According to U.S. code, all journals requesting payment of author page charges in order to defray the cost of publication are required to publish a disclaimer. This article must, therefore, be marked “advertisement” in compliance with U.S.C. Section 1734 solely to indicate this fact.
Both Subcutaneous and Visceral Adipose Tissue Correlate Highly with Insulin Resistance in African Americans
Article first published online: 6 SEP 2012
2004 North American Association for the Study of Obesity (NAASO)
Volume 12, Issue 8, pages 1352–1359, August 2004
How to Cite
Tulloch-Reid, M. K., Hanson, R. L., Sebring, N. G., Reynolds, J. C., Premkumar, A., Genovese, D. J. and Sumner, A. E. (2004), Both Subcutaneous and Visceral Adipose Tissue Correlate Highly with Insulin Resistance in African Americans. Obesity Research, 12: 1352–1359. doi: 10.1038/oby.2004.170
- Issue published online: 6 SEP 2012
- Article first published online: 6 SEP 2012
- Received for review November 13, 2003; Accepted in final form June 16, 2004
- abdominal obesity;
- insulin resistance complications;
- African American
- Top of page
- Research Methods and Procedures
Objective: The contribution of visceral adipose tissue (VAT) to insulin resistance is well-established; however, the role of subcutaneous abdominal adipose tissue (SAT) in insulin resistance remains controversial. Sex may determine which of these two components of abdominal obesity is more strongly related to insulin resistance and its consequences. The aim of this study was to determine whether both VAT and SAT contribute to insulin resistance in African Americans and to examine the effects of sex on this relationship.
Research Methods and Procedures: This was a cross-sectional study of 78 nondiabetic African-American volunteers (44 men, 35 women; age 33.8 ± 7.3 years; BMI 30.9 ± 7.4 kg/m2). VAT and SAT volumes were measured using serial computerized tomography slices from the dome of the diaphragm to the iliac crest. The insulin sensitivity index (SI) was determined from the minimal model using data obtained from the frequently sampled intravenous glucose tolerance test.
Results: In men, both VAT and SAT were negatively correlated with SI (r for both correlations = −0.57; p < 0.01). In women, the correlation coefficient between VAT and SI was −0.50 (p < 0.01) and between SAT and SI was −0.67 (p < 0.01). In women, the correlation coefficient for SI with SAT was significantly greater than the correlation coefficient with VAT (p = 0.02).
Discussion: Both SAT and VAT are strongly correlated with insulin resistance in African Americans. For African-American women, SAT may have a greater effect than VAT on insulin resistance.
- Top of page
- Research Methods and Procedures
Obesity, particularly abdominal obesity, is strongly associated with insulin resistance and its complications such as hypertriglyceridemia, low high-density lipoprotein (HDL)1-cholesterol, altered lipoprotein metabolism, elevated serum glucose, and hypertension (1, 2). Several studies have shown a strong association between visceral adiposity and insulin resistance and its metabolic consequences (3, 4, 5). However, other studies have shown that subcutaneous abdominal adipose tissue (SAT) may have an equally strong or even superior relationship to insulin resistance than visceral adipose tissue (VAT) (5, 6, 7).
The portal theory for the pathogenesis of the insulin resistance syndrome suggests that the syndrome is caused by increased delivery of free fatty acids to the liver from intraabdominal fat (8). According to this theory, free fatty acids are more easily mobilized from VAT and are delivered directly into the portal circulation resulting in altered hepatic metabolism and insulin resistance. Therefore, VAT should be a stronger determinant of the metabolic anomalies associated with insulin resistance than SAT. However, it has also been hypothesized that an increase in free fatty acid delivery to the liver from SAT could produce a similar effect on hepatic metabolism (9, 10). SAT in women is considerably larger than VAT; therefore, on the basis of size alone, SAT is likely to contribute to insulin resistance by the release of free fatty acids into the systemic circulation (9). Furthermore, systemic free fatty acids may be an especially important contributor to insulin resistance in the presence of upper abdominal obesity [defined as a waist-to-hip ratio (WHR) > 0.85] (10).
