Diabetes Care 1

Prince Dadson,1 Linda Landini,1,2 Effect of Bariatric Surgery Mika Helmio,¨ 3 Jarna C. Hannukainen,1 Heidi Immonen,1,4 Miikka-Juhani Honka,1 on Adipose Tissue Glucose Marco Bucci,1 Nina Savisto,1 Minna Soinio,4 Paulina Salminen,3 Metabolism in Different Depots Riitta Parkkola,5,6 Jussi Pihlajamaki,¨ 7,8 Patricia Iozzo,2 Ele Ferrannini,2 and in Patients With or Without Pirjo Nuutila1,4 Type 2 Diabetes DOI: 10.2337/dc15-1447

OBJECTIVE We investigated fat distribution and tissue-specific insulin-stimulated glucose uptake (GU) in seven fat compartments (visceral and subcutaneous) and skeletal muscle in morbidly obese patients with (T2D) and without (ND) type 2 diabetes before and 6 months after bariatric surgery.

RESEARCH DESIGN AND METHODS A total of 23 obese patients (BMI 43.0 6 3.6 kg/m2;9T2Dand14ND)were recruited from a larger, randomized multicenter SLEEVEPASS study. MRI (for fat PATHOPHYSIOLOGY/COMPLICATIONS 1 distribution) and 18F-fluorodeoxyglucose PET (for GU) studies were performed for Turku PET Centre, University of Turku, Turku, Finland the obese before and 6 months postsurgery; 10 lean subjects served as control 2Institute of Clinical Physiology, National Re- subjects and were studied once. search Council, Pisa, Italy 3Division of Digestive Surgery and Urology, De- RESULTS partment of Acute and Digestive Surgery, Turku At baseline, visceral fat GU was 30 6 7% of muscle GU in control subjects and 57 6 University Hospital, Turku, Finland 4Department of Endocrinology, Turku University 5% in obese patients. Visceral and deep subcutaneous fat were more abundant Hospital, Turku, Finland (despite same total fat mass) and less insulin sensitive in T2D than ND; in both, GU 5Medical Imaging Center, Turku University Hos- was impaired compared with control subjects. Postsurgery, visceral fat mass de- pital, Turku, Finland 6 creased (∼40%) more than subcutaneous fat (7%). Tissue-specific GU was improved, Department of Radiology, University of Turku and Turku University Hospital, Turku, Finland but not normalized, at all sites in T2D and ND alike. The contribution of visceral fat to 7Institute of Public Health and Clinical Nutrition, whole-body GU was greater in T2D than ND but decreased similarly with surgery. University of Eastern Finland, Kuopio, Finland Subcutaneous fat made a fourfold greater contribution to whole-body GU in obese 8Department of Clinical Nutrition and Obesity versus lean subjects (15% vs. 4%) both before and after surgery. Center, University of Eastern Finland, Kuopio, Finland CONCLUSIONS Corresponding author: Pirjo Nuutila, pirjo. nuutila@utu.fi. Bariatric surgery leads to sustained weight loss and improves tissue-specific glucose 1 Received 4 July 2015 and accepted 20 October metabolism in morbidly obese patients. We conclude that ) enhanced visceral fat 2015. accumulation is a feature of T2D, 2) severe obesity compromises muscle insulin Clinical trial reg. no. NCT00793143, clinicaltrials sensitivity more than fat insulin sensitivity, and 3) fat mass expansion is a sink for .gov. plasma glucose. This article contains Supplementary Data online at http://care.diabetesjournals.org/lookup/ suppl/doi:10.2337/dc15-1447/-/DC1. Visceral fat (VF) and subcutaneous fat (SC) are structurally, metabolically, and func- tionally distinct, albeit both contribute to obesity (1). Abdominal SC fat has clearly © 2015 by the American Diabetes Association. Readers may use this article as long as the work is defined, metabolically distinct deep and superficial layers separated by the Scarpa properly cited, the use is educational and not for fascia (2,3). Research suggests that the deep layers are metabolically more active profit, and the work is not altered. Diabetes Care Publish Ahead of Print, published online December 17, 2015 2 Bariatric Surgery, Fat Mass, and GU Diabetes Care

and, hence, heavily linked to metabolic before they were included in the study. clamp technique as previously described complications, particularly to insulin re- The exclusion and inclusion criteria fol- (15). Insulin infusion rate was 1 mU z sistance and type 2 diabetes (T2D) (4,5). lowed international guidelines as previ- min21 z kg21 (Actrapid; Novo Nordisk, Co- The concentration of saturated fatty ously reported (13). The obese patient penhagen, Denmark). During hyperinsuli- acids in the deep layer is higher com- population consisted of 3 men and 20 nemia, plasma glucose levels were pared with the superficial layer, provid- women, of whom 9 were postmeno- maintained at 5 mmol/L using a variable ing additional evidence of its higher pausal. By the criteria of the American infusion of 20% glucose based on arterial- metabolic activity (6). Diabetes Association(14),9patients ized glucose measurements every 5–10 Accumulation of abdominal VF is an had T2D and 14 did not have diabetes min. The rate of whole-body GU was cal- independent risk factor for obesity- (ND) (of whom 5 had impaired glucose culated over the same time period that related metabolic abnormalities (7). VF tolerance and 4 impaired fasting glu- the measurements of adipose tissue GU can be partitioned into intraperitoneal cose). Four of the nine patients with were made and expressed both in total and extraperitoneal compartments using T2D had newly diagnosed T2D and amount (mmol/min) and normalized by anatomical reference points such as the were treated with metformin, and five kilogram of fat-free mass (Mffm)(in 21 21 kidneys and ascending and descending patients had been treated for an average mmol z min z kgffm ). colon (8). The release of free fatty acids of 3 years with combinations of oral glu- (FFAs) from the intraperitoneal depot cose–lowering drugs (metformin in two PET Protocol and Data Acquisition drains into the portal system interfering patients and metformin/sulphonylurea/ The PET studies were conducted both in with hepatic glucose metabolism (9), gliptin, metformin/pioglitazone/gliptin, the fasting condition and during the leading to hyperinsulinemia and insulin and metformin/pioglitazone for each of clamp (15) on separate days ,2weeks resistance. Extraperitoneal fat, on the the remaining three patients). Thirteen apart using the GE advanced PET camera other hand, drains into the systemic cir- (7 T2D and 6 ND) obese patients under- (General Electric Medica Systems, culation and thus is not directly involved went Roux-en-Y gastric bypass (RYGB) Milwaukee, WI). Subjects were in a in hepatic glucose metabolism (8). and10(2T2Dand8ND)patientsre- supine-lying position, and two cathe- Absolute quantification of fat mass is ceived sleeve gastrectomy. Ten lean ters, one for infusion of glucose and in- achieved by MRI, while tissue-specific healthy volunteers with normal OGTT re- sulin and for tracer injection and the glucose metabolism can be noninvasively cruited through advertisement in local other for arterialized blood sampling, assessed using positron emission tomog- newspapers served as control subjects. were inserted into an antecubital vein raphy (PET) with [18F]-fluorodeoxyglu- of each arm. At the time point 100 6 18 cose (FDG) (10). Combined PET/MRI Study Design 10 min of the clamp, [ F]-FDG (187 6 thus provides anatomical and functional Studies in obese patients were per- 9 MBq i.v.) was injected over 15 s. Dynamic information on specific tissue compart- formed twice: before and 6 months af- PET scanning was started in the thoracic ments. Bariatric surgery leads to major, ter bariatric surgery; the baseline region at 60 min (5 3 180 s frames), sustained weight loss and improves met- studies were carried out before they followed by abdominal region at 80 min abolic and lipid profiles in obese indivi- started a 4-week very–low caloric diet (5 3 180 s frames) and then femoral re- duals with or without T2D (11). We (VLCD). Healthy subjects were studied gion at 100 min (3 3 300 s frames). After hypothesized that bariatric surgery may only once. Two MRI/PET acquisitions, the injection of [18F]-FDG, blood samples have differential effects on metabolism one in fasting condition and the other were obtained throughout the PET scan- in different adipose tissue compart- during euglycemic clamp, were per- ning time. Plasma radioactivity level was ments in severely obese patients. We formed at baseline and repeated post- measured using an automatic g counter therefore combined PET/MRI to assess surgery. MRI was performed after 2–3h (Wizard 1480; Wallac, Turku, Finland). the effect of bariatric surgery on fat dis- of fasting and PET after an overnight All PET image data were corrected for tribution and glucose uptake (GU) in dif- fast. Vigorous physical activity was pro- time decay and photon attenuation. ferent fat compartments in severely hibited from the preceding evening. Carimas (version 2.8; Turku PET Center) obese patients with or without diabetes. Glucose-lowering treatment and all di- was used for the analysis of the PET im- abetes medications were withheld 24– ages. PET and MRI images were coregis- RESEARCH DESIGN AND METHODS 72 h before the onset of metabolic studies. tered using the normalized mutual Patients The flowchart of the study design is shown information (NMI) (16).Three-dimensional Patient population and study design in Supplementary Fig. 1. The PET/MRI re- volumes of interest were manually drawn have previously been described (12). sults in the obese group were compared in the different adipose tissue compart- Briefly, a total of 23 morbidly obese pa- with those from the lean healthy control ments while avoiding bone, muscle, and tients (BMI $40 kg/m2 or $36 kg/m2 subjects. All studies were performed in ac- skin and in skeletal muscles without in- with an additional obesity-related co- cordance with the principles of the Decla- termuscular fat. The fractional uptake morbid condition) were recruited from ration of Helsinki and approved by the rate, like the influx constant (Ki), was cal- the SLEEVEPASS study, a larger randomized ethics committee of the Hospital District culated for each voxel drawn using the controlled clinical trial comparing different of the Southwestern Finland. tissue and plasma inputs time activity surgical techniques for the treatment curves (17). GU for adipose tissue and of morbid obesity (clinicaltrials.gov, Whole-Body GU skeletal muscle were calculated at the NCT00793143). All subjects gave in- Whole-body GU rate was determined voxel level by multiplying the Ki by the formed consent and were screened using the euglycemic hyperinsulinemic plasma glucose concentration and dividing care.diabetesjournals.org Dadson and Associates 3

