Ethnicity-specific anthropometric predictors of metabolic risk in women

Abstract

Objective

The objective of this study was to determine associations of anthropometric measures of thigh and abdominal adipose tissue with metabolic risk factors, and whether these associations differed with ethnicity. We hypothesized that thigh circumference (ThC) would have an independent favorable association with insulin sensitivity, lipids, and blood pressure, whereas waist circumference (WC) would have an independent deleterious association with these variables in both African Americans (AA) and European Americans (EA).

Methods

Subjects were 228 healthy, overweight, premenopausal AA and EA women. Insulin sensitivity was assessed by intravenous glucose tolerance test and minimal modeling. Simple relationships between anthropometric measures and risk factors were determined by Pearson correlation analysis. Partial correlation coefficients were determined for circumference measures adjusted for thigh and abdominal skinfolds to differentiate relationships between thigh and abdominal subcutaneous fat from thigh muscle and deeper abdominal fat, respectively.

Results

In EA but not AA, ThC was positively associated with insulin sensitivity, independent of thigh skinfold. In both EA and AA, ThC was associated with a desirable lipid profile. In AA but not EA, WC was associated with lower insulin sensitivity and a less desirable metabolic profile.

Conclusion

Results suggest that thigh muscle (ThC adjusted for thigh skinfold) may be metabolically protective in EA but not AA. In contrast, WC was a better indicator of insulin sensitivity and metabolic health in AA. Further investigation is needed to verify the association between thigh muscle and metabolic health, and to probe the reason for the observed ethnic specificity of the associations between anthropometric measures and metabolic risk factors.

Introduction

Differences in body fat patterning have long been associated with metabolic disease risk (1). Increased abdominal fat deposition, particularly intra-abdominal adipose tissue (IAAT), has been linked to increased risk for type 2 diabetes and cardiovascular disease (2–4). However, increased peripheral fat deposition, particularly thigh fat, appears to have an inverse relationship to disease risk (5–10).

Advanced techniques such as dual-energy X-ray absorptiometry (DXA) and computed tomography (CT) are often used to assess body fat distribution. However, these sophisticated and costly techniques are not always feasible, particularly in large-scale population studies. Simple anthropometric measures are often preferable for a number of reasons, including their non-invasive nature, low cost, lack of radiation exposure, and ease of administration. In addition, anthropometric measures such as waist circumference (WC) and thigh circumference (ThC) have shown strong associations with disease risk factors (11–13). Indeed, elevated WC is associated with dyslipidemia, higher fasting insulin, lower insulin sensitivity, and elevated blood pressure and is considered part of the diagnostic criteria for metabolic syndrome (14–17). Although WC is thought to be a fairly accurate indirect measure of IAAT (18), other studies suggest it is a better reflection of subcutaneous abdominal adipose tissue [SAAT, (19)].

Research regarding the protective effects of preferential thigh fat deposition, assessed by ThC, are not as prevalent as those focused on the deleterious effects of excess abdominal fat. Regardless, a few studies show that ThC is related to a reduced risk for type 2 diabetes and a more desirable metabolic profile including lower triglycerides (TG), lower low-density lipoprotein cholesterol (LDL-C), and higher high-density lipoprotein cholesterol (HDL-C) (20;21). These relationships are maintained even after controlling for measures of abdominal obesity. In addition, studies also show that thigh fat deposition, assessed via ThC, DXA, or CT, is related to more desirable glucose and insulin levels taken from either fasting values or oral glucose tolerance test (OGTT) measures (22–25). However, the relationship between ThC and insulin sensitivity assessed via direct measurement such as with frequently sampled intravenous glucose tolerance testing (FSIGT) and minimal modeling has not been determined.

Ethnic differences among African Americans (AA) and European Americans (EA) with regards to metabolic risk factors as well as body composition and fat partitioning have been noted by several investigators. AA have lower insulin sensitivity (26), higher systolic and diastolic blood pressure (27), lower triglyceride and higher HDL-C concentrations compared to their EA counterparts (28). Some studies indicate that AA have lower IAAT, higher SAAT, and higher fat-free and lean tissue mass than EA (26;29–31). In addition, WC may provide a better estimate of IAAT in AA compared to EA (32;33). Ethnic differences in the relationship between metabolic risk factors and circumference measures, particularly ThC, in AA and EA have not been fully explored.

