DIABETICMedicine DOI: 10.1111/dme.12512
Research: Treatment Impact of rosiglitazone on body composition, hepatic fat, fatty acids, adipokines and glucose in persons with impaired fasting glucose or impaired glucose tolerance: a sub-study of the DREAM trial Z. Punthakee1,2, N. Alm eras3, J.-P. Despr es3,4, G. R. Dagenais3,5, S. S. Anand1,2,6, D. L. Hunt2, A. M. Sharma7, H. Jung1, S. Yusuf1,2 and H. C. Gerstein1,2 1 Population Health Research Institute, Hamilton Health Sciences, 2Department of Medicine, McMaster University, Hamilton, ON, 3Quebec Heart and Lung Institute, 4Department of Social and Preventive Medicine, 5Department of Medicine, Laval University, Quebec, QC, 6Department of Epidemiology, McMaster University, Hamilton, ON, and 7Department of Medicine, University of Alberta, Edmonton, AB, Canada
Accepted 30 May 2014
Abstract Aims Thiazolidinediones reduce ectopic fat, increase adiponectin and reduce inflammatory adipokines, fatty acids and glucose in people with Type 2 diabetes. We aimed to measure these effects in people with impaired fasting glucose and/ or impaired glucose tolerance. Methods After approximately 3.5 years of exposure to rosiglitazone 8 mg (n = 88) or placebo (n = 102), 190 DREAM trial participants underwent abdominal computed tomography and dual-energy X-ray absorptiometry scans. Visceral and subcutaneous adipose tissue areas, estimated hepatic fat content, total fat and lean mass were calculated and changes in levels of fasting adipokines, free fatty acids, glucose and post-load glucose were assessed.
Compared with the placebo, participants on rosiglitazone had no difference in lean mass, had 4.1 kg more body fat (P < 0.0001) and 31 cm2 more subcutaneous abdominal adipose tissue area (P = 0.007). Only after adjusting for total fat, participants on rosiglitazone had 23 cm² less visceral adipose tissue area (P = 0.01) and an 0.08-unit higher liver:spleen attenuation ratio (i.e. less hepatic fat; P = 0.02) than those on the placebo. Adiponectin increased by 15.0 lg/ml with rosiglitazone and by 0.4 lg/ml with placebo (P < 0.0001). Rosiglitazone’s effect on fat distribution was not independent of changes in adiponectin. Rosiglitazone’s effects on fasting (–0.36 mmol/l; P = 0.0004) and 2-h post-load glucose (–1.21 mmol/l; P = 0.0008) were not affected by adjustment for fat distribution or changes in adiponectin or free fatty acids. Results
Conclusions In people with impaired fasting glucose/impaired glucose tolerance, rosiglitazone is associated with relatively less hepatic and visceral fat, increased subcutaneous fat and increased adiponectin levels. These effects do not appear to explain the glucose-lowering effect of rosiglitazone.
Diabet. Med. 31, 1086–1092 (2014)
Introduction Obesity is a powerful risk factor for Type 2 diabetes [1] and much of this risk may be mediated by ectopic deposition of fat in visceral adipose tissue, liver and muscle [2]. Peroxisome proliferator-activated receptor gamma (PPAR-c) agonists have salutary effects on adipose tissue over the short term. In people with Type 2 diabetes, thiazolidinediones increase total body fat with a preferential increase in subcutaneous fat with or without a decrease in visceral fat, Correspondence to: Zubin Punthakee. E-mail: [email protected]
1086
leading to a lower proportion of fat in the visceral compartment [3–7]. This is reflected by reduced ectopic fat in the liver [3–6]. Thiazolidinediones also increase circulating adiponectin and reduce interleukin 6 (IL-6) and/or tumour necrosis factor-alpha (TNF-a) levels in people with diabetes [8–10]. However, the effect of thiazolidinediones on directly measured body composition and adipokines beyond 1 year of therapy, and in people without diabetes, particularly those with impaired fasting glucose or impaired glucose tolerance has not been previously reported. In the DREAM trial (Diabetes REduction Assessment with ramipril and rosiglitazone Medication), rosiglitazone
ª 2014 The Authors. Diabetic Medicine ª 2014 Diabetes UK
Research article
What’s new? • Most available studies show the effect of thiazolidinediones on body composition after 1 year in people with diabetes. • After 3.5 years of exposure to rosiglitazone compared with placebo, people with impaired fasting glucose or impaired glucose tolerance have more subcutaneous fat with relatively less visceral fat and liver fat, and a greater increase in adiponectin. • The long-term glucose-lowering effect of rosiglitazone is independent of the effects on body composition and adipokines. significantly reduced the incidence of diabetes by 60% and increased regression to normoglycaemia by 70% compared with placebo among people with impaired fasting glucose and/or impaired glucose tolerance [11]. These striking benefits on glucose homeostasis were accompanied by a significant gain in body weight, but a reduction in waist–hip ratio, because of a 2-cm increase in hip circumference and no effect on waist circumference [11]. There was also a reduction in circulating alanine aminotransferase (ALT), a marker associated with ectopic hepatic fat deposition [12]. These gross changes in body weight and shape noted during the DREAM trial prompted a more accurate characterization of body fat distribution, with detailed imaging studies in a subset of trial participants at the end of the trial. We hypothesized that long-term use of rosiglitazone by people with impaired fasting glucose and/or impaired glucose tolerance would be associated with lower amounts of directly measured visceral fat and liver fat, more subcutaneous fat, increases in circulating adiponectin and reductions in free fatty acids, IL-6 and TNF-a, as well as glucose levels.
Research design and methods The design and results of the DREAM trial are described elsewhere [11,13]. For the DREAM trial, 5269 men and women of at least 30 years of age with impaired fasting glucose and/or impaired glucose tolerance and without a history of coronary, cerebral or peripheral artery disease were recruited from 2001 to 2003. Participants were randomized to rosiglitazone (4 mg daily for the first 2 months, then 8 mg daily) or placebo, as well as to ramipril (15 mg daily) or placebo using a factorial design. Study visits were every 6 months, and blood samples were taken in the fasting state and 2 h after a 75-g oral glucose load at baseline, and at study end for central analysis. The effects of rosiglitazone on body composition at the end of the trial and on adipokine levels during the trial were assessed in a DREAM sub-study conducted at two Canadian study sites with the required imaging equipment (Hamilton
ª 2014 The Authors. Diabetic Medicine ª 2014 Diabetes UK
DIABETICMedicine
and Quebec City). All non-pregnant participants at these sites who weighed ≤ 136 kg (300 lbs) were asked to participate, and consenting participants had an abdominal computed tomography (CT) scan and a whole-body dual-energy X-ray absorptiometry (DXA) scan up to 30 days after their final DREAM visit. All analyses were carried out without knowledge of participants’ treatment allocation, adherence or diabetes status. Ethics approval and written informed consent were obtained at both participating institutions.
Body composition measures
Abdominal CT scan images were obtained from one scanner at each site using the same protocol to ensure standardized attenuations and image sizes for evaluating fat areas (140 kV, 220 mA, field of view 480 9 480 mm, matrix 512 9 512). All scans were obtained between January and June 2006. CT Quality Control Units (phantoms) developed at the Quebec Heart and Lung Institute were scanned at each site before the first participant and after the last participant to ensure reliability over time at each site. Computed tomography images were analysed centrally by a trained individual blind to treatment allocation using SliceOmatic software (Tomovision, Montreal, QC, Canada) [14]. For assessment of visceral and subcutaneous abdominal fat areas, a transverse 10-mm CT scan slice was acquired at the L4–L5 intervertebral space. An attenuation window of –190 to –30 Hounsfield units (HU) was used to identify abdominal adipose tissue [15]. The boundary separating subcutaneous and visceral fat was traced manually and visceral abdominal fat, subcutaneous abdominal fat and the visceral:subcutaneous abdominal fat ratio [16,17] were calculated. For assessment of hepatic fat, a transverse 10-mm CT scan slice was acquired at the T11–T12 intervertebral space. The mean HU of the liver and the spleen were calculated, as well as the ratio of the two mean attenuations [14,18]. Lower values for the mean HU reflect greater fat content. The ratio of liver HU:spleen HU has been validated against liver histology and it correlates highly (r = –0.86) with hepatic fat content by histomorphometry [19,20]. Whole-body DXA scans were completed at both study sites using GE Lunar Prodigy scanners (GE Healthcare, Madison, WI, USA) and the manufacturer’s standard protocol. A variable composition phantom (Lunar, Madison, WI, USA) was scanned before the first participant and after the last participant at each site to ensure reliability between sites and over time. Data files were analysed centrally using Encore Software, version 8.80, 2004 (Lunar) by a single trained individual blind to treatment allocation to estimate total body fat and leg fat mass.
