1. Clinical Characteristics
Subject characteristics are shown in . There were significant differences between the male and female subjects for age and all anthropometric variables. Thus, the impact of these indices of insulin sensitivity, body weight and total and regional adiposity on serum lipids were analyzed separately for men and women.
Clinical characteristics of enrolled subjects.
2. Insulin Sensitivity Distribution
Following an overnight fast, an oral glucose tolerance test (OGTT) was performed. Blood glucose and serum insulin values at 0, 30, 60, 90, and 120 minutes following glucose challenge were employed to calculate an insulin sensitivity index (ISI) value for each subject (23
). ISI values distributed over a 7-fold range, with no significant difference in mean ISI values for men vs. women (10.6 ± 0.7 vs
. 11.6 ± 0.6, P
= NS) ().
Distribution of insulin sensitivity in non-obese subjects
3. Relationship between ISI and total and regional adiposity
To explore the impact of adiposity on the serum lipid profile, we assessed total and regional fat stores by several different methods. We selected percent total fat (total fat) and percent truncal fat (truncal fat) measurements obtained by DXA as the primary indices for total body fat and central fat stores. Truncal fat determined by DXA was highly correlated with total abdominal fat as measured by MRI (r = 0.875, P < 0.0001) and subcutaneous abdominal fat volume (r=0.866, p<0.0001). Truncal fat was significantly less predictive of visceral fat stores (r = 0.282, P = 0.20), suggesting that truncal fat is a more appropriate marker of total central fat than visceral fat. These relationships were not different between men and women. These data therefore support the use of truncal fat as an appropriate index of central adiposity in both sexes that is more accurate than common anthropometric measures. There were significant differences between the men and women subjects for age and measurements of generalized and regional stores ().
We assessed the contribution of total and regional adipose stores on insulin sensitivity by correlational analysis (). We observed that in men, higher values for indices of total and central adipose stores were associated with decreasing insulin sensitivity. ISI was negatively correlated with BMI (r = −0.39, P < 0.05), total fat (r = −0.41, P < 0.05), waist circumference (r = −0.48, P < 0.05), and truncal fat mass (r = −0.40, P < 0.05). In contrast, ISI values in women were not associated with any variables related to total or regional adiposity.
Correlations between indices of adiposity and ISI
4. The relationship between insulin sensitivity and serum lipid values adjusted for total and central adiposity
Due to the non-linear relationship between these variables, the effects of ISI and adiposity on serum lipids were examined across quartiles for these parameters. In men, ISI was significantly associated with TG (P < 0.01) when adjusted for age and truncal fat. Adjusted TG levels in the most resistant quartile were 1.49 fold higher than in the most sensitive quartile (P = 0.01) (). The relationship between ISI and TG was similar when adjusting for age and total fat, rather than truncal fat (P = 0.01).
Serum triglycerides are elevated in insulin resistant men but not women
In contrast, in women the association between ISI and TG levels was less pronounced. There was no significant effect of ISI on TG when adjusted for age and truncal fat (P
= 0.14). Adjusted triglyceride levels were 1.18 fold increased in the lowest vs. the highest quartile of ISI, but this difference did not reach statistical significance (P
= 0.15) (). While ISI had a smaller absolute influence on triglyceride levels in women compared to men, the interaction effect between gender and ISI on TG levels did not reach statistical significance (P
= 0.08) LDL subclass pattern B identifies small dense LDL particles. Pattern A is predominant large LDL particles and A/B is intermediate (25
). Low ISI values were associated with a more atherogenic LDL subtype in men but not women (). However, neither LDL, HDL, remnant, nor total cholesterol levels were significantly associated with ISI in either sex. Similarly, homocysteine, hsCRP levels were not significantly associated with differences in insulin sensitivity in either sex ().
