Relationship of Abdominal Visceral and Subcutaneous Adipose Tissue With Lipoprotein Particle Number and Size in Type 2 Diabetes

Subjects for the current analysis were Caucasian and African-American participants in the CHICAGO trial, a prospective study of the effects of pioglitazone compared with glimepiride on carotid intima-media thickness in subjects with type 2 diabetes recruited from 28 clinical sites in Chicago (31). The details of the study have been previously reported (31,32). Data included in this report were obtained before randomization to treatment groups. All subjects were asymptomatic for coronary artery disease at baseline. The study was approved by central and local institutional review board committees, and all participants provided written informed consent. All subjects underwent measurements of height, weight, and waist and hip circumference by a trained nurse at the baseline visit. Waist circumference was measured at the smallest circumference between the ribs and iliac crest, and hip circumference was measured at maximum circumference between the iliac crest and crotch to the nearest 0.1 cm. BMI was calculated as weight in kilograms divided by the square of height in meters.

Subjects underwent an abdominal CT scan for determination of VAT and SAT. Abdominal adipose tissue content and distribution were quantified by CT scan at the level of L4-L5 vertebra when subjects were in supine position with both arms stretched above the head (33–35). A single 6-mm slice was taken during suspended respiration after a normal expiration. Total abdominal adipose tissue (TAT) area was measured by delineating the body surface with a receiver operator instrument (ROI) and then computing the adipose tissue volume using an attenuation range of −190 to −30 HU (33–35). VAT area was quantified by delineating the abdominal cavity at the internal aspect of the abdominal wall and the posterior aspect of the vertebral body with an ROI (33–35). SAT was calculated by subtracting VAT from TAT area. To obtain VAT and SAT volumes, the area for each fat component was multiplied by the slice thickness. Fasting blood samples were obtained at the baseline visit for measurement of lipids, A1C, and lipoprotein profile (31,32). Lipoproteins were analyzed using NMR technology by LipoScience (Raleigh, NC) (1).

Homeostatis model assessment insulin resistance (HOMA-IR) was calculated according to the following formula: [fasting glucose (mmol/l) × fasting insulin (U/ml)]/22.5. Log transformation of the data was performed when necessary to achieve homogeneity of variance. Pearson correlation coefficients were used after adjustment for age and sex to assess the relationship between each subclass lipoprotein particle number and size and BMI, waist circumference, A1C, SAT, and VAT. Each subclass lipoprotein particle number and size was further examined in relation to BMI, SAT, and VAT areas using a general linear model with sex, statin use, and diabetes therapy as fixed variables and age, duration of diabetes, years of smoking, and A1C levels as continuous variables. The multivariable analyses were repeated with the addition of BMI to the models when assessing the relation of VAT or SAT to lipoprotein number and size or with the addition of both VAT and SAT to the model when assessing the relation of BMI to lipoprotein parameters. To evaluate the impact of insulin resistance on these relationships, HOMA-IR was further added to the models. The associations between VAT and lipoprotein particle number and size were further examined by multivariable models after addition of SAT or of hip circumference to models that included BMI. Similar analyses were performed to evaluate the association of SAT to lipoprotein parameters before and after adjustment for BMI or for BMI and VAT. Analyses were performed for the entire group and then separately for statin-treated and untreated subjects. The relation of A1C to BMI, VAT, and SAT was examined using a general linear model with sex, statin use, and diabetes therapy as fixed variables and age, duration of diabetes, and years of smoking as continuous variables. Additional models were performed after the addition of BMI and BMI plus VAT and SAT. General linear modeling was also used to examine the relation of non-HDL cholesterol to BMI, VAT, and SAT areas with sex, statin use, and diabetes therapy as fixed variables and age, duration of diabetes, years of smoking, and A1C levels as continuous variables. Analyses were performed using the 11.0 PC package of SPSS statistical software (SPSS, Chicago, IL). A P < 0.01 was considered significant to adjust for evaluation of multiple lipoprotein parameters (VLDL, LDL, and HDL particle number and size) for their relation to VAT (which was our primary analysis) and other adiposity measures (as secondary analyses).

The baseline characteristics of study subjects are presented in Table 1. The mean age was 61 years. Thirty-eight percent of subjects were women, 55% were on statin therapy, and 65% were current or former smokers. Subjects were on the following diabetes therapies at the time of participation in the study: 15% of subjects were not taking any medications for diabetes, 15% were taking sulfonylureas alone, 29% were taking metformin alone, 31% were taking a combination of metformin and sulfonlyureas, and 10% were on insulin therapy. The average BMI was 32.5 kg/m², the mean duration of type 2 diabetes was 92 months, and the mean A1C was 7.4%. Mean HDL cholesterol was 1.2 nmol/l, LDL cholesterol 2.8 nmol/l, and triglyceride 1.8 nmol/l. In age- and sex-adjusted correlations, VAT was positively associated with VLDL particle number, LDL particle number, and VLDL size and negatively associated with LDL size and HDL size (Table 2). BMI and SAT were not associated with particle size or number for any lipoprotein species (Table 2). Waist circumference was negatively associated with LDL and HDL size. A1C was positively associated with VLDL and LDL particle number and negatively associated with LDL size (Table 2).

