Effects of Insulin Resistance and Obesity on Lipoproteins and Sensitivity to Egg Feeding

Insulin resistance is a condition of impaired effectiveness of insulin. Insulin resistance is common¹ and is a major predictor of coronary artery disease.^2–5 Obesity and insulin resistance are highly associated,^1,6,7 but insulin resistance can be present in subjects meeting the criteria for being normal weight and overweight.^1,7,8

Because insulin resistance and obesity exaggerate cardiovascular disease risk, one might hypothesize that these 2 conditions would exaggerate the cholesterol-raising effect of a high saturated fat diet. Surprisingly, investigations on the influence of insulin resistance on the lipoprotein response to diet are few, indirect, and somewhat contradictory. In our Dietary Alternatives Study, men with elevated plasma insulin and triglyceride levels and low HDL cholesterol (HDL-C) (all associated with insulin resistance) had diminished LDL-C–lowering response to a low-fat diet.^9,10 In keeping with these observations, Beynen and Katan¹¹ found greater reductions in serum cholesterol with removal of egg sources of cholesterol from the diet in persons with low body mass index (BMI) and high HDL-C levels likely to be insulin sensitive. In contrast, subjects with combined hyperlipidemia fed 2 eggs per day for 12 weeks in 2 different studies had greater increases in LDL-C than persons with simple hypercholesterolemia.^12,13 The only prospective evaluation of the effect of egg feeding on LDL levels in subjects in whom insulin sensitivity was quantified by the insulin clamp found no effect of insulin resistance but also little LDL rise with egg feeding.¹⁴

To address these diverging lines of evidence, we studied 197 healthy individuals dichotomized based on insulin sensitivity and BMI. Based on the published relationship of insulin resistance to BMI,¹ midpoint values of BMI <27.5 and ≥27.5 kg/m² and insulin sensitivity quantified as the insulin-sensitivity index (S_I) of ≥4.2 and <4.2 (10⁻⁴ min⁻¹ μU/mL) were used to divide study subjects into insulin-sensitive (IS), insulin-resistant (IR), and obese insulin-resistant (OIR) groups. A fourth group of IS, obese subjects was too small to study because of the curvilinear relationship of insulin sensitivity to BMI.^1,8 We sought to determine if the LDL response to egg feeding was exaggerated or diminished by insulin resistance alone or insulin resistance combined with obesity.

Methods

Study Design

The study was a double-blinded, randomized, 3-period crossover clinical trial. It consisted of 3 1-month intervention periods with intervening 1-month washout periods during which study subjects continued to follow the Step I diet. The intervention consisted of daily ingestion of 0, 2, or 4 eggs per day or an equivalent placebo, as described in the diet section below. Subjects were randomized to the order in which they would receive the egg preparation.

Study Subjects

Two hundred seventy-nine healthy subjects responding to public advertising were invited to participate. Subjects were excluded if they had cholesterol >300 mg/dL and triglyceride >500 mg/dL, diabetes, blood pressure >150/100 mm Hg, renal, liver, or unstable thyroid disease (stable thyroid hormone use allowed), or coronary or peripheral vascular insufficiency or anemia. Use of lipid-altering medications was prohibited, including lipid-lowering medications, β-blockers, thiazides, corticosteroids, oral contraceptives, and anticonvulsants. Continuous postmenopausal hormone replacement therapy was allowed. Women with an irregular menstrual period, exceeding an interval of 24 to 32 days, were excluded. Ingesting more than 2 alcoholic drinks daily, history of mental instability or mental illness, or inability to comply with study requirements were also exclusionary. The study was approved by the University of Washington Human Subjects Review Committee, and all subjects gave written informed consent before participating.

Study Visits

Subjects made 10 visits over a 7-month period. Visit 1 comprised a study orientation, consent form signing, and teaching the National Cholesterol Education Program (NCEP) Step I diet.¹⁵ At visit 2, vital signs, body weight and height, lipoprotein profile, and a fingerstick glucose measurement were obtained. A fingerstick value >110 mg/dL confirmed by a laboratory value >115 mg/dL was exclusionary. A derived lipoprotein quantification was obtained.¹⁵ At visits 2 and 3, each subject’s evaluated dietary intake had to be within Step I guidelines to proceed. At visit 4, a frequently sampled, tolbutamide-modified intravenous-glucose tolerance test (FSIGT) and abdominal CT scan were performed.^16,17

Subjects were randomized to the order of intervention before visit 5, the baseline visit. At visit 5, vital signs, waist to hip ratio, diet evaluation, fasting glucose, insulin, lipoproteins, and apoprotein (apo) B were measured. Vertical angle spin for LDL size was performed, and DNA was stored.

