Introduction. EG Kamel 1 *, G McNeill 1, TS Han 1, FW Smith 2, A Avenell 3, L Davidson 1 and P Tothill 4

Transcription

1 International Journal of Obesity (1999) 23, 686±692 ß 1999 Stockton Press All rights reserved 0307±0565/99 $ Measurement of abdominal fat by magnetic resonance imaging, dual-energy X-ray absorptiometry and anthropometry in non-obese men and women EG Kamel 1 *, G McNeill 1, TS Han 1, FW Smith 2, A Avenell 3, L Davidson 1 and P Tothill 4 1 Department of Medicine and Therapeutics, University of Aberdeen, Aberdeen, UK; 2 Department of Biomedical Physics, University of Aberdeen, Aberdeen, UK; 3 Health Services Research Unit, University of Aberdeen, Aberdeen, UK and 4 Department of Medical Physics, University of Edinburgh, Edinburgh, UK OBJECTIVE: To investigate whether the dual-energy X-ray absorptiometry (DEXA) was a better predictor of abdominal fat, measured by magnetic resonance imaging (MRI) at the level of L2-L3, than anthropometric measurements in non-obese men and non-obese women. DESIGN: Observational, cross sectional study. SUBJECTS: 34 healthy subjects (17 men and 17 women) aged 20 ± 53 y with a body mass index (BMI) < 30 kg=m 2. MEASUREMENTS: Fat distribution parameters including waist circumference and waist-to-hip ratio (WHR); intra abdominal fat (IAF) by MRI; and central abdominal fat (CAF) by DEXA. RESULTS: Measurement of IAF by MRI, was highly correlated to the CAF measured by DEXA. In men, both waist circumference and WHR had similar correlation to IAF as DEXA. In women, waist circumference was less well correlated with IAF than DEXA, and the WHR had a weaker non signi cant correlation with IAF. CONCLUSIONS: In non obese men; DEXA, waist circumference and WHR can predict IAF equally well, while in non obese women, DEXA is superior to waist circumference and much better than WHR. Keywords: body fat distribution; L2-L3; obesity; methodology; waist circumference; waist-to-hip ratio Introduction Increased concern about obesity and its rising prevalence, has stimulated interest in identifying or predicting those at greatest health risk. 1,2 There is increasing recognition of the independent importance of body fat distribution, and an increased risk of myocardial infarction, angina pectoris, stroke and non-insulin dependent diabetes mellitus (NIDDM) which has been shown to be signi cantly associated with the amount of visceral adipose tissue. 2,3 A number of methods and techniques are used for the measurement of fat distribution including anthropometry, dual-energy X-ray absorptiometry (DEXA), magnetic resonance imaging (MRI) and computed tomography (CT). These methods differ in terms of cost, experience required, reproducibility, safety for the subject and accuracy for total body fat 4,5 and abdominal fat. 6,7 CT has been demonstrated to be an excellent technique for measuring visceral fat. 7,8 However, because of the radiation required for the CT examination, the number of measurements that can be performed on one individual is limited, making *Correspondence: Dr Emad G. Kamel, Department of Medicine and Therapeutics, University of Aberdeen, Aberdeen, AB25 2ZD, UK. Received 21 September 1998; revised 18 January 1999; accepted 10 Febraury 1999 this method unsuitable for studies requiring repeated measurements on the same subject. 9 MRI compares well with CT measured adipose tissue area, and both techniques assessed adipose tissue volume with a comparable degree of accuracy 10 ± 12 by comparison with chemical analysis. For a simple but accurate determination of the mass of abdominal adipose tissue, several studies showed that in both men and women, only one slice at the level between the second and third lumbar vertebrae gives a high and consistent predictive value for abdominal fat. 1,13,14 In order to estimate abdominal fat, simple anthropometric measurements can also be used. The waistto-hip ratio (WHR) is the traditional method of identifying subjects at increased risk due to the accumulation of excess intra abdominal fat (IAF). However, a recent study in non-obese women indicates, that measurement of waist circumference is preferable because it re ects the IAF mass 1 and is more strongly related than WHR to the health risks associated with obesity in men and women. 15 Furthermore, waist circumference is better correlated to IAF measured by CT than WHR in non-obese men and non-obese women aged > 40 y, while in younger men, both waist circumference and WHR showed the same correlation with IAF, but in younger women, the correlations were considerably weaker. 16 DEXA is becoming increasingly popular for the measurement of soft tissue composition as well as

