Visceral and Subcutaneous Adiposity Measurements in Adults: Influence of Measurement Site

Transcription

1 Human Physiology Visceral and Subcutaneous Adiposity Measurements in Adults: Influence of Measurement Site Kenneth J. Ellis,* Birgit Grund, Fehmida Visnegarwala,* Lisa Thackeray, Collin G. Miller, Charles E. Chesson, Wafaa El-Sadr, and Andrew Carr for the Strategies for Management of Anti- Retroviral Therapy (SMART) Study Group Abstract ELLIS, KENNETH J., BIRGIT GRUND, FEHMIDA VISNEGARWALA, LISA THACKERAY, COLLIN G. MILLER, CHARLES E. CHESSON, WAFAA EL-SADR, AND ANDREW CARR FOR THE STRATEGIES FOR MANAGEMENT OF ANTI-RETROVIRAL THERAPY (SMART) STUDY GROUP. Visceral and subcutaneous adiposity measurements in adults: influence of measurement site. Obesity. 2007;15: Objective: Excess abdominal adiposity is a known risk factor for cardiovascular diseases. Computed tomography can be used to examine the visceral (VAT) and subcutaneous (SAT) components of abdominal adiposity, but it is unresolved whether single-slice or multi-slice protocols are needed. Research Method and Procedures: Nine computed tomography scans were obtained in the lumbar spine region of 24 adults. The nine slices were obtained at three intervertebral positions (L2 L3, L3 L4, and L4 L5) and at 7 mm above and below these locations. Intra-site and inter-site differences in SAT, VAT, total adipose tissue, and the VAT/SAT ratio were examined using ANOVA and confidence intervals for pairwise differences between means. Results: Intervertebral SAT values increased from Received for review May 26, Accepted in final form November 28, The costs of publication of this article were defrayed, in part, by the payment of page charges. This article must, therefore, be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. *USDA/ARS Children s Nutrition Research Center, Baylor College of Medicine, Houston, Texas; School of Statistics, University of Minnesota, Minneapolis, Minnesota; Bio- Imaging Technologies, Newtown, Pennsylvania; Social and Scientific Systems, Silver Springs, Maryland; Department of Medicine, Columbia University, New York, New York; and HIV, Immunology and Infectious Diseases Clinical Services Unit, St. Vincent s Hospital, Sydney, Australia. Address correspondence to Kenneth J. Ellis, Body Composition Laboratory, USDA/ARS Children s Nutrition Research Center, Baylor College of Medicine, 1100 Bates Street, Houston, TX kellies@bcm.tmc.edu Copyright 2007 NAASO 50.9 (standard deviation) cm 2 at L2 L3 to cm 2 at L4 L5, whereas the corresponding VAT values decreased from to cm 2. The VAT/SAT ratio was not constant, decreasing from to Repeated-measures ANOVA indicated significant inter- and intra-site differences (p 0.02) for SAT, VAT, and the VAT/SAT ratio at L3 L4 and L4 L5 (p 0.001). Discussion: These differences show the limitation of using a single-slice assessment of abdominal fat distribution, both for a subject and between subjects. Furthermore, the sizeable differences in the intra-site scans indicate that precise repositioning is needed for longitudinal studies. In summary, our findings suggest that a multi-site imaging protocol may provide a more complete assessment of abdominal fat stores and distribution than use of a single site. Key words: computed tomography, human immunodeficiency virus, abdominal adiposity, lipodystrophy Introduction Excess abdominal adiposity has been related to a number of diseases affecting both quality of life and longevity (1 5). Furthermore, the visceral adipose tissue (VAT) 1 component of abdominal fat seems to have a stronger association with these diseases than does subcutaneous adipose tissue (SAT). Fat distribution in the abdomen can be assessed using computed tomography (CT); its accuracy has been confirmed in cadaver studies (6 8). For most studies, it has been customary to report the mean VAT and SAT values and occasionally the VAT/SAT ratio. Although the newest CT instruments are capable of obtaining simultaneous multi-slice 1 Nonstandard abbreviations: VAT, visceral adipose tissue; SAT, subcutaneous adipose tissue; CT, computed tomography; HIV, human immunodeficiency virus; TAT, total adipose tissue; SD, standard deviation. OBESITY Vol. 15 No. 6 June

