predictor of the severity of atherosclerosis [3, 41, or of the presence of atherosclerosis [5]. In these studies, plasma IDL or Sf 12-60 lipoprotein cholesterol concentrations were only lo-25% that of LDL cholesterol; this probably explains why in one of the studies [3] LDL cholesterol was a more important independent predictor than cholesterol in remnant lipoproteins in both men and women, and that in another [4] this was the case in men. In our rabbit studies, mean IDL and Sf 12-60 lipoprotein cholesterol levels were 50 and 85% of the respective LDL cholesterol levels. An additional advantage of the present studies, is that the extent of atherosclerosis was graded directly on the removed arteries, and not from arteriograms as in the human studies [3-51. Accelerated atherosclerosis in man is a feature of type III hyperlipopro-
44 TABLE
4
CORRELATION
COEFFICIENTS
AMONG
PLASMA
AND
LIPOPROTEIN
Cholesterol
Group
Triglyceride
VLDL
IDL
LDL
HDL
Plasma
VLDL
IDL
LDL
HDL
0.90
0.97 0.90
0.85 0.57 0.75
NS NS NS NS
NS 0.55 NS NS - 0.64
NS 0.56 NS NS - 0.58
0.56 0.75 0.56 NS - 0.55
0.46 0.58 NS NS - 0.70
NS NS NS NS NS
0.98
0.92
0.87 0.78 0.86
1 (n = 25)
Cholesterol Plasma VLDL IDL LDL HDL Triglyceride Plasma VLDL IDL LDL
Group
0.89
0.46 NS NS NS
Sf > 60
Sf 12-60
LDL
HDL
Plasma
Sf > 60
Sf 12-60
LDL
HDL
0.59
0.96 0.62
0.96 0.43 0.87
NS NS NS NS
NS 0.77 NS NS - 0.54
NS 0.62 NS NS -0.51
0.53 0.76 0.67 NS - 0.46
0.66 0.54 0.73 0.57 NS
NS NS NS - 0.50 NS
0.88
0.85 0.53
0.54 NS 0.77
0.47 0.54 NS NS
2 (n = 20)
Cholesterol Plasma Sf > 60 Sf 12-60 LDL HDL Triglyceride Plasma Sf > 60 Sf 12-60 LDL
NS = no statistically
TABLE
LIPIDS
significant
correlation
OF LIPOPROTEIN
LIPIDS
(P > 0.05).
5
RANKING
AS PREDICTORS
Arch
Group 1 (n = 25) IDL cholesterol Group 2 (n = 20) Sf 12-60 cholesterol LDL cholesterol Group 1 + 2 (n = 45) LDL cholesterol VLDL + IDL cholesterol
a
OF EXTENT Thoracic
OF AORTIC
ATHEROSCLEROSIS Whole aorta
Abdominal
Rank
Cum&
Rank
CumR*
Rank
Cum R2
Rank
Cum R*
1
0.56
1
0.58
1
0.54
1
0.63
1
0.82
1
0.72
1
0.75
1
0.58
1 2
0.56 0.63
1 2
0.64 0.73
1 2
0.67 0.75
2 1
0.66 0.56
Step-up multiple linear regression adjusted for age was performed; only predictors that were independently statistically (P -c 0.05) were ranked. Cum RZ = Cumulative (for number 1 predictor alone, or for number 1 + 2 predictors together) adjusted for age. a Equal to Sf > 60 + Sf 12-60 lipoprotein cholesterol.
significant partial R*
45 teinemia (in which Sf 12-60 lipoprotein levels are grossly elevated while LDL cholesterol is typically low) [7], of chronic renal failure in which IDL levels are elevated [17], and of familial combined hyperlipidemia, in which the lipoprotein abnormalities include elevation of IDL [18]. Furthermore, it has been reported that in humans, changes in IDL levels are a better predictor of progression of atherosclerosis, than are changes in LDL levels 161. That HDL cholesterol levels were not inversely related to atherosclerosis severity in the present study was of interest. The variation in HDL cholesterol was smaller than that of plasma levels of the apolipoprotein B containing lipoproteins, possibly overriding any effect of different plasma levels of HDL. Nevertheless the findings exclude the possibility that in this rabbit strain the association between levels of triglyceride-rich lipoproteins and extent of atherosclerosis is indirect and mediated by an inverse relation between levels of these particles and HDL cholesterol. Whether plasma triglyceride is a risk factor for atherosclerosis and coronary heart disease in humans remains controversial [1,2]. Severe hypertriglyceridemia, as in humans with lipoprotein lipase deficiency or apolipoprotein C-II deficiency or in alloxan-diabetic, cholesterol-fed rabbits, is not associated with accelerated atherosclerosis [19-211. The mechanism of this effect may be that the lipoprotein particles that contain most of the circulating cholesterol are too large to enter the arterial intima [22]; lipoproteins with diameters > 75 nm are practically excluded from entering the arterial intima in diabetic, cholesterol-fed rabbits [13]. Moderate hypertriglyceridemia (2-10 mmol/l) on the other hand, may be associated with increased atherogenic risk; possibly through the atherogenic potential of the triglyceride-rich IDL or Sf 12-60 lipoprotein. Both forms of hypertriglyceridemia are associated with low HDL cholesterol levels. We have recently shown in humans that Sf 12-60 lipoprotein, under conditions where concentrations in plasma of these lipoproteins are low compared with LDL concentrations, enter and leave the arterial intima at similar fractional rates as LDL [23]. Under conditions where Sf 12-60 lipoprotein is elevated in plasma, there-
fore, it is possible that and Sf 12-60 lipoprotein
similar amounts of LDL enter the arterial intima.
