International Journal of Obesity (2007) 31, 147–152 & 2007 Nature Publishing Group All rights reserved 0307-0565/07 $30.00 www.nature.com/ijo
ORIGINAL ARTICLE Regional differences in abdominal fat loss JA Kanaley, I Giannopoulou and LL Ploutz-Snyder Department of Exercise Science, Syracuse University, Syracuse, NY, USA Objective: This study determined if the magnetic resonance imaging (MRI) protocol used alters the estimation of change in abdominal fat with weight loss in obese type 2 diabetic women. This study also examined if there is a uniform fat loss across the abdomen. Methods and procedures: Thirty-three obese postmenopausal women with type 2 diabetes (age 50–70 years, body mass index430 kg/m2) had a total abdominal MRI scan pre- and post weight loss intervention. Three different MRI analysis protocols were used and compared: a single slice at L2–L3 vs five slices (centered at L4–L5) vs all abdominal slices. In addition, the total abdominal scan was divided into four regions (four slices each) with region 3 (critical region) including the traditionally studied L2–L3, and regions 1 and 2 superior and region 4 inferior to critical region 3. Analysis of variance (ANOVA) with repeated measures was used to compare the influence of weight loss on abdominal fat measured both regionally and using the varying number of MR slices. Results: At baseline, the ratio of visceral adipose tissue:subcutaneous adipose tissue (VAT:SAT) was significantly lower using the single-slice method compared to five slices and the total abdomen (Po0.01). Using the single-slice method, a lower %VAT was found than with the other methods (Po0.01). In regions 1, 2, 3, and 4, the absolute change in total fat was 122750, 182748, 182755, and 155740 cm3, respectively. The regional difference in abdominal fat patterning revealed that the critical region (region 3) had a smaller VAT:SAT ratio than regions 1 and 2 (Po0.05), and the ratio at region 4 was smaller than region 3 (Po0.05). Weight loss resulted in a decrease in the VAT:SAT ratio (Po0.05) for regions 3 and 4 but not for regions 1 and 2. Conclusions: The number of MR slices analyzed yields differential result in relative VAT distribution. Regional differences in abdominal fat loss occur with a greater relative VAT loss in the critical region, thus if only the critical region is analyzed the overall VAT loss induced by weight loss intervention may be overestimated. International Journal of Obesity (2007) 31, 147–152. doi:10.1038/sj.ijo.0803359; published online 25 April 2006 Keywords: intra-abdominal fat; single slice; multiple-slice scanning; MRI
Introduction Quantification of abdominal fat and its regional distribution has been of considerable interest in the past decade because of its link to numerous health complications. Although there are many methods to measure total body fat, computed tomography (CT) and magnetic resonance imaging (MRI) are currently the best methods for analyzing regional fat distribution. Numerous protocols using CT and MRI have been developed in an attempt to quantify the abdominal fat, yet not exposing the subject to excessive radiation and/or to minimize the cost. These protocols include whole body imaging, multi-slice imaging of the entire abdominal area, or a single slice at a predetermined area of the abdomen (L2–L3 or L4–L5).1–5 Recently, Thomas and Bell6 reported that single-
Correspondence: Dr J Kanaley, Exercise Science, 820 Comstock Ave, Rm. 201, Syracuse University, Syracuse, NY 13244, USA. E-mail:
[email protected] Received 29 November 2005; revised 16 February 2006; accepted 5 March 2006; published online 25 April 2006
slice images may be less likely to detect small changes in abdominal adiposity or may require a larger cohort of subjects to detect similar changes than multi-slice images of the total abdomen. Further, when comparing relative fat distribution, the assumption is made that the relative proportion of fat at different levels of the abdomen is consistent between subjects. Greenfield et al.7 quantified the variability between adjacent abdominal levels and reported that the pattern of variability was not uniform between subjects. Moreover using the single-slice method, the assumption is made that during weight loss, all areas of the abdomen will lose fat in the same relative proportion, and this assumption may be misleading. Thus, the purpose of this study was to examine and compare the changes in abdominal fat with weight loss as determined by three commonly used MRI protocols: total abdominal MRI vs multi-slice MRI vs single-slice MRI. In addition, this study examined if the pattern of variability was uniform between subjects at the adjacent abdominal levels and if the fat loss following an intervention was similar between these levels. We hypothesized that the various MRI
Abdominal fat loss JA Kanaley et al
148 protocols would result in different changes in fat loss and that the adjacent regions of the abdomen would lose fat differently.
