Low Calorie Diet-induced Weight Loss is Associated with Altered Brain Connectivity and Food Desire in Obesity

Abstract

Objective:

The main objective of this study is to better understand the effects of diet-induced weight loss on brain connectivity in response to changes in glucose levels in individuals with obesity.

Methods:

A total of 25 individuals with obesity among which 9 had a diagnosis of type 2 diabetes (T2D), underwent fMRI scans before and after an 8-week low calorie diet. We used a 2-step hyper-euglycemia clamp approach to mimic the changes in glucose levels observed in the post-prandial period in combination with task-mediated fMRI intrinsic connectivity distribution (ICD) analysis.

Results:

After the diet, participants lost an average of 3.3% body weight. Diet-induced weight loss led to a decrease in leptin levels, an increase in hunger and food intake, and greater brain connectivity in the parahippocampus, right hippocampus and temporal cortex (limbic-temporal network). Group differences (with vs without T2D) were noted in several brain networks. Connectivity in the limbic-temporal and frontal-parietal brain clusters inversely correlated with hunger.

Conclusion:

Short-term low-calorie diet led to a multifaceted body response in patients with obesity, with an increase in connectivity in the limbic-temporal network (emotion and memory), and hormone and eating behavior changes, that may be important for recovering the weight lost.

Introduction

One of the major challenges in individuals with obesity is weight loss maintenance. Current understanding of the homeostatic regulation of energy metabolism and obesity suggests that the brain plays a central role(1) by integrating multiple peripheral metabolic inputs related to energy status, such as nutrients(glucose) and feeding-related hormones (e.g., leptin, insulin, ghrelin), which are then combined with external signals such as time of day and social context, to influence motivation and eating behavior(2). Key brain regions, including the hypothalamus (feeding center), limbic system (desire to eat), striatum (motivation), and frontal cortex (executive function)(3) orchestrate this interplay between hedonic and homeostatic systems to modulate feeding behavior(4).

Altered brain function has been well documented in obesity with the use of functional magnetic resonance imaging (fMRI)(1, 5, 6). Individuals with obesity, in comparison to individuals with normal weight, have increased responses to food cues in reward-related brain regions(7), and decreased responses in brain regions involved with executive-control regions(8). In addition, changes in circulating glucose levels have been shown to participate in acute regulation of eating behavior(9–11) and brain activity(12–14), which has been shown to be altered in subjects with obesity(12, 15–17). In our previous work utilizing the euglycemic-hyperglycemic clamp model(17), we observed that visual food cues resulted in greater activity in brain areas associated with motivation and executive control (insula, putamen, and prefrontal cortex) in subjects with obesity compared to normal weight during hyperglycemia. These results suggested that individuals with obesity have enhanced motivation and less self-control over food cues, along with diminished responsiveness to homeostatic signals (hyperglycemia). Investigating the effect of hyperglycemia on brain networks may offer further insights into the pathogenesis of obesity. Moreover, rather than single isolated brain regions, functional neural networks involved in food and hormone regulation, such as the default mode network and temporal lobe network, also play a role in obesity (6, 18). However, less is known about the effects of hyperglycemia on brain connectivity in individuals with obesity.

Studies have shown that abnormal brain responses in obesity(6, 18) may be reversed with weight loss(19), behavioral therapy(20) and bariatric surgery(21, 22). While long-term impacts of weight loss interventions have been well studied, less is known about the effects on brain activity after a short-term dietary intervention. Therefore, this study investigates the effects of an 8-week low-calorie diet program on brain connectivity in participants with obesity. We used the novel fMRI technique, Intrinsic Connectivity Distribution (ICD), to measure synchronous brain activity during visual task, identifying patterns of correlated neural activity, and providing insights into how brain regions interact with each other in response to difference conditions(23, 24). ICD uses a whole-brain voxel-wise approach that does not require a prior selection of pre-defined brain regions (seed-based)(23, 25) to determine brain connectivity and can be performed across different types of tasks(24, 26). The absence of seed region selection ensures a data-driven approach, allowing the identification of connectivity patterns that may not have been considered in advance. ICD analysis can provide insights into how different brain regions interact and communicate with each other, and how this communication changes over time or in response to different conditions(24).

