Insulin Resistance Is Associated with Lipid andLipoprotein Abnormalities in Subjects with Varying

Degrees of Glucose Tolerance

Markku Laakso, Helena Sarlund, and Leena Mykkanen

We tested the hypothesis that Insulin resistance, rather than high Insulin level, isassociated with lipid and lipoprotein changes favoring atherosclerosis Independently ofthe glucose tolerance status. To this aim, 50 subjects with normal glucose tolerance,28 subjects with Impaired glucose tolerance, and 54 subjects with nonlnsulln-dependentdiabetes (NIDDM) were studied. Subjects wtth low glucose disposal rate (GDR) or a highdegree of Insulin resistance as measured by the euglycemlc hyperinsullnemlc damptechnique had lower high density lipoprotein (HDL) cholesterol and higher total and verylow density lipoprotein (VLDL) triglycerldea than did subjects wtth high GDR (highest GDRfertile). These associations were Independent of fasting insulin level and other confound-ing factors. In stepwlse multiple linear regression analysis, GDR was the most importantsingle variable associated with HDL cholesterol and VLDL trigh/ceride level Independentlyof age, obesity, distribution of obesity (waist/hip ratio), 2-hour glucose level, and free fattyacid concentration. We conclude: 1) Insulin resistance measured by the euglycemlc clamptechnique Is associated with adverse llpkj and lipoprotein changes favoring atheroscle-rosis not only In nondiabettc subjects (as shown In previous studies) but also In Impairedglucose tolerance and NIDDM subjects; 2) the association of high insulin level withadverse lipid and lipoprotein changes Indirectly reflects the association of Insulin resis-tance wtth lipid and lipoprotein levels; and 3) HDL cholesterol and VLDL triglycerldes areIndependently associated wtth Insulin-mediated glucose uptake, which may indicate thatthese llpoprotelns have separate sites of Interaction with Insulin action.(Arteriosclerosis 10:223-231, March/April 1990)

Evidence from experimental, clinical, and epidemio-logical studies suggests that hyperinsulinemia is

linked with the development of atherosclerotic vascularcomplications.12 Three prospective population studies,the Helsinki Policemen Study, the Paris ProspectiveStudy, and the Busselton Study, have independentlyshown that high plasma insulin level is associated with anincreased risk of coronary heart disease in nondiabeticmale subjects.345 Information on the relationship ofplasma insulin to atherosclerosis in noninsulin-depen-dent diabetes (NIDDM) is scanty.6

The mechanisms by which hyperinsulinemia promotesatherosclerosis are, however, unknown. Several studiesindicate that high insulin concentration is associated withadverse changes in serum lipid and lipoprotein levels,characterized by elevated total and very low densitylipoprotein (VLDL) triglyceride and decreased high den-sity lipoprotein (HDL) cholesterol levels in subjects withnormal glucose tolerance (NGT)7-11 and NIDDM.11 Sincefasting insulin level correlates closely with insulin-mediated glucose uptake,12 it is possible that high insulin

From the Department of Medicine, Kuopio University Centra)Hospital, Kuopio, Finland.

This work was supported by a grant from the Academy ofFinland.

Address for correspondence: Markku Laakso, M.D., Depart-ment of Medicine, Kuopio University Central Hospital, 70210Kuopio, Finland.

Received December 29, 1988; revision accepted October 5,1989.

concentration Is only an indirect indicator of the linkbetween insulin resistance and adverse changes in lipidand lipoprotein levels. Indeed, two recent studies insubjects with NGT have demonstrated that low insulin-mediated glucose uptake is associated with lipid andlipoprotein changes favoring atherosclerosis.1314 Sub-jects with impaired glucose tolerance (IGT) and NIDDMare more insulin-resistant than are subjects with NGT,16

and in both there is an increased risk of atherosclerosis.2

Therefore, if insulin resistance is a general risk factor foratherosclerosis, low insulin-mediated glucose uptakeshould be associated with adverse lipid and lipoproteinchanges irrespective of the glucose tolerance status. Totest this hypothesis, we studied lipid and lipoproteinlevels and their association with insulin-mediated glucoseuptake in a total of 132 subjects with varying degrees ofglucose tolerance.

MethodsSubjects

The subjects for this study, ages 50 years and older,were recruited from the previous population studies doneby our department.1617 Random samples of 50 subjectswith NGT, 28 subjects with IGT, and 54 patients withNIDDM (duration of diabetes 10±1 years) were studied.The diagnosis of diabetes was based on a previoushistory of diabetes and a 2-hour oral glucose tolerancetest (75 g of glucose) according to the World Health

223

by guest on July 10, 2017http://atvb.ahajournals.org/

Dow

nloaded from

224 ARTERIOSCLEROSIS VOL 10, No 2, MARCH/APRIL 1990

Organization criteria.18 Among diabetic subjects, onlypatients with NIDDM and those treated with diet only orwith oral antidiabetic drugs were included. Eight of thediabetic subjects were treated with diet only, one withmetformin, 13 with sutfonylureas, and 28 with a combi-nation therapy of metformin and sulfonylureas. The defi-nition of IGT and NGT was based on a 2-hour oralglucose tolerance test according to the Worid HealthOrganization criteria.18 All subjects were in good healthas judged by medical history, physical examination, androutine blood and urine tests. All patients had normalliver, kidney, and thyroid function, and no subject wasreceiving drugs for hyperlipidemia. The use of diureticsand beta-blocking agents was more frequent in patientswith NIDDM (28% and 57%, respectively, p<0.05) than insubjects with NGT (8% and 18%, respectively) or insubjects with IGT (11% and 36%, respectively).

Study Protocol

The subjects were admitted to the hospital's metabolicward for 2 to 3 days. During their stay, a weight-maintainingdiet (30 kcal/kg/day) was given; the diet contained 50%carbohydrate, 30% fat, and 20% protein. A 2-hour oralglucose tolerance test and euglycemic hyperinsuiinemicclamp studies were performed after a 12-hour fast

The study protocol was approved by the Ethics Com-mittee of the University of Kuopio. Informed consent wasobtained from each subject.

