Advertisement
Circulating glucose levels modulate neural control of desire for high-calorie foods in humans
Kathleen A. Page,1,2 Dongju Seo,3 Renata Belfort-DeAguiar,1 Cheryl Lacadie,4 James Dzuira,5 Sarita Naik,1 Suma Amarnath,1 R. Todd Constable,4 Robert S. Sherwin,1 and Rajita Sinha3
1Section of Endocrinology, Yale University School of Medicine, New Haven, Connecticut, USA. 2Division of Endocrinology, University of Southern California Keck School of Medicine, Los Angeles, California, USA. 3Department of Psychiatry, 4Department of Radiology, and 5Yale Center for Clinical Investigation, Yale University School of Medicine, New Haven, Connecticut, USA.
Address correspondence to: Robert S. Sherwin, C.N.H. Long Professor of Medicine, Director, Yale Center for Clinical Investigation, Yale School of Medicine, The Anlyan Center S141, PO Box 208020 New Haven, Connecticut 06520-8020, USA. Phone: 203.785.4183; Fax: 203.737.5558; E-mail: robert.sherwin@yale.edu.
Authorship note: Kathleen A. Page and Dongju Seo contributed equally to this work and serve as joint first authors. Rajita Sinha and Robert S. Sherwin contributed equally and serve as joint senior authors.
Find articles by Page, K. in: JCI | PubMed | Google Scholar
1Section of Endocrinology, Yale University School of Medicine, New Haven, Connecticut, USA. 2Division of Endocrinology, University of Southern California Keck School of Medicine, Los Angeles, California, USA. 3Department of Psychiatry, 4Department of Radiology, and 5Yale Center for Clinical Investigation, Yale University School of Medicine, New Haven, Connecticut, USA.
Address correspondence to: Robert S. Sherwin, C.N.H. Long Professor of Medicine, Director, Yale Center for Clinical Investigation, Yale School of Medicine, The Anlyan Center S141, PO Box 208020 New Haven, Connecticut 06520-8020, USA. Phone: 203.785.4183; Fax: 203.737.5558; E-mail: robert.sherwin@yale.edu.
Authorship note: Kathleen A. Page and Dongju Seo contributed equally to this work and serve as joint first authors. Rajita Sinha and Robert S. Sherwin contributed equally and serve as joint senior authors.
Find articles by Seo, D. in: JCI | PubMed | Google Scholar
1Section of Endocrinology, Yale University School of Medicine, New Haven, Connecticut, USA. 2Division of Endocrinology, University of Southern California Keck School of Medicine, Los Angeles, California, USA. 3Department of Psychiatry, 4Department of Radiology, and 5Yale Center for Clinical Investigation, Yale University School of Medicine, New Haven, Connecticut, USA.
Address correspondence to: Robert S. Sherwin, C.N.H. Long Professor of Medicine, Director, Yale Center for Clinical Investigation, Yale School of Medicine, The Anlyan Center S141, PO Box 208020 New Haven, Connecticut 06520-8020, USA. Phone: 203.785.4183; Fax: 203.737.5558; E-mail: robert.sherwin@yale.edu.
Authorship note: Kathleen A. Page and Dongju Seo contributed equally to this work and serve as joint first authors. Rajita Sinha and Robert S. Sherwin contributed equally and serve as joint senior authors.
Find articles by Belfort-DeAguiar, R. in: JCI | PubMed | Google Scholar
1Section of Endocrinology, Yale University School of Medicine, New Haven, Connecticut, USA. 2Division of Endocrinology, University of Southern California Keck School of Medicine, Los Angeles, California, USA. 3Department of Psychiatry, 4Department of Radiology, and 5Yale Center for Clinical Investigation, Yale University School of Medicine, New Haven, Connecticut, USA.
Address correspondence to: Robert S. Sherwin, C.N.H. Long Professor of Medicine, Director, Yale Center for Clinical Investigation, Yale School of Medicine, The Anlyan Center S141, PO Box 208020 New Haven, Connecticut 06520-8020, USA. Phone: 203.785.4183; Fax: 203.737.5558; E-mail: robert.sherwin@yale.edu.