this by a lumped constant of 1.14 for 0.05 considered statistically significant. 43.0 6 9.2 years, P = 0.01). Altogether, adipose tissue (10) and 1.2 for skeletal Continuous variables are expressed as the MRI-scanned fat regions constituted muscle (18). The depot-specific GU was mean 6 SD. Normality of distribution .90% of the total fat mass as assessed calculated by multiplying tissue-specific was assessed using the Shapiro-Wilk by bioelectric impedance. Compared GU (mmol z L21 z min21) by the size of the test. Variables that were not normally with sex-matched control subjects, the fat depot (in kg) corrected for tissue den- distributed were log transformed before obese group as a whole had expanded sity. Supplemental Fig. 2 is a representa- analysis. Pearson correlation analyses fat depots at all locations, subcutaneous tive example of MRI and insulin-stimulated were performed to investigate the univar- and visceral alike, and displayed insulin [18F]-FDG PET images in abdominal fat re- iate associations between fat distribution resistance, hyperinsulinemia, dyslipidemia, gions of obese before (left) and 6 months and tissue-specific GU with metabolic and elevated hs-CRP as a marker of subclin- after (right) surgery. Regions of interest and lipid variables. Two-way repeated- ical inflammation. In the T2D group, VF mass are outlined with black in SC and VF. MRI measures ANOVA was used to assess was larger (Table 1) than in ND (7.4 6 0.7 vs. and PET images were obtained at the the interaction of T2D status with surgery. 4.7 6 0.3 kg, P , 0.01) and insulin resistance level of the umbilicus. In the lower panel was worse (Mffm 19.2 6 1.2 vs. 28.0 6 21 21 are PET images with scale bar showing RESULTS 0.8 mmol z min z kgffm , P , 0.03). levels of activity (Bq/mL) in the various Anthropometric and Metabolic T2D patients were classified as re- regions. Characteristics mitters if they achieved 2-h glucose There was no difference in age between levels ,7.8 mmol/L on the OGTT and Measurement of Adipose Tissue Mass obese and control subjects, but the fasting glucose levels ,7.0 mmol/L Body fat content was measured using obese T2D patient group was older without diabetes medication (14). bioelectric impedance (Omron BF400). than the ND group (53.3 6 4.4 vs. Based on this criterion, five of the The volumes of different adipose depots were measured using 1.5 Tesla MRI (Tesla Intera system; Philips Medical Systems, Best, the Netherlands). Thorax Table 1—Anthropometric and metabolic characteristics of control and obese and upper arm, abdominal, and femoral subjects Obese patients (n =23) regions were acquiredwithaxialT1- Control subjects weighted dual fast fieldechoimages (n =10) Before surgery After surgery P (echo time 2.3 and 4.6, repetition time Age (years) 47.3 6 6.0 46.5 6 9.0#,a 46.8 6 8.9 d 120 ms, slice thickness 10 mm without Sex (female/male) 8/2 19/4 19/2 ns gap). The different fat regions were an- Body weight (kg) 69.2 6 6.7 121.9 6 11.6† 93.2 6 12.9† ,0.001 alyzed using sliceOmatic (Tomovision, BMI (kg/m2)23.76 1.8 43.1 6 3.6† 33.2 6 4.1† ,0.001 Montreal, Quebec, Canada). SC adipose Waist (cm) 81.4 6 10.1 123.9 6 8.3† 104.3 6 10.6† ,0.001 tissue compartment was defined as the WHR (cm/cm) 0.8 6 0.1 0.9 6 0.1 0.9 6 0.1 ns adipose tissue between the skin and the Percent body fat 31.4 6 6.5 48.0 6 7.0† 42.0 6 6.4† ,0.0018 outermost regions of the muscle wall. Fat mass (kg) 21.7 6 4.5 57.8 6 8.7† 39.1 6 7.6† ,0.001 Femoral SC mass was calculated from Fat-free mass (kg) 47.6 6 7.5 63.2 6 12.2† 54.2 6 10.8 ,0.001 the femoral head to the patella surface, 6 6 † 6 † , humeral SC from humeral head to the Total abdominal SC (kg) 6.8 2.4 26.7 6.0 15.9 4.7 0.001 6 6 a 6 , 8 region of the capitulum, thoracic SC Total abdominal VF (kg) 2.2 1.2 5.6 2.0#, 3.5 2.0 0.001 from the clavicle to the diaphragm, Thoracic SC (kg) 1.3 6 0.4 4.7 6 0.7† 3.2 6 0.8† ,0.001 and abdominal SC from diaphragm to Femoral SC (kg) 7.1 6 2.8 16.4 6 4.2† 11.5 6 4.0† ,0.001 the head of the femur. The abdominal Humeral SC (kg) 0.5 6 0.2 1.0 6 0.2† 0.7 6 0.1† ,0.001 SC compartment was divided in anterior Posterior deep SC (kg) 1.3 6 0.6 5.7 6 1.9#,b,† 3.4 6 1.3† ,0.0001 and posterior regions by drawing a line Fasting FFA (mmol/L) 0.5 6 0.2 0.6 6 0.2 0.5 6 0.1 ,0.04 along the coronal plane passing through Triglycerides (mmol/L) 0.8 6 0.3 1.3 6 0.5† 0.9 6 0.2 ,0.001 the anterior surface of the vertebral HDL cholesterol (mmol/L) 2.0 6 0.5 1.3 6 0.5† 1.5 6 0.3† 0.001 bodies (19). The posterior region was LDL cholesterol (mmol/L) 2.7 6 0.6 2.4 6 0.6 2.2 6 0.6† ns a further divided into deep and superficial HbA1c (%) 5.7 6 0.2 5.9 6 0.7#, 5.6 6 0.3 0.018 a areas using the Scarpa fascia as a guide HbA1c (mmol/mol) 39.0 6 2.4 47.1 6 7.4#, 37.8 6 3.7 0.018 (20). Abdominal VF region was sepa- Fasting glucose (mmol/L) 5.5 6 0.4 6.5 6 1.4#,a 5.4 6 0.6 ,0.0018 rated into the intraperitoneal and Fasting insulin (mmol/L) 6.7 6 3.7 17.4 6 13.1† 6.5 6 3.0 ,0.001 extraperitoneal compartments using hs-CRP (mg/L) 1.1 6 0.8 4.6 6 4.1† 1.7 6 1.5 ,0.001 fi speci c anatomical reference points 2-h glucose (mmol/L) 5.7 6 1.2 9.4 6 3.4#,a,† 6.4 6 2.9 ,0.0018 3 (2). Fat volumes (cm ) were converted 21 21 b Mffm (mmol z min z kgffm )58.96 13.6 24.6 6 11.8#, ,† 40.9 6 15.4† ,0.001 to mass (kg) by taking into account the Whole-body M (mmol/min) 2.77 6 0.61 1.50 6 0.66#,b,† 2.13 6 0.66† ,0.001 tissue density (8). Data are mean 6 SD or n/n. 2-h glucose, levels after a standard oral glucose tolerance test. P Statistical Analysis value for presurgery vs. postsurgery comparisons. #P # 0.05 for the difference between ND and Data analysis was performed using SPSS T2D at baseline; †P # 0.05 for the comparison with control subjects; 8P # 0.05 for the interaction surgery 3 T2D; aT2D . ND; bT2D , ND. ns, nonsignificant. (version 22; SPSS, Chicago, IL) with P # 4 Bariatric Surgery, Fat Mass, and GU Diabetes Care

