- Email: [email protected]
Relation of angiographically defined coronary artery disease and plasma concentrations of insulin, lipid, and apolipoprotein in normolipidemic subjects with varying degrees of glucose tolerance
Relation of Angiographically Defined Coronary Artery Disease and Plasma Concentrations of Insulin., lipid, and Apolipoprotein in Normolipldemic Subjects with Varying Degrees of Glucose Tolerance Ryuichi Fuiiwara, MD, Yasunori Kutsumi, MD, Takio Hayashi, MD, Hiroyuki Nishio, MD, Yusuke Koshino, MD, Yoshifumi Shimada, MD, Tsuguhiko Nakai, MD, and Susumu Miyabo, MD We investi ated the association between h perinsulinemia an 1 than es in lipid, lipoprotein, an dyapolipoprotein that woul 3 increase the risk of coronary artery disease (CAD) independent of glucose tolerance. A coronary angiogram was recorded in 127 male subjects, includin 41 with normal glucose tolerance, 41 with im air J glucose tolerance, and 45 with noninsulin- crependent diabetes mellitus (NIDDM) . Subjects were divided into 2 groups according to results: the group with CAD (n = 94) and the group with normal coronary arteries (n = 33). All subjects were normolipidemic (total cholesterol ~230 mg/dl and triglycerides cl50 mg/dl). The CAD roup had a significantly lower plasma level of high- 8 ensity lipoprotein (HDL) cholesterol and apolipoprotein A-l (apo A-l) and a higher level of apolipo rotein B (a o B) than the normal group with norma P glucose to Perance. in considering subjects with im aired glucose tolerance or NIDDM, the CAD roup ha s a significantly lower plasma level of HDL ci! olesterol and a A-l and a significantly higher plasma level of total cr olesterol, triglycerides, and apo
B than the normal grou . In each of the sub’ects with normal and impaired g Pucose tolerance, an cl NIDDM, the elevation of plasma insulin concentration during both the complete test period and the early hase of an oral glucose challenge was significantly Righer in the CAD than in the normal group. In all subiects, graded reductions in HDL cholesterol and apo A-l and graded increases in plasma total cholesterol, trig1 cerides, and apo B were observed with increasing terti res of the postglucose challenge measurements of insulinemia. Multivariate analysis of the data confirmed the independent effect of plasma levels of apo A-l and apo B on the severity of CAD. Thus, hyperinsulinemia appeared to be associated with changes in lipid and apoli oprotein that predisposed toward coronary atheroscPerosis not only in nondiabetic subiects, but also in those with impaired glucose tolerance and NIDDM. The plasma levels of both apo A-l and apo B, both nontraditional risk factors, were better predictors of CAD than were plasma levels of lipids in normolipidemic men. (Am J Cardiol 1995;75: 122-l 26)
oronary metabolic risk factors may differ in subjects with normolipidemic coronary artery disease(CAD) C compared with hyperlipidemic subjects. We recently
status. We studied lipid and apolipoprotein levels and their association with postglucosechallenge measuresof insulinemia in angiographically assessedmen with varying degreesof glucose tolerance.
showed that plasma levels of apolipoprotein A-I (apo AI) and apolipoprotein B (apo B) are powerful discriminators in normolipidemic patients with CAD and that the hyperinsulinemic response may indicate an enhanced susceptibility to CAD.l Several studies indicate that hyperinsulinemia or insulin resistanceis associatedwith unfavorable changesin plasma lipid and lipoprotein concentrations characterizedby elevated total and very-lowdensity lipoprotein (VLDL), triglycerides, and decreased high-density lipoprotein (HDL) cholesterol.2-l If insulin resistance is a general risk factor for CAD, hyperinsulinemia may be associatedwith unfavorable changesin lipid and apolipoprotein irrespective of glucose tolerance
METHODS
From The Third Department of Internal Medicine, Fukui Medical School, MatsuokaCho, Fukui, Japan. This study was supported by Grant-in-Aid for Scientific Research (C) from the Ministry of Education, Science, and Culture of Japan, and by Grant-in-Aid for Scientific Research from the Ministry of Health and Welfare of Japan. Manuscript received Mav 23, 1994; revised manuscriot received and accepted August 36, 1994. Address for reorints: Rvuichi Fuiiwara. MD. The Third Deoartment I of Internal Medicine, Fukbi Medical School, MatsuokaCho, Fukui, 910-l 1, Japan.
Three hundred forty-two consecutive male subjects who presented for coronary angiographic assessment were studied. To determine if the relation between abnormalities of lipid metabolism and CAD also exists in subjects whose plasma lipid levels are within the normal range, 98 subjects with hyperlipidemia were excluded. Sixty-three subjects with overt diabetes mellitus receiving treatment with oral antidiabetic agents or insulin were excluded becausewe wanted participants who had a degree of hyperinsulinemia directly related to the degreeof insulin resistance.We determined the presence and severity of coronary arterial stenosis in the 15 segments of the artery designated by the American Heart Association.5 CAD was considered to be present if 21 lesions narrowed the lumen of any of the 15 arterial segments by 275%. Patients with such lesions were consideredto have CAD, whereasthose without narrowings or with ~25% narrowing of the lumen were considered to have normal coronary arteries. To ensure characteristics of metabolic abnormalities in CAD in this study, 54
122
IAN.
THE AMERICAN
IOURNAL
OF CARDIOLOGY@
VOL. 75
15.
