Caloric Restriction and Weight Loss Are Primary Factors in the Early Tissue-Specific Metabolic Changes After Bariatric Surgery

Bariatric surgeries including Roux-en-Y gastric bypass (RYGB) and vertical sleeve gastrectomy (VSG) cause substantial and sustained weight loss in obese patients and are superior in the reversal of type 2 diabetes (T2D) compared with medical treatment (1). Improvements in T2D and insulin sensitivity are observed in the initial days and weeks after bariatric surgery (2,3) while weight loss is minimal (4), giving rise to the hypothesis that weight loss–independent mechanisms contribute to the metabolic improvements after bariatric surgery. A recent study contrasting outcomes from RYGB after 16 weeks and from caloric restriction at 23 weeks showed that after an equivalent 18% body weight reduction the rates of glucose disposal and metabolic variables were similarly improved (5). We tested the hypothesis that improvements in insulin resistance observed early after bariatric surgery are due to caloric restriction–dependent weight loss.

Participants in the RYGB and VSG groups were recruited from the Vanderbilt Center for Surgical Weight Loss and studied after informed, written consent was obtained. Participants in the diet group were recruited from the local community. Exclusion criteria included age <18 years or >65 years; smoking; pregnancy or breastfeeding; significant renal, hepatic, or cardiovascular disease; recent history of cancer; and previous malabsorptive or restrictive gastrointestinal surgery. This study was approved by the Vanderbilt Institutional Review Board and registered with ClinicalTrials.gov (NCT01474785).

Bariatric surgery participants completed study visits 2–3 before and ∼1–2 weeks after surgery. Participants in the diet group completed study visits before and 7–9 days after a post–bariatric surgery diet without bariatric surgery. Each study visit consisted of a hyperinsulinemic-euglycemic clamp, adipose tissue microdialysis, and a body composition measurement. Details of the study methods can be found in Supplementary Material. A subset of subjects underwent hepatic MRI (6). T2D was identified according to prior medical diagnosis or fasting plasma glucose ≥126 mg/dL during the baseline study. Subjects on oral antidiabetes medications discontinued usage 5 days prior to study.

Weight loss occurred in all three groups, with a significant difference among the groups. Mean ± SD weight loss for each group was 4.7 ± 1.4 kg for diet, 7.0 ± 3 kg for RYGB, and 7.2 ± 2.4 kg for VSG (P = 0.013). Subjects in the diet group lost significantly less weight compared with the surgery groups (P ≤ 0.049). Percent total body weight loss was 5.2 ± 1.5 for RYGB, 5.6 ± 1.7 for VSG, and 3.9 ± 0.9 for diet (P = 0.007) and only significantly different between the VSG and diet groups (P = 0.006). The postintervention study was 8 ± 1 days after diet, 10 ± 3 days after RYGB, and 11 ± 2 days after VSG (P ≥ 0.161); differences were related to scheduling.

Fasting plasma glucose, insulin, free fatty acid concentrations, and basal endogenous glucose production (EGP) were reduced significantly by equivalent amounts among the groups (Supplementary Table 1). Hepatic insulin sensitivity also improved after RYGB, VSG, and diet (P < 0.0001) with no differences among groups (P = 0.730) (Fig. 1A). EGP was on average suppressed by ∼80% during insulin infusion and improved slightly in all groups. Insulin-stimulated glucose disposal (M) did not significantly change over time or among groups. During the clamp procedure, mean ± SD plasma insulin concentrations were 193 ± 39 µU/mL at the baseline visit and significantly decreased to 170 ± 30 µU/mL at the postintervention study visit with no differences among the groups, corresponding with an increased clearance rate of insulin (MCR-I) (Supplementary Table 1). When M values were normalized to steady-state plasma insulin concentrations (M/I), values were significantly increased over time but not different among groups. Insulin-stimulated glucose uptake over basal was significantly increased over time without a group-by-time interaction (Fig. 1B). Adipose tissue insulin sensitivity also equally improved after each intervention (Fig. 1C).

In all three groups, plasma leptin levels significantly decreased and were associated with significant increases in IL-8 and TNFα over time (Supplementary Table 2). There were no significant changes after RYGB, VSG, or diet intervention in plasma concentrations of adiponectin, resistin, MCP-1, PAI-1, IL-6, or IL-1β over time. Plasma concentrations of acyl ghrelin decreased after RYGB and VSG but were not significantly different after 1 week of diet. Desacyl ghrelin concentrations only decreased after VSG. The acyl–to–desacyl ghrelin ratio decreased over time, with no differences among groups. The changes in adipose tissue interstitial fluid levels were more apparent, with demonstrated significant decreases over time in leptin, adiponectin, and PAI-1 that were not detected in the plasma (Supplementary Table 3).