Procedures

Oral Glucose Tolerance Test

After an overnight fast, the oral glucose tolerance testwas performed with 75 g of glucose. Blood was drawn at0, 60, and 120 minutes for the measurement of bloodglucose and plasma insulin concentrations. The 2-hourarea under the glucose and insulin curve was calculated.

Studies of Insulin Action

Insulin action was assessed with the euglycemic clamptechnique.19 At 7:30 A.M. after a 12-hour overnight fast, anintravenous catheter was placed In an antecubital vein forinfusion of insulin, 20% glucose, and (3-3H) glucose (NewEngland Nuclear, Boston, MA). Another cannula wasinserted into a wrist vein surrounded by a heated box(70°C) for blood sampling. After insertion of the catheters,(3-3H) glucose was infused as a primed (40 jiCi) constant(0.40 jiCi/min) infusion for 120 minutes before the start-ing of insulin infusion. A priming dose of insulin (VelosulinHuman, Nordisk Insulin, Gentofte, Denmark) was admin-istered during the initial 10 minutes in a logarithmicallydecreasing manner to acutely raise serum insulin to thedesired level, where it was maintained by a continuousinsulin infusion of 40 mU/mJ!/min. Steady-state insulinconcentrations were 98±2, 100±4, and 91 ±5 mU/l insubjects with NGT, IGT, and NIDDM, respectively (p=NSbetween the groups). Blood glucose was maintained at5.5 mmoVl in each subject for 120 minutes by infusing20% glucose at varying rates according to blood glucosemeasurements done at 5-minute intervals (mean coeffi-cient of variation of blood glucose <4% in each group).

Before the initiation of glucose infusion in subjects withNIDDM, blood glucose was allowed to fall to 5.5 mmol/lwith insulin infusion. During euglycemic hyperinsuiinemicclamp studies, the rates of glucose appearance (Ra) anddisappearance (Rd) were quantified from the serum(3-'H) glucose specific activities and calculated from theSteele's equations in their modified derivative form be-cause the tracer exhibits non-steady-state kinetics underthese conditions.20 The isotope dilution technique hasbeen recently criticized for underestimating the rates ofRa and Rd. Although the reasons for this are not com-pletely dear, it is generally recognized that this underes-timation is greatest in non-steady-state and at high glu-cose turnover rates.21 Since all subjects were studied insteady-state conditions and the mean glucose disposalrate (GDR) was <6 mg/kg/min, the underestimation of Rais small in such study conditions. The data were calcu-lated for each 20-minute interval. The mean value for theperiod 60 to 120 minutes was used to calculate the GDR.

UpkJ and Upoproteln Measurements

Serum lipids and lipoproteins were determined fromfresh serum samples drawn after a 12-hour overnight fastLJpoprotein fractionation was performed using ultracentrifu-gabon and selective precipitation with a modification22 of themethod originally described by Havel et al.23 All spinningswere done at 1 fJC by using a Kontron TGA-65 ultracentri-fuge (Kontron International, Switzerland). Serum sampleswere centrifuged at density (d)=1.006 (105 000 g, for18 hours), and VLDL triglycerides (d<1.006) were recov-ered as the top fraction. Total HDL cholesterol was deter-mined directly with dextrane sulphate and magnesiumchloride precipitation, which correlates closely with theresults obtained by ultracentrifugation.22 Low density lipo-protein (LDL) cholesterol (d=1.006 to 1.063, includingintermediate density lipoprotein) was calculated as a dif-ference between the bottom fractions. On the average, themean day-to-day variation in HDL cholesterol measure-ment was 3.3%, and the daily variation was 0.95%. Cho-lesterol and triglycerides from the whole serum and fromlipoprotein fractions were assayed by automated enzy-matic methods (Boehringer-Mannheim, FRG).

Analytic Methods

Blood glucose was measured by the glucose oxidasemethod (Glucose Auto & Stat HGA-1120 analyzer, Daiici,Kyoto, Japan). Serum free fatty acids (FFA) were deter-mined by the enzymatic method of Wako ChemicalsGmbH (Newss, FRG). Plasma insulin was determined bya radioimmunoassay (Phadeseph Insulin RIA 100, Phar-macia Diagnostics AB, Uppsala, Sweden). Coefficients ofvariation within and between insulin assays were 5% and4%, respectively. The (3-3H) glucose specific activity inblood samples was determined after precipitation ofproteins with perchloric add. The supernatant was evap-orated to dryness, dissolved in water, and counted inaqueous sdntillation fluid, as previously described.24

Data Analysis

All calculations were performed by using the SPSSXprograms. The data are presented as the means±SEM.

by guest on July 10, 2017http://atvb.ahajournals.org/

Dow

nloaded from

INSULIN RESISTANCE AND UPOPROTEINS Laakso et al. 225

Table 1. Clinical and Metabolic Data on Subjects with Normal Glucose Tolerance, ImpairedGlucose Tolerance, and Nonlnsulln-dependent Diabetes

Men/womenAge (years)Weight (kg)Body mass index (kg/m2)Waist/hip ratioFasting glucose (mmol/I)2-hr glucose (mmol/I)2-hr glucose area (mmol/l-h)Fasting Insulin (mll/1)2-hr Insulin (mU/l)2-hr Insulin area (mU/Hi)Free fatty acids (mmol/I)Glucose disposal rate

(mg/kg/min)

Normal glucosetolerance

23/2765±176±228±1

0.96±0.014.7±0.15.2±0.1

11.9±0.313.2±0.864.0±6.0

147.5±12.20.65±0.04

5.9±0.3

Impaired glucosetolerance

15/1364±276±228±1

0.98±0.015.0±0.1*7.7±0.2+

15.2±0.4*16.0+1.3

108.0±20.3*178.5±23.80.67±0.04

4.6±0.4*

Noninsulin-dependentdiabetes28/2661 ±279±229±1

0.97±0.0110.2±0.4*18.5±0.6*31.6±1.0+13.2±0.835.2±3.9*69.8±6.3*0.84±0.05t

3.8±0.2t

*p<0.05, tp<0.01, +p<0.001glucose tolerance).