The aim of this study was to assess the relationships of ThC and WC with metabolic risk factors including insulin sensitivity assessed by FSIGT and minimal modeling in both AA and EA premenopausal women and to determine if these relationships differed by ethnicity. We tested the hypotheses that, 1) ThC would be independently associated with a more favorable metabolic profile including greater insulin sensitivity in both AA and EA, and 2) WC would be independently associated with a less favorable metabolic profile including lower insulin sensitivity in both AA and EA. Because circumference measures include both subcutaneous fat and non-fat tissue (34), we also tested these hypotheses using thigh and abdominal skinfold measures, which reflect only subcutaneous fat, in order to differentiate the relationship between subcutaneous thigh fat and thigh muscle and between SAAT and deeper abdominal adipose tissue depots such as IAAT or deep SAAT.

Methods and Procedures

Subjects

Study participants were 228 healthy, overweight AA (52%) and EA (48%) women aged 25 – 45 years who were originally recruited as part of a weight loss study. All subjects for whom baseline data (prior to any weight loss) were available were included in this current study. Participant’s ethnicity was self-reported and included both parents and grandparents having the same AA or EA ancestry. All subjects were nonsmokers, sedentary (defined as <1 exercise session per week over the past year), premenopausal with no known history of irregular menstrual cycles, overweight (BMI 27–30 kg/m²), and normoglycemic. Subjects were not taking medications known to affect metabolism or body composition. All testing was performed during the follicular phase of the menstrual cycle for all women. Data was obtained during an inpatient stay at the General Clinical Research Center (GCRC) at the University of Alabama at Birmingham (UAB). The study was approved by UAB’s Institutional Review Board for Human Use, and all subjects were consented before any testing was performed. We certify that all applicable institutional and governmental regulations concerning the ethical use of human volunteers were followed during this research.

Protocol

Prior to baseline testing, all subjects were required to follow a weight maintenance diet for 4 weeks with food provided by the GCRC during weeks 3 and 4. To ensure weight maintenance, subjects were weighed 3–5 times a week, and their caloric intake was adjusted as needed. At the end of the weight maintenance period, subjects were admitted for an overnight stay at the GCRC where anthropometric measurements, insulin sensitivity, fasting lipid concentrations, TG, total cholesterol (TC), HDL-C, LDL-C, total cholesterol:HDL ratio (TC:HDL-C), and systolic and diastolic blood pressure (SBP and DBP, respectively) were assessed.

Insulin sensitivity

Insulin sensitivity was assessed by an insulin-modified FSIGT, which accounts for both insulin-stimulated glucose uptake and insulin-inhibited hepatic glucose production (35). Testing was performed in the morning after a 12-hour overnight fast. Three blood samples were collected over a 40-minute period prior to glucose administration. These samples were separated, pooled, and stored for subsequent lipid analysis. An additional three blood samples were obtained over a 20-minute period prior to glucose administration for assessment of basal insulin and glucose concentrations. These values were averaged to obtain basal “fasting” levels of insulin and glucose. At minute “zero”, an intravenous injection of glucose (50% dextrose; 11.4 g/m²) was administered. Eleven blood samples were then collected from minutes “2” to “19”. Twenty minutes after glucose injection, an intravenous bolus of insulin (0.02 units/kg) was administered. Eighteen additional blood samples were then collected from minutes “22” to “180”. The time points for all blood samples collected during the FSIGT are relative to the minute “zero” glucose injection. Insulin and glucose values were determined for all time points. These values were used for the determination of insulin sensitivity via minimal modeling (version 3.0, ©Richard N. Bergman, (36;37).

Anthropometrics

Subjects were weighed (Scale-tronix 6702W; Scale-tronix, Carol Stream, IL) to the nearest 0.1 kg in minimal clothing without shoes. Height was recorded without shoes using a digital stadiometer (Heightronic 235; Measurement Concepts, Snoqualmie, WA). For all subjects, anthropometric measurements including ThC, WC, thigh skinfold, and abdominal skinfold were obtained by the same registered dietitian during their inpatient stay at the GCRC. Both ThC and thigh skinfold were measured halfway between the inguinal crease and the proximal patella border (38;39). WC was measured between the ribs and the iliac crest at the “narrowest part of the torso” or what is considered the natural waist (38). Abdominal skinfold was measured 3 cm over (lateral) and 1 cm down (inferior) from the midpoint of the umbilicus (39). Both circumference measures were obtained using a flexible tape measure (Gulick II; Country Technology, Inc., Gays Mills, WI) and were recorded to the nearest 0.1 cm. Both skinfold measures were obtained using Lange Skinfold Calipers (Beta Technology Incorporated, Cambridge, MD) and were recorded to the nearest mm.