Biochemical measures
Blood samples from baseline and study end were stored at – 120 °C until analysed centrally. Free fatty acids were
1087
DIABETICMedicine
Rosiglitazone, fat and glucose metabolism in DREAM Z. Punthakee et al.
measured using a colorimetric method that has a detection limit of 5.0 mmol/l and coefficient of variation of 2.7%. TNF-a was determined using ELISA (R&D Systems Inc., Minneapolis, MN, USA), with a detection limit of 0.12 ng/l and coefficient of variation of 16.7%. IL-6 was measured by ELISA (R&D Systems Inc.), with a detection limit of 0.04 ng/l and a coefficient of variation of 9.6%. Adiponectin was assayed by ELISA (Linco Research Inc., St Charles, MO, USA) with a detection limit of 1 lg/l and a coefficient of variation of 8.4%. Locally measured ALT was recorded at baseline and at 1 year.
Statistical analysis
Participants were analysed according to the intention-to-treat principle. Characteristics of participants at baseline, at the end of the study and at changes in adipokine levels are presented for rosiglitazone and placebo groups as mean SD or n (%) and were compared by t-tests or v2-tests, respectively. Linear regression was used to assess the relationship between rosiglitazone and changes in adipokine levels, measures of body fat and glucose levels, adjusted for ramipril group allocation, age, sex and baseline BMI. Models for measures of body fat were additionally adjusted for total body fat at study end and changes in free fatty acids, adiponectin, TNF-a and IL-6. Regression models for glucose levels were adjusted for ramipril group allocation, age, sex and baseline BMI, change in adiponectin and body fat measures. All analyses were conducted using SAS version 9.1 (SAS Institute, Cary, NC, USA).
Results Of the 264 individuals who were randomized at the two sites and who attended the final visit, 190 participated in this sub-study (88 allocated to rosiglitazone and 102 allocated to placebo; see Fig. 1). The proportions excluded or declined were not different between the groups (2 d.f., v2 = 2.77, P = 0.25). A total of 153 participants had adequate CT and DXA scans to complete all measurements and 128 participants also had sufficient remaining blood samples for all biochemical measures. The proportions lacking CT and blood data were not different between the two groups (2 d.f., v2 = 4.54, P = 0.10). Participants had been followed for a mean of 3.5 years (SD 0.6) after randomization. Baseline characteristics of participants are shown in Table 1. There were no significant differences between the rosiglitazone and placebo groups. Adherence data were reported by 181 participants at the end of the study, and 79/82 (96%) of the rosiglitazone group and 96/99 (97%) of the placebo group were at least 80% adherent to study medication. At the end of the study, the mean daily dose of rosiglitazone was 7.66 mg (SD 1.12), and nobody in the placebo group was receiving open-label thiazolidinedione.
1088
264 randomized in DREAM 1 died 0 were lost 0 withdrew consent 263 attended final DREAM visit
129 had been randomized to rosiglitazone
134 had been randomized to placebo
Excluded: 3 >136 kg 2 pregnancy risk 36 declined
Excluded: 2 >136 kg 0 pregnancy risk 30 declined
88 sub-study participants
102 sub-study participants
Completed assessments 88 DXA 72 DXA and CT 65 DXA and CT and blood
Completed assessments 102 DXA 81 DXA and CT 63 DXA and CT and blood
FIGURE 1 Enrolment and evaluation of study participants. CT, computed tomography; DXA, dual-energy X-ray absorptiometry.
Effect on body fat compartments
Total body weight gain and total body weight were 3.4 kg (P = 0.005) and 6.0 kg (P = 0.02) higher, respectively, in the rosiglitazone group than the placebo group at the end of the study. Total fat mass by DXA was 4.1 kg (95% CI 2.1–6.2; P < 0.0001) higher in the rosiglitazone group than the placebo group at the end of the study after adjustment for age, sex, baseline BMI and ramipril allocation. Lean mass by DXA did not differ (P = 0.9). The effects of rosiglitazone on visceral fat and ectopic hepatic fat are shown in Fig. 2. After adjustment for age, sex, baseline BMI and ramipril allocation, there was no difference in visceral abdominal fat area between the rosiglitazone and placebo groups. After additional adjustment for final total body fat mass to account for the increase in total fat with rosiglitazone, visceral abdominal fat area was 23 cm² lower (95% CI 5–41; P = 0.01) in the rosiglitazone group. In similar models, rosiglitazone was also associated with a lower visceral:subcutaneous abdominal fat ratio by 0.104 (95% CI 0.027–0.181; P = 0.008) compared with placebo and a higher liver:spleen ratio by 0.08 (95% CI 0.01–0.15; P = 0.02), suggesting less hepatic fat relative to the total amount of body fat. The effects of rosiglitazone on these measures of fat distribution were eliminated after adjustment for the change in fasting free fatty acids or the change in fasting adiponectin from baseline. Measures of subcutaneous fat, subcutaneous abdominal fat assessed by CT and leg fat assessed by DXA were higher in the rosiglitazone group by 31 cm2 (95% CI 9–53;
ª 2014 The Authors. Diabetic Medicine ª 2014 Diabetes UK
Research article
DIABETICMedicine
P = 0.007) and 1.9 kg (95% CI 1.0–2.7; P < 0.0001), respectively, after adjustment for age, sex, baseline BMI and ramipril allocation. The associations of rosiglitazone with subcutaneous fat compartments were attenuated after adjustment for total body fat mass. Addition of adipokines into the models did not have an important effect on the observed relationship between rosiglitazone and subcutaneous fat (Fig. 2).
Effect on adipokines
There was a large and significant increase in adiponectin in the rosiglitazone group (Table 2). After adjustment for age, sex, baseline BMI and ramipril allocation, rosiglitazone Table 1 Baseline characteristics* Rosiglitazone n (%)/mean (SD) Number of participants Ramipril arm allocation Ramipril Placebo Age (years) Female > 3 alcoholic drinks/week Isolated impaired fasting glucose† Isolated impaired glucose tolerance‡ Impaired fasting glucose + impaired glucose tolerance§ Weight (kg) BMI (kg/m2) Waist circumference (cm) Men Women Hip circumference (cm) Men Women Waist–hip ratio Men Women Systolic blood pressure (mmHg) Diastolic blood pressure (mmHg) Fasting plasma glucose (mmol/l) 2-h plasma glucose (mmol/l)¶ HOMA-IR Free fatty acids (lmol/l) Adiponectin (lg/ml) TNF-a (pg/ml) IL-6 (pg/ml) ALT (U/l)
88 49 39 53.3 44 36 15 52
Placebo n (%)/mean (SD)
50 52 52.8 62 33 13 68
(49%) (51%) (10.2) (61%) (32%) (13%) (67%)
21 (24%)
21 (21%)
85.4 (16.2) 30.5 (5.2)
82.7 (17.9) 30.1 (5.6)
105.4 (12.2) 93.9 (12.2)
103.1 (12.5) 91.9 (13.1)
108.9 (7.1) 110.8 (12.1)
107.3 (10.1) 108.7 (12.4)
0.97 0.85 135 80 5.7 8.6 3.8 495 6.4 1.82 1.80 28
(0.07) (0.05) (17) (10) (0.7) (1.3) (2.5) (139) (3.0) (2.43) (1.27) (16)
0.96 0.84 134 81 5.6 8.7 4.6 525 6.1 1.72 1.71 25
(0.06) (0.07) (18) (10) (0.7) (1.3) (4.4) (170) (3.2) (0.98) (1.08) (13)
*There were no significant differences between groups †6.1 mmol/l ≤ fasting plasma glucose < 7.0 mmol/l and 2-h plasma glucose < 7.8 mmol/l. ‡Fasting plasma glucose < 6.1 mmol/l and 7.8 ≤ 2-h plasma glucose < 11.1 mmol/l. §6.1 mmol/l ≤ fasting plasma glucose < 7.0 mmol/l and 7.8 mmol/l ≤ 2-h plasma glucose < 11.1 mmol/l. –Plasma glucose 2 h after a 75-g oral glucose load. ALT, alanine aminotransferase; HOMA-IR, homeostasis model assessment of insulin resistance; IL-6, interleukin 6; TNF-a, tumour necrosis factor alpha
ª 2014 The Authors. Diabetic Medicine ª 2014 Diabetes UK
Effect on glucose metabolism
Rosiglitazone reduced both fasting and 2-h post-load glucose levels (Table 2). After adjusting for age, sex, baseline BMI and ramipril allocation, rosiglitazone lowered fasting glucose by 0.36 mmol/l more than placebo (95% CI 0.16–0.56; P = 0.0004) and 2-h post-load glucose by 1.21 mmol/l (95% CI 0.51–1.91; P = 0.0008) more than placebo. These reductions were independent of the effects of rosiglitazone on total body fat, visceral, hepatic or subcutaneous fat and adiponectin (Fig. 2).
Discussion
102 (56%) (44%) (9.2) (50%) (41%) (17%) (59%)
raised adiponectin levels 14.9 lg/ml more than placebo (95% CI 11.1–18.7; P < 0.0001). Rosiglitazone had no significant effect on free fatty acids, TNF-a or IL-6 (Table 2). These relationships were observed before and after adjustment for age, sex, baseline BMI and ramipril allocation (data not shown).