Materials and methods Subjects Thirty-three obese women with type 2 diabetes, body mass index (BMI)430 kg/m2, participated in a larger study examining abdominal fat loss with diet and/or exercise,8 and their MRI scans of the abdomen were used for this analysis. The women signed an informed consent approved by both the Syracuse University and SUNY Upstate Medical University Institutional Review Boards. All women were postmenopausal (50–70 years old), for a minimum of 1 year (8.971.3 years), obese, and diagnosed with type 2 diabetes for at least 1 year (3.070.6 years). None of the women were on thiazolidinediones, insulin, or b-blockers. Oral hypoglycemic agents were used by 22 of the subjects, whereas 11 women were not taking any medication for diabetes. Ten of the women were on hormone replacement therapy (HRT) (7.672.5 years). All women had been weight stable for at least 6 months, were in good health with no major complications related to diabetes, such as cardiovascular disease and neuropathies. During the course of study participation, there were no changes in the subjects’ medications. More details regarding subject selection and methodology have been previously published in detail.8
Experimental design The subjects had an MRI scan of the whole abdomen at the beginning and end of a 14-week weight loss intervention. The women were randomly assigned to one of three interventions, diet alone, exercise alone, and diet þ exercise. The women also had total body fat measured by either underwater weighing or the BodPod.8
Abdominal fat measurements Abdominal fat distribution was determined by MRI with techniques previously reported.9 To reiterate, using a GE Sigma 1.5T MRI scanner, the MR images were obtained with standard T1-weighted spin-echo imaging with respiratory compensation. The sequence parameters used were time relaxation (TR) ¼ 400 ms, time echo (TE) ¼ 20 ms, field of view (FOV) ¼ 40–48, acquisition matrix ¼ 256 256, with nex ¼ 2. Approximately 35 consecutive, 1 cm slices per total scan were acquired. Initially, a scout scan was conducted to determine the specific region of interest, and then scanning started at the superior portion of the head of the femur and including the most superior portion of the kidneys, which was B25 cm in most subjects depending on their height. At each MR image slice level, fat pixel intensity was segmented and used to determine the total volume of fat for the slice. International Journal of Obesity
Visceral fat volume was also determined and then subtracted from the total fat visceral adipose tissue (VAT) plus subcutaneous adipose tissue (SAT) of that slice to calculate the total subcutaneous fat of that slice. Summation of all the 1 cm slices provided total abdominal fat volume. The digital image data were analyzed using a fat segmentation program running on a SUN workstation (SUN Microsystems, Santa Clara, CA, USA).8 Test–retest for this method was r ¼ 0.9999 (Po0.0001).10
Data analysis The adipose tissue volume for the whole abdominal region was calculated by summating the volumes of each slice using the slices between the top of the kidneys to the top of the head of the acetabulum. The single-slice analysis was conducted using the slice at L2–L3, whereas the five-slice analysis included L4–L5 as well as two slices above this point and two slices below this point (Figure 1).3 To conduct an analysis of the fat loss in various abdominal areas, the abdomen was divided into four regions with four slices each. Region 3 was considered the critical region as it contained L2–L3. Regions 1 and 2 were superior to this region and region 4 was inferior to this region. The ratio between VAT and SAT volumes (VAT:SAT) was calculated to determine if the method of analysis (all slices of the abdomen vs L2–L3 single slice vs five slices at L4–L5) would produce different results. Both absolute abdominal fat volume and relative VAT volume (%VAT) were calculated. The data were log transformed and a Bland–Altman plot was used to highlight the differences between methods and total error was calculated. A 2 3 analysis of variance (AVOVA) with repeated measures (time protocol) was used to determine if differences existed between the three methods for both the pre- and post data. For comparison of the various regions, a 2 4 ANOVA with repeated measures (time by region) was used to compare the change in fat loss by region. The data are expressed as mean7s.e. SPSS (ver13.0) was used for the data analysis.
Results The mean age of the subjects was 57.170.98 years, the mean weight 91.673.3 kg, and the mean BMI of 34.671.1 kg/m2. Their percent body fat was 44.571.4%. On average, these women had type 2 diabetes for 3.470.6 years. As a result of the diet and/or exercise intervention, there was an B4.5 kg weight loss. Exercise alone resulted in o2 kg weight loss.8 We examined the ratio of VAT to SAT volume as measured by each MRI analysis method at baseline. There was a significant difference by method such that the single-slice method gave a lower VAT:SAT ratio than using all slices or five slices (Po0.01, Figure 2). There was a trend for the main effect of time (P ¼ 0.08), such that with five slices and single slice, the VAT:SAT ratio decreased with the intervention but
Abdominal fat loss JA Kanaley et al
149
Figure 1 Scout scan of the abdominal region indicating the various regions used for the single slice, five slices, and all slices of the abdomen.
