Differences in Regional Brain Responses to Food Ingestion After Roux-en-Y Gastric Bypass and the Role of Gut Peptides: A Neuroimaging Study
OBJECTIVE Improved appetite control, possibly mediated by exaggerated gut peptide responses to eating, may contribute to weight loss after Roux-en-Y gastric bypass (RYGB). This study compared brain responses to food ingestion between post-RYGB (RYGB), normal weight (NW), and obese (Ob) unoperated subjects and explored the role of gut peptide responses in RYGB.
RESEARCH DESIGN AND METHODS Neuroimaging with [18F]-fluorodeoxyglucose (FDG) positron emission tomography was performed in 12 NW, 21 Ob, and 9 RYGB (18 ± 13 months postsurgery) subjects after an overnight fast, once FED (400 kcal mixed meal), and once FASTED, in random order. RYGB subjects repeated the studies with somatostatin infusion and basal insulin replacement. Fullness, sickness, and postscan ad libitum meal consumption were measured. Regional brain FDG uptake was compared using statistical parametric mapping.
RESULTS RYGB subjects had higher overall fullness and food-induced sickness and lower ad libitum consumption. Brain responses to eating differed in the hypothalamus and pituitary (exaggerated activation in RYGB), left medial orbital cortex (OC) (activation in RYGB, deactivation in NW), right dorsolateral frontal cortex (deactivation in RYGB and NW, absent in Ob), and regions mapping to the default mode network (exaggerated deactivation in RYGB). Somatostatin in RYGB reduced postprandial gut peptide responses, sickness, and medial OC activation.
CONCLUSIONS RYGB induces weight loss by augmenting normal brain responses to eating in energy balance regions, restoring lost inhibitory control, and altering hedonic responses. Altered postprandial gut peptide responses primarily mediate changes in food-induced sickness and OC responses, likely to associate with food avoidance.
Roux-en-Y gastric bypass (RYGB) causes weight loss, improves metabolic parameters, and reduces premature mortality (1), with little evidence of restriction of meal size or malabsorption (2). Sensations of fullness are increased and food consumption reduced (3,4). Understanding changes in gut-to-brain signaling and brain function mediating these effects would improve understanding of weight control and weight loss and may help develop novel approaches to preventing and treating obesity.
Altered gut-to-brain signaling after RYGB may be mediated by gut peptides (2). Glucagon-like peptide 1 (GLP-1) and peptide YY (PYY) provide satiety signals. RYGB increases their postprandial responses (2,5). The somatostatin analog octreotide inhibits gut peptide secretion, increases food intake, reduces satiety, and alters appetitive behavior after RYGB (6,7).
In functional neuroimaging, surrogates are used to image regional brain activity. In [18F]-fluorodeoxyglucose positron emission tomography (FDG-PET), 18FDG is taken up as native glucose and phosphorylated but not metabolized further, accumulating within cells (8). Brain FDG uptake correlates with brain glucose metabolic rate, a marker for brain activity. Comparing FDG-PET images can therefore identify regions of altered neuronal activation without preconceived hypotheses. FDG-PET is suited to imaging responses to slowly changing physiological stimuli, such as eating, but has not previously been used for this. Other functional neuroimaging modalities have been used to investigate bariatric surgery (mainly RYGB). Two dopamine receptor radioligand PET studies, imaging only pathways using the ligand receptor, gave conflicting results (9,10). Functional MRI (fMRI) studies of responses to food cues (rather than food ingestion) have described reduced responses in regions including the dorsal striatum (reward), dorsolateral frontal cortex (DLFC) (inhibitory control), precuneus, and posterior cingulate in the fed state (11,12); and, examining predefined regions of interest in the fasted state, in the orbitofrontal cortex (OFC) and amygdala, with evidence that octreotide increases responses to food pictures in the fed state without affecting fullness (7,13). One fMRI study reported differences in response to oral glucose between lean and obese subjects in the hypothalamus, OFC, and somatosensory cortex were partially reversed after RYGB (4).
