A Role for GLP-1 in Treating Hyperphagia and Obesity

Abstract

Obesity is a chronic recurring disease whose prevalence has almost tripled over the past 40 years. In individuals with obesity, there is significant increased risk of morbidity and mortality, along with decreased quality of life. Increased obesity prevalence results, at least partly, from the increased global food supply that provides ubiquitous access to tasty, energy-dense foods. These hedonic foods and the nonfood cues that through association become reward predictive cues activate brain appetitive control circuits that drive hyperphagia and weight gain by enhancing food-seeking, motivation, and reward. Behavioral therapy (diet and lifestyle modifications) is the recommended initial treatment for obesity, yet it often fails to achieve meaningful weight loss. Furthermore, those who lose weight regain it over time through biological regulation. The need to effectively treat the pathophysiology of obesity thus centers on biologically based approaches such as bariatric surgery and more recently developed drug therapies. This review highlights neurobiological aspects relevant to obesity causation and treatment by emphasizing the common aspects of the feeding-inhibitory effects of multiple signals. We focus on glucagon like peptide-1 receptor (GLP-1R) signaling as a promising obesity treatment target by discussing the activation of intestinal- and brain-derived GLP-1 and GLP-1R expressing central nervous system circuits resulting from normal eating, bariatric surgery, and GLP-1R agonist drug therapy. Given the increased availability of energy-dense foods and frequent encounters with cues that drive hyperphagia, this review also describes how bariatric surgery and GLP-1R agonist therapies influence food reward and the motivational drive to overeat.

A Role for GLP-1 in Treating Hyperphagia and Obesity

Obesity’s Prevalence Is High and Its Negative Impact on Human Health Is Profound

Between 1960 and 1980, obesity prevalence was moderate, ranging between 13% and 15% (1). Over the next 4 decades, obesity prevalence has nearly tripled to a level of 42.4%, affecting more than 100 million US adults (2). An additional 33% of US adults are overweight and are at risk of further weight gain (3), and even larger increases in obesity prevalence are predicted by 2030 (4). There are numerous pathophysiologic consequences of the excessive adiposity that characterizes obesity and thus the rapid and dramatic rise in obesity prevalence has significantly increased the risk of morbidity and mortality in affected individuals (3). For example, as body weight or body mass index (BMI) significantly increases, so do the risks for type 2 diabetes, cardiovascular disease, hypertension, nonalcoholic fatty liver disease, gall bladder disease, osteoarthritis, and sleep apnea (5). Further, elevated BMI is an established risk factor for several causes of death, including ischemic heart disease, stroke, and various cancers (6). Obesity was the second strongest predictor of hospitalization and the need for mechanical ventilation after old age in early findings on the COVID-19 pandemic (7). Beyond this, obesity reduces the quality of life through its pathophysiologic sequelae including but not limited to its effects on: joint pain that limits ambulation, sleep apnea that reduces sleep quality, stigmatization that limits social engagement, and also serves as a risk factor for depression (8) and as a contributor to cognitive decline (9, 10).

The recent designation of obesity as a chronic and recurring disease by national/regional medical bodies (11) and scientific societies (3) has increased attention to the need to effectively treat individuals with medically significant obesity and thereby has drawn attention to the limitations of behavior-based obesity therapy. Behavior therapy, also known as diet and exercise, or lifestyle therapy is the recommended initial therapy for improving weight and cardiovascular disease risk in patients with obesity (12). Although behavioral therapy results in an average of 5% to 8% loss of initial weight after 4 to 6 months of high-intensity, supervised treatment, it is clear that a substantial percentage (35%-50%) of patients undergoing behavior therapy fail to lose any significant weight (13, 14). Even more troublesome is that those who do lose weight regain much of the lost weight over time (15).

These profound limitations of behavioral therapy treatment outcome are explained by powerful compensatory neuroendocrine processes that are engaged by a vigilant energy balance control system that senses and reacts to the biological correlates of reduced energy intake. A variety of studies show that diet-driven weight loss triggers a perfect storm that includes reduced energy expenditure, reduced levels of feeding inhibitory gastrointestinal hormones, elevated levels of the orexigenic hormone ghrelin, increased appetite and hunger ratings, and reduced postprandial fullness (3, 16-18). In short, in response to weight or energy loss, biological regulatory processes, along with other actions, stimulate food intake and thereby effectively diminishing a patient’s diet adherence (19, 20). Polidori et al (19). conclude that feedback control-driven food intake appears to play an even larger role in reducing the efficacy of behavior therapy than energy expenditure adaptations, generally considered as the primary explanation for its limited weight loss and weight regain (21). Importantly, increased attention to the limitations of behavior therapy have provided: (1) compelling support for the recurring and chronic nature of obesity; (2) increased attention on the value of biologically based therapies (i.e., bariatric surgery and drugs) whose actions oppose the feedback regulatory driven increased hunger and reduced fullness resulting from dieting-driven weight loss; and (3) recognition that the development of new and more efficacious biologically based therapies will arise from an increased focus on the gastrointestinal (GI) and central nervous system (CNS) signals and circuits that mediate the motivational pull of the food environment as well as the homeostatic push toward food. These efforts underscore the need for more mechanistic studies to help better understand the potent food intake inhibitory signals arising from the GI tract and how they affect the central neural control of food intake.