2 images, the most commonly used protocol for assessment of abdominal adiposity has been a single-slice scan, typically obtained at the L4 L5 region of the lumbar spine (9 15). Only a few studies have examined the differences in SAT and VAT when obtained at more than one abdominal site (15 19). In this study, we obtained a total of nine single-slice axial scans of the abdomen in the lumbar spine region between L2 and L5. We wanted to compare the results for VAT, SAT, and the VAT/SAT ratio at the intervertebral positions of L2 L3, L3 L4, and L4 L5. Our second objective was to determine the magnitude of the errors that are produced if the slice is inaccurately positioned at any of these locations. This was determined on the basis of scans that were obtained 7 mm inferior and superior to the true intervertebral positions. Our hypotheses were 1) VAT and SAT values would be different among the intervertebral sites; 2) the VAT/SAT ratio would remain relatively unchanged; and 3) small inaccuracies in the slice location at an intervertebral site would not significantly alter the values for subcutaneous or visceral adiposity. Research Methods and Procedures Subjects The participants in this study were part of the Body Composition substudy of the Strategies for Management of Anti-retroviral Therapy study (SMART, org), which was an international randomized trial comparing the long-term health consequences of two different strategies for use of anti-retroviral drugs in human immunodeficiency virus (HIV)-positive adults. Written informed consent was obtained from all subjects. The CT scans for the initial 27 subjects participating in the Body Composition substudy were examined. Three subjects were excluded from further analyses because their nine scan images could not be fully analyzed, therefore providing an incomplete dataset. Of the remaining 24 subjects (20 men and 4 women), there were 12 whites, 7 blacks, and 5 Hispanics, with a mean age of years, who were HIV positive for years. Mean body weight and BMI were kg and kg/m 2, respectively. Three men had BMI values 30 kg/m 2, the lower limit often used to define obesity in adults. CT Scans and Analyses The Strategies for Management of Anti-retroviral Therapy study is an international program; thus, subjects for this study were recruited at several clinical centers in the United States and Australia. A common scan protocol was followed at all locations, and unanalyzed scans and quality control scans were sent to a central reading center (Bio-Imaging Technologies, Newtown, PA) for analysis. For the scans, the subjects lay in a supine position on the CT table, with hands stretched above the head and legs fully extended downward. Based on a lateral scout scan of the abdomen (lowest rib to the pelvic area), the intervertebral positions of L2 L3, L3 L4, and L4 L5 of the lumbar spine were located. To obtain scans perpendicular to the intervertebral spaces, the CT s gantry was rotated to off-vertical angles for each position. Three axial scans were obtained at each site: superior (7 mm above the intervertebral space), middle (at the intervertebral space), and inferior (7 mm below the intervertebral space). This sequence was repeated for the three intervertebral sites, resulting in a total of nine axial slices. The subcutaneous and visceral adipose regions were obtained by manually tracing these regions on each image. The skin defined the outer boundary for the SAT region, and abdominal muscle and bone were used to trace the interboundary. This boundary also served as the outer boundary for the visceral region. The pixel density (in arbitrary units) for the SAT region was first determined and used to identify pixels with a similar density in the visceral region in order to define VAT. The boundaries for non-adipose tissues within the visceral region, i.e., bone, muscle, organs, blood vessels, and bowels, were traced out, and these regions were excluded from the calculation of the VAT area. In all cases, the inner