Authorship note: Kathleen A. Page and Dongju Seo contributed equally to this work and serve as joint first authors. Rajita Sinha and Robert S. Sherwin contributed equally and serve as joint senior authors.
Find articles by Lacadie, C. in: JCI | PubMed | Google Scholar
1Section of Endocrinology, Yale University School of Medicine, New Haven, Connecticut, USA. 2Division of Endocrinology, University of Southern California Keck School of Medicine, Los Angeles, California, USA. 3Department of Psychiatry, 4Department of Radiology, and 5Yale Center for Clinical Investigation, Yale University School of Medicine, New Haven, Connecticut, USA.
Address correspondence to: Robert S. Sherwin, C.N.H. Long Professor of Medicine, Director, Yale Center for Clinical Investigation, Yale School of Medicine, The Anlyan Center S141, PO Box 208020 New Haven, Connecticut 06520-8020, USA. Phone: 203.785.4183; Fax: 203.737.5558; E-mail: robert.sherwin@yale.edu.
Authorship note: Kathleen A. Page and Dongju Seo contributed equally to this work and serve as joint first authors. Rajita Sinha and Robert S. Sherwin contributed equally and serve as joint senior authors.
Find articles by Dzuira, J. in: JCI | PubMed | Google Scholar
1Section of Endocrinology, Yale University School of Medicine, New Haven, Connecticut, USA. 2Division of Endocrinology, University of Southern California Keck School of Medicine, Los Angeles, California, USA. 3Department of Psychiatry, 4Department of Radiology, and 5Yale Center for Clinical Investigation, Yale University School of Medicine, New Haven, Connecticut, USA.
Address correspondence to: Robert S. Sherwin, C.N.H. Long Professor of Medicine, Director, Yale Center for Clinical Investigation, Yale School of Medicine, The Anlyan Center S141, PO Box 208020 New Haven, Connecticut 06520-8020, USA. Phone: 203.785.4183; Fax: 203.737.5558; E-mail: robert.sherwin@yale.edu.
Authorship note: Kathleen A. Page and Dongju Seo contributed equally to this work and serve as joint first authors. Rajita Sinha and Robert S. Sherwin contributed equally and serve as joint senior authors.
Find articles by Naik, S. in: JCI | PubMed | Google Scholar
1Section of Endocrinology, Yale University School of Medicine, New Haven, Connecticut, USA. 2Division of Endocrinology, University of Southern California Keck School of Medicine, Los Angeles, California, USA. 3Department of Psychiatry, 4Department of Radiology, and 5Yale Center for Clinical Investigation, Yale University School of Medicine, New Haven, Connecticut, USA.
Address correspondence to: Robert S. Sherwin, C.N.H. Long Professor of Medicine, Director, Yale Center for Clinical Investigation, Yale School of Medicine, The Anlyan Center S141, PO Box 208020 New Haven, Connecticut 06520-8020, USA. Phone: 203.785.4183; Fax: 203.737.5558; E-mail: robert.sherwin@yale.edu.
Authorship note: Kathleen A. Page and Dongju Seo contributed equally to this work and serve as joint first authors. Rajita Sinha and Robert S. Sherwin contributed equally and serve as joint senior authors.
Find articles by Amarnath, S. in: JCI | PubMed | Google Scholar
1Section of Endocrinology, Yale University School of Medicine, New Haven, Connecticut, USA. 2Division of Endocrinology, University of Southern California Keck School of Medicine, Los Angeles, California, USA. 3Department of Psychiatry, 4Department of Radiology, and 5Yale Center for Clinical Investigation, Yale University School of Medicine, New Haven, Connecticut, USA.