Vital signs and lipoprotein measurements were obtained, and dietary intake was monitored with 3-day food records at the remaining 5 visits, which occurred at the beginning and end of the 4-week intervention periods. One month separated each of the intervention months. Of the 279 subjects screened, 12 were obese and IS and were excluded because of their small number. We randomized 221 subjects, and 197 completed all feeding sequences. Twenty-four subjects dropped out, primarily for changes in personal life (n=9), failure to attend visits on time (n=6), and intolerance to the egg preparation (n=5).

Diet

Egg or egg substitute preparations consisted of homogenized natural eggs or Egg Beaters prepared in the metabolic kitchen of the University of Washington General Clinical Research Center. Egg placebo weighed 108 g, as did the 2 egg per day preparation, which consisted of 34 g egg yolks, 64 g Egg Beaters, and 10 g water. The 4 egg per day preparation consisted of 68 g egg yolks, 20 g Egg Beaters, and 20 g water. The 0, 2, and 4 egg per day preparations contained 45, 171, and 298 calories, 0, 10, and 20 g fat, and 0, 425, and 850 mg cholesterol, respectively. Daily portions of egg preparation were provided frozen, sufficient for 1 month, to be eaten with ad libitum food choices at home within Step I guidelines.

Three-day food records kept at 6 visits were used for counseling on the NCEP Step I diet. Dietary intake was reviewed by a registered dietitian and entered for analyses using the Nutrition Data Systems database.

The FSIGT

The FSIGT was performed as described by Beard et al¹⁶ and modified by the reduction of the tolbutamide dose to 125 mg/m² and extension of the protocol to 240 minutes. The minimal model of glucose kinetics of Bergman et al¹⁸ was used. Day to day reproducibility of S_I is 16.9%.¹⁹

Abdominal Fat by Computed Tomography

A CT image of the abdomen at the level of the umbilicus was obtained according to the method of Shuman and Fujimoto.¹⁷ From this scan, the cross-sectional area of fat was determined using a density-contour program available in the standard GE 8800 computer software, as previously described.¹⁷ The error of reproducibility of the area measurements is 1.5%.¹⁷

Plasma Analyses

The derived lipoprotein quantification was determined using the Friedewald equation and measurement of fasting total triglyceride, total cholesterol, and HDL-C. Plasma apo B levels were analyzed using immunonephelometry as standardized by Marcovina and Albers.²⁰ A vertical angle profile of LDL cholesterol was obtained in all subjects at baseline and after the first egg feeding regimen (visit 7), meaning that one third of subjects in each group had this assessment after egg feeding. This technique uses nonequilibrium density gradient ultracentrifugation, collecting 38 fractions in which cholesterol was measured.²¹ Results are expressed as a percent of the total cholesterol in the 38 fractions. The relative flotation rate of LDL, a measure of LDL particle buoyancy, was determined by dividing the number for the fraction containing the peak LDL-C concentration by 38. Plasma glucose was measured using a coupled glucose oxidase method, and plasma insulin was measured by the method of Morgan and Lazarow.²²

Statistics

The sample size of 67 subjects per group was based on a significance level of 0.05, power of 0.80, and a mean±SD difference of 9.3±17.6 mg/dL LDL-C observed in our previous egg-feeding study in hyperlipidemic adults.¹² Percent change in the lipoprotein values was calculated on measurements taken at the beginning and end of each egg-feeding period. Descriptive statistics include mean±SDs and group percentages. One-way ANOVA was used to test for statistically significant differences among the 3 groups, and the Games-Howell test,²³ which adjusts for unequal sample sizes and group variances, was used to make multiple pairwise comparisons. Baseline items with outliers or marginally normal distributions in 1 or more groups (triglycerides, LDL-C, non-HDL-C, insulin, BMI, and insulin sensitivity) were also tested using nonparametric methods (Kruskal-Wallis and the rank-sums multiple pairwise comparison tests).²⁴ Differences in the 2 testing approaches were minor, and results are presented as ANOVA and Games-Howell analyses. χ² tests were used to test for differences for categorical data. Tests on baseline triglyceride and dietary cholesterol data were performed on log normalized values. Significance testing was 2-sided.

Results

Baseline Characteristics

Subject characteristics are presented in Table 1. The IS group was 5 to 6 years younger than the other 2 groups. A slight majority of the subjects in all groups were women. Body weight was incrementally higher in the 3 groups, with the greatest increment in the OIR group over the IS and IR groups. BMI, by definition, was only marginally higher in IR over IS subjects. Intra-abdominal, subcutaneous, and total abdominal fat areas were higher incrementally and statistically in the 2 IR groups (all P<0.001), with the greatest increments from IR to OIR. Systolic and diastolic blood pressures were also progressively higher across the 3 groups. The IS subjects were more frequent aerobic exercisers.