and outer boundaries for the adipose regions and the tracings around organs could touch or overlap. All CT analyses were independently over-read by an experienced investigator, and small adjustments (changes in boundaries) were made as needed. It was at this stage that the three subjects were excluded from the statistical analyses. One subject s SAT region in several images was too small to accurately trace, while for the other two subjects, the quality of the scan images was too poor to be reliably read. For the remaining 24 subjects, the adiposity values were obtained for each of the nine slices. These values were obtained for VAT, SAT, total adipose tissue (TAT; TAT VAT SAT), and the VAT/SAT ratio. Statistical Procedures The results for the 24 subjects who had a complete set of adiposity measurements for all nine scans were used in the statistical analyses. For each of the nine slices, the mean, standard deviation (SD), and median values were calculated for VAT, SAT, TAT, and VAT/SAT ratio. For each of the three lumbar spine locations (L2 L3, L3 L4, and L4 L5), differences between the middle and the superior and inferior slices were compared using F-tests in a repeated-measures mixed effects ANOVA model (intra-site comparisons). p values, SDs of residuals, and 95% confidence intervals for differences between adjacent slices at each site are reported. For inter-site comparisons, the middle (true intervertebral) slices were compared using F-tests in a repeated-measures 1442 OBESITY Vol. 15 No. 6 June 2007

3 Table 1. Mean, SD, and median values for SAT, VAT, VAT/SAT, and TAT for the superior, middle, and inferior slices at three intervertebral sites (n 24) Mean SD (Median) Superior slice Middle slice Inferior slice SD of residuals* p SAT (cm 2 ) L2-L (94.6) (96.8) (100.4) L3 L (116.9) (122.7) (127.6) L4 L (123.1) (128.1) (133.9) VAT (cm 2 ) L2 L (162.9) (160.5) (163.9) L3 L (152.0) (144.8) (141.1) L4 L (124.1) (118.3) (104.3) TAT (cm 2 ) L2 L (252.2) (237.0) (250.9) L3 L (238.6) (229.3) (253.9) L4 L (268.0) (260.2) (263.2) VAT/SAT L2 L (1.3) (1.4) (1.4) L3 L (1.2) (1.1) (1.0) L4 L (0.8) (0.7) (0.6) SD, standard deviation; SAT, subcutaneous adipose tissue; VAT, visceral adipose tissue; TAT, total adipose tissue. * Estimates within-subject variability at each site, adjusted for slice effect. F-test for equal means of the superior, middle, inferior slices, in a repeated-measures mixed model. mixed effects ANOVA model. All analyses were performed using SAS version 8.2 software (SAS Institute, Inc., Cary, NC). Results The mean, median, and SD values for SAT, VAT, TAT, and the VAT/SAT ratio for all nine locations are given in Table 1. The between-subject (biological) variability is reflected by the magnitude of the SD values. There were inter-site differences in VAT, SAT, and the VAT/SAT ratio among the L2 L3, L3 L4, and L4 L5 sites (middle slices, p for all comparisons), but no significant difference in TAT (p 0.37). The SDs of the within-subject measurement errors for each adiposity index are given in Table 1, as SDs of residuals. For SAT, these values ranged from 4.6 cm 2 at L2 L3 to 10.0 cm 2 at L4 L5. The corresponding values for VAT varied from 9.7 cm 2 at L2 L3 to 11.3 cm 2 at L4 L5. The SDs of residuals at L4 L5 for SAT, VAT, and TAT were substantially larger than those at the other two sites. The range and distribution of SAT and VAT values for the nine sites are shown in Figure 1. The general pattern was for SAT values to increase when moving lower on the abdomen relative to the L2 L3 location. There was a significant increase (p 0.001) of 50% in the mean SAT values between the upper (L2 L3, cm 2 ) and lower (L4 L5, cm 2 ) abdomen. The general pattern for within-site changes in SAT indicated a significant increase Figure 1: Mean values for SAT and VAT for the superior, middle, and inferior slices at each intervertebral location. OBESITY Vol. 15 No. 6 June