Address correspondence to: Robert S. Sherwin, C.N.H. Long Professor of Medicine, Director, Yale Center for Clinical Investigation, Yale School of Medicine, The Anlyan Center S141, PO Box 208020 New Haven, Connecticut 06520-8020, USA. Phone: 203.785.4183; Fax: 203.737.5558; E-mail: robert.sherwin@yale.edu.
Authorship note: Kathleen A. Page and Dongju Seo contributed equally to this work and serve as joint first authors. Rajita Sinha and Robert S. Sherwin contributed equally and serve as joint senior authors.
Find articles by Constable, R. in: JCI | PubMed | Google Scholar
1Section of Endocrinology, Yale University School of Medicine, New Haven, Connecticut, USA. 2Division of Endocrinology, University of Southern California Keck School of Medicine, Los Angeles, California, USA. 3Department of Psychiatry, 4Department of Radiology, and 5Yale Center for Clinical Investigation, Yale University School of Medicine, New Haven, Connecticut, USA.
Address correspondence to: Robert S. Sherwin, C.N.H. Long Professor of Medicine, Director, Yale Center for Clinical Investigation, Yale School of Medicine, The Anlyan Center S141, PO Box 208020 New Haven, Connecticut 06520-8020, USA. Phone: 203.785.4183; Fax: 203.737.5558; E-mail: robert.sherwin@yale.edu.
Authorship note: Kathleen A. Page and Dongju Seo contributed equally to this work and serve as joint first authors. Rajita Sinha and Robert S. Sherwin contributed equally and serve as joint senior authors.
Find articles by Sherwin, R. in: JCI | PubMed | Google Scholar
1Section of Endocrinology, Yale University School of Medicine, New Haven, Connecticut, USA. 2Division of Endocrinology, University of Southern California Keck School of Medicine, Los Angeles, California, USA. 3Department of Psychiatry, 4Department of Radiology, and 5Yale Center for Clinical Investigation, Yale University School of Medicine, New Haven, Connecticut, USA.
Address correspondence to: Robert S. Sherwin, C.N.H. Long Professor of Medicine, Director, Yale Center for Clinical Investigation, Yale School of Medicine, The Anlyan Center S141, PO Box 208020 New Haven, Connecticut 06520-8020, USA. Phone: 203.785.4183; Fax: 203.737.5558; E-mail: robert.sherwin@yale.edu.
Authorship note: Kathleen A. Page and Dongju Seo contributed equally to this work and serve as joint first authors. Rajita Sinha and Robert S. Sherwin contributed equally and serve as joint senior authors.
Find articles by Sinha, R. in: JCI | PubMed | Google Scholar
Published September 19, 2011 - More info
J Clin Invest. 2011;121(10):4161–4169. https://doi.org/10.1172/JCI57873.
© 2011 The American Society for Clinical Investigation
Received: March 4, 2011; Accepted: July 27, 2011
-
Introduction
Glucose is an important regulatory signal and the primary fuel source for the brain (1). Specialized glucose-sensing neurons located in the hypothalamus, hindbrain, and forebrain are important in the control of glucose homeostasis and feeding behavior (1, 2). Transient declines in blood glucose increase hunger and therefore mobilize an individual toward food consumption (3, 4), particularly high-sugar (5) and high-fat foods (6). Further, hypoglycemia provokes a physiological stress response to mobilize the individual toward seeking food and restoring glucose levels (6). While the role of hindbrain and hypothalamic neuronal responses in hypoglycemia and maintaining energy homeostasis is well characterized (1, 2, 7), the specific neural mechanisms mediating the motivational drive for food under mild hypoglycemic conditions are not known. We hypothesized that a reduction in circulating glucose, to levels commonly observed several hours after glucose ingestion in healthy individuals (8), would activate brain reward and motivation pathways, including the striatum and insula, while concomitantly increasing desire for high-calorie foods.