(Impaired glucose tolerance or noninsulin-dependent diabetes versus normal

Analysis of variance (ANOVA) was used in the comparisonof more than two groups, and the adjustment for con-founding factors was performed with the analysis of cova-riance (ANCOVA). Student's two-tailed f test for indepen-dent samples was used in the comparison of two groups.Correlations were calculated with Pearson correlation co-efficients. Multiple stepwise linear regression analy-ses were performed to investigate the association ofdifferent variables with HDL cholesterol and VLDL triglyc-eride concentrations and with GDR. In all statistical analy-ses, logarithmic values for total and VLDL triglycerides and2-hour insulin area were used, including these variables.

ResultsAs shown in Table 1, the study groups did not differ

with respect to sex, age, weight, body mass index (BMI),or waist/hip ratio. Fasting insulin was similar between thegroups, but 2-hour insulin was higher in subjects with IGTand lower in subjects with NIDDM than in subjects withNGT. Patients with NIDDM had lower 2-hour insulin areasin an oral glucose tolerance test and higher levels of FFAthan did subjects with NGT. GDR was significantly lowerin subjects with IGT (p<0.05) and in subjects withNIDDM (p<0.01) compared with that in subjects withNGT. In each study group, GDR was similar in men andwomen (NGT: 6.0±0.4 vs. 5.8±0.5 mg/kg/min; IGT:4.5±0.5 vs. 4.8±0.6 mg/kg/mln; NIDDM: 4.1 ±0.3 vs.3.6±0.4 mg/kg/min). Therefore, in all the following statis-tical analyses the results were combined by sex.

GDR correlated inversely with total and VLDL triglycer-ides and positively with HDL cholesterol when all subjectswere included in the analyses (Table 2). No consistentcorrelations between GDR and other lipoprotein fractionswere observed, and therefore Table 2 reports only corre-lations with respect to HDL cholesterol and total andVLDL triglycerides. The correlation between GDR andHDL cholesterol was positive in each group, but only insubjects with IGT did it reach the conventional level of

statistical significance. Fasting glucose and 2-hour glu-cose area in an oral glucose tolerance test correlatedsignificantly with total and VLDL triglycerides, and 2-hourglucose area correlated negatively with HDL cholesterol.However, within each study group (NGT, IGT, NIDDM)these correlations were not statistically significant. Fast-ing insulin and 2-hour insulin area did not correlatesignificantly with liplds and lipoproteins in any subject,but in subjects with NGT and IGTs, fasting and 2-hourinsulin area correlated significantly and inversely withHDL cholesterol. In addition, in subjects with NGT, cor-relations of total and VLDL triglycerides with 2-hourinsulin area, and in subjects with IGT, correlations of totaland VLDL triglycerides with fasting insulin level werepositive and statistically significant.

To investigate the possibility that the relationship be-tween GDR and liplds and lipoproteins could be curvilin-ear, we also calculated liptd and lipoprotein concentrationsin the tertiles of GDR (Table 3) (lowest tertile <3.32 mg/kg/min, highest tertile >5.43 mg/kg/min). The tertiles wereformed on the basis of GDR in the whole series of subjects,and therefore the limits are similar for subjects with NGT,IGT, and NIDDM. Total, LDL, and VLDL cholesterol andHDL triglycerides did not differ between the GDR tertiles inany of the study groups. HDL cholesterol was significantlyhigher in the highest GDR tertile than in the lowest tertile insubjects with NGT. In subjects with NGT and NIDDM, totaland VLDL triglycerides differed between GDR tertiles andwere significantly higher in the highest GDR tertile com-pared with the lowest tertile. In patients with NIDDM, LDLtriglycerides were lower in the highest GDR tertile than inthe lowest tertile.

Table 4 reports unadjusted and adjusted lipid andlipoprotein levels by GDR tertile and includes all studysubjects irrespective of the glucose tolerance status.Adjustment was done separately for fasting insulin (A)and for age, BMI, fasting insulin, and 2-hour glucose in anoral glucose tolerance test and FFA (B). These variables

by guest on July 10, 2017http://atvb.ahajournals.org/

Dow

nloaded from

226 ARTERIOSCLEROSIS VOL 10, No 2, MARCH/APRIL 1990

Table 2. Correlations of Upld and Upoprotein Concentrations with Glucose Disposal RateMeasured by the Euglycemic Clamp Technique and Glucose and Insulin Concentrations duringan Oral Glucose Tolerance Test In Subjects with Normal Glucose Tolerance, Impaired GlucoseTolerance, and Nonlnsulln-dependent Diabetes, and In All Subjects Combined

Glucose disposal rate(mg/kg/min)

HDL cholesterol

Total trlglycerldes

VLDL trlglycerldes

Fasting glucose (mmol/1)

HDL cholesterol

Total triglycerides

VLDL triglycerides

2-hr glucose area (mmol/l-h)

HDL cholesterol

Total triglycerides

VLDL triglycerides

Fasting Insulin (mU/l)

HDL cholesterol

Total triglycerides

VLDL triglycerides

2-hr Insulin area (mU/l-h)

HDL cholesterol

Total triglycerides

VLDL triglycerides

Normalglucose tolerance

0.26

-0.29*

-0.35*

-0.08

-0.06

-0.03

-0.15

-0.06

0.06

-0.29*

0.18

0.25

-0.431"

0.32*

0.43t

Impairedglucose tolerance

0.50*

-0.40*

-0.40*

-0.13

0.33

0.32

-0.39

-0.06

-0.11

-0 .41*

0.50t

0.53t

-0.55t0.32

0.33

Nonlnsulln-dependentdiabetes

0.15

-0.42t-0.37*

0.04

0.23

0.23

0.00

0.19

0.19

0.06

0.05

0.02

0.01

0.08

0.05

Allsubjects

0.37t

-0.46*

-0.47*

-0.20

0.43t

0.42t

-0.27*

0.43t

0.43t

-0.20

0.16

0.18

-0.08

-0.11

-0.07

*p<0.05, tp<0.01, tp<0.001.HDL=hlgh density Upoprotein, VLDL=very tow density Upoprotein.