Due to well-described racial differences in central body fat distribution between VAT and SAT compartments, assumptions made in whites about the association between insulin sensitivity and VAT and insulin sensitivity and SAT need to be independently established in African Americans. For example, African-American men and women have lower VAT but greater insulin resistance than whites and Hispanic Americans (5, 11, 12, 13, 14, 15). Additionally, African Americans have greater SAT than whites (12, 16, 17). Therefore, it is especially important to examine the effect of SAT on insulin resistance in African Americans.
The aim of this study was to evaluate the relationship in African Americans among VAT, SAT, and insulin resistance. Additionally, we wanted to determine the clinical impact of the relationship among insulin resistance and each of these fat depots by examining the correlations of both VAT and SAT with the consequences of insulin resistance, specifically, lower HDL-cholesterol, higher triglycerides (TGs), increased low-density lipoprotein (LDL) particle number, small LDL and HDL particle size, higher systolic and diastolic blood pressure, and increased fasting glucose (18, 19, 20).
Research Methods and Procedures
- Top of page
- Research Methods and Procedures
Seventy-eight African-American men and women participated in a protocol to evaluate cardiovascular risk in normal healthy volunteers at NIH (Bethesda, MD). All participants were born in the United States and identified both parents as being of African descent. All women were premenopausal and were studied in the follicular phase of the menstrual cycle. The volunteers were recruited by advertisements in the local newsletters, posters, and the NIH web site. The study was approved by the Institutional Review Board of the National Institute of Diabetes and Digestive and Kidney Diseases. Informed consent was obtained from each subject.
Subjects made four visits to the Clinical Center over a 3-month period. At Visit 1, a history and physical examination were performed. Visits 2 to 4 occurred in the morning after the subject had fasted for 12 hours. At each of these visits, a fasting lipid profile was obtained. The lipid results reported for TG, HDL, and apolipoprotein B (apo B) are the mean of these three measurements. At Visit 2, a 2-hour, 75-gram oral glucose tolerance test (Trutol 75; Custom Laboratories, Baltimore, MD) was performed, with glucose determined at 0 and 2 hours. Blood was also collected for LDL particle size and number and HDL particle size. At Visit 3, an abdominal computerized tomography (CT) scan and DXA were obtained. At Visit 4, an insulin-modified frequently sampled intravenous glucose tolerance test was performed.
The Frequently Sampled Intravenous Glucose Tolerance Test
Intravenous lines were placed in each antecubital vein. The arm used for phlebotomy was wrapped in a heating blanket. At time 0, glucose (0.3 g/kg) was injected over 1 minute. Insulin (4 mU/kg per minute) was administered from 20 to 25 minutes. Samples were taken at: −10, −1, 0, 1, 2, 3, 4, 5, 6, 7, 8, 10, 12, 14, 16, 20, 22, 23, 24, 25, 27, 30, 40, 50, 60, 70, 80, 90, 100, 120, 150, and 180 minutes. Glucose and insulin levels were measured in all samples. Glucose was determined at the bedside, and blood for insulin determination was drawn into serum separator tubes.
Anthropometric measurements were performed according to standard procedures (23). Body weight was measured to the nearest 0.1 kg using a platform digital scale, and standing height was obtained to the nearest 0.1 cm using a stadiometer. The circumference measurements were performed by a single observer (N.G.S.) using a nonstretch tape. The waist circumference was measured just above the anterior iliac crest. The hip circumference was measured at the level of maximum extension of the buttocks. Circumference measurements were performed three times and the mean value reported.
Visceral and Subcutaneous Adipose Tissue Measurement
CT was performed with a HiSpeed Advantage CT/i scanner (GE Medical Systems, Milwaukee, WI) and analyzed using the MEDx image analysis software package (Sensor System, Inc., Sterling, VA) on a SUN workstation. VAT and SAT volumes were measured with images from 10-mm contiguous slices from the dome of the diaphragm to the iliac crest. Each subject had 20 to 25 slices performed, with the exact number dependent on height. A cursor was placed on the abdominal wall, which allowed the tracing of the periphery of the abdominal cavity to be automatically performed using edge detection techniques. The perimeter of the visceral cavity below the subcutaneous fat was then traced manually. A density mask inclusive of pixels with attenuation values ranging from −150 to −50 Hounsfield units (which includes adipose tissue without including other soft tissue, bone, and air) was used to measure the areas. The system calculated the visceral and subcutaneous fat areas separately for each slice. The results from the contiguous slices were added to give the VAT and SAT volumes.