To test this hypothesis, we performed functional MRI (fMRI) studies in 14 healthy (9 nonobese and 5 obese) subjects 2 hours after ingestion of a standardized lunch. Subjects viewed high-calorie food, low-calorie food, and non-food images while lying in the scanner during a stepped hyperinsulinemic euglycemic-hypoglycemic clamp. To control for potential session order effects, 7 additional subjects viewed the same pictures during a hyperinsulinemic euglycemic clamp under identical conditions (Figure 1A). Behavioral ratings of wanting and liking were presented after each food and non-food image (Figure 1B), and hunger ratings were assessed at the beginning and end of each phase. This approach allowed us to identify how a standardized reduction in circulating glucose, independent of changes in circulating insulin, interacts with external food cues to modulate the neural circuitry that controls feeding behavior.
Figure 1Study procedure. (A) While subjects were in the scanner, a hyperinsulinemic clamp was performed with a constant infusion of insulin along with variable amounts of glucose to maintain euglycemic conditions for the first 60 minutes. For the stepped clamp study (n = 14), plasma glucose was then lowered to approximately 65 mg/dl for the hypoglycemic phase. For the euglycemic control study (n = 7), plasma glucose was maintained at approximately 90 mg/dl (dotted line). During both conditions, functional scans were performed while subjects viewed images that were projected onto a screen in the scanner. (B) Time course of a single trial. Each trial consisted of 3 events. First, a high-calorie food, low-calorie food, or non-food picture appeared for 6 seconds. Second, two rating scales were presented for 3 seconds each and consisted of liking and wanting scales with values 1–9, where a rating of 1 indicated “not at all” and a rating of 9 indicated “very much.” At the end of a trial, a fixation cross appeared, and subjects relaxed until the onset of the next trial.
-
Results
Metabolic changes
Plasma glucose was maintained at 88 ± 2 mg/dl during the euglycemic phase and reduced to 67 ± 1 mg/dl during the hypoglycemic phase of the study. Glucose levels were maintained at 92 ± 4 mg/dl throughout the euglycemic control study (Figure 2A). Plasma insulin levels were equivalent during both study sessions (Figure 2B). Plasma cortisol levels were significantly higher during the hypoglycemic versus the euglycemic phase (P = 0.003) but were not different throughout the euglycemic control study (Supplemental Figure 1; supplemental material available online with this article; doi: 10.1172/JCI57873DS1). The mild hypoglycemic stimulus did not significantly alter plasma epinephrine, glucagon, leptin, or ghrelin levels. Growth hormone levels rose (hypoglycemic: 11.5 ± 1; euglycemic: 5.9 ± 2 ng/dl, P < 0.001) and C-peptide levels declined (hypoglycemic: 0.50 ± 0.07; euglycemic: 0.87 ± 0.1 ng/ml, P < 0.001) during the hypoglycemic versus the euglycemic phase, and they remained unchanged during the euglycemic control study.
Figure 2Plasma glucose and insulin levels. (A) Plasma glucose and (B) insulin levels (mean ± SEM) during the stepped euglycemic-hypoglycemic (black circles) and euglycemic control (white squares) study sessions. *P < 0.001 unpaired t test, euglycemic versus hypoglycemic session.
Neural activation and behavioral ratings
Main effects of condition. In the total group, a significant effect was detected for euglycemic versus hypoglycemic conditions across all food and non-food visual stimuli (see Table 1 for all brain regions that were significantly affected by euglycemia relative to hypoglycemia). There was greater activation of the prefrontal cortex (PFC) and anterior cingulate cortex (ACC) under euglycemia relative to hypoglycemia, whereas the nucleus accumbens, insula, hypothalamus, thalamus, caudate, and putamen were preferentially activated under hypoglycemia relative to euglycemia (P < 0.05, corrected, covaried for BMI, Figure 3A). Hunger ratings were greater under hypoglycemic (5.7 ± 0.5) versus euglycemic (4.5 ± 0.5) conditions (P = 0.009) and did not significantly vary as a function of BMI. However, significant BMI group interactions were seen with condition. Obese individuals showed greater left substantia nigra/ventral tegmental area (SN/VTA) activation and greater bilateral activation of the hypothalamus, thalamus, striatum, and insula during hypoglycemia relative to euglycemia (P < 0.05, corrected). Furthermore, increased activation of the medial PFC or ACC under euglycemia relative to hypoglycemia did not occur in obese individuals (Figure 3B). Nonobese individuals demonstrated greater right insula and putamen activity during hypoglycemia relative to euglycemia and greater activity in the left SN/VTA, bilateral hippocampus, and medial PFC and ACC during euglycemia relative to hypoglycemia (Figure 3C).