The main goal of this study is to investigate the impact of weight loss on brain connectivity in response to increments in blood glucose levels, as previously shown(27, 28). This is the first study examining intrinsic connectivity before and after 8-week diet in obesity using a multimodal approach that concurrently monitors metabolic, hormonal, food motivation, and neural activities. As an exploratory study, we investigated the effects of weight loss in a subgroup of individuals with obesity and type 2 diabetes (T2D). We hypothesized that diet-induced weight loss will improve glucose metabolism and restore, at least partially, the altered brain connectivity in networks associated with emotion and control in individuals with obesity.

Methods

Participants

A total of 25 individuals with obesity (15 women/10 men) completed this study with a mean age of 46±9 years old and body mass index (BMI) of 33.4±2.8. Among these, 9 (5 women/4 men) had a diagnosis of T2D (hemoglobin A1c [HbA1c] 6.6±0.7 on day of screening). All participants were obese (BMI>30 kg/m²) and were weight stable for the past 3 months. Participants in the obesity group (OB(-T2D)) had normal OGTT and HbA1c<5.7% (no diagnosis of diabetes or prediabetes as defined by American Diabetes Association Guidelines)(29). Participants in the obesity and T2D group (OB(+T2D)) had prior history of T2D, or a screening OGTT and/or HbA1c diagnostic for T2D. Exclusion criteria included history of major medical/psychiatric disorders, pregnancy, contraindications for an MRI scan, BMI>40 kg/m² (scanner weight limit), and T2D on insulin or glucagon-like peptide 1 (GLP-1) agonists (Figure S1). Participants selection in the final analyses are further detailed in the supplemental information.

Study Design

Screening Day:

Participants arrived at the Hospital Research Unit (HRU) in the morning of the study after at least an 8 hour overnight fast. Measurements of BMI and vital signs were obtained. Females were assessed for pregnancy with a urine human chorionic gonadotropin (hCG) test.

Oral glucose tolerance test:

During the screening visit, after obtaining baseline glucose and insulin levels, subjects ingested 7.5 oz of glucola (75g glucose in flavored water). Blood samples were taken at −15, 0, 10, 20, 30, 60, 90, and 120 minutes to measure plasma glucose concentrations to exclude the diagnosis of diabetes.

Magnetic Resonance Imaging Visit:

Qualifying participants returned to the HRU approximately 2–3 weeks later for the fMRI-clamp visit (Figure 1). The participants were instructed to avoid any strenuous activity the day before the MRI visit. Participants from the OB(-T2D) group were admitted in the morning of the fMRI visit, while patients from the OB(+T2D) group were admitted the night before the fMRI scan (around 10pm) and fasted after 11pm. Plasma glucose levels were monitored every hour, and an insulin drip was started with a goal to keep overnight fasting glucose levels within a target range of 100–120 mg/dL. Prior to transfer to the MRI suite, insulin drip was discontinued.

Figure 1:

MRI with 2-Step Clamp Study Design

Magnetic Resonance Imaging Visit: The visit consisted of a MRI scanning together with a two-step hyperglycemic-euglycemic clamp. fMRI was performed while participants were presented with a series of visual stimuli (food: high calorie (HC) and low calorie (LC), and non-food (NF) images). After viewing each image, participants provided liking and wanting ratings. Hunger was measured before and after each fMRI run. After the fMRI-clamp procedure, participants were provided an ad libitum buffet meal. This visit was performed twice: before and after an 8-week low calorie diet.

Two-Step Hyperglycemic-Euglycemic Clamp:

We utilized the 2-step hyperglycemic-euglycemic clamp approach to mimic the early post-prandial state, which has been shown to be a critical period in which plasma glucose levels influence hunger and modulate food intake(9–11). This test was performed by increasing blood glucose to ~200mg/dL (hyperglycemia) followed by decreasing glucose levels to ~100 mg/dL (euglycemia). A fasting blood sample was collected for measurement of baseline levels of glucose, lipids, HbA1c and hormones (insulin, leptin, ghrelin, GLP-1). Research participants were then transferred to the MRI suite. Blood was drawn throughout the scan for glucose and hormone measurements. Participants were given intravenous (IV) 20% dextrose solution to raise blood glucose levels to ~200mg/dL before undergoing the visual task while getting the fMRI scan. After completion of the visual task, dextrose infusion was stopped, and IV insulin at a rate of 2mU/kg/min was infused to reduce serum glucose levels to ~100 mg/dL, before repeating the same visual task (Figure 1).