4 Table 2. Mean differences and 95% confidence intervals for SAT, VAT, VAT/SAT, and TAT for adjacent slices at each intervertebral position Difference of means (95% confidence interval) Superior middle Middle inferior Figure 2: Mean values for the VAT/SAT ratio and TAT for the superior, middle, and inferior slices at each intervertebral location. (p 0.001) when moving from the inferior to the superior position at each of the three intervertebral locations. For VAT, the overall pattern was reversed from what was observed for SAT, but the differences were not as striking. The mean VAT value at L4 L5 (126.0 cm 2 ) was 24% lower than at L2 L3 (164.3 cm 2 ). This general pattern of decreasing VAT values was also evident for intra-site values for VAT, except at L2 L3, where there was a small increase from the superior to the inferior slice. ANOVA comparisons indicated both intra-site and inter-site differences (p ) from L2 L3 to L4 L5. As was evident for the SAT values, the between-subject variability in VAT values was also large, as indicated by the large SDs. The mean VAT/SAT and TAT values are also given in Table 1 and are shown in Figure 2 for each of the nine slices. At L2 L3, there were no intra-site differences (p 0.7) in the VAT/SAT ratio. However, for the remaining six slices from the superior position at L3 L4 to the inferior position at L4 L5, the mean VAT/SAT ratios decreased significantly (p 0.001), indicating both intra-site and inter-site differences. The mean VAT/SAT ratio at L2 L3 was 1.8, or double the 0.9 value observed for L4 L5. As observed for the VAT and SAT values separately, the between-subject variability for the VAT/SAT ratio was also large. A summary of the results for the pairwise comparisons of adjacent slices at the same intervertebral location is given in Table 2. The mean differences between adjacent slices (superior vs. middle and middle vs. inferior) are provided, along with their 95% confidence intervals. These differences represent the average error (bias) that would be introduced if an intervertebral scan was inaccurately positioned 7 mm off center from the true intervertebral location, with the 95% confidence interval describing the uncertainty in the estimated bias. For example, at L2 L3, the mean SAT difference was 6.2 cm 2 between the middle and inferior SAT (cm 2 ) L2-L3 0.6 ( 3.3; 2.1) 6.2 ( 8.9; 3.5)* L3 L4 4.3 ( 7.3; 1.3)* 8.7 ( 11.8; 5.7)* L4 L5 6.8 ( 12.6; 1.0)* 8.4 ( 14.2; 2.6)* VAT (cm 2 ) L2 L3 0.1 ( 5.6; 5.7) 10.7 ( 16.3; 5.0)* L3 L4 7.1 (1.2; 12.9)* 0.4 ( 5.4; 6.3) L4 L5 7.5 (0.9; 14.1)* 6.3 ( 0.3; 12.9) TAT (cm 2 ) L2 L3 0.5 ( 7.4; 6.4) 16.9 ( 23.8; 9.9)* L3 L4 2.8 ( 4.0; 9.5) 8.3 ( 15.1; 1.6)* L4 L5 0.7 ( 8.4; 9.9) 2.1 ( 11.3; 7.0) VAT/SAT L2 L ( 0.09; 0.05) 0.02 ( 0.04; 0.10) L3 L (0.01; 0.17)* 0.13 (0.05; 0.20)* L4 L (0.02; 0.13)* 0.10 (0.05; 0.15)* SD, standard deviation; SAT, subcutaneous adipose tissue; VAT, visceral adipose tissue; TAT, total adipose tissue. * Difference was statistically significant (p 0.03 to 0.001). slices, with a 95% confidence interval from 8.9 to 3.5 cm 2 ; the difference was statistically significant. The corresponding values for the superior vs. middle slice at L2 L3 are 0.6 cm 2, with a 95% confidence interval from 3.3 to 2.1 cm 2. More than one half of the pairings given in Table 2 showed statistically significant differences (p ) between adjacent slices, although the adjacent slices were only 7 mm apart. Thus, there would be a bias introduced if an adjacent scan was substituted for the middle slice or if a follow-up scan, such as in a longitudinal study, was 7 mm displaced from the previous scan. Discussion We observed between-site differences for SAT, VAT, and the VAT/SAT ratio among the three intervertebral locations examined using single-slice CT scans. Thus, our first hypothesis was verified, in that there were significant differences in the subcutaneous and visceral adiposity estimates among these three intervertebral positions (Figure 1). However, contrary to our second hypothesis, where we expected no significant differences in the VAT/SAT ratio, 1444 OBESITY Vol. 15 No. 6 June 2007

