Longer-term liraglutide administration at the highest dose approved for obesity increases reward-related orbitofrontal cortex activations to food cues: Implications for plateauing weight loss in response to anti-obesity therapies

Abstract

Aims:

GLP-1 analogs have recently risen to the forefront as effective medications for lowering weight through actions in the central nervous system (CNS). However, their actions in the CNS have not yet been studied in the human brain after longer-term administration at the highest dose approved for obesity (liraglutide 3.0mg).

Materials and Methods:

20 subjects with obesity were treated with placebo and liraglutide (3.0mg) in the context of a randomized, placebo-controlled, double-blind, cross-over trial after five weeks of dose escalation. Neurocognitive and neuroimaging (fMRI) responses to food cues were examined at the clinical research center of Beth Israel Deaconess Medical Center.

Results:

While on liraglutide, patients lost more weight (placebo-subtracted −2.7%; p<.001), had decreased fasting glucose (p<.001), and showed improved cholesterol levels. In an uncontrolled analysis, brain activations to food images were not altered in response to liraglutide vs. placebo. When controlled for BMI/weight, liraglutide increased activation of the right orbitofrontal cortex (OFC) to food cues (p<.016, corrected for multiple comparisons).

Conclusions:

In contrast to prior studies, we demonstrate for the first time herein that liraglutide treatment administered over a longer period at the highest doses approved for obesity does not alter brain activations to food cues. A counter-regulatory increase in reward-related OFC activation to food cues can be observed when neuroimaging data are controlled for BMI changes, indicating changes in CNS that could lead to later plateaus of weight loss. These data point to a promising focus for additional interventions which, by contributing to the CNS reward system, could provide tangible benefits towards reversing the plateauing phenomenon and promoting further weight loss.

CT Registration:

ClinicalTrials.gov

Obesity is a worldwide problem with approximately a third of adults in developed countries having obesity and an additional third being overweight [1]. Obesity contributes and/or leads to several comorbid conditions such as type 2 diabetes, cardiovascular diseases, and cancers, which account for four of the ten leading causes of death in the United States (US) [2]. While several factors lead to obesity, the central nervous system (CNS) has been implicated in the development and treatment of obesity [3].

Glucagon-like peptide 1 (GLP-1) analogs such as liraglutide are effective at lowering body weight and are typically well-tolerated with the most common side effect being transient, mild to moderate nausea[4]. Native GLP-1 is derived from the pro-glucagon gene and is secreted by the intestinal L cells[5]. It is now well-known that GLP-1 and its analogs cause weight loss in lean, obese, and diabetic subjects [6]. While this was originally thought to be due to slowed gastric emptying[7], it has been determined that GLP-1 crosses the blood brain barrier to decrease food intake, thus having important CNS effects[8].

Originally approved for type 2 diabetes at lower doses, 3.0mg liraglutide was approved for the treatment of obesity with or without diabetes in both the US and Europe. Our group and others have examined how liraglutide, at doses approved for diabetes (up to 1.8mg), or exenatide, currently approved for diabetes, act in the human brain to decrease appetite and cause weight loss [9–13]. Liraglutide (1.8mg) and exenatide both decreased reward-related activations[10–12], and liraglutide (1.8mg) decreased attention-related activations to food cues[11]. Notably, these studies were mostly short-term with doses being given once[9, 10], for 10 days[12], or for 17 days[11]. In the one longer-term study of 12 weeks, liraglutide (1.8mg) did not alter brain activations to food cues[12, 13], but this was also compared with insulin glargine, which might also affect activity in the brain, and not a matched placebo.

Thus, we designed a randomized, placebo-controlled, cross-over, double-blind clinical trial which aimed to determine whether and how liraglutide modified brain activations at the highest dose approved for obesity (3.0mg) as well as at 35 days, after the titration of liraglutide to reduce side effects was completed (7 days of 0.6mg, 7 days of 1.2mg, 7 days of 1.8mg, 7 days of 2.4mg and 7 days of 3.0mg). The underlying hypothesis was that we would start observing altered activation of brain areas in response to longer treatment duration which would precede the beginning of the plateauing phenomenon of body weight loss predictably seen in response to each and every anti-obesity treatment.