2 bone mineral. It appears to offer a precise and simple way of measuring total and regional body fat and lean masses. 17,18 The effective dose equivalent per whole body measurement is < 5 msv and this low radiation exposure allows repeated examinations. 19 However, as it is two dimensional, DEXA cannot differentiate between IAF and subcutaneous fat, therefore DEXA has been used to assess central abdominal fat (CAF) that is IAF in addition to anterior and posterior subcutaneous fat excluding approximately 30% of the abdominal subcutaneous fat. 20 In 1992, Svendsen et al 17 found that abdominal fat measured by CT in postmenopausal women, was better correlated to the CAF estimated by DEXA than either waist circumference or WHR. 17 None of these studies compared whether DEXA measured IAF better than anthropometry in non obese men and non-obese women, and as DEXA is more expensive and time consuming than anthropometry, it is important to evaluate the usefulness of using DEXA to estimate the IAF. The aim of the present study was to investigate whether the DEXA was a better predictor of abdominal fat, measured by MRI at the level of L2-L3, than anthropometric measurements in nonobese men and non-obese women. Methods Subjects There were 34 healthy subjects (17 women and 17 men), recruited through local media. Women and men, aged between 20 ± 51 y and 25 ± 53 y, respectively, body weight 46 ± 84.6 kg and 58.5 ± 87 kg, respectively, and body mass index (BMI) 19.6 ± 29.8 kg=m 2 and 20.4 ± 28.3 kg=m 2, respectively. The study was approved by the Joint Ethical Committee of Grampian Health Board and the University of Aberdeen, and all subjects gave informed written consent to participate. MRI In women, the Mark II imager developed by the Department of Biomedical Physics and Bioengineering, University of Aberdeen was used for the total body MRI examinations. Magnetic eld strength was 0.08 T 22 and a pulse sequence technique using an interleaved inversion=saturation recovery method, repetition time 1000 ms and slice thickness 20 mm, was used to synthesise the images in order to distinguish fat from other tissues. Four images were used: at the xiphisternum, anterior iliac crest and another two equally spaced between these levels. Adipose tissue volume was calculated using a truncated cone model. 23 Fat areas of the images were traced by hand using computer mouse, the areas were obtained in pixels and converted to cm 2. In men, the Mark II imager was no longer available, so the MRI images were obtained from a Siemens Magnetom Impact System (Siemens AG, Erlangen, Germany). Magnetic eld strength was 0.95 T and the whole body was examined in contiguous 10 mm transaxial slices using a T1-weighted spin-echo sequence (TR ˆ 1150 ms, TE ˆ 12 ms). Field of view was 500 mm and acquisition time was 3 min 18 s for a group of 25 slices. MRI images at level of L2 ± L3 intervertebral space of 5 cm thickness were collected and analysed with a histogram technique 24 using ANALYZE image analysis software program (CN Software, Rochester, MN) to analyse the intra abdominal, subcutaneous and total abdominal adipose tissue in this area. As shown in Figure 1, a line can be drawn manually to separate between the intra abdominal and subcutaneous abdominal fat, then a different threshold can be applied to each compartment to separate the adipose and non-adipose tissues. Intra abdominal adipose tissues were selected according to standard anatomical knowledge. The number of pixels within the intra abdominal area was automatically calculated, and a histogram curve showing the number of pixels 687 Anthropometry Weight in light clothes was measured to the nearest 50 g, and height was measured barefoot with the head in the horizontal Frankfurt plane to the nearest 0.1 cm. Measurement of waist circumference was performed midway between the lateral lower ribs and the iliac crests while the subject was standing, after a moderate expiration, and the hip circumference was measured at the widest part over the greater trochanters. Waist and hip circumferences were recorded to the nearest millimetre, and the average of two measurements was used for analysis. Skinfold thickness was measured on the left side of the body to the nearest millimetre at the biceps, triceps, subscapular and suprailiac sites using standard skinfold calipers. The sum of the four skinfold thicknesses was used to estimate the total body fat in all subjects. 21 Figure 1 A typical magnetic resonance imaging (MRI) image with outlined intra abdominal adipose tissue compartment at L2- L3 level, the arrows show the intra abdominal adipose tissue.

