Effects of Pioglitazone Versus Diet and Exercise on Metabolic Health and Fat Distribution in Upper Body Obesity

Metabolic Syndrome/Insulin Resistance Syndrome/Pre-Diabetes ORIGINAL ARTICLE

SAMYAH SHADID, MD regional variability in TZD responsive- MICHAEL D. JENSEN, MD ness has been demonstrated; preadipocytes from subcutaneous fat differentiate more in response to TZD in vitro than visceral adipose tissue (6). If the same phenomenon occurs in vivo, one would OBJECTIVE — Insulin resistance is associated with visceral adiposity, and interventions that expect selective adipocyte proliferation reduce this depot, e.g., diet and exercise, improve insulin resistance. Thiazolidinediones (TZDs) and thus body fat redistribution. also improve insulin action but paradoxically increase total fat mass, perhaps through remod- However, studies of animals and dia- eling (recruitment of smaller fat cells) and redistribution of adipose tissue. We assessed the effects betic humans have reported increasing, of pioglitazone versus diet and exercise on fat distribution and the relationship between fat distribution and insulin sensitivity in upper body obesity. decreasing, and unchanged visceral and subcutaneous fat depots after TZD ad- RESEARCH DESIGN AND METHODS — Thirty-nine upper body obese, insulin- ministration despite improved insulin resistant, nondiabetic men and premenopausal women were randomly assigned to receive either sensitivity (3–5,7,8). This suggests that 30 mg/day pioglitazone or a diet and exercise program for 20 weeks. Before and after the TZD effects on visceral adiposity, if intervention, insulin sensitivity, body composition, body fat distribution (waist-to-hip ratio present in vivo, might not contribute to [WHR], computed tomography abdomen, and dual-energy X-ray absorptiometry), and abdom- their insulin sensitization. inal and femoral fat cell size were assessed. We assessed the effects of pioglitazone on fat distribution, fat cell size, and RESULTS Ϯ — Diet and exercise resulted in an 11.8 1.1 kg weight loss. Both diet and exercise the relationship between fat distribution and pioglitazone improved insulin sensitivity, but only the former was associated with loss of intra-abdominal fat. Pioglitazone increased total body fat, which preferentially accumulated in and insulin sensitivity in upper body obe- the lower body depot in both men and women. WHRs decreased in both groups. Abdominal fat sity, a known insulin-resistant state. cell size decreased (P ϭ 0.06) after diet and exercise. No statistically significant changes in fat cell Comparison of the effects of pioglitazone size were observed in pioglitazone-treated volunteers. with diet and exercise, a standard intervention to improve insulin sensitivity, CONCLUSIONS — In nondiabetic upper body obese subjects, increasing insulin sensitivity was performed to place the results in via diet and exercise accompanies reductions in visceral fat. Pioglitazone treatment also improves context. insulin sensitivity and lowers WHR, but this is due to a selective increase in lower body fat. This confirms a site-specific responsiveness of adipose tissue to TZD and suggests that improvements in insulin sensitivity by pioglitazone are achieved independent of changes in intra-abdominal fat. RESEARCH DESIGN AND METHODS — Written informed con- Diabetes Care 26:3148–3152, 2003 sent was obtained from 68 healthy upper body obese men and premenopausal women. Subjects were nondiabetic, sed- ody fat distribution is an important depot via diet, exercise, or surgery (1,2). entary, and weight stable up to at least 6 variable in the relationship between Thiazolidinediones (TZDs) are also months before entering the research pro- B overweight and insulin resistance. known to improve insulin sensitivity de- gram. Inclusion criteria consisted of a BMI Intra-abdominal or visceral fat accumula- spite paradoxically increasing total fat of 28–36 kg/m2 and one of the following tion is more strongly associated with in- mass (3–5). It has been suggested that re- three items: 1) waist-to-hip ratio (WHR) sulin resistance, and insulin resistance distribution of body fat may contribute to Ͼ0.85 (women) or Ͼ0.95 (men); 2) com- can be improved by decreasing this fat their insulin-sensitizing qualities. Indeed, puted tomography (CT)-measured vis- ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ceral fat area Ͼ120 cm2 (women) or From the Endocrine Research Unit, Mayo Clinic, Rochester, Minnesota. Ͼ180 cm2 (men); or 3) a ratio of visceral Address correspondence and reprint requests to Michael D. Jensen, MD, Endocrine Research Unit, 5-194 to total fat Ͼ0.30 (women) or Ͼ0.40 Joseph, Mayo Clinic, Rochester, MN 55905. E-mail: [email protected]. Received for publication 30 April 2003 and accepted in revised form 17 June 2003. (men). If WHR exceeded 0.80 (women) S.S. was sponsored in part by Novo Nordisk. M.D.J. has received grant support from Takeda Pharma- or 0.90 (men), and fasting glucose was ceuticals. between 100–126 mg/dl, volunteers were Abbreviations: CT, computed tomography; DEXA, dual-energy X-ray absorptiometry; FFM, fat-free included regardless of the CT results. Ex- mass; Si, insulin sensitivity index; TZD, thiazolidinedione; WHR, waist-to-hip ratio. clusion criteria were a history of coronary A table elsewhere in this issue shows conventional and Syste`me International (SI) units and conversion factors for many substances. heart disease, atherosclerosis, known © 2003 by the American Diabetes Association. systemic illness, renal or liver failure, clin- See accompanying editorial, p. 3184. ically diagnosed type 2 diabetes, hyper-