were selected as covariates because age, BMI, fastinginsulin, 2-hour glucose, and FFA concentrations werehigher in subjects in the highest GDR tertile than insubjects in the lowest GDR tertile (data not shown). Allanalyses were also done after the exclusion of subjectsreceiving antihypertensive medication (diuretics or betablocking agents) (C). HDL cholesterol was higher andtotal and VLDL triglycerides lower in subjects in thehighest GDR tertile than In subjects in the lowest GDRtertile, and these differences persisted even after theadjustment for insulin (A) and other related variables (B).The differences in total and VLDL triglycerides persistedeven after the exclusion of subjects with anti hypertensivemedication (C). All these analyses were also performedseparately in obese (BMI£27.0 kg/m2) and nonobese(BMI<27.0 kg/m2) subjects. The results showed that lowHDL cholesterol and high total and VLDL triglycerideswere associated with low GDR similarly in both obese andnonobese groups (data not shown). The unadjustedmeans for VLDL cholesterol and HDL triglycerides werelower in subjects with high GDR compared to subjectswith low GDR, but the differences between these subjectsdisappeared after the adjustment for related factors (B).

Stepwise multiple linear regression analyses were per-formed to assess the independent contributions of age,BMI, waist/hip ratio, and 2-hour glucose level in an oralglucose tolerance test, fasting FFA concentration, GDR,and fasting Insulin to the variation in HDL cholesterol andVLDL triglycerides (Table 5). GDR was the most impor-tant single variable associated positively wtth HDL cho-lesterol and negatively with VLDL triglycerides and ex-plained 14% and 24%, respectively, of the variation inHDL cholesterol and VLDL triglycerides. Also, 2-hourglucose associated significantly with HDL cholesterol andVLDL triglycerides, and FFA were associated significantlywith HDL cholesterol. When these analyses were per-formed by substituting fasting Insulin with 2-hour insulinor 2-hour insulin area, the results remained essentiallyunchanged (data not shown).

Analyses similar to those reported in Table 5 wereperformed to assess the contribution of age, BMI, waist/hip ratio, 2-hour glucose, FFA, and HDL cholesterol(Model A) or VLDL triglycerides (Model B) to the variationin GDR. In both models, BMI and 2-hour glucose wereassociated inversely with GDR and explained 34% and 16%,respectively, of the variation in GDR. VLDL triglycerides werestrongly and inversely associated with GDR and explained

by guest on July 10, 2017http://atvb.ahajournals.org/

Dow

nloaded from

INSUUN RESISTANCE AND UPOPROTEINS Laakso et al. 227

Table 3. Upld and Upoprotein Levels by Glucose Disposal Rate Tertlle In Subjects with NormalGlucose Tolerance, Impaired Glucose Tolerance, and Nonlnsulin-dependent Diabetes

Cholesterol

Total

NGT

IQT

NIDDM

HDL

NGT

IGT

NIDDM

LDL

NGT

IGT

NIDDM

VLDL

NGT

IGT

NIDDM

Trlglyce rides

Total

NGT

IGT

NIDDM

HDL

NGT

IGT

NIDDM

LDL

NGT

IGT

NIDDM

VLDL

NGT

IGT

NIDDM

Low

5.97±0.18

5.56±0.47

6.56±0.23

1.07±0.08

1.03±0.08

1.01 ±0.03

3.99±0.17

3.66±0.40

3.91 ±0.19

0.92±0.15

0.86±0.25

1.65±0.17

2.16±0.28

2.20±0.53

3.75±0.34

0.21 ±0.03

0.21 ±0.03

0.29±0.02

0.42 ±0.04

0.35±0.07

0.60±0.05

1.54±0.25

1.64±0.45

2.87±0.30

Middle

6.69±0.37

6.05±0.25

6.00±0.31

1.15±0.08

1.02±0.06

1.06±0.05

4.41 ±0.36

3.93±0.20

3.75±0.26

1.13±0.20

1.10±0.10

1.19±0.16

2.30±0.35

2.36±0.27

2.80±0.38*

0.20±0.03

0.18±0.02

0.28±0.04

0.52±0.08

0.49±0.06

0.49±0.04

1.58±0.29

1.68±0.23

2.02±0.36*

High

6.29±0.25

5.94±0.28

6.36±0.65

1.32 ±0.06*

1.24±0.08

1.11 ±0.09

4.14±0.23

3.75±0.20

3.95±0.37

0.83±0.09

0.95±0.19

1.31 ±0.34

1.58±0.14*

1.58±0.20

2.14±0.33t

0.18±0.02

0.15±0.03

0.26±0.04

0.45±0.04

0.38±0.04

0.35±0.07t

0.96±0.12t1.05±0.17

1.53±0.28t

ANOVAp value

NS

NS

NS

0.049

0.083

NS

NS

NS

NS

NS

NS

NS

0.033

NS

0.004

NS

NS

NS

NS

NS

0.035

0.008

NS

0.006

Values are given in mmol/l.*p<0.05, tp<0.01 (middle or high tertile vs. tow fertile).NGT-nomial glucose tolerance, IGT=lmpaired glucose tolerance, NIDDM=noninsulln-dependent diabetes,

HDL=hlgh density lipoprotein, LDL-low density lipoprotein, VLDL=very low density lipoprotein.

24% of the variation in GDR. HDL cholesterol and FFA levelswere weakly associated with GDR and explained 4% and 2%,respectively, of the variation in GDR.