In this sub-study of the DREAM trial, exposure to rosiglitazone compared with placebo for a mean of 3.5 years had significant effects on body fat distribution and adiponectin levels in people with impaired fasting glucose/impaired glucose tolerance. Rosiglitazone led to a marked increase in adiponectin levels. While other studies have shown a rise in adiponectin up to 1 year [7,21], the current study indicates that the effect of rosiglitazone is sustained over at least 3.5 years. This was not the case for IL-6 and TNF-a, which were not different between the two groups. Shorter studies have been inconsistent in their findings regarding these pro-inflammatory adipokines [21,22], which suggests that thiazolidinediones may not have direct anti-inflammatory effects, but inflammation may be temporarily reduced while rosiglitazone restores adipose tissue physiology, or that other unmeasured markers of inflammation may be mediating any ongoing anti-inflammatory effect. In this study, participants taking rosiglitazone had significantly more subcutaneous fat measured in the abdomen and legs than those in the placebo group. Adjusting for total fat essentially eliminated the difference, suggesting the increase in fat with rosiglitazone is subcutaneous fat. There was no absolute difference in visceral fat area or liver fat between the groups. However, the visceral:subcutaneous fat ratio was lower after treatment with rosiglitazone, suggesting a redistribution of fat. Furthermore, total body-fat adjusted differences in visceral fat and hepatic fat content were approximately half of the differences noted between obese people with and without diabetes [23–26]. While this observed adjusted difference may be primarily attributable to increases in total body fat in the rosiglitazone group, among people with Type 2 diabetes, differences in subcutaneous and visceral fat of similar magnitude have been seen within 24 weeks in placebo-controlled trials of troglitazone,
1089
Rosiglitazone, fat and glucose metabolism in DREAM Z. Punthakee et al.
DIABETICMedicine
(a) Effect of rosiglitazone after adjustment for:
Visceral abdominal fat/ Subcutaneous abdominal fat
Visceral abdominal fat
Liver:spleen
I: Age, sex, baseline BMI II: I + Total body fat mass III: II + ΔFree fatty acids IV: II + ΔAdiponectin V: II + ΔIL-6 VI: II + ΔTNF-α VII: II + All the above -50
0
50
cm2
(b)
-0.2
-0.1
Less abdominal fat in visceral depot
0
0.1
-0.1
0
More hepatic fat
More abdominal fat in visceral depot
0.1
0.2
Less hepatic fat
Leg fat
Subcutaneous abdominal fat
I: Age, sex, baseline BMI II: I + Total body fat mass III: II + ΔFree fatty acids IV: II + ΔAdiponectin V: II + ΔIL-6 VI: II + ΔTNF-α VII: II + All the above -50
0
50
-2
0
cm2
(c)
2
4
kg
ΔFasting plasma glucose
Δ2-h plasma glucose
I: Age, sex, baseline BMI II: I + Total body fat mass III: II + Visceral abdominal fat IV: III + ΔAdiponectin V: II + Liver:spleen ratio VI: V + ΔAdiponectin -1
-0.5
mmol/l
0
-3
-2
-1
0
mmol/l
FIGURE 2 Effect of rosiglitazone vs. placebo on: (a) visceral and ectopic fat compartments, (b) subcutaneous fat compartments and (c) change in fasting and 2-h post-load glucose levels. Mean difference (rosiglitazone – placebo) and 95% CI. All models are adjusted for ramipril allocation. Visceral abdominal fat, abdominal visceral adipose tissue area; subcutaneous abdominal fat, abdominal subcutaneous adipose tissue area; liver: spleen, ratio of computed tomography (CT) attenuation of liver and spleen; D, change (final – baseline); 2-h plasma glucose, plasma glucose 2 h after 75-g oral glucose load; IL-6, interleukin 6; TNF-a, tumour necrosis factor alpha.
pioglitazone and rosiglitazone [3–5,27], and both rosiglitazone and pioglitazone reduce liver fat by 30–50% within 3–6 months [28]. These observations support the hypothesis that, at various degrees of dysglycaemia, all of the thiazolidinediones ‘shunt’ fat to the subcutaneous depot and away from metabolically sensitive organs, including visceral adipose tissue and the liver. This may have an impact on hepatocellular damage as reflected in the ALT reductions noted in the DREAM trial itself [11], and similar but non-significant reductions in ALT seen during the first year in the current small subsample. The analyses in Fig. 2 suggest that the effects of rosiglitazone on visceral and liver fat are linked to its effect on adiponectin levels. As such, the rise in adiponectin may be a useful marker of rosiglitazone’s effect on ectopic fat in the long term. However, data from this study are not able to indicate whether the effects on fat
1090
distribution mediate adipokine changes, or vice versa, or if there is an unmeasured common antecedent. Thiazolidinediones act mainly through PPAR-c receptors in adipose tissue. Despite this tissue being the main site of action, the current data suggest that the effect of rosiglitazone on fasting and 2-h glucose is not explained by differences in total body fat or visceral or hepatic fat. Furthermore, reductions in homeostasis model assessment of insulin resistance (HOMA-IR) and free fatty acids (thought to be the main mechanism of thiazolidinedione insulin sensitization) were not apparent in this study. Rosiglitazone’s glucose-lowering effect parallels its adiponectin-raising effect, but is not completely explained by it. Indeed, studies in isolated muscle and liver suggest that direct and PPAR-c-independent mechanisms may contribute to the glucose-lowering effects of thiazolidinediones via increased
ª 2014 The Authors. Diabetic Medicine ª 2014 Diabetes UK
Research article
DIABETICMedicine
Table 2 Unadjusted body composition at study end, and anthropometric and biochemical changes from baseline Rosiglitazone n/mean (SD) Number of participants CT scan Visceral abdominal fat area (cm2) Subcutaneous abdominal fat area (cm2) Visceral abdominal fat:subcutaneous abdominal fat ratio Liver attenuation (HU) Spleen attenuation (HU) Liver:spleen HU ratio DXA scan Total fat mass (kg) Total per cent body fat (%) Leg fat (kg) DWeight (kg) DBMI (kg/m2) DWaist circumference (cm) Men Women DHip circumference (cm) Men Women DWaist–hip ratio Men Women DFasting plasma glucose (mmol/l) D2-h plasma glucose (mmol/l) DHOMA-IR DFree fatty acids (lmol/l) DTNF-a (pg/ml) DIL-6 (pg/ml) DAdiponectin (lg/ml) DALT at 1 year (U/l)*
88
Placebo n/mean (SD)
P-value
102
191 333 0.570 53.6 43.4 1.236
(95) (111) (0.303) (8.2) (3.4) (0.205)
183 313 0.635 51.9 43.9 1.178
(80) (119) (0.302) (11.4) (4.6) (0.277)
0.55 0.27 0.17 0.26 0.41 0.12
33.8 39.9 10.96 +2.6 +0.9
(11.7) (8.9) (4.80) (10.8) (3.8)
29.8 37.9 9.37 –0.8 –0.3
(11.2) (10.5) (4.29) (5.0) (1.8)
0.02 0.16 0.02 0.005 0.004
–1.1 (7.8) –0.2 (7.3)
2.1 (6.0) +0.5 (7.1)
0.52 0.61
+2.3 (5.3) +2.6 (7.2)
–1.4 (5.0) –0.3 (6.8)
0.002 0.04
–0.030 –0.022 –0.17 –1.63 –0.9 –109 –0.02 +0.50 +15.0 –5.7
(0.055) (0.041) (0.74) (2.46) (2.6) (198) (1.61) (6.81) (17.8) (12.3)
0.009 +0.008 +0.21 –0.46 –0.2 –67 +0.03 –0.04 +0.4 –2.7
(0.039) (0.056) (0.59) (2.29) (3.7) (206) (1.15) (1.17) (1.8) (9.5)
0.04 0.003 0.0003 0.001 0.14 0.19 0.80 0.44 < 0.0001 0.07
*DALT is reported at 1 year after randomization because ALT was only measured systematically up to 1 year. ALT, alanine aminotransferase; CT, computed tomography; DXA, dual-energy X-ray absorptiometry; HOMA-IR, homeostasis model assessment of insulin resistance; HU, Hounsfield units; IL-6, interleukin 6; SD, standard deviation; TNF-a, tumour necrosis factor alpha.
glycolysis, glycogen synthesis and/or PPAR-a agonism in these organs [29]. Strengths of this study include long duration of treatment and the large number of people with impaired fasting glucose and/or impaired glucose tolerance studied with both detailed imaging and biochemical evaluation. The study had several limitations as well. Participants were recruited into this study at the end of the treatment phase; however, they were analysed according to their randomly assigned groups and, while there were no identified imbalances, we adjusted for baseline characteristics. A major limitation of the trial was the lack of body composition measures at baseline. Thus, we were unable to assess directly the change in body composition measures attributable to rosiglitazone; however, as participants were randomized, had a high rate of participation, and the two groups had similar baseline weight, BMI and waist–hip ratio, there is no reason to suspect a difference in baseline fat distribution. Assessing changes in adipokine levels was possible; however, larger than expected variances in the measures may have reduced the power to detect small differences between the groups. In conclusion, rosiglitazone is associated with long-term changes in adiponectin (but not free fatty acids or inflamma-
ª 2014 The Authors. Diabetic Medicine ª 2014 Diabetes UK
tory adipokines), subcutaneous fat and, after adjustment for total fat, the relative amounts of visceral fat and liver fat. However, none of these are accurate markers of the improved glucose metabolism seen in people with impaired fasting glucose/impaired glucose tolerance treated with rosiglitazone. Thiazolidinediones were never approved for diabetes prevention, and their use as diabetes medications has declined for a variety of reasons [30]. Nevertheless, these findings continue to be relevant to currently available drugs as well as newer drugs with PPAR-c activity. Furthermore, understanding how modulation of visceral and liver fat impacts diabetes risk and glucose metabolism clearly continues to be important for both diabetes prevention and treatment.