3148 DIABETES CARE, VOLUME 26, NUMBER 11, NOVEMBER 2003 Shadid and Jensen tension requiring medication that could not be safely stopped 2 weeks before the study, smoking, pregnancy, and breast- feeding. Thirteen volunteers were excluded on the basis of screening laboratory results, 10 volunteers withdrew before starting the program and 6 dropped out for a variety of reasons. One volunteer was excluded for noncompliance with the diet and exercise program.

Study protocol The remaining 39 volunteers underwent blood testing (complete blood count, chemistry panel, and lipid profile), an insulin-modified intravenous glucose tolerance test, CT measures of visceral fat area (L level) (9), and dual-energy X-ray ab- 2–3 Figure 1—Determination of fat compartments using DEXA and computed tomography. Lower sorptiometry (DEXA) (DPX-IQ; Lunar body fat is considered fat caudal to the inguinal ligaments; upper body fat is equal to total fat minus Radiation, Madison, WI) for body com- lower body fat. To estimate visceral fat mass, the DEXA and CT data are both used. Horizontal position assessment before and after the lines are used at the level of the symphysis pubis and the diaphragm on the DEXA image to intervention. Adipose tissue biopsies encompass abdominal fat. Arms are positioned during the scan to allow diagonal lines to be placed were taken from femoral and abdominal to separate them from the abdomen. The ratio of visceral-to-total abdominal fat from CT is subcutaneous areas. Oxygen consump- multiplied by abdominal fat from DEXA to estimate visceral fat. Upper body fat minus visceral fat ˙ is upper body nonvisceral fat. tion (VO2peak) and maximum heart rate were determined by a graded exercise test performed on a Quinton (Seattle, WA) motor-driven treadmill using a modified Assessment of fat compartment Assays Bruce protocol (10). Heart rate and volume The following assays were used: glucose: rhythms were monitored continuously The CT images were analyzed to distin- Hitachi 912 Chemistry Analyzer using the via a 10-lead electrocardiogram. The dif- guish compartmental fat volumes, as pre- hexokinase reagent (Boehringer Mann- ference between resting and maximal viously described (9). Results were heim, Indianapolis, IN) or the Beckman combined with data from DEXA (9) to Glucose Analyzer (Beckman Instruments, heart rate defined the heart rate reserve, calculate the following fat compartments: Fullerton, CA); insulin: chemiluminis- which was used to determine exercise in- total body fat, lower body fat, upper body cence method with Access Ultrasensitive tensity goals for those in the diet and ex- nonvisceral fat, and visceral (intra- Immunoenzymatic Assay system (Beck- ercise program (see below). abdominal) fat (Fig. 1). man, Chaska, MN); C-peptide: direct, After the baseline measurements, the double antibody sequential radioimmu- volunteers were randomized to receive 30 noassay (Linco Research, St. Louis, MO); mg pioglitazone daily or a diet and exer- Fat biopsies and fat cell size Subcutaneous fat was aspirated from fem- triglycerides: Hitachi 912 chemistry ana- cise program for 18–20 weeks. The pio- lyzer using Technicon triglyceride re- glitazone treated volunteers were moni- oral and abdominal depots under sterile conditions using local anesthesia. Fat was agent (Bayer, Tarrytown, NY). tored every 4 weeks for weight, liver immediately rinsed with saline and di- Intravenous glucose tolerance test. An function tests, and pill counts. Diet and intravenous injection of 0.33 g/kg dex- gested in a HEPES/collagenase solution ϭ exercise volunteers were instructed in a trose (D50)attime(t) 0 min was fol- (37°C; Sigma Type II C-6885). Adipo- ϭ 500-kcal deficit diet and in an exercise cytes were isolated (centrifuge) and lowed by 0.02 units/kg insulin at t 20 program starting with 15 min of aerobic stained with methylene blue to visualize min. Blood samples for plasma glucose exercise three times per week at 50% of (Beckman Glucose Analyzer) and insulin the nuclei. Digital photographs were ϭ each individual’s heart rate reserve. This taken at 10ϫ magnification , after which concentrations were taken at t 0, 2, 4, was gradually increased to 45 min aerobic the diameters of Ն150 cells per site were 8, 10, 18, 24, 32, 40, 60, 70, 120, and 180 exercise four times a week at 60–70% of measured using National Institutes of min. Data were analyzed using Bergman’s heart rate reserve. In addition, the diet Health Image Analysis software for PCs minimal model (12,13). and exercise group participated in a be- from Scion (Frederick, MD). Histograms havior modification (modified LEARN) were graphically and numerically dis- Statistical analysis program biweekly and worked with a played. Cell volumes were calculated Values are expressed as means Ϯ SE. Sta- general clinical research center dietitian using the Goldrick formula (11). Adi- tistical comparisons of the two groups every 4 weeks. After 18–20 weeks, all pocellular lipid content was calculated as (pioglitazone versus diet and exercise) tests and biopsies were repeated. fat cell volume times 0.95. and the responses of the groups to the