The aim of our study was to use FDG-PET neuroimaging to identify regions where brain responses to food ingestion were different between post-RYGB (RYGB), normal weight (NW), and obese (Ob) unoperated subjects and investigate the effect of using somatostatin to inhibit gut peptide responses in RYGB.
Research Design and Methods
This research was approved by The Royal Marsden Research Ethics Committee (08/H0801/152) and the Administration of Radioactive Substances Advisory Committee (261-1945) and conducted in accordance with the Declaration of Helsinki (2008).
Participants and Recruitment
Right-handed adults were recruited from obesity and bariatric surgery clinics at King’s College Hospital and by e-mail advertisement to students and staff at King’s College London in three groups: NW (BMI 20–25 kg/m2), Ob (BMI 30–40 kg/m2), and RYGB (≥3 months after RYGB, ≥10% excess weight loss, current BMI 25–40 kg/m2). Exclusion criteria included contraindications to PET or MRI; pregnancy, planning pregnancy, breastfeeding; glucose >15 mmol/L during 75 g oral glucose tolerance test (NW and Ob) or >11 mmol/L after 400 kcal test meal (RYGB); glucose-lowering medications (metformin permitted); significant brain disorder; use of psychotropic medication.
NW and Ob underwent five visits: screening, 75 g oral glucose tolerance test, two PET scanning visits (FASTED and FED in random order), and a structural MRI brain scan (Philips Achieva 3.0 T scanner). RYGB underwent seven visits: screening, 400 kcal test meal (to determine capacity to consume the meal and glucose response), four PET scanning visits (placebo-FASTED, placebo-FED, somatostatin-FASTED, and somatostatin-FED in random order) and a MRI brain scan. Subjects underwent a dummy PET scan to diminish the effect of the first study (14).
FDG-PET visits were performed after overnight fasting (>9 h), with water allowed. Premenopausal women were scanned in the first 10 days of their cycle. Arm intravenous catheters were sited. At RYGB-somatostatin (RYGBss) visits, intravenous infusions of somatostatin (Actavis or Eumedica, 0.1 μg/kg/min) (15,16) and soluble human insulin (Actrapid, in 4% autologous blood solution, 3.6 mU/m2body surface area/min; Novo Nordisk) were started at −95 min and continued throughout. At RYGB-placebo (RYGBpl) visits, 0.9% saline was infused. Participants were blind to infusion content. If nausea developed, the somatostatin infusion was reduced to 70%. If venous plasma glucose (VPG) fell below 3.8 mmol/L, 20% glucose was infused to maintain at 4–4.5 mmol/L.
For FED studies, subjects consumed a 400 kcal meal (Häagen-Dazs vanilla ice cream, fat 27 g, carbohydrate 32 g, protein 8 g) starting at −5 min (NW and Ob after 20-min rest, RYGB after 90-min infusion). Three RYGB subjects unable to consume the 400 kcal test meal were given the amount tolerated (220–256 kcal). FDG (90 MBq) was injected intravenously 15 min after meal completion in FED studies or equivalent time in FASTED. Scanning commenced at +55 min for 15 min (GE Discovery PET scanner, 15.8-cm axial field of view; GE Medical System). A low-dose computed tomography brain scan was taken for attenuation correction.
After each PET scan, subjects underwent a 1-h ad libitum meal (6) in which 100 kcal ice cream was presented every 5 min, and subjects were instructed to eat until they felt full. Subjects rated fullness and sickness on visual analog scales (VAS) at −105 min (RYGB only), −7 min, +10 min, and +80 min (6). Venous blood was taken for insulin, GLP-1, PYY, glucose-dependent insulinotropic polypeptide (GIP), and glucagon at −100 min (RYGB), −10 min, +30 min, and +80 min and for glucose every 5 to 15 min. Plasma glucose was analyzed immediately (YSI 2300 Stat analyzer). Serum insulin was measured by chemiluminometric immunoassay (Advia Centaur; Seimens) and gut peptides GLP-1 and GIP by ELISA (Millipore) and PYY and glucagon by radioimmunoassay (Millipore).
