articles
nature publishing group
Methods and Techniques
Dual-energy X-ray Absorptiometry and Anthropometric Estimates of Visceral Fat in Black and White South African Women Lisa K. Micklesfield1,2, Juliet Evans1, Shane A. Norris2, Estelle V. Lambert1, Courtney Jennings1, Yael Joffe1, Naomi S. Levitt3 and Julia H. Goedecke1,4 Visceral adipose tissue (VAT) is associated with increased risk for cardiovascular disease, and therefore, accurate methods to estimate VAT have been investigated. Computerized tomography (CT) is the gold standard measure of VAT, but its use is limited. We therefore compared waist measures and two dual-energy X-ray absorptiometry (DXA) methods (Ley and Lunar) that quantify abdominal regions of interest (ROIs) to CT-derived VAT in 166 black and 143 white South African women. Anthropometry, DXA ROI, and VAT (CT at L4–L5) were measured. Black women were younger (P < 0.001), shorter (P < 0.001), and had higher body fat (P < 0.05) than white women. There were no ethnic differences in waist (89.7 ± 18.2 cm vs. 90.1 ± 15.6 cm), waist:height ratio (WHtR, 0.56 ± 0.12 vs. 0.54 ± 0.09), or DXA ROI (Ley: 2.2 ± 1.5 vs. 2.1 ± 1.4; Lunar: 2.3 ± 1.4 vs. 2.3 ± 1.5), but black women had less VAT, after adjusting for age, height, weight, and fat mass (76 ± 34 cm2 vs. 98 ± 35 cm2; P < 0.001). Ley ROI and Lunar ROI were correlated in black (r = 0.983) and white (r = 0.988) women. VAT correlated with DXA ROI (Ley: r = 0.729 and r = 0.838, P < 0.01; Lunar: r = 0.739 and r = 0.847, P < 0.01) in black and white women, but with increasing ROI android fatness, black women had less VAT. Similarly, VAT was associated with waist (r = 0.732 and r = 0.836, P < 0.01) and WHtR (r = 0.721 and r = 0.824, P < 0.01) in black and white women. In conclusion, although DXA-derived ROIs correlate well with VAT as measured by CT, they are no better than waist or WHtR. Neither DXA nor anthropometric measures are able to accurately distinguish between high and low levels of VAT between population groups. Obesity (2010) 18, 619–624. doi:10.1038/oby.2009.292
Introduction
Visceral adipose tissue (VAT) has been shown to be a risk factor for type 2 diabetes and cardiovascular disease (1), leading to the development of various methodological strategies aimed at quantifying and characterizing VAT. Multiple-slice computerized tomography (CT) has been suggested as the gold standard to measure VAT volume, but concern with the high radiation exposure associated with this technique has limited its use and acceptability. Thus, single-slice CT measures of VAT area, which correlate strongly with VAT volume, have been used (2). Magnetic imaging resonance (MRI) is an alternative method used to quantify adipose tissue (3), with the added advantage that it does not expose individuals to radiation, and recent research by Demerath et al. (4) suggests that a single MRI may closely approximate VAT. However, the use of both CT and MRI is restricted due to limited availability and accessibility, as well as the high cost associated with these methods.