Figure 2 The change in the visceral adipose tissue (VAT): subcutaneous adipose tissue (SAT) ratio by magnetic resonance imaging (MRI) protocol (all slices vs five slices at L4–L5 vs single slice at L2–L3) and over time. *Po0.001 vs the all slice and five slices protocols.
Figure 3 The comparison of the relative distribution of visceral adipose tissue (VAT) by magnetic resonance imaging (MRI) protocol. P ¼ 0.000 significant effect of protocol; *P ¼ 0.02 all slice vs single slice at L2–L3; w P ¼ 0.000 vs single slice at L2–L3.
not when all slices were included in the analysis. Further examining the relative distribution of VAT and SAT with each method (e.g. %VAT or %SAT), a significant main effect of method (Po0.01) was found. Again, the method using all slices and five slices gave a higher percent VAT compared to the single slice method (Figure 3). Examining the differences between the three methods in detecting the changes in the %VAT after weight loss revealed no differences between the methods for VAT, but a significant difference was found
between the three methods for SAT. Further highlighting the differences between the methods, the Bland–Altman plots demonstrate that there is a 73.9% coefficient of variance (CV) between the log of the difference between all slices and five slices and the mean of the five slices, with a total error (log) of 3.47 cm3 (Figure 4). The CV between the log of the difference between all slices and single slice was 75.4% and the total error (log) was 2.86 cm3. International Journal of Obesity
Abdominal fat loss JA Kanaley et al
150
Figure 4 Bland–Altman plots comparing the log total abdominal fat (all slices) with protocols that only use single slice or five slices.
It is theoretically possible that regional differences in the amount of fat loss can occur. Figure 5 shows the absolute volumes of total abdominal fat, VAT, and SAT by region. There was a significant main effect for (Po0.01) region and time for total abdominal fat and SAT volume, whereas a significant time by region interaction was found for VAT (Po0.01). Region 3 was significantly different (Po0.01) from regions 2 and 4, and region 4 was significantly different than all other regions (Po0.01). At baseline, a significant regional difference was found in the ratio of VAT:SAT (Po0.001), such that there was a larger ratio of VAT:SAT volume for regions 1 and 2 than regions 3 and 4 (Figure 6). The critical region (region 3) had a higher VAT:SAT ratio (Po0.05) than region 4. Further, a time effect was found (Po0.05), such that a significantly lower ratio was found post weight loss for both regions 3 and 4 but not for regions 1 and 2. Expressed as a percent change over time, VAT volume in regions 3 (the critical region) and 4 showed a significantly greater decrease than seen in regions 1 and 2 (Po0.01) (Table 1). There was no difference between regions 3 and 4.
Figure 5 The absolute volumes of total abdominal fat, visceral fat (VAT), and subcutaneous fat (SAT) by region. Region 3 is the critical region at L4–L5. w Po0.01 regions 2,4 from region 1; yPo0.01 regions 1,3,4 from region 2; z Po0.001 regions 2,4 from region 3; Po0.001 regions 1,2,3 from region 4.
Figure 6 The comparison of the ratio of visceral adipose tissue (VAT):
Discussion Previous literature6 has suggested that MRI analysis using a single slice is less likely to detect changes in abdominal fat loss pre- and post intervention compared to analysis of all abdominal slices. Consistent with previous literature,6 we International Journal of Obesity
subcutaneous adipose tissue (SAT) by abdominal region pre- and post intervention. zPo0.05 significant time effect (pre vs post); *Po0.001 vs regions 1 and 2; wPo0.05 vs region 3.
found that the single-slice analysis method was not in agreement with the multiple-slice method in the determination of content and the loss of abdominal fat with an
Abdominal fat loss JA Kanaley et al
151 Table 1 Percent change (post vs pre) by region for total abdominal fat, visceral fat and subcutaneous fat (n ¼ 33) Region 1
Region 2
Region 3
Region 4
Total abdominal fat 3.1872.83 6.3972.20 6.1072.35 6.8571.42 Visceral fat* 2.5572.58 6.0973.41 11.2072.23** 9.99872.16** Subcutaneous fat 3.0673.49 5.6572.17 3.0472.99 4.9971.40 *Po0.01 significant main effect by region; **Po0.01 vs region 1. Region 3 – critical region at L2–L3..