All groups were within the limits of the NCEP Step I diet of 30% of total calories (en%), with fat intakes of 24.9, 26.9, and 27.8 en%. Likewise, saturated fat intake was within guidelines and clustered about halfway between Step 1 and Step 2 guidelines but was significantly lower in IS than IR (P<0.05) and OIR groups (P<0.01). The amount of cholesterol in the baseline diets was 186, 192, and 258 mg/d in the respective groups, OIR being significantly higher than IS and IR.

Plasma glucose and insulin concentrations were progressively higher across the 3 groups, with OIR significantly greater than IS and IR groups. Correspondingly, insulin sensitivity was lower by definition in IR and OIR groups, but the OIR group was still significantly lower than the IR subjects.

Lipoprotein levels are shown in Table 1 and Figure 1. Total plasma cholesterol, LDL-C, apo B, and non-HDL-C concentrations were similarly and significantly higher in IR and OIR compared with IS subjects. There were no statistical differences between IR and OIR groups for total cholesterol, LDL-C, non-HDL-C, or apo B. The pattern of LDL-C persisted after adjustment for age and sex (P=0.001) and age, intra-abdominal fat area, total and saturated dietary fat, and dietary cholesterol intake (P=0.006). In contrast, plasma triglyceride levels showed a stepwise increase across the 3 groups whereas HDL-C levels and LDL buoyancy were lower stepwise across the 3 groups (Table 1 and Figure 1).

• Download figure
• Download PowerPoint

Lipoprotein Response to Egg Feeding

The plasma lipoprotein responses to egg ingestion, expressed as a percentage change from baseline, are shown in Table 2 and as concentration change from baseline in Figure 2 (4 eggs per day). Total plasma cholesterol levels were increased significantly by ingesting 4 eggs per day for 1 month in all 3 groups, but most in the IS group. A lower triglyceride level was seen in the IS group fed 4 eggs per day.

• Download figure
• Download PowerPoint

LDL-C increased significantly from baseline in the IS and IR but not in the OIR groups (7.8%, 3.3%, and 2.4%, respectively). Calculated increments above the 0 egg control period were 6.8%, 5.5%, and 0.3%, respectively. Absolute increases in LDL-C showed similar trends, as follows: 0.21±0.40, 0.093±0.45, and 0.060±0.41 mmol/L (8.3±15.6, 3.6±17.4, and 2.3±15.9 mg/dL) (Figure 2). Analysis of covariance revealed no significant effect of baseline differences in LDL-C levels or the intake of cholesterol or saturated and polyunsaturated fat in the background diet to these increments in LDL-C.

The effect of egg feeding on LDL density is depicted as a plot of differences in the percent of total cholesterol in each of 38 density fractions (Figure 3) after egg feeding for 1 month minus the baseline profile. IS subjects fed 0 eggs for 1 month showed no change in the buoyancy distribution of LDL or any other lipoprotein fraction from baseline. However, IS subjects fed 2 and 4 eggs per day (data pooled) showed an increase in buoyant LDL (fractions 19-12) and a decrease in IDL (fractions 30-20), as demonstrated by the error bars clearing 0 in the difference plots. Subjects in the IR and OIR groups showed a similar tendency toward increased buoyant LDL (fractions 19-12) and decreased small-dense LDL (fractions 11-7) for all feeding periods, possibly related to the increased intake of egg protein.

• Download figure
• Download PowerPoint

The HDL-C percent change with egg feeding shows the same trend as in LDL and with statistically significant increases of 8.8%, 5.2%, and 3.6% in IS, IR, and OIR groups, respectively, fed 4 eggs per day.

Discussion

This investigation makes 2 main points. First, LDL-C is meaningfully and similarly elevated (0.47 to 0.59 mmol/L or 18 to 23 mg/dL) in persons with insulin resistance regardless of obesity status, whereas triglycerides are higher and HDL-C levels are lower with insulin resistance and more so with mean BMI in the obese range. These results differ from most other accounts of the effect of insulin resistance and obesity on lipoproteins, which emphasize the abnormalities in triglyceride and HDL.^2–5,25,26 However, a rising LDL-C trend was previously reported in NHANES III,²⁷ with increasing BMI to a maximum of 27.1 kg/m². At higher BMIs, LDL-C was not higher, whereas triglyceride and HDL became progressively more abnormal.²⁷ The same lipoprotein trends were seen in a previous examination of baseline data from the present study across a similar sample and across 4 quartiles of S_I and abdominal obesity.⁸ Mechanistically, insulin-resistance could elevate LDL-C by impairing LDL receptor–mediated LDL removal.²⁸

An alternative mechanism for the greater LDL-C level in insulin resistance and obesity is a greater intake of cholesterol and fat. However, baseline cholesterol intake was similar in nonobese subjects, regardless of insulin resistance status, and saturated fat intake was only 1 en% higher in IR over IS, although statistically significant. On the other hand, intakes of both cholesterol and saturated fat were higher in OIR than in IR subjects, but LDL-C was not different. Neither of these factors nor age had important effects on baseline LDL-C level in multivariate analysis.