Dual-energy X-ray absorptiometry (DXA) is becoming more widely recognized as a useful and precise tool to measure body composition in research studies (5). Although two-dimensional in nature, various DXA methods, measuring a specific abdomi nal region of interest (ROI), have been proposed as an alternative to three-dimensional CT scans for quantifying VAT (6–9). Although a number of these studies have shown good correlations between DXA ROI and VAT, they have many and varied limitations. For example, a large proportion of these methods distinguish the ROI as the quadrilateral box drawn around L1–L4 (8), which includes bone (ribs), behind which the composition of the tissue is only assumed rather than measured directly (10). Further, other studies are limited by small subject numbers (9), a narrow range of participant BMI (11) and age (7), thereby suggesting that their results may not be applicable to a wider population. In addition, few studies have examined the consistency of the relationship between DXA ROI and
UCT/MRC Research Unit for Exercise Science and Sports Medicine, Department of Human Biology, University of Cape Town, Cape Town, South Africa; 2Wits/MRC Research Unit for Mineral Metabolism, Department of Paediatrics, University of Witwatersrand, Johannesburg, South Africa; 3Endocrine Unit, Department of Medicine, University of Cape Town, Cape Town, South Africa; 4South African Medical Research Council, Cape Town, South Africa. Correspondence: Lisa K. Micklesfield (
[email protected]) 1
Received 20 February 2009; accepted 9 July 2009; published online 17 September 2009. doi:10.1038/oby.2009.292 obesity | VOLUME 18 NUMBER 3 | march 2010
619
articles Methods and Techniques VAT in populations of different ethnic origins who are known to have different amounts of VAT for the same level of body fat (12,13). Moreover, discrepancies in height between ethnic groups have been reported (14), which may have implications for the ROI itself. One method that accounts for differences in truncal height by positioning the subregion box with the lower border at the pelvis cut and the upper border above the pelvis cut by 20% of the distance between the pelvis and neck cuts is the standard method included in recent Lunar software (Lunar Radiation, Madison, WI; software version >3.4). To our knowledge, no studies have compared this method to the more traditional method which relies on bony landmarks to determine the specific ROI, particularly in populations with varying height and body fat distribution. Anthropometric measures such as waist circumference are routinely used as an indirect measure of VAT in clinical practice, as well as for large epidemiological studies for which DXA may not be available. Although there is a strong correlation between waist circumference and VAT (15), the strength of this relationship may be altered by a number of methodological issues such as intra- and intertester variability (16,17) and the variation in the site of the measurement (for example, minimal or umbilical waist circumference) (18). In addition, ethnic differences in VAT may alter the relationship between waist circumference and VAT, and consequently different waist cutoffs for the metabolic syndrome have been proposed for different ethnic groups (19). In order to obviate the need for ethnic-specific waist cutpoints, the waist:height ratio (WHtR), has been proposed (20), and no studies appear to have examined the relationship between WHtR and VAT. A comparison of the relationships between VAT and waist circumference, WHtR, and DXA ROI, in populations of different ethnic origins, is warranted. Therefore, in a sample of black and white women of different heights and visceral fat areas, we aimed to (i) compare two DXA ROI methodologies; (ii) compare these methods to CT-derived VAT; (iii) to determine whether the DXA ROI methods are better than waist circumference and WHtR in predicting CT-derived VAT; and (iv) how these associations differ in black and white women. Methods and Procedures Subjects The study sample consisted of 166 black and 143 white premenopausal South African women, recruited from church groups, community centers, universities, and through the local press. Inclusion criteria were: (i) age 18–45 years; (ii) no known diseases or taking medication for any metabolic disorders; and (iii) not currently pregnant, lactating, or postmenopausal. The Research Ethics Committee of the Faculty of Health Sciences of the University of Cape Town gave approval for the study and written informed consent was obtained from all participants. Testing procedures Height (cm) and weight (kg) were measured with subjects wearing light clothing and no shoes. Waist and hip circumference were measured at the level of the umbilicus and largest gluteal area, respectively. Waist circumference (cm) was divided by height (cm) to calculate a WHtR. Whole body composition (fat mass, fat-free soft tissue mass, and bone 620