The results from multivariable regression models are shown in Table 3. After adjustment for age, sex, diabetes therapy, duration of diabetes, statin use, A1C, and smoking years, VAT was positively associated with VLDL and LDL particle number and VLDL particle size and negatively associated with LDL and HDL size. These associations remained significant after adjustment for BMI. Addition of SAT with BMI to the model did not change the strength of the associations (data not shown). SAT was not associated with lipoprotein particle size or number before or after adjustment for BMI (Table 3) or after adjustment for BMI and VAT (data not shown). BMI was borderline associated with VLDL and HDL particle number, and the borderline association with HDL particle number persisted after adjustment for SAT and VAT (Table 3). Addition of HOMA-IR did not change the associations between VAT and VLDL particle number or size or VAT and LDL or HDL size (Table 4).

It has been suggested that lower-body subcutaneous fat may have beneficial effects on insulin sensitivity and cardiometabolic risk. We therefore added hip circumference, as an index of lower-body subcutaneous fat mass, to the multivariable model for the relationship of VAT to lipoprotein particle number and size. After addition of hip circumference to the multivariable model, we noted a 4–10% increase in regression coefficient for the models examining the relation between VAT and each lipoprotein particle number and size (data not shown).

Because of the potent effect of statins on lipoprotein metabolism, we assessed the relationship between VAT and lipoprotein parameters in statin users and nonusers separately (Table 5). The association between VAT and LDL particle number was only present among statin nonusers, but VAT was strongly and significantly related to VLDL particle number and size and LDL and HDL particle size among statin users (Table 5).

The data in Table 2 show that glycohemoglobin level was significantly associated with VLDL and LDL particle number and LDL size. After adjustment for age, sex, baseline diabetes therapy, duration of diabetes, smoking years, and statin use, A1C level remained positively associated with VLDL particle number and LDL particle number and negatively associated with LDL particle size. These significant associations remained intact after addition of BMI or after addition of BMI, VAT, and SAT to the model (data not shown). In multivariable models, neither BMI nor SAT was a significant predictor of non-HDL cholesterol (not shown). VAT had a borderline significant association with non-HDL cholesterol; P = 0.02, r² = 0.15, 0.03 (0.02–0.04) mmol/l increase for each 10 cm³ increase in VAT.

In this population of middle- and older-aged men and women with type 2 diabetes, higher VAT was strongly associated with changes in lipoprotein particle number and size (1,2). These associations were independent of overall adiposity as measured by BMI and were independent of SAT content. The associations persisted after adjustment for HOMA-IR as a measure of insulin resistance for all of the lipoprotein parameters except for LDL particle number. The associations were strongest in subjects who were not on statin therapy but remained significant for VLDL particle number and size and LDL and HDL size in subjects who were treated with statins. We did not observe associations between SAT and lipoprotein particle number or size. We observed a borderline negative association between BMI and HDL particle number, as higher BMI was associated with a decrease in HDL particle number. Our data suggest that in type 2 diabetes, adverse changes in lipoprotein particle number and size are most strongly related to accumulation of visceral fat rather than overall adiposity.

Whereas the harmful impact of central fat on insulin sensitivity and metabolic disorders is well accepted (15), the individual contribution of VAT and SAT to metabolic risk remains uncertain (15–17). Visceral fat is considered to have greater lipolytic activity compared with subcutaneous fat and has favored access to the liver through the portal vein. Thus, it has been proposed that the high free fatty acid (FFA) flux from visceral fat may reduce hepatic insulin sensitivity, favor hepatic fat accumulation, and thereby promote an atherogenic lipid profile (36). However, a number of studies have suggested that SAT may have as strong or even stronger deleterious impact on insulin sensitivity and metabolic risk than VAT (28–30,37). SAT comprises a larger fat depot than visceral fat (38) and contributes ∼75% of the total FFA to the peripheral circulation (39,40). Other studies have suggested a less important role for SAT compared with VAT (20,27,41). In a recent report, surgical removal of SAT in obese subjects did not result in metabolic improvements or beneficial changes in cardiovascular risk factors (42). In our study, SAT was not associated with the changes in lipoprotein particle size or number typically observed with insulin resistance. Unlike most previous studies, however, our study focused exclusively on subjects with type 2 diabetes, a large proportion of whom were obese (20,27–29). The relationship between central fat depots and metabolic risk could be modified in diabetes either by the more severe degree of insulin resistance in type 2 diabetic compared with obese nondiabetic subjects or by the failure of insulin secretion to compensate for the metabolic derangements produced by insulin resistance (43,44). Increasing evidence indicates that adipose tissue, especially VAT, is the source of a number of hormones (45), cytokines, and inflammatory factors (46) that can impact substrate flux and lipid metabolism in distant tissues. It is possible that this secretory pattern is altered by the presence of type 2 diabetes (47). Recently, an association between small LDL particles and plasminogen activator inhibitor-1 in subjects with type 2 diabetes has been shown to be related to VAT (48).