Insulin resistance associated with more atherogenic LDL subtypes in men
5. The relationship between central and total adiposity and lipid values adjusted for insulin sensitivity
In contrast to ISI, there was a significant independent effect of adiposity on multiple serum lipid parameters in women when adjusted for age and ISI that was not observed in men. In men, adjusted triglyceride levels were not associated with total or truncal percent fat (P=0.59, P=0.68, respectively). Accordingly, adjusted triglyceride values were not different between the lowest and highest quartiles of truncal percent fat (1.17 fold increase, 95% CI: 0.81, 1.69, P = 0.40) (). Similarly, age and ISI-adjusted values for LDL cholesterol, HDL cholesterol, cholesterol remnants, total cholesterol, homocysteine, hs-CRP were not significantly associated with difference in total or truncal percent fat (data not shown). In women, however total and truncal percent fat were independent predictors of several serum lipid parameters. Quartile analysis by ANOVA indicated that, when adjusted for age and ISI, there was a significant association between truncal fat and TG (p=0.02), total cholesterol (P=0.04), LDL cholesterol (p=0.02), and cholesterol remnants (P= 0.01). Age and ISI-adjusted triglyceride levels were 1.42 fold increased in the highest truncal percent fat quartile compared to the lowest quartile, an effect that just missed statistical significance (95% CI: 0.99, 2.05) ().
Serum triglycerides not independently associated with truncal fat in men
Similarly, women in the highest quartile for truncal fat had age and ISI-adjusted LDL cholesterol levels 29 mg/dl higher than those in the lowest quartile (95% CI: 2.3, 55.7 mg/dl, P= 0.03). Interestingly, the increased LDL content accompanying increased truncal fat was associated with an increased prevalence of less atherogenic LDL subtypes (data not shown). However, age and ISI-adjusted levels of the atherogenic cholesterol remnants were 9 mg/dl higher in the highest truncal fat quartile (95% CI: 2.2, 16.4 mg/dl, P = 0.01).
The association of age and ISI-adjusted serum lipids with percent total fat in women followed a similar trend (data not shown), but was less marked; only cholesterol remnants were significantly independently associated with the quartile of total body fat (P=0.04).
6. Prevalence and determinants of cardiovascular risk in the non-obese population
We examined the prevalence of our cases where the lipid values exceeded thresholds for cardiovascular risk as defined by the World Health Organization (26
); Adult Treatment Panel III (27
) and International Diabetes Federation (28
). We found that these markers of CV risk were relatively rare in this population. In women, there were no cases of hypertriglyceridemia (>150mg/dl) or elevated homocysteine levels (> 10.4 µmol/L). There were two cases (3%) of high LDL (>160 mg/dl) and six cases (10%) each of low HDL (< 50mg/dl) and high hs-CRP (>3 mg/L). Chi square analysis of ISI quartiles indicates that the prevalence of these risk markers does not increase with insulin resistance in women. The prevalence of high LDL levels was significantly (P
= 0.05) associated with being in the highest quartile for truncal fat in women. The association of truncal fat with LDL risk, and the general low prevalence of cardiac risk markers in this these women with BMI < 27, underscores the critical role of adiposity in the incidence of hyperlipidemia in women.
In men, there was greater prevalence of several risk markers compared to women. There were 4 cases (9% incidence rate) of TG over 150 and elevated hs-CRP, 5 cases (11%) of elevated LDL cholesterol, 9 cases (20%) of elevated homocysteine and 10 (22%) cases of low HDL cholesterol.
In general high TG, and low HDL cholesterol incidence in men tended to be more prevalent in the insulin resistant group, but these trends did not meet statistical significance. The prevalence of LDL values >160 mg/dl was significantly greater in the most insulin resistant quartile (2/12) compared to the other quartiles combined (3/34) (P < 0.05). The highest quartile of truncal fat was associated with increased incidence of elevated hs-CRP (3/11 cases) (P < 0.05) but not hypertriglyceridemia, elevated LDL or homocysteine, or low HDL cholesterol levels.
Overall, in the absence of obesity there was minimal evidence for clustering of CV risk factors in this population. In men with HDL cholesterol levels below 40 mg/dl, the prevalence of hypertriglyceridemia was increased (3/4) compared to the men with normal HDL cholesterol (7/42) (P < 0.05). Low HDL was also associated with increased incidence of CRP > 3 mg/L (3/10 vs. 1/36) (P < 0.05).