Food Motivation:

Visual stimuli task: We followed our prior validated fMRI-metabolic study protocol, the details of which have been previously described(16, 17, 30). Each task consisted of 3 events: picture viewing (food and nonfood), rating (“how much do you like the item shown in the picture?”, “how much do you want the item shown in the picture?”), and inter-stimulus interval (blank screen). Each run consisted of 30 pictures shown with a total of 8 runs (Figure 1). This is further detailed in the Supplementary Information. Hunger ratings were measured with a visual analogue scale from 1 to 9 (1= “not at all”; 9= “very much”) performed before and after each run (10 time points in total).

Food Buffet:

After completion of the scan, a buffet lunch was offered, which included choice of hamburger (with/without cheese, lettuce, tomatoes, onions, bacon, ketchup, mustard), macaroni and cheese, chicken, peanut butter and grape jelly, salad (ranch or light vinaigrette dressing), drink (water, coke, diet-coke, orange juice), banana, apple, pound cake, cookie. Without their knowledge, amounts of food consumed were measured and recorded.

Diet:

Research participants were instructed to follow an 8-week low-calorie diet (Males: 1800 Cal/day; Females: 1500 Cal/day) (Macronutrient ratio: 55% carbohydrates, 30% fats, 15% proteins) with the goal to lose ~5% of initial body weight (as recommended in the Position from the Academy of Nutrition and Dietetics)(31). Registered Dietitian formulated diet plans and followed up every 2 weeks to ensure adherence to diet and measurement of body weight/fat with Tanita scale^®. After the 8-week low calorie diet, participants underwent a second fMRI-clamp visit.

Study Outcomes:

Our primary outcome was to evaluate the effect of short-term diet-induced weight loss on brain connectivity during hyperglycemia and euglycemia. As secondary outcomes, we examined the impact of weight loss on 1) Metabolic profile: plasma glucose, HbA1c, hormones (leptin, ghrelin, and GLP-1), and glucose infusion rate (GIR) during the 2-step clamp; 2) Food motivation (hunger) and food intake. As an exploratory outcome, we investigated whether the effects of diet-induced weight loss were impacted by the presence of T2D.

Data Analysis:

The fMRI scan protocols were performed as previously described(16, 30). In short, visual stimuli consisted of 168 pictures of food (56 high calorie (HC); 56 low calorie (LC)) and 56 non-food (NF) pictures using E-Prime software (Psychological Software Tools Inc.). The task was divided into 2 sessions comprising of 4 functional runs. Functional connectivity was calculated across raw task data without regressing out task.

Functional MRI Signal Acquisition

Functional MRI Signal Acquisition was obtained with a 3.0 Tesla Siemens Trio MRI scanner. Functional images were collected with a standard quadrature head coil, using a T2*-sensitive gradient-recalled single shot echo planar pulse and spin echo sequence. A high-resolution 3D sequence (MPRAGE) was used to acquire sagittal images for multi-subject registration. MRI parameters: TR = 1.5 ms, echo time = 30 ms, FOV = 235 mm, matrix size = 84×84, slice thickness = 3 mm, flip angle = 90°, bandwidth = 2126 Hz/pixel with 44 slices) with 266 volumes per run. More details can be found in the supplementary section.

Intrinsic Connectivity Distribution analyses (ICD):

After preprocessing (details in Supplementary Material), ICD was performed at the voxel level for each subject as previously described in Scheinost et al.(23). Functional runs were concatenated for each session (hyperglycemia and euglycemia) separately, and the functional connectivity of each voxel was then calculated in each subject’s individual space. To acquire connectivity patterns, ICD analysis was conducted for the entire 6 minutes of each fMRI run, across images (HC, LC, NF). A gray matter mask was applied and only voxels in the gray matter were used for calculation of connectivity (MNI brain).

Statistical analyses:

Statistical analyses were performed using SPSS version 22.0 (IBM). All data are presented as means and standard deviations, unless otherwise specified. Comparisons within subjects were determined by the two tailed Student’s t-test for paired samples and between subjects by the unpaired Student’s t-test for equality of means. Pearson correlations were performed between ICD and hormonal and eating behavior variables. A value of P<0.05 was used as the statistically significant threshold for all analyses.