5 there were clear trends indicating differences in the relative changes in SAT and VAT among the three intervertebral locations, as well as for adjacent slices at the L3 L4 and L4 L5 locations. Hence, our second hypothesis, that the VAT/SAT ratio would be relatively constant across the sites, was not valid, because this ratio changed by a factor of 2 (Figure 2) between the L2 L3 and L4 L5 abdominal positions. Furthermore, our third hypothesis was also found to be invalid, because our results clearly showed significant differences between adjacent slices in at least 3 of 4 adipose tissue measurements at each location (Table 2). That is, a small inaccuracy in selection of the axial slice location, relative to the true intervertebral position, may introduce significant errors in the adiposity estimates. An unexpected result was that TAT remained relatively constant for at least seven of the nine locations (Figure 2). This would imply that any of the sites would be acceptable for estimating total abdominal adiposity but not the VAT/SAT ratio. Taken together, these findings do not support the interchangeability of SAT, VAT, or VAT/SAT values among the three lumbar spine positions we examined. These differences, therefore, will limit comparisons among studies where different abdominal sites were scanned. It is also clear that precise repositioning is needed for the assessment of changes in abdominal adiposity for the individual, such as in longitudinal studies. In our review of the literature, we found that the L4 L5 location has been the most frequently used single-slice site for the measurement of intra-abdominal adiposity (9 15), although several studies have reported that the L2 L3 site may be more appropriate (20,21). In our subjects, the within-subject variability of VAT and SAT values at the L4 L5 location was the highest, indicating that this location was less reliable than L2 L3 or L3 L4. Other investigators have also argued that a single-slice scan obtained at L4 L5 may have limited accuracy for the assessment of fat distribution in the abdomen (15,16,18,19). Our observation that the VAT/SAT ratio changed significantly across and within the three abdominal sites, whereas TAT remained relatively constant among the same sites, confirms this claim. When Shen et al. (15,22) re-examined using the L4 L5 location for single-slice measurements, they determined that positions higher on the abdomen (5 cm superior for women and 10 cm superior for men relative to the L4 L5 position) were better choices for assessing abdominal adiposity. If one takes into consideration the physical size of a single vertebra for an adult and the usual intervertebral spacing in the lumbar region, their new locations correspond closely to the L2 L3 position that we measured in our study. Greenfield et al. (16) used four equally spaced CT slices between the midpoints of L2 and L4 to examine the intra-subject variability in abdominal adiposity. These investigators reached the conclusion that a single-slice CT scan for the assessment of abdominal adiposity has limited application. Similar concerns have been reported in a weight loss study involving women on an exercise program, where the relative reduction in adiposity values at L2 L3 were substantially higher than at L4 L5 (23,24). Although not a specific focus of our study, others have also reported limitations when using a single-slice CT protocol to monitor the superficial vs. deeper layers within the subcutaneous adiposity region (25,26). There are several limitations of this study. Some HIVpositive adults can have an abnormal abdominal fat distribution (27 29). This could limit the translation of our findings to HIV-negative adults. As noted in the Research Methods and Procedures section, 1 of our original 27 subjects had several CT images in which the subcutaneous fat layer was so thin that it could not be accurately traced. It is highly likely that this subject may have had HIV-related lipodystrophy, which is characterized by a major loss of subcutaneous fat (27 29). This type of anatomical limitation would be a concern whenever the boundary of the subcutaneous fat layer has to be traced, including in the use of commercially available software for the automated analyses (30). Another limitation