Percentage Body Fat Measurement
Percentage body fat was determined with a Hologic QDR 4500A DXA (Hologic, Inc., Bedford, MA) in the array mode using software version 5.71A. Five subjects (three men and two women) who were above the weight limits for the DXA machine (136 kg) had their percentage fat determined from a regression equation derived in this population (24) using the percentage body fat calculation (Weight Manager software, version 2.2, RJL Systems, Detroit MI) from the bioelectrical impedance measurement (Quantum II BIA Analyzer, RJL Systems).
Laboratory Analytic Methods
Glucose was measured using the glucose oxidase method (Glucostat; Yellow Springs Instrument, Yellow Springs, OH). Insulin assays were performed using the double antibody chemiluminescent sandwich assay (Diagnostic Products, Los Angeles, CA). TG was analyzed using standard enzymatic methods by an automated analyzer (Boehringer Mannheim/Hitachi 917, Roche Diagnostic, Indianapolis, IN). HDL-cholesterol was isolated according to the method of Warnick (25). Coefficients of variation for the TG and HDL-cholesterol assays were 1.5% and 2.2%, respectively. Apo B was assayed turbidimetrically using commercial antibodies (Beckman kit/Beckman Array; Beckman Coulter, Fullerton, CA). The LDL particle number and the LDL and HDL particle diameters were determined by LipoScience (Raleigh, NC) using nuclear magnetic spectroscopy (26, 27). The technique involves measurement of the plasma nuclear magnetic resonance spectrum for each lipoprotein particle followed by computer deconvolution of the spectral data and calculation of the subclass fractions (27). Normal values for LDL particle number and size of LDL and HDL particles are not yet established. Based on studies performed in the Framingham Offspring Study, the normal mean ± SD for the LDL particle number is 1483 ± 432 nM and for LDL and HDL particle size are 20.9 ± 0.5 and 9.2 ± 0.5 nM, respectively (J.D. Otvos, personal communication).
Unpaired Student's t tests were used to detect sex differences. Variables without a normal distribution were log transformed before this analysis. The Spearman rank correlation coefficient was used to examine the relationship among the obesity measurements (VAT and SAT), SI, and the lipid and glucose consequences of insulin resistance. To determine whether the difference in the correlation coefficients among SI, VAT, and SAT was significant, a z statistic was generated (28). This statistic was derived in each sex from the correlation coefficients among SI, VAT, and SAT, as well as the correlation coefficient between VAT and SAT. Using a 95% confidence limit, the difference between coefficients was significantly different if the z statistic was either less than −1.96 or more than 1.96 (corresponding to p < 0.05). The correlation coefficients between VAT and SAT and each of the consequences of insulin resistance were similarly compared using this method.
The WHR was used to explore whether upper abdominal obesity (WHR > 0.85) might modify the relationship between VAT or SAT and SI. An interaction term consisting of the product of WHR and VAT or SAT was included in sex-specific regression models containing age and using SI as the dependent variable. The likelihood ratio (LR) test was used to compare models with the main effects and the interaction term with models containing the main effects alone. All statistical tests were considered significant for p ≤ 0.05. Analysis was performed using SAS (version 8, SAS Corporation, Cary, NC).
- Top of page
- Research Methods and Procedures
The anthropometric and metabolic characteristics of the 78 subjects (44 men and 34 women) are presented in Table 1. Fourteen men (32%) and five women (15%) had glucose intolerance. All analyses were performed including and excluding the glucose intolerant subjects. Because the results did not vary significantly, the findings from the analysis of the combined population are presented in this paper.