Results

Subject Characteristics:

A total of 25 participants completed the diet and fMRI-clamp visits (Table 1 shows the subject characteristics and hormone levels). There were no group differences in demographics except for diabetes-related indicators (glucose, HbA1c, Insulin). In the OB(+T2D), 8 participants were taking metformin at stable dose for more than 3 months, and 1 was not on any anti-diabetes medications.

Table 1 –

Subject Characteristics of the study participants with obesity without T2D (OB(-T2D)) and with T2D (OB(+T2D)).

	OB(-T2D)		OB(+T2D)
	Before Diet	After Diet	Before Diet	After Diet	P-Value Between groups (before diet)
	N = 16 (10F/6M)		N = 9 (5W/4M)
Age (years)	44 ± 8		48 ± 9		0.320
Weight (Kg)	90.7 ± 12.9	87.7 ± 13.0 **	93.2 ± 15.5	89.8 ± 14.6 *	0.674
BMI (kg/m2)	32.8 ± 2.1	31.7 ± 2.5 **	34.0 ± 2.7	32.7 ± 2.5 *	0.225
Glucose (mg/dL)	94.2 ± 5.5	90.8 ± 9.3	118.3 ± 15.2	111.4 ± 16.7	<0.001
HbA1c (%)	5.3 ± 0.3	5.3 ± 0.4	6.4 ± 0.8	6.4 ± 0.7	<0.001
Insulin (uU/mL)	12.9 ± 8.1	11.4 ± 5.5	24.5 ± 6.2	32.1 ± 23.8	<0.001
Ghrelin (ng/mL)	810 ± 416	822 ± 349	566 ± 158	617 ± 149	0.107
Leptin (ng/mL)	43.1 ± 26.0	31.2 ± 22.2 **	37.5 ± 27.1	32.6 ± 24.7 *	0.620
GLP-1 (pm/L)	5.3 ± 7.7	9.6 ± 19.7	6.2 ± 11.0	6.0 ± 9.7	0.813

Blood samples were collected during fasting on fMRI visit before and after the 8-week weight loss diet. Data are presented as mean ± SD. Significance in bold at *P<0.05 and **P<0.001 comparing within subjects before and after the diet.

After the 8-week low-calorie diet, participants lost on average 3.1 kg (3.3% of initial body weight), with both groups losing similar amount of weight (P=0.88 for group comparison). No differences in weight loss were observed by sex (P=0.63). Approximately one third of the participants did not lose or lost less than 1% of their initial body weight: 31% in the OB(-T2D) group and 33% in the OB(+T2D) group. HbA1c levels did not change with the diet in either group (P >0.05).

Non-Imaging Results

Glucose and Hormones (from fMRI-clamp visit, shown in Table 1 and Figure 2):

Figure 2:

Glucose and Hormone Levels during 2-Step Clamp

Plasma glucose (Figure 2A), insulin (Figure 2B), glucose infusion rate (GIR) (Figure 2C), leptin (Figure 2D), ghrelin (Figure 2E), and GLP-1 (Figure 2F) levels during the 2-step (hyperglycemia-euglycemia) hyperinsulinemic clamp in OB(-T2D) (blue circle) and OB(+T2D) participants (red square) before (dashed lines) and after the diet (solid line). Data are presented as mean ± SD.

No statistically significant changes were observed in fasting glucose, insulin, and GLP-1 levels before and after weight loss. GLP-1 levels during the clamp did not change after diet. Fasting ghrelin levels did not change with diet, (average levels: Pre=556±216, Post=611±249, P=0.02) (Figure 2e).

Glucose infusion rate (GIR) calculated during the last 25–30 minutes of each step of the clamp, an indicator of insulin sensitivity, was higher in the OB(-T2D) compared to the OB(+T2D) group (P<0.01). GIR in both groups did not significantly change after the diet (P >0.10).

Food motivation and Food Intake (Figure 3):

Figure 3:

Hunger and Food intake.

Figure 3A depicts average hunger ratings during the MRI-clamp procedure before (dashed line) and after the diet (solid line). After each MRI scan, participants were offered a food buffet lunch meal in which the amounts of food consumed was measured and recorded. Figure 3B shows the total calories consumed of the buffet meal before and after the diet for all 25 study participants divided by amount of protein (prot), carbohydrate (carb), and fat (in kilocalories). Data are presented as mean ± SD. Significance at *P<0.05 and **P<0.001.