Increased Obesity Prevalence Results at Least Partly From an Increased Food Supply That Provides Easy Access to Energy-Dense Foods

Obesity prevalence has increased dramatically over the past 40 years, concomitant with changes in the global food system that have resulted in dramatic increases in access to tasty, low-cost, energy dense, ultraprocessed foods that in addition to their intrinsic appeal are marketed aggressively (22-24). Vandevijvere et al (22) correlated changes in available food energy supply with changes in body weight between 1971 and 2010 for 69 countries and concluded that the increased food supply is sufficient to explain the increased average population body weight. Hall et al (25) used a validated mathematical model of human energy expenditure to examine whether increases in average food intake underly the observed increase of average adult body weight in the United States between 1974 and 2005. Their analysis showed that over this 30-year period, average daily energy intake increased by ~600 kcal (from ~2100 kcal/d to ~2700 kcal/d) when adjusted for food waste.

Humans typically eat in discrete bouts and whether such events are labeled meals or snacks or more generically as feeding occasions, there are 2 principal variables that determine daily caloric intake, the size and frequency of these bouts. These parameters are also quite relevant to assessing the efficacy of biologically based obesity treatments whose weight loss effects derive from reducing food intake, discussed later. There is evidence that both meal size and meal frequency have increased over the past 30 to 40 years. Popkin and Duffey (26) conclude that US adults as well as children are eating more frequently over the day than they had 30 years ago. Using large data sets covering 3 time intervals between 1977 and 2006, their study found that (1) the number of feeding bouts per day increased, (2) the time between feeding occasions decreased (frequency increased), and (3) energy intake from snacking increased for children and adults. One manifestation of the combined effects of increased food supply, reduced cost of food, increased access to preprepared food, and more aggressive food marketing has been notable increases in portion size. Many studies have examined whether increased portion size affects daily caloric intake with general agreement about a positive relationship between the two. However, extrapolating from many of these studies is constrained by study design features like limiting assessments to a single meal of varying size that is consumed in a laboratory setting (27). A study by Kelly et al (28) stands out because it used a randomized, within-subject, crossover design where subjects were presented in 2 phases with “standard” or “large” food portions of the same foods and beverages on 4 occasions per day over a 4-day period with all meals presented in a residential setting. The results are clear and compelling. In response to large portions, men and women significantly increased caloric intake by 17% and 10%, respectively, relative to standard portions. Importantly, mean daily intake did not significantly differ across study days for either portion condition indicating that no compensation was made for the increased energy/food consumed over the 16 feeding occasions examined in the study. Of direct relevance, body weight of individuals in the large portion condition increased significantly (0.9 kg [SD 1.1]) for men and (0.6 kg [SD 0.6]) for women with no weight change observed in the standard portion phase of the study.

The Sensory Features of Tasty, Calorically Dense Foods and Cues That Predict Them Activate the Brain Appetitive Control Circuits

The association between increases in the food supply and increases in human obesity prevalence is widely recognized. A key, yet important missing, element in this discussion is defining the biological factors that mediate the increased food intake that results from increases in the food supply. It is worth mentioning that there are dramatic differences between the current food environment that offers ubiquitous access to hedonically compelling, high-calorie foods and snacks and food environments of the past that were marked by higher probability access to lower energy dense options and more variable access to energy-dense foods. The latter food environment persisted for the majority of human evolution and can still be modeled to some degree by the study of extant hunter-gatherer societies and the foraging strategies they use (29). Physiological control systems evolved to ensure the acquisition of energy resources from the environment (i.e., food) that are sufficient to maintain energy balance; this regulatory control is an absolute requirement for both individual survival and reproductive success. Therefore, under the constraints imposed by living in lower energy-dense environments where access to energy-dense food resources was rare, opportunistic feeding was a useful strategy as excess consumed calories were stored in adipose tissue and catabolized as needed in future. Observations from hunter-gatherer societies reveal that on occasions that an animal is killed or a beehive taken, that the energy-dense fat extracted from the animal or the energy-rich honey from the beehive was quickly consumed by taking meals of enormous size (29). In the current energy-rich food environment, human feeding continues to be opportunistic and remains driven by hedonic and calorie predicting sensory features such as the fat and sugar components of foods as well as nonfood, reward predictive cues. In the modern context however, the consequence of chronic, excess feeding is the continued expansion of adipose mass until at some point, for most individuals, the pathophysiologic consequences of obesity become manifest.

Hedonic feeding is typically characterized as the motivational pull of reward predictive cues while homeostatic feeding is described by the push of detected metabolic need toward food. Rather than viewing these as distinct motivations to eat with separable neurologies (30), it is more practical to view them as overlapping and interacting (31, 32). Metabolic need does not operate in 2 states but as a continuum. In addition, foods that humans describe as sweet or fatty remain attractive in states of greater as well as lesser hunger as witnessed by the interest in consuming energy dense foods with those sensory features at the end of meals.