|Men (n = 44)||Women (n = 34)||p value|
|Age (years)||35.1 ± 7.9||32.1 ± 6.2||0.07|
|BMI (kg/m2)||30.6 ± 6.4||30.9 ± 8.6||0.97|
|Percentage body fat||22.9 ± 6.5||36.9 ± 9.0||<0.01|
|Waist circumference (cm)||99.1 ± 16.2||96 ± 16||0.37|
|WHR||0.90 ± 0.07||0.86 ± 0.06||0.01|
|VAT volume (mm3)||1580 ± 1030||939 ± 776||<0.01|
|SAT volume (mm3)||4085 ± 2960||6237 ± 4563||0.02|
|SI (L/mU per minute)||3.52 ± 1.98||3.77 ± 2.86||0.71|
|Fasting insulin (pM)||55.1 ± 11.0||63.3 ± 40.5||0.30|
|Fasting glucose (mM)||5.0 ± 0.6||4.6 ± 0.5||<0.01|
|Cholesterol (mM)||4.5 ± 1.0||4.4 ± 1.0||0.56|
|HDL-cholesterol (mM)||1.21 ± 0.25||1.39 ± 0.26||<0.01|
|TGs (mM)||0.93 ± 0.33||0.66 ± 0.38||<0.01|
|Apo B (g/L)||0.89 ± 0.27||0.83 ± 0.25||0.34|
|LDL particle number (nM)||1253 ± 410||1250 ± 422||0.98|
|LDL particle size (nm)||21.0 ± 0.6||21.3 ± 0.6||0.04|
|HDL particle size (nm)||8.8 ± 0.5||9.2 ± 0.4||<0.01|
|Systolic blood pressure (mm Hg)||119 ± 11||112 ± 14||0.01|
|Diastolic blood pressure (mm Hg)||70 ± 10||69 ± 8||0.65|
|Number with IGT (%)||14 (32%)||5 (15%)||0.11|
The BMI of the subjects studied ranged from 21.9 to 54.7 kg/m2 in men and 20.9 to 49.7 kg/m2 in women. There were no sex differences in age, waist circumference, or BMI. Women had a higher percentage body fat (p < 0.01) and a lower WHR than men (P = 0.01). Women also had lower VAT (p < 0.01) and higher SAT (p = 0.02) than men.
Despite the sex differences in WHR, VAT, and SAT, the mean SI of the men and women studied was similar (p = 0.71). The mean fasting glucose, systolic blood pressure, and TG were all significantly higher in men than women; however, there was no sex difference in the proportion of men or women with impaired glucose tolerance. There were no sex differences in LDL particle number and apo B; however, women had a higher HDL and LDL particle size compared with men. VAT and SAT were highly correlated with each other in men (r = 0.82; p < 0.01) and women (r = 0.84; p < 0.01).
Sex-Specific Correlations for VAT and SAT with Insulin Resistance and Its Consequences
The sex-specific correlation coefficients for VAT and SAT with insulin resistance and its consequences are presented in Table 2. In African-American men, VAT and SAT were negatively correlated with SI and HDL particle size. VAT and SAT were positively correlated with TG and fasting glucose. VAT but not SAT was also positively correlated with LDL particle number and apo B. There was no correlation between blood pressure and either VAT or SAT.
|Variable||SI||TG||HDL||HDL particle size||LDL particle size||LDL particle number||apo B||Fasting glucose||Systolic blood pressure||Diastolic blood pressure|
In African-American women, both VAT and SAT were negatively correlated with SI, HDL-cholesterol, HDL particle size, and LDL particle size. VAT and SAT were positively correlated with TG, LDL particle number, apo B, and the fasting glucose. Positive correlations were noted between systolic blood pressure and VAT and SAT in women. Neither VAT nor SAT was correlated with the diastolic blood pressure.
Comparison of Sex-Specific Correlation Coefficients for VAT and SAT
Figure 1 presents by sex the relationship between SI and VAT and SI and SAT. In men, when correlation coefficients for VAT vs. SI and SAT vs. SI were compared, neither measurement was more highly correlated statistically with SI (difference in r = 0; p = 0.50). In women, SAT was more highly correlated with SI than VAT (difference in r = 0.17; p = 0.02). In African-American men and women, no significant difference in the correlation coefficients for VAT and SAT with the glucose and lipid consequences of insulin resistance was detected (data not shown).
Insulin Sensitivity, VAT, SAT, and Upper Body Obesity
The presence or absence of upper body obesity (assessed by WHR) did not modify the relationship between either VAT or SAT and SI. In women, the LR test from the inclusion of an interaction term between WHR and VAT or SAT was not significant (p for LR test for regression models with and without either the VAT-WHR or SAT-WHR interaction term were 0.85 and 0.83, respectively). In men, interaction terms for VAT, SAT, and SI were also not significant (p for LR test for difference in model with inclusion of VAT-WHR interaction term was 0.49 and 0.70 for the SAT-WHR interaction term). There was no improvement in the regression models when the waist circumference was used instead of WHR to generate the interaction terms in both sexes.