Funding sources
This sub-study was supported by a grant from the Canadian Institutes of Health Research. ZP was partly supported by a McMaster University Department of Medicine Career Award. SSA holds the May Cohen Eli Lilly Chair in Women’s Health and the Michael G. DeGroote Heart and Stroke Foundation of Ontario Chair in Population Health Research.
1091
DIABETICMedicine
Rosiglitazone, fat and glucose metabolism in DREAM Z. Punthakee et al.
Competing interests
The Population Health Research Institute has received research grants for the conduct of clinical trials that include rosiglitazone from GlaxoSmithKline. ZP, J-PD, GRD, DLH, SY and HCG have received honoraria from GlaxoSmithKline for advice and/or speaking. SSA has received an unrestricted grant from GlaxoSmithKline to study genetic associations in EpiDREAM screenees. AMS has received honoraria from GlaxoSmithKline for consulting and expert testimony. NA and HJ have nothing to declare.
References 1 Hu FB, Manson JE, Stampfer MJ, Colditz G, Liu S, Solomon CG et al. Diet, lifestyle, and the risk of type 2 diabetes mellitus in women. N Engl J Med 2001; 345: 790–797. 2 Hill MJ, Metcalfe D, McTernan PG. Obesity and diabetes: lipids, ‘nowhere to run to’. Clin Sci (Lond) 2009; 116: 113–123. 3 Carey DG, Cowin GJ, Galloway GJ, Jones NP, Richards JC, Biswas N et al. Effect of rosiglitazone on insulin sensitivity and body composition in type 2 diabetic patients. Obes Res 2002; 10: 1008–1015. 4 Katoh S, Hata S, Matsushima M, Ikemoto S, Inoue Y, Yokoyama J et al. Troglitazone prevents the rise in visceral adiposity and improves fatty liver associated with sulfonylurea therapy—a randomized controlled trial. Metabolism 2001; 50: 414–417. 5 Kelly IE, Han TS, Walsh K, Lean ME. Effects of a thiazolidinedione compound on body fat and fat distribution of patients with type 2 diabetes. Diabetes Care 1999; 22: 288–293. 6 Smith SR, De JL, Volaufova J, Li Y, Xie H, Bray GA. Effect of pioglitazone on body composition and energy expenditure: a randomized controlled trial. Metabolism 2005; 54: 24–32. 7 Bertrand OF, Poirier P, Rodes-Cabau J, Rinfret S, Title LM, Dzavik V et al. Cardiometabolic effects of rosiglitazone in patients with type 2 diabetes and coronary artery bypass grafts: a randomized placebo-controlled clinical trial. Atherosclerosis 2010; 211: 565–573. 8 Hirose H, Kawai T, Yamamoto Y, Taniyama M, Tomita M, Matsubara K et al. Effects of pioglitazone on metabolic parameters, body fat distribution, and serum adiponectin levels in Japanese male patients with type 2 diabetes. Metabolism 2002; 51: 314–317. 9 Miyazaki Y, Mahankali A, Matsuda M, Mahankali S, Hardies J, Cusi K et al. Effect of pioglitazone on abdominal fat distribution and insulin sensitivity in type 2 diabetic patients. J Clin Endocrinol Metab 2002; 87: 2784–2791. 10 Shadid S, Stehouwer CD, Jensen MD. Diet/exercise versus pioglitazone: effects of insulin sensitization with decreasing or increasing fat mass on adipokines and inflammatory markers. J Clin Endocrinol Metab 2006; 91: 3418–3425. 11 Gerstein HC, Yusuf S, Bosch J, Pogue J, Sheridan P, Dinccag N et al. Effect of rosiglitazone on the frequency of diabetes in patients with impaired glucose tolerance or impaired fasting glucose: a randomised controlled trial. Lancet 2006; 368: 1096–1105. 12 Klain J, Fraser D, Goldstein J, Peiser J, Avinoah E, Ovnat A et al. Liver histology abnormalities in the morbidly obese. Hepatology 1989; 10: 873–876. 13 Gerstein HC, Yusuf S, Holman R, Bosch J, Pogue J. Rationale, design and recruitment characteristics of a large, simple international trial of diabetes prevention: the DREAM trial. Diabetologia 2004; 47: 1519–1527.
1092
14 Despres JP, Ross R, Boka G, Almeras N, Lemieux I. Effect of rimonabant on the high-triglyceride/low-HDL-cholesterol dyslipidemia, intraabdominal adiposity, and liver fat: the ADAGIO– Lipids trial. Arterioscler Thromb Vasc Biol 2009; 29: 416–423. 15 Sjostrom L, Kvist H, Cederblad A, Tylen U. Determination of total adipose tissue and body fat in women by computed tomography, 40K, and tritium. Am J Physiol 1986; 250: E736–E745. 16 von Eyben FE, Mouritsen E, Holm J, Montvilas P, Dimcevski G, Suciu G et al. Intra-abdominal obesity and metabolic risk factors: a study of young adults. Int J Obes Relat Metab Disord 2003; 27: 941–949. 17 Kelley DE, McKolanis TM, Hegazi RA, Kuller LH, Kalhan SC. Fatty liver in type 2 diabetes mellitus: relation to regional adiposity, fatty acids, and insulin resistance. Am J Physiol Endocrinol Metab 2003; 285: E906–E916. 18 Davidson LE, Kuk JL, Church TS, Ross R. Protocol for measurement of liver fat by computed tomography. J Appl Physiol 2006; 100: 864–868. 19 Longo R, Ricci C, Masutti F, Vidimari R, Croce LS, Bercich L et al. Fatty infiltration of the liver. Quantification by 1H localized magnetic resonance spectroscopy and comparison with computed tomography. Invest Radiol 1993; 28: 297–302. 20 Ricci C, Longo R, Gioulis E, Bosco M, Pollesello P, Masutti F et al. Non-invasive in vivo quantitative assessment of fat content in human liver. J Hepatol 1997; 27: 108–113. 21 Shadid S, Stehouwer CD, Jensen MD. Diet/exercise versus pioglitazone: effects of insulin sensitization with decreasing or increasing fat mass on adipokines and inflammatory markers. J Clin Endocrinol Metab 2006; 91: 3418–3425. 22 Miyazaki Y, Mahankali A, Matsuda M, Mahankali S, Hardies J, Cusi K et al. Effect of pioglitazone on abdominal fat distribution and insulin sensitivity in type 2 diabetic patients. J Clin Endocrinol Metab 2002; 87: 2784–2791. 23 Anjana M, Sandeep S, Deepa R, Vimaleswaran KS, Farooq S, Mohan V. Visceral and central abdominal fat and anthropometry in relation to diabetes in Asian Indians. Diabetes Care 2004; 27: 2948–2953. 24 Boyko EJ, Fujimoto WY, Leonetti DL, Newell-Morris L. Visceral adiposity and risk of type 2 diabetes: a prospective study among Japanese Americans. Diabetes Care 2000; 23: 465–471. 25 Kanaya AM, Harris T, Goodpaster BH, Tylavsky F, Cummings SR. Adipocytokines attenuate the association between visceral adiposity and diabetes in older adults. Diabetes Care 2004; 27: 1375–1380. 26 Kelley DE, McKolanis TM, Hegazi RA, Kuller LH, Kalhan SC. Fatty liver in type 2 diabetes mellitus: relation to regional adiposity, fatty acids, and insulin resistance. Am J Physiol Endocrinol Metab 2003; 285: E906–E916. 27 Shah PK, Mudaliar S, Chang AR, Aroda V, Andre M, Burke P et al. Effects of intensive insulin therapy alone and in combination with pioglitazone on body weight, composition, distribution and liver fat content in patients with type 2 diabetes. Diabetes Obes Metab 2011; 13: 505–510. 28 Yki-Jarvinen H. Thiazolidinediones and the liver in humans. Curr Opin Lipidol 2009; 20: 477–483. 29 Furnsinn C, Waldhausl W. Thiazolidinediones: metabolic actions in vitro. Diabetologia 2002; 45: 1211–1223. 30 Leal I, Romio SA, Schuemie M, Oteri A, Sturkenboom M, Trifiro G. Prescribing pattern of glucose-lowering drugs in the UK in the last decade: a focus on the effects of safety warnings about rosiglitazone. Br J Clin Pharmacol 2013; 75: 861–868.