3 688 of different intensity was displayed. The bi-model distribution was used to determine a cut-off value for image intensity of adipose tissue (high intensity) and non-adipose tissue (low intensity). The number of pixels above the cut-off intensity was calculated and converted to a cross sectional area which was measured to re ect the adipose tissue. The MRI analysis was repeated twice and the average of the three values was used for the analysis. Adipose tissue was assumed to contain 80% fat, 18% water and 2% protein, and fat mass was calculated in kilograms, assuming the density of the adipose tissue to be kg=l. 25 All subjects wore comfortable clothes without any metal in the clothes or any jewellery attached to their bodies. During MRI examination, the subjects stayed absolutely still to reduce interference with the images. In both sexes, measurements did not include the head, feet, forearms and hands. DEXA Whole body DEXA measurements were made with a Norland XR26 Mark II HS Scanner (Norland Corporation, Winsconsin, USA), equipped with dynamic ltration. A standard whole body DEXA examination includes total body and three regional measurements of trunk (chest, abdomen and pelvis), arms and legs, to analyse body composition according to a three compartmental model; including fat mass, lean tissue and bone mineral content. The standard soft tissue analysis was performed using software version 2.4 supplied by the manufacturer. Total body fat was estimated for each subject in kg. CAF was measured using the region of interest facility to de ne a rectangular area of abdominal tissue of 5 cm thickness at the level of L2 ± L3. This included the IAF at this region plus anterior and posterior subcutaneous fat. The measurements for CAF were calculated three times with three different widths of the region of interest; rstly with vertical sides of the rectangle extending to the lateral margins of the image (CAF1); secondly, with vertical sides as the continuation of the lateral sides of the rib cage 20 (CAF2); thirdly, with xed width of 15 cm 26 (CAF3) (Figure 2). Only 12 women had DEXA, the time interval between MRI and DEXA measurements was approximately two months, and the weight of the subjects did not change by > 2 kg in this period. Five of the women were not available for the DEXA measurement. In the men, the interval between MRI and DEXA was not more than two weeks. Statistical analysis Data are presented as mean s.d. and differences between men and women were tested by unpaired t- test. Linear regression and correlation coef cients were used to assess the relationship between variables. Multiple regression analyses were applied to identify the best predictors of IAF. Analysis was carried out Figure 2 Print-out of dual-energy X-ray absorptiometry (DEXA) scan, showing regions of interest: (A) The region of interest includes abdominal fat (IAF) plus all the subcutaneous fat at this region (CAF1); (B) The lateral borders of the region of interest are the continuation of the rib cage (CAF2); and (C) The width of the region of interest is 15 cm (CAF3). CAF ˆ central abdominal fat. using SPSS Advanced Statistics software version 6.1 (SPSS Inc., Chicago, USA). Results Characteristics of the samples of men and women are presented in Table 1. An initial comparison of the results shows that men were taller, heavier and had higher waist circumference and WHR than women. The DEXA women (n ˆ 12) were similar in anthropometric characteristics to the MRI women group (n ˆ 17). Total body fat measured by DEXA indicates that women had more fat than men. This high fat content in women is mainly in the subcutaneous area as measured by MRI (P < 0.05). The mean value of BMI indicates that men were slightly more overweight than women (P > 0.05) and they had more IAF as measured by MRI than women (P < 0.001). As DEXA measures CAF, which includes anterior and posterior subcutaneous fat in addition to the IAF, it was found that CAF was higher than IAF as measured by MRI for both women and men (Table 1). To detect the best width in the abdominal region of interest for estimation of the CAF by DEXA, we measured the fat in this area by using three different widths. The correlation coef cient between IAF and each method was calculated using linear regression. All methods were also highly correlated with IAF by MRI with correlation coef cients ranging from 0.83 ±