nine T2D subjects achieved remission with postsurgery total VF mass (b 20.70, We did not find differences in tissue- at 6 months postsurgery. P , 0.001, R2 =50%)butnotwiththe specific GU in any of the studied regions Six months after surgery, T2D and ND other fat regions. Drug regimen did between the normoglycemic and predi- patients had lost similar amounts of not influence postoperative fat distri- abetic (impaired glucose tolerance and body weight (30 6 3and276 2kg, bution and tissue-specificGUinthe impaired fasting glucose) groups. Men- respectively), and fat depots and fat- obese group. opause in obese women did not influ- free mass were all reduced in size, al- ence insulin-stimulated GU in skeletal though they were still larger than in con- Insulin-Stimulated GU muscle or any fat regions after the sur- trol subjects. At this time, metabolic Before surgery, all obese subjects had gical intervention. parameters had improved significantly severe skeletal muscle insulin resis- in both groups; HbA and glucose levels tance, which improved markedly after Fasting GU 1c Presurgery, fasting tissue-specificGU declined more in the T2D group, while surgery, though remaining lower than rates were lower than in the insulin- whole-body insulin sensitivity remained in control subjects (Table 2 and Fig. stimulated state at all sites and did not subnormal in both groups. 1A). Presurgery, insulin-stimulated tissue- show major differences between obese At baseline, extraperitoneal fat mass specific GU was impaired in obese pa- patients and control subjects (except for was significantly larger in T2D than ND tients compared with control subjects significantly lower extraperitoneal VF (2.8 6 1.2 vs. 1.8 6 0.6 kg, P =0.02). in all fat compartments (though falling uptake in obese vs. control subjects); Similarly, T2D patients had increased short of statistical significance for hu- surgery did not alter these rates of GU amounts of intraperitoneal fat than ND meral and thoracic depots) and increased significantly. On the other hand, depot- (4.5 6 0.8 vs. 2.9 6 0.7 kg, P , 0.001). significantly in all of them postsurgery (re- specific GU was larger in obese than After surgery, total VF mass remained maining subnormal only in extraperito- control subjects in all compartments higher in T2D than ND (4.7 6 0.9 vs. neal VF). Within the obese group, VF and was higher in T2D compared with 2.8 6 0.3 kg, P = 0.02), while abdominal and abdominal deep SC adipose tissue ND (18.7 6 7.0 vs. 11.1 6 4.9, P = deepSCwashigherinNDthanT2D GU was lower in T2D than ND (Fig. 1B). 0.01); surgery decreased depot-specific (6.3 6 2.1 vs. 4.5 6 0.8 kg, P =0.03). Because of fat mass expansion, pre- GU at all sites (except for the humeral There were no differences between surgery depot-specific GU was increased depot) (Supplementary Table 1). the two types of surgery on the remission in obese compared with control subjects of T2D postsurgery. Preoperatively, the in all compartments (except for extrap- Contribution of Different Fat Depots prediabetic group (impaired glucose toler- eritoneal VF and femoral SC sites). Com- to Whole-Body Glucose Disposal ance and impaired fasting glucose) had paredwithNDsubjects,T2Dpatients Before surgery, the contribution of VF higher intraperitoneal (3.5 6 0.7 vs. had higher uptake in the intraperitoneal GU to whole-body insulin-mediated glu- 2.5 6 0.4, P = 0.01) and extraperitoneal VF depot and lower uptake in abdominal cose disposal was twice as high in ND (2.3 6 0.6 vs. 1.5 6 0.4, P = 0.01) fat deep and femoral SC depots. After sur- and four times as high in T2D patients mass than the normoglycemic group. gery, depot-specific GU in intraperito- (P , 0.05 between ND and T2D) com- These differences in fat mass were no neal VF decreased significantly in both pared with control subjects (Fig. 1C). All longer there postoperatively. Menopausal T2D and ND, whereas GU in all other SC depots, on the other hand, contrib- status of obese women was associated depots was unchanged. uted four times more in patients, ND