Another potential mechanism for the elevated LDL-C level in insulin resistance and obesity is heightened sensitivity to dietary cholesterol and saturated fat. This hypothesis is addressed by the second main result of the study. IR and obese subjects were less sensitive to the LDL-raising effect of high cholesterol and saturated fat intake in the form of 4 eggs per day for 1 month, disproving the hypothesis. This result seems counterintuitive to the notion that insulin resistance and obesity aggravate cardiovascular disease risk.^2–5 However, our results are consistent with previous research. In the study of Beynen and Katan,¹¹ lean subjects with elevated HDL-C and HDL₂-C levels had a greater LDL-C response to a reduction cholesterol feeding compared with obese persons with lower HDL levels. Similarly, in our long-term study of increasing degrees of fat restriction on LDL (the Dietary Alternatives Study), men with simple hypercholesterolemia had a greater LDL decrease than men with elevated plasma cholesterol, triglyceride, insulin, and body weight at 3 levels of fat intake.^9,10 A similarly diminished LDL responsiveness to NCEP Step I diet has also been seen in obese compared with nonobese young men in Spain²⁹ and in fatter versus leaner men in the Chicago Western Electric Study.³⁰

Why are LDL and HDL cholesterol levels in lean subjects more sensitive to egg feeding than in IR and OIR subjects? A plausible explanation is found in recent studies of cholesterol absorption by Miettinen and colleagues.^31–33 They observed decreased cholesterol absorption in IR states^31,32 and overt diabetes.³³ When overweight persons underwent a weight loss diet, their cholesterol absorption increased.³¹

Like the LDL-C response, the increase in HDL-C in IS subjects with egg feeding is also consistent with an intestinal etiology. Intestinal apo A-I formation is increased with cholesterol and fat feeding,^34,35 and plasma apo A-I levels increase with egg feeding.¹² The slightly greater and more statistically robust percentage increase in HDL-C relative to LDL-C is also in keeping with an intestinal regulation of the HDL-C rise with egg feeding.

The use of egg yolk as a dietary perturbation is not strictly a test of the effect of cholesterol feeding alone. Four eggs per day comprise approximately 3 en% of saturated fat in the daily diet, representing the difference in saturated fat goals between the NCEP Step I and Step II diets of 10 and 7 en%, respectively.³⁶ For this reason, the results may be generalized from egg feeding to the general condition of high fat and cholesterol intake in IS, IR, and obese subjects.

The significance of the egg-induced increment in LDL-C concentrations for atherosclerosis risk is uncertain. The LDL-C rise is accompanied by an equal or greater percent increase in HDL-C level in all 3 groups of subjects. Additionally, the LDL-C rise in the most responsive, IS group was in the buoyant LDL fraction, a less atherogenic particle. Finally, a significant LDL-C increase was observed only in IS subjects and with 4 eggs daily, an uncommonly high intake in daily living. The LDL-C rise may be limited by the low background intake of saturated fat in these subjects who followed a diet pattern between Step I and II, an effect seen previously.^12,37 Although the mean LDL-C increases observed here are less than reported in meta-analyses,^37,38 the dispersion around the mean indicates that LDL-C will still increase or even decline with egg feeding in some subjects.

These results differ from the observations of Reaven et al,¹⁴ who did not see a significant increase in LDL-C with a cholesterol-rich diet or an effect of insulin-resistance in an all-female study group. Their smaller subject number and maximum level of cholesterol intake of ≈950 mg/d versus our ≈1200 mg/d maximum at a level of 4 eggs per day plus background diet cholesterol intake may explain the lack of LDL-C differences. More difficult to explain is the greater LDL-C rise with 2 eggs per day for 12 weeks that we⁹ and Garber et al¹³ observed in subjects with combined hyperlipidemia compared with simple hypercholesterolemia. However, both studies were conducted in persons with established hyperlipidemia. This distinction underscores the complexity of genetic factors impinging on lipoprotein regulation.

In conclusion, the dyslipidemias of insulin resistance and obesity include an elevation of LDL-C as well as elevated triglyceride and low HDL-C levels. These abnormalities seem to be inherent to insulin resistance and not attributable to dietary intake, because LDL-C response to dietary cholesterol and saturated fat is diminished in IR states. Because diminished absorption of cholesterol from the intestine is a likely mechanism of reduced dietary sensitivity to egg feeding, diet composition seems to be less important in the management of IR and obese individuals than caloric restriction itself. Thus, optimal diet therapy for such persons may best emphasize total calorie restriction, allowing a greater proportion of fat to improve satiety and the prospect of successful weight loss.

Footnotes