4 Table 1 Subject characteristics Women (n ˆ12) a Women (n ˆ17) Men (n ˆ17) 689 Mean s.d. Range Mean s.d. Range Mean s.d. Range P b Age (y) ± ± ± 53.0 NS Weight (kg) ± ± ± Height (cm) ± ± ± BMI (kg=m 2 ) ± ± ± 28.3 NS Waist circumference (cm) ± ± ± Hip circumference (cm) ± ± ± NS WHR ± ± ± TBF by DEXA (kg) ± 32.5 ± ± ± 30.5 NS TAF by MRI (kg) ± ± ± 1.3 NS IAF by MRI (kg) ± ± ± SAF by MRI (kg) ± ± ± CAF2 by DEXA (kg) ± 1.96 ± ± ± 1.8 NS BMI ˆ body mass index; WHR ˆ waist-to-hip ratio; TBF ˆ total body fat; DEXA ˆ dual energy X-ray absorptiometry; TAF ˆ total abdominal fat; MRI ˆ magnetic resonance imaging; IAF ˆ intra abdominal fat; SAF ˆ subcutaneous abdominal fat; CAF2 ˆ central abdominal fat, the vertical sides of the rectangle were the continuation of the lateral sides of the rib cage; NS ˆ not statistically signi cant. a Five women did not have a DEXA examination. The characters of the other 12 women who had both DEXA and MRI examinations are show as a separate group. b Comparison by unpaired t-test between women (n ˆ 17) and men (n ˆ 17) except for TBF and CAF2 by DEXA (n ˆ 12 women). Table 2 Correlation coef cients of measurements of central abdominal fat (CAF) by dual-energy X-ray absorptiometry (DEXA) at level of L2=L3 in a slice of 5 cm thickness, but different widths of the area of interest, with intra abdominal fat (IAF) measured by magnetic resonance imaging (MRI) Women (n ˆ12) Men (n ˆ17) Fat content(kg) Width(cm) r P Fat content (kg) Width (cm) r P CAF CAF CAF For de nition of CAF1, CAF2 and CAF3 see text. Table 3 Comparison of tertiles of intra abdominal fat (IAF) and central abdominal fat (CAF) in women (Table 3a) and men (Table 3b). Table 3a Women (n ˆ 12) CAF by DEXA High > 0.25 kg IAF by MRI Moderate 0.15 ^ 0.25 kg Low < 0.15 kg High > 1.0 kg 3 1 Moderage 0.6 ± 1.0 kg Low < 0.6 kg 1 3 Table 3b Men (n ˆ 17) CAF by DEXA High > 0.5 kg IAF by MRI Moderate 0.42 ^ 0.5 kg Low < 0.42 kg High > 1.04 kg 4 1 Moderate 0.9 ± 1.04 kg Low < 0.9 kg 1 5 MRI ˆ magnetic resonance imaging; DEXA ˆ dual-energy X-ray absorptiometry (Table 2). Measurement of CAF2 and CAF3 included 84% and 46.5%, respectively, of the total regional fat measured by CAF1. In order to compare MRI and DEXA, the data were divided into tertiles. Of the 12 women 8 were correctly classi ed by DEXA and no subject was grossly misclassi ed, while 13 men out of the 17 were correctly classi ed by DEXA and no subject was grossly misclassi ed (Table 3a and Table 3b). Table 4, and Figure 3, Figure 4, Figure 5 and Figure 6, show the correlation coef cient of the other measurements with IAF as measured by MRI. In men, waist circumference and WHR show the same correlation with IAF as CAF (r ˆ 0.89, 0.90 and 0.87, respectively). In women, CAF shows a better correlation coef cient with IAF than waist circumference (r ˆ 0.88 and 0.77, respectively). Furthermore, WHR was not signi cantly correlated with IAF (r ˆ 0.44, P ˆ 0.08). Using multiple regression, the addition of waist circumference, WHR or waist height ratio to CAF by DEXA did not improve the correlation coef cients in women, while in men the correlation increased to 0.90 (P < 0.001) using DEXA and waist circumference, and 0.93 (P < 0.001) for DEXA and WHR (Table 5). Table 6 shows that WHR, waist circumference and CAF2 by DEXA in men, explain nearly the same

5 690 Table 4 Correlation coef cient with intra abdominal fat (IAF) measured by magnetic resonance imaging (MRI) Women (n ˆ17) Men (n ˆ17) Variable r P r P CAF measured by DEXA a Weight Height 0.06 NS 0.40 NS BMI Waist circumference Hip circumference Waist-to-hip ratio 0.44 NS CAF ˆ central abdominal fat; DEXA ˆ dual-energy X-ray absorptiometry; BMI ˆ body mass index; NS ˆ not statistically signi cant. a (n ˆ 12 women). Figure 5 Correlation between intra abdominal fat (IAF) mass as measured by magnetic resonance imaging (MRI) and waist circumference in men (r ˆ 0.89, P < 0.001). Figure 3 Correlation between intra abdominal fat (IAF) mass