1995
subjects who had moderate coronary arterial stenosis cumulative consumption of tobacco.Plasmalipoproteins (maximum between 25% and 75%) on angiography were and apolipoproteins were determined using plasma samexcluded. Thus, results are based on 127 Japanesemen ples drawn after a 14-hour overnight fast. Plasma cho(plasma cholesterol ~230 mg/dl and triglycerides ~150 lesterol and triglyceride concentrations were measured mg/dl) aged 40 to 78 years. The severity of atheroscle- enzymatically. For HDL cholesterol determination, rotic changeswas also assessedby grading the stenotic VLDL was fist removed by ultracentrifugation at a denlesions as follows: 0 = luminal reduction ~25%; 1 = sity of 1.006 g/ml. Cholesterol concentration was mealuminal reduction of 25% to 50%; 2 = luminal reduc- sured enzymatically in the supernatant after precipitation of 50% to 75%; 3 = luminal reduction of 75% to tion of low-density lipoprotein with sodium heparin and 90%; 4 = luminal reduction of 90% to 95%; 5 = lumi- manganese chloride* using the VLDL-free fraction. nal reduction of 95% to 99%; and 6 = total occlusion. Plasma apo A-I and apo B levels were determined by The points used to grade each branch of the coronary single radial immunodiffusion.9 Data are presentedas mean It: SEM. Data were stored artery were added and a coronary score was obtained. The total coronary score was used to reflect the severi- in a Statistical Analysis System(SAS, Gary, North Carty of CAD. Patients were divided into 2 groups accord- olina) database.Distributions of categorical data were ing to the results of coronary angiography: the group compared using the chi-square test with Yates’ correcwith normal coronary arteries (n = 33) and the group tion. Group differences for continuous variables were with CAD (n = 94). The diagnosis of diabetes mellitus evaluated by 2-tailed t tests. Correlations were calculatand impaired glucose tolerance was contied by a 2- ed with Pearson’s correlation coefficients. Differences hour oral glucose tolerance test (75 g of glucose) accord- across tertiles of different insulinemic measurements ing to the World Health Organization Expert Commit- were examinedby l-way analysis of variance using Duntee on Diabetes Mellitus.6 Only diabetic patients with can’s procedure to estimate which tertile differed signon-insulin-dependent diabetes mellitus (NIDDM) and nificantly from others. Stepwise multiple regression those treated only with diet were included in the study. analysis was performed to analyze the independent relaVenous blood sampleswere collected for determination tion of the risk factors to the coronary score. The variof blood glucose and plasma insulin levels before, and able with the highest standard partial regression coeffi30, 60, and 120 minutes after glucose ingestion. Plasma cient was entered at each step until no variable remained glucose concentrations were measured with an auto- with an F value (F to enter) of 22. A p level <0.05 was mated analyzer. Plasma immunoreactive insulin was considered statistically significant. determined by a double antibody radioimmunoassay.7The glucose and insulin TABLE I Clinical and Biochemical Features of 127 Subjects Studied response to the glucose load was calcuUnivoriote Comparison of lated as the sum of glucose values and Plasma Concentrations the sum of plasma insulin levels during the test, respectively. The early responsCoronary Arteries es of glucose and insulin to a glucose Normal Narrowed load were calculated as the differences (n = 941 In = 33) p Value between 0 and 30 minutes of glucose valAge (years) 59.8 i 1.7 58.9 i 1 .O 0.648 ues and plasma insulin levels, respecBody mass index (kg/m*] 23.1 t 0.4 23.0 i 0.3 0.858 tively. The ratios of the insulin releaseto Cumulative tobacco 551 i86 885 i 47 0.0006 the glucose values for the corresponding consumption (cigareiieyeors] Hypertension [no. of pts.) 14 40 0.925 periods of time were also calculated. No Total cholesterol [mg/dl] 167+5 186t3 0.0006 patient had taken P-adrenergic blockers, Triglycerides (mg/dl) lOOk 118 t 3 0.0007 lipid-lowering drugs, or other drugs that HDL cholesterol [mg/dl) 47 + 2 36+ 1 0.0001 may affect the metabolism of plasma Apolipoproieins A-l (mg/dl) 114i4 88 + 2 0.0001 lipids, glucose tolerance, and insulin reB lmg/4 89 i 3 117i2 0.0001 sponseto glucose ingestion. Height and Basal plasma glucose (mg/dl) 96 * 2 971 1 0.629 weight without outer clothing and shoes Basal insulin @J/ml] 9.4 f 0.6 9.5 i 0.4 0.896 were determined with a balance beam Sum of glucose values [mg/dl) 628 + 17 632 t 11 0.850 scale. Body mass index was calculated Sum of plasma insulin 152 * 13 229 * 10 0.0001 values [~U/ml) as weight (kg) divided by height squared (m2). With the patient in the sitting posiSum of plasma ~nsulln values 249 f 25 363 + 15 x103 0.0002 tion after a 5-minute rest, blood pressure Sum of plasma glucose values was measured using a mercury sphygDifference between 0 and 30 min 81 i4 80 + 3 0.859 plasma glucose values (mg/dl) momanometer. In the normal group, 3 Difference between 0 and 30 min 24.4 + 4.2 50.9 + 3.6 0.0001 patients were receiving calcium antagoplasma insulin values ($J/ml) nists and 5 were receiving angiotensinDifference between 0 and 30 min converting enzyme inhibitors. In the plasma glucose values x10’ 31.5 + 5.5 67.6 i 4.6 0.0001 Difference between 0 and 30 min CAD group, 8 patients were treated with plasma insulin values calcium antagonistsand 6 with angiotensin-converting enzyme inhibitors. “CigValues ore expressed us mean + SEM. HDL = hlgh~densih/ lipoprotein cholesterol. arette-years” was used to estimate the CORONARY
ARTERY
DISEASE/CAD
RISK
FACTORS
AND
GLUCOSE
TOLERANCE
123
RESULTS Characteristics of the study population are summarizedin Table I. The cumulative lifetime consumption of tobacco was significantly higher in the CAD group. The score of the CAD group averaged 13.1 f 1.4. Total cholesterol and triglyceride levels were significantly higher in subjects with than without CAD. Plasma concentration of the HDL cholesterol in the CAD group was significantly lower than that in the normal group. Plasma apo B levels were significantly higher in the CAD than in the normal group, and plasma apo A-I levels were significantly lower in the CAD group. A hyperinsulinemic responseto an oral glucose load was observedin subjects with CAD. In subjectswith normal glucose tolerance, no significant differencesin total cholesterol and triglyceride levels were observed between the 2 groups (Table II); however, HDL cholesterol and plasma apo A-I levels were significantly lower in the CAD than in the normal group. The plasma apo B level was significantly higher in the CAD group. The elevation of plasma insulin concentrations during the test period and the early phaseof the glucose challenge were significantly higher in the CAD than in the normal group. Conventional coronary risk factors, as well as plasma lipid and apolipoprotein concentrations, and the insulinemic measurementsin subjects with impaired glucose tolerance were included in the analyses.Total cholesterol and triglyceride levels were significantly higher in subjectswith than without CAD. The plasma concentration of the HDL cholesterol in the CAD group was significantly lower than that in the normal group. The level of plasma apo A-I was significantly lower in the CAD than in the normal group, whereas the plasma apo B level was significantly higher in the CAD group. A hyperinsulinemic response to an oral glucose load was observed in subjects with CAD. Table II also shows the clinical and metabolic characteristics of the subjects with NIDDM. Total cholesterol and triglyceride levels were higher in subjects with than without CAD. There were significant differences in the levels of HDL cholesterol, plasma apo A-I, and plasma apo B between the 2 groups. The CAD group had significantly higher insulinemic measuresduring the glucose challenge. To investigate the possibility that the relation betweenhyperinsulinemia and clinical features,lipid concentrations, and apolipoprotein concentrations could be curvilinear, we calculated the clinical and metabolic data in the tertiles of the sum of 124
THE AMERICAN JOURNAL OF CARDIOLOGYa
VOL. 7.5
JAN. 15, 1995
TABLE III Characteristics Glucose Tolerance
of 1’27 Normoliptdemic
Men Grouped
in Tertiles of the Sum of Plasma Insulin levels in Response to a 7.5 g Oral
Test Tertifes of Sum of Plasma Insulin levels
Sum of plasma insulin values ($J/ml] Age (years] Body mass index [kg/m’] Cumulative tobacco consumption (cigaretteyears) Hypertension (no. of pts.) Total cholesterol [mg/dl) Triglycerides [mg/dl) HDL cholesterol (mg/dl) Apolipoprotein (mg/dl) A-l B Coronary artery disease (no. of pts.] tire number HDL = high-density
of subjects lipoprotein.