Acknowledgements We thank Peter Lumb for skillful technical assistance. This study was supported by the Danish Heart Foundation and the Danish Medical Research Council. References 1 Austin, M.A., Plasma triglyceride as a risk factor for coronary heart disease. The epidemiologic evidence and beyond. Am. J. Epidemiol., 129 (1989) 249. 2 Grundy, S.M. and Vega, G.L., Hypertriglyceridemia: causes and relation to coronary heart disease. Semin. Thromb. Hemost., 14 (1988) 149. 3 Tatami, R., Mabuchi, H., Ueda, K., et al., Intermediatedensity lipoprotein and cholesterol-rich very low density lipoprotein in angiographically determined coronary artery disease. Circulation, 64 (1981) 1174. 4 Reardon, M.F., Nestel, P.J., Craig, I.H. and Harper, R.W., Lipoprotein predictors of the severity of coronary artery disease in men and women. Circulation 71 (1985) 881. 5 Steiner, G., Schwartz, L., Shumak, S. and Poapst, M., The association of increased levels of intermediate-density lipoproteins with smoking and with coronary artery disease. Circulation 75 (1987) 124. 6 Krauss, R.M., Lindgren, F.T., Williams, P.T., et al., Intermediate-density lipoproteins and progression of coronary artery disease in hypercholesterolaemic men. Lancet, II (1987) 62. 7 Brown, M.S., Goldstein, J.L., Fredrickson, D.S., Familial type 3 hyperlipoproteinemia (Dysbetalipoproteinemia). In: Stanbury, J.B., Wyngaarden, J.B., Fredrickson, D.S., Goldstein, J.L. and Brown, MS. (Eds.), The Metabolic Basis of Inherited Disease, 5th edn., McGraw-Hill, New York, 1983, pp. 655-671. 8 La Ville, A., Turner, P.R., Pittilo, M., et al., Hereditary hyperlipidemia in the rabbit due to overproduction of lipoproteins. I. Biochemical studies. Arteriosclerosis, 7 (1987) 105. 9 Seddon, A.M., Woolf, N., La Ville, A., et al., Hereditary hyperlipidemia and atherosclerosis in the rabbit due to overproduction of lipoproteins. II. Preliminary report of arterial pathology. Arteriosclerosis 7 (1987) 113. 10 Reardon, M.F., Fidge, N.H. and Nestel, P.J., Catabolism of very low density lipoprotein B apoprotein in man. J. Clin. Invest., 61 (1978) 850. 11 Gustafson, A., Alaupovic, P., and Furman, R.H., Studies of the composition and structure of serum lipoproteins: isolation, purification, and characterization of very low density lipoproteins of human serum. Biochemistry, 4 (1965) 596.
12 La Ville, A.E., Seddon, A.M., Shaikh, M., Rowles, P.M., Woolf, N. and Lewis, B., Primary prevention of atherosclerosis by lovastatin in a genetically hyperlipidaemic rabbit strain. Atherosclerosis, 78 (1989) 205. 13 Nordestgaard, B.G. and Ziiversmit, D.B., Large lipoproteins are excluded from the arterial wall in diabetic cholesterol-fed rabbits. J. Lipid Res., 29 (1988) 1491. 14 Nordestgaard, B.G. and Zilversmit, D.B., Comparison of arterial intimal clearances of LDL from diabetic and nondiabetic cholesterol-fed rabbits. Differences in intimal clearance explained by size differences. Arteriosclerosis, 9 (1989) 176. 15 Ryan, B.F., Joiner, B.L. and Ryan, T.A., Jr., Minitap Handbook, 2nd edn., Duxbury Press, Boston, MA, 1985, pp. 218-259. 16 Snedecor, G.W. and Co&ran, W.G., Statistical Methods, 7th edn., Iowa State University Press, Ames, IA, 1980, pp. 334-364. 17 Nestel, P.J., Fidge, N.H. and Tan, M.H., Increased lipoprotein-remnant formation in chronic renal failure. N. Engl. J. Med., 307 (1982) 329. 18 Grundy, S.M., Chait, A. and Brunzell, J.D., Familial combined hyperlipidemia workshop. Arteriosclerosis, 7 (1987) 203.