as measured by magnetic resonance imaging (MRI) and waist circumference in women (r ˆ 0.77, P < 0.001). WC ˆ waist circumference. Figure 6 Correlation between intra abdominal fat (IAF) mass as measured by magnetic resonance imaging (MRI) and central abdominal fat (CAF) as estimated by dual-energy X-ray absorptiometry (DEXA) in men (r ˆ 0.87, P < 0.001). Table 5 Multiple regression of central abdominal fat (CAF) and anthropometric measures with intra abdominal fat (IAF) as measured by magnetic resonance imaging (MRI) Women (n ˆ12) Men (n ˆ17) Variable R* P R P DEXA waist circumference DEXA WHR DEXA waist=height ratio Waist circumference WHR DEXA ˆ dual-energy X-ray absorptiometry; WHR ˆ waist-to-hip ratio. *R ˆ multiple regression. Figure 4 Correlation between intra abdominal fat (IAF) mass as measured by magnetic resonance imaging (MRI) and central abdominal fat (CAF) as estimated by dual-energy X-ray absorptiometry (DEXA) in women (r ˆ 0.88, P < 0.001). amount of variance in the IAF (80%, 79% and 76%, respectively), while in women CAF2 by DEXA explains more of the variance than waist circumference (78% and 59%, respectively). Moreover, WHR in women explains a non signi cant amount of variance. Discussion Despite the fact that women have a higher percentage of body fat than men (30.7% and 23.2% of their body weight, respectively, based on DEXA measurements in this study), male subjects had signi cantly higher mean waist circumference and higher mean WHR, in agreement with the greater tendency of men to accu-

6 Table 6 Variances of different indices of adiposity and central abdominal fat (CAF) explained by intra abdominal fat (IAF) measured by magnetic resonance imaging (MRI) using linear regression analysis Women Men adjusted r 2 P adjustedr 2 P BMI Waist circumference WHR 0.19 NS Weight CAF a BMI ˆ body mass index; WHR ˆ waist-to-hip ratio; NS ˆ not statistically signi cant. a n ˆ 12 women. mulate excess fat in the abdominal region. 2 In addition, the men had a substantially lower mean total abdominal fat than women, but they had higher IAF, in agreement with the previously documented greater tendency of men to store fat within the abdominal cavity. 11 Furthermore, by comparing total abdominal fat and IAF by MRI, it can be calculated that women had just 24.5% of the total abdominal fat as visceral fat, while men had 47.2% as IAF. Some of these differences may be due to the use of different MRI images and different analysis techniques between men and women, but it seems unlikely that such a large difference can be explained by the difference in techniques only. In this study, we compared the waist circumference, the WHR and DEXA with regard to their relations with total body fat as measured by skinfold thickness. Waist circumference and DEXA were strongly associated with total body fat in both sexes, while the association between WHR and total body fat mass was substantially weaker (data not presented). Waist circumference was signi cantly correlated to intra IAF measured by MRI in men (r ˆ 0.89) and in women (r ˆ 0.77) in agreement with Seidell et al 16 who reported signi cant correlation, in non obese subjects, ranging from 0.81 ± 0.90 in men and 0.49 ± 0.86 in women, according to their age; while Pouliot et al 15 estimated a correlation of 0.77 in men with wide ranges of age and BMI, including obese subjects. The correlation between waist circumference and IAF in the women of this study, was less than other studies 1,15,17,27 in which the range of the correlation was 0.84 ± 0.87, while Ross et al 11 found a correlation between IAF and waist circumference less than ours (0.65). The differences between the results of this study and the other studies, may be due to the differences in age and BMI ranges or measurement techniques. Weaker correlation between IAF and WHR than with waist circumference in women, was found in this study (0.44), in agreement with other studies which reported a correlation range between 0.59 ± 0.69, 1,15,16,27,28 while Svendsen et al 17 reported a correlation of 0.81 in postmenopausal women. In men, we found a high correlation between IAF and WHR (0.90), in agreement with Seidell et al 16 who