Figure 3Differences between euglycemic and hypoglycemic conditions. Axial slices with (A) whole group, covaried for BMI (n = 14), (B) obese group (n = 5), and (C) nonobese group (n = 9) averages, showing brain response to euglycemia compared with mild hypoglycemia across visual cue tasks (threshold of P < 0.05, 2 tailed, FWE whole brain corrected). Red and yellow areas show greater activity during euglycemia, and blue areas indicate greater activity during hypoglycemia. The color scale gives the t value of the functional activity. Eu, euglycemia; Hypo, hypoglycemia; NAcc, nucleus accumbens; Hyp, hypothalamus; VMPFC, ventromedial prefrontal cortex; Hipp, hippocampus; L, left; R, right. MNI coordinates were used to define brain regions.
Condition × task effects. Next we compared how the level of plasma glucose influenced the ability of visual food stimuli (high-calorie and low-calorie foods) to affect brain activation and wanting and liking of food (Figure 4). During euglycemia compared with hypoglycemia, there was significantly greater activation in the PFC and ACC, whereas hypoglycemia compared with euglycemia provoked greater activation in the insula and striatum in response to images of high- and low-calorie foods (P < 0.05, corrected, covaried for BMI, Figure 4A). Hypoglycemia compared with euglycemia also caused a greater wanting (P = 0.02, covaried for BMI) but no difference in liking of food (Figure 4B).
Figure 4Condition × task effects. (A) Axial slices with group averages (n = 14), covaried for BMI, showing brain response to food (high-calorie and low-calorie) cues under euglycemia compared with mild hypoglycemia (threshold of P < 0.05, 2-tailed, FWE whole brain corrected). (B) Wanting and liking ratings for food during euglycemia (gray bars) and mild hypoglycemia (black bars). *P = 0.02. (C) Brain response specifically to high-calorie food images under euglycemia compared with mild hypoglycemia (threshold of P < 0.05, 2-tailed, FWE whole brain corrected). (D) Wanting and liking ratings for high-calorie foods during euglycemia (gray bars) and mild hypoglycemia (black bars); **P = 0.006. Red/orange areas show greater activity, and blue areas indicate more suppressed activity during euglycemia relative to hypoglycemia. MNI coordinates were used to define brain regions.
Mild hypoglycemia preferentially increased activation of the striatum and insula in response to high-calorie food stimuli (Figure 4C) and provoked a greater wanting (P = 0.006, covaried for BMI) but no difference in liking of high-calorie foods (Figure 4D). In contrast, low-calorie food cues did not provoke a differential brain response (data not shown) or a significant difference in the behavioral response to mild hypoglycemia compared with euglycemia (Supplemental Figure 2). These results could not be attributed to nonspecific time-related effects, since no differences in brain activation to food cues or in hunger ratings were observed during the second half of the euglycemic control study session compared with the first half (Supplemental Figure 3). Thus, mild hypoglycemia sets in motion adaptive mechanisms in motivational pathways to specifically increase wanting of high-energy and glucose-rich foods.
Neural activation and neuroendocrine correlations
The association of changes in circulating glucose and hormones with brain activation to high-calorie food images was assessed using whole-brain, voxel-based correlation analyses (Figure 5). Higher plasma glucose levels correlated with greater brain activity in executive control centers in the ACC and ventromedial PFC, whereas higher levels of plasma cortisol, but not other hormones, were correlated with greater activation in reward regions, such as the insula and putamen (P < 0.01, corrected), in response to high-calorie food cues.