DIABETES CARE, VOLUME 26, NUMBER 11, NOVEMBER 2003 3149 Effects of pioglitazone versus diet and exercise

Table 1—Demographic and body composition before and after intervention

Diet/exercise Pioglitazone Pre Post Pre Post P-⌬ Age (years) 41 Ϯ 2 — 36 Ϯ 2 —— Weight (kg) 97.5 Ϯ 3.3 85.8 Ϯ 3.1* 98.2 Ϯ 2.7 100.9 Ϯ 3.0† 0.0001 BMI (kg/m2) 32.1 Ϯ 0.7 27.7 Ϯ 0.8* 33.4 Ϯ 0.6 34.0 Ϯ 0.7‡ 0.0001 Waist circumference (cm) 105 Ϯ 295Ϯ 2* 107 Ϯ 2 107 Ϯ 2 0.0001 Hip circumference (cm) 111 Ϯ 1 105 Ϯ 2* 115 Ϯ 1 119 Ϯ 2* 0.0001 WHR 0.95 Ϯ 0.02 0.90 Ϯ 0.01* 0.94 Ϯ 0.01 0.90 Ϯ 0.01* 0.42 Body fat (%) 40 Ϯ 1.4 32 Ϯ 1.8* 41 Ϯ 1.4 41 Ϯ 1.5 0.0001 Body fat (kg) 35.8 Ϯ 1.5 26.5 Ϯ 1.9* 38.0 Ϯ 1.5 39.3 Ϯ 1.7† 0.0001 FFM (kg) 59.0 Ϯ 2.5 58.5 Ϯ 2.4 58.0 Ϯ 2.2 59.1 Ϯ 2.2 0.07 Data are means Ϯ SE. Baseline values are not significantly different between diet and exercise and pioglitazone groups. *P Ͻ 0.001; †P Ͻ 0.01; and ‡P Ͻ 0.05 before versus after the intervention. P-⌬ signifies the difference between the effect of the two interventions. interventions were done using repeated- for women, 13.9 Ϯ 1.5 for men) by loss of ing plasma glucose and C-peptide con- measures ANOVA, followed by t tests fat, not fat-free mass (FFM). Because adi- centrations were similar in both groups. (paired or nonpaired) if needed. P val- pose tissue mass is 85% lipid and 15% The changes in serum lipid concentra- ues Ͻ0.05 were considered statistically water (14), and because DEXA measure- tions were more marked in the diet and significant. ment of FFM includes this adipose tissue exercise group; the only significant be- water, the loss of adipose without loss of tween-group difference, however, was for RESULTS FFM suggests favorable changes in mus- serum total cholesterol (Table 2). The cle mass. Proportionately more abdomi- greater decrease in insulin in the pioglita- Subject characteristics nal than femoral and more visceral than zone group is confounded by higher base- Nineteen volunteers (10 men and 9 subcutaneous abdominal fat was lost, as line concentrations. In the diet and women) completed the diet and exercise evidenced by the changes in WHR, the exercise group, blood pressure decreased intervention and 20 volunteers (10 men ratio of visceral fat to subcutaneous ab- from 128 Ϯ 4/82 Ϯ 2to122Ϯ 3/76 Ϯ 2 and 10 women) completed the pioglita- dominal fat area, and the various fat com- (P Ͻ 0.0001 for systolic and P ϭ 0.07 for zone intervention. The two groups were partments measured by DEXA (Tables 1 diastolic), and in the pioglitazone group, well matched for age, BMI, WHR, insulin and 3). Fat cell lipid content decreased in blood pressure decreased from 129 Ϯ sensitivity parameters, and body compo- the abdominal (P ϭ 0.06) but not the 4/80 Ϯ 3to127Ϯ 3/75 Ϯ 2(P Ͻ 0.05 for sition (Tables 1–3). The