reported a correlation range between 0.84 ± 0.89, depending on the age group, while Pouliot et al 15 estimated a correlation of 0.71 between IAF and WHR. WHR is often used in the clinical setting as an indicator for metabolic complication of obesity, as con rmed in this study in men. The adjusted r 2 between WHR and IAF in women was low (0.19). Moreover, adjusted r 2 in men between IAF and either waist circumference, WHR or DEXA was similar to each other. Thus WHR and waist circumference are good and simple indicators for IAF only in men while in women, it is suggested that a measurement such as DEXA should be considered by clinicians instead. Carey et al 20 in 1996 excluded some subcutaneous fat by applying the lateral margins of the abdominal region of interest as a continuation of the outer edge of the rib cage. They mentioned that this method excludes approx. 30% of the total abdominal subcutaneous fat. In the current study we found that this method (CAF2), as well as the other two methods (CAF1 and CAF3), are highly correlated to each other and to IAF measured by MRI, therefore, using any of these methods can predict IAF mass accurately. Measurement of IAF by MRI as a gold standard, showed a good correlation with DEXA predicted abdominal fat (r ˆ 0.87 in men and 0.88 in women). Furthermore, by dividing the data into tertiles, no subject was grossly misclassi ed. Jensen et al 7 measured IAF by CT in men and Svendsen et al 17 measured IAF by CT in women, they found high correlation with central abdominal fat measured by DEXA (0.985 and 0.998, respectively). On the other hand, Treuth et al 27 reported a correlation of 0.72 between DEXA and CT in women with different BMI. We also compared the waist circumference, WHR and DEXA with regard to their relations with abdominal visceral fat. CAF by DEXA was strongly associated with the level of abdominal visceral fat in both sexes. In addition, waist circumference and WHR give the same correlation as DEXA in men. In women, waist circumference showed slightly weaker correlation than DEXA and WHR showed a much weaker correlation than DEXA. These results suggest that DEXA is superior to waist circumference or WHR in women, while these measurements are the same as DEXA in men. Therefore, for a simple and easy method to estimate the abdominal fat mass, waist circumference or WHR are more useful in men, while in women DEXA is more useful. Unfortunately, the combination of DEXA and anthropometry did not improve the correlation in women, while in men there was a slight improvement. Jensen et al, 7 who used a group of men and women with a wide BMI range, found that the combination of DEXA and anthropometry, was not a better predictor of IAF than each one alone. 691

7 692 Conclusion IAF measured by MRI and CAF estimated by DEXA were highly correlated in both sexes. However, in men, both waist circumference and WHR showed similar correlation with IAF by MRI, while in women, DEXA was a little better than waist circumference and much better than WHR. Further research is needed to assess whether the CAF measured by DEXA is a better predictor of IAF than anthropometric measurements in obese men and obese women. Acknowledgements We thank Dr David Reid, Department of Medicine and Therapeutics, University of Aberdeen for his help with DEXA examinations. Our thanks are also due to Dr Tom Redpath, Department of Medical Physics, Aberdeen Royal In rmary for his help with MRI examinations, and to all the subjects for their participation in the study. References 1 Han TS, McNeill G, Seidell JC, Lean MEJ. Predicting intrauterine fatness from anthropometric measures: the in uence of status. Int J Obes 1997; 21: 587 ± BjoÈrntorp IP, Brodoff BN. Regional obesity. In: Bjorntorp IP, Brodoff BN (eds) Obesity. J.B. Lippincott Co, Pennsylvania, 1992, pp 579 ± Lemieux S, Prud'homme D, Bouchard C, Tremblay A, DespreÂs J. A single threshold value of waist girth identi es normal-weight and overweight subjects with excess visceral adipose tissue. Am J Clin Nutr 1996; 64: 685 ± McNeill G, Fowler PA, Maughan RJ, McGaw BA, Gvozdanovic D, Fuller MF. Body fat in lean and obese women measured by six methods. Proc Nutr Soc 1989; 48: 23A. 5 Fowler PA, Fuller MF, Glasbey MA, Cameron GG, McNeill G, Maughan RJ. 