DiscussionOur study showed that low HDL cholesterol and high

total and VLDL triglycerides were associated with insulin

resistance measured by the eugtycemic clamp techniquein subjects with NGT (as reported in previous studies1314)and also in subjects with abnormal glucose tolerance andsimilar ages and obesity. Therefore, impaired insulin-mediated glucose uptake is related to adverse lipid andlipoprotein changes irrespective of the glucose tolerancestatus.

by guest on July 10, 2017http://atvb.ahajournals.org/

Dow

nloaded from

228 ARTERIOSCLEROSIS VOL 10, No 2, MARCH/APRIL 1990

Table 4. Unadjusted andTertile Including All Study

Cholesterol

Total

Unadjusted mean

Adjusted mean A

Adjusted mean B

Adjusted mean C

HDL

Unadjusted mean

Adjusted mean A

Adjusted mean B

Adjusted mean C

LDL

Unadjusted mean

Adjusted mean A

Adjusted mean B

Adjusted mean C

VLDL

Unadjusted mean

Adjusted mean A

Adjusted mean B

Adjusted mean C

Trig lyce rides

Total

Unadjusted mean

Adjusted mean A

Adjusted mean B

Adjusted mean C

HDL

Unadjusted mean

Adjusted mean A

Adjusted mean B

Adjusted mean C

LDL

Unadjusted mean

Adjusted mean A

Adjusted mean B

Adjusted mean C

VLDL

Unadjusted mean

Adjusted mean A

Adjusted mean B

Adjusted mean C

Adjusted Upld and UpoprotelnSubjects Irrespective of Glucose

Low

6.19

6.24

6.09

5.99

1.03

1.04

1.06

1.10

3.84

3.86

3.93

3.80

1.32

1.35

1.09

1.08

3.03

3.10

2.63

2.53

0.26

0.26

0.24

0.23

0.51

0.52

0.49

0.45

2.28

2.32

1.91

1.85

Middle

6.22

6.22

6.22

6.46

1.06

1.06

1.08

1.24

4.01

4.01

4.02

4.34

1.15

1.15

1.12

0.87

2.55

2.55

2.48

1.54*

0.22

0.22

0.22

0.16

0.51

0.52

0.51

0.40

1.82

1.82

1.76

0.98*

Levels by GlucoseTolerance Status

High

6.25

6.20

6.23

6.03

1.27

1.26

1.23*

1.27*

4.03

4.01

3.96

3.90

0.95*

0.92*

1.05

0.85

1.70

1.64

2.06*

1.84*

0.19

0.18

0.20

0.18

0.42

0.41

0.43

0.43

1.09

1.04

1.43*

1.23*

Disposal Rate

ANOVA/ANCOVAp value

NS

NS

NS

NS

<0.001

<0.001

0.050

NS

NS

NS

NS

NS

0.028

0.052

NS

NS

<0.001

<0.001

0.055

0.011

0.012

0.006

NS

NS

NS

NS

NS

NS

<0.001

<0.001

0.034

0.006

Values are given In mmol/1.*p<0.05, tp<0.01, *p<0.001 (middle or high tertile vs. low tertile).A=adjusted for fasting Insulin level (ANCOVA).B=adjusted for age, body mass Index, fasting insulin, 2-hr glucose, and fasting free fatty acids (ANCOVA).C=adjusted for age, body mass Index, fasting insulin, 2-hr glucose, and fasting free fatty acids (ANCOVA) in

subjects not receiving antihypertenslve medication (glucose disposal rate tertlles: lowest tertJle<4.62 mg/kg/min,highest tertile>6.71 mg/kg/min).

Logarithmic transformations for total and VLDL triglycerides were used in the comparison of tertile groupsincluding these variables.

See legend to Table 3 for explanation of abbreviations.

by guest on July 10, 2017http://atvb.ahajournals.org/

Dow

nloaded from

INSULIN RESISTANCE AND LIPOPROTEINS Laakso et al. 229

Table 5. Multiple Stepwise Linear Regression Analyses of Variables Associated with HDLCholesterol and VLOL Triglycerlde Concentrations In All Study Subjects Irrespective of GlucoseTolerance Status

Independent variable

Age

Body mass Index

Walst/hlp ratio

2-hr glucose

Free fatty acids

Glucose disposal rate

Fasting Insulin

R2 for the model (%)

HDL cholesterol

Beta

0.011

-0.022

-0.145

-0.250t

0.231t0.333*

-0.138

Dependent variable

0

0

0

3.5

4.7

13.5

0

21.7

VLDL triglycerides

Beta

0.018

0.128

-0.048

0.293*

-0.140

-0.368*

0.084

0

0

0

7.1

0

23.8

0

30.9

+p<0.01, 4p<0.001.Beta=standardized regression coefficient, fl!=proporHon explained by a given Independent variable (%).HDL=>hlgh density llpoprotetn, VLDL=very tow density llpoprotein.

GDR was positively correlated with HDL cholesteroland negatively correlated with total and VLDL triglycer-ides (Table 2). Analyses based on the comparison of lipidand lipoprotein levels over the GDR tertlles showed thatthe relationship between GDR and abnormal lipid andlipoprotein levels was linear with respect to total andVLDL triglycerides (Table 3). In contrast, low HDL cho-lesterol seemed to associate with both the lowest and themiddle GDR tertjles. These associations were indepen-dent of fasting insulin level and other related factors(Table 4). These results suggest that previously reportedcorrelations between fasting insulin and lipid and lipopro-tein concentrations are only indirect indicators of theassociation between insulin resistance and adverse lipidand lipoprotein changes. Since the adjustment was donealso for 2-hour glucose, these findings show that low HDLcholesterol and high total and VLDL triglyceride concen-trations are associated with insulin resistance (low GDR)independently of the glucose tolerance status (NGT, IGT,and NIDDM).

At least three mechanisms must be considered toexplain the association between Insulin resistance andlipid and lipoprotein abnormalities: 1) Insulin resistancecauses adverse changes in lipids and lipoproteins.2) Abnormalities in lipids and lipoproteins cause an im-pairment in insulin action. 3) The association of insulinresistance and abnormal lipid and lipoprotein levels iscaused by factors that influence both glucose and lipidmetabolism.