preintervention femoral depot (P ϭ 0.33) with diet and diastolic only). Overnight postabsorptive fasting plasma insulin concentrations exercise. plasma free fatty acid concentrations were were greater (P Ͻ 0.05) in the pioglita- In the pioglitazone group, the average not different between the two groups be- zone than in the diet and exercise group, weight gain of 2.7 Ϯ 0.7 kg was attributed fore or after treatment. and the subcutaneous fat area by CT was to an increase in fat (1.3 kg), predomi- greater (P Ͻ 0.05) in the pioglitazone nantly in the leg depot, and to increased Relationship between change in group. FFM (1.1 kg, P ϭ 0.07). There was no body composition and change in change in abdominal fat compartments metabolic variables Relationship between body (visceral or upper body nonvisceral). Changes in Si were not significantly cor- composition and metabolic variables Consistent with the DEXA and CT data, related with changes in body composition At baseline, fasting plasma insulin con- WHR decreased due to increased hip, but in either group (results not shown). In centrations correlated with weight (r ϭ not waist, circumference. The average addition, changes in glucose, C-peptide, Ͻ 0.37, P 0.05), waist circumference adipocyte lipid content after the pioglita- or Si did not correlate with changes in (0.35, P Ͻ 0.05), and WHR (0.37, P Ͻ zone treatment was less in both the fem- femoral or abdominal fat cell size in either 0.05). C-peptide was significantly corre- oral and abdominal sites but the group. lated with these parameters as well as with difference from baseline was not statisti- BMI (0.33, P Ͻ 0.05) and visceral fat area cally significant (decrease by 0.09 ␮g lip- CONCLUSIONS — We compared (0.33, P Ͻ 0.05). At baseline, insulin sen- id/cell in femoral, P ϭ 0.15; decrease by the effects of two insulin-sensitizing regi- ␮ ϭ sitivity index (Si) correlated positively 0.06 g lipid/cell in abdominal, P mens, pioglitazone versus diet and exer- with HDL but negatively with WHR 0.23). cise, on body composition, body fat (Ϫ0.36, P Ͻ 0.05) and total abdominal distribution, and insulin sensitivity. Non- fat area by CT (Ϫ0.36, P Ͻ 0.05). Metabolic response to intervention diabetic upper body obese adults were The increase in Si was greater in the diet studied because of the high prevalence of Body composition changes in and exercise than in the pioglitazone insulin resistance in this population. The response to intervention group, although the difference between anticipated improvement in Si occurred The diet and exercise program induced a the two treatments was not statistically with each treatment, and the change in weight loss of 11.8 Ϯ 1.1 kg (9.5 Ϯ 1.0 significant (P ϭ 0.15). Decreases in fast- body fat compartments in response to diet