mineral content) was measured using DXA (Hologic Discovery-W, software version 12.1; scan region 195 × 65 cm2 and weight limit 204 kg) according to standard procedures, with a coefficient of variation of 0.7% for fat-free tissue mass and 1.67% for fat mass. Manually determined android waist ROIs of fat mass measured in kg, were determined for each individual by the same operator using two different methods. The first method (Ley ROI), described previously by Ley et al. (21), positioned the subregion box with the lower border above the iliac crest and the upper border at the lower levels of the ribs (excluding any bone). The lateral borders were positioned so that all soft tissue was included. The second method (Lunar ROI) is the default method used by Lunar Prodigy (GE Healthcare, Madison, WI; version 8.1), with the subregion box positioned with the lower border at the pelvis cut and the upper border above the pelvis cut by 20% of the distance between the pelvis and neck cuts, thereby taking truncal height into account. The lateral borders are represented by the arm cuts. VAT and abdominal subcutaneous adipose tissue (SAT) area were measured using a single-slice CT scan (Toshiba Xpress Helical Scanner; Toshiba Medical Systems, Tokyo, Japan) at the level of the L4–L5 lumbar vertebrae, as previously described (2). VAT areas were converted to mass (kg) using regression equations by Smith et al. (2). Statistical analysis Data are presented as means ± s.d. One-way analysis of covariance was used to compare body composition of black and white women. When comparing waist circumference, Ley ROI, VAT, and SAT, we adjusted for age, height, weight, and fat mass, and when comparing the heightadjusted variables, WHtR, and Lunar ROI, we covaried for age, weight, and fat mass. Univariate analyses were performed using Pearson’s correlations, and confidence intervals were calculated. To examine ethnic differences in the slope and intercept of the regressions between VAT and the DXA and anthropometric measures of android fat, multiple regression was used including an interaction term with ethnicity. Limits of agreement between measurements were determined using the technique of Bland and Altman (22,23). Data were analyzed using STATISTICA version 8 (StatSoft, Tulsa, OK). Results Subject characteristics
The subject characteristics and body composition data of the black and white women are presented in Table 1. The black women were significantly younger than the white women (P < 0.001), and hence all subsequent ethnic comparisons were adjusted for age. After adjusting for age, there were no ethnic differences in body weight or total fat mass; however, black women were shorter and had a significantly higher BMI and percent body fat than white women. There were no differences between the groups for waist and DXA ROI measurements, however, when comparing the CT measurements white women had significantly greater VAT and less SAT than black women. After covarying for ethnic differences in age, height, weight, and fat mass, waist circumference and Ley ROI were still not different between the groups. Similarly, Lunar ROI was not different between the groups after adjusting for age, weight, and fat mass, however, WHtR was significantly lower in the white women (0.54 ± 0.05 vs. 0.57 ± 0.05; P < 0.001). VAT (cm2 and kg) were still significantly lower in the black women compared to the white women after adjusting for age, height, weight, and fat mass (76 ± 34 cm2 vs. 98 ± 35 cm2; P < 0.001 and 2.4 ± 1.3 kg vs. 3.3 ± 1.2 kg; P < 0.001, respectively), however, SAT was no longer different between the ethnic groups. VOLUME 18 NUMBER 3 | march 2010 | www.obesityjournal.org
articles Methods and Techniques Table 1 Subject characteristics and body composition of black and white women Black (n = 166)
White (n = 143)
Mean ± s.d.
Range
Mean ± s.d.
Age (years)
26.7 ± 7.6
18.0–49.0
31.8 ± 8.0**
18.0–45.0
Weight (kg)
74.9 ± 19.4
43.0–119.7
77.3 ± 19.5
49.7–140.1
Height (m)
1.59 ± 0.06
1.47–1.77
1.67 ± 0.06**
1.54–1.86
BMI
29.3 ± 7.5
17.7–46.1
27.4 ± 6.7*
19.0–46.1
Fat mass (kg)
29.4 ± 13.3
9.7–65.0
29.1 ± 14.0
10.6–68.6
Body fat (%)
38.1 ± 8.5
20.4–55.4
36.1 ± 8.9*
16.6–53.7
Waist (cm)
89.7 ± 18.2
60.0–135.0
90.1 ± 15.6
65.0–129.0
Waist:height
0.56 ± 0.12
0.37–0.89
0.54 ± 0.09
0.38–0.76
Ley ROI (kg)
2.2 ± 1.5
0.3–6.4
2.1 ± 1.4
0.2–6.2
Lunar ROI (kg)
2.3 ± 1.4
0.4–6.4
2.3 ± 1.5
0.2–6.3
Range
DXA-derived abdominal fat
CT-derived abdominal fat area VAT (cm2)
73 ± 37
11–216
102 ± 60**
21–320
SAT (cm2)
370 ± 211
64–883
328 ± 189*
29–786
Data presented as unadjusted mean ± s.d. CT, computerized tomography; DXA, dual-energy X-ray absorptiometry; Ley ROI, android region of interest using Ley et al., (21) methodology; Lunar ROI, android region of interest using Lunar Prodigy default methodology; SAT, subcutaneous adipose tissue; VAT, visceral adipose tissue. *P < 0.05; **P < 0.001 compared to the black group.