in group
Differences
Between Tertiles
Middle [n = 41)
High (n = 45)
1 “S 2
2 vs 3
1 “S 3
333 50 58.1 i 1.6 21.9 * 0.3 640 * 70
151-250 60.8 zt 1.5 23.2 in 0.4 984 + 80
251-446 58.2 I 1 .5 23.9 I 0.4 793 -e 65
0.221 0.012 0.002
0.223 0.225 0.065
0.964 0.001 0.112
12
23 194*4 126+4 34 + 1
0.111 0 292 0 112 0 037
0.659 0.006 0.007 0.024
0.039
45 f 2
19 178 +4 112+3 39 * 2
104i 3 98 + 3 26
94 + 4 106*3 30
87 z~ 3 126*4 38
0.048 0.062 0.342
0.160 0.001 0.199
0.001
172+4 104i4
01 mean
Significant
Low (n = 41)
0.001 0.001 0.001
0001
0.026
f SEM
plasma insulin values during the test period of the glucose challenge (Table III). Tertiles were based on the sum of plasma insulin levels during the glucose challenge in all subjects; therefore, the limits were similar for subjectswith normal glucose tolerance, impaired glucose tolerance, and NIDDM. Increasing tertiles of the sum of postglucose insulin levels were associatedwith an increase in body mass index, in cumulative tobacco consumption, and in the prevalence of hypertension and CAD. Increasing tertiles of the sum of postglucose insulin levels were also associatedwith increasesin total cholesterol, triglycerides, and apo B, and with reductions in HDL cholesterol and apo A-I. The relation between the sum of plasma insulin levels during the glucose challenge and abnormal lipid and apolipoprotein levels was positive and linear with respect to total cholesterol, triglycerides, and apo B levels, and negative and linear with respectto HDL cholesterol and apo A-I levels. Stepwise multiple linear regression analyses were performed to assessthe contribution of plasma lipid and apolipoprotein concentrations, body mass index, and the sum of postglucoseinsulin values to the variation in CAD score (Table IV). The analysesindicated that plasma apo A-I and apo B concentrations were more accuratepredictors of the CAD score than the corresponding levels of plasma lipids in this group of subjects.
DISCUSSION We found that the risk of CAD in normolipidemic male subjectswhose CAD had been diagnosedby coronary angiography was higher in those with endogenous hyperinsulinemia than in those with relatively less insulin secretion. In all subjects,postglucose hyperinsulinemia was associatedwith higher plasma total cholesterol, triglycerides, and apo B levels, and with lower HDL cholesterol and apo A-I levels in subjectswith normal and abnormal glucose tolerance, similar age, and obesity. Therefore, an impaired uptake of glucose mediated by insulin may be related to unfavorable changes in plasma lipid and apolipoprotein irrespective of glucosetolerance status.Analyses basedon the comparison CORONARY
TABLE IV Multiple Stepwise Linear Regression Analyses of Variables Associated with Coronary Artery Disease Score in 1 1 Study Subjects Irrespective of Glucose Tolerance Status
Risk Factors
Standard Partial Regression Coefficient
Total cholesterol Triglycerides HDL cholesterol Apo A-l Apo B Body mass index XIRI
0.014 0.093 -0.052 -0.209 0.247 0.164 0.068
R
0.578
F
3.948
Apo = cholesterol, coefficient
95% Confidence Interval for Regression Coefficient -0.089, -0.057, -0.201, -0.208, 0.028, -1.084, -0.065,
0.102 0.15 1 0.326 -0.009 0.235 2.256 0.083
0.888 0.374 0.639 0.044 0.006 0.891 0.805 0.0002
apolipoproiein; F = F stotlstic, HDL = highdensih/ ZIRI = sum of plasma
p Value
insulin
levels;
R = multiple
lipoprotein correlation
of lipid and apolipoprotein levels over tertiles of the postglucose insulin levels showed that the abnormal metabolism of lipids and apolipoproteins that carry an increased risk of CAD is associatedwith a hyperinsulinemic response independent of the glucose tolerance status in subjects with angiographically defined CAD. Our data suggest that hyperinsulinemia has a greater impact on the postprandial metabolism of lipids and apolipoproteins.‘Insulin can influence the activity of lipid transfer protein, which is especially pronounced after meals and is affected by HDL composition.1o It is important to determine how hyperinsulinemia exerts a variable influence on the metabolism of lipids and apolipoproteins. The iirst possibility, that hyperinsulinemia is the primary etiologic mechanism responsible for abnormal metabolism of lipids and apolipoproteins, is particularly supported by kinetic studies of triglyceride metabolism. Several studies”J2 have demonstratedthat hyperinsulinemia can induce an overproduction of triglyceride-rich VLDL in the liver by increasing the availability of free fatty acids, which are
ARTERY DISEASE/CAD
RISK FACTORS
AND
GLUCOSE
TOLERANCE
125
the most important, if not the only, precursorsof de novo triglyceride synthesis. The hepatic synthesis of triglycerides can result from hyperinsulinemia.13In addition to increased synthesis, the rate of VLDL removal by peripheral tissues Tom the circulation can be reduced in subjects with hyperinsulinemia, and both an increased synthesis and a decreasedclearance rate may coexist.14 A factor that may contribute to the association between hypertriglyceridemia and hyperinsulinemia is the resistance of adipose tissue lipolysis to inhibition by insulin. This factor and the resistance of glucose elimination to stimulation by insulin may bc linked.15The association between higher levels of apo B and hyperinsulincmia also may be explained by an increasedhepatic secretion of apo B in response to insulin resistance.t6,t7Hepatic insulin resistance may therefore alter both the rates of lipoprotein synthesisand catabolism in the liver. We previously reported that the increased rate of triglyceride secretion induced by hyperinsulincmia contributes to the hypertriglyceridemia observed in obeserats treated with monosodium glutamate.Ix The lipids and protein concentrations of VLDL and plasma free fatty acid concentration were significantly higher in the monosodium glutamate-treated rats than in the controls. HDL is relatcd to VLDL metabolism in that surface components from remnantsof triglyceride-rich lipoproteins are transferred to HDL during metabolism of VLDL. It is possible that the relation between HDL cholesterol and the insulinemic measurementswas due to a reduction in adipose tissue or in muscle lipoprotein triglyceride clearance, and/or an increased activity of hepatic lipase. Hyperinsulinemia also correlates with an increase in the fractional catabolic rate of iodine-125 apo A-I/HDL and a decreasein plasma HDL cholesterol concentration.t9 One must consider that abnormalities in lipid and apolipoprotein concentrations impair the action of insulin. High concentrations of VLDL lead to an impairment of insulin action.20 With respect to long-term effects of elevated VLDL concentration on glucose metabolism, it is interesting that a high plasma VLDL level was found to be a risk factor for glucose intolerance in the Framingham Study.21Obesity is associated with both higher insulin levels and higher level of lipids. In our study subjects, obesity did not influence lipid metabolism but did influence insulin secretion (data not shown).