ª 2014 The Authors. Diabetic Medicine ª 2014 Diabetes UK
Research: Treatment Impact of rosiglitazone on body composition, hepatic fat, fatty acids, adipokines and glucose in persons with impaired fasting glucose or impaired glucose tolerance: a sub-study of the DREAM trial Z. Punthakee1,2, N. Alm eras3, J.-P. Despr es3,4, G. R. Dagenais3,5, S. S. Anand1,2,6, D. L. Hunt2, A. M. Sharma7, H. Jung1, S. Yusuf1,2 and H. C. Gerstein1,2 1 Population Health Research Institute, Hamilton Health Sciences, 2Department of Medicine, McMaster University, Hamilton, ON, 3Quebec Heart and Lung Institute, 4Department of Social and Preventive Medicine, 5Department of Medicine, Laval University, Quebec, QC, 6Department of Epidemiology, McMaster University, Hamilton, ON, and 7Department of Medicine, University of Alberta, Edmonton, AB, Canada
Accepted 30 May 2014
Abstract Aims Thiazolidinediones reduce ectopic fat, increase adiponectin and reduce inflammatory adipokines, fatty acids and glucose in people with Type 2 diabetes. We aimed to measure these effects in people with impaired fasting glucose and/ or impaired glucose tolerance. Methods After approximately 3.5 years of exposure to rosiglitazone 8 mg (n = 88) or placebo (n = 102), 190 DREAM trial participants underwent abdominal computed tomography and dual-energy X-ray absorptiometry scans. Visceral and subcutaneous adipose tissue areas, estimated hepatic fat content, total fat and lean mass were calculated and changes in levels of fasting adipokines, free fatty acids, glucose and post-load glucose were assessed.
Compared with the placebo, participants on rosiglitazone had no difference in lean mass, had 4.1 kg more body fat (P < 0.0001) and 31 cm2 more subcutaneous abdominal adipose tissue area (P = 0.007). Only after adjusting for total fat, participants on rosiglitazone had 23 cm² less visceral adipose tissue area (P = 0.01) and an 0.08-unit higher liver:spleen attenuation ratio (i.e. less hepatic fat; P = 0.02) than those on the placebo. Adiponectin increased by 15.0 lg/ml with rosiglitazone and by 0.4 lg/ml with placebo (P < 0.0001). Rosiglitazone’s effect on fat distribution was not independent of changes in adiponectin. Rosiglitazone’s effects on fasting (–0.36 mmol/l; P = 0.0004) and 2-h post-load glucose (–1.21 mmol/l; P = 0.0008) were not affected by adjustment for fat distribution or changes in adiponectin or free fatty acids. Results
Conclusions In people with impaired fasting glucose/impaired glucose tolerance, rosiglitazone is associated with relatively less hepatic and visceral fat, increased subcutaneous fat and increased adiponectin levels. These effects do not appear to explain the glucose-lowering effect of rosiglitazone.
Diabet. Med. 31, 1086–1092 (2014)
Introduction Obesity is a powerful risk factor for Type 2 diabetes [1] and much of this risk may be mediated by ectopic deposition of fat in visceral adipose tissue, liver and muscle [2]. Peroxisome proliferator-activated receptor gamma (PPAR-c) agonists have salutary effects on adipose tissue over the short term. In people with Type 2 diabetes, thiazolidinediones increase total body fat with a preferential increase in subcutaneous fat with or without a decrease in visceral fat, Correspondence to: Zubin Punthakee. E-mail: [email protected]
1086
leading to a lower proportion of fat in the visceral compartment [3–7]. This is reflected by reduced ectopic fat in the liver [3–6]. Thiazolidinediones also increase circulating adiponectin and reduce interleukin 6 (IL-6) and/or tumour necrosis factor-alpha (TNF-a) levels in people with diabetes [8–10]. However, the effect of thiazolidinediones on directly measured body composition and adipokines beyond 1 year of therapy, and in people without diabetes, particularly those with impaired fasting glucose or impaired glucose tolerance has not been previously reported. In the DREAM trial (Diabetes REduction Assessment with ramipril and rosiglitazone Medication), rosiglitazone
ª 2014 The Authors. Diabetic Medicine ª 2014 Diabetes UK
Research article
What’s new? • Most available studies show the effect of thiazolidinediones on body composition after 1 year in people with diabetes. • After 3.5 years of exposure to rosiglitazone compared with placebo, people with impaired fasting glucose or impaired glucose tolerance have more subcutaneous fat with relatively less visceral fat and liver fat, and a greater increase in adiponectin. • The long-term glucose-lowering effect of rosiglitazone is independent of the effects on body composition and adipokines. significantly reduced the incidence of diabetes by 60% and increased regression to normoglycaemia by 70% compared with placebo among people with impaired fasting glucose and/or impaired glucose tolerance [11]. These striking benefits on glucose homeostasis were accompanied by a significant gain in body weight, but a reduction in waist–hip ratio, because of a 2-cm increase in hip circumference and no effect on waist circumference [11]. There was also a reduction in circulating alanine aminotransferase (ALT), a marker associated with ectopic hepatic fat deposition [12]. These gross changes in body weight and shape noted during the DREAM trial prompted a more accurate characterization of body fat distribution, with detailed imaging studies in a subset of trial participants at the end of the trial. We hypothesized that long-term use of rosiglitazone by people with impaired fasting glucose and/or impaired glucose tolerance would be associated with lower amounts of directly measured visceral fat and liver fat, more subcutaneous fat, increases in circulating adiponectin and reductions in free fatty acids, IL-6 and TNF-a, as well as glucose levels.
Research design and methods The design and results of the DREAM trial are described elsewhere [11,13]. For the DREAM trial, 5269 men and women of at least 30 years of age with impaired fasting glucose and/or impaired glucose tolerance and without a history of coronary, cerebral or peripheral artery disease were recruited from 2001 to 2003. Participants were randomized to rosiglitazone (4 mg daily for the first 2 months, then 8 mg daily) or placebo, as well as to ramipril (15 mg daily) or placebo using a factorial design. Study visits were every 6 months, and blood samples were taken in the fasting state and 2 h after a 75-g oral glucose load at baseline, and at study end for central analysis. The effects of rosiglitazone on body composition at the end of the trial and on adipokine levels during the trial were assessed in a DREAM sub-study conducted at two Canadian study sites with the required imaging equipment (Hamilton
ª 2014 The Authors. Diabetic Medicine ª 2014 Diabetes UK
DIABETICMedicine
and Quebec City). All non-pregnant participants at these sites who weighed ≤ 136 kg (300 lbs) were asked to participate, and consenting participants had an abdominal computed tomography (CT) scan and a whole-body dual-energy X-ray absorptiometry (DXA) scan up to 30 days after their final DREAM visit. All analyses were carried out without knowledge of participants’ treatment allocation, adherence or diabetes status. Ethics approval and written informed consent were obtained at both participating institutions.
Body composition measures
Abdominal CT scan images were obtained from one scanner at each site using the same protocol to ensure standardized attenuations and image sizes for evaluating fat areas (140 kV, 220 mA, field of view 480 9 480 mm, matrix 512 9 512). All scans were obtained between January and June 2006. CT Quality Control Units (phantoms) developed at the Quebec Heart and Lung Institute were scanned at each site before the first participant and after the last participant to ensure reliability over time at each site. Computed tomography images were analysed centrally by a trained individual blind to treatment allocation using SliceOmatic software (Tomovision, Montreal, QC, Canada) [14]. For assessment of visceral and subcutaneous abdominal fat areas, a transverse 10-mm CT scan slice was acquired at the L4–L5 intervertebral space. An attenuation window of –190 to –30 Hounsfield units (HU) was used to identify abdominal adipose tissue [15]. The boundary separating subcutaneous and visceral fat was traced manually and visceral abdominal fat, subcutaneous abdominal fat and the visceral:subcutaneous abdominal fat ratio [16,17] were calculated. For assessment of hepatic fat, a transverse 10-mm CT scan slice was acquired at the T11–T12 intervertebral space. The mean HU of the liver and the spleen were calculated, as well as the ratio of the two mean attenuations [14,18]. Lower values for the mean HU reflect greater fat content. The ratio of liver HU:spleen HU has been validated against liver histology and it correlates highly (r = –0.86) with hepatic fat content by histomorphometry [19,20]. Whole-body DXA scans were completed at both study sites using GE Lunar Prodigy scanners (GE Healthcare, Madison, WI, USA) and the manufacturer’s standard protocol. A variable composition phantom (Lunar, Madison, WI, USA) was scanned before the first participant and after the last participant at each site to ensure reliability between sites and over time. Data files were analysed centrally using Encore Software, version 8.80, 2004 (Lunar) by a single trained individual blind to treatment allocation to estimate total body fat and leg fat mass.