The perspective taken here and supported by a broad range of scholarship is that obesity results, at least in part, from the hyperphagia that is driven by the activation of appetitive behavior-generating brain circuits by food environment-related stimuli. The ubiquitous energy-dense food stimuli and associated nonfood, reward predictive cues are among the key food intake-promoting stimuli in focus in this review. Because it is clear that human obesity is genetically associated (33, 34) and that roughly one-third of US adults have BMIs below 25, increased obesity prevalence appears most likely as a consequence of a gene by environment interaction (35). There are some who emphasize that modern environments and associated changes in the nature of human work result in reduced energy output and expenditure and that reduced energy expenditure rather than increased energy intake should be viewed is the primary driver of obesity prevalence (36). Support for this perspective is, however, limited (29, 37, 38). The paragraphs that follow highlight data supporting the neurobiological mediation of food environment-driven hyperphagia. The discussion identifies causative mechanisms of obesity and are also of value in assessing the efficacy of biologically based obesity treatments whose actions oppose the stimulating properties of energy-dense foods and reward predictive cues on food intake.

Simply viewing photographs of energy-dense foods (i.e., reward predictive cues) activates appetitive and reward circuits in various regions of the human brain compared with viewing pictures of nonenergy dense foods. These results derive from a frequently employed paradigm using functional magnetic resonance imaging that evaluates blood oxygen level–dependent (BOLD) signals and uses energy status and obese status (BMI) as key variables (39-41). Across-subject BMI viewing high-calorie food images in a modest deprivation state increased BOLD signals (a proxy for increased neural activity) in the dopamine-rich midbrain nuclei, the striatal projections of these midbrain neurons, insula, aspects of hypothalamus and thalamus, and, in some studies, orbitofrontal cortex, prefrontal cortex, and hippocampus. When examined in the sated or euglycemic state, clear differences were observed in the responses of subjects with obesity and without obesity. For subjects with obesity, the same appetitive control brain areas were activated by viewing high-energy food pictures as was observed when tested under modest deprivation. For sated subjects without obesity, however, the pattern differed strikingly: prefrontal cortex and anterior cingulate cortex, areas associated with behavior inhibitory control, were activated in those subjects. A wealth of preclinical and clinical studies show that neurons in the regions activated by viewing images of high-calorie food are functionally associated with the activation of goal-directed appetitive behaviors including food seeking and feeding motivation and in response to food reward or food reward prediction or anticipation (42, 43).

The evidence that reward predictive cues, like food pictures, trigger neural activation of behavior generating appetitive control circuits and increases dopamine signaling is extensive (44, 45). A variety of studies show that tasting hedonic energy-dense food rapidly elevates dopamine release in the striatum where dopamine receptor activation stimulates behavior. The use of positron emission tomography combined with application of the dopamine receptor antagonist [¹¹C]-raclopride revealed that tasting one’s favorite food releases endogenous dopamine in the human striatum (46). Tasting palatable foods also increased dopamine release from the rat striatum using in vivo brain microdialysis (47). In preclinical studies, animals tasting sucrose or corn oil, 2 of the of most common ingredients in hedonic, energy-dense foods and snacks, significantly increased striatal dopamine release measured by fast scan cyclic voltammetry and microdialysis respectively (48, 49). The robust BOLD activation of brain appetitive control regions by food predictive cues and the release of dopamine from midbrain neurons to striatal targets by taste or other reward-predictive cues demonstrate that food and food cues activate anatomically distributed brain circuits (50-52). Input from sensory cues that are predictive of the postabsorptive availability of calorie-rich, energetically useful metabolic fuels is sufficient to activate neural circuits that drive goal-directed, food-motivated behavior (53, 54). Visual cues that through training were associated with tasting a pleasant glucose stimulus increased BOLD signal activation in midbrain dopamine regions, striatum, posterior dorsal amygdala, and orbitofrontal cortex of humans (43). Similarly, in rodent models, association of visual or auditory cues with oral sucrose access transformed the response from what were neutral environmental stimuli into reward-predictive cues that released dopamine and potently drove feeding behavior (55, 56). Not surprisingly, food marketers routinely harness the neural- and behavior-inducing power of food predictive cues in their recurring presentation of brand logos, in restaurant signage, food packaging, and advertising and also in the visual-spatial cues arising from the uniformity of design features (furniture, color, shapes, odors) of chain restaurants that through repeated association with ingesting hedonic energy-dense food become neural activating, reward predictive cues 57-61). Through repeated association with ingesting energy-dense food, these features become reward predictive cues.