A subset of subjects underwent abdominal MRI/MRS for quantification of hepatic fat (Supplementary Fig. 1A). In a combined analysis, mean ± SD hepatic fat content significantly decreased after surgical or dietary intervention by 19.0 ± 17.3%, from 12.9 ± 11.4% fat content at baseline (range 2.7–38.9) to 10.8 ± 10.2% after intervention (2.5–32.5) (P = 0.0071) (Supplementary Fig. 1B). At baseline, 9 of the 20 subjects had ≤5% hepatic fat; the decrease in steatosis remained significant after exclusion of these subjects (20.4 ± 10.3% vs. 17.19 ± 9.8%, P = 0.022). At baseline, percent hepatic fat positively correlated with fasting glucose and insulin, negatively correlated with hepatic insulin sensitivity and M, and did not correlate with weight, BMI, body fat, MCR-I, or EGP. The change in hepatic fat after intervention did not correlate with changes in any of these parameters.

Immediately after bariatric surgery, patients are instructed to consume a significantly restricted caloric diet within the range of 500–800 kcal/day. Numerous studies have implicated this reduction in caloric intake in the postoperative improvements in overall insulin sensitivity observed following bariatric procedures, especially RYGB (2,7–13). Except for one (8), findings of studies are in agreement that bariatric surgery is associated with decreased EGP and increased hepatic insulin sensitivity but variable improvements in peripheral insulin sensitivity (2,9,11,13). Our data support an improvement in insulin-stimulated glucose disposal at 1 week after RYGB and VSG, attributed to significant increases in insulin clearance rates. Effects of VSG on glucose disposal were less robust than those of RYGB and may be due to the greater preoperative adiposity in the VSG group or to body water/hydration status.

Improvements in hepatic insulin sensitivity after bariatric surgery are commonly reported to be secondary to reductions in hepatic fat. Previous studies demonstrated hepatic fat reductions were not observed at 4 weeks postoperative (14). Similarly, decrements in hepatic fat content (∼4%) were not observed 2 weeks after a low-calorie diet (800–1,100 kcal/day) (15). In this study, improvements in hepatic insulin sensitivity were similar among groups, with a 2% average reduction in hepatic fat. Our data suggest that early improvements (within 12 days) in hepatic insulin sensitivity after bariatric surgery or caloric restriction are due to factors beyond changes in hepatic fat, suggesting additional signaling pathways as drivers of metabolic improvement. We observed significant increases in plasma leptin and decreases in IL-8 and TNFα, while in adipose interstitial fluid levels of leptin, adiponectin, and PAI-1 were decreased. Whether changes in levels of interstitial fluid adipokines reflect altered function or are secondary to altered adipose tissue blood flow is unclear.

In summary, weight loss and the associated improvements in insulin sensitivity observed early after bariatric surgery can be replicated by equivalent, short-term caloric restriction. The observed metabolic improvements are characterized by overall improved hepatic and peripheral insulin sensitivities and consistent with a resolution of adipose tissue inflammation unique to bariatric procedures.

Acknowledgments. The authors thank VUMC investigators Kareem Jabbour, Emily E. Harmada, and Pam Marks-Shulman for acquisition of data and Bruce D. Gaylinn, University of Virginia, for technical support. The authors also thank the volunteers and patients who took part in these studies.

Funding. This research was supported by the National Institutes of Health: National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) grants DK091748 (to N.N.A.) and DK020593 (to the Vanderbilt Diabetes Research and Training Center), Vanderbilt Institute for Clinical and Translational Research grants UL1 TR000445 and TR002243 (Clinical and Translational Science Award [CTSA] program, CTSA grant), and NIDDK grant DK058404 (to the Vanderbilt Digestive Disease Research Center).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. R.A.T. and N.N.A. contributed to the study concept and design. C.R.F., R.A.T., J.A., R.M.S., B.W., M.D.S., W.J.E., E.B.W., and N.N.A. contributed to acquisition of data and technical support. All authors contributed to analysis and interpretation of data. C.R.F., R.A.T., and N.N.A. contributed to drafting of the manuscript. All authors contributed to critical revision of the manuscript. C.R.F., R.A.T., S.S., and N.N.A. contributed to data/statistical analysis. C.R.F., R.A.T., and N.N.A. are guarantors of data integrity and accuracy of data analysis.