Methods

Twenty-eight men and women with obesity (body mass index [BMI]≥ 30 kg/m²) provided written informed consent to participate in this double-blind, cross-over, placebo-controlled trial approved by the Institutional Review Board at Beth Israel Deaconess Medical Center (BIDMC). Participants first came to the BIDMC clinical research center (CRC) for a screening visit where they were examined for inclusion/exclusion criteria. Participants were included in the study if they had a BMI≥30kg/m² and were excluded if they had taken any anti-obesity or anti-diabetes (aside from metformin) medications within 3 months or had a history of bariatric surgery, had any changes in body weight in the preceding 3 months, were unable or unwilling to participate for any reason, were unable to have an MRI, had any history of suicidality or were taking any psychiatric medications, or were unable/unwilling to take the study medication. Based on our previous study of 18 patients with DM in a cross-over, placebo-controlled, double-blind design, we can calculate that a paired comparison between treatment and placebo in this study will require only 14 participants to have 95% power to detect the same difference in parietal cortex brain activation to highly versus less desirable food cues with α=0.05.

After the screening visit, participants were randomized 1:1 to receive liraglutide or placebo for their first phase, after which they had a minimum 3 week wash-out, and then received the opposite in their second phase. Each phase consisted of six visits, each separated by a week, at which patients had anthropometry, a blood draw, physical exam/medical history, and increased their dose by 0.6mg (resulting in 0.6mg for 1 week, 1.2mg for 1 week, 1.8mg for 1 week, 2.4mg for 1 week, and 3.0mg for 1 week). At the final visit of each phase after participants had taken 3.0mg liraglutide or placebo for 6 days, patients came into the controlled CRC environment in the evening where they had their seventh dose before undergoing testing the following morning, which included (in addition to above) a fasting fMRI scan and neurocognitive testing. Subjects also rated their hunger, satiety, and the palatability of food on visual analog scales (VAS) before and after the fMRI scan. Fasting blood was drawn by venipuncture by a registered nurse. Glucose and cholesterol levels were analyzed by LabCorp, a CLIA-certified laboratory. Between visits, participants kept a detailed food record three days per week (two weekdays and one weekend day), which was discussed with and analyzed by registered dieticians.

Cognitive Testing

Cognitive testing was performed using CANTAB (Cambridge Cognition, London, UK), and tests performed were: Stop-Signal Task (SST): The SST is a classic behavioral inhibition test, which uses a staircase procedure to generate an estimate of the stop signal reaction time, which measures an individual’s ability to inhibit a prepotent response as a measure of inhibitory control. Intra-Extra Dimensional Set Shift (IED): The IED is a test of rule acquisition and reversal. It features visual discrimination and attentional set formation, as well as maintenance, shifting and flexibility of attention. Verbal Recognition Memory (VRM): The VRM test assesses immediate and delayed memory of verbal information under free recall and forced choice recognition conditions. Spatial Span (SSP): The SSP assesses general working memory capacity. Spatial Working Memory (SWM): The SWM provides a measure of spatial working memory capacity as participants are required to remember where items have appeared.

Data Analysis

Data were analyzed using the Statistical Package for Social Sciences (SPSS), v.19 and first summarized with descriptive statistics. Data for categorical variables are presented as numbers and/or percentages. Kolmogorov-Smirnov test and frequency histograms were used to check the normality of distribution of the continuous variables. Mixed models analysis or paired t-tests (for variables collected only at visit 6 and 12) were performed with liraglutide or placebo as a between subjects factor. Results which passed a threshold of p<.05 were considered significant.