126
IHE AMERICA\
JOURNAL OF CARIOLOGY=
VCI
75
1. Fujiwara R, Kutsurm Y, Hayashi T. Kim S-S, Misawa ‘I‘, Tada II, Nishio Il. ‘Toyora K, Tamai T, bkai T, Miyabo S. Metabolic risk factors in the normolipidemic malt patients with angiographically defined coronary ancry disease. Jpn Gin J 199X4:493 -500. 2. Staldcr M, Pomctta D. Suenram A. Relationship behveen plasma insulin lcvclr 1981: and high density lipoprotein cholesterol levels in healthy men. Di&erolugin 2 I :.54&548. 3. Orchard TJ. Becker DJ, Bales M, Kuller LH, Dmsh AL. Plasma insulin and hpoprotein concentrations: an arhcrogcnic asxxiation? Am J Epidemiol 1983;118: 326337. 4. Modan M, llalkin H, Lusky A, Segal P, Fuchs 7, Chetrit A. Hyperinsulinemia ir characterbed by jointly disturbed plasma VLDL, LDL, and HDL levels. A population-ha.sed study. Arreriosrlrrosrs 1988;8:227--236. 5. Austen WC. Edward E, Frye RC. Gcnsini GC. Gott VL, Gnflith CCS, Goon DC, Murphy ML, Rowe BB. A reporting system on patients evaluated for coronary artery disease. Kepon of the Ad Hw Commiwe for Grading of Coronary Awry Disease. Counsel on Cardiouw~lar Surgery, American Heart Association. Ci,zularron 1975;5 I(suppl 1):SdO. 6. World Health Organization: Diabetes mellitus: repon of a WI10 study group. Geneva. World Health Organization, technical report series no. 727, 1985: l&12. 7. Iicrbe~t V, I.au K. Gottlieb CW, Bleicher SJ. Coated charcoal immunoassay of insulin. J C/in Endocnnol .Mefah 1965;25: 1375-1384. 8. Burstein Y. Scholnick HR. Marlin R. Rapid method for the isolation of lipoproteins from human semm by precipitation with polyanions. J Lipid Res 1970~11: 583-59.5. 9. Goto Y, Akanuma Y, Harano Y. Hara Y, Itakura H, Kajiyama G. Kawade MI. Koga M, Kuzuya F, Mamhama Y, Matsuzawa Y. Murai A, Murase T, Nailo C, Nakai ‘I‘, Noma A, Saitoh Y. Sasaai J. Takeuchi N. Yama/aki S, Yarugi T. Yukawa S. Determination hy the SRID method of normal values of setum apolipoproteins (A-l. A-II. B, C-II. C-III and E) in normolipidemic healthy Japanese subjects. J C/in Biochem NW 19X6: 1:7>8X. 10. Patxh JR. Karlin JB. Scott LW, Smith LC, Gotto AM. Inverse relationship between hlood level of high density lipoprotein subfraction 2 and magnitude of post-pnandial lipemia. Proc Nail Acad Sci LfSA 1983;80:1449 -1453. 11. Swiner G. Vranic M. Hyperinsulinemia and hypertriglyceridemia: a vicious cycle with atherogenic potential. In/J Obcs 19X2:6: I 17-124. 12. Tohey TA. Greenfield M, Kraemer I:, Reaven GM. Relationship between insulin resistance. insulin secretion, very low density lipoprotein kin&$ and plasma triglyccridc levels in nonnotriglyccridemic man. .Mrrabolirm 1981;30:165-171. 13. Reaven GM. Chen Y-D. Role of insulin in regulation of lipoprotein metaboliam in diabctcs. Diubheres Merub Rev 3988;4:639652. 14. Grundy SY. Pathogenesis of hypcrhiglyceridemia in man. In: Carlson I.A, Olsson AG, eds. Trcaunent of Hypcrlipoproteinemia. I\;ew York, Raven Press, 1984: 21-34. 15. Chen Y-D, Golay A, Swislocki A, Reaven GM. Keristance to insulin suppresbion of plasma free fatty and concentrations and insulin btimulation of glucose uptake in noninsulindepcndent diabetes mellitus. J C/h Endocrinnl Mrtab 1987:64:17.-21. 16. Dashti X. Wolfbauer G. Secretion of lipids. apolipoproteins, and lipoproteins hy human hepatoma cell line, HepG2: effcctr of oleic acid and insulin. J Liprd Res 1987;28:423426. 17. Dashti X Williamb DL, Alaupwic P. Effects of oleate and insulin on the pmduction and cellular mRNA concentrations of apolipoproteins in flepG2 cells. J
Lipid Rcs 1989;30:1365-1373. 18. Oida K, Sakai T. Hayashi T. bliyaho S. Takeda R. Plasma lipoproteins of monosodium gluramate-induced obese rats. Inr J Otws 19844:8:385-391. 19. Golay A, Zech 1.. Shi M-Z, Jeng C-Y, Chiou Y-AM, Reaven GM, Chen Y-DI. J Lipid Res Role of insulin in qulation of high density lipoprotein metabolism. 1987:28:1&18. 20. Bieger WI’, Michel G. Banvich D. Biehl K, Wirth A. Diminished insulin rccepton on monocytes and cryxhrocy~es in hypuuiglyccridemia. Meraholirnr 19X4:33: 9X2- 9X7. 21. Wilson PW. McGee DL. Kannel WB. Obesity, very low density lipoproteins, and glucose inrolerance over fourteen years. Am J Epidrmiol I98 I:1 14:697-704.