intervention. Also, this study is the first to demonstrate that the fat loss from the abdominal region is not uniform and that greater relative VAT loss was found in the lower regions of the abdomen than was found in the higher regions of the abdomen. As there is no consistency in the change in fat loss across regions, weight loss studies using only single slice or five slices may draw different conclusions than studies that examine the whole abdominal region. Considerable research has been conducted comparing the multi-slice to single-slice images. Different researchers have shown that a strong predictive ability of the abdominal fat depot can be achieved by using a slice at L2–L31,11 or at L4–L5; and using fewer slices to minimize radiation exposure to the subject and minimize costs.2,3 Abate et al.11 reported that the slice at the level of L2–L3 contained the highest amount of intra-peritoneal and retroperitoneal adipose tissue and was the best predictor of total intra-abdominal fat volume. Kvist et al.2 observed that a reduction in the number of images used to calculate total adipose tissue volume resulted in only a small error, which was consistent with the findings of Ross et al.3 Recently, Shen et al.5 using a large sample of subjects noted the highest correlation between VAT area and the volume B5–10 cm above L4–L5. The L4–L5 level is popular among many investigators because it is at the level of the umbilicus and is the same location as the waist measurement site.2 The landmarks of this region are easier to locate particularly when multiple scans over time are required. In many of the previous studies, strong correlations have been found when comparing the single slice to multiple slice of the total abdomen but using these two variables in the same regression model results in collinearity of the data, thus potentially inflating the correlations. Our data using single and multiple slices correlated well with total abdominal fat (r ¼ 0.96, r ¼ 0.91, respectively; Po0.01); however, using the ratio of VAT:SAT highlighted the differences between the methods. The single-slice method resulted in a lower VAT:SAT ratio compared to the other methods used, and hence using the single-slice method may cause spurious interpretation. The assumption that the distribution of SAT and VAT in the abdomen is similar between individuals is made when using single-slice MRI, such that quantification at one level of the abdomen (L4–L5) is the same as that obtained at another level of the abdomen (L2–L3). Our data demonstrate that although the largest volume of abdominal fat was found in the critical region, this region actually had a greater SAT
volume than the other regions. The critical region had a smaller absolute VAT than in regions 1 and 2. Likewise, previous evidence has shown the assumption of a uniform VAT distribution to be incorrect. Han et al.1 demonstrated in men that intra-abdominal fat areas were higher in the upper scans in the L1–L5 region than in the lower scans of this region. Lee et al.12 noted that the ranking of men for abdominal SAT and VAT deposition was dependent on the level of the abdomen at which these depots were measured. Yet, these authors reported that using single images at either L4–L5 or L3–L4 was comparable despite interindividual variation in ranking the ability to predict SAT and VAT mass. Greenfield et al.7 demonstrated marked intra-subject variability in cross-sectional CT-measured intra-abdominal fat and SAT areas by comparing results from four predetermined anatomically based lumbar set-points. They found an intrasubject variability ranging from 8 to 61% with greater variability when the area of intra-abdominal fat was expressed as a percentage of the total intra-abdominal area (19–124%). In agreement with our findings, Thomas and Bell6 demonstrated that erroneous conclusions may result when using a single slice of the abdomen compared to multiple slices when studying individuals over time. They found that despite the strong correlation between single slice and all slices of the total abdomen, that single slice methods may be less likely to detect small changes in abdominal adiposity or require a larger cohort of subjects to detect similar changes. The present study is the first to demonstrate that abdominal fat loss with a diet and/or exercise intervention varies by region and the selection of only one region may lead to spurious conclusions. A greater decrease (10%) in the critical region and region 4 was observed compared to a smaller fat loss found in regions 1 and 2. Thus if only this region was examined pre/post weight loss, an overestimation of VAT loss would occur. A study using all the slices in the abdomen would report smaller losses because of the inclusion of the regions, which do not lose VAT at the same rate. Interestingly, this greater relative loss of VAT occurred despite a lower absolute VAT content in regions 3 and 4 compared to the regions superior to it. Therefore, although using just one slice may reduce radiation exposure (e.g. CT scanning) and reduce data analysis time of the slices, it may be misleading about the amount of the VAT loss with an intervention. This may explain the existing controversy in the literature with regards to the relationship between fat loss and insulin sensitivity or changes in cardiovascular risk factors. It should also be noted that some of the women were on HRT in this study. Hormone replacement therapy use has been implicated in causing differences in abdominal fat gain,13 thus it is potentially possible that this may also impact fat loss. We found no differences from the reported results when we adjusted for HRT use; however, it should be noted that the mean weight loss was B4.5 kg, and more weight loss may be needed to see an effect of HRT. Further, these results are applicable to postmenopausal women with International Journal of Obesity
Abdominal fat loss JA Kanaley et al
152 type 2 diabetes, but we speculate that they are applicable to men and to non-diabetic individuals. In the present study, the women responded with a decrease in abdominal fat with exercise alone (and minimal weight loss) which is similar to earlier findings by Ross et al.14,15 in obese men. As men frequently have more visceral fat than women, it is possible that the pattern of fat loss may be even more disparate in men than women. Further, Han et al.1 observed that intraabdominal fat areas were higher in the upper region of the L1–L5 area than the lower region in men, whereas in women the intra-abodminal fat areas were similar at all levels. Taken together with the earlier studies, we conclude that MRI measurements of total abdominal fat provides a more accurate assessment of abdominal fat content and of fat loss over time than single-slice measurements. The changes in fat loss in the abdominal region are not uniform and the VAT loss, induced by the intervention, may be overestimated if only critical region 3 is considered. Thus using only a few abdominal samples may not provide an accurate reflection of changes in abdominal fat loss with an intervention and demonstrate the need for whole abdominal imaging to provide a complete assessment of the abdominal fat.