fMRI protocol

Participants viewed food and non-food cues within a 3-Tesla GE MR750 scanner at BIDMC while fasting using a GE 32-channel head array coil using a protocol similar to our previous publications [11, 14–18]. First, a T1-weighted MPRAGE (Magnetization Prepared Rapid Gradient Echo) structural MR image was acquired. Next, five 7-minute gradient-echo T2-weighted echo planar images depicting blood oxygenation level-dependent (BOLD) contrast were acquired from non-contiguous near axial planes (repetition time, TR= 3.2s, echo time, TE= 21 ms, flip angle= 90, in-plane resolution= 1.9×1.9 mm, matrix size= 128×128, field of view= 24 × 24 cm, voxel bandwidth= 250kHz, slice thickness= 2.5mm). During the five runs, subjects viewed highly desirable food (e.g. desserts, burgers, fried foods), less desirable food (vegetables, fruits), and control (non-food) stimuli, presented with a block design consisting of either 5 successive highly desirable food, 5 successive less desirable food or 5 successive non-food pictures. Each block was presented in a counterbalanced order and interspersed with periods of visual fixation. As previously described, food desirability was confirmed in a previous study and visual appearance of food images was matched for visual properties[11, 14–17].

BOLD images were preprocessed using Statistical Parametric Mapping 12 (SPM12; The Wellcome Trust Centre of Neuroimaging, London, UK). In brief, for each individual subject, images were flipped, motion-corrected (realigned), normalized to an EPI template with affine registration, nonlinearly transformed, and then smoothed with a Gaussian kernel of 6mm. A general linear model (GLM) was constructed for each individual subject, using the onsets of the food or non-food image blocks with realignment parameters in 6 dimensions. The data were high-pass filtered to remove low-frequency signal drifts. The contrast images (highly desirable > less desirable food images; all food (highly and less desirable) > nonfood images) of the first-level analysis were used for the second-level group statistics. Paired t-tests were used to compare images from the liraglutide vs. placebo phases and subsequent analyses included covariates. Given the nature of fMRI analysis, activations which pass a corrected threshold of p<.05, family-wise error (FWE) corrected for multiple comparisons for the peak activation are reported.

Results

Eleven men and nine women (55±2years of age) completed visits in both the liraglutide (n=20) and placebo (n=20) conditions (Supplementary Figure 1). Weight, BMI, waist circumference, and fasting glucose were decreased and their cholesterol levels improved while on liraglutide (Table 1). Patients lost a higher percent of weight on liraglutide (−2.50±1.89%) as compared to placebo (0.19±1.93%; p<.001).

Table 1.

Metabolic effects of five weeks of liraglutide (n=20) as compared to placebo (n=20) in the context of the randomized, placebo-controlled, double-blind clinical trial.

	Placebo		Liraglutide		p-value
	Baseline	5 weeks	Baseline	5 weeks
BMI (kg/m2)	35.13 ± 1.26	35.40 ± 1.26	36.39 ± 1.59	34.67 ± 1.24	<.001
Weight (kg)	101.29 ± 3.63	101.45 ± 3.64	104.50 ± 4.31	99.66 ± 3.67	<.001
Waist circumference (cm)	117.03 ± 3.19	118.06 ± 3.38	119.62 ± 3.65	116.78 ± 3.31	0.02
Waist-to-hip ratio	1.00 ± 0.02	1.03 ± 1.35	1.02 ± 0.02	1.01 ± 1.22	0.16
Heart rate (bpm)	74.15 ± 2.48	97.93 ± 0.06	75.86 ± 2.66	97.99 ± 0.09	0.61
Systolic blood pressure (mmHg)	133.25 ± 2.80	131.15 ± 3.70	133.81 ± 2.64	127.90 ± 3.34	0.37
Diastolic blood pressure (mmHg)	82.50 ± 2.73	78.80 ± 2.26	80.38 ± 2.49	77.05 ± 2.17	0.44
Fasting glucose (mg/dL)	105.25 ± 3.32	101.75 ± 2.38	102.19 ± 3.56	91.79 ± 1.49	<.001
Fasting insulin (uIU/mL)	17.39 ± 2.34	14.73 ± 1.65	19.19 ± 4.15	15.41 ± 2.12	0.75
Total cholesterol (mg/dL)	183.73 ± 7.35	184.74 ± 7.56	176.88 ± 6.78	162.71 ± 6.86	<.001
Triglycerides (mg/dL)	110.27 ± 9.44	122.89 ± 12.98	107.69 ± 12.55	103.00 ± 7.50	0.01
HDL cholesterol (mg/dL)	55.67 ± 4.10	54.42 ± 4.11	54.06 ± 3.96	50.71 ± 3.84	0.08
VLDL cholesterol (mg/dL)	22.00 ± 1.89	24.63 ± 2.61	21.56 ± 2.52	20.58 ± 1.51	0.01
LDL cholesterol (mg/dL)	106.07 ± 5.75	105.68 ± 5.92	101.25 ± 5.47	91.41 ± 5.94	<.001
Total Grams (g/day)	2127.14 ± 196.80	2129.55 ± 247.66	2292.46 ± 251.26	1884.53 ± 144.97	0.32
Energy (kcal/day)	1901.23 ± 110.14	1890.00 ± 158.89	1829.52 ± 161.34	1640.12 ± 108.39	0.06
Total Fat (g/day)	80.70 ± 6.29	78.42 ± 7.00	74.43 ± 8.11	61.38 ± 6.18	0.03
Total Carbohydrate (g/day)	205.69 ± 15.23	204.33 ± 23.31	205.85 ± 20.50	190.57 ± 17.97	0.43
Total Protein (g/day)	82.29 ± 4.65	84.53 ± 6.57	79.13 ± 7.70	73.79 ± 5.80	0.20