JAN
15, 1995
B than the normal grou . In each of the sub’ects with normal and impaired g Pucose tolerance, an cl NIDDM, the elevation of plasma insulin concentration during both the complete test period and the early hase of an oral glucose challenge was significantly Righer in the CAD than in the normal group. In all subiects, graded reductions in HDL cholesterol and apo A-l and graded increases in plasma total cholesterol, trig1 cerides, and apo B were observed with increasing terti res of the postglucose challenge measurements of insulinemia. Multivariate analysis of the data confirmed the independent effect of plasma levels of apo A-l and apo B on the severity of CAD. Thus, hyperinsulinemia appeared to be associated with changes in lipid and apoli oprotein that predisposed toward coronary atheroscPerosis not only in nondiabetic subiects, but also in those with impaired glucose tolerance and NIDDM. The plasma levels of both apo A-l and apo B, both nontraditional risk factors, were better predictors of CAD than were plasma levels of lipids in normolipidemic men. (Am J Cardiol 1995;75: 122-l 26)
oronary metabolic risk factors may differ in subjects with normolipidemic coronary artery disease(CAD) C compared with hyperlipidemic subjects. We recently
status. We studied lipid and apolipoprotein levels and their association with postglucosechallenge measuresof insulinemia in angiographically assessedmen with varying degreesof glucose tolerance.
showed that plasma levels of apolipoprotein A-I (apo AI) and apolipoprotein B (apo B) are powerful discriminators in normolipidemic patients with CAD and that the hyperinsulinemic response may indicate an enhanced susceptibility to CAD.l Several studies indicate that hyperinsulinemia or insulin resistanceis associatedwith unfavorable changesin plasma lipid and lipoprotein concentrations characterizedby elevated total and very-lowdensity lipoprotein (VLDL), triglycerides, and decreased high-density lipoprotein (HDL) cholesterol.2-l If insulin resistance is a general risk factor for CAD, hyperinsulinemia may be associatedwith unfavorable changesin lipid and apolipoprotein irrespective of glucose tolerance
METHODS
From The Third Department of Internal Medicine, Fukui Medical School, MatsuokaCho, Fukui, Japan. This study was supported by Grant-in-Aid for Scientific Research (C) from the Ministry of Education, Science, and Culture of Japan, and by Grant-in-Aid for Scientific Research from the Ministry of Health and Welfare of Japan. Manuscript received Mav 23, 1994; revised manuscriot received and accepted August 36, 1994. Address for reorints: Rvuichi Fuiiwara. MD. The Third Deoartment I of Internal Medicine, Fukbi Medical School, MatsuokaCho, Fukui, 910-l 1, Japan.
Three hundred forty-two consecutive male subjects who presented for coronary angiographic assessment were studied. To determine if the relation between abnormalities of lipid metabolism and CAD also exists in subjects whose plasma lipid levels are within the normal range, 98 subjects with hyperlipidemia were excluded. Sixty-three subjects with overt diabetes mellitus receiving treatment with oral antidiabetic agents or insulin were excluded becausewe wanted participants who had a degree of hyperinsulinemia directly related to the degreeof insulin resistance.We determined the presence and severity of coronary arterial stenosis in the 15 segments of the artery designated by the American Heart Association.5 CAD was considered to be present if 21 lesions narrowed the lumen of any of the 15 arterial segments by 275%. Patients with such lesions were consideredto have CAD, whereasthose without narrowings or with ~25% narrowing of the lumen were considered to have normal coronary arteries. To ensure characteristics of metabolic abnormalities in CAD in this study, 54
122
IAN.
THE AMERICAN
IOURNAL
OF CARDIOLOGY@
VOL. 75
15.
1995
subjects who had moderate coronary arterial stenosis cumulative consumption of tobacco.Plasmalipoproteins (maximum between 25% and 75%) on angiography were and apolipoproteins were determined using plasma samexcluded. Thus, results are based on 127 Japanesemen ples drawn after a 14-hour overnight fast. Plasma cho(plasma cholesterol ~230 mg/dl and triglycerides ~150 lesterol and triglyceride concentrations were measured mg/dl) aged 40 to 78 years. The severity of atheroscle- enzymatically. For HDL cholesterol determination, rotic changeswas also assessedby grading the stenotic VLDL was fist removed by ultracentrifugation at a denlesions as follows: 0 = luminal reduction ~25%; 1 = sity of 1.006 g/ml. Cholesterol concentration was mealuminal reduction of 25% to 50%; 2 = luminal reduc- sured enzymatically in the supernatant after precipitation of 50% to 75%; 3 = luminal reduction of 75% to tion of low-density lipoprotein with sodium heparin and 90%; 4 = luminal reduction of 90% to 95%; 5 = lumi- manganese chloride* using the VLDL-free fraction. nal reduction of 95% to 99%; and 6 = total occlusion. Plasma apo A-I and apo B levels were determined by The points used to grade each branch of the coronary single radial immunodiffusion.9 Data are presentedas mean It: SEM. Data were stored artery were added and a coronary score was obtained. The total coronary score was used to reflect the severi- in a Statistical Analysis System(SAS, Gary, North Carty of CAD. Patients were divided into 2 groups accord- olina) database.Distributions of categorical data were ing to the results of coronary angiography: the group compared using the chi-square test with Yates’ correcwith normal coronary arteries (n = 33) and the group tion. Group differences for continuous variables were with CAD (n = 94). The diagnosis of diabetes mellitus evaluated by 2-tailed t tests. Correlations were calculatand impaired glucose tolerance was contied by a 2- ed with Pearson’s correlation coefficients. Differences hour oral glucose tolerance test (75 g of glucose) accord- across tertiles of different insulinemic measurements ing to the World Health Organization Expert Commit- were examinedby l-way analysis of variance using Duntee on Diabetes Mellitus.6 Only diabetic patients with can’s procedure to estimate which tertile differed signon-insulin-dependent diabetes mellitus (NIDDM) and nificantly from others. Stepwise multiple regression those treated only with diet