Statistical analyses used SPSS 22 software (IBM). P ≤ 0.05 was considered significant. Uncorrected P values are reported. Continuous demographic data were compared using one-way ANOVA with post hoc comparisons, and categorical data were compared using the Fisher exact test. Mixed ANOVA was used for analysis of VAS fullness, ad libitum consumption, glucose, insulin, and gut peptide data. For significant interactions between fed state and group, post hoc comparisons for differences between groups in “FED effect” (FED minus FASTED) were performed using the Fisher least significant difference test. If there was no interaction, main effects of fed state and group are reported. For the effect of somatostatin in RYGB, within–within-subjects ANOVA was used. Kruskal-Wallis and Wilcoxon signed-rank tests were used to compare nonnormally distributed sickness VAS.
FDG-PET Neuroimaging Analysis
Differences in FDG uptake between scans were analyzed using Statistical Parametric Mapping (SPM8) (www.fil.ion.ucl.ac.uk). Images were reconstructed using the filtered back-projection algorithm. Images were acquired dynamically (15 × 1 min frames), and frames showing motion were removed. Images were spatially normalized to Montreal Neurological Institute space using each subject’s structural MRI. MRI was not available in one NW and one Ob subject, and mean PET images were warped directly to Montreal Neurological Institute space using the SPM PET template. Images were smoothed with a Gaussian kernel of 8 mm. The cerebellum (Tziortzi atlas ) was excluded from further analysis. Global differences in FDG uptake between scans were removed by normalizing voxel values to the mean gray-matter value for each scan, scaled to 100. White matter was masked out by including voxels with values >60% mean and gray-matter probability >30%. The pituitary (defined by MRI template) and hypothalamus (Baroncini et al. ) were masked in.
Images were compared to identify clusters with significant differences using mixed ANOVA in SPM. For paired tests (effect of fed state), clusters of voxels were considered to show significant effect at voxel level P < 0.001 and cluster level P < 0.05 (corrected for family-wise error). For interactions between fed state and group, clusters were considered to show significant interaction with cluster size >100 voxels and two voxel level thresholds: P < 0.01 (liberal) and P < 0.001 (stringent). Clusters were localized using the Tziortzi atlas (17) modified to include the pituitary and hypothalamus.
For clusters identified in SPM with a significant interaction between fed state and group, mean normalized voxel values were extracted for each cluster for all scans and food-evoked signal change (FESC) (FED minus FASTED) was calculated. For each cluster, the nature of the difference in FESC between groups was analyzed using the Fisher least significant difference test in SPSS and the effects of somatostatin in RYGB using paired t tests. Exploratory Spearman correlational analyses were performed between FESC and ad libitum meal consumption at the FASTED visit (i.e., in the fasted state), FED effect +10 min fullness and sickness, and FED effect +30 min insulin and gut peptides.
The study included 12 NW, 21 Ob, and 9 RYGB subjects, with a mean ± SD age of 34.4 ± 11.5 years (RYGB subjects were older), with no significant between-group differences in sex, ethnicity, or systolic blood pressure (Table 1). RYGB subjects were 18 ± 12.6 months after RYGB, having lost 30.9 ± 8.5% of their preoperative weight. BMI was not different between Ob and RYGB. HOMA2-insulin resistance, reflecting fasting insulin resistance (19), was not different between NW and RYGB but was higher in Ob. Medications included metformin (one Ob, one RYGB), orlistat (two Ob, one RYGB), and topiramate (one RYGB).
VAS for Fullness and Sickness
At −7 min, VAS scores for fullness were higher in RYGBpl than in NW or Ob, which were not different (Fig. 1). Across groups, fullness was higher in FED versus FASTED at +10 and +80 min. The numerically greater FED effect on fullness at +10 min in RYGBpl did not reach significance (P = 0.14), although fullness scores at +10 min were higher, irrespective of fed state, in RYGBpl compared with NW and Ob. Somatostatin had no significant effect on fullness at −7 min or on responses to food ingestion at +10 or +80 min.