- Top of page
- Research Methods and Procedures
SAT and VAT volumes were both highly correlated with SI in African Americans. In men, the correlation coefficients for SAT with SI and VAT with SI were similar, whereas in women, the correlation between SAT and SI was significantly greater than the correlation between SI and VAT. Furthermore, SI was similar in men and women, despite the significant sex differences in WHR, SAT, and VAT. In women, both VAT and SAT were correlated with most of the consequences of insulin resistance: higher TG, LDL particle number, fasting glucose, and systolic blood pressure, lower HDL, and smaller LDL and HDL particle size. In men, both VAT and SAT were correlated only with higher TG and fasting glucose and smaller LDL particle size.
Insulin Resistance and Central Body Fat Distribution
The medical literature suggests that both SAT and VAT contribute to insulin resistance. Abate et al. (6), in a study of middle-aged men, showed that the SAT had a stronger correlation with the glucose disposal rate than VAT did. Goodpaster et al. (7), in a study of men and premenopausal women, concluded that even though both VAT and SAT areas had strong correlations with insulin sensitivity, SAT had an independent effect on insulin sensitivity.
In contrast, other investigators have found stronger associations between the VAT and insulin resistance compared with SAT (4, 29, 30, 31). Cnop et al. (31), in a study of men and pre- and postmenopausal women, concluded that the VAT (compared with SAT, BMI, and WHR) was the strongest predictor of SI. When lean and obese insulin resistant subjects from this study were matched for sex and VAT area, there was no difference in SI among the subjects. The absence of a difference in SI among subjects matched for VAT area has been demonstrated in other studies (29, 30), whereas matching subjects for SAT area has not been shown to remove differences in SI (29, 30). None of these studies described the effect of race on their findings.
Banerji et al., in a study of 52 African Americans (32 men and 20 women, mean age 48 and 54 years, respectively) with type 2 diabetes, found a significant correlation between VAT and insulin-mediated glucose disposal, a measure of insulin sensitivity (32). In contrast to our study, the correlation between SAT and insulin-mediated glucose disposal was not statistically significant in either sex despite similarities in the technique used to measure intraabdominal fat. This study differs from ours in a number of areas. We studied a younger population of African Americans, all of the women were premenopausal, and none of the subjects had type 2 diabetes. The significant heterogeneity among our study populations may explain the differences in our conclusions.
Similar to our work, Lovejoy et al. (5) found a significant association between SI and both VAT and SAT in African-American women. However, these investigators found that in African-American women, both VAT and SAT areas were equally correlated with the SI. However, we determined that although SI in African-American women was associated with VAT and SAT, the association between SI and SAT was significantly greater than the association between SI and VAT. There may be a methodological reason to account for why we found a relatively greater association between SI and SAT than Lovejoy et al. First, we had a more optimal measure of VAT and SAT. Specifically, we measured total VAT and SAT volume, whereas the other investigators had access to only a single CT area measurement at L4 to L5. Having only area measurements may have reduced the ability to detect a difference in the correlation between VAT and SAT with SI. To compare our data with Lovejoy et al., we looked at the CT slice at L4 to L5 and calculated the areas for VAT and SAT. When we did this, we also found that the correlation coefficients for SI with VAT and SAT areas at L4 to L5 were similar (data not shown).
Importantly, the results of our study are consistent with the findings of the recently published Insulin Resistance Atherosclerosis Study Family Study (15). This population, which consisted of African Americans and Hispanic Americans, also demonstrated that, independent of ethnicity and sex, both SAT and VAT areas had a negative association with SI. In fact, the SI was lowest in the subjects with the largest VAT and SAT areas.
In our study, upper abdominal obesity, assessed using the WHR, did not modify the relationship of either VAT or SAT with SI in African-American women. This suggests that in African-American women, SAT contributes to SI regardless of the amount of upper abdominal obesity. In African-American men, SAT and VAT contributed equally to SI regardless of upper abdominal obesity.