Table 2—Insulin-stimulated GU in fat compartments and skeletal muscle* Obese patients (n =23) Control subjects (n = 10) Before surgery After surgery P Tissue-specificGU(mmol z L21 z min21) Extraperitoneal VF 26.7 6 11.7 8.3 6 2.7#† 13.5 6 5.1† ,0.0001 Intraperitoneal VF 17.9 6 7.5 10.1 6 2.7#,b,† 13.8 6 3.9 0.0002 Abdominal deep SC 17.3 6 9.1 6.3 6 1.9#,b,† 11.7 6 4.9 ,0.0001 Abdominal superficial SC 11.8 6 5.5 5.8 6 1.7† 9.3 6 3.4 ,0.0001 Femoral SC 10.2 6 3.9 5.0 6 1.2† 8.9 6 4.6 0.0007 Humeral SC 8.6 6 4.4 6.7 6 1.9 9.4 6 3.7 0.0041 Thoracic SC 8.8 6 4.4 7.2 6 2.3 9.2 6 3.8 ,0.03 Skeletal muscle 72.5 6 26.2 22.0 6 17.6† 45.7 6 31.6† 0.003 Depot-specific GU(mmol z min21) Extraperitoneal VF 27.3 6 14.6 19.4 6 7.7 20.9 6 10.4 ns Intraperitoneal VF 17.5 6 5.8 36.6 6 11.2#,a,† 28.1 6 12.5† ,0.0001 Abdominal deep SC 23.6 6 16.6 39.8 6 21.3#,b,† 44.4 6 23.7† ns Abdominal superficial SC 19.4 6 13.5 41.3 6 21.2† 40.7 6 21.8† ns Femoral SC 89.9 6 54.5 93.0 6 34.1#,b 113.3 6 53.7 ns Humeral SC 4.3 6 3.7 6.9 6 2.5† 7.4 6 3.5† ns Thoracic SC 12.9 6 7.6 36.1 6 11.7† 31.9 6 12.3† ns Data are mean 6 SD. P value for presurgery vs. postsurgery comparisons. #P # 0.05 for the difference between ND and T2D at baseline; †P # 0.05 for the comparison with control subjects; 8P # 0.05 for the interaction surgery 3 T2D; aT2D . ND; bT2D , ND. ns, nonsignificant. care.diabetesjournals.org Dadson and Associates 5

Figure 1—Skeletal muscle (A) and VF (B) insulin-mediated GU in lean control subjects and in obese ND and T2D patients before and 6 months after bariatric surgery. Percentage contribution of VF depots (C) and SC depots (D) to whole-body insulin-mediated glucose uptake in lean control subjects and in obese ND and T2D patients before and 6 months after bariatric surgery. Data are mean 6 SEM. *P # 0.05 compared with control subjects; §P # 0.05 for the pre- to postsurgery comparison.

and T2D alike, than in control subjects. CONCLUSIONS The second major finding is that VF Combined VF and SC fat contribution to The first major finding of the current GU was 30 6 7% of skeletal muscle GU total GU was 5.6 6 2.5% in control sub- study is that prior to surgery, obese in control subjects but twice as much jects and 20.2 6 9.5% in the whole pa- T2D patients had more VF mass than (57 6 5%) in obese patients whether tient group (P , 0.0001). After surgery, ND patients (13 6 7% vs. 8 6 3% of total T2D or ND. Thus, the presence of severe the contribution of VF fat was significantly fat mass, P , 0.02) despite virtually obesity compromises skeletal muscle in- reduced (but still higher in T2D than in identical total fat mass (57 6 4vs. sulin sensitivity more than adipose tis- control subjects), whereas that of SC tis- 58 6 2kg,P = nonsignificant). Further- sue insulin sensitivity. The cellular basis sue was unchanged (Fig. 1C and D). more, insulin-stimulated GU in VF tis- of this discrepancy, which to our knowl- In the whole data set, GU rate into VF sueswasmoreimpairedinT2Dthan edge has not been reported before, re- (intraperitoneal plus extraperitoneal) ND (Table 2). Therefore, the presence mains to be investigated. In any event, was inversely related to both 2-h plasma of T2D (and ;10-year older age) was as a combined result of less compro- glucose and suppression of lipolysis dur- associated with a selective visceral de- mised insulin sensitivity and the ex- ing insulin stimulation (Fig. 2). Upon cat- position of extra adipose tissue more re- panded VF pool size, the contribution egorizing obese subjects in T2D and ND sistant to insulin action. In addition to of VF to whole-body glucose disposal groups, we found that clamped FFA lev- the endogenous glucose overproduc- under insulinized conditions was twice els significantly correlated with VF GU in tion and b-cell dysfunction typical of as much in ND obese and four times as ND (r = 20.77, P = 0.001) and not in the T2D, the metabolic impact of this abnor- much in T2D obese than in control sub- T2D group (P =0.22).Nosignificant cor- mal adipose tissue phenotype is re- jects (Fig. 1B). Moreover, SC fat GU av- relation was registered between 2-h flected in the reciprocal relation of VF eraged 15% of whole-body glucose plasma glucose and VF GU in ND or GU to both 2-h glucose levels and circu- disposal in the obese patients versus 4% T2D groups. lating FFA concentrations (Fig. 2). of control subjects (Fig. 1D). Therefore, 6 Bariatric Surgery, Fat Mass, and GU Diabetes Care

Of note is that percentagewise, bari- hepatic insulin resistance in men but not atric surgery induced a greater loss of VF women. It is therefore conceivable that (;40% of presurgery size) than SC fat sex-specific differences in the distribution (7%) equally in ND and T2D. This phe- of abdominal SC fat subcompartments nomenon was also reported by Gray may account for these discrepant findings. et al. (22), who found that subjects with To our knowledge, there are relatively more intra-abdominal fat at baseline few studies that have explored the use tended to lose more fat from this depot of dynamic FDG PET to quantitatively during weight loss. Studies by Kim et al. analyze tissue-specificGUinVFandSC (23) likewise suggested that weight loss fat (10,30) and skeletal muscle (31). after bariatric surgery preferentially tar- Based on the current data, obese T2D gets VF in T2D patients. Although there have decreased GU in skeletal muscles are well-established links between in- compared with ND and control subjects; creased intra-abdominal fat accumulation this confirms previous results by our group and T2D (5,22), previous studies did not (31). The decreased skeletal muscle GU in separate intraperitoneal from extraperi- T2D in response to insulin may be due, at toneal fat compartments and did not least in part, to increased FFA levels (32), quantify intra-abdominal fat for the i.e., a model of substrate competition. whole abdominal region. Upon accurately With regard to fat glucose metabo- compartmentalizing