Blood pressure, insulin, glucose and lipid assays

Blood pressure (Dinamap Pro 200 automated cuff, GE Medical Systems) was measured after subjects were awake and in the supine position. Three daily measurements were obtained and averaged to yield final SBP and DBP values.

All samples were analyzed at the Core Laboratory of the GCRC and Clinical Nutrition Research Center at UAB. Insulin was measured in duplicate using Linco Research Products Inc. (St. Charles, MO) reagents. In this Core Laboratory, the assay sensitivity is 3.35μIU/ml. The mean intra and inter-assay coefficient of variation are 3.49% and 5.57%, respectively. Variation over time is monitored by including low, medium and high insulin concentrations from commercial quality control sera in every assay. The Ektachem DT II System (Johnson and Johnson Clinical Diagnostics) was used to measure glucose, TC, HDL-C, and TG. Glucose was measured in 10 μl of sera and the mean intra and inter-assay coefficient of variation in the Core Laboratory are 0.61% and 1.45%, respectively. HDL-C was determined after the precipitation of LDL-C and very-low-density lipoprotein cholesterol via dextran sulfate and magnesium chloride. For glucose and lipids, control sera with both low and high substrate concentrations were measured and values shown to be within acceptable ranges for each group of samples analyzed. The Ektachem DT II System is calibrated every six months using manufacturer supplied reagents. The Friedewald equation was used to derive LDL-C concentrations (40).

Statistics

Because there are ethnic differences in body fat distribution (26;41;42) and insulin sensitivity (26) among AA and EA, analyses for each group were performed separately. Demographic data and descriptive statistics (percents and mean ± SD) were determined for variables of interest. Two-group t-tests were used to determine differences in descriptive statistics for AA and EA. Pearson correlation analysis was used to assess the simple relationships between anthropometric measures and metabolic variables for each ethnic group. Partial correlation coefficients for each ethnic group were calculated to determine the independent associations between ThC and WC measures and metabolic variables adjusting for the corresponding skinfold measure (i.e. thigh or abdominal skinfold). This was performed in an effort to disentangle the relationship between thigh subcutaneous fat and thigh muscle as well as between SAAT and IAAT or deep SAAT. Likewise, partial correlation coefficients for each ethnic group were calculated to determine the independent associations between thigh and abdominal skinfold measures and metabolic variables adjusting for the corresponding circumference measure (i.e. ThC or WC). Partial correlation coefficients for HDL-C and circumference and skinfold measures were also adjusted for TG due to the close mechanistic link between the two (43). However, adjustment for TG did not alter the significance of the results; therefore, the TG-adjusted data are not presented.

All metabolic variables (insulin, glucose, insulin sensitivity, lipids, and blood pressure) were log₁₀ transformed to ensure normality of distribution. Preliminary analysis indicated the presence of two outliers (>3 standard deviations above the mean) for fasting insulin, which were consequently excluded from all subsequent analyses. Not all variables were available on all subjects. Sample sizes for individual variables ranged from 100 to 118. All statistical tests were two-tailed with a 5% significance level. All analyses were performed using SAS software package (version 9.1; SAS Institute, Cary, NC).

Results

Baseline demographic and descriptive information are presented in Table 1. Results from two-group t-tests indicate that EA had significantly higher insulin sensitivity, fasting glucose, triglycerides, TC:HDL-C ratio, WC, and abdominal skinfold thickness and significantly lower HDL-C, DBP, and ThC compared to AA.

Table 1.