VAS scores for sickness were higher in NW versus RYGBpl at −7 min. At +10 min, sickness was higher in FED versus FASTED in RYGBpl but not in NW or Ob. Somatostatin was reduced in two RYGB subjects due to nausea. Despite this, sickness scores were higher with somatostatin at −7 min. However, somatostatin attenuated the increase in sickness at +10 min in FED versus FASTED (median FED effect RYGBpl +26 points, RYGBss +5 points; P = 0.05).
Ad Libitum Meal
Subjects consumed less at FED versus FASTED, with no significant between-group differences in the effect of the fed state on the amount consumed (Fig. 1). Regardless of fed state, RYGBpl consumed less than NW and Ob. Ad libitum consumption in RYGBpl was (mean ± SE) 272 ± 38 kcal vs. 371 ± 99 kcal in RYGBss (P = 0.27 for main effect of somatostatin).
Glucose, Insulin, and Gut Peptides
Mean VPG between 0 to +80 min was higher in FED versus FASTED, with no between-group differences (P = 0.214 for interaction, P < 0.001 for main effect fed state, and P = 0.166 for main effect group): (mean ± SD) NW-FASTED, 4.9 ± 0.4 mmol/L; NW-FED, 5.1 ± 0.5 mmol/L; Ob-FASTED, 5.0 ± 0.3 mmol/L; Ob-FED, 5.5 ± 0.6 mmol/L; RYGBpl-FASTED 4.7 ± 0.5 mmol/L; RYGBpl-FED, 5.2 ± 0.9 mmol/L; RYGBss-FASTED, 4.3 ± 0.3 mmol/L; and RYGBss-FED, 5.4 ± 0.7 mmol/L. There was no difference between somatostatin and insulin versus placebo on the FED effect (P = 0.152 for interaction) or VPG irrespective of fed state (P = 0.675 for main effect). The highest VPG was 7.9 mmol/L.
Insulin was higher at −10 min in Ob versus NW or RYGBpl, and at +30 min in FED versus FASTED across all groups (Fig. 2). Insulin was not different at +80 min between NW and RYGBpl, but remained higher in Ob, with a greater FED effect and higher concentrations overall.
There were no between-group differences in GLP-1, PYY, GIP (Fig. 2) or glucagon (not shown) at −10 min. Between-group differences were found in FED effect on GLP-1 and PYY at +30 and +80 min, which were larger in RYGBpl than in NW or Ob. GLP-1 was higher in FED versus FASTED at +30 and +80 min in all groups (not significant in NW +30 min). The FED effect on PYY was small in NW and Ob and statistically significant only in NW at +80 min. GIP was higher in FED versus FASTED at +30 min, with no between-group differences, and remained higher in all groups at +80 min, with a smaller FED effect in RYGBpl. Glucagon was higher in FED versus FASTED at +30 min, with no between-group differences in the FED effect and no between-state or between-group differences by +80 min.
Somatostatin with insulin in RYGB achieved no significant difference in insulin between placebo and somatostatin at −10 min (85 min into infusions). GLP-1, PYY, GIP, and glucagon were lower with somatostatin at −10 min. Somatostatin abolished insulin, GLP-1, PYY, GIP, and glucagon responses to food ingestion.
SPM analysis for the main effect of the fed state across groups showed a single large cluster (K) where FDG uptake was higher in FED versus FASTED and five clusters (L-P) where FDG uptake was lower in FED versus FASTED (Fig. 3).
Cluster K (17,485 voxels) included the hypothalamus, ventral cingulate subcallosal gyrus, anterior cingulate gyrus, bilateral ventral striatum, globus pallidus, temporal thalamus, insular cortex, orbital cortex (OC), extensive regions in the temporal lobes (including amygdala and hippocampus), and midbrain, pons, and medulla (Supplementary Table 1).