3150 DIABETES CARE, VOLUME 26, NUMBER 11, NOVEMBER 2003 Shadid and Jensen

Table 2—Effects of diet/exercise and pioglitazone on insulin sensitivity and lipids

Diet/exercise Pioglitazone Pre Post Pre Post P-⌬ Fasting glucose (mg/dl) 96 Ϯ 291Ϯ 2* 97 Ϯ 292Ϯ 2† 0.69 Fasting insulin (␮U/ml) 8 Ϯ 15Ϯ 1* 11 Ϯ 17Ϯ 1* 0.06 Fasting C-peptide (nmol/l) 0.60 Ϯ 0.04 0.45 Ϯ 0.03* 0.68 Ϯ 0.04 0.50 Ϯ 0.03* 0.61 Serum triglyceride (mg/dl) 162 Ϯ 31 102 Ϯ 12† 171 Ϯ 16 156 Ϯ 16 0.26 HDL cholesterol (mg/dl) 39 Ϯ 238Ϯ 234Ϯ 235Ϯ 2 0.08 Total cholesterol (mg/dl) 201 Ϯ 18 159 Ϯ 7‡ 195 Ϯ 7 190 Ϯ 7 0.01 Ϫ1 Ϫ1 Ϯ Ϯ Ϯ Ϯ Si (mU/l) min 5.31 0.97 10.34 1.88* 4.23 0.61 6.60 1.09† 0.15 Baseline values are not significantly different between diet and exercise and pioglitazone except in fasting insulin, which were higher in the pioglitazone group than in the diet and exercise group (P ϭ 0.04). *P Ͻ 0.001; †P Ͻ 0.01; and ‡P Ͻ 0.05 before versus after the intervention. P-⌬ signifies the difference between the effect of the two interventions. and exercise was consistent with previous in visceral fat combined TZDs with ener- responsible for the leg fat gain. This reports. We unexpectedly found that pio- gy-restricted diets (7,16) or other medi- would be consistent with the peroxisome glitazone resulted in the preferential accu- cation (3), which may have modulated proliferator–activated receptor-␥ agonist mulation of lower body fat rather than the TZD effects. These investigators stud- effects of pioglitazone on preadipocytes. loss of visceral fat. Thus, both diet and ied the effects of TZDs in type 2 diabetic If the improvements in insulin sensitivity exercise and pioglitazone resulted in a re- adults, a different study population from we observed are related to changes in ad- duced WHR but the mechanism was quite our insulin-resistant nondiabetic volun- ipose tissue metabolism, these data sug- different. The shift toward a lower body teers. Although it is possible that the re- gest an independent role for the relative fat distribution by pioglitazone via gain of sponse to TZDs is different between amount of lower body fat. Alternatively, leg fat, not loss of visceral fat, is consistent diabetic and nondiabetic humans, we pioglitazone may improve Si and increase with adipose depot–specific responses, note that a trend toward decreasing WHR lower body fat independently. but not of the type previously reported. despite an increasing waist diameter was A number of investigators have exam- The lack of change in intra-abdominal noted in adults with type 2 diabetes (3), ined the site-specific actions of TZD on adipose tissue area with pioglitazone is suggesting that our finding is not entirely adipose tissue remodeling. In vitro re- consistent with some, but not all, previ- unique to nondiabetic volunteers. sponsiveness of abdominal subcutaneous findings. Four reports described no Pioglitazone increased leg fat without ous, but not omental human preadi- change (4,5,7,15) and three a decrease influencing upper body fat mass. The pocytes to TZD, has been reported (6). In (3,7,16). Most (4,5,7,15,16), but not all trend toward smaller femoral fat cell size vivo, ovarian adipose tissue was more (4), reported a decrease in the visceral-to- in the pioglitazone group in the face of sensitive to TZD than retroperitoneal or subcutaneous abdominal fat ratio. How- increased leg fat mass suggests adipocyte subcutaneous abdominal fat Zucker rats ever, the investigators reporting reductions proliferation rather than hypertrophy was (8). Although not specifically supportive