Demographic and descriptive statistics by ethnicity

Variable	European American (n=109)	African American (n=119)	P1
Age	34.34 ± 6.56	34.13 ± 5.92	NS
BMI (kg/m2)	28.20 ± 1.46	28.37 ± 1.34	NS
Insulin Sensitivity [× 10−4min−1/(μIU/ml)]	3.49 ± 1.82	2.51 ± 1.89	<0.001
Fasting Insulin (μIU/ml)	11.25 ± 3.29	12.05 ± 4.08	NS
Fasting Glucose (mg/dl)	88.50 ± 6.26	86.44 ± 6.44	<0.05
Triglycerides (mg/dl)	115.16 ± 53.20	71.43 ± 27.85	<0.001
Total Cholesterol (mg/dl)	159.62 ± 30.92	156.16 ± 33.08	NS
HDL-C (mg/dl)	36.05 ± 9.23	42.92 ± 11.04	<0.001
LDL-C (mg/dl)	100.54 ± 27.28	98.96 ± 30.63	NS
Total Cholesterol:HDL-C Ratio	4.69 ± 1.41	3.84 ± 1.16	<0.001
Systolic Blood Pressure (mmHg)	112.49 ± 8.33	114.48 ± 9.55	NS
Diastolic Blood Pressure (mmHg)	62.39 ± 6.39	65.52 ± 7.97	<0.01
Thigh Circumference(cm)	60.01 ± 3.44	61.65 ± 3.40	<0.001
Waist Circumference (cm)	88.01 ± 6.00	86.02 ± 6.78	<0.05
Thigh Skinfold (mm)	45.12 ± 7.79	44.95 ± 7.13	NS
Abdominal Skinfold (mm)	34.31 ± 7.31	30.20 ± 8.04	<0.001

Abbreviations: BMI, Body Mass Index; HDL-C, High Density Lipoprotein; LDL-C, Low Density Lipoprotein.

¹P values for ethnic differences determined by two-group t-test.

Simple and partial correlation coefficients for ThC and WC with metabolic variables are presented in Table 2. In EA, but not AA, ThC was significantly and positively associated with insulin sensitivity, and significance remained after adjusting for thigh skinfold (Figure 1). Thigh circumference wasinversely associated with TC, LDL-C a nd TC:HDL-C ratio in EA. However, the significance of the relationships with TC and LDL-C were attenuated after adjusting for thigh skinfold. In AA, as with EA, ThC was significantly, inversely associated with LDL-C and TC:HDL-C ratio. Adjusting for thigh skinfold did not alter the significance of these associations.

Table 2.

Simple and partial correlation coefficients for thigh and waist circumference measures and metabolic risk factors by ethnicity.¹

	EA				AA
	Thigh Circumference		Waist Circumference		Thigh Circumference		Waist Circumference
	Simple	Partial	Simple	Partial	Simple	Partial	Simple	Partial
	r	r	r	r	r	r	r	r
Insulin Sensitivity	0.30*	0.29*	−0.13	−0.10	0.08	0.08	−0.33*	−0.25*
Fasting Insulin	−0.02	0.05	0.15	0.16	−0.04	−0.06	0.38*	0.30*
Fasting Glucose	−0.03	−0.06	−0.01	−0.03	0.02	−0.02	0.29*	0.19
Triglyceride	−0.12	−0.11	0.10	0.08	−0.15	−0.10	0.19*	0.17
Total Cholesterol	−0.20*	−0.14	0.02	0.04	−0.15	−0.17	0.19*	0.12
HDL-C	0.09	0.12	−0.05	0.01	0.17	0.15	−0.21*	−0.16
LDL-C	−0.21*	−0.16	−0.01	0.01	−0.21*	−0.23*	0.24*	0.15
Total Cholesterol:HDL Ratio	−0.21*	−0.20*	0.05	0.02	−0.26*	−0.26*	0.33*	0.23*
Systolic Blood Pressure	−0.04	0.03	0.17	0.20*	0.05	0.06	0.22*	0.20*
Diastolic Blood Pressure	−0.09	0.02	−0.04	−0.01	−0.00	−0.02	0.04	−0.02

Abbreviations: HDL-C, High Density Lipoprotein Cholesterol; LDL-C, Low Density Lipoprotein Cholesterol.

¹For partial correlations, thigh circumference adjusted for thigh skinfold and waist circumference adjusted for abdominal skinfold.

^*P < 0.05.

Figure 1.

The association of thigh circumference with insulin sensitivity in European American and African American women adjusted for thigh skinfold.

WC was significantly, inversely associated with insulin sensitivity in AA, but not EA, and significance remained after adjusting for abdominal skinfold (Table 2, Figure 2). Likewise, WC was significantly associated with all additional metabolic risk factors except for DBP in the AA group. However, adjusting for abdominal skinfold attenuated the significance of some of these relationships. The only significant association observed for WC in EA was with SBP.

Figure 2.

The association of waist circumference with insulin sensitivity in European American and African American women adjusted for abdominal skinfold.

Table 3 presents simple and partial correlations for thigh and abdominal skinfold measures with metabolic risk factors. Partial correlations for thigh skinfold were adjusted for ThC and partial correlations for abdominal skinfold were adjusted for WC. Thigh skinfold was significantly and inversely associated with DBP in EA and with TG in AA. No other significant relationships were observed for thigh skinfold in either ethnic group.