intra-abdominal VF, lism, research suggests that VF depots we found that T2D patients had increased tends to have higher rate of insulin- amounts of both intra- and extraperito- stimulated GU compared with SC com- Figure 2—Nonlinear, reciprocal relationship neal fat compared with obese ND pa- partments (33). In obese individuals, GU between VF GU and 2-h plasma glucose con- tients or control subjects. in visceral and SC compartments is centrations (A) and steady-state plasma FFA That increased amounts of total VF lower than in nonobese subjects (10). on the clamp (B) in the pooled pre- and postsurgery data set. Lines are best fitwith heighten risk of developing insulin resis- Our findings confirm that expanded fat 95% CIs. tance seems to be well established (24). mass leads to decreased GU, particularly However, Kobayashi et al. (25) found no in the VF fat regions. difference in intra-abdominal VF be- Dividing the abdominal SC fat into despite the insulin resistance, fat expan- tween T2D men and women with similar deep and superficial regions and VF sion in the obese provides a sink for glu- BMI. Abate et al. (26) stated that intra- into intraperitoneal and extraperitoneal cose, which would otherwise accumulate peritoneal fat volume in males with T2D yielded interesting results. We observed in the bloodstream and further impair tends to be similar to that in healthy that obese ND patients tended to have glucose tolerance. The converse problem, control subjects with similar BMI. The significantly higher GU in intraperito- namely, the inability of adipose tissue to apparent contradictions with our results neal VF fat and deep SC fat than T2D. expand, is also conducive to reduced glu- may stem from gender differences, as The relatively higher metabolic activity cose tolerance and dyslipidemia (21). Abate et al. studied only males, while in abdominal VF fat compared with SC Surgically induced major weight loss 80% of the current study subjects were fat depot is well documented (34), but brought about improved insulin sensi- female. the metabolic potential in deep SC de- tivity and, in the T2D group, improved Differences in the partition of abdom- pots has been likened to VF fat (29,35). glucose tolerance, as expected. How- inal SC fat into deep and superficial This metabolic similarity is attributed to ever, skeletal muscle insulin-mediated layers in obese and lean subjects and the restricted capacity of peripheral fat GU was still subnormal in the obese their metabolic implications have previ- depot to store excessive energy, leading group, especially in T2D patients, and ously been elucidated by He, Engelson, to overflow of energy into other fat de- VF insulin sensitivity was still impaired and Kotler (20). In our data, obese ND pots such as deep SC and VF compart- in the T2D group (Fig. 1B). Whether in- patients have increased posterior deep ments (36). Research has shown that sulin action would eventually normalize SC fat mass compared with T2D (6.3 6 adipocytes from the abdominal deep with further weight loss cannot be de- 2.1 vs. 4.5 6 0.8, P , 0.05). One could depots are metabolically superior to su- termined from the current data, as therefore speculate that expanded deep perficial SC compartments (29), thereby these very obese patients rarely regain SC fat might be metabolically protective, contributing significantly to global GU. lean body weight. What is remarkable is although few studies have established Indeed, Abate et al. (26) have reported that despite the massive reduction in fat direct association between abdominal that intraperitoneal, but not extraperito- mass, SC depots continued to make a deep SC depot (27,28), posterior SC ab- neal, fat is a better correlate of metabolic threefold greater contribution to total dominal fat (2), and metabolic complica- complications. It is worth mentioning that insulin-mediated glucose disposald tions such as insulin resistance. According the expanded VF in T2D is accompanied compared with lean control subjectsdas to Marinou et al. (29), expanded deep SC by increased adipocyte size, which tends a result of more efficient glucose utiliza- fat is a major feature of increased adipos- to correlate positively with insulin tion. Thus, in the postsurgical state resid- ity in men but not women. Furthermore, resistance (37). ual obesity continues to protect glycemia Miyazaki et al. (5) showed that deep SC The use of [18F]-FDG PET for the as- from rising higher. fat is associated with peripheral and sessment of inflammation in adipose care.diabetesjournals.org Dadson and Associates 7