Means ± standard error shown from the final visit of the appropriate phase. P-value is from a paired t-test or mixed models analysis between liraglutide and placebo phases. BMI, body mass index; DEXA, dual energy x-ray absorptiometry; HDL, high density lipoprotein; LDL, low density lipoprotein; VLDL, very low density lipoprotein.

Patients also report being able to eat less food (“How much do you think you could eat right now?” placebo: 6.29±0.66cm; liraglutide: 4.60±0.58cm; p<.006) and that it would be less pleasant to eat (“How pleasant would it be to eat right now?” placebo: 6.33±0.62cm; liraglutide: 4.87±0.61cm; p<.01) on liraglutide versus placebo, both of which remained significant when controlling for changes in weight or BMI. At five weeks, there was no difference in reported nausea between liraglutide versus placebo (“How nauseous do you feel right now?” placebo: 0.34±0.10cm; liraglutide: 0.92±0.37cm; p<.10). Both the ratings of the quantity of food they could eat and the pleasantness of eating were associated with the change in body weight (r=.39; p<.01; r=.37; p<.02; respectively).

In reported food intake, patients reported a trend towards eating fewer calories per day on liraglutide (p<.06) as well as less amount of fat in food (p<.03).

Neurocognitive testing

Patients did not show changes on intra/extra-dimensional set shift, spatial span, spatial working memory, or verbal memory neurocognitive testing outcomes when taking liraglutide versus placebo (Supplementary Table 1). On the stop signal task, patients had a great proportion of successful stops when taking liraglutide (placebo: 49±8%; liraglutide: 55±6%; p<.004; Supplementary Table 1).

Neuroimaging

No changes were observed between liraglutide and placebo conditions to food cues. When images were corrected for BMI or weight, patients presented with greater activations while on liraglutide vs. placebo in the right orbitofrontal cortex (OFC) to food as compared to non-food images (cluster size=1254mm³; peak (X,Y,Z) at 24, 32, −16; z=5.17; p<.016, Family-Wise Error [FWE] corrected for peak; Figure 1).

Figure 1.

Brain activation in right orbitofrontal cortex increases with 3.0mg liraglutide treatment to food as compared to non-food images in a whole-brain paired samples t-test when BMI is controlled (p<.05, FWE-corrected for peak). BOLD contrasts are superimposed on a T1 structural image in neurological orientation. The color bar represents voxel T value.

Discussion

This randomized, placebo-controlled, cross-over, double-blind trial demonstrates that a five-week dose escalation of liraglutide to 3.0mg improves body weight, fasting glucose and cholesterol levels, and trends toward decreased reported food intake. Although brain activations are not apparent at five weeks without corrections, when controlled for BMI or body weight, which has decreased over the preceding 35 days, the reward-related right orbitofrontal cortex (OFC) is increased in response to food as compared to non-food images. Since patients have lost a placebo-subtracted 2.7% body weight, this is likely a counter-regulatory measure which would later curb/plateau weight loss.