19 Nikkill, E.A., Familial lipoprotein lipase deficiency and related disorders of chylomicron metabolism. In: Stanbury, J.B., Wyngaarden, J.B., Fredrickson, D.S., Goldstein, J.L. and Brown, M.S. (Eds.), The Metabolic Basis of Inherited Disease, 5th edn. McGraw-Hill, New York, 1983, pp. 622642. 20 Saku, K., Cedres, C., McDonald, B., et al., C-II anapolipoproteinemia and severe hypertriglyceridemia. Report of a rare case with absence of C-II apolipoprotein isoforms and review of the literature. Am. J. Med., 77 (1984) 457. 21 Duff, G.L. and Payne, T.P.B., The effect of alloxan diabetes on experimental cholesterol atherosclerosis in the rabbit. III. The mechanism of the inhibition of experimental cholesterol atherosclerosis in alloxan-diabetic rabbits. J. Exp. Med., 92 (1950) 299. 22 Nordestgaard, B.G., Stender, S. and Kjeldsen, K., Severe hypertriglyceridemia, large lipoproteins and protection against atherosclerosis. Stand. J. Clin. Lab. Invest., 47 (1987) (suppl. 186) 7. 23 Shaikh, M., Wootton, R., Nordestgaard, B.G. et al., Quantitative studies of transfer in vivo of low density, Sf 12-60 and Sf 60-400 lipoproteins between plasma and arterial intima in man. Arteriosclerosis, 11 (1991) in press.
39
ATHERO 04603
Intermediate density lipoprotein levels are strong predictors of the extent of aortic atherosclerosis in the St. Thomas’s Hospital rabbit strain Bsrge G. Nordestgaard
‘, * and Barry Lewis 2
’Department of Chemical Pathology and Metabolic Disorders, and ’Rayne Institute, St. Thomas’s Hospital, London (U.K.) (Received 26 June, 1990) (Revised, received 1 November, 1990) (Accepted 5 November, 1990)
sunlmaly
This study assessed nonfasting cholesterol and triglyceride in plasma and in lipoproteins as predictors of the extent of aortic atherosclerosis in 2 similar groups of rabbits from the St. Thomas’s Hospital strain; the lipoprotein classes studied in the 2 groups were very low (VLDL), intermediate (IDL), low (LDL), and high density lipoprotein(HDL), and Sf > 60 lipoprotein, Sf 12-60 lipoprotein, LDL and HDL, respectively. These rabbits exhibit elevated plasma levels of VLDL, IDL, and LDL, with plasma cholesterol and triglyceride of up to 23 mmol/l and 7 mmol/l, respectively, and with up to 100% of the aortic intima bearing atherosclerosis-like lesions. In group 1 rabbits (n = 25), univariate linear regression showed that cholesterol in plasma, LDL, IDL and in VLDL each were positively associated with the extent of aortic atherosclerosis. In group 2 rabbits (n = 20), cholesterol in plasma, LDL and Sf 12-60, but not in Sf > 60 lipoprotein, was consistently positively associated with the extent of aortic atherosclerosis. Neither plasma triglyceride, triglyceride in lipoprotein fractions nor HDL cholesterol was associated consistently with the extent of atherosclerosis. Using step-up multiple linear regression among lipoprotein lipids, IDL and Sf 12-60 lipoprotein cholesterol were the most powerful independent predictors of the extent of aortic atherosclerosis in the 2 groups of rabbits. LDL cholesterol was the only other independent predictor. The results suggest that remnant lipoproteins, whether defined as IDL or Sf 12-60 lipoprotein, play an important causal role in atherosclerosis under conditions where plasma levels of these lipoproteins are elevated.
* Present address Department of vej 9, DK-2100 33 98 ext. 4116;
and correspondence: Barge G. Nordestgaard, Cardiology 2142, Rigshospitalet, BlegdamsCopenhagen, Denmark. Phone: + 45 (35) 45 Telefax: +45 (35) 37 55 40
40
Key words: Atherosclerosis; Cholesterol; Familial combined hyperlipidemia; Hypercholesterolemia; Hypertriglyceridemia; Intermediate density lipoprotein; Low density lipoprotein; Remnant lipoproteins; St. Thomas’s Hospital rabbit strain; Very low density lipoprotein
Introduction