tissue has been reported on previously Academy of Finland, University of Turku, Abo˚ 10. Virtanen KA, Peltoniemi P, Marjamaki¨ P, (38). Higher fasting FDG uptake represents Akademi University and Turku University Hospital. et al. Human adipose tissue glucose uptake de- fl fl glucose utilization not only by adipocytes Duality of interest. No potential con icts of termined using [(18)F]- uoro-deoxy-glucose interest relevant to this article were reported. ([(18)F]FDG) and PET in combination with mi- fi but also by in ltrating macrophages and Author Contributions. P.D. analyzed fat dis- crodialysis. Diabetologia 2001;44:2171–2179 other cells in the stromovascular fraction tribution from MRI and performed FDG-PET 11. Schauer PR, Kashyap SR, Wolski K, et al. (39). Based on this assumption, inflamma- image analysis in abdominal, thoracic, and hu- Bariatric surgery versus intensive medical ther- tory signals from intraperitoneal fat were meral regions. L.L. analyzed FDG uptake in the apy in obese patients with diabetes. N Engl J femoral region. P.D. and L.L. wrote the manu- – higher in ND, whereas T2Ds had higher Med 2012;366:1567 1576 script. M.H. and P.S. were involved in the study 12. Immonen H, Hannukainen JC, Iozzo P, et al. inflammatory signals from abdominal SC design, patient recruitment, and surgical proce- Effect of bariatric surgery on liver glucose me- fat. In addition, T2D had enhanced inflam- dures. J.C.H., H.I., N.S., M.S., and R.P. were tabolism in morbidly obese diabetic and non- matory signals from skeletal muscle com- involved in data acquisition and contributed to diabetic patients. J Hepatol 2014;60:377–383 pared with ND, in line with reports of the discussion of the results. M-J.H. and M.B. 13. Helmio¨ M, Victorzon M, Ovaska J, et al. fl contributed to and made corrections to the SLEEVEPASS: a randomized prospective multi- increased levels of in ammation in skeletal manuscript. J.P. and P.I. contributed to and center study comparing laparoscopic sleeve muscle of T2D patients (40). edited the manuscript. E.F. compiled the results, gastrectomy and gastric bypass in the treatment Some strengths and limitations of this performed the statistical analysis and contrib- of morbid obesity: preliminary results. Surg En- study should be acknowledged. Strengths uted to and edited the manuscript. P.N. was dosc 2012;26:2521–2526 are that within the same study, we involved in the project planning, analysis, and 14. Expert Committee on the Diagnosis and writing of the manuscript. P.N. is the guarantor Classification of Diabetes Mellitus. Report of performed a hyperinsulinemic-euglycemic of this work and, as such, had full access to all the the expert committee on the diagnosis and clas- clamp to assess whole-body insulin sen- data in the study and takes responsibility for the sification of diabetes mellitus. Diabetes Care sitivity; MRI for absolute quantification integrity of the data and the accuracy of the data 2003;26(Suppl. 1):S5–S20 of fat in thoracic, humeral, abdominal, analysis. 15. DeFronzo RA, Tobin JD, Andres R. Glucose and femoral regions; and FDG PET to Prior Presentation. Parts of this study were clamp technique: a method for quantifying insulin presented in abstract form at the 50th Annual fi secretion and resistance. Am J Physiol 1979;237: quantify tissue-speci cGUinskeletal Meeting of the European Association for the E214–E223 muscle in the above-mentioned fat re- Study of Diabetes, Vienna, Austria, 15–19 Sep- 16. Hill DL, Maurer CR Jr, Studholme C, gions. Limitations of the current study tember 2014. Fitzpatrick JM, Hawkes DJ. Correcting scaling are the small sample size of 23 obese pa- errors in tomographic images using a nine de- tients. The patients and control groups References gree of freedom registration algorithm. J Com- put Assist Tomogr 1998;22:317–323 1. Bays HE, Gonzalez-Campoy´ JM, Bray GA, were predominantly comprised by fe- 17. Patlak CS, Blasberg RG. Graphical evalua- et al. Pathogenic potential of adipose tissue males, and as such fat distribution and tion of blood-to-brain transfer constants from and metabolic consequences of adipocyte hy- multiple-time uptake data. Generalizations. J metabolic activities may have been differ- pertrophy and increased visceral adiposity. Ex- Cereb Blood Flow Metab 1985;5:584–590 ent if equal number of both sexes were pert Rev Cardiovasc Ther 2008;6:343–368 studied. There is lack of data for the as- 2. Misra A, Garg A, Abate N, Peshock RM, Stray- 18. Peltoniemi P, Lonnroth¨ P, Laine H, et al. Lumped constant for [(18)F]fluorodeoxyglucose in sessment of GU at the molecular and cel- Gundersen J, Grundy SM. Relationship of ante- rior and posterior subcutaneous abdominal fat skeletal muscles of obese and nonobese humans. lular level. Bioelectric impedance analysis – to insulin sensitivity in nondiabetic men. Obes Am J Physiol Endocrinol Metab 2000;279:E1122 may not be an accurate measure of body Res 1997;5:93–99 E1130 fat content in morbidly obese subjects. 3. Smith SR, Lovejoy JC, Greenway F, et al. Con- 19. Ross R, Freeman J, Hudson R, Janssen I. The weight of the patient was not fully tributions of total body fat, abdominal subcutane- Abdominal obesity, muscle composition, and insulin resistance in premenopausal stabilized at 6 months of follow up. ous adipose tissue compartments, and visceral adipose tissue to the metabolic complications of women. J Clin Endocrinol Metab 2002;87: fi – In conclusion, our study con rms the obesity. Metabolism 2001;50:425–435 5044 5051 metabolic variations in different fat com- 4. Kelley DE, Thaete FL, Troost F, Huwe T, 20. He Q , Engelson ES, Kotler DP. A comparison partments in obese patients. Bariatric sur- Goodpaster BH. Subdivisions of subcutaneous ab- of abdominal subcutaneous adipose tissue pat- gery improves insulin-stimulated glucose dominal adipose tissue and insulin resistance. Am tern in obese and lean HIV-infected women. J – Nutr 2005;135:53–57 in skeletal muscle and in different fat J Physiol Endocrinol Metab 2000;278:E941 