Table 3—Effects of diet/exercise and pioglitazone on fat distribution and fat cell size

Diet/exercise Pioglitazone Pre Post Pre Post P-⌬ CT abdomen SC fat (cm2) 259 Ϯ 15 196 Ϯ 18* 317† Ϯ 18 328 Ϯ 23 0.04 Visceral (cm2) 203 Ϯ 29 123 Ϯ 22* 154 Ϯ 15 154 Ϯ 17 0.0001 Visceral/SC 0.80 Ϯ 0.11 0.66 Ϯ 0.11* 0.50† Ϯ 0.05 0.48 Ϯ 0.05 0.007 DEXA fat (kg) Total body 35.8 Ϯ 1.5 26.5 Ϯ 1.9* 38.0 Ϯ 1.5 39.3 Ϯ 1.7‡ 0.0001 Leg 11.7 Ϯ 0.8 8.7 Ϯ 0.8* 13.8 Ϯ 0.8 14.8 Ϯ 0.9* 0.0001 CT ϩ DEXA fat (kg) UBNV 18.3 Ϯ 0.7 14.5 Ϯ 1.0* 19.8 Ϯ 0.9 20.2 Ϯ 0.8 0.0001 Visceral 5.8 Ϯ 0.6 3.3 Ϯ 0.5* 4.4 Ϯ 0.4 4.3 Ϯ 0.4 0.0001 Fat cell size (␮g lipid/cell) Abdominal 0.84 Ϯ 0.19 0.68 Ϯ 0.16 0.81 Ϯ 0.18 0.75 Ϯ 0.17 0.52 Femoral 0.85 Ϯ 0.19 0.79 Ϯ 0.18 0.85 Ϯ 0.19 0.76 Ϯ 0.17 0.60 Data are means Ϯ SE. Baseline values are not signiﬁcantly different between diet and exercise and pioglitazone except as noted. *P Ͻ 0.001 before versus after the intervention; †P Ͻ 0.05 diet and exercise versus piolitazone at baseline; and ‡P Ͻ 0.01 before versus after the intervention; P-⌬ signiﬁes the difference between the effect of the two interventions. SC, subcutaneous; UBNV, upper body nonvisceral.