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 ± Van der Kooy, Seidell JC. Techniques for the measurement of visceral fat: a practical guide. Int J Obes 1993; 17: 187 ± 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 ± Enzi G, Gasparo M, Biondetti PR, Fiore D, Semisa M, Zurlo F. Subcutaneous and visceral fat distribution according to the sex, age and overweight, evaluated by computed topography. Am J Clin Nutr 1986; 44: 739 ± Sohlstrom A, Wahlund L, Forsum E. Adipose tissue distribution as assessed by magnetic resonance imaging and total body fat by magnetic resonance imaging, underwater weighing and body-water dilution in healthy women. Am J Clin Nutr 1993; 58: 830 ± Ross R, Leger L, Guardo R, D Guise J, Pike BG. Adipose tissue volume measured by magnetic resonance imaging and computerised tomography in rats. J App Physiol 1991; 70: 2164 ± Ross R, Shaw KD, Martel Y, Guise J, Avruch L. Adipose tissue distribution measured by resonance imaging in obese women. Am J Clin Nutr 1993; 57: 470 ± Abate N, Burns D, Peshoch RM, Garg A, Grundy SM. Estimation of adipose tissue mass by magnetic resonance imaging: validation against dissection in human cadavers. J Lipid Res 1994; 35: 1490 ± Abate N, Garg A, Coleman R, Grundy SM, Peshock RM. Prediction of total subcutaneous abdominal, intraperitoneal and retroperitoneal adipose tissue masses in men by a single axial magnetic resonance imaging slice. Am J Clin Nutr 1997; 65: 403 ± Han TS, Kelly IE, Walsh K, Greene RME, Lean MEJ. Relationship between volumes and areas from single transverse scans of intra-abdominal fat measured by magnetic resonance imaging. Int J Obes 1997; 21: 1161 ± Pouliot MC, DespreÂs JP, Lemieux S, Moorjani S, Bouchard C, Tremblay A, Nadeau A, Lupien PJ. Waist circumference and abdominal sagittal diameter: best simple anthropometric indices of abdominal visceral tissue accumulation and related cardiovascular risk in men and women. Am J Cardiol 1994; 74: 460 ± Seidell JC, Oosterlee A, Deurenberg P, Hautvast JGAJ, Ruijs JHJ. Abdominal fat depots measured with computed topography: effect of degree of obesity, sex and age. Euro J Clin Nutr 1988; 42: 805 ± Svendsen OL, Hassager C, Bergmann I, Christiansen C. Measurement of abdominal and intra-abdominal fat in post menopausal women by dual energy X-ray absorptiometry and anthropometry: comparison with computerised tomography. Int J Obes 1993; 17: 45 ± Tothill P, Han TS, Avenell A, McNeill G, Reid DM. Comparisons between fat measurements by dual-energy X-ray absorptiometry, underwater weighing and magnetic resonance imaging in healthy women. Eur J Clin Nutr 1996; 50: 747 ± Jebb SA. Measurement of soft tissue composition by dual X- ray absorptiometry. Br J Nutr 1997; 77: 151 ± Carey DGP, Nguyen TV, Campbell LV, Chisholm DJ, Kelly P. Genetic in uences on central abdominal fat: a twin study. Int J Obes 1996; 20: 722 ± Durnin JVGA, Womersley J. Body fat assessed from total body density and its estimation from skin fold thickness: measurements on 481 men and women aged from 16 to 72 years. Br J Nutr 1974; 32: 77 ± Redpath TW, Hutchison JMS, Eastwood LM, Selbie RD, Johnson G, Jones RA, Mallard JR. A low eld NMR imager for clinical use. J Phys E:Sci Instrum 1987; 20: 1228 ± Kvist H, Chowdbury B, Grangard U, Tylen U, Sjostrom L. Total and visceral adipose tissue volumes derived from measurements with computed tomography in adult men and women. Am J Clin Nutr 1988; 48: 153 ± Lancaster JL, Ghiatas AA, Alyassin A, Kilcoyne RF, Bonora E, DeFronzo RA. Measurement of abdominal fat with T1- weight MR Images. JMRI 1991; 1: 363 ± Garrow JS. Energy balance and obesity in men. Elsevier: New York, Schlemmer A, Hassager C, Haarbo J, Christiansen C. Direct measurement of abdominal fat by dual photon absorptiometry. Int J Obes 1990; 14: 603 ± Treuth MS, Hunter GR, Kekes-Szabo T. Estimating intra abdominal tissue in women by dual X-ray absorptiometry. Am J Clin Nutr 1995; 62: 527 ± Ashwell M, Cole TJ, Dixon AK. Obesity: new insight into the anthropometric classi cation of fat distribution shown by computed tomography. BMJ 1985; 290: 1692 ± 1694.