Evidence from many sources indicates that elevated plasma levels of cholesterol and of LDL are causally associated with atherosclerosis, and epidemiological data suggest that elevated plasma levels of HDL may have a protective effect. Much less consensus exists on the atherogenic potential of elevated plasma triglyceride and of elevated levels of the triglyceride-rich lipoproteins IDL and VLDL [1,2]. However, some evidence has accumulated that remnant lipoproteins may play a role in atherogenesis, in humans without major genetic forms of hyperlipidemia [3-61 as well as in humans with type III hyperlipoproteinemia [7]. Recently, a rabbit model, the St. Thomas’s Hospital strain, has been described [8]; it has many features in common with the human disorder familial combined hyperlipidemia. Affected rabbits from this strain, fed normal rabbit chow, exhibit hypercholesterolemia, hypertriglyceridemia, or combined hyperlipidemia. Hyperlipidemic rabbits have elevated plasma levels of VLDL, IDL, and LDL due to overproduction of these lipoproteins, and such rabbits develop intimal lesions resembling human atherosclerosis [9]. We used rabbits from this strain to study the relative atherogenicity of VLDL, IDL and LDL under conditions where all 3 lipoprotein fractions are elevated in plasma, and where no other cardiovascular risk factor except male sex is present. Since plasma concentrations of IDL and VLDL after an overnight fast may not represent the average lipoprotein levels to which the arterial wall is exposed, the mean of 2 non-fasting plasma samples of the rabbits fed ad libitum were analysed. Using univariate and multiple linear regression, we examined the predictive value of cholesterol and triglyceride in VLDL, IDL, LDL, and HDL for the development of aortic atherosclerosis. In one group of rabbits, the conventional limit between VLDL and IDL (d = 1.006 g/ml or Sf = 20) was used. Since Sf 12-60 lipoprotein may
better represent the functional entity described as remnant lipoproteins, than does conventional IDL (= Sf 12-20 lipoprotein) [lo], in another, otherwise similar, group of rabbits the Sf = 60 upper cutpoint was chosen. Methods
Animals The 45 rabbits, 29 females and 16 males, were descendants of the founder male of the St. Thomas’s Hospital rabbit strain [8]; the parents of the rabbits were either 2 hyperlipidemic rabbits or one hyperlipidemic and one normolipidemic rabbit, the latter also a New Zealand White. The rabbits were divided into 2 groups; except for the difference in lipoprotein fractionation described below, these 2 groups were similar. SG 1 pellets (Grain Harvesters Ltd., Wigham, Kent, U.K.) were fed ad libitum. Experimental protocols were in accordance with UK guidelines for experiments with animals. Lipoproteins Two 4-ml non-fasting blood samples for lipoprotein fractionation were drawn at least 2 h apart between noon and 6 p.m. during the day before, or the day of killing. The blood containing the anticoagulant and antioxidant Na, EDTA (1.2 mg/ml) (Sigma, Poole, Dorset, U.K.), the antibiotics chloramphenicol (80 pg/ml) (Sigma) and gentamycin sulfate (80 pg/ml) (Sigma), and the protease inhibitors r-amino-n-caproic acid (2.6 mg/ml) (Sigma), benzamidine (10 pg/ml) (Sigma) and aprotinin (5-9 Kallikrein units/ml) (Sigma) was kept at 4” C until plasma was separated at 4” C. Lipoproteins were separated promptly at 4O C using salt solutions with the above mentioned concentrations of the 2 antibiotics and the 3 protease inhibitors and with 0.1 mg/ml of Na, EDTA. In the first group of rabbits (n = 25), 18 females and 7 males, VLDL (Sf > 20 lipoprotein: d < 1.006 g/nil), IDL (Sf 12-20 lipoprotein: 1.006
41 g/ml < d < 1.019 g/ml), LDL (Sf O-12 lipoprotein: 1.019 g/ml < d < 1.063 g/ml), and HDL (d > 1.063 g/ml) were isolated from plasma by sequential ultracentrifugation in a Kontron TFT 45.6 rotor (Watford, Hertfordshire, U.K.) for 2.4 x lo”, 2.6 x 106, and 2.9 x lo6 g,, x h in solvent densities of 1.006, 1.019, and 1.063 g/ml, respectively. In the second group of rabbits (n = 20) 11 females and 9 males, Sf > 60 lipoprotein (large VLDL) was first separated from plasma followed by sequential ultracentrifugation at solvent densities of 1.019 and 1.063 g/ml to isolate Sf 12-60 lipoprotein (small VLDL + IDL), LDL, and HDL. To isolate Sf > 60 lipoprotein, 3 ml plasma was overlayered with 2 ml of d = 1.006 g/ml salt solution in 13 x 64 mm polyallomer tubes (Kontron). After centrifugation for 1.1 X 10’ g,, X min [ll] in a Kontron TFT 45.6 rotor the top 1.5 ml was harvested to provide the Sf > 60 lipoprotein. Enzymatic methods were used to measure cholesterol (CHOD-PAP, Boehringer Mannheim, Mannheim, Germany) and