The first possibility, that insulin resistance is the pri-mary pathogenetic mechanism responsible for abnormal-ities in lipid and lipoprotein metabolism, is supportedparticularly by kinetic studies of triglyceride metabolism.Most studies26-26 have indicated that insulin resistanceand hyperinsulinemia can induce an overproduction oftriglyceride-rich VLDL in the liver by increasing the avail-ability of FFA. However, some studies have failed to showany significant correlation between insulin level and VLDLproduction,27 and studies2829 in isolated hepatocyteshave even demonstrated an inhibitory effect of insulin onVLDL production. In addition to increased synthesis,

VLDL clearance from the circulation can be reduced insubjects with insulin resistance, particularly in NIDDM,and both increased synthesis and decreased removalrate may coexist.30 Furthermore, hyperglycemia maystimulate VLDL production.27

Another possible mechanism is reduced adipose tis-sue lipoprotein lipase activity, which could restrain theremoval of triglyceride-rich particles3132 and lead to al-tered HDL cholesterol metabolism. In addition to reducedactivity of lipoprotein lipase, increased activity of hepaticlipase33 could contribute to an inverse association of HDLcholesterol and GDR. Our results, based on regressionanalyses, showed that GDR was inversely associated withVLDL triglyceride and directly with HDL cholesterol con-centrations independently of other confounding factorsincluding fasting insulin level (Table 5). Therefore, ourfindings are in agreement with the possibility that insulinresistance might induce abnormalities in lipid and lipopro-tein levels. Although HDL cholesterol and VLDL triglycer-ide metabolism are tightly linked, our findings may furtherindicate that HDL cholesterol and VLDL triglycerides haveseparate sites of interaction with insulin action.

A second mechanism is that abnormal lipid and lipo-protein concentrations could induce an impairment ininsulin action. In vitro and in vivo studies support thisnotion. In vitro studies3430-36 have shown that high con-centrations of VLDL result in an impairment of insulinaction, which is in part due to the inhibition of the glucosetransport system.38 On the other hand, a previous studyon normoglycemic subjects13 has reported that all mea-sured indices of insulin action (insulin-mediated glucoseuptake and oxidative and nonoxidative glucose metabo-lism) correlate significantly with the VLDL secretion rate.Other in vivo studies37-38 have demonstrated that anacute infusion of intralipid leads to the reduction ofinsulin-mediated glucose uptake. With respect to thelong-term effects of elevated VLDL concentration onglucose metabolism, it is interesting to note that theFramingham Study has shown that a high VLDL level is arisk factor for glucose intolerance.39 It remains unknownon the basis of in vivo studies whether the impairment in

by guest on July 10, 2017http://atvb.ahajournals.org/

Dow

nloaded from

230 ARTERIOSCLEROSIS VOL 10, No 2, MARCH/APRIL 1990

Table 6. Multiple Stepwise Linear Regression Analyses of Variables Associated with GlucoseDisposal Rate In All Study Subjects Irrespective of Glucose Tolerance Status

Independent variable

Age

Body mass Index

Waist/hip ratio

2-hr glucose

Free fatty acids

HDL cholesterol

VLDL triglycerides

R2 for the model (%)

Model A

Beta

-0.122

-0.366*

-0.090

-0.224*

-0 .161*

0.245t—

fl2(%)

0

23.2

0

11.1

2.2

4.0-

40.5

Model B

Beta

-0.084

-0.356*

-0.134

-0.229t

-0.147—

-0.2711

fl2(%)

0

11.7

0

4.2

0—

23.8

39.7

*p<0.05, tp<0.01, tp<0.001.Beta=standardlzed regression coefficient, R2<= proportion explained by a given Independent variable (%).Model A Includes HDL cholesterol as an independent variable. Model B includes VLDL triglycerides (logarithmic)

as an Independent variable.See legend to Table 3 for explanation of abbreviations.

insulin action is mediated through triglycerides or FFA.Our study demonstrated that a high VLDL triglyceridelevel was associated strongly with a low GDR indepen-dently of possible confounding factors (Table 6). Simi-larly, a low HDL cholesterol concentration and high FFAlevels were related to insulin resistance, although theseassociations were much weaker than those of VLDLtriglycerides and GDR. Therefore, it is possible that bothVLDL triglyceride and FFA levels can independentlyinfluence insulin action. No previous evidence exists toshow that low HDL cholesterol could induce an impair-ment in insulin-mediated glucose uptake.

The third mechanism that may explain the associationbetween insulin action and lipids and lipoproteins is thenotion that this association is caused by factors thatinfluence both glucose and llpid metabolism. Obesitycould be one of the most important factors causing thisassociation, since it may induce insulin resistance,40 in-crease VLDL triglyceride production,4142 and lower HDLcholesterol,4344 thus indirectly strengthening the associa-tion of adverse lipid and lipoprotein changes and Impairedinsulin action. However, we found that low HDL cholesteroland high total and VLDL triglycerides were similarly asso-ciated with low GDR in both obese (BMI&27.0 kg/m2) andnonobese (BMI<27.0 kg/m2) subjects. Furthermore, inregression analyses, BMI was not significantly associatedwith HDL cholesterol or VLDL triglycerides independentlyof other confounding factors (Table 5). In our study sub-jects, obesity was not an Important factor influencing bothglucose and lipid metabolism. Our study subjects wereequally obese in each glucose tolerance category, and therange of BMI was small. However, this does not excludethe possibility that, in other study subjects with a widerrange of BMI, obesity could be an important variableexplaining the association between insulin resistance andabnormal lipid and lipoprotein concentrations. In contrastto previous reports which have demonstrated that theabdominal distribution of obesity is correlated with im-paired insulin action45 and abnormal lipid and lipoproteinlevels,4647 we did not find a statistically significant associ-ation between the waist/hip ratio and GDR independentlyof other variables including BMI.

Although the mechanisms explaining the relationshipbetween insulin resistance and lipid and lipoproteinchanges remain unclear and could be due to any of thethree mechanisms discussed or their combinations, ourresults have implications in the understanding of an in-creased risk of atherosclerosis, particularly in NIDDM. Anincreased frequency of atherosclerotic events has beenreported in subjects with IGT, and at the time of diagnosis,the patients with NIDDM have more atherosclerotic com-plications than do nondiabetic subjects of correspondingages.2 Since subjects with IGT and NIDDM are insulin-resistant and since low insulin-mediated glucose uptake isassociated with unfavorable lipoprotein changes (highVLDL triglyceride and low HDL cholesterol levels) and withelevated blood pressure levels,48 insulin resistance shouldbe considered as an important potential risk factor foratherosclerosis. Prospective studies, however, are neededto confirm this association.