Figure 5Whole-brain, voxel-based correlation analyses. Axial brain slices and corresponding scatter plots showing correlations between (A) plasma glucose levels and VMPFC/ACC response to high-calorie food images; and (B) plasma cortisol levels and left and right insula/striatal response to high-calorie food images. P < 0.01, 2-tailed, FWE whole brain corrected. Areas in yellow indicate brain regions positively correlated to plasma glucose and plasma cortisol levels. There were no outliers in these associations. MNI coordinates were used to define brain regions.
-
Methods
Study participants. Twenty-one healthy male (n = 12) and female (n = 9) volunteers with a mean (±SD) age of 31.4 ± 7.9 years and BMI of 25.2 ± 4 kg/m2 who were weight stable for the 3 months before recruitment participated in this study. Fourteen subjects (9 male, 5 female; mean [±SD] age, 30.4 ± 7.9 years; BMI, 25.6 ± 4.6 kg/m2) participated in a stepped euglycemic-hypoglycemic study; 9 of these subjects were nonobese (BMI, 22.8 ± 2.9 kg/m2), and 5 were obese (BMI, 30.9 ± 1.4 kg/m2). Seven subjects (3 male, 4 female; mean [±SD] age, 33.6 ± 8.0 years; BMI, 24.3 ± 2.4 kg/m2) participated in a euglycemic-euglycemic control study. We recruited subjects with a range of BMIs that allowed us to investigate a distribution representative of the general population; however, whole group statistical analyses were covaried for BMI. All subjects participated in a screening visit, which included a medical history, physical examination, and blood tests. Female subjects were studied during the follicular phase of their menstrual cycle. We excluded individuals with a history of smoking, medical illness, hematocrit less than 33%, elevated hemoglobin A1C, claustrophobia, and metal implants, as well as vegetarians and those taking medications known to influence metabolism.
Study protocol. On the fMRI study day, subjects reported to the Yale–New Haven Hospital Research Unit at noon, where they ingested a standardized lunch. Then, an intravenous catheter was inserted into a distal arm or hand vein: this arm was gently heated, allowing for sampling of arterialized venous blood. A second intravenous catheter was established for the administration of insulin and glucose. One hour before the scanning session, participants received pre-training for the task. Subjects were given a demonstration of the two rating scales and completed a block of 15 practice trials. Subjects were informed that they would receive insulin during the study and that their glucose levels would be reduced below normal, which could lead to symptoms of low blood sugar. They were, however, blinded to the timing of changes in glycemic state.
Two hours after the standardized lunch, subjects began the 120-minute fMRI study session, which included a stepped euglycemic-hypoglycemic study or euglycemic-euglycemic control study. Blood oxygen level–dependent (BOLD) acquisitions were obtained while subjects viewed high-calorie food, low-calorie food, and non-food visual images during euglycemia and hypoglycemia. Insulin was infused at a constant rate of 2 mU/kg/min, with a variable infusion of 20% glucose adjusted to achieve euglycemia or mild hypoglycemia. This approach allowed us to examine the effects of mild hypoglycemia compared with euglycemia on brain activation and behavioral responses to food cues. Blood sampling was performed at approximately 5-minute intervals for glucose measurements and at 10- to 20-minute intervals for measurements of plasma insulin, C-peptide, epinephrine, cortisol, glucagon, growth hormone, ghrelin, and leptin.