triglyceride (GPO-PAP, Wako Chemicals GmbH, Neuss, Germany) in plasma and lipoprotein fractions. Average recoveries of cholesterol and triglyceride for the three consecutive ultracentrifugations together were 81% and 83%. The mean values, in the two plasma samples from each animal, of plasma lipids and of lipoprotein lipids, corrected for loss during ultracentrifugation, were used in the statistical analysis. A therosclerosis The percentage of the aortic intimal surface that was covered with atherosclerosis-like lesions, was used as a measure of atherosclerosis. The rabbits were killed by i.v. pentobarbitone sodium (50-100 mg/kg) (May & Baker Ltd., Dagenham, UK), and the aorta was excised, opened longitudinally, and rinsed thoroughly with saline. The vessel was fixed with pins on a corkboard, the area was outlined on graph paper, and the aorta was divided into the aortic arch (from the heart to the first intercostal arteries), the thoracic aorta (to the diaphragm), and the abdominal aorta (to the bifurcation). In each of these aortic regions the fraction of the total aortic intimal surface that was covered with lesions was graded visually as 0, 2, 5, 10, 20, 30,40, 50,60,70,80, 90, or 100%. Based on the areas and on the % of lesions in each of the 3
aortic sites, the average % lesion for the whole aorta was calculated. The rabbits were always graded by the same person (BGN) without knowledge of the rabbits’ lipoprotein levels. Most of the rabbits used in the present study were also used for studying lipoprotein-arterial wall interactions (to be reported elsewhere), which necessitated immediate processing of arterial tissue, and thus excluded more time-consuming measurements of aortic lesions. To validate the visual grading of aortic lesions, the arches, thoracic and abdominal aortas from another set of rabbits from the St. Thomas’s Hospital strain [12], were graded visually by BGN and compared with % lesions determined independently by planimetry after staining with Oil red 0. The linear regression of one grading method versus the other had an R2 of 94% (P < 0.001, n = 78) and the line had an intercept not significantly different from zero. As a further validation, cholesterol contents in intimas of arches, thoracic and abdominal aortas (pg/cm2) m cholesterol-fed rabbits [13,14] were compared with visual grading as described above and also performed by BGN. Linear regression of log (intimal cholesterol content) versus percentage of the total aortic surface covered with lesions had an R2 of 80% (P < 0.001, n = 83). Statistics Univariate and multiple linear regression analysis were performed with the Minitap programme [15]. For the linear regressions shown in Tables 2, 3, and 5, age was entered before plasma or lipoprotein lipids as predictors of the extent of aortic atherosclerosis. To identify the best subset of predictors of the extent of aortic lesions, step-up multiple linear regression as described by Snedecar and Co&ran [16] was performed. The statistical analysis were performed both with unadjusted and with logarithmically transformed plasma triglyceride values. Results The characteristics of the 45 rabbits are in Table 1. Plasma cholesterol ranged from mmol/l, plasma triglyceride from 0.5 mmol/l, and the fraction of the aortic
shown 1 to 23 to 6.5 intimal
42 TABLE 1 AGE, NONFASTING PLASMA AND LIPOPROTEIN LIPIDS, AND EXTENT GROUPS OF RABBITS FROM THE ST. THOMAS’S HOSPITAL STRAIN Group 1 (n = 25)
Age (months)
ATHEROSCLEROSIS
IN 2
Group 2 (n = 20)
Mean k SD
Range
16
11
rt 4
OF AORTIC
-
31
Mean f SD
Range
17
9
+ 9
-
35
Cholesterol (mmol/l) Plasma VLDL IDL Sf > 60 Sf 12-60 LDL HDL
9.4+ 6.2 2.3& 2.7 2.0+ 1.9
1.9o.oo.o-
23.1 8.5 7.0
4.1+ 1.1*
2.4 0.5
0.60.4-
8.0 2.3
1.9+ 0.9* 0.3+
1.4 0.9 0.2
0.50.2o.o-
6.5 4.0 1.0
5.9*
4.4
1.2_ _
16.6
0.7* 0.5 1.9+2.3 2.2+ 2.0 1.1* 0.5
o.o1.9 0.1-7.7 0.27.3 0.41.9
2.1+ _
1.3
0.5-
0.8+ 0.5* 0.4+ 0.3*
0.8 0.4 0.3 0.2
Triglyceride (mmol/l) Plasma VLDL IDL Sf > 60 Sf 12-60 LDL HDL Lesions (W of area) Arch Thoracic Abdominal Whole aorta
_ 0.5* 0.2*
55 33 30 38
0.3 0.2
+37 +31 +28 *30
0.2o.o-
0 0 0 0
-100 -100 -100 -100
1.3 0.8
4.5
_
25 14 13 16
+29 +24 *20 k23
O.Oo.oO.lo.o-
0 0 0 0
2.9 1.7 1.3 0.7
-100 -100 - 90 - 96
SD = standard deviation.