were included in the study. analysis was performed to analyze the independent relaVenous blood sampleswere collected for determination tion of the risk factors to the coronary score. The variof blood glucose and plasma insulin levels before, and able with the highest standard partial regression coeffi30, 60, and 120 minutes after glucose ingestion. Plasma cient was entered at each step until no variable remained glucose concentrations were measured with an auto- with an F value (F to enter) of 22. A p level <0.05 was mated analyzer. Plasma immunoreactive insulin was considered statistically significant. determined by a double antibody radioimmunoassay.7The glucose and insulin TABLE I Clinical and Biochemical Features of 127 Subjects Studied response to the glucose load was calcuUnivoriote Comparison of lated as the sum of glucose values and Plasma Concentrations the sum of plasma insulin levels during the test, respectively. The early responsCoronary Arteries es of glucose and insulin to a glucose Normal Narrowed load were calculated as the differences (n = 941 In = 33) p Value between 0 and 30 minutes of glucose valAge (years) 59.8 i 1.7 58.9 i 1 .O 0.648 ues and plasma insulin levels, respecBody mass index (kg/m*] 23.1 t 0.4 23.0 i 0.3 0.858 tively. The ratios of the insulin releaseto Cumulative tobacco 551 i86 885 i 47 0.0006 the glucose values for the corresponding consumption (cigareiieyeors] Hypertension [no. of pts.) 14 40 0.925 periods of time were also calculated. No Total cholesterol [mg/dl] 167+5 186t3 0.0006 patient had taken P-adrenergic blockers, Triglycerides (mg/dl) lOOk 118 t 3 0.0007 lipid-lowering drugs, or other drugs that HDL cholesterol [mg/dl) 47 + 2 36+ 1 0.0001 may affect the metabolism of plasma Apolipoproieins A-l (mg/dl) 114i4 88 + 2 0.0001 lipids, glucose tolerance, and insulin reB lmg/4 89 i 3 117i2 0.0001 sponseto glucose ingestion. Height and Basal plasma glucose (mg/dl) 96 * 2 971 1 0.629 weight without outer clothing and shoes Basal insulin @J/ml] 9.4 f 0.6 9.5 i 0.4 0.896 were determined with a balance beam Sum of glucose values [mg/dl) 628 + 17 632 t 11 0.850 scale. Body mass index was calculated Sum of plasma insulin 152 * 13 229 * 10 0.0001 values [~U/ml) as weight (kg) divided by height squared (m2). With the patient in the sitting posiSum of plasma ~nsulln values 249 f 25 363 + 15 x103 0.0002 tion after a 5-minute rest, blood pressure Sum of plasma glucose values was measured using a mercury sphygDifference between 0 and 30 min 81 i4 80 + 3 0.859 plasma glucose values (mg/dl) momanometer. In the normal group, 3 Difference between 0 and 30 min 24.4 + 4.2 50.9 + 3.6 0.0001 patients were receiving calcium antagoplasma insulin values ($J/ml) nists and 5 were receiving angiotensinDifference between 0 and 30 min converting enzyme inhibitors. In the plasma glucose values x10’ 31.5 + 5.5 67.6 i 4.6 0.0001 Difference between 0 and 30 min CAD group, 8 patients were treated with plasma insulin values calcium antagonistsand 6 with angiotensin-converting enzyme inhibitors. “CigValues ore expressed us mean + SEM. HDL = hlgh~densih/ lipoprotein cholesterol. arette-years” was used to estimate the CORONARY
ARTERY
DISEASE/CAD
RISK
FACTORS
AND
GLUCOSE
TOLERANCE
123
RESULTS Characteristics of the study population are summarizedin Table I. The cumulative lifetime consumption of tobacco was significantly higher in the CAD group. The score of the CAD group averaged 13.1 f 1.4. Total cholesterol and triglyceride levels were significantly higher in subjects with than without CAD. Plasma concentration of the HDL cholesterol in the CAD group was significantly lower than that in the normal group. Plasma apo B levels were significantly higher in the CAD than in the normal group, and plasma apo A-I levels were significantly lower in the CAD group. A hyperinsulinemic responseto an oral glucose load was observedin subjects with CAD. In subjectswith normal glucose tolerance, no significant differencesin total cholesterol and triglyceride levels were observed between the 2 groups (Table II); however, HDL cholesterol and plasma apo A-I levels were significantly lower in the CAD than in the normal group. The plasma apo B level was significantly higher in the CAD group. The elevation of plasma insulin concentrations during the test period and the early phaseof the glucose challenge were significantly higher in the CAD than in the normal group. Conventional coronary risk factors, as well as plasma lipid and apolipoprotein concentrations, and the insulinemic measurementsin subjects with impaired glucose tolerance were included in the analyses.Total cholesterol and triglyceride levels were significantly higher in subjectswith than without CAD. The plasma concentration of the HDL cholesterol in the CAD group was significantly lower than that in the normal group. The level of plasma apo A-I was significantly lower in the CAD than in the normal group, whereas the plasma apo B level was significantly higher in the CAD group. A hyperinsulinemic response to an oral glucose load was observed in subjects with CAD. Table II also shows the clinical and metabolic characteristics of the subjects with NIDDM. Total cholesterol and triglyceride levels were higher in subjects with than without CAD. There were significant differences in the levels of HDL cholesterol, plasma apo A-I, and plasma apo B between the 2 groups. The CAD group had significantly higher insulinemic measuresduring the glucose challenge. To investigate the possibility that the relation betweenhyperinsulinemia and clinical features,lipid concentrations, and apolipoprotein concentrations could be curvilinear, we calculated the clinical and metabolic data in the tertiles of the sum of 124
THE AMERICAN JOURNAL OF CARDIOLOGYa
VOL. 7.5
JAN. 15, 1995
TABLE III Characteristics Glucose Tolerance
of 1’27 Normoliptdemic
Men Grouped
in Tertiles of the Sum of Plasma Insulin levels in Response to a 7.5 g Oral
Test Tertifes of Sum of Plasma Insulin levels
Sum of plasma insulin values ($J/ml] Age (years] Body mass index [kg/m’] Cumulative tobacco consumption (cigaretteyears) Hypertension (no. of pts.) Total cholesterol [mg/dl) Triglycerides [mg/dl) HDL cholesterol (mg/dl) Apolipoprotein (mg/dl) A-l B Coronary artery disease (no. of pts.] tire number HDL = high-density
of subjects lipoprotein.