Comparing DXA-derived abdominal fat methods
obesity | VOLUME 18 NUMBER 3 | march 2010
6 Lunar ROI (kg)
Ley ROI and Lunar ROI were highly correlated in both black (r = 0.983; 95% confidence interval 0.955–1.011) and white (r = 0.988; 95% confidence interval 0.963–1.014) women (Figure 1). Limits of agreement (LOA) for the two DXA-derived methods for the black and white women are presented in Figure 2. In black women, the mean difference between the two methods was 63 g, with a relatively large LOA (533 g) due to the high heteroscedasticity with increasing ROI fatness. In white women, the mean difference between the two methods was larger than in black women, with the Lunar method measuring, on average, 232 g higher than the Ley method. However, the LOA were similar in white and black women (475 g vs. 533 g) as the Lunar method measured consistently higher than the Ley method in the majority of the subjects in both groups (77% vs. 54%, respectively). We compared the black women whose mean difference was negative (n = 90) to those whose mean difference was positive (n = 29), and to those who showed no mean difference (n = 50) between the Lunar compared to the Ley method (Figure 2). Although the average ROI fat mass was not different on the Bland–Altman plot, the group that measured higher on the Lunar compared to the Ley method had a significantly lower body weight (72.7 ± 19.2 kg vs. 82.8 ± 21.3 kg; P = 0.018), waist circumference (86.9 ± 18.1 cm vs. 97.8 ± 18.2 cm; P = 0.005), % body fat (36.4 ± 8.8% vs. 41.4 ± 7.1%; P = 0.008), and total body fat (27.4 ± 13.5 kg vs. 34.3 ± 12.6 kg; P = 0.019). The group that showed no mean difference between methods was not significantly different to the other two groups. In white women only (Figure 2), the higher measures on the Lunar compared to the Ley method (negative mean difference) in the majority of the subjects (n = 110; 77%), increased with
White: r = 0.988; 95% CI 0.963–1.014 Black: r = 0.983; 95% CI 0.955–1.011
7
5 4 3 2 1 0 0
1
2
3 4 Ley ROI (kg)
5
6
7
Figure 1 The association between Ley ROI and Lunar ROI in black (n = 166) and white (n = 143) women. CI, confidence interval; ROI, region of interest.
increasing ROI fat mass. We compared the body composition data of these women to those whose mean difference was 0 (n = 32). Although average ROI fat mass was not different, the group that measured higher on the Lunar method were younger (31.2 ± 8.0 years vs. 34.3 ± 7.6 years; P < 0.05), had lower waist circumferences (88.7 ± 15.0 cm vs. 94.8 ± 17.1 cm; P = 0.05) and lower VAT areas (95.4 ± 54.9 cm2 vs. 122.7 ± 71.5 cm2; P < 0.05) than the other group. Comparing CT-derived VAT to DXA and anthropometric measures in black and white women
CT-derived measures of VAT (cm2 and kg) were significantly correlated with Lunar and Ley ROI methods in white and black women (Figure 3a). However, the strength of the associations was higher in the white compared to the black women for both ROI methods. The intercept and slope of the relationship between DXA-derived Ley and Lunar ROI, and VAT, were 621
articles Methods and Techniques Mean difference: −0.06 kg s.d. of difference: 0.27 kg LOA: 0.54 kg
Mean difference Ley-Lunar ROI (kg)
1.2 0.8
Mean + 2 s.d
0.4 −0.0
Mean
−0.4
Mean − 2 s.d
−0.8
Mean difference: −0.23 kg s.d. of difference: 0.24 kg LOA: 0.48 kg
White 1.2 Mean difference Ley-Lunar ROI (kg)
Black
−1.2
0.8 0.4
Mean + 2 s.d
−0.0
Mean
−0.4
Mean − 2 s.d
−0.8 −1.2
0
1 2 3 4 5 6 7 Average Ley ROI and Lunar ROI (kg)
0
1 2 3 4 5 6 7 Average Ley ROI and Lunar ROI (kg)
Figure 2 Bland–Altman plots of the differences between Ley and Lunar ROI against their mean values in black (n = 166) and white (n = 143) women. LOA, limits of agreement; ROI, region of interest.