Identifying Signals in the GI Tract and Brain That Inhibit Feeding Provide Clues for Obesity Treatment

To design better antiobesity treatments, it is logical to focus in detail on the normal physiology of food intake inhibition. An obvious and appropriate starting place is to focus on the GI signals released by food ingestion because the actions of these signals on the central nervous system constitute the major determinants of ending meals (satiation) (62). Such signals are referred to as satiation signals and include gastric mechanosensation, intestinal caloric detection, and intestinal peptide hormone release (e.g., glucagon like peptide-1 [GLP-1], peptide YY [PYY_3-36], cholecystokinin) (63). The discussion that follows highlights the contribution of 1 of these signals, GLP-1. It is imperative, however, to appreciate that while manipulating levels of an individual signal like GLP-1 can have an effect on feeding and weight control, it is the integrated action of various GI, systemic, and central signals that collectively mediate the endogenous control of food intake inhibition. There are 5 tissues that express the preproglucagon gene, which encodes GLP-1: (1) pancreatic alpha cells with glucagon as its primary posttranslational product; (2) the intestinal enteroendocrine cells; (3) NTS neurons with GLP-1 and GLP-2 as primary gene products; and (4) taste cells and (5) and the olfactory bulb interneurons with GLP-1 as gene product (64, 65). GLP-1 is released from intestinal enteroendocrine cells in response to the direct calorie stimulation resulting from digestive processing of ingested food. GLP-1 acts in a paracrine fashion to excite GLP-1 receptor (GLP-1R) expressing vagal afferent neurons whose cell bodies are located in the nodose ganglion (66). Central axon projections of these activated vagal afferent neurons release glutamate to excite postsynaptic, nucleus tractus solitarius (NTS) neurons of particular phenotypes including proglucagonergic (PPG; GLP-1 positive), catecholaminergic (tyrosine hydroxylase and dopamine beta hydroxylase [DBH positive]), and prolactin-releasing peptide expressing neurons (67-70) as well as adjacent glia/astrocytes (71). The axons of NTS^DBH and NTS^PPG neurons project monosynaptically to anatomically distributed CNS targets to affect the excitability of neurons in identified nuclei including the pontine parabrachial nucleus (PBN), midbrain ventral tegmental area (VTA), lateral dorsal tegmental nucleus (LDTg), hypothalamic paraventricular nucleus, lateral hypothalamic area, paraventricular thalamic nucleus (PVT), lateral septum (LS), central nucleus of the amygdala, bed nucleus stria terminalis, and nucleus accumbens (NAc, ventral striatum) (72, 73). Vagal activation of NTS neurons also affects the excitability of neurons in other locations through polysynaptic connections (e.g., ventral hippocampal neurons (74)) that also contribute to food intake inhibitory control (75).

A wealth of evidence supports the contribution of particular neurons within these anatomically distributed nuclei to the inhibitory control of feeding including results from experiments using an array of sophisticated behavioral methods in conjunction with modern neural techniques, such as parenchymal delivery of receptor ligands, selective neurotoxic lesions, chemo- and optogenetic excitation and inhibition, fiber photometry, and receptor knockdown (68, 76-82). GLP-1R expressing neurons are highlighted here for their relevance to the endogenous control of feeding inhibition as well as their roles in bariatric and drug treatments for obesity. Multiple, anatomically distributed, GLP-1R expressing neurons receive endogenously released GLP-1 and glutamate from the axon terminals of vagally activated NTS^PPG neurons (83-85). Direct intraparenchymal delivery of low doses of the GLP-1R agonist exendin-4 (Ex₄) clearly and consistently reduces meal size and food intake when applied directly to a surprisingly large number of GLP-1R expressing CNS nuclei including NTS, PBN, LDTg, VTA, hypothalamic paraventricular nucleus, lateral hypothalamic area, PVT, LS, bed nucleus stria terminalis, or NAc (86-93). Complementing the feeding suppressive effects of agonist-induced increases in GLP-1R signaling, are results showing increased food intake and meal size in response to reducing endogenous GLP-1R signaling with intraparenchymal delivery of GLP-1R antagonist, Ex_9-39, into the same nuclei. The intake stimulatory effect of parenchymal antagonist injection can be interpreted as an action opposing the increased CNS GLP-1R signaling endogenously driven by food ingestion resulting from vagal activation of NTS PPG neurons and release of GLP-1 from their axonal terminals, highlighting the physiological relevance of this system.

Evaluating the Efficacy of Biologically Based Obesity Treatments and Their Actions on Feeding Behavior

Given the importance of treating the chronic, recurring disease of human obesity and the major limitations of behavior therapy already noted, an emphasis on applying biologically based treatments to inhibit the feeding stimulatory effects of the food environment and to also activate endogenous satiation systems is warranted. The effectiveness of these biologically based obesity therapies derives from their ability to emulate some of the signals and circuits that endogenously inhibit food intake and thereby control body weight. Given the hyperphagic drive of environmental cues to be effective, obesity treatments must also attenuate the activation of appetitive brains circuits by reward predictive cues arising from the modern food environment. Described next is evidence addressing the effects of drug and bariatric surgery on these 2 neurobiologically controlled functions as well as a discussion of the mechanisms that contribute to their food intake inhibitory actions. These mechanisms include the induction of visceral malaise, the amplification of the effects of satiation signals, and the reduction of appetitive behavior.