Fasting hunger did not change with the diet (before: 5.6±2.1, after:6.2±1.8; P=0.212), although there was a progressive and steady increase in hunger throughout the fMRI-clamp study after weight loss, with subjects reporting higher hunger at the start of the euglycemic phase (before:6.1±2.1, after:7.2±1.7; P=0.045) and at the end of the clamp (before: 6.9±1.6, after: 7.9±1.6; P=0.001). There was no statistically significant group interaction in hunger, wanting, or liking before and after weight loss.

After each MRI scan, participants were offered a food buffet lunch (Figure 1). There was significant increase in total calories consumed from the buffet meal after the diet (before:939±443 kilocalories, after:1185±532 kilocalories; P=0.02). (Figure 3b), with a significant increase in calorie consumption from fat (before:38.5±22.8, after:52.9±30.4; P=0.009) but not from carbohydrate or protein (P=0.13 and 0.11, respectively) (Figure 3b). Baseline hunger and food intake were not different between the groups, and no statistically significant interaction of diet by group was noted (P>0.05).

Correlations:

Analyzing the 2 groups together, average hunger throughout the 2-step clamp correlated at trend-level significance with food intake both before (Pearson correlation: total gram r= 0.41, P=0.04 (uncorrected), and fat intake (in calories): r= 0.42, P=0.04, but not with protein nor carbohydrate) and after the diet (total gram: r= 0.43, P=0.03 (uncorrected), but not with fat intake, protein nor carbohydrates). No statistically significant correlations were observed between total calorie and total gram intake from food buffet with hormone levels, either before or after the diet (P >0.05, both groups together).

Imaging Results:

Intrinsic Connectivity Distribution (ICD)

ICD during task analysis identified a main effect of diet (Figure 4, higher FWE corrected threshold at P<0.01) with an increase in connectivity in limbic-temporal network including right parahippocampus, right hippocampus and temporal cortex (BA22 and BA40). The diet by group interaction revealed statistically significant group differences in several networks including limbic-temporal, frontal-parietal cluster, and another cluster involving brain stem and cerebellum (Figure 5a, and Supplemental Table 1, initial threshold of P<0.001, whole brain, FWE corrected at P<0.001, α <0.05). When comparing before and after diet, weight loss led to greater connectivity in the OB(+T2D) group and less connectivity in the OB(-T2D) group in these networks. (Figure 5b). Main effect of session and group were not statistically significant, as well as diet by session interaction.

Figure 4:

Main Effect of Diet on Brain Connectivity

Whole brain map of main effect of diet showing in red/yellow statistically significant connectivity in limbic network including right hippocampus and parahippocampus and BA22 (temporal cortex) and BA40 (parietal cortex) (initial threshold of P<0.001, whole brain, FWE corrected at P<0.05).

Figure 5:

Brain Connectivity

Figure 5A: Group X Diet brain maps (Group: OB(-T2D) x OB(+T2D); Diet: Before x After) showing in red/yellow. Significantly increased connectivity was found comparing before and after diet in healthy OB(-T2D) (n = 16) than OB(-T2D) with T2D (n = 9) participants. Regions identified as significantly different showing in these brain slices were amygdala, hippocampus, hypothalamus, and anterior prefrontal cortex (PFC). Significance at P<0.05 initial threshold of P<0.001. Figure 5B depicts beta-values from ICD in hypothalamus-temporal lobe regions during hyperglycemia and euglycemia in OB(-T2D) (blue) and OB(+T2D) (red) groups before (pale bars) and after the diet (solid bars). Figures 5C and 5D: Correlations between hypothalamus-temporal lobes brain connectivity and average hunger during hyper and euglycemia before (r= 0.05, P=0.27) and after the diet (r=0.36, P<0.005).