Two FED<FASTED clusters (L, 5,079 voxels; M, 4,571 voxels) included bilateral anterior and posterior DLFC, extending into bilateral precentral gyrus, bilateral frontal operculum, and right lateral OC. Cluster N (388 voxels) included anterior cingulate gyrus and dorsal anterior cingulate gyrus. Clusters O (19,248 voxels) and P (200 voxels) included posterior cingulate gyrus, bilateral precuneus, and cuneus, extending posteriorly to include bilateral calcarine cortex, lingual gyrus, occipital pole, and occipital fusiform gyrus and laterally to include bilateral parietal lobule, angular gyrus, supramarginal gyrus, parietal operculum, central operculum (right), and posterior temporal cortex (Supplementary Table 2).
SPM analysis for interaction between fed state and group showed 10 clusters (A–J, voxel level P < 0.01, cluster size threshold 100 voxels) (Supplementary Fig. 3 and Supplementary Table 4). A more stringent statistical threshold (voxel level P < 0.001) showed three clusters corresponding to clusters C, F, and G (data not shown). Including age as a covariate did not materially affect the interaction clusters identified (data not shown). For clusters A–J, FESC in NW, Ob, and RYGBpl are shown in Fig. 4 (representative clusters) and in Supplementary Fig. 5 (numerical data are reported in Supplementary Table 4). In E (hypothalamus) and F (pituitary), FESC was larger in RYGBpl than in NW or Ob, with no difference between NW and Ob. In A, C, and D (right DLFC, anterior medial frontal cortex, medial and lateral OC, frontal operculum, and insular cortex), there was a similar negative FESC in NW and RYGBpl, absent in Ob. In B (left medial OC), there was a negative FESC in NW, with small positive FESC in Ob and a larger positive FESC in RYGBpl. In G, H, and I (posterior cingulate gyrus, bilateral precuneus, angular gyrus, occipital pole, right cuneus, posterior superior and middle temporal gyri, and left parietal lobule), there was a larger negative FESC in RYGBpl than in both NW and Ob, with no difference between NW and Ob (G and I) or larger negative FESC in NW than in Ob (H). In cluster J (lingual gyrus), FESC was positive in NW and negative in Ob and RYGBpl.
For each cluster where there was a significant difference in responses to food ingestion among the three groups, FESC was calculated for RYGBss and compared with RYGBpl (Fig. 4 and Supplementary Fig. 5). In cluster B (left medial OC), somatostatin abolished the positive FESC seen in RYGBpl. Somatostatin had no effect on FESC elsewhere.
Exploratory Correlational Analyses
Exploratory correlational analyses were performed for clusters A–J (Supplementary Tables 6–8). These analyses should be viewed with caution. Analyses including all subjects were required to achieve sufficient power to detect even strong correlations; therefore, weaker correlations may have been missed, and there is a potential effect of group separation. Positive correlations were found between FESC and ad libitum consumption in the fasted state in NW in cluster C (rs = 0.910, P < 0.001), and across all subjects in clusters A (rs = 0.378, P = 0.014), G (rs = 0.461, P = 0.002), H (rs = 0.640, P < 0.001), and I (rs = 0.662, P < 0.001), with greater deactivation associated with lower ad libitum consumption. Negative correlations were found across all subjects in clusters E (rs = −0.596, P < 0.001) and F (rs = −0.539, P < 0.001), with greater activation associated with lower ad libitum consumption.
VAS scores showed significant negative correlations between FESC and FED effect fullness in cluster I (rs = −0.331, P = 0.034) and FED effect sickness in cluster G (rs = −0.321, P = 0.041).
In cluster B, there were significant positive correlations between FESC and FED effect insulin (rs = 0.472, P = 0.015), GLP-1 (rs = 0.708, P < 0.001), and PYY (rs = 0.468, P = 0.018). In cluster E, there was a significant positive correlation between FESC and FED effect GLP-1 (rs = 0.632, P = 0.001) and in cluster F between FESC and FED effect GLP-1 (rs = 0.709, P < 0.001) and PYY (rs = 0.562, P = 0.003). Significant negative correlations between FESC and FED effect GLP-1 were found in clusters G (rs = −0.725, P < 0.001), H (rs = −0.729, P < 0.001), and I (rs = −0.662, P < 0.001).