in group
Differences
Between Tertiles
Middle [n = 41)
High (n = 45)
1 “S 2
2 vs 3
1 “S 3
333 50 58.1 i 1.6 21.9 * 0.3 640 * 70
151-250 60.8 zt 1.5 23.2 in 0.4 984 + 80
251-446 58.2 I 1 .5 23.9 I 0.4 793 -e 65
0.221 0.012 0.002
0.223 0.225 0.065
0.964 0.001 0.112
12
23 194*4 126+4 34 + 1
0.111 0 292 0 112 0 037
0.659 0.006 0.007 0.024
0.039
45 f 2
19 178 +4 112+3 39 * 2
104i 3 98 + 3 26
94 + 4 106*3 30
87 z~ 3 126*4 38
0.048 0.062 0.342
0.160 0.001 0.199
0.001
172+4 104i4
01 mean
Significant
Low (n = 41)
0.001 0.001 0.001
0001
0.026
f SEM
plasma insulin values during the test period of the glucose challenge (Table III). Tertiles were based on the sum of plasma insulin levels during the glucose challenge in all subjects; therefore, the limits were similar for subjectswith normal glucose tolerance, impaired glucose tolerance, and NIDDM. Increasing tertiles of the sum of postglucose insulin levels were associatedwith an increase in body mass index, in cumulative tobacco consumption, and in the prevalence of hypertension and CAD. Increasing tertiles of the sum of postglucose insulin levels were also associatedwith increasesin total cholesterol, triglycerides, and apo B, and with reductions in HDL cholesterol and apo A-I. The relation between the sum of plasma insulin levels during the glucose challenge and abnormal lipid and apolipoprotein levels was positive and linear with respect to total cholesterol, triglycerides, and apo B levels, and negative and linear with respectto HDL cholesterol and apo A-I levels. Stepwise multiple linear regression analyses were performed to assessthe contribution of plasma lipid and apolipoprotein concentrations, body mass index, and the sum of postglucoseinsulin values to the variation in CAD score (Table IV). The analysesindicated that plasma apo A-I and apo B concentrations were more accuratepredictors of the CAD score than the corresponding levels of plasma lipids in this group of subjects.
DISCUSSION We found that the risk of CAD in normolipidemic male subjectswhose CAD had been diagnosedby coronary angiography was higher in those with endogenous hyperinsulinemia than in those with relatively less insulin secretion. In all subjects,postglucose hyperinsulinemia was associatedwith higher plasma total cholesterol, triglycerides, and apo B levels, and with lower HDL cholesterol and apo A-I levels in subjectswith normal and abnormal glucose tolerance, similar age, and obesity. Therefore, an impaired uptake of glucose mediated by insulin may be related to unfavorable changes in plasma lipid and apolipoprotein irrespective of glucosetolerance status.Analyses basedon the comparison CORONARY
TABLE IV Multiple Stepwise Linear Regression Analyses of Variables Associated with Coronary Artery Disease Score in 1 1 Study Subjects Irrespective of Glucose Tolerance Status
Risk Factors
Standard Partial Regression Coefficient
Total cholesterol Triglycerides HDL cholesterol Apo A-l Apo B Body mass index XIRI
0.014 0.093 -0.052 -0.209 0.247 0.164 0.068
R
0.578
F
3.948
Apo = cholesterol, coefficient
95% Confidence Interval for Regression Coefficient -0.089, -0.057, -0.201, -0.208, 0.028, -1.084, -0.065,
0.102 0.15 1 0.326 -0.009 0.235 2.256 0.083
0.888 0.374 0.639 0.044 0.006 0.891 0.805 0.0002
apolipoproiein; F = F stotlstic, HDL = highdensih/ ZIRI = sum of plasma
p Value
insulin
levels;
R = multiple
lipoprotein correlation
of lipid and apolipoprotein levels over tertiles of the postglucose insulin levels showed that the abnormal metabolism of lipids and apolipoproteins that carry an increased risk of CAD is associatedwith a hyperinsulinemic response independent of the glucose tolerance status in subjects with angiographically defined CAD. Our data suggest that hyperinsulinemia has a greater impact on the postprandial metabolism of lipids and apolipoproteins.‘Insulin can influence the activity of lipid transfer protein, which is especially pronounced after meals and is affected by HDL composition.1o It is important to determine how hyperinsulinemia exerts a variable influence on the metabolism of lipids and apolipoproteins. The iirst possibility, that hyperinsulinemia is the primary etiologic mechanism responsible for abnormal metabolism of lipids and apolipoproteins, is particularly supported by kinetic studies of triglyceride metabolism. Several studies”J2 have demonstratedthat hyperinsulinemia can induce an overproduction of triglyceride-rich VLDL in the liver by increasing the availability of free fatty acids, which are
ARTERY DISEASE/CAD
RISK FACTORS
AND
GLUCOSE
TOLERANCE
125
the most important, if not the only, precursorsof de novo triglyceride synthesis. The hepatic synthesis of triglycerides can result from hyperinsulinemia.13In addition to increased synthesis, the rate of VLDL removal by peripheral tissues Tom the circulation can be reduced in subjects with hyperinsulinemia, and both an increased synthesis and a decreasedclearance rate may coexist.14 A factor that may contribute to the association between hypertriglyceridemia and hyperinsulinemia is the resistance of adipose tissue lipolysis to inhibition by insulin. This factor and the resistance of glucose elimination to stimulation by insulin may bc linked.15The association between higher levels of apo B and hyperinsulincmia also may be explained by an increasedhepatic secretion of apo B in response to insulin resistance.t6,t7Hepatic insulin resistance may therefore alter both the rates of lipoprotein synthesisand catabolism in the liver. We previously reported that the increased rate of triglyceride secretion induced by hyperinsulincmia contributes to the hypertriglyceridemia observed in obeserats treated with monosodium glutamate.Ix The lipids and protein concentrations of VLDL and plasma free fatty acid concentration were significantly higher in the monosodium glutamate-treated rats than in the controls. HDL is relatcd to VLDL metabolism in that surface components from remnantsof triglyceride-rich lipoproteins are transferred to HDL during metabolism of VLDL. It is possible that the relation between HDL cholesterol and the insulinemic measurementswas due to a reduction in adipose tissue or in muscle lipoprotein triglyceride clearance, and/or an increased activity of hepatic lipase. Hyperinsulinemia also correlates with an increase in the fractional catabolic rate of iodine-125 apo A-I/HDL and a decreasein plasma HDL cholesterol concentration.t9 One must consider that abnormalities in lipid and apolipoprotein concentrations impair the action of insulin. High concentrations of VLDL lead to an impairment of insulin action.20 With respect to long-term effects of elevated VLDL concentration on glucose metabolism, it is interesting that a high plasma VLDL level was found to be a risk factor for glucose intolerance in the Framingham Study.21Obesity is associated with both higher insulin levels and higher level of lipids. In our study subjects, obesity did not influence lipid metabolism but did influence insulin secretion (data not shown).