a
White: r = 0.838; 95% CI 0.747–0.929 Black: r = 0.729; 95% CI 0.623–0.834
350
150
4
100
2
50
0
1
2
3 4 5 Ley ROI (kg)
6
6
150
4
100
2 0
0
7
0
White: r = 0.836; 95% CI 0.744–0.927 Black: r = 0.732; 95% CI 0.627–0.837
400
300
1
2
3 4 5 Lunar ROI (kg)
6
7
White: r = 0.824; 95% CI 0.729–0.918 Black: r = 0.721; 95% CI 0.615–0.828
300
250
VAT (cm2)
VAT (cm2)
8
200
50
0
0
350
VAT (cm2)
VAT (cm2)
6
10
250
VAT (kg)
200
VAT (kg)
8
12
300
10
250
b
350
12
300
White: r = 0.847; 95% CI 0.758–0.935 Black: r = 0.739; 95% CI 0.635–0.842
200 150 100
200 100
50 0
0 50
75
100 Waist (cm)
125
150
0.3
0.4
0.5
0.6 0.7 WHtR
0.8
0.9
1.0
Figure 3 The associations between computerized tomography–derived visceral adipose tissue (VAT) and (a) dual-energy X-ray absorptiometry– derived Ley ROI and Lunar ROI and (b) waist and waist:height ratio (WHtR) in black (n = 166) and white (n = 143) women. ROI, region of interest.
significantly different for black and white women (P < 0.001). With increasing ROI fat mass (Lunar and Ley), black women had less VAT than white women. Similar associations were found between VAT (cm2) and waist circumference, and VAT (cm2) and WHtR in both black and white women (Figure 3b). These associations were also not different to those observed with the DXA ROI methods, and similar differences in the slope and intercept between black and white women were observed. To explore the sensitivity of DXA-derived ROIs to correctly classify VAT (as measured by CT, cm2), we divided CT-derived VAT, as well as the DXA and anthropometric measures of android fat, into quintiles. In the combined sample, we showed that both DXA and anthropometric derived measures misclassified CT-derived VAT similarly (Ley: 57%; Lunar: 56%; waist: 57%; WHtR: 58%). There was no appreciable difference in misclassification across quintiles of VAT, or when the black (Ley: 56%; Lunar: 56%; waist: 54%; WHtR: 57%) and white 622
(Ley: 46%; Lunar: 44%; waist: 49%; WHtR: 49%) groups were analyzed separately. The LOA between CT-derived VAT (kg) and DXA-derived ROI measures of android fatness were similar within the ethnic groups (Figure 4). In black women, the mean difference between VAT, and the two methods, was small (70–130 g), however, there was increased variance between the measures at a mean VAT >2 kg. The mean difference between the methods was greater in the white women (1,110–1,350 g), with evidence of heteroscedasticity at a VAT >3 kg. With increasing VAT, the DXA-derived measures underestimated VAT to a larger extent. Discussion
The main findings of this study were that the two DXA-derived ROI methods for measuring android fatness were closely correlated with each other and were both similarly associated with VAT. However, these DXA-derived measures were not able to distinguish between low and high levels of VAT, and their VOLUME 18 NUMBER 3 | march 2010 | www.obesityjournal.org
articles Methods and Techniques Black
Mean difference: 0.13 kg s.d. of difference: 1.05 kg LOA: 2.10 kg
Mean difference VAT-Ley ROI (kg)
10.0 7.5 5.0 2.5 0.0
Mean + 2 s.d. Mean
−2.5
Mean − 2 s.d.