Visual stimuli. Tasks were presented using E-Prime software (Psychological Software Tools Inc.). Picture stimuli for fMRI consisted of high-calorie food, low-calorie food, and non-food neutral pictures. Pictures were selected from various Web sites and from the International Affective Picture System (37). High-calorie and low-calorie food pictures were balanced to ensure equivalent levels of emotional valence across glycemia sessions. All food pictures had been previously rated on affective valence in a pilot study performed outside the MRI scanner, and there was no statistical difference in affective valence between high- and low-calorie food items. Participants in the pilot study were of age and BMI similar to those of the participants in the fMRI study. High-calorie food pictures included items such as hamburgers, French fries, cookies, ice cream, chocolate, and pizza. Low-calorie food pictures included items such as salads, broccoli, bean sprouts, tofu, and fruits. High-calorie food items were significantly greater than low-calorie food items in caloric density (352 ± 118 vs. 79 ± 23 kcal/100 g) and fat content (16.6 ± 9 vs. 1.2 ± 2.5 g/100 g) (P < 0.001, Table 2). The USDA National Nutrient Database was used to determine the nutritional content of food items ( http://www.nal.usda.gov/fnic/foodcomp/search). The control stimuli consisted of neutral, non-food pictures (e.g., building, basket, book, bicycle, and door). Non-food pictures did not include utensils, so as not to confuse participants by associating them with food stimuli. There were 8 fMRI runs, and each run included 21 randomized images of 7 high-calorie foods, 7 low-calorie foods, and 7 non-food items, resulting in a total of 168 pictures (56 high-calorie, 56 low-calorie, and 56 non-food). Each glycemic session (euglycemic and hypoglycemic) consisted of 4 runs with 84 pictures (28 high-calorie, 28 low-calorie, 28 non-food). Each picture was presented only once for a participant, and stimuli were counterbalanced and randomized across participants.
Task and behavioral ratings. A series of pictures were presented during the euglycemic and hypoglycemic phases of the task. A fast event-related design was used that included a “time-jitter” technique to effectively differentiate BOLD activity into separate events of interest within a trial (38). Each trial consisted of 3 events. First, a high-calorie food, low-calorie food, or non-food picture appeared for 6 seconds. After the picture presentation, two rating scales (liking and wanting) appeared for 3 seconds. Rating scales consisted of numbers 1–9, with a rating of 1 indicating “not at all” and a rating of 9 indicating “very much.” To select their ratings, the participant pressed buttons using a non-ferromagnetic button box. At the end of a trial, a fixation cross appeared with a jittered inter-trial interval (mean, 6 seconds; range, 3–9 seconds), during which participants relaxed until the onset of the next trial. Eight runs (4 during euglycemic phase and 4 during hypoglycemic phase) were acquired, each lasting 6 minutes 36 seconds. Pictures were presented in random order and counterbalanced during each phase of the study. Each picture was presented only once for each participant.
fMRI acquisition. Images were obtained using a 3-T Siemens Trio MRI system equipped with a standard quadrature head coil, using T2*-sensitive gradient-recalled single-shot echo planar pulse sequence. Subjects were positioned in the coil, and head movements were restrained using foam pillows. Anatomical images of the functional slice locations were next obtained with spin echo imaging in the axial plane parallel to the AC-PC line with repetition time (TR) = 300 ms, echo time (TE) = 2.46 ms, bandwidth = 310 Hz/pixel, flip angle = 60°, field of view = 220 × 220 mm, matrix = 256 × 256, 32 slices with slice thickness = 4 mm and no gap. Functional BOLD signals were then acquired with a single-shot gradient echo planar imaging (EPI) sequence. Thirty-two axial slices parallel to the AC-PC line covering the whole brain were acquired with TR = 2,000 ms, TE = 25 ms, bandwidth = 2,520 Hz/pixel, flip angle = 85°, field of view = 220 × 220 mm, matrix = 64 × 64, 32 slices with slice thickness = 4 mm and no gap, 198 measurements. At the end of the functional imaging, a high-resolution 3D magnetization-prepared rapid gradient echo (MPRAGE) sequence (TR = 2530 ms; TE = 2.62 ms; bandwidth = 240 Hz/pixel; flip angle = 9°; slice thickness = 1 mm; field of view = 256 × 256 mm; matrix = 256 × 256) was used to acquire sagittal images for multi-subject registration.