Using imaging techniques that identify differences in regional brain activation independent of prior hypotheses and a protocol that examines the response to food ingestion per se, we have shown an effect of RYGB on the response to ingestion of a 400 kcal meal in brain regions subserving signal processing relating to food ingestion and energy balance (hypothalamus), reward evaluation (medial OC), inhibitory control (DLFC), and the default mode network (DMN). Our control subjects were NW and Ob adults, the latter matched for BMI to RYGB subjects, allowing us to investigate effects of obesity.
The RYGB group showed normal fasting insulin sensitivity. The meal size, chosen to be tolerable after RYGB, was sufficient in all groups to affect fullness, eating behavior, and gut peptides for at least the duration of the FDG-PET scanning. The physiological data from our RYGB group are consistent with the literature, with higher fullness, reduced ad libitum consumption (3), and exaggerated postprandial GLP-1 and PYY responses (5,20).
Regional brain responses to food ingestion were extensive in regions known to be involved in central regulation of food intake (21). FDG uptake increased, reflecting activation, in the hypothalamus and brainstem (signal processing relating to food ingestion and energy balance); insula (interoception); ventral striatum, globus pallidus, OC, ventral cingulate subcallosal gyrus, anterior cingulate gyrus, amygdala, and hippocampus (reward); and in the temporal lobes. FDG uptake decreased, reflecting deactivation, in bilateral DLFC (inhibitory control), anterior cingulate gyrus, and a large posterior cluster. The consistency of these regions with those identified in previous functional neuroimaging studies in NW individuals investigating response to food ingestion (22–25) or effect of food ingestion on the response to food cues (26–29) demonstrates the utility of FDG-PET in imaging responses to food ingestion. The activation in the temporal lobes and deactivation in the posterior cluster were unexpected. The latter is compatible with the structures of the DMN (30).
Hypothalamic and pituitary activation to food ingestion was exaggerated after RYGB, suggesting that food ingestion represents a greater physiological stimulus after RYGB. The exploratory analyses suggest correlation between activation in these regions and limitation of food intake. The only other study looking at brain responses to nutrient ingestion after RYGB showed partial restoration of hypothalamic responses from the obese toward lean (4). Discrepancies in difference patterns may relate to our use of a mixed meal versus a pure glucose stimulus.
Our RYGB subjects showed activation in the left medial OC versus deactivation in NW control subjects. The above study also showed differences in OC after RYGB (4). The medial OC is involved in evaluation of reward, and the pattern in our study is consistent with pleasant sensation in response to eating in NW subjects versus unpleasant sensation after RYGB.
Our data showed deactivation in right DLFC in NW and RYGB subjects, absent from the unoperated Ob subjects. The correlational data in our NW subjects supports an association between DLFC deactivation and inhibitory control of food intake. Attenuated DLFC responses to food ingestion in Ob versus lean subjects in [15O]-water PET studies were interpreted as loss of inhibitory control (31). Changes in response to food cues have been described in DLFC after RYGB (11,12) but not in response to nutrient ingestion. Our data are consistent with loss of the normal “stop eating” signal in obesity and/or insulin resistance, restored after RYGB and with clinical observations of restored inhibitory control after RYGB, suggesting altered DLFC activity after RYGB may contribute to weight loss.
Our data showed exaggerated deactivation in regions mapping to the DMN in RYGB. These changes are consistent with the response to eating being a greater brain “task” after RYGB. Alternatively, RYGB may reduce effects of previous eating experience. The exploratory analyses suggest correlation between deactivation in these regions and limitation of food intake. There is some evidence for an effect of RYGB on DMN regions (4,11,12).