White 10.0 Mean difference VAT-Ley ROI (kg)
a
−5.0
5.0
Mean + 2 s.d.
2.5
Mean
0.0
Mean − 2 s.d.
−2.5
2 4 6 8 10 Average VAT and Ley ROI (g)
2.5
Mean + 2 s.d.
0.0
Mean Mean − 2 s.d.
−5.0
2 4 6 8 10 Average VAT and Ley ROI (kg)
White
Mean difference: 0.07 kg s.d. of difference: 1.02 kg LOA: 2.04 kg
5.0
−2.5
0
10.0 Mean difference VAT-Lunar ROI (kg)
Black 7.5
Mean difference VAT-Lunar ROI (kg)
7.5
−5.0 0
b
Mean difference: 1.35 kg s.d. of difference: 1.33 kg LOA: 2.66 kg
Mean difference: 1.11 kg s.d. of difference: 1.28 kg LOA: 2.56 kg
7.5 5.0
Mean + 2 s.d.
2.5
Mean
0.0
Mean − 2 s.d.
−2.5 −5.0
0
2 4 6 8 10 Average VAT and Lunar ROI (kg)
0
2 4 6 8 10 Average VAT and Ley ROI (kg)
Figure 4 Bland–Altman plots of the differences between computerized tomography– and dual-energy X-ray absorptiometry–derived measures of visceral adipose tissue (VAT), using the (a) Ley ROI and (b) Lunar ROI methods, against their mean values in black (n = 166) and white (n = 143) women. LOA, limits of agreement; ROI, region of interest.
association with VAT was also no better than that for waist, and WHtR. Adjusting for the ethnic differences in height by using the WHtR did not improve the association with VAT. We showed a very strong relationship between the Lunar and Ley ROI methods for the measurement of android fatness in both black and white women, despite the Lunar ROI method being height dependant. However, the Ley ROI method measured lower than the Lunar ROI method in the majority of women in both groups who, on average, were similar in height but who had lower levels of android fatness than the other women. The discrepancy between the two ROI methods increased with greater android fatness, possibly due to the measurement error when positioning larger individuals (5). Further, the differences between the android ROIs may be due in part, to the inclusion of some ribs in the Lunar ROI. The composition of soft tissue in areas underlying bone is based on assumptions made by the machine manufacturers and are not known (10). The presence of three distinct groups within each ethnic group could be explained, in part, by differences in body composition and distribution, however, we cannot exclude other unknown factors that may have contributed to this finding. Despite these minor differences between the DXA-derived methodologies, android fatness as measured by Ley and Lunar ROIs was similarly associated with VAT, as measured by single-slice CT, in both black and white women. The Lunar ROI is directly height dependant (20% of the distance between the neck and pelvic cuts), whereas the Ley ROI is indirectly determined by height as it is dependant on the absolute size of the region, which may differ with different body proportions. Differences in body proportions have been reported between ethnic groups (24), however, to our knowledge, no studies have compared black and white women. These DXAderived methods were selected specifically for this study due obesity | VOLUME 18 NUMBER 3 | march 2010
to previously reported ethnic differences in height (14) which may affect the size of the ROI. Although black women in this study were on average 10 cm shorter than white women, the height dependant measure of android fatness (Lunar ROI) was no more sensitive than the Ley ROI method. Although we showed a significant difference in VAT between the ethnic groups as reported previously (12), there were no ethnic differences in either of the DXA-derived measures of android fatness. The DXA-derived ROI measures were closely associated with VAT in both groups, but the slope of the relationship was different between black and white women. For the same level of android fatness, black women had less VAT than white women, particularly at higher levels of fatness. The DXA-derived ROI methods were not able to distinguish between the low and high levels of VAT in black and white women, respectively. Based on the Bland–Altman plots, both Lunar and Ley ROI underestimated VAT in white women, especially for VAT greater than ~3 kg. In black women, there was no heteroscedasticity, however, the scatter increased with VAT greater than ~2 kg. These findings