Correlations:

Significant inverse correlations were observed between connectivity strength of the limbic-temporal and frontal-parietal brain clusters and average hunger during hyper and euglycemia after diet (P<0.005) but not before diet (r=0.05, P=0.27) (Figures 5c and 5d, frontal-parietal network data not shown). The limbic-temporal cluster includes bilateral hypothalamus, bilateral amygdala, and left hippocampus. The frontal-parietal cluster includes bilateral prefrontal cortex (BA10), left motor cortex (BA6), left parietal lobes including angular gyrus (BA39), and inferior parietal lobe (BA40). No statistically significant correlations were observed between ROI in the limbic-temporal, frontal-parietal and cerebellum-brain stem clusters with food intake (from buffet meal), glucose, hormones, GIR, and wanting/liking

Discussion

This was a pilot study investigating the effects of short-term weight loss on the relationship between glucose levels and brain connectivity in individuals with obesity. Employing a multimodal approach, we examined changes in diet-related feeding behaviors in obesity, including hormones, hunger, food consumption (buffet meal) and neural connectivity. For this purpose, we studied 25 individuals with obesity before and after an 8-week low-calorie diet. We found that diet-induced weight loss led to increase in connectivity in the limbic-temporal regions. Together with these changes in brain connectivity in neural networks involved with food motivation, we found noticeable diet-related changes including increased hunger, increased food intake, decreased leptin, and diminished glucose-induced suppression of ghrelin.

Altered brain connectivity has been well documented in individuals with both obesity without(6, 18) and with T2D(32–34). Studies have shown greater connectivity in brain regions (temporal lobe, insular cortex, basal ganglia) associated with reward and food regulation in obesity compared to lean subjects (6, 18). In our study we found a main effect of the diet that was evident by an increase in connectivity in the limbic-temporal regions. These results are consistent with a prior study(19) that showcased heightened resting-state connectivity between the hippocampus and frontal-parietal network, accompanied by improved memory recognition due to short-term caloric restriction-induced weight loss. However, changes in connectivity were not observed in another study comparing individuals with obesity before and after a 8-week low calorie diet(35). This latter study used whole brain resting-state fMRI, with amygdala and hypothalamus as seed regions. Differences in fMRI data analysis may explain the discordant results between these two studies.

As an exploratory feasibility study, we investigated the impact of T2D on brain connectivity after weight loss. When comparing OB(-T2D) and OB(+T2D) groups, differential brain connectivity patterns were noted in limbic-temporal, frontal-parietal, and brain stem-cerebellum areas. Interestingly, our study demonstrated opposite patterns in brain connectivity in the OB(-T2D) and OB(+T2D) groups after weight loss, where weight loss lessened brain connectivity in OB(-T2D) but amplified brain connectivity in OB(+T2D) (Figure 5b). These findings suggest that OB(+T2D) may have differentiated neural mechanisms in comparison to healthy OB(-T2D) in response to weight loss. In support of these results, Hwang et al(28) showed that individuals with obesity and T2D have decreased intracerebral glucose levels in response to hyperglycemia, in comparison to individuals with obesity without T2D. The authors(28) suggested that differential glucose entry and metabolism in the brain could impact brain glucose sensing and feeding behavior in patients with T2D. Differential changes in the limbic-temporal network, an area of feeding and hormone regulation, may indicate different patterns of vulnerability to homeostatic and environmental signals that may converge to comparable feeding and diet-related behaviors. These findings should be explored in future studies.

Prior studies have shown that circulating glucose levels have a direct effect in humans’(16, 17) and rodents’ brains(36), although, we did not observe any significant interactions between diet or group with session (glycemia) in our study. However, it is possible that glucose-induced changes in hormones levels may have played a role in modulating feeding behavior. Feeding-related hormones were measured during the fMRI-clamp studies: insulin, leptin, ghrelin, GLP-1. We observed that 3% weight loss resulted in a 24% decrease in fasting leptin levels. Leptin is a satiety hormone, but high leptin levels are typically seen in obesity because of leptin resistance (37, 38). Decreased leptin levels after weight loss may result in reduced satiety and decreased metabolic rate, which may be a physiological mechanism to conserve energy and restore the weight lost(39, 40). No significant changes in insulin or GIR were noted in this study, most likely due to the small amount of weight loss (<5%) and short duration of diet(41). Weight loss also influenced ghrelin levels (an orexigenic hormone), reducing its suppression from hyperglycemia (initial step of clamp). Ghrelin has been implicated in modulating brain regions associated with regulation of feeding and appetitive responses to food cues(42). Although diet-induced changes in hormones may influence eating behavior and food intake, no direct correlation was observed between hormones and food intake or hunger.