Apart from the OC, our data showed loss of normal responses in obesity, restored after RYGB, or exaggeration of normal responses after RYGB. This is consistent with data showing fMRI responses to oral glucose revert toward normal after RYGB (4). Studies looking at food cues generally find greater responsivity to food cues in the obese attenuated after RYGB (11–13,32–34). Discrepancies may relate to differences between food cues and food ingestion or represent activation/deactivation of stimulatory/inhibitory pathways in functionally similar brain regions.
Somatostatin in RYGB suppressed basal (fasting) GLP-1, PYY, GIP, and glucagon and abolished postprandial peptide responses, and concomitant insulin infusion successfully replaced basal insulin. Although postprandial sickness was attenuated by somatostatin, fullness and ad libitum consumption did not change significantly. Previous studies found octreotide (without basal insulin replacement) reduced the effect of food ingestion on fullness and increased ad libitum consumption after RYGB (6,35). There may be differences in the effects of octreotide and somatostatin. Octreotide may have a direct action on the brain (36). However, Goldstone et al. (7) found no direct effect of octreotide on food reward behaviors in NW subjects or, when given with insulin, on postprandial hunger or fullness after RYGB. Insulin reduces food intake (37–39), albeit at higher concentrations, and regional differences in brain FDG uptake were found in reward regions with somatostatin with versus without low-dose insulin infusion (15), suggesting absence of basal insulin may be important.
Of the 10 clusters where differences between groups in brain response to food ingestion were identified, somatostatin only impacted in cluster B, the left medial OC, where it abolished the activation seen after RYGB (not seen in NW or Ob subjects). In exploratory correlational analyses, this was the only cluster where activation correlated with increase in insulin, GLP-1, and PYY. These data suggest gut peptides may mediate the altered OC response after RYGB but are not key mediators of the differences seen in the hypothalamus, DLFC, or DMN. Postprandial sickness in RYGB subjects, also attenuated with somatostatin, may be part of a food-avoidance response to calorie-dense meals mediated by exaggerated gut peptide responses.
In summary, the RYGB group studied here showed expected increased fullness, reduced food consumption, and exaggerated postprandial GLP-1 and PYY responses. Differences in brain responses to food ingestion were exaggeration of normal hypothalamic activation, consistent with food ingestion being a greater physiological stimulus; reversed responses in the left medial OC, consistent with unpleasant, rather than pleasant, sensation; restoration of normal responses in inhibitory control regions, lost in obesity; and exaggerated deactivation in DMN, consistent with food ingestion being a greater task. These changes in brain responses would be expected to contribute to weight loss. The somatostatin data suggest exaggerated gut peptide responses after RYGB mediate changes in medial OC activity and in postprandial nausea but may not be the major mediator of increased fullness and reduced food ingestion and do not mediate the other differences in brain responses to food ingestion after RYGB.
Acknowledgments. The authors thank the participants; research nurses Andrew Pernet and Bula Wilson; Tracy Dew; the laboratory staff at Viapath, King’s College Hospital, London; the radiographers and administrative staff at the PET Imaging Centre; and the staff of the Clinical Research Facility, St Thomas’ Hospital, London.
Funding. This study was funded by a project grant from The Diabetes Foundation UK.
Duality of Interest. No potential conflicts of interest relevant to this article were reported.
Author Contributions. K.F.H. and J.T.D. collected data. K.F.H., J.T.D., C.W.l.R., P.K.M., and S.A.A. wrote the manuscript. K.F.H., J.T.D., and L.J.R. analyzed the data. K.F.H. and A.G.P. recruited participants. All authors contributed to conception and design of the research and to interpreting the data. S.A.A. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Prior Presentation. Parts of these data were presented at the 3rd World Congress on Interventional Therapies for Type 2 Diabetes, London, U.K., 28–30 September 2015. The results of the new correlational analyses were presented in abstract form at the 52nd European Association for the Study of Diabetes Annual Meeting, Munich, Germany, 12–16 September 2016.
This article contains Supplementary Data online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc15-2721/-/DC1.
- Received December 16, 2015.
- Accepted June 8, 2016.
- © 2016 by the American Diabetes Association.
Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at http://www.diabetesjournals.org/content/license.