Acknowledgements This study was partially funded by the NY State Diabetes Bridge Grant, the ACSM Graduate Student Research Grant, and the Woodrow Wilson Research Grant.
References 1 Han T, Kelly I, Walsh K, Greene R, Lean M. 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: 1161–1166. 2 Kvist H, Sjostrom L, Tylen U. Adipose tissue volume determinations in women by computed tomography: technical considerations. Int J Obes Relat Metab Disord 1986; 10: 53–67.
International Journal of Obesity
View publication stats
3 Ross R, Leger L, Morris D, de Guise J, Guardo R. Quantification of adipose tissue by MRI: relationship with anthropometric variables. J Appl Physiol 1992; 72: 787–795. 4 Ross R, Shaw K, Martel Y, de Guise J, Avruch L. Adipose tissue distribution measured by magnetic resonance imaging in obese women. Am J Clin Nutr 1993; 57: 470–475. 5 Shen W, Punyanitya M, Wang Z, Gallagher D, St-Onge M, Albu J et al. Total body skeletal muscle and adipose tissue volumes: estimation from a single abdominal cross-sectional image. J Appl Physiol 2004; 97: 2333–2338. 6 Thomas E, Bell J. Influence of undersampling on magnetic resonance imaging measurements of intra-abdominal adipose tissue. Int J Obes Relat Metab Disord 2003; 27: 211–218. 7 Greenfield J, Samaras K, Chisholm D, Campbell L. Regional intrasubject variability in abdominal adiposity limits usefulness of computed tomography. Obes Res 2002; 10: 260–265. 8 Giannopoulou I, Ploutz-Snyder L, Carhart R, Weinstock R, Fernhall B, Goulopoulou S et al. Exercise is required for visceral fat loss in postmenopausal women with type 2 diabetes. J Clin Endocrinol Metab 2005; 90: 1511–1518. 9 Kanaley JA, Giannopoulou I, Tillapage-Fay G, Nappi J, PloutzSnyder LL. Racial differences in subcutaneous and visceral fat distribution in postmenopausal black and white women. Metabolism 2003; 52: 186–191. 10 Kanaley J, Sames C, Swisher L, Swick A, Ploutz-Snyder L, Steppan C et al. Abdominal fat distribution in pre- and postmenopausal women: The impact of physical activity, age and menopausal status. Metabolism 2001; 50: 976–982. 11 Abate N, Garg A, Coleman R, Grundy S, Peshock R. 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: 403–408. 12 Lee S, Janssen I, Ross R. Interindividual variation in abdominal subcutaneous and visceral adipose tissue: influence of measurement site. J Appl Physiol 2004; 97: 948–954. 13 Haarbo J, Marslew U, Gotfredsen A, Christiansen C. Postmenopausal hormone replacement therapy prevents central distribution of body fat after menopause. Metabolism 1991; 40: 1323– 1326. 14 Ross R, Dagnone D, Jones P, Smith H, Paddags A, Hudson R et al. Reduction in obesity and related comorbid conditions after diet-induced weight loss or exercise-induced weight loss in men: a randomized controlled trial. Ann Intern Med 2000; 1333: 92–103. 15 Ross R, Janssen I, Dawson J, Kungle A-M, Kuk J, Wong S et al. Exercise-induced reduction in obesity and insulin resistance in women: a randomized controlled trial. Obes Res 2004; 12: 789–798.