Insulin Resistance and Disorders of Glucose and Lipoprotein Metabolism
In subjects of a variety of racial backgrounds, disorders of glucose and lipoprotein metabolism have been strongly associated with insulin resistance (19, 33). Most studies that have examined the association between the metabolic consequences of insulin resistance and VAT and SAT have concluded that VAT is the more important contributor (34, 35, 36). However, in our study of African Americans, both VAT and SAT volumes were found to be highly correlated with fasting glucose, HDL particle size, and serum TGs. When these correlation coefficients were compared, there was no significant statistical difference among them. VAT and SAT were not correlated with all of the complications of insulin resistance in African-American men. However, in African-American women, both VAT and SAT were equally correlated with all of the consequences of insulin resistance (with the exception of diastolic blood pressure). Therefore, we suggest that in African Americans, SAT has an important role in the pathogenesis of insulin resistance and its metabolic consequences. Our findings also suggest that SAT may be of potentially greater importance to insulin resistance than VAT in African-American women.
We are unaware of any other metabolic study of African Americans that has examined the relationship between intraabdominal fat and lipoprotein particle size and number. The sexual dimorphism in the relationship between VAT, SAT, and LDL particle number and apo B is of particular interest. In men, the VAT but not SAT was correlated with these variables. This suggests a stronger relationship between VAT and dyslipoproteinemia than SAT and dyslipoproteinemia in African-American men despite both abdominal fat depots being equally correlated with SI. It would be interesting to see whether this relationship is replicated in other studies of African-American men.
The strong correlation between VAT and SAT presents a significant problem if one attempts to simultaneously compare the effect of these measurements on any dependent variable in a regression model or by using partial Spearman analysis because of colinearity (37). To avoid this difficulty, we quantified the correlation of both VAT and SAT with SI and consequences of insulin resistance, and we assessed the statistical significance of the difference among the correlation coefficients using the method suggested by Olkin (28).
Other advantages of our study include the availability of direct measurements of VAT and SAT volumes (superior to area measurements performed at a single lumbar vertebrae) and the ability to examine the effect of VAT and SAT on some of the newer markers of insulin resistance. Kelley et al. (38) has demonstrated the importance of divisions of subcutaneous adipose tissue to insulin resistance in lean and obese glucose-tolerant men and women. Given the importance of SAT to SI in African Americans, the contribution of deep and superficial subcutaneous fat to insulin resistance and its metabolic consequences will need further exploration in future studies.
In African Americans, both SAT and VAT contribute to insulin resistance and its complications. In African-American women, because insulin resistance is more highly correlated with SAT than VAT, SAT may be an even greater contributor to the development of insulin resistance than VAT. Therefore, further investigations are needed to determine the metabolic significance of SAT in African Americans, particularly African-American women.
- Top of page
- Research Methods and Procedures
We thank all of the study participants, Aramesh Saremi for her assistance with the data analysis, and Jonathan Krakoff for his thoughtful review of the manuscript. This study was supported by intramural funding from the National Institute of Diabetes and Digestive and Kidney Diseases, NIH.
Nonstandard abbreviations: HDL, high-density lipoprotein; SAT, subcutaneous abdominal adipose tissue; VAT, visceral adipose tissue; WHR, waist-to-hip ratio; TG, triglyceride; LDL, low-density lipoprotein; apo B, apolipoprotein B; CT, computerized tomography; SI, insulin sensitivity index; LR, likelihood ratio.
- Top of page
- Research Methods and Procedures
- 102002) Adipose tissue and fatty acid metabolism in humans. J R Soc Med. 95(suppl): 3–7. 42.(
- 18Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults. (2001) Executive Summary of the Third Report of The National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, And Treatment of High Blood Cholesterol In Adults (Adult Treatment Panel III). JAMA. 285: 2486–2497.
- 231998) Anthropometric Standardization Reference Manual. Human Kinetics Books: Champaign, IL., , (
- 281967) Correlation revisited. In Stanely, JC (ed). Improving Experimental Design and Statistical Analysis. Rand McNally: Chicago, IL.(
- 371998) Regression Diagnostics: Applied Regression Analysis and Multivariable Methods. Duxbury Press: Pacific Grove, CA. 212–280., , , (