In addition to brain connectivity, we also explored changes in eating behavior following diet. Hunger at baseline (fasting) before the fMRI scan wasn’t affected by weight loss. However, after diet, hunger increased significantly throughout the clamp, particularly towards the end of the scan. In addition, greater food intake was observed after the diet (measured with ad libitum Buffet Meal). This was observed despite no significant changes in insulin resistance (either baseline insulin levels or GIR during the clamps). These results suggest that short-term weight loss diet may stimulate hunger and trigger increased food consumption.

These results indicate that diet-induced weight loss leads to a whole-body response, including hormones, eating behavior, and brain connectivity. However, some limitations were noted: 1) due to time constraints (fMRI scan lasting ~2.5 hours), brain activity was not measured at baseline before hyperglycemia. However, in our previous study(17), we demonstrated normoglycemia followed by hyperglycemia affected brain activity in obesity, with a significant session effect (altered response to hyperglycemia in individuals with obesity but not individuals with healthy weight). 2) Brain responses in OB(+T2D) may have been minimized by the degree of hyperglycemia. T2D individuals (who are chronically exposed to higher blood glucose levels) may sense hyperglycemia (~200 mg/dL) and increment in glucose levels from ~110 mg/dL (OB(-T2D)) and ~90 (OB(+T2D)) differently than individuals with obesity without T2D (28); in addition, lowering glucose levels to ~100 mg/dL may be perceived as hypoglycemia in individuals with T2D. 3) We used the clamp technique to manipulate blood glucose levels and we cannot rule out that brain connectivity in response to oral glucose may be different from IV glucose. However, the clamp technique allows measuring brain connectivity in response to similar glucose and insulin levels before and after diet. 4) Although fMRI was obtained while viewing food and non-food cues, ICD analysis did not allow us to differentiate brain connectivity in response to food and non-food cues. Connectivity needs at least 2.5 minutes of continuous scanning, but pictures were only presented for 6 seconds. Therefore, all the tasks had to be collapsed, and connectivity was obtained for the entirety of each fMRI run. 5) Brain connectivity correlated with hunger after diet, but not with food intake. Hunger was measured while participants were still inside the scan; while food intake occurred after scans ended. It is possible that other factors may have impacted food intake, such as desirability for the Buffet Meal, time of day, and volunteers’ concern of being observed while eating. However, type of food, food presentation and timing of eating were done in the same fashion before and after the diet. 6) Absence of a healthy control group (individuals without obesity) confounds if the effect of weight loss is specific to those with obesity and/or T2D. Furthermore, the inclusion of diabetic patients taking metformin introduces a confounding variable, as the observed effects could be attributed to the medication rather than diabetes alone. However, those who were on metformin were on a stable dose before and after weight loss. 7) Finally, the study experienced a notable drop in the number of participants, decreasing from 86 initially enrolled to a final sample size of 25 due to inclusion/exclusion criteria (Figure S1). Despite these limitations, it should be noted that this is the first study examining intrinsic connectivity before and after 8-week diet in obesity using a multimodal neuroimaging technique that concurrently monitors metabolic, hormonal, and neural activities.

In summary, our study’s primary focus on brain connectivity elucidates the impact of short-term weight loss on the intricate neural networks underlying responses to weight restoration. The observed changes in brain connectivity, coupled with hormonal fluctuations and behavioral changes provide a comprehensive insight into the multifaceted nature of weight loss interventions. These findings collectively underscore the challenges faced by individuals with obesity, particularly those with T2D, when adhering to low-calorie diets. Further exploration into these neural mechanisms holds promising implications for designing more effective and tailored interventions for obesity management and weight loss maintenance.

Study Importance:

Previous research has demonstrated altered brain connectivity in individuals with obesity along with diminished responsiveness to homeostatic signals (hyperglycemia). Additionally, abnormal brain responses observed in obesity and type 2 diabetes have shown potential reversibility through weight loss, behavioral therapy, and bariatric surgery.

Our study’s main finding on brain connectivity investigated the impact of short-term weight loss on the intricate neural networks that may be involved with weight regulation.

This data helps further understand how the human body responds to diet-induced weight loss, resulting in a coordinated brain and peripheral responses to increase hunger and food consumption, and consequently recover the weight lost. These findings collectively underscore the challenges faced by individuals with obesity, particularly those with T2D, when adhering to low-calorie diets. Further exploration into these neural mechanisms holds promising implications for designing more effective and tailored interventions for obesity management and weight loss maintenance.