E948 5. Miyazaki Y, Glass L, Triplitt C, Wajcberg E, 21. Rutkowski JM, Stern JH, Scherer PE. The cell compartments of obese subjects inde- Mandarino LJ, DeFronzo RA. Abdominal fat dis- biology of fat expansion. J Cell Biol 2015;208: pendent of diabetes status. tribution and peripheral and hepatic insulin re- 501–512 sistance in type 2 diabetes mellitus. Am J Physiol 22. Gray DS, Fujioka K, Colletti PM, et al. Mag- Endocrinol Metab 2002;283:E1135–E1143 netic-resonance imaging used for determining 6. Lundbom J, Hakkarainen A, Lundbom N, fat distribution in obesity and diabetes. Am J Acknowledgments. The authors thank the Taskinen MR. Deep subcutaneous adipose tissue Clin Nutr 1991;54:623–627 authorities and the staff of Turku PET Centre is more saturated than superficial subcutaneous 23. Kim MK, Lee HC, Kwon HS, et al. Visceral and Turku University Hospital for performing adipose tissue. Int J Obes 2013;37:620–622 obesity is a negative predictor of remission of laboratory measurements, PET, computed to- 7. Giorgino F, Laviola L, Eriksson JW. Regional diabetes 1 year after bariatric surgery. Obesity mography,andMRIimaging.TheyalsothankKari differences of insulin action in adipose tissue: (Silver Spring) 2011;19:1835–1839 Kalliokoski, Turku PET Center, for invaluable insights from in vivo and in vitro studies. Acta 24. Snijder MB, van Dam RM, Visser M, Seidell advice on PET image analysis. Physiol Scand 2005;183:13–30 JC. What aspects of body fat are particularly Funding. This work was supported by grants 8. Abate N, Burns D, Peshock RM, Garg A, hazardous and how do we measure them? Int from the Finnish Diabetes Association, Finnish Grundy SM. Estimation of adipose tissue mass J Epidemiol 2006;35:83–92 Heart Foundation, Finnish Foundation for Car- by magnetic resonance imaging: validation 25. Kobayashi J, Maruyama T, Watanabe H, diovascular Research, and Albert Renold Fellow- against dissection in human cadavers. J Lipid et al. Gender differences in the effect of ship from European Foundation for the Study of Res 1994;35:1490–1496 type 2 diabetes on serum lipids, pre-heparin Diabetes as L.L.’s funding source. This study was 9. Bjorntorp¨ P. “Portal” adipose tissue as a gen- plasma lipoprotein lipase mass and other met- conducted at the Center of Excellence in Cardio- erator of risk factors for cardiovascular disease abolic parameters in Japanese population. Dia- vascular and Metabolic Diseases supported by and diabetes. Arteriosclerosis 1990;10:493–496 betes Res Clin Pract 2003;62:39–45 8 Bariatric Surgery, Fat Mass, and GU Diabetes Care

26. Abate N, Garg A, Peshock RM, Stray- adipose tissue FDG uptake by PET/CT in metabol- 36. McQuaid SE, Hodson L, Neville MJ, et al. Gundersen J, Adams-Huet B, Grundy SM. Rela- ically healthy obese subjects. Obesity (Silver Downregulation of adipose tissue fatty acid traf- tionship of generalized and regional adiposity to Spring) 2015;23:286–289 ficking in obesity: a driver for ectopic fat depo- insulin sensitivity in men with NIDDM. Diabetes 31. Virtanen KA, Iozzo P, Hallsten¨ K, et al. In- sition? Diabetes 2011;60:47–55 1996;45:1684–1693 creased fat mass compensates for insulin resis- 37. Hoffstedt J, Arner E, Wahrenberg H, et al. 27. Enevoldsen LH, Simonsen L, Stallknecht B, tance in abdominal obesity and type 2 diabetes: Regional impact of adipose tissue morphology Galbo H, Bulow¨ J. In vivo human lipolytic activity a positron-emitting tomography study. Diabe- on the metabolic profile in morbid obesity. Dia- in preperitoneal and subdivisions of subcutane- tes 2005;54:2720–2726 betologia 2010;53:2496–2503 ous abdominal adipose tissue. Am J Physiol En- 32. Boden G. Fatty acid-induced inflammation 38. Kang S, Kyung C, Park JS, et al. Subclinical docrinol Metab 2001;281:E1110–E1114 and insulin resistance in skeletal muscle and vascular inflammation in subjects with normal 28. Sadananthan SA, Prakash B, Leow MK, et al. liver. Curr Diab Rep 2006;6:177–181 weight obesity and its association with body fat: Automated segmentation of visceral and sub- 33. Ibrahim MM. Subcutaneous and visceral an 18 F-FDG-PET/CT study. Cardiovasc Diabetol. cutaneous (deep and superficial) adipose tis- adipose tissue: structural and functional differ- 2014;13:70 sues in normal and overweight men. J Magn ences. Obes Rev 2010;11:11–18 39. Torriani M, Zanni MV, Fitch K, et al. In- Reson Imaging 2015;41:924–934 34. Stolic M, Russell A, Hutley L, et al. Glucose creased FDG uptake in association with reduced 29. Marinou K, Hodson L, Vasan SK, et al. Struc- uptake and insulin action in human adipose tis- extremity fat in HIV patients. Antivir Ther 2013; tural and functional properties of deep abdominal sue–influence of BMI, anatomical depot and 18:243–248 subcutaneous adipose tissue explain its associa- body fat distribution. Int J Obes Relat Metab 40. Perseghin G, Petersen K, Shulman GI. tion with insulin resistance and cardiovascular risk Disord 2002;26:17–23 Cellular mechanism of insulin resistance: in men. Diabetes Care 2014;37:821–829 35. Tchernof A, Despres´ JP. Pathophysiology of potential links with inflammation. Int J 30. Oliveira AL, Azevedo DC, Bredella MA, human visceral obesity: an update. Physiol Rev Obes Relat Metab Disord 2003;27(Suppl. 3): Stanley TL, Torriani M. Visceral and subcutaneous 2013;93:359–404 S6–S11