fMRI analysis. All data were converted from Digital Imaging and Communication in Medicine (DICOM) format to Analyze format using XMedCon ( xmedcon.sourceforge.net) (39). During the conversion process, the first 3 images at the beginning of each of the 6 functional series were discarded to enable the signal to achieve steady-state equilibrium between radio frequency pulsing and relaxation, leaving 195 measurements for analysis. Images were motion corrected for 3 translational and 3 rotational directions (40). Trials with linear motion in excess of 1.5 mm or rotation greater than 2 degrees were discarded. The first run of each phase was removed (leaving 3 runs for each phase) because some subjects had not achieved the target mild hypoglycemic level prior to the first run. Individual subject data were analyzed using a general linear model (GLM) on each voxel in the entire brain volume with 3 regressors specific for the task. The regressors were those images that pertained to the time when the subject was viewing the high-calorie food image, low-calorie food image, or non food image. The resulting functional images for each image type were spatially smoothed with a 6-mm Gaussian kernel to account for variations in the location of activation across subjects. The output maps were normalized beta-maps, which were in the acquired space (3.44 mm × 3.44 mm × 4 mm).
To take these data into a common reference space, 3 registrations were calculated within the Yale BioImage Suite software package ( http://www.bioimagesuite.org) (41, 42). The first registration performs a linear registration between the individual subject raw functional image and that subject’s 2D anatomical image. The 2D anatomical image is then linearly registered to the individual’s 3D anatomical image. The 3D differs from the 2D in that it has a 1 × 1 × 1–mm resolution, whereas the 2D z-dimension is set by slice thickness and its x-y dimensions are set by voxel size. Finally, a nonlinear registration is computed between the individual 3D anatomical image and a reference 3D image. The reference brain used was the Colin27 Brain (43), which is in Montreal Neurological Institute (MNI) space (44) and is commonly applied in SPM and other software packages. All 3 registrations were applied sequentially to the individual normalized beta-maps to bring all data into the common reference space.
Data were converted to AFNI format (ref. 45; http://afni.nimh.nih.gov) for group analysis. We applied a linear mixed effects (LME) model (condition, which was either euglycemia or hypoglycemia, by task, which was either non-food, low-calorie food, or high-calorie food images) in which the subject was treated as a random factor, covarying for BMI using the LME modeling program 3dLME from AFNI ( http://afni.nimh.nih.gov/sscc/gangc/lme.html). Results were masked and converted back into ANALYZE format for viewing in BioImage Suite. Whole-brain, voxel-based correlation analyses with plasma glucose and cortisol levels were implemented using BioImageSuite (41, 42) with the application of AFNI AlphaSim family-wise error (FWE) correction for multiple comparisons. All data analyses were based on 2-tailed tests.
Biochemical analysis. Plasma glucose was measured enzymatically using glucose oxidase (YSI). Double-antibody radioimmunoassay was used to measure plasma insulin, ghrelin, leptin, and glucagon (Millipore) and cortisol (Diagnostic Products Corp.). Plasma epinephrine was measured by high-performance liquid chromatography (ESA).
Statistics. Analyses of ratings for liking, wanting, and hunger were conducted using LME models (PROC MIXED version 9.2, SAS) (46). Fixed factors in the models included session type (euglycemia or hypoglycemia), food task (non-food, low-calorie, high-calorie), and their interaction for liking and wanting ratings. For hunger ratings the model included session type and time of rating and their interaction. Post hoc linear contrasts (2-tailed 0.05 significance level) were corrected for multiple comparisons using a modified Bonferroni procedure. A random subject effect was included to account for the correlation of repeated assessments from each subject. All analyses were adjusted for BMI. Residual analysis was performed to confirm that outcomes did not depart from the assumption of normality. Paired t tests were performed on hormonal data averaged across euglycemia and hypoglycemia. A P value less than 0.05 was considered significant. Unless otherwise stated, data are presented as mean ± SEM.
Study approval. All participants provided informed consent prior to participation in this study. All aspects of the study were approved by the Human Investigation Committee at the Yale School of Medicine, New Haven, Connecticut, USA.
Copyright © 2024 American Society for Clinical Investigation
ISSN: 0021-9738 (print), 1558-8238 (online)