is that we had only four women in our study; thus, we could not adequately address the possibility of sex differences in adiposity distribution. Finally, because we obtained only a cross-sectional sampling of our population, we cannot say with certainty how these general patterns might change with time for the individual. For the 20 men and the 4 women in our study, their subcutaneous and visceral fat values were within the range reported by others (19,31) for the general U.S. adult population. In a study by Kuk et al. (19) for 85 adult men, the average VAT was 130 cm 2 at L4 L5 compared with cm 2 for the men in our study. At L2 L3, the value of Kuk et al. for VAT increased to 180 cm 2, while the corresponding value for our men was cm 2. For SAT, the corresponding paired values were 220 vs cm 2 at L4 L5 and 180 vs cm 2 at L2 L3. Although the 80- to 90-cm 2 differences in the mean SAT values could possibly reflect pre-clinical lipodystrophy in some of our men, it is important to note that this same difference was reported by Kuk et al. (19) when they separated their subjects into those with and without the metabolic syndrome. Stanforth et al. (31) reported single-slice CT measurement of VAT and SAT at L4 L5 for 692 adults, 17 to 65 years of age. At the same age and level of adiposity, black men and women had less VAT than white men and women, with the difference being greater in men than in women. Compared with our subjects and those of Kuk et al. (19), the adiposity values reported by Stanforth et al. (31) were, on average, 50 cm 2 lower for VAT and 80 to 120 cm 2 higher for VAT. These differences may be related to the older ages of our subjects and to differences in the ethnic and sex percentages. OBESITY Vol. 15 No. 6 June

6 In summary, we showed that estimates of abdominal adiposity based on a single-slice CT scan were highly site specific. This implies that single-slice measurements for SAT, VAT, or the VAT/SAT ratio cannot be easily generalized to other sites. Although the data suggest that the L2 L3 position may be better than L4 L5, we still recommend that a multi-slice protocol should be used whenever possible. Furthermore, our findings clearly show that precise positioning for the CT scans is needed for both intersubject comparisons and longitudinal monitoring of intrasubject changes. Therefore, it seems prudent to monitor multiple sites when assessing abdominal subcutaneous or visceral adiposity. Acknowledgments This work was supported by National Institute of Allergy and Infectious Diseases Grants U01AI and U01AI46362 and the U.S. Department of Agriculture under cooperative agreement with Baylor College of Medicine. References 1. Olshansky SJ, Passaro DJ, Hershow RC, et al. A potential decline in life expectancy in the United States in the 21st century. N Engl J Med. 2005;352: Wang Y, Rimm EB, Stampfer MJ, Willett WC, Hu FB. Comparison of abdominal adiposity and overall obesity in predicting risk of type 2 diabetes among men. Am J Clin Nutr. 2005;81: Nicklas BJ, Penninx BW, Cesari M, et al. Association of visceral adipose tissue with incident myocardial infarction in older men and women: the Health, Aging and Body Composition Study. Am J Epidemiol. 2004;160: Onat A, Avci GS, Barlan MM, Uyarel H, Uzunlar B, Sansoy V. Measures of abdominal obesity assessed for visceral adiposity and relation to coronary risk. Int J Obes Relat Metab Disord. 2004;28: Tirkes AT, Gottlieb RH, Voci SL, Waldman DL, Masetta J, Conover DL. Risk of significant coronary artery disease as determined by CT measurement of the distribution of abdominal adipose tissue. J Comput Assist Tomogr. 2002;26: Martin AD, Janssens V, Caboor D, Clarys JP, Marfell- Jones MJ. Relationships between visceral, trunk and wholebody adipose tissue weights by cadaver dissection. Ann Hum Biol. 2003;30: Janssens V, Thys P, Clarys JP, et al. Post-mortem limitations of body composition analysis by computed tomography. Ergonomics. 1994;37: Mitsiopoulos N, Baumgartner RN, Heymsfield SB, Lyons W, Gallagher D, Ross R. Cadaver validation of skeletal muscle measurement by magnetic resonance imaging and computerized tomography. J Appl Physiol. 