References1. Stout RW. The relationship of abnormal circulating Insulin

levels to atherosclerosis. Atherosclerosis 1977;27:1-132. Pyorala K, Laakso M, Uutltupa M. Diabetes and athero-

sclerosis: an epidemiological view. Diabetes Metab Rev1987;2:463-524

3. Pyorala K. Relationship of glucose tolerance and plasmainsulin to the incidence of coronary heart disease: resultsfrom two population studies in Finland. Diabetes Care1979;2:131-141

4. DudmetJoro P, Eschwoge E, Papoz L, Richard JL, ClaudeJR, Rosaelln G. Relationship of plasma Insulin levels to theincidence of coronary heart disease mortality in a middle-aged population. Diabetologia 1980:19:205-210

5. Walbom TA, Weame K. Coronary heart disease incidenceand cardiovascular mortality in Busselton with reference toglucose and insulin concentrations. Diabetes Care 1979;2:154-160

6. Pyorala K, Uusltupa M, Laakso M, Slttonan O, NlskanenL, RAnnemaa T. Macrovascular complications in relation tohyperinsullnemia in non-Jnsulin-dependent diabetes melli-tus. Diabetes Metab 1987:13:345-349

7. Hornlck CA, Fellmeth BD. High density lipoprotein choles-terol, Insulin and obesity In Samoans. Atherosclerosis 1981;39:321-328

by guest on July 10, 2017http://atvb.ahajournals.org/

Dow

nloaded from

INSUUN RESISTANCE AND UPOPROTEINS Laakso et al. 231

8. Stalder M, Pometta D, Suenram A. Relationship betweenplasma insulin levels and high density lipoproteln choles-terol levels in healthy men. Dlabetologla 1981;21:544-548

9. Orchard TJ, Becker DJ, Bates M, Kuller LH, Drash AL.Plasma Insulin and lipoprotein concentrations: an athero-genic association? Am J Epidemiol 1983;118:326-337

10. Zavaronl I, Dall'Agllo E, Alpl O, et al. Evidence for anindependent relationship between plasma Insulin and con-centration of high density lipoproteln cholesterol and triglyc-eride. Atherosclerosis 1985;55:259-266

11. Laakso M, Pyorfilfl K, Voutllalnen E, Mamleml J. PlasmaInsulin, and serum lipids and lipoprotelns in middle-agednon-insulin-dependent diabetic and non-dlabetlc subjects.Am J Epidemiol 1987;125:611-621

12. Oleftky JM. Insulin resistance and Insulin action. Diabetes1981:30:148-162

13. Abbott WHO, Ullloja S, Young AA, et al. Relationshipsbetween plasma lipoprotein concentrations and insulin ac-tion In an obese hyperlnsullnemlc population. Diabetes1987;36:897-904

14. Garg A, He4derman JH, Koffler M, Ayuso R, RosanstockJ, Raskin P. Relationship between lipoproteln levels and invivo Insulin action In normal young white men. Metabolism1988;37:982-987

15. Olefsky JM, Kolterman OG, Scarlett A. Insulin action andresistance In obesity and nonlnsulln-dependent type II dia-betes mellitus. Am J Physiol 1982;243:E15-E30

16. Laakso M, R6nnemaa T, Pyfirftla K, Kalllo V, Puukka P,PentUli I. Atherosclerotic vascular disease and Its riskfactors In non-insulin dependent diabetic and nondlabeticsubjects in Finland. Diabetes Care 1988; 11:449-463

17. Laakso M, Mykkanen L Pyoralft K. Prevalence of diabetesmellitus in elderly subjects In East Finland. Dtabetologia1987;30:195-196

18. World Health Organization Expert Committee on Diabe-tes Mellttus. Second report. Technical report series 646.Geneva: World Health Organization, 1980:11-13

19. DeFronzo RA, TobJn JD, Andres R. Glucose clamp tech-nique: A method for quantifying insulin secretion and resis-tance. Am J Physiol 1979;237:E214-E223

20. DeBodo RC, Steels R, Altszuler N, Dunn A, Bishop JS.On the hormonal regulation of carbohydrate metabolism:Studies with C14-glucose. Recent Prog Horm Res 1963;19:445-482

21. Flnegood DT, Bergman RN, Vranlc M. Estimation of en-dogenous glucose production during hyperinsulinemlc glu-cose clamps: comparison of unlabelled and labelled exog-enous glucose infusates. Diabetes 1987;36:914-924

22. Penttlla IM, Voutllalnen E, Laltinen P, Juutllalnen P.Comparison of different analytical and precipitation methodsfor the direct estimation of serum high-density lipoproteincholesterol. Scand J Clin Lab Invest 1981;41:353-360

23. Havel RJ, Eder HA, Bragdon HJ. The distribution andchemical composition of ultracentrifugally separated lipo-protelns In human serum. J Clln Invest 1955;34:1345-1353

24. Rtzza R, Cryer PE, Haymond MW, Gerich J. Adrenergicmechanisms for the effect of eplnephrlne on glucose produc-tion and clearance in man. J Clin Invest 1980;65:682-689

25. Stelner Q, Vranlc M. Hyperinsulinemia and hypertrigtycer-idemla: a vicious cycle with atherogenlc potential. Int JObesity 1982;6:117-124

26. Tobey TA, Greenfletd M, Kraemer F, Reaven GM. Relation-ship between insulin resistance, Insulin secretion, very lowdensity lipoproteln kinetics and plasma trigtyceride levels inrxxrnotriglyceridemic man. Metabolism 1981;30:165-171