Obesity drug therapy

In the United States, there are now 4 FDA-approved drugs that can be prescribed for obesity therapy; a fifth, lorcaserin, was recently withdrawn. Two are monotherapies; 1 is phentermine, a norepinephrine-releasing agent (and to a lesser extent a releaser of dopamine and serotonin) and the other an agonist of the GLP-1R (liraglutide, Saxenda). Two are combination therapies; naltrexone/bupropion links an opiate receptor antagonist with a dopamine/norepinephrine reuptake inhibitor (Contrave) and phentermine/topiramate combines a norepinephrine/dopamine/serotonin-releasing agent with a GABA-A receptor agonist (Qsymia). Given the role of brain systems in mediating the appetitive drive to eat along with the inhibition of feeding, it is noteworthy that the CNS is the site of action for all of these drugs. The focus here will be on GLP-1R agonist therapy as roles for endogenous GLP-1 release in the intestine and in brain and the involvement of GLP-1R signaling at multiple CNS expressing nuclei in the endogenous control of feeding inhibition have already been introduced. Space limits a consideration of the other approved drugs. It is worth mentioning that the pharmaceutical industry’s focus on increasing GLP-1R signaling as a treatment strategy for obesity developed in parallel with its development of treatments to enhance ß-cell function for type 2 diabetes. Data support roles for peripheral GLP-1R (94-97) as well as for CNS GLP-1R agonism in feeding and weight reduction. Therefore, it became important to discern the relative roles of peripheral and central receptor populations in mediating the effects of systemic delivered longer acting GLP-1R agonists. Loss of function experiments targeting peripheral GLP-1R (complete subdiaphragmatic vagal deafferentation sparing 50% of vagal efferents) or of central GLP-1R antagonism (widespread brain delivery of GLP-1R antagonist, Ex_9-39,) revealed that, although the peripheral receptor population contributes to anorexic and weight-reducing actions of the higher agonist doses, CNS GLP-1Rs mediate of the food intake and body weight reducing effects of a larger range of liraglutide and Ex₄ doses (98). These findings have been confirmed by others using other methods (99, 100).

Multiple studies in human subjects, supported by parallel data from rodents, show that chronic, systemic delivery of liraglutide reduces body weight with its primary action on food intake inhibition. For example, subjects with obesity who lost ~5% of body weight during a tightly monitored, low-calorie diet run-in period lost on average an additional 6.2% of body weight and maintained the combined weight loss during the 56 weeks of daily systemic delivery of liraglutide (3 mg) (101). This study confirmed 2 things: that GLP-1R agonist drug therapy reduces body weight and also enhances weight loss maintenance in contrast to the weight regain typically observed with behavior therapy without drug, noted previously. Consistent with results from behavior therapy, once drug was terminated weight was regained; see also (102). Another study by the same group (103) examined the effect of 52 weeks of liraglutide treatment combined with intense and highly structured behavior therapy and reported an 11.8% weight loss. The drug-treated group reported a significant reduction in hunger and a significant increase in fullness. In contrast, subjects receiving intense behavioral therapy without liraglutide showed decreased fullness, no change in hunger, and lesser weight loss. van Can et al (104) showed that the liraglutide-induced reduction of food intake in obese subjects was driven by decreased hunger and increased fullness and satiety. Although not yet approved for obesity, semaglutide, the Novo Nordisk follow-up compound to liraglutide, shares liraglutide’s profile of action: food intake inhibition, reduced hunger, increased fullness, and satiety effects (105). Semaglutide’s weight-reducing effects are significantly greater than liraglutide; notably and importantly, the weight loss achieved with neither drug can be explained by increased energy expenditure (105-107).

As previously mentioned, GLP-1 is 1 of a variety of GI signals released by food ingestion that, in addition to other non-GI signals (e.g., leptin, melanocortin, oxytocin) contribute collectively to the endogenous inhibitory control of food intake. For these reasons, there is a compelling logic for future obesity drugs to use combination therapies to better emulate the combined action of multiple gut or other signals and thereby to hopefully enhance effects on feeding and weight (108). Such an example is tirzepatide (Lilly), a dual GLP-1 and glucose-dependent insulinotropic polypeptide (GIP) receptor agonist, that reduces food intake and appetite and importantly offers weight loss superior to liraglutide (109). Although GIP was known to influence glycemia, until recently there was a lack of clarity about whether GIP-R agonism exerted a food intake inhibitory effect. An important study (110) revealed that GIP-R is expressed in multiple hypothalamic cell groups, that it is expressed in both GLP-1R+ as well as in GLP–1R– neurons, and that chemogenetically activating theses CNS receptors reduces food intake. The mechanistic basis of the combinatorial actions of GLP-1R and GIP-R and its enhanced effects on food intake and weight loss relative to liraglutide will require additional study (111).

What processes account for the feeding inhibitory and body weight reducing effects of GLP-1R agonists?

An evaluation of the large literature on GLP-1R agonist treatment reveals that weight loss for this drug class is driven by reductions in caloric intake, not by elevated energy expenditure, and that the feeding inhibition is attributable to 3 processes: visceral malaise, increased satiation/fullness, and reduced appetitive drive.