surface that was covered with atherosclerosis-like lesions ranged from 0 to 100%. In group 1 rabbits (Table 2) univariate linear regression adjusted for differences in age, consistently showed that cholesterol in plasma, VLDL, IDL, and in LDL were positively associated with the extent of lesions in aortic intima. There were no consistent associations between plasma or lipoprotein triglyceride, or between HDL cholesterol, and extent of aortic atherosclerosis. In group 2 rabbits (Table 3), univariate analysis allowing for age, showed consistent positive associations between extent of aortic lesions and cholesterol in plasma, Sf 12-60 lipoprotein, and in LDL, whereas less consistent associations were found for cholesterol in Sf > 60 lipoprotein and for triglyceride in Sf 12-60 lipoprotein and in LDL. Plasma triglyceride, and HDL cholesterol
and triglyceride, were not associated with extent of atherosclerosis. Table 4 shows intercorrelations among plasma and lipoprotein lipids in the 2 groups of rabbits. VLDL, IDL, and LDL cholesterol, and cholesterol in Sf > 60, Sf 12-60 lipoprotein and in LDL, were correlated in the 2 groups of rabbits, respectively. Using step-up multiple linear regression adjusted for differences in age (only including lipoprotein lipids, not plasma lipids), LDL and Sf 12-60 lipoprotein cholesterol were the most powerful independent predictors of the extent of aortic atherosclerosis in the 2 groups of rabbits (Table 5). To achieve more statistical power, all 45 rabbits were considered together: cholesterol and triglyceride in triglyceride-rich lipoproteins (d -c 1.019 g/ml = VLDL + IDL = Sf > 60 + Sf 12-60 lipoprotein), in LDL and in HDL as predictors of
43 TABLE
2
UNIVARIATE CORRELATION COEFFICIENTS OF EXTENT OF AORTIC ATHEROSCLEROSIS AS A FUNCTION OF NONFASTING LIPIDS IN GROUP 1 RABBITS (n = 25) Independent variables
Cholesterol Plasma VLDL IDL LDL HDL Triglyceride Plasma a VLDL IDL LDL HDL
Dependent
variable:
Arch
Thoracic
Abdominal
Whole aorta
0.81
0.76 0.64 0.76 0.63 NS
0.76 0.64 0.73 0.67 NS
0.82 0.69 0.79 0.71 NS
NS
NS NS NS NS NS
NS NS NS NS NS
0.67 0.75 0.73 NS
NS NS NS NS - 0.46
NS
NS NS NS
58 intima with lesions
NS = no statistically significant correlation (P > 0.05). Partial correlation coefficients adjusted for age are shown. a No significant correlation whether unadjusted or log-transformed plasma triglyceride values were used.
TABLE
3
UNIVARIATE CORRELATION COEFFICIENTS OF EXTENT OF AORTIC ATHEROSCLEROSIS AS A FUNCTION OF NONFASTING LIPIDS IN GROUP 2 RABBITS (n=20) Independent variables
Dependent
variable:
8 intima with lesions
Arch
Thoracic
Abdominal
Whole aorta
Cholesterol Plasma Sf > 60 Sf 12-60 LDL HDL
0.90 0.51 0.91 0.86 NS
0.84 NS 0.85 0.83 NS
0.75 NS 0.75 0.76 NS
0.86 NS 0.87 0.84 NS
Triglyceride Plasma ” Sf > 60 Sf 12-60 LDL HDL
NS
NS 0.60 0.66 NS
NS NS NS 0.48 NS
NS NS NS NS NS
NS NS NS 0.52 NS
NS = no statistically significant correlation (P > 0.05). Partial correlation coefficients adjusted for age are shown. a No significant correlation whether unadjusted or log-transformed plasma triglyceride values were used.
extent of atherosclerosis was examined. LDL cholesterol and cholesterol in the triglyceride-rich lipoproteins were both independent predictors. To exclude that differences in sex distribution might affect the above results, gender was entered in the various multiple linear regression analyses after age and the independent lipoprotein lipid predictors. Sex did not influence any of the independent predictors shown in Table 5. Discussion The present result that cholesterol in remnant lipoproteins, whether defined as IDL or Sf 12-60 lipoprotein, was the strongest independent predictor of extent of aortic atherosclerosis, suggests that these particles are at least as important as LDL particles in causing atherosclerosis in the St. Thomas’s Hospital rabbit strain. The groups of rabbits used in the present paper were well suited to study the relative atherogenicity of VLDL, IDL, and LDL: within the groups of rabbits large variation in concentrations was present for all 3 lipoprotein fractions as well as for the extent of atherosclerosis. The lipoprotein levels were similar to those commonly seen in hyperlipidemic humans. In humans without major genetic forms of hyperlipidemia, there is some evidence from within patient comparison and a case control study, that mildly elevated plasma levels of remnant lipoproteins, defined as IDL or Sf 12-60 lipoprotein, are a predictor of the severity of atherosclerosis [3, 41, or of the presence of atherosclerosis [5]. In these studies, plasma IDL or Sf 12-60 lipoprotein cholesterol concentrations were only lo-25% that of LDL cholesterol; this probably explains why in one of the studies [3] LDL cholesterol was a more important independent predictor than cholesterol in remnant lipoproteins in both men and women, and that in another [4] this was the case in men. In our rabbit studies, mean IDL and Sf 12-60 lipoprotein cholesterol levels were 50 and 85% of the respective LDL cholesterol levels. An additional advantage of the present studies, is that the extent of atherosclerosis was graded directly on the removed arteries, and not from arteriograms as in the human studies [3-51. Accelerated atherosclerosis in man is a feature of type III hyperlipopro-
44 TABLE
4
CORRELATION
COEFFICIENTS
AMONG
PLASMA
AND
LIPOPROTEIN
Cholesterol
Group
Triglyceride
VLDL
IDL
LDL
HDL
Plasma
VLDL
IDL
LDL
HDL
0.90
0.97 0.90
0.85 0.57 0.75
NS NS NS NS
NS 0.55 NS NS - 0.64
NS 0.56 NS NS - 0.58
0.56 0.75 0.56 NS - 0.55
0.46 0.58 NS NS - 0.70
NS NS NS NS NS
0.98
0.92
0.87 0.78 0.86
1 (n = 25)
Cholesterol Plasma VLDL IDL LDL HDL Triglyceride Plasma VLDL IDL LDL
Group
0.89
0.46 NS NS NS
Sf > 60
Sf 12-60
LDL
HDL
Plasma
Sf > 60
Sf 12-60
LDL
HDL
0.59
0.96 0.62
0.96 0.43 0.87
NS NS NS NS
NS 0.77 NS NS - 0.54
NS 0.62 NS NS -0.51
0.53 0.76 0.67 NS - 0.46
0.66 0.54 0.73 0.57 NS
NS NS NS - 0.50 NS
0.88
0.85 0.53
0.54 NS 0.77
0.47 0.54 NS NS
2 (n = 20)
Cholesterol Plasma Sf > 60 Sf 12-60 LDL HDL Triglyceride Plasma Sf > 60 Sf 12-60 LDL
NS = no statistically
TABLE
LIPIDS
significant
correlation
OF LIPOPROTEIN
LIPIDS
(P > 0.05).