This is the first neuroimaging (fMRI) study that has examined liraglutide at the dose approved for obesity. We extend previous findings by our group and others to demonstrate that, at five weeks, after a dose escalation, liraglutide does not show differential activations to food cues. However, when we control for body weight or BMI, which have decreased, we observe increased activation in the right OFC. Within the literature of GLP-1 analogs, studies which tested their effects before weight loss at a single dose, at 10 days, or at 17 days, liraglutide and exenatide at the doses approved for diabetes were shown to decrease reward-related activations to food cues[10–12]. This effect disappeared at 12 weeks, where no brain activations were seen with liraglutide at the dose approved for diabetes as compared to insulin[12, 13]. Our findings fill in the gap in between these time points, after some weight loss but before the plateau of body weight loss. In uncontrolled, whole brain analyses, we have similar effects as in the 12 week study[12, 13], where we do not see any brain activations anymore. However, when we control for the change in body weight, we observe an increase in OFC activation, notably the opposite direction of the early, short-term findings, indicating the beginning of counter-regulatory changes in response to weight loss which would tend to negate weight loss and start bringing body weight up over time. The 12-week study did not control for weight or BMI changes[12, 13], and it is thus impossible to tell whether they might have had similar brain changes when controlled. The OFC is well known to increase in response to rewarding stimuli as well as to pleasantness ratings of food images[19, 20]. Fasting, motivation for food, and evaluation of food also increase OFC activation in healthy individuals without obesity[21–23]. Obese subjects have less gray and white matter volume in OFC[19] and show increased responses of OFC to food cues in imaging studies[24, 25]. Thus, we believe that this is a counter-regulatory mechanism where weight loss is causing early increases in activations of the OFC to food cues, leading to the eventual weight loss plateau observed with this and other weight loss medications.

Interestingly, we also observe an increased proportion of stops on the stop signal task (SST) in patients taking liraglutide. The SST is a task which measures inhibitory control or the ability to stop oneself from a prepotent action, such as eating high calorie foods, and is impaired in obesity[26]. While our results do not show a significant change in stop signal reaction time (SSRT), a measure of inhibitory control where a lower score indicates greater control, with liraglutide, they do trend in this direction and may be significant with a higher number of subjects. This may indicate that liraglutide improves cognitive control, but this remains to be confirmed by larger studies. No other changes in neurocognition were observed in our study. Liraglutide improves spatial memory in rodent models of Alzheimer’s disease[27–30], but we may not have observed similar changes due to the healthy nature of our study population.

This study examined the highest dose of liraglutide approved for obesity in humans, after the requisite dose escalation, which is done to decrease gastrointestinal side effects and is currently the only approved prescribing mechanism. However, by the time patients receive the highest dose, they have typically lost weight, as in our study herein. Thus, it is difficult to determine whether the effects of this highest dose would be similar to previous studies in patients with diabetes performed in the short-term. On the other hand, our results were not confounded by nausea, as they would be in single dose studies, which may impact how patients view food cues. Future studies may determine, using a single sample, the impacts of GLP-1 analogs at different doses and times, and most importantly even more potent analogues such as semaglutide, controlling for nausea and weight loss, as relevant. Additionally, future studies which examine post-prandial brain activations using milkshakes or similar food delivery designs would be helpful to confirm the findings observed with food cues.

In summary, our study is the first to examine the CNS impacts of liraglutide at the dose approved for obesity (3.0mg) and at five weeks i.e. just before any plateauing weight loss phenomenon would start. We found counter-regulatory activations of the OFC with liraglutide to food cues when we controlled for weight/BMI in a randomized, placebo-controlled, cross-over, double-blind trial. This study fills gaps in the literature on the neural impacts of GLP-1 analogs in terms of time and dose. Future studies may use combination therapies to target these counter-regulatory activations and promote further weight loss in patients taking liraglutide.