Visceral malaise

The most commonly reported adverse events from GLP-1R agonist-treatment are GI events, in particular, nausea and vomiting, referred to here as visceral malaise. In an attempt to mitigate these effects, study designs include up-titration involving incrementing dose level in multiple, equivalent dose steps over weeks. Up-titration is effective in reducing adverse GI events and sustaining a high percentage of subjects participating in clinical trials (e.g., 4% withdrew (102)). For example, in studies that examine subjects treated liraglutide 3.0 mg, approximately 40% of subjects reported nausea early in the dose-escalation phase but by 40 weeks, the percentage declined to approximately 12% (102). In rodent models, visceral malaise induced by the administration of lithium chloride (LiCl), cisplatin chemotherapy or other agents known to promote nausea and emesis in humans reduces food intake (112) and is accompanied by pica, elevated consumption of kaolin clay, a broadly accepted proxy of visceral malaise in nonvomiting rats. Peripheral administration of LiCl induces the expression of the early immediate gene cFos, a marker of neuronal activity, in the lateral division of PBN, NTS, and area postrema (113) in the hindbrain. This expression pattern overlaps with the cFos activity pattern induced by centrally administered native GLP-1 (113). Although such comparisons between the behavioral and neural effects of GLP-1R agonism and LiCl are suggestive, other data challenge the perspective that visceral malaise is the principal mediator of the effects of GLP-1R agonism on feeding and body weight. Notably, GLP-1R signaling in the PVT, LDtg, ventral hippocampus, LS, VTA, and NAcc reduces food intake but does not increase pica (75, 86-88, 93). Also notable is that GLP-1R agonist-treated subjects who do not report GI adverse events lose significantly more weight than controls (102).

Increased satiation

Satiation describes the process integrated by the brain that ends a meal and determines it size (114). Satiation signals such as gastric distension and the release of GI hormones like GLP-1 activate the afferent vagus nerve that in turn excites NTS^PPG and NTS^DBH neurons. NTS^PPG neurons such as PPG-positive intestinal enteroendocrine cells release GLP-1. Vagal input-activated NTS^PPG neurons release endogenous GLP-1 from their axons that terminate on multiple CNS GLP-1R expressing sites. These GLP-1R signaling elements can be considered a necessary part of endogenous satiation control, resulting in meal termination. Of direct interest, GLP-1R agonist (Ex₄) applied exogenously to each of the many recently named CNS sites decreases meal size and reduces food intake (86-88, 90-93). Similarly, systemically delivered GLP-1R agonist acts on brain GLP-1R to reduce meal size as well as overall food intake (115). Consistent with this, food intake reduction following energy-dense food consumption is attenuated by GLP-1R antagonism (Ex_9-39) targeted to individual CNS GLP-1R expressing sites (87, 115).

Reduced appetitive drive

The activation of appetitive behavior control brain areas such as the striatum, insula, and orbitofrontal cortex by viewing reward-predictive, energy-dense food pictures was significantly reduced by GLP-1R agonist treatment in subjects with obesity or diabetes (116). Relatedly, the functional magnetic resonance imaging BOLD signal activation of appetitive control areas observed in response to viewing tasty food pictures was attenuated by consuming a meal (resulting in increased GLP-1 release and increased CNS GLP-1R signaling) and was restored by systemic GLP-1R antagonist, Ex_9-39 treatment (117). Consistent with these data, results from numerous animal studies show that GLP-1R signaling reduces appetitive processes including food seeking, feeding motivation, and food reward. The progressive ratio test that requires rodents to work harder (increase the number of lever presses) to obtain each successive sugar pellet is used to assess feeding motivation and reward value. In multiple studies examining the effects of delivering GLP-1R agonist to individual GLP-1R expressing sites, rats significantly reduced the number of lever presses that they were willing to make to obtain sugar pellets compared with their feeding motivation under vehicle treatment (79, 87-89, 118, 119). Complementary results from experiments that genetically reduce GLP-1R expression demonstrate that various central GLP-1R populations contribute to the motivation to work for food. Rats with bilateral, partial knockdown of GLP-1R in lateral hypothalamus (118) or in NTS (79) dramatically increased feeding motivation/reward value; they made more lever presses and obtained greater numbers of sucrose pellets when GLP-1R expression was reduced. Reward predictive cues associated with hedonic, energy-dense food like sugar pellets increase dopamine release in the nucleus accumbens providing an additional metric of reward value. Cue-evoked dopamine activity was reduced in an Ex₄ dose-related manner following lateral ventricle injection that provided GLP-1R agonist to multiple GLP-1R sites in rat brain (82). Collectively, data are consistent with the hypothesis that CNS GLP-1R signaling inhibits the motivation for palatable food and its reward value. Conditioned place preference testing assesses food seeking by measuring the time animals spend in the presence of cues that though training predict access to energy-dense food. Over a series of training sessions, 1 set of environmental cues was associated with access to and consumption of high-sugar/high-fat food, whereas a different set of environmental cues was linked to no food access or low-fat chow access. Before training rats spent equivalent time in the 2 different cue environments but when place preference was reexamined after training, without food present, rats spent 20% to 30% more time in the environment that had been associated with energy-dense food access than the other environment. Strikingly, when a low dose of exenatide was injected into 1 of many different GLP-1R-expressing CNS sites, rats no longer sought out the high-energy food-associated environment (88, 89), indicating that local elevations in GLP-1R signaling reduce food seeking. Collectively, the data reviewed in this section are consistent with the hypothesis that CNS GLP-1R signaling reduces appetitive drive.