Table 3.

Simple and partial correlation coefficients for thigh and abdominal skinfold measures and metabolic risk factors by ethnicity.¹

	EA				AA
	Thigh Skinfold		Abdominal Skinfold		Thigh Skinfold		Abdominal Skinfold
	Simple	Partial	Simple	Partial	Simple	Partial	Simple	Partial
	r	r	r	r	r	r	r	r
Insulin Sensitivity	0.10	−0.03	−0.17	−0.22*	0.02	−0.01	−0.27*	−0.30*
Fasting Insulin	−0.16	−0.17	−0.02	−0.05	0.07	0.08	0.28*	0.15
Fasting Glucose	0.08	0.10	0.08	0.09	0.15	0.15	0.31*	0.22*
Triglyceride	−0.05	0.00	0.10	0.08	−0.23*	−0.20*	0.08	0.01
Total Cholesterol	−0.19	−0.12	−0.11	−0.12	0.05	0.08	0.20*	0.13
HDL-C	−0.07	−0.11	−0.25*	−0.24*	0.09	0.05	−0.16	−0.08
LDL-C	−0.17	−0.09	−0.09	−0.09	0.06	0.11	0.26*	0.18
Total Cholesterol:HDL Ratio	−0.07	0.01	0.13	0.12	−0.04	0.02	0.29*	0.17
Systolic Blood Pressure	−0.16	−0.16	−0.11	−0.15	−0.01	−0.02	0.11	0.01
Diastolic Blood Pressure	−0.27*	−0.26*	−0.14	−0.13	0.07	0.08	0.15	0.14

Abbreviations: HDL-C, High Density Lipoprotein Cholesterol; LDL-C, Low Density Lipoprotein Cholesterol.

¹For partial correlations, thigh skinfold adjusted for thigh circumference and abdominal skinfold adjusted for waist circumference.

^*P < 0.05.

Although abdominal skinfold was significantly associated with lower insulin sensitivity in both EA and AA, the relationship in EA only became significant after adjusting for WC (Table 3). In addition to the relationship with insulin sensitivity, abdominal skinfold was also inversely associated with HDL-C in EA but not AA. In AA, but not EA, abdominal skinfold was positively associated with fasting insulin, fasting glucose, TC, LDL-C, and TC:HDL-C ratio. After adjusting for WC however, only fasting glucose remained significantly associated with abdominal skinfold in AA, although trends were observed for LDL-C and TC:HDL-C ratio.

Discussion

The objective of this study was to determine associations of anthropometric measures of thigh and abdominal tissue with metabolic risk factors, and whether these associations differed with ethnicity in EA and AA premenopausal women. Both circumferences and skinfolds were used in order to tease apart associations with subcutaneous adipose tissue from those with muscle and deeper adipose stores.

Results showed that ThC was positively related to insulin sensitivity in EA but not AA. Studies have shown ThC, leg fat (by DXA), and thigh subcutaneous fat (by CT) to be inversely associated with proxy measures of insulin resistance such as fasting glucose, post-load glucose, or the homeostasis model assessment of insulin resistance in white European men and women (44;45), as well as a group of EA and AA men (46). We are the first to show the relationship between ThC and insulin sensitivity using FSIGT, a direct measure of insulin sensitivity. To our knowledge, we are also the first to show that there is a racial divergence in this relationship such that it is only present in EA women and not AA women.

The positive relationship between ThC and insulin sensitivity in EA was maintained after adjusting for thigh skinfold, an indirect measure of subcutaneous thigh fat. Furthermore, there was no significant association of thigh skinfold with insulin sensitivity. Together, these findings suggest that thigh skeletal muscle, as opposed to subcutaneous thigh fat, might be a mediating factor in the association between ThC and insulin sensitivity in our EA population. While this relationship needs to be verified in future studies with direct measures of thigh muscle, this premise seems logical considering that skeletal muscle is the predominant site for insulin-stimulated glucose uptake (47;48). Consistent with this hypothesis, Snijder et al (49) found significant or nearly significant inverse associations of leg lean tissue mass (via DXA) with fasting glucose and homeostasis model assessment of insulin resistance, respectively, in older men. Although subcutaneous adipose tissue has been implicated as the tissue mediating the beneficial effect of the thigh on insulin action, (50;51), Goodpaster et al (52) did not find an association between subcutaneous thigh adipose tissue and insulin sensitivity determined by hyperinsulinemic-euglycemic clamp. Taken together, these observations suggest that perhaps thigh muscle, rather than subcutaneous thigh fat, may be the tissue responsible for the “protective” effect of ThC on insulin sensitivity.