The role of adipose tissue distribution in insulin sensitivity and metabolic risk extends beyond abdominal fat. Increases in hepatic and skeletal muscle fat have also been associated with insulin resistance (30,49). Furthermore, lower body fat has been shown to at least partially counteract the influence of abdominal fat and protect against insulin resistance (50). Peripheral body fat, mainly stored in subcutaneous thigh and gluteal regions, has lower lipolytic activity compared with central fat and may serve as a metabolic sink by taking up excess circulating FFA and even preventing ectopic fat accumulation (51). Recent data indicate that waist and hip circumference have opposite associations with coronary artery disease; the risk for developing coronary artery disease increased with waist circumference, but the risk was lowered by increasing hip circumference (52). A recent study has shown that subjects with type 2 diabetes had less leg fat mass and greater liver and trunk fat mass compared with similarly obese subjects without type 2 diabetes (30). Interestingly, in our study adjustment for hip circumference in multivariable analyses tended to strengthen the associations between VAT and atherogenic lipoprotein particle number and size parameters, suggesting that higher fat content in the gluteal-femoral area could mitigate the negative impact of visceral fat on lipoprotein metabolism. However, overall changes were small, and we did not perform imaging studies to quantify lower body fat depots.

Studies using NMR technology to analyze subclass lipoprotein profile have demonstrated that progressive insulin resistance is associated with an increase in VLDL size and large VLDL particle concentration, a decrease in LDL size reflecting marked increase in small LDL particles and a reduction in large LDL, an overall increase in the number of LDL particles, and a decrease in HDL size as a result of reduction of large HDL particles and a modest increase in small HDL (1). These alterations in lipoprotein particle number and size are considered to be atherogenic and to predispose to CVD (3–11). In this study, we demonstrate that these unfavorable changes in lipoprotein profile were related to increasing VAT independent of overall adiposity or of SAT, suggesting that VAT has serious negative consequences on lipoprotein metabolism leading to increased risk for CVD in subjects with type 2 diabetes. Statin therapy abolished the relation between VAT and LDL particle number but was not able to abolish the relationship between VAT and VLDL particle number or between VAT and VLDL, LDL, and HDL particle size. These data could indicate that statin therapy only partially addresses lipoprotein-related CVD risk in subjects with type 2 diabetes. Residual excess CVD risk in subjects with diabetes who have been treated with statins has been observed (53) and may be addressed by therapies that impact residual lipoprotein abnormalities. Our data also emphasize that the adverse effects of VAT on lipoprotein parameters are not completely reversed by statin therapy, underlining the importance of interventions that produce weight loss, particularly in the visceral fat depot.

Strengths of our study include inclusion of a large sample of well-characterized subjects with type 2 diabetes, the use of NMR technology for determination of subclass lipoprotein profile, and the use of CT scanning to quantitate VAT and SAT. We were also able to adjust for a number of potential confounders, including smoking history, sex, and statin use. With respect to limitations, our study does not permit firm conclusions regarding causality. In addition, we do not have measures of overall truncal, hepatic, or lower-extremity fat that could have provided additional information on the association between body fat distribution and lipoprotein particle number and size. Our study included only diabetic subjects, and relationships may be different in those without diabetes. In addition, our subjects had BMIs that clustered in the obese range, as is typical for type 2 diabetes. The influence of adipose tissue distribution on lipoprotein parameters could be different in those with lower BMI.

In summary, increasing VAT independent of BMI and SAT was associated with an atherogenic lipoprotein profile in subjects with type 2 diabetes. In contrast, in our study we were unable to show an association between SAT or BMI and lipoprotein parameters. The data suggest that the atherogenic lipoprotein profile associated with type 2 diabetes is related to VAT accumulation. Prospective studies will be needed to provide more information regarding the causal nature of these associations.

T.M. has received National Institutes of Health Grant DK-71711. The CHICAGO study was sponsored and funded by Takeda Global Research & Development. This work has been supported by an institutional award from the University of Illinois at Chicago.