Bariatric surgical treatment of obesity

In comparison to drug treatments, bariatric surgical approaches for treating obesity have existed for far longer, beginning in the late 1960s (120), have been used to treat far greater numbers of humans with obesity (US 2014 annual estimate ~200 000 bariatric procedures (121)) and produce greater weight loss. In a recent analysis of 1738 subjects who underwent Roux-en-Y gastric bypass (RYGB) surgery, the 7-year mean weight loss was 28.4% with a weight regain of 3.9% after 3 years (122). The authors described considerable variability in the overall mean weight loss but noted that 75% of RYGB patients maintained a 20% weight loss over the 7-year study period. In a frequently cited Swedish study, average total weight loss 10 years after RYGB was 18% (123). Sleeve gastrectomy, a procedure gaining in popularity produces similar weight loss (124).

The body weight loss induced by bariatric surgery results from a combination of reduced energy intake and alterations in metabolic variables (124, 125). RYGB-induced weight loss is not due to caloric malabsorption (126). Mechanical restriction also does not explain the reduced food intake observed after bariatric surgery because ingested food arriving in the small residual gastric pouch moves directly to the jejunum due to the absence of the pyloric sphincter. There does not appear to be any consistent evidence for continuous visceral malaise following bariatric surgery. Although visceral malaise is reported immediately after bariatric surgery (127), beyond the first few weeks after surgery, the majority of patients do not report visceral malaise. The inhibition of food intake produced by bariatric surgery is associated with reduced hunger and increased fullness/satiety, a pattern similar to that observed with GLP-1R agonist drug therapy. Six weeks postoperatively the meal size of bariatric surgery patients was 42% of their preoperative values and while meal size increased after 1 and 2 years, it was still significantly lower than preoperative meal sizes by 57% and 66%, respectively (128). When food preferences are assessed with an ad libitum buffet meal test, there were no changes in food preferences following bariatric surgery (129). This finding has an additional relevance because it is inconsistent with a view that bariatric surgery facilitates conditioned food aversions that would alter food preferences adversely.

Bariatric surgery accelerates the arrival of ingesta to the distal bowel, triggering an earlier and exaggerated secretion of GLP-1 and PYY_3-36 (130, 131). There is general agreement that a principal driver of the reduced food intake, increased fullness, and decreased hunger effects of bariatric surgery are elevated levels of multiple GI hormonal signals whose effects on vagal afferents and specific CNS targets have been described previously. A variety of studies support for this view. Combining GLP-1R antagonism with DPP-4 inhibitor mediated lowering of PYY_3–36 increased food intake by ~20% in RYGB patients, whereas neither GLP-1R blockade alone nor DPP-4 inhibition alone affected food intake (132). Comparable results were observed in RYGB patients and in a rat model of RYGB where octreotide, a treatment that suppressed postprandial plasma levels of both PYY_3-36, and GLP-1, increased food intake of RYGB-treated subjects compared to control treatment (133, 134). Further, levels of postprandial PYY_3-36 and GLP-1 were reduced in surgical patients who had a poor weight loss response compared with patients whose weight loss was judged as good (134). The same study showed that as soon as 2 days after surgery when significant reductions in assessed hunger and increases in fullness were observed, levels of GLP-1 and PYY_3-36 were significantly elevated.

Reduced appetitive drive provides a mediating mechanism for the observed reduction in food intake after RYGB in addition to the increased fullness/satiation already mentioned. A progressive ratio task to assess feeding motivation/reward value in humans before and within 8 weeks of surgery revealed a 50% reduction in the amount of effort (mouse clicks) that subjects made to obtain a sugary food reward (135). Complementary data showed that octreotide-treatment reversed the reduced food motivation in RYGB patients; octreotide increased the reward value of sugary treats and the motivation to obtain these snacks compared to saline-treated patients (136). Furthermore, RYGB reduced the average BOLD signal across 5 areas of interest (ventral and dorsal striatum, insula, orbitofrontal cortex, and amygdala) compared with responses of BMI-matched controls (137). RYGB also reduced BOLD signal to viewing calorie-dense food pictures in the striatum and in the reduced BOLD signal to tasting a chocolate milkshake in the insula (138). Treatment with the GLP-1R antagonist, Ex_9-39, attenuated each of these effects of RYGB, indicating a role for GLP-1R in mediating BOLD responses to visual and taste/olfactory cues. Thus, there is collective support for the hypothesis that bariatric surgery reduces appetitive drive feeding motivation, and food reward, and that these alterations likely derive from the surgically induced elevations in GLP-1 and PYY_3-36 and increases in CNS GLP-1R signaling.