It is not clear why ThC was not associated with insulin sensitivity among AA. However, compared to EA, AA have greater intermuscular adipose tissue (53), lesser skeletal muscle mitochondrial function (54;55), and fewer type I muscle fibers (56;57). These aspects of muscle quality could influence the association between ThC and insulin sensitivity, and could potentially explain the ethnicity-specific association between ThC and insulin sensitivity observed in this study. It is also possible that differences in muscle quality may contribute to the well-documented lower insulin sensitivity in AA compared to EA (26;58;59).

Results from a number of studies have suggested that greater thigh fat deposition is related to a more favorable lipid profile (60–64). Likewise, we found that ThC was inversely related to LDL-C in both EA and AA, and to TC in EA. In general, these relationships in EA were eliminated by adjustment for thigh skinfold, indicating that the protective effect of ThC was due to subcutaneous adipose tissue, which is speculated to act as a “metabolic sink” to reduce circulating free fatty acids and their associated effects on lipid/lipoprotein metabolism (65). Among AA, subcutaneous fat as determined by thigh skinfold showed a protective association with TG. Although we cannot explain the ethnic specificity of this association, higher lipoprotein lipase activity and lower hormone-sensitive lipase activity in AA relative to EA may play a role (66–68). Taken together, these findings indicate a racial divergence in the protective role of subcutaneous fat on disease risk.

The detrimental effects of abdominal fat on disease risk have long been established (69). Studies have shown greater WC, trunk fat (by DXA), and/or IAAT (by CT) to be significantly associated with lower insulin sensitivity and HDL-C, and greater fasting insulin, circulating lipids, and blood pressure in EA and AA women (61;70–73). Consistent with these findings, we also report here that an elevated WC was associated with a poorer metabolic profile. However, these results were observed primarily in AA. Similarly, previous studies have demonstrated that there is a differential relationship between abdominal adipose tissue depots determined by CT and metabolic risk in EA and AA individuals (74–76). Many of the associations observed here with WC remained significant after adjusting for abdominal skinfold suggesting that a deeper adipose tissue depot such as IAAT may be exerting some influence on these associations. Although we also found significant associations of abdominal skinfold with disease risk factors in AA, many of these were attenuated after adjusting for WC. These findings suggest that either IAAT or deep SAAT mediated the observed associations between WC and risk factors in AA women. It has been shown that WC is a better predictor of IAAT in AA vs EA women (77;78). Further research is needed to identify the specific depot responsible for the observed association between WC and risk in AA women.

In contrast to AA women, among EA women, SAAT may be metabolically important. In agreement with previous studies showing an association between SAAT and disease risk in both men and women (79–81), we observed a significant inverse association of abdominal skinfold with insulin sensitivity after adjusting for WC, and a significant inverse association with HDL-C with or without adjustment. These observations implicate subcutaneous fat rather than IAAT in disease risk in EA women. However, further research is needed to clarify the role of superficial SAAT on metabolic health in EA women, and the ethnicity-specific association between this depot and disease risk.

A strength of this study was the use of a direct measure of insulin sensitivity, as opposed to proxy indices, such as measures based on fasting insulin and glucose. Additionally, we had access to a number of metabolic risk factors in a relatively large sample of premenopausal AA and EA women. However, because this study consisted of a relatively homogenous population of healthy, overweight, premenopausal women, it may not be appropriate to generalize the findings to other populations. Likewise, due to the cross-sectional nature of this study, a cause-and-effect relationship between anthropometric measures and risk factors cannot be determined. Lastly, we acknowledge that the use of DXA and CT are optimal techniques for assessing body composition and fat distribution. However, in field studies and large, population based studies where these sophisticated methods are not practical, the use of anthropometric measures such as ThC and WC as an estimate of body composition may be a viable option as they are inexpensive, do not require radiation exposure, and are fairly simple to administer by a trained technician.

In summary, we demonstrated that greater ThC was a strong determinant of greater insulin sensitivity in EA. Further investigation is required to determine if this association is due to thigh muscle, and why this association is specific to EA. In contrast, WC was associated with lower insulin sensitivity and a less favorable metabolic profile in AA, suggesting that WC may be a better indicator of metabolic risk in AA. Additional research is needed to elucidate the reasons for the differential associations between WC and metabolic risk factors in AA and EA.