Biochemical measures
Blood samples from baseline and study end were stored at – 120 °C until analysed centrally. Free fatty acids were
1087
DIABETICMedicine
Rosiglitazone, fat and glucose metabolism in DREAM Z. Punthakee et al.
measured using a colorimetric method that has a detection limit of 5.0 mmol/l and coefficient of variation of 2.7%. TNF-a was determined using ELISA (R&D Systems Inc., Minneapolis, MN, USA), with a detection limit of 0.12 ng/l and coefficient of variation of 16.7%. IL-6 was measured by ELISA (R&D Systems Inc.), with a detection limit of 0.04 ng/l and a coefficient of variation of 9.6%. Adiponectin was assayed by ELISA (Linco Research Inc., St Charles, MO, USA) with a detection limit of 1 lg/l and a coefficient of variation of 8.4%. Locally measured ALT was recorded at baseline and at 1 year.
Statistical analysis
Participants were analysed according to the intention-to-treat principle. Characteristics of participants at baseline, at the end of the study and at changes in adipokine levels are presented for rosiglitazone and placebo groups as mean SD or n (%) and were compared by t-tests or v2-tests, respectively. Linear regression was used to assess the relationship between rosiglitazone and changes in adipokine levels, measures of body fat and glucose levels, adjusted for ramipril group allocation, age, sex and baseline BMI. Models for measures of body fat were additionally adjusted for total body fat at study end and changes in free fatty acids, adiponectin, TNF-a and IL-6. Regression models for glucose levels were adjusted for ramipril group allocation, age, sex and baseline BMI, change in adiponectin and body fat measures. All analyses were conducted using SAS version 9.1 (SAS Institute, Cary, NC, USA).
Results Of the 264 individuals who were randomized at the two sites and who attended the final visit, 190 participated in this sub-study (88 allocated to rosiglitazone and 102 allocated to placebo; see Fig. 1). The proportions excluded or declined were not different between the groups (2 d.f., v2 = 2.77, P = 0.25). A total of 153 participants had adequate CT and DXA scans to complete all measurements and 128 participants also had sufficient remaining blood samples for all biochemical measures. The proportions lacking CT and blood data were not different between the two groups (2 d.f., v2 = 4.54, P = 0.10). Participants had been followed for a mean of 3.5 years (SD 0.6) after randomization. Baseline characteristics of participants are shown in Table 1. There were no significant differences between the rosiglitazone and placebo groups. Adherence data were reported by 181 participants at the end of the study, and 79/82 (96%) of the rosiglitazone group and 96/99 (97%) of the placebo group were at least 80% adherent to study medication. At the end of the study, the mean daily dose of rosiglitazone was 7.66 mg (SD 1.12), and nobody in the placebo group was receiving open-label thiazolidinedione.
1088
264 randomized in DREAM 1 died 0 were lost 0 withdrew consent 263 attended final DREAM visit
129 had been randomized to rosiglitazone
134 had been randomized to placebo
Excluded: 3 >136 kg 2 pregnancy risk 36 declined
Excluded: 2 >136 kg 0 pregnancy risk 30 declined
88 sub-study participants
102 sub-study participants
Completed assessments 88 DXA 72 DXA and CT 65 DXA and CT and blood
Completed assessments 102 DXA 81 DXA and CT 63 DXA and CT and blood
FIGURE 1 Enrolment and evaluation of study participants. CT, computed tomography; DXA, dual-energy X-ray absorptiometry.
Effect on body fat compartments
Total body weight gain and total body weight were 3.4 kg (P = 0.005) and 6.0 kg (P = 0.02) higher, respectively, in the rosiglitazone group than the placebo group at the end of the study. Total fat mass by DXA was 4.1 kg (95% CI 2.1–6.2; P < 0.0001) higher in the rosiglitazone group than the placebo group at the end of the study after adjustment for age, sex, baseline BMI and ramipril allocation. Lean mass by DXA did not differ (P = 0.9). The effects of rosiglitazone on visceral fat and ectopic hepatic fat are shown in Fig. 2. After adjustment for age, sex, baseline BMI and ramipril allocation, there was no difference in visceral abdominal fat area between the rosiglitazone and placebo groups. After additional adjustment for final total body fat mass to account for the increase in total fat with rosiglitazone, visceral abdominal fat area was 23 cm² lower (95% CI 5–41; P = 0.01) in the rosiglitazone group. In similar models, rosiglitazone was also associated with a lower visceral:subcutaneous abdominal fat ratio by 0.104 (95% CI 0.027–0.181; P = 0.008) compared with placebo and a higher liver:spleen ratio by 0.08 (95% CI 0.01–0.15; P = 0.02), suggesting less hepatic fat relative to the total amount of body fat. The effects of rosiglitazone on these measures of fat distribution were eliminated after adjustment for the change in fasting free fatty acids or the change in fasting adiponectin from baseline. Measures of subcutaneous fat, subcutaneous abdominal fat assessed by CT and leg fat assessed by DXA were higher in the rosiglitazone group by 31 cm2 (95% CI 9–53;
ª 2014 The Authors. Diabetic Medicine ª 2014 Diabetes UK
Research article
DIABETICMedicine
P = 0.007) and 1.9 kg (95% CI 1.0–2.7; P < 0.0001), respectively, after adjustment for age, sex, baseline BMI and ramipril allocation. The associations of rosiglitazone with subcutaneous fat compartments were attenuated after adjustment for total body fat mass. Addition of adipokines into the models did not have an important effect on the observed relationship between rosiglitazone and subcutaneous fat (Fig. 2).
Effect on adipokines
There was a large and significant increase in adiponectin in the rosiglitazone group (Table 2). After adjustment for age, sex, baseline BMI and ramipril allocation, rosiglitazone Table 1 Baseline characteristics* Rosiglitazone n (%)/mean (SD) Number of participants Ramipril arm allocation Ramipril Placebo Age (years) Female > 3 alcoholic drinks/week Isolated impaired fasting glucose† Isolated impaired glucose tolerance‡ Impaired fasting glucose + impaired glucose tolerance§ Weight (kg) BMI (kg/m2) Waist circumference (cm) Men Women Hip circumference (cm) Men Women Waist–hip ratio Men Women Systolic blood pressure (mmHg) Diastolic blood pressure (mmHg) Fasting plasma glucose (mmol/l) 2-h plasma glucose (mmol/l)¶ HOMA-IR Free fatty acids (lmol/l) Adiponectin (lg/ml) TNF-a (pg/ml) IL-6 (pg/ml) ALT (U/l)
88 49 39 53.3 44 36 15 52
Placebo n (%)/mean (SD)
50 52 52.8 62 33 13 68
(49%) (51%) (10.2) (61%) (32%) (13%) (67%)
21 (24%)
21 (21%)
85.4 (16.2) 30.5 (5.2)
82.7 (17.9) 30.1 (5.6)
105.4 (12.2) 93.9 (12.2)
103.1 (12.5) 91.9 (13.1)
108.9 (7.1) 110.8 (12.1)
107.3 (10.1) 108.7 (12.4)
0.97 0.85 135 80 5.7 8.6 3.8 495 6.4 1.82 1.80 28
(0.07) (0.05) (17) (10) (0.7) (1.3) (2.5) (139) (3.0) (2.43) (1.27) (16)
0.96 0.84 134 81 5.6 8.7 4.6 525 6.1 1.72 1.71 25
(0.06) (0.07) (18) (10) (0.7) (1.3) (4.4) (170) (3.2) (0.98) (1.08) (13)
*There were no significant differences between groups †6.1 mmol/l ≤ fasting plasma glucose < 7.0 mmol/l and 2-h plasma glucose < 7.8 mmol/l. ‡Fasting plasma glucose < 6.1 mmol/l and 7.8 ≤ 2-h plasma glucose < 11.1 mmol/l. §6.1 mmol/l ≤ fasting plasma glucose < 7.0 mmol/l and 7.8 mmol/l ≤ 2-h plasma glucose < 11.1 mmol/l. –Plasma glucose 2 h after a 75-g oral glucose load. ALT, alanine aminotransferase; HOMA-IR, homeostasis model assessment of insulin resistance; IL-6, interleukin 6; TNF-a, tumour necrosis factor alpha
ª 2014 The Authors. Diabetic Medicine ª 2014 Diabetes UK
Effect on glucose metabolism
Rosiglitazone reduced both fasting and 2-h post-load glucose levels (Table 2). After adjusting for age, sex, baseline BMI and ramipril allocation, rosiglitazone lowered fasting glucose by 0.36 mmol/l more than placebo (95% CI 0.16–0.56; P = 0.0004) and 2-h post-load glucose by 1.21 mmol/l (95% CI 0.51–1.91; P = 0.0008) more than placebo. These reductions were independent of the effects of rosiglitazone on total body fat, visceral, hepatic or subcutaneous fat and adiponectin (Fig. 2).
Discussion
102 (56%) (44%) (9.2) (50%) (41%) (17%) (59%)
raised adiponectin levels 14.9 lg/ml more than placebo (95% CI 11.1–18.7; P < 0.0001). Rosiglitazone had no significant effect on free fatty acids, TNF-a or IL-6 (Table 2). These relationships were observed before and after adjustment for age, sex, baseline BMI and ramipril allocation (data not shown).