are relevant to other studies that use DXA-derived measures to estimate VAT in populations with varying ethnicities. Although DXA is becoming widely accepted as a state-of-theart tool to measure body composition, we found that the association between the DXA-derived ROI measures of android fatness and VAT were no stronger than the association between VAT and waist, or WHtR. Further, the use of DXA requires expensive equipment, a trained technician and exposes participants to radiation, and does not appear to offer any additional benefit over simple anthropometric measures for quantifying VAT. In addition, the ability of the DXA to precisely define a ROI is limited by the resolution of the scan, with one pixel being equivalent to ~1 cm. WHtR has been proposed as a useful index to estimate abdominal fat deposition in different populations that vary by 623
articles Methods and Techniques age and sex (20), and some studies have suggested that this may be particularly applicable in ethnic groups of different body sizes (25). Due to height differences between black and white women in our study, we hypothesized that this measure may be more closely associated with VAT than waist alone. This was not the case in our study as we found a similar association with VAT, using waist and WHtR, in both black and white women. To our knowledge, this is the first study to examine the association between VAT and WHtR in women of different ethnic origin. Further studies are, however, required to examine differences in the association between these anthropometric measures and metabolic risk across different ethnic groups. The strength of this study is the large sample of both black and white women with different levels of VAT, which was used as a model to assess the ability of DXA ROI and simple anthropo metric measures, to estimate VAT. Although we performed singleslice CT scans to measure VAT area as opposed to multiple-slice scans that yield volume measures, the strong correlation between CT-derived VAT areas and VAT volume (2) suggests that our results and conclusions are valid and applicable. We do, however, acknowledge that the CT-derived estimate of VAT mass is a limitation of the study as it was calculated from equations based on seven-slice CT scans in a small sample of only 18 individuals (2) and therefore may have overestimated VAT mass. These estimates were merely used to explore the agreement between CT and DXA using the same units of measurement. Previous DXAderived ROI methodologies have included a measure of skinfold to estimate subcutaneous fat (6,26). However, due to the inherent problems associated with the measurement of abdominal skinfolds using calipers in obese subjects (17), we did not include this method in our study. An accurate measure of subcutaneous fat may, however, have assisted us in discriminating between low and high levels of VAT in black and white women. In conclusion, although DXA-derived measures of android fatness correlate well with VAT as measured by CT, they are no better than waist circumference or WHtR. Neither DXA nor anthropometric measures are able to accurately distinguish between high and low levels of VAT between population groups. Therefore the results of this study suggest that direct measures of VAT, such as CT or MRI, may be required when examining the specific effects of VAT on metabolic risk, which has been shown to differ widely between ethnic groups. Acknowledgments We thank the research volunteers for their participation in this study and Nandipha Sinyanya for her excellent field work. Jack Bergman, Naomi Fenton of Symington Radiology, and Linda Bewerunge are thanked for performing the computerized tomography and dual-energy X-ray absorptiometry scans, respectively. This study was funded by the South African Medical Research Council, the International Atomic Energy Agency, the National Research Foundation of South Africa, and the University of Cape Town.
Disclosure The authors declared no conflict of interest. © 2009 The Obesity Society
References 1. Després JP. Cardiovascular disease under the influence of excess visceral fat. Crit Pathw Cardiol 2007;6:51–59.