5
RANKING
AS PREDICTORS
Arch
Group 1 (n = 25) IDL cholesterol Group 2 (n = 20) Sf 12-60 cholesterol LDL cholesterol Group 1 + 2 (n = 45) LDL cholesterol VLDL + IDL cholesterol
a
OF EXTENT Thoracic
OF AORTIC
ATHEROSCLEROSIS Whole aorta
Abdominal
Rank
Cum&
Rank
CumR*
Rank
Cum R2
Rank
Cum R*
1
0.56
1
0.58
1
0.54
1
0.63
1
0.82
1
0.72
1
0.75
1
0.58
1 2
0.56 0.63
1 2
0.64 0.73
1 2
0.67 0.75
2 1
0.66 0.56
Step-up multiple linear regression adjusted for age was performed; only predictors that were independently statistically (P -c 0.05) were ranked. Cum RZ = Cumulative (for number 1 predictor alone, or for number 1 + 2 predictors together) adjusted for age. a Equal to Sf > 60 + Sf 12-60 lipoprotein cholesterol.
significant partial R*
45 teinemia (in which Sf 12-60 lipoprotein levels are grossly elevated while LDL cholesterol is typically low) [7], of chronic renal failure in which IDL levels are elevated [17], and of familial combined hyperlipidemia, in which the lipoprotein abnormalities include elevation of IDL [18]. Furthermore, it has been reported that in humans, changes in IDL levels are a better predictor of progression of atherosclerosis, than are changes in LDL levels 161. That HDL cholesterol levels were not inversely related to atherosclerosis severity in the present study was of interest. The variation in HDL cholesterol was smaller than that of plasma levels of the apolipoprotein B containing lipoproteins, possibly overriding any effect of different plasma levels of HDL. Nevertheless the findings exclude the possibility that in this rabbit strain the association between levels of triglyceride-rich lipoproteins and extent of atherosclerosis is indirect and mediated by an inverse relation between levels of these particles and HDL cholesterol. Whether plasma triglyceride is a risk factor for atherosclerosis and coronary heart disease in humans remains controversial [1,2]. Severe hypertriglyceridemia, as in humans with lipoprotein lipase deficiency or apolipoprotein C-II deficiency or in alloxan-diabetic, cholesterol-fed rabbits, is not associated with accelerated atherosclerosis [19-211. The mechanism of this effect may be that the lipoprotein particles that contain most of the circulating cholesterol are too large to enter the arterial intima [22]; lipoproteins with diameters > 75 nm are practically excluded from entering the arterial intima in diabetic, cholesterol-fed rabbits [13]. Moderate hypertriglyceridemia (2-10 mmol/l) on the other hand, may be associated with increased atherogenic risk; possibly through the atherogenic potential of the triglyceride-rich IDL or Sf 12-60 lipoprotein. Both forms of hypertriglyceridemia are associated with low HDL cholesterol levels. We have recently shown in humans that Sf 12-60 lipoprotein, under conditions where concentrations in plasma of these lipoproteins are low compared with LDL concentrations, enter and leave the arterial intima at similar fractional rates as LDL [23]. Under conditions where Sf 12-60 lipoprotein is elevated in plasma, there-
fore, it is possible that and Sf 12-60 lipoprotein
similar amounts of LDL enter the arterial intima.