126
IHE AMERICA\
JOURNAL OF CARIOLOGY=
VCI
75
1. Fujiwara R, Kutsurm Y, Hayashi T. Kim S-S, Misawa ‘I‘, Tada II, Nishio Il. ‘Toyora K, Tamai T, bkai T, Miyabo S. Metabolic risk factors in the normolipidemic malt patients with angiographically defined coronary ancry disease. Jpn Gin J 199X4:493 -500. 2. Staldcr M, Pomctta D. Suenram A. Relationship behveen plasma insulin lcvclr 1981: and high density lipoprotein cholesterol levels in healthy men. Di&erolugin 2 I :.54&548. 3. Orchard TJ. Becker DJ, Bales M, Kuller LH, Dmsh AL. Plasma insulin and hpoprotein concentrations: an arhcrogcnic asxxiation? Am J Epidemiol 1983;118: 326337. 4. Modan M, llalkin H, Lusky A, Segal P, Fuchs 7, Chetrit A. Hyperinsulinemia ir characterbed by jointly disturbed plasma VLDL, LDL, and HDL levels. A population-ha.sed study. Arreriosrlrrosrs 1988;8:227--236. 5. Austen WC. Edward E, Frye RC. Gcnsini GC. Gott VL, Gnflith CCS, Goon DC, Murphy ML, Rowe BB. A reporting system on patients evaluated for coronary artery disease. Kepon of the Ad Hw Commiwe for Grading of Coronary Awry Disease. Counsel on Cardiouw~lar Surgery, American Heart Association. Ci,zularron 1975;5 I(suppl 1):SdO. 6. World Health Organization: Diabetes mellitus: repon of a WI10 study group. Geneva. World Health Organization, technical report series no. 727, 1985: l&12. 7. Iicrbe~t V, I.au K. Gottlieb CW, Bleicher SJ. Coated charcoal immunoassay of insulin. J C/in Endocnnol .Mefah 1965;25: 1375-1384. 8. Burstein Y. Scholnick HR. Marlin R. Rapid method for the isolation of lipoproteins from human semm by precipitation with polyanions. J Lipid Res 1970~11: 583-59.5. 9. Goto Y, Akanuma Y, Harano Y. Hara Y, Itakura H, Kajiyama G. Kawade MI. Koga M, Kuzuya F, Mamhama Y, Matsuzawa Y. Murai A, Murase T, Nailo C, Nakai ‘I‘, Noma A, Saitoh Y. Sasaai J. Takeuchi N. Yama/aki S, Yarugi T. Yukawa S. Determination hy the SRID method of normal values of setum apolipoproteins (A-l. A-II. B, C-II. C-III and E) in normolipidemic healthy Japanese subjects. J C/in Biochem NW 19X6: 1:7>8X. 10. Patxh JR. Karlin JB. Scott LW, Smith LC, Gotto AM. Inverse relationship between hlood level of high density lipoprotein subfraction 2 and magnitude of post-pnandial lipemia. Proc Nail Acad Sci LfSA 1983;80:1449 -1453. 11. Swiner G. Vranic M. Hyperinsulinemia and hypertriglyceridemia: a vicious cycle with atherogenic potential. In/J Obcs 19X2:6: I 17-124. 12. Tohey TA. Greenfield M, Kraemer I:, Reaven GM. Relationship between insulin resistance. insulin secretion, very low density lipoprotein kin&$ and plasma triglyccridc levels in nonnotriglyccridemic man. .Mrrabolirm 1981;30:165-171. 13. Reaven GM. Chen Y-D. Role of insulin in regulation of lipoprotein metaboliam in diabctcs. Diubheres Merub Rev 3988;4:639652. 14. Grundy SY. Pathogenesis of hypcrhiglyceridemia in man. In: Carlson I.A, Olsson AG, eds. Trcaunent of Hypcrlipoproteinemia. I\;ew York, Raven Press, 1984: 21-34. 15. Chen Y-D, Golay A, Swislocki A, Reaven GM. Keristance to insulin suppresbion of plasma free fatty and concentrations and insulin btimulation of glucose uptake in noninsulindepcndent diabetes mellitus. J C/h Endocrinnl Mrtab 1987:64:17.-21. 16. Dashti X. Wolfbauer G. Secretion of lipids. apolipoproteins, and lipoproteins hy human hepatoma cell line, HepG2: effcctr of oleic acid and insulin. J Liprd Res 1987;28:423426. 17. Dashti X Williamb DL, Alaupwic P. Effects of oleate and insulin on the pmduction and cellular mRNA concentrations of apolipoproteins in flepG2 cells. J
Lipid Rcs 1989;30:1365-1373. 18. Oida K, Sakai T. Hayashi T. bliyaho S. Takeda R. Plasma lipoproteins of monosodium gluramate-induced obese rats. Inr J Otws 19844:8:385-391. 19. Golay A, Zech 1.. Shi M-Z, Jeng C-Y, Chiou Y-AM, Reaven GM, Chen Y-DI. J Lipid Res Role of insulin in qulation of high density lipoprotein metabolism. 1987:28:1&18. 20. Bieger WI’, Michel G. Banvich D. Biehl K, Wirth A. Diminished insulin rccepton on monocytes and cryxhrocy~es in hypuuiglyccridemia. Meraholirnr 19X4:33: 9X2- 9X7. 21. Wilson PW. McGee DL. Kannel WB. Obesity, very low density lipoproteins, and glucose inrolerance over fourteen years. Am J Epidrmiol I98 I:1 14:697-704.
JAN
15, 1995