1998;85: Grauer WO, Moss AA, Cann CE, Goldberg HI. Quantification of body fat distribution in the abdomen using computed tomography. Am J Clin Nutr. 1984;39: Kvist H, Sjöström L, Tylen U. Adipose tissue volume determinations in women by computed tomography: technical considerations. Int J Obes. 1986;10: Kvist H, Chowdhury B, Sjöström L, Tylen U, Cederblad A. Adipose tissue volume determination in males by computed tomography and 40K. Int J Obes. 1988;12: Katznelson L, Rosenthal DI, Rosol MS, et al. Using quantitative CT to assess adipose distribution in adult men with acquired hypogonadism. AJR Am J Roentgenol. 1998;170: Sumner AE, Farmer NM, Tulloch-Reid MK, et al. Sex differences in visceral adipose tissue volume among African Americans. Am J Clin Nutr. 2002;76: Borkan GA, Gerzof SG, Robbins AH, Hults DE, Silbert CK, Silbert JE. Assessment of abdominal fat content by computed tomography. Am J Clin Nutr. 1982;36: Shen W, Punyanitya M, Wang Z, et al. Visceral adipose tissue: relations between single-slice areas and total volume. Am J Clin Nutr. 2004;80: Greenfield JR, Samaras K, Chisholm DJ, Campbell LV. Regional intra-subject variability in abdominal adiposity limits usefulness of computed tomography. Obes Res. 2002;10: Lee SJ, Janssen I, Ross R. Interindividual variation in abdominal subcutaneous and visceral adipose tissue: influence of measurement site. J Appl Physiol. 2004;97: Warren M, Schreiner PJ, Terry JG. The relation between visceral fat measurement and torso level is one level better than another? The Atherosclerosis Risk in Communities Study, Am J Epidemiol. 2006;163: Kuk JL, Blair SN, Church TS, Ross R. Does measurement site for visceral and adominal subcutaneous adipose tissue alter associations with the metabolic syndrome? Diabetes Care. 2006;29: Han TS, Kelly IE, Walsh K, Greene RM, Lean ME. Relationship between volumes and areas from single transverse scans of intra-abdominal fat measured by magnetic resonance imaging. Int J Obes Relat Metab Disord. 1997;21: Abate N, Garg A, Coleman R, Grundy SM, Peshock RM. Prediction of total subcutaneous abdominal, intraperitoneal, and retroperitoneal adipose tissue masses in men by single axial magnetic resonance imaging slice. Am J Clin Nutr. 1997;65: Shen W, Punyanitya M, Wang Z, et al. Total body skeletal muscle and adipose tissue volumes: estimation from a single abdominal cross-sectional image. J Appl Physiol. 2004;97: Thomas EL, Saeed N, Hajnal JV, et al. Magnetic resonance imaging of total body fat. J Appl Physiol. 1998;85: Thomas EL, Bell JD. Influence of undersampling on magnetic resonance imaging measurements of intra-abdominal adipose tissue. Int J Obes Relat Metab Disord. 2003;27: Kelley DE, Thaete FL, Troost F, Huwe T, Goodpaster BH. Subdivisions of subcutaneous abdominal adipose tissue and insulin resistance. Am J Physiol Endocrinol Metab. 2000;278: E OBESITY Vol. 15 No. 6 June 2007

7 26. Deschenes D, Couture P, Dupont P, Tchernof A. Subdivision of the subcutaneous adipose tissue compartment and lipid-lipoprotein levels in women. Obes Res. 2003;11: Dinges WL, Chen D, Snell PG, Weatherall PT, Peterson DM, Garg A. Regional body fat distribution in HIV-infected patients with lipodystrophy. J Investig Med. 2005; 53: Carr A, Samaras K, Thorisdottir A, Kaufmann GR, Chisholm DJ, Cooper DA. Diagnosis, prediction, and natural course of HIV-1 protease-inhibitor-associated lipodystrophy, hyperlipidaemia, and diabetes mellitus: a cohort study. Lancet. 1999;353: Saint-Marc T, Partisani M, Poizot-Martin I, et al. A syndrome of peripheral fat wasting (lipodystrophy) in patients receiving long-term nucleoside analogue therapy. AIDS. 1999; 13: Liou TH, Chan WP, Pan LC, Lin PW, Chou P, Chen CH. Fully automated large-scale assessment of visceral and subcutaneous abdominal adipose tissue by magnetic resonance imaging. Int J Obes (Lond). 2006;30: Stanforth PR, Jackson AS, Green JS, et al. Generalized abdominal visceral fat prediction models for black and white adults age y: the HERITAGE Family Study. Int J Obes Relat Metab Disord. 2004;28: OBESITY Vol. 15 No. 6 June