In this sub-study of the DREAM trial, exposure to rosiglitazone compared with placebo for a mean of 3.5 years had significant effects on body fat distribution and adiponectin levels in people with impaired fasting glucose/impaired glucose tolerance. Rosiglitazone led to a marked increase in adiponectin levels. While other studies have shown a rise in adiponectin up to 1 year [7,21], the current study indicates that the effect of rosiglitazone is sustained over at least 3.5 years. This was not the case for IL-6 and TNF-a, which were not different between the two groups. Shorter studies have been inconsistent in their findings regarding these pro-inflammatory adipokines [21,22], which suggests that thiazolidinediones may not have direct anti-inflammatory effects, but inflammation may be temporarily reduced while rosiglitazone restores adipose tissue physiology, or that other unmeasured markers of inflammation may be mediating any ongoing anti-inflammatory effect. In this study, participants taking rosiglitazone had significantly more subcutaneous fat measured in the abdomen and legs than those in the placebo group. Adjusting for total fat essentially eliminated the difference, suggesting the increase in fat with rosiglitazone is subcutaneous fat. There was no absolute difference in visceral fat area or liver fat between the groups. However, the visceral:subcutaneous fat ratio was lower after treatment with rosiglitazone, suggesting a redistribution of fat. Furthermore, total body-fat adjusted differences in visceral fat and hepatic fat content were approximately half of the differences noted between obese people with and without diabetes [23–26]. While this observed adjusted difference may be primarily attributable to increases in total body fat in the rosiglitazone group, among people with Type 2 diabetes, differences in subcutaneous and visceral fat of similar magnitude have been seen within 24 weeks in placebo-controlled trials of troglitazone,
1089
Rosiglitazone, fat and glucose metabolism in DREAM Z. Punthakee et al.
DIABETICMedicine
(a) Effect of rosiglitazone after adjustment for:
Visceral abdominal fat/ Subcutaneous abdominal fat
Visceral abdominal fat
Liver:spleen
I: Age, sex, baseline BMI II: I + Total body fat mass III: II + ΔFree fatty acids IV: II + ΔAdiponectin V: II + ΔIL-6 VI: II + ΔTNF-α VII: II + All the above -50
0
50
cm2
(b)
-0.2
-0.1
Less abdominal fat in visceral depot
0
0.1
-0.1
0
More hepatic fat
More abdominal fat in visceral depot
0.1
0.2
Less hepatic fat
Leg fat
Subcutaneous abdominal fat
I: Age, sex, baseline BMI II: I + Total body fat mass III: II + ΔFree fatty acids IV: II + ΔAdiponectin V: II + ΔIL-6 VI: II + ΔTNF-α VII: II + All the above -50
0
50
-2
0
cm2
(c)
2
4
kg
ΔFasting plasma glucose
Δ2-h plasma glucose
I: Age, sex, baseline BMI II: I + Total body fat mass III: II + Visceral abdominal fat IV: III + ΔAdiponectin V: II + Liver:spleen ratio VI: V + ΔAdiponectin -1
-0.5
mmol/l
0
-3
-2
-1
0
mmol/l
FIGURE 2 Effect of rosiglitazone vs. placebo on: (a) visceral and ectopic fat compartments, (b) subcutaneous fat compartments and (c) change in fasting and 2-h post-load glucose levels. Mean difference (rosiglitazone – placebo) and 95% CI. All models are adjusted for ramipril allocation. Visceral abdominal fat, abdominal visceral adipose tissue area; subcutaneous abdominal fat, abdominal subcutaneous adipose tissue area; liver: spleen, ratio of computed tomography (CT) attenuation of liver and spleen; D, change (final – baseline); 2-h plasma glucose, plasma glucose 2 h after 75-g oral glucose load; IL-6, interleukin 6; TNF-a, tumour necrosis factor alpha.
pioglitazone and rosiglitazone [3–5,27], and both rosiglitazone and pioglitazone reduce liver fat by 30–50% within 3–6 months [28]. These observations support the hypothesis that, at various degrees of dysglycaemia, all of the thiazolidinediones ‘shunt’ fat to the subcutaneous depot and away from metabolically sensitive organs, including visceral adipose tissue and the liver. This may have an impact on hepatocellular damage as reflected in the ALT reductions noted in the DREAM trial itself [11], and similar but non-significant reductions in ALT seen during the first year in the current small subsample. The analyses in Fig. 2 suggest that the effects of rosiglitazone on visceral and liver fat are linked to its effect on adiponectin levels. As such, the rise in adiponectin may be a useful marker of rosiglitazone’s effect on ectopic fat in the long term. However, data from this study are not able to indicate whether the effects on fat
1090
distribution mediate adipokine changes, or vice versa, or if there is an unmeasured common antecedent. Thiazolidinediones act mainly through PPAR-c receptors in adipose tissue. Despite this tissue being the main site of action, the current data suggest that the effect of rosiglitazone on fasting and 2-h glucose is not explained by differences in total body fat or visceral or hepatic fat. Furthermore, reductions in homeostasis model assessment of insulin resistance (HOMA-IR) and free fatty acids (thought to be the main mechanism of thiazolidinedione insulin sensitization) were not apparent in this study. Rosiglitazone’s glucose-lowering effect parallels its adiponectin-raising effect, but is not completely explained by it. Indeed, studies in isolated muscle and liver suggest that direct and PPAR-c-independent mechanisms may contribute to the glucose-lowering effects of thiazolidinediones via increased
ª 2014 The Authors. Diabetic Medicine ª 2014 Diabetes UK
Research article
DIABETICMedicine
Table 2 Unadjusted body composition at study end, and anthropometric and biochemical changes from baseline Rosiglitazone n/mean (SD) Number of participants CT scan Visceral abdominal fat area (cm2) Subcutaneous abdominal fat area (cm2) Visceral abdominal fat:subcutaneous abdominal fat ratio Liver attenuation (HU) Spleen attenuation (HU) Liver:spleen HU ratio DXA scan Total fat mass (kg) Total per cent body fat (%) Leg fat (kg) DWeight (kg) DBMI (kg/m2) DWaist circumference (cm) Men Women DHip circumference (cm) Men Women DWaist–hip ratio Men Women DFasting plasma glucose (mmol/l) D2-h plasma glucose (mmol/l) DHOMA-IR DFree fatty acids (lmol/l) DTNF-a (pg/ml) DIL-6 (pg/ml) DAdiponectin (lg/ml) DALT at 1 year (U/l)*
88
Placebo n/mean (SD)
P-value
102
191 333 0.570 53.6 43.4 1.236
(95) (111) (0.303) (8.2) (3.4) (0.205)
183 313 0.635 51.9 43.9 1.178
(80) (119) (0.302) (11.4) (4.6) (0.277)
0.55 0.27 0.17 0.26 0.41 0.12
33.8 39.9 10.96 +2.6 +0.9
(11.7) (8.9) (4.80) (10.8) (3.8)
29.8 37.9 9.37 –0.8 –0.3
(11.2) (10.5) (4.29) (5.0) (1.8)
0.02 0.16 0.02 0.005 0.004
–1.1 (7.8) –0.2 (7.3)
2.1 (6.0) +0.5 (7.1)
0.52 0.61
+2.3 (5.3) +2.6 (7.2)
–1.4 (5.0) –0.3 (6.8)
0.002 0.04
–0.030 –0.022 –0.17 –1.63 –0.9 –109 –0.02 +0.50 +15.0 –5.7
(0.055) (0.041) (0.74) (2.46) (2.6) (198) (1.61) (6.81) (17.8) (12.3)
0.009 +0.008 +0.21 –0.46 –0.2 –67 +0.03 –0.04 +0.4 –2.7
(0.039) (0.056) (0.59) (2.29) (3.7) (206) (1.15) (1.17) (1.8) (9.5)
0.04 0.003 0.0003 0.001 0.14 0.19 0.80 0.44 < 0.0001 0.07
*DALT is reported at 1 year after randomization because ALT was only measured systematically up to 1 year. ALT, alanine aminotransferase; CT, computed tomography; DXA, dual-energy X-ray absorptiometry; HOMA-IR, homeostasis model assessment of insulin resistance; HU, Hounsfield units; IL-6, interleukin 6; SD, standard deviation; TNF-a, tumour necrosis factor alpha.
glycolysis, glycogen synthesis and/or PPAR-a agonism in these organs [29]. Strengths of this study include long duration of treatment and the large number of people with impaired fasting glucose and/or impaired glucose tolerance studied with both detailed imaging and biochemical evaluation. The study had several limitations as well. Participants were recruited into this study at the end of the treatment phase; however, they were analysed according to their randomly assigned groups and, while there were no identified imbalances, we adjusted for baseline characteristics. A major limitation of the trial was the lack of body composition measures at baseline. Thus, we were unable to assess directly the change in body composition measures attributable to rosiglitazone; however, as participants were randomized, had a high rate of participation, and the two groups had similar baseline weight, BMI and waist–hip ratio, there is no reason to suspect a difference in baseline fat distribution. Assessing changes in adipokine levels was possible; however, larger than expected variances in the measures may have reduced the power to detect small differences between the groups. In conclusion, rosiglitazone is associated with long-term changes in adiponectin (but not free fatty acids or inflamma-
ª 2014 The Authors. Diabetic Medicine ª 2014 Diabetes UK
tory adipokines), subcutaneous fat and, after adjustment for total fat, the relative amounts of visceral fat and liver fat. However, none of these are accurate markers of the improved glucose metabolism seen in people with impaired fasting glucose/impaired glucose tolerance treated with rosiglitazone. Thiazolidinediones were never approved for diabetes prevention, and their use as diabetes medications has declined for a variety of reasons [30]. Nevertheless, these findings continue to be relevant to currently available drugs as well as newer drugs with PPAR-c activity. Furthermore, understanding how modulation of visceral and liver fat impacts diabetes risk and glucose metabolism clearly continues to be important for both diabetes prevention and treatment.