DIABETES CARE, VOLUME 26, NUMBER 11, NOVEMBER 2003 3151 Effects of pioglitazone versus diet and exercise of our observations regarding leg fat, both body subcutaneous fat occurred with diet of troglitazone on body fat distribution in findings suggest a regional difference in and exercise. Understanding the depot type 2 diabetic patients. Diabetes Care 22: TZD sensitivity. specific action of TZD may help define the 908–912, 1999 8. de Souza CJ, Eckhardt M, Gagen K, Dong As expected, Si correlated with WHR insulin-sensitizing properties of this class and total abdominal fat at baseline. This of compounds. M, Chen W, Laurent D, Burkey BF: Effects of pioglitazone on adipose tissue remod- was expected given the known associa- eling within the setting of obesity and in- tion between insulin resistance and vis- sulin resistance. Diabetes 50:1863–1871, ceral fat accumulation (17). The relative Acknowledgments— This study was sup- ported by grants DK40484, DK50456, and 2001 weakness of the correlation coefficients 9. Jensen MD, Kanaley JA, Reed JE, Sheedy between S and anthropometric/body RR00585 from the U.S. Public Health Service, i the Mayo Foundation, and Takeda Pharma- PF: Measurement of abdominal and vis- composition parameters in our popula- ceuticals. ceral fat with computed tomography and tion are likely due to the selection of the We thank Dr. Scott Crow from the Univer- dual-energy x-ray absorptiometry. Am J participants; the narrow range of Si and sity of Minnesota for his help in the behavioral Clin Nutr 61:274–278, 1995 body fat/fat distribution variables re- modification program. 10. Doan AE, Peterson DR, Blackmon JR, duced the strength of the associations. Bruce RA: Myocardial ischemia after max- Given the weak correlation coefficients imal exercise in healthy men. Am Heart J between S and body composition at base- References 69:11–25, 1965 i 11. Di Girolamo M, Fine JB: Cellularity mea- line, it is not surprising that changes in S 1. Ross R, Rissanen J, Pedwell H, Clifford J, i surements. Methods Mol Biol 155:65–75, were not significantly correlated with Shragge P: Influence of diet and exercise on skeletal muscle and visceral adipose 2001 changes in the parameters we assessed 12. Cobelli C, Caumo A, Omenetto M: Mini- with regards to adipose tissue mass, dis- tissue in men. J Appl Physiol 81:2445– mal model SG overestimation and SI un- tribution, or cellularity. The relatively 2455, 1996 2. Barzilai N, She L, Liu B, Vuguin P, Cohen derestimation: improved accuracy by a small interindividual variation in fat and P, Wang J, Rossetti L: Surgical removal of Bayesian two-compartment model. Am J visceral fat loss in the diet and exercise visceral fat reverses hepatic insulin resis- Physiol 277:E481–E488, 1999 group, combined with inherent individ- tance. Diabetes 48:94–98, 1999 13. Bergman RN, Ider YZ, Bowden CR, Co- belli C: Quantitative estimation of insulin ual differences in the Si response to weight 3. Kelly IE, Han TS, Walsh K, Lean MEJ: loss, may limit the ability to detect a cor- Effects of a thiazolidinedione compound sensitivity. Am J Physiol 236:E667–E677, relation between fat loss and improve- on body fat and fat distribution of patients 1979 ment in S . Alternatively, improvements with type 2 diabetes. Diabetes Care 22: 14. Wang J, Pierson RN: Disparate hydration i of adipose and lean tissue require a new in S from 5 months of exercise and energy 288–293, 1999 i 4. Kawai T, Takei I, Oguma Y, Ohashi N, model for body water distribution in man. restriction may be independent of and J Nutr 106:1687–1693, 1976 greater than the effects of regional fat loss, Tokui M, Oguchi S, Katsukawa F, Hirose H, Shimada A, Watanabe K, Saruta T: Ef- 15. Katoh S, Hata S, Matsushima M, Ikemoto such that the underlying relationship is fects of troglitazone on fat distribution in S, Inoue Y, Yokoyama J, Tajima N: Trogli- undetectable. It is also possible that the the treatment of male type 2 diabetes. Me- tazone prevents the rise in visceral adipos- precision of measurement of insulin ac- tabolism 48:1102–1107, 1999 ity and improves fatty liver associated tion with the intravenous glucose toler- 5. Akazawa S, Sun F, Ito M, Kawasaki E, with sulfonylurea therapy: a randomized ance test is insufficient for detecting a Eguchi K: Efficacy of troglitazone on body controlled trial. Metabolism 50:414–417, relationship despite the relatively accu- fat distribution in type 2 diabetes. Diabe- 2001 rate measures of fatness. tes Care 23:1067–1071, 2000 16. Miyazaki Y, Mahankali A, Matsuda M, In conclusion, weight loss via diet and 6. Adams M, Montague CT, Prins JB, Holder Mahankali S, Hardies J, Cusi K, Man- exercise and pioglitazone improve insulin JC, Smith SA, Sanders L, Digby JE, Sewter darino LJ, DeFronzo RA: Effect of pioglitazone on abdominal fat distribution and sensitivity and shift adipose tissue toward CP, Lazar MA, Chatterjee KK, O’Rahilly S: Activators of peroxisome proliferator- insulin sensitivity in type 2 diabetic pa- a lower body fat distribution in upper activated receptor ␥ have depot-specific tients. J Clin Endocrinol Metab 87:2784– body obese nondiabetic adults. Pioglita- effects on human preadipocyte differenti- 2791, 2002 zone selectively increased lower body fat, ation. J Clin Invest 100:3149–3153, 1997 17. Smith SR, Zachwieja JJ: Visceral adipose apparently via adipocyte proliferation, 7. Mori Y, Murakawa Y, Okada K, Horikoshi tissue: a critical review of intervention whereas preferential loss of visceral/upper H, Yokoyama J, Tajima N, Ikeda Y: Effect strategies. Int J Obes 23:329–335, 1999

3152 DIABETES CARE, VOLUME 26, NUMBER 11, NOVEMBER 2003