Comparing treatments and acknowledging the differences in their efficacy

There are notable similarities in the qualitative effects of bariatric surgical and GLP-1R agonist drug therapies for obesity. Their body weight-reducing effects involve reduced energy intake and that outcome is mediated by increased fullness as well as reduced appetitive drive. Increased GLP-1R signaling is a contributor to these effects in both treatment types.

The most notable difference between bariatric surgery and the current GLP-1R agonist therapy, liraglutide, is in the magnitude of weight loss achieved: ~7% to 11% for liraglutide and ~18% to 25% for bariatric surgery. This difference is referred to as the treatment efficacy gap. It is not surprising that there are differences between the outcomes of a drug treatment affecting a single receptor and surgical therapies that impact multiple gut signals of relevance to feeding control as well as control of metabolic endpoints (e.g., bile acids). In addition to surgically induced elevations in GLP-1, it is clear that elevated postprandial PYY_3-36 is a necessary contributor to the feeding suppression resulting from bariatric surgery, as shown clearly by antagonist studies (132-134). GIP and other gut hormone levels are also affected by bariatric surgery; the beneficial impact on weight loss of combining GIP and GLP-1 as a dual agonist drug therapy for obesity was considered previously.

This efficacy gap between bariatric surgery and GLP-1R monotherapy is large and drives the development of new drugs (108). For example, the weight-reducing effect of tirzapatide is significantly greater than that of liraglutide (109). Once approved for obesity therapy, combination therapies of 2 or more agents like tirzapatide and other drugs underdevelopment including semaglutide combined with another feeding inhibitory GI or other signal will likely narrow the efficacy gap between drug and surgical therapies. Obviously, predictions about continued increases in weight loss efficacy are fraught with limitations and some new compounds heralded for their efficacy may ultimately disappoint or be withdrawn.

Individual differences

There are individual differences in weight loss outcome for each of the currently available therapies. An important challenge going forward is to determine whether there are distinguishable subtypes of human obesity and whether there are measurable parameters that can be used to reliably identify such subtypes. If and when such data materialize, there could be an opportunity to define obesity treatments that yield maximal subgroup-specific efficacy 139-141), raising the possibility for precision medicine approaches in the future.

Conclusions

The case for applying biologically based therapy to treat the chronic, recurring disease of obesity through the alteration of normal physiology with surgery or chronic drug delivery is strong. Indeed, this review emphasizes how these biologically based treatments inhibit feeding as well as the feeding stimulatory influence of environment-based cues on behavior. The increase in the food supply and its aggressive marketing has ensured greater contact with these potent environmental stimuli. Connections were drawn between signals, such as GLP-1, and the treatment-based effects of elevated endogenous levels of GLP-1 and PYY_3-36 and other GI signals via surgery or obesity pharmacotherapies. Support for the inhibitory action of drug and bariatric surgery therapies on the appetitive drive to eat exerted by contact with reward predictive cues was considered and provides a useful, even if partial, explanation for treatment efficacy.

Bariatric surgery is at present the most efficacious treatment for treating severe obesity (121). It is important to acknowledge that relative to bariatric surgery, obesity drug therapy is in its relative infancy. Although the efficacy of “first-generation drugs” is substantially less than that of surgery, there are encouraging signs that the efficacy of second and successive generations of obesity drugs will improve and that with such improvements the efficacy gap between drugs and surgery will narrow. For obesity drug therapy to represent a far greater share of the treatment options for human obesity than at present, reducing the efficacy gap must be combined with other efforts. For example, attention to a range of important infrastructural regulatory issues such as greatly expanding coverage and reimbursement for chronic drug treatment, will be necessary. Although this review focused on the value of harnessing inhibitory feeding control signals as drug targets, improving obesity treatment outcomes may also benefit from pursuing drug targets affecting other relevant aspects of biology, such as energy expenditure (i.e., beta-3 adrenergic receptor agonism) and metabolism (i.e., FGF21). Nonetheless, the ability to more effectively inhibit feeding and body weight over time with pharmacotherapies will revolutionize the state of obesity treatment.

Glossary

Abbreviations

BMI

body mass index

BOLD

blood oxygen level–dependent

CNS

central nervous system

DBH

dopamine beta hydroxylase

Ex4

exendin-4

GI

gastrointestinal

GIP

glucose-dependent insulinotropic polypeptide

GLP-1

glucagon like peptide-1

GLP-1

glucagon like peptide-1 receptor

LiCl

lithium chloride

LDTg

lateral dorsal tegmental nucleus

LS

lateral septum

NAc

nucleus accumbens

NTS

nucleus tractus solitarius

PBN

pontine parabrachial nucleus

PPG

proglucagonergic

PVT

paraventricular thalamic nucleus

PYY

peptide YY

RYGB

Roux-en-Y gastric bypass

VTA

ventral tegmental area.

Additional Information

Disclosure Summsty: H.J.G. serves on the Global Obesity Advisory Board for Novo Nordisk, has served as a consultant for Janssen and Pfizer on other topics, and has received research funding from Novo Norisk.

Data Availability: Data sharing is not applicable to this article as no datasets were generated or analyzed during the current study.