Funding sources
This sub-study was supported by a grant from the Canadian Institutes of Health Research. ZP was partly supported by a McMaster University Department of Medicine Career Award. SSA holds the May Cohen Eli Lilly Chair in Women’s Health and the Michael G. DeGroote Heart and Stroke Foundation of Ontario Chair in Population Health Research.
1091
DIABETICMedicine
Rosiglitazone, fat and glucose metabolism in DREAM Z. Punthakee et al.
Competing interests
The Population Health Research Institute has received research grants for the conduct of clinical trials that include rosiglitazone from GlaxoSmithKline. ZP, J-PD, GRD, DLH, SY and HCG have received honoraria from GlaxoSmithKline for advice and/or speaking. SSA has received an unrestricted grant from GlaxoSmithKline to study genetic associations in EpiDREAM screenees. AMS has received honoraria from GlaxoSmithKline for consulting and expert testimony. NA and HJ have nothing to declare.
References 1 Hu FB, Manson JE, Stampfer MJ, Colditz G, Liu S, Solomon CG et al. Diet, lifestyle, and the risk of type 2 diabetes mellitus in women. N Engl J Med 2001; 345: 790–797. 2 Hill MJ, Metcalfe D, McTernan PG. Obesity and diabetes: lipids, ‘nowhere to run to’. Clin Sci (Lond) 2009; 116: 113–123. 3 Carey DG, Cowin GJ, Galloway GJ, Jones NP, Richards JC, Biswas N et al. Effect of rosiglitazone on insulin sensitivity and body composition in type 2 diabetic patients. Obes Res 2002; 10: 1008–1015. 4 Katoh S, Hata S, Matsushima M, Ikemoto S, Inoue Y, Yokoyama J et al. Troglitazone prevents the rise in visceral adiposity and improves fatty liver associated with sulfonylurea therapy—a randomized controlled trial. Metabolism 2001; 50: 414–417. 5 Kelly IE, Han TS, Walsh K, Lean ME. Effects of a thiazolidinedione compound on body fat and fat distribution of patients with type 2 diabetes. Diabetes Care 1999; 22: 288–293. 6 Smith SR, De JL, Volaufova J, Li Y, Xie H, Bray GA. Effect of pioglitazone on body composition and energy expenditure: a randomized controlled trial. Metabolism 2005; 54: 24–32. 7 Bertrand OF, Poirier P, Rodes-Cabau J, Rinfret S, Title LM, Dzavik V et al. Cardiometabolic effects of rosiglitazone in patients with type 2 diabetes and coronary artery bypass grafts: a randomized placebo-controlled clinical trial. Atherosclerosis 2010; 211: 565–573. 8 Hirose H, Kawai T, Yamamoto Y, Taniyama M, Tomita M, Matsubara K et al. Effects of pioglitazone on metabolic parameters, body fat distribution, and serum adiponectin levels in Japanese male patients with type 2 diabetes. Metabolism 2002; 51: 314–317. 9 Miyazaki Y, Mahankali A, Matsuda M, Mahankali S, Hardies J, Cusi K et al. Effect of pioglitazone on abdominal fat distribution and insulin sensitivity in type 2 diabetic patients. J Clin Endocrinol Metab 2002; 87: 2784–2791. 10 Shadid S, Stehouwer CD, Jensen MD. Diet/exercise versus pioglitazone: effects of insulin sensitization with decreasing or increasing fat mass on adipokines and inflammatory markers. J Clin Endocrinol Metab 2006; 91: 3418–3425. 11 Gerstein HC, Yusuf S, Bosch J, Pogue J, Sheridan P, Dinccag N et al. Effect of rosiglitazone on the frequency of diabetes in patients with impaired glucose tolerance or impaired fasting glucose: a randomised controlled trial. Lancet 2006; 368: 1096–1105. 12 Klain J, Fraser D, Goldstein J, Peiser J, Avinoah E, Ovnat A et al. Liver histology abnormalities in the morbidly obese. Hepatology 1989; 10: 873–876. 13 Gerstein HC, Yusuf S, Holman R, Bosch J, Pogue J. Rationale, design and recruitment characteristics of a large, simple international trial of diabetes prevention: the DREAM trial. Diabetologia 2004; 47: 1519–1527.
1092
14 Despres JP, Ross R, Boka G, Almeras N, Lemieux I. Effect of rimonabant on the high-triglyceride/low-HDL-cholesterol dyslipidemia, intraabdominal adiposity, and liver fat: the ADAGIO– Lipids trial. Arterioscler Thromb Vasc Biol 2009; 29: 416–423. 15 Sjostrom L, Kvist H, Cederblad A, Tylen U. Determination of total adipose tissue and body fat in women by computed tomography, 40K, and tritium. Am J Physiol 1986; 250: E736–E745. 16 von Eyben FE, Mouritsen E, Holm J, Montvilas P, Dimcevski G, Suciu G et al. Intra-abdominal obesity and metabolic risk factors: a study of young adults. Int J Obes Relat Metab Disord 2003; 27: 941–949. 17 Kelley DE, McKolanis TM, Hegazi RA, Kuller LH, Kalhan SC. Fatty liver in type 2 diabetes mellitus: relation to regional adiposity, fatty acids, and insulin resistance. Am J Physiol Endocrinol Metab 2003; 285: E906–E916. 18 Davidson LE, Kuk JL, Church TS, Ross R. Protocol for measurement of liver fat by computed tomography. J Appl Physiol 2006; 100: 864–868. 19 Longo R, Ricci C, Masutti F, Vidimari R, Croce LS, Bercich L et al. Fatty infiltration of the liver. Quantification by 1H localized magnetic resonance spectroscopy and comparison with computed tomography. Invest Radiol 1993; 28: 297–302. 20 Ricci C, Longo R, Gioulis E, Bosco M, Pollesello P, Masutti F et al. Non-invasive in vivo quantitative assessment of fat content in human liver. J Hepatol 1997; 27: 108–113. 21 Shadid S, Stehouwer CD, Jensen MD. Diet/exercise versus pioglitazone: effects of insulin sensitization with decreasing or increasing fat mass on adipokines and inflammatory markers. J Clin Endocrinol Metab 2006; 91: 3418–3425. 22 Miyazaki Y, Mahankali A, Matsuda M, Mahankali S, Hardies J, Cusi K et al. Effect of pioglitazone on abdominal fat distribution and insulin sensitivity in type 2 diabetic patients. J Clin Endocrinol Metab 2002; 87: 2784–2791. 23 Anjana M, Sandeep S, Deepa R, Vimaleswaran KS, Farooq S, Mohan V. Visceral and central abdominal fat and anthropometry in relation to diabetes in Asian Indians. Diabetes Care 2004; 27: 2948–2953. 24 Boyko EJ, Fujimoto WY, Leonetti DL, Newell-Morris L. Visceral adiposity and risk of type 2 diabetes: a prospective study among Japanese Americans. Diabetes Care 2000; 23: 465–471. 25 Kanaya AM, Harris T, Goodpaster BH, Tylavsky F, Cummings SR. Adipocytokines attenuate the association between visceral adiposity and diabetes in older adults. Diabetes Care 2004; 27: 1375–1380. 26 Kelley DE, McKolanis TM, Hegazi RA, Kuller LH, Kalhan SC. Fatty liver in type 2 diabetes mellitus: relation to regional adiposity, fatty acids, and insulin resistance. Am J Physiol Endocrinol Metab 2003; 285: E906–E916. 27 Shah PK, Mudaliar S, Chang AR, Aroda V, Andre M, Burke P et al. Effects of intensive insulin therapy alone and in combination with pioglitazone on body weight, composition, distribution and liver fat content in patients with type 2 diabetes. Diabetes Obes Metab 2011; 13: 505–510. 28 Yki-Jarvinen H. Thiazolidinediones and the liver in humans. Curr Opin Lipidol 2009; 20: 477–483. 29 Furnsinn C, Waldhausl W. Thiazolidinediones: metabolic actions in vitro. Diabetologia 2002; 45: 1211–1223. 30 Leal I, Romio SA, Schuemie M, Oteri A, Sturkenboom M, Trifiro G. Prescribing pattern of glucose-lowering drugs in the UK in the last decade: a focus on the effects of safety warnings about rosiglitazone. Br J Clin Pharmacol 2013; 75: 861–868.
ª 2014 The Authors. Diabetic Medicine ª 2014 Diabetes UK