27. Howard BV. Lipoproteln metabolism In diabetes mellitus. JLipld Res 1987;28:613-628

28. Durrlngton PN, Newton RS, Welnsteln DB, Steinberg D.Effects of Insulin and glucose on very low density lipoproteln

triglyceride secretion by cultured rat hepatocytes. J ClinInvest 1982;70:63-73

29. Patsch W, Franz S, Schonfeld G. Role of insulin In lipopro-tein secretion by cultured rat hepatocytes. J Clin Invest1983:71:1161-1174

30. Grundy SM. Pathogenesis of hypertriglyceridemia In man.In: Carlson LA, Olsson AG, eds. Treatment of hyperiipopro-teinemia. New York: Raven, 1984:21-34

31. Sadur CN, Yost TJ, Eckel RH. Insulin responsiveness ofadipose tissue lipoproteln lipase is delayed but preserved Inobesity. J Clin Endocrinoi Metab 1984;59:1176-1182

32. Pykallstd OJ, Smith PH, Brunzell JD. Determinants ofhuman adipose tissue lipoproteln lipase: Effect of diabetesand obesity on basal and diet induced activity. J Clln Invest1975:56:1108-1117

33. Applebaum-Bowden D, Haffner SM, Wahl PW, et al.Postheparin plasma triglyceride Upases: relationships withvery low density lipoproteln triglyceride and high densitylipoprotein 2 cholesterol. Arteriosclerosis 1985:5:273-282

34. Berliner JA, Frank HJL, Karaslc D, Capdevllle M.Upoprotein-induced Insulin resistance In aortic endothe-llum. Diabetes 1984;33:1039-1044

35. Bleger WP, Michel G, Barwlch D, Blehl K, Wlrth A.Diminished insulin receptors on monocytes and erythro-cytes In hypertriglyceridemia. Metabolism 1984;33:982-987

36. YM-Jaxvfrwn H, TasMnen M-R. Interrelationships among In-sulin's arrtilipotytk: and glucoregulatory effects and plasmatrlgtycerides In nondlabetic and diabetic patients with endog-enous hypertriglyceridemia Diabetes 1988;37:1271 -1278

37. Thlebaud D, DeFronzo RA, Jacot E, et al. Effect of longchain triglyceride infusion on glucose metabolism In man.Metabolism 1982;31:1128-1136

38. Bevllacqua S, Bonadonna R, Buzzlgoll G, et al. Acuteelevation of free fatty acid levels leads to hepatic Insulinresistance in obese subjects. Metabolism 1987:36:502-506

39. Wilson PW, McGee DL, Kannel WB. Obesity, very lowdensity lipoprotelns, and glucose intolerance over fourteenyears. Am J Epidemiol 1981 ;114:697-704

40. Kolterman OG, Insel J, Saekow M, Olefsky JM. Mecha-nisms of Insulin resistance in human obesity. Evidence forreceptor and postreceptor defects. J Clln Invest 1980;65:1272-1284

41. Grundy SM, Mok HYI, Zech L, Steinberg D, Berman M.Transport of very low density llpoprotein-triglycerides Invarying degrees of obesity and hypertriglyceridemia J ClinInvest 1979:63:1274-1283

42. Egusa G, Beta WF, Grundy SM, Howard BV. Influence ofobesity on the metabolism of apolipoprotein B In humans. JClln Invest 1985;76:596-603

43. Stre|a DA, Boyko E, RabkJn SW. Changes In plasmahigh-density lipoprotein cholesterol concentration afterweight reduction In grossly obese subjects. Br Med J1980;281:770-772

44. Brownell KD, Stunkard AJ. Differential changes in plasmahigh-density llpoprotein-cholesterol levels in obese menand women during weight reduction. Arch Intern Med 1981;141:1142-1146

45. Fujloka S, Matsuzawa Y, Tokunaga K, Tarul S. Contribu-tion of Irrtra-abdomlnal fat accumulation to the impairment ofglucose and lipid metabolism In human obesity. Metabolism1987;36:54-59

46. Foster CJ, Welnsler RL, Birch R, et al. Obesity and serumlipids: an evaluation of the relative contribution of body fatand fat distribution to lipid levels. Int J Obesity 1987;11:151-161

47. Anderson AJ, Sobodnskl KA, Freedman DS, BarborlakJJ, Rlmm AA, Gruchow HW. Body fat distribution, plasmalipids, and llpoproteins. Arteriosclerosis 1988;8:88-94

48. Ferrannlnl E, Buzzlgoll G, Bonadonna R, Glortco MA,Olegglnl M, Grazladel L, Pedrlneill R. Insulin resistance inessential hypertension. N Engl J Med 1987;317:350-357

Index Terms: insulin resistance • diabetes • lipids • lipoproteins

by guest on July 10, 2017http://atvb.ahajournals.org/

Dow

nloaded from

M Laakso, H Sarlund and L Mykkänenvarying degrees of glucose tolerance.

Insulin resistance is associated with lipid and lipoprotein abnormalities in subjects with

Print ISSN: 1079-5642. Online ISSN: 1524-4636 Copyright © 1990 American Heart Association, Inc. All rights reserved.

Avenue, Dallas, TX 75231is published by the American Heart Association, 7272 GreenvilleArteriosclerosis, Thrombosis, and Vascular Biology

doi: 10.1161/01.ATV.10.2.2231990;10:223-231Arterioscler Thromb Vasc Biol. 

http://atvb.ahajournals.org/content/10/2/223World Wide Web at:

The online version of this article, along with updated information and services, is located on the

  http://atvb.ahajournals.org//subscriptions/

at: is onlineArteriosclerosis, Thrombosis, and Vascular Biology Information about subscribing to Subscriptions:

  http://www.lww.com/reprints

Information about reprints can be found online at: Reprints: 

document.Permissions and Rights Question and AnswerFurther information about this process is available in theis being requested is located, click Request Permissions in the middle column of the Web page under Services.Clearance Center, not the Editorial Office. Once the online version of the published article for which permission

can be obtained via RightsLink, a service of the CopyrightArteriosclerosis, Thrombosis, and Vascular Biology Requests for permissions to reproduce figures, tables, or portions of articles originally published inPermissions:

by guest on July 10, 2017http://atvb.ahajournals.org/

Dow

nloaded from