Acknowledgements We thank Peter Lumb for skillful technical assistance. This study was supported by the Danish Heart Foundation and the Danish Medical Research Council. References 1 Austin, M.A., Plasma triglyceride as a risk factor for coronary heart disease. The epidemiologic evidence and beyond. Am. J. Epidemiol., 129 (1989) 249. 2 Grundy, S.M. and Vega, G.L., Hypertriglyceridemia: causes and relation to coronary heart disease. Semin. Thromb. Hemost., 14 (1988) 149. 3 Tatami, R., Mabuchi, H., Ueda, K., et al., Intermediatedensity lipoprotein and cholesterol-rich very low density lipoprotein in angiographically determined coronary artery disease. Circulation, 64 (1981) 1174. 4 Reardon, M.F., Nestel, P.J., Craig, I.H. and Harper, R.W., Lipoprotein predictors of the severity of coronary artery disease in men and women. Circulation 71 (1985) 881. 5 Steiner, G., Schwartz, L., Shumak, S. and Poapst, M., The association of increased levels of intermediate-density lipoproteins with smoking and with coronary artery disease. Circulation 75 (1987) 124. 6 Krauss, R.M., Lindgren, F.T., Williams, P.T., et al., Intermediate-density lipoproteins and progression of coronary artery disease in hypercholesterolaemic men. Lancet, II (1987) 62. 7 Brown, M.S., Goldstein, J.L., Fredrickson, D.S., Familial type 3 hyperlipoproteinemia (Dysbetalipoproteinemia). In: Stanbury, J.B., Wyngaarden, J.B., Fredrickson, D.S., Goldstein, J.L. and Brown, MS. (Eds.), The Metabolic Basis of Inherited Disease, 5th edn., McGraw-Hill, New York, 1983, pp. 655-671. 8 La Ville, A., Turner, P.R., Pittilo, M., et al., Hereditary hyperlipidemia in the rabbit due to overproduction of lipoproteins. I. Biochemical studies. Arteriosclerosis, 7 (1987) 105. 9 Seddon, A.M., Woolf, N., La Ville, A., et al., Hereditary hyperlipidemia and atherosclerosis in the rabbit due to overproduction of lipoproteins. II. Preliminary report of arterial pathology. Arteriosclerosis 7 (1987) 113. 10 Reardon, M.F., Fidge, N.H. and Nestel, P.J., Catabolism of very low density lipoprotein B apoprotein in man. J. Clin. Invest., 61 (1978) 850. 11 Gustafson, A., Alaupovic, P., and Furman, R.H., Studies of the composition and structure of serum lipoproteins: isolation, purification, and characterization of very low density lipoproteins of human serum. Biochemistry, 4 (1965) 596.
12 La Ville, A.E., Seddon, A.M., Shaikh, M., Rowles, P.M., Woolf, N. and Lewis, B., Primary prevention of atherosclerosis by lovastatin in a genetically hyperlipidaemic rabbit strain. Atherosclerosis, 78 (1989) 205. 13 Nordestgaard, B.G. and Ziiversmit, D.B., Large lipoproteins are excluded from the arterial wall in diabetic cholesterol-fed rabbits. J. Lipid Res., 29 (1988) 1491. 14 Nordestgaard, B.G. and Zilversmit, D.B., Comparison of arterial intimal clearances of LDL from diabetic and nondiabetic cholesterol-fed rabbits. Differences in intimal clearance explained by size differences. Arteriosclerosis, 9 (1989) 176. 15 Ryan, B.F., Joiner, B.L. and Ryan, T.A., Jr., Minitap Handbook, 2nd edn., Duxbury Press, Boston, MA, 1985, pp. 218-259. 16 Snedecor, G.W. and Co&ran, W.G., Statistical Methods, 7th edn., Iowa State University Press, Ames, IA, 1980, pp. 334-364. 17 Nestel, P.J., Fidge, N.H. and Tan, M.H., Increased lipoprotein-remnant formation in chronic renal failure. N. Engl. J. Med., 307 (1982) 329. 18 Grundy, S.M., Chait, A. and Brunzell, J.D., Familial combined hyperlipidemia workshop. Arteriosclerosis, 7 (1987) 203.
19 Nikkill, E.A., Familial lipoprotein lipase deficiency and related disorders of chylomicron metabolism. In: Stanbury, J.B., Wyngaarden, J.B., Fredrickson, D.S., Goldstein, J.L. and Brown, M.S. (Eds.), The Metabolic Basis of Inherited Disease, 5th edn. McGraw-Hill, New York, 1983, pp. 622642. 20 Saku, K., Cedres, C., McDonald, B., et al., C-II anapolipoproteinemia and severe hypertriglyceridemia. Report of a rare case with absence of C-II apolipoprotein isoforms and review of the literature. Am. J. Med., 77 (1984) 457. 21 Duff, G.L. and Payne, T.P.B., The effect of alloxan diabetes on experimental cholesterol atherosclerosis in the rabbit. III. The mechanism of the inhibition of experimental cholesterol atherosclerosis in alloxan-diabetic rabbits. J. Exp. Med., 92 (1950) 299. 22 Nordestgaard, B.G., Stender, S. and Kjeldsen, K., Severe hypertriglyceridemia, large lipoproteins and protection against atherosclerosis. Stand. J. Clin. Lab. Invest., 47 (1987) (suppl. 186) 7. 23 Shaikh, M., Wootton, R., Nordestgaard, B.G. et al., Quantitative studies of transfer in vivo of low density, Sf 12-60 and Sf 60-400 lipoproteins between plasma and arterial intima in man. Arteriosclerosis, 11 (1991) in press.