624
2. Smith SR, Lovejoy JC, Greenway F et al. Contributions of total body fat, abdominal subcutaneous adipose tissue compartments, and visceral adipose tissue to the metabolic complications of obesity. Metab Clin Exp 2001;50:425–435. 3. Fowler PA, Fuller MF, Glasbey CA et al. Total and subcutaneous adipose tissue in women: the measurement of distribution and accurate prediction of quantity by using magnetic resonance imaging. Am J Clin Nutr 1991;54:18–25. 4. Demerath EW, Shen W, Lee M et al. Approximation of total visceral adipose tissue with a single magnetic resonance image. Am J Clin Nutr 2007;85:362–368. 5. Laskey MA. Dual-energy X-ray absorptiometry and body composition. Nutrition 1996;12:45–51. 6. Hill AM, LaForgia J, Coates AM, Buckley JD, Howe PR. Estimating abdominal adipose tissue with DXA and anthropometry. Obesity (Silver Spring) 2007;15:504–510. 7. Snijder MB, Visser M, Dekker JM et al. The prediction of visceral fat by dual-energy X-ray absorptiometry in the elderly: a comparison with computed tomography and anthropometry. Int J Obes Relat Metab Disord 2002;26:984–993. 8. Glickman SG, Marn CS, Supiano MA, Dengel DR. Validity and reliability of dual-energy X-ray absorptiometry for the assessment of abdominal adiposity. J Appl Physiol 2004;97:509–514. 9. Jensen MD, Kanaley JA, Reed JE, Sheedy PF. Measurement of abdominal and visceral fat with computed tomography and dual-energy x-ray absorptiometry. Am J Clin Nutr 1995;61:274–278. 10. Pietrobelli A, Formica C, Wang Z, Heymsfield SB. Dual-energy X-ray absorptiometry body composition model: review of physical concepts. Am J Physiol 1996;271:E941–E951. 11. Kamel EG, McNeill G, Van Wijk MC. Usefulness of anthropometry and DXA in predicting intra-abdominal fat in obese men and women. Obes Res 2000;8:36–42. 12. Goedecke JH, Dave JA, Faulenbach MV et al. Insulin response in relation to insulin sensitivity: an appropriate beta-cell response in black South African women. Diabetes Care 2009;32:860–865. 13. Weinsier RL, Hunter GR, Gower BA et al. Body fat distribution in white and black women: different patterns of intraabdominal and subcutaneous abdominal adipose tissue utilization with weight loss. Am J Clin Nutr 2001;74:631–636. 14. Rush EC, Goedecke JH, Jennings C et al. BMI, fat and muscle differences in urban women of five ethnicities from two countries. Int J Obes (Lond) 2007;31:1232–1239. 15. Janssen I, Heymsfield SB, Allison DB, Kotler DP, Ross R. Body mass index and waist circumference independently contribute to the prediction of nonabdominal, abdominal subcutaneous, and visceral fat. Am J Clin Nutr 2002;75:683–688. 16. van der, Kooy K, Seidell JC. Techniques for the measurement of visceral fat: a practical guide. Int J Obes Relat Metab Disord 1993;17:187–196. 17. Wang J, Thornton JC, Kolesnik S, Pierson RN Jr. Anthropometry in body composition. An overview. Ann NY Acad Sci 2000;904:317–326. 18. Willis LH, Slentz CA, Houmard JA et al. Minimal versus umbilical waist circumference measures as indicators of cardiovascular disease risk. Obesity (Silver Spring) 2007;15:753–759. 19. Alberti KG, Zimmet P, Shaw J. The metabolic syndrome—a new worldwide definition. Lancet 2005;366:1059–1062. 20. Parikh RM, Joshi SR, Menon PS, Shah NS. Index of central obesity—a novel parameter. Med Hypotheses 2007;68:1272–1275. 21. Ley CJ, Lees B, Stevenson JC. Sex- and menopause-associated changes in body-fat distribution. Am J Clin Nutr 1992;55:950–954. 22. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;1:307–310. 23. Altman DG, Bland JM. Measurement in medicine: the analysis of method comparison studies. Statistician 1983;32:307–317. 24. Rush EC, Freitas I, Plank LD. Body size, body composition and fat distribution: comparative analysis of European, Maori, Pacific Island and Asian Indian adults. Br J Nutr 2009;102:632–641. 25. Kagawa M, Byrne NM, Hills AP. Comparison of body fat estimation using waist:height ratio using different ‘waist’ measurements in Australian adults. Br J Nutr 2008;100:1135–1141. 26. Bertin E, Marcus C, Ruiz JC, Eschard JP, Leutenegger M. Measurement of visceral adipose tissue by DXA combined with anthropometry in obese humans. Int J Obes Relat Metab Disord 2000;24:263–270.
VOLUME 18 NUMBER 3 | march 2010 | www.obesityjournal.org
