Gut Hormones and Related Concepts

At a symposium at the 2006 American Diabetes Association (ADA) Scientific Sessions on the implications of taste perception, Robert F. Margolskee (New York, NY) discussed the molecular biology of taste. Multiple taste buds are contained within papillae of the tongue. Taste receptors are present in cells of the front, side, and posterior portion of the tongue, with different taste responses in different areas. The innervation of the taste buds involves the facial nerves and vagus. Each taste bud contains 50–100 taste receptor cells, specialized epithelial cells with neuron-like properties making contact with taste stimuli in the oral cavity, leading to activation of an intracellular cascade with subsequent neurotransmitter release. Two taste qualities, salty and sour, are mediated by microvilli in the superior portions of the taste receptor cells, with salty taste involving sodium-specific channels and sour taste involving a number of different mechanisms (either receptors or via blockade of apical potassium entry). The three other tastes are bitter, sweet, and umami (Japanese for “delicious,” the taste of monosodium glutamate), all mediated by G-protein receptors. Bitter taste receptors have 25–30 different structures coupled to a specific G-protein, while umami and sweet depend on a different family of taste receptors—T1R2 and T1R3 for sweet, and T1R1 and T1R3 (as well as other complex combinations) for the taste of glutamate. Both the sweet and umami receptors have large extracellular domains.

Sweet taste genetics have been studied in inbred mice strains to determine the saccharine taste locus, with some strains tasting and others not tasting this. In humans, similar studies allowed identification of loci on chromosome 1p38.33 with a candidate gene identified for T1R3, which was previously thought to be an orphan receptor. In mice not expressing T1R3, there is no compensatory increase in T1R2 and T1R1, and these mice show decrease, both in neural response to and in preference for artificial sweeteners.

A puzzle has been the diversity of chemically distinct sweet compounds, including small molecule artificial sweeteners, large proteins such as brazzein, and mono- and disacccharides. Multiple regions of T1R2–T1R3 allow effective interaction with different sweet compounds, appearing to explain the recognition of multiple different sweet-tasting substances. Heterodimerized T1R2 plus T1R3 forms a sweet receptor, which can recognize sweet substances such as d-tryptophan, N-saccharine, sucrose, aspartame, and brazzein. It is not always possible to extrapolate from the animal model to humans; human T1R3 determines responsiveness to cyclamate, while the mouse receptor does not respond to this, although it does respond to d-tryptophan. Margolskee showed a model of the three-dimensional structure of these receptors, allowing identification of the sites of action of various sweet-tasting substances.

Understanding the mechanisms of taste may give fascinating insights into the mechanisms of gastrointestinal chemosensation. More than two decades ago, cells were found in the gastrointestinal tract with similarity to gustatory cells of the tongue, with recognition that such chemosensory cells are the endocrine cells of the gastrointestinal tract and share receptor characteristics with taste buds. T1R2 and T1R3 are expressed in the gastrointestinal tract, as are other taste receptors. The taste signaling elements are coexpressed in enteroendocrine cells of the small intestine, particularly by L-cells, which express glucagon-like peptide (GLP)-1, GLP-2, PYY, and additional hormones mediating aspects of glucose homeostasis. GLP-1 release from the GLUTag cell line, a mouse L-cell model, is stimulated by sucralose (Splenda) and blocked by T1R3 inhibitor. In studies of an animal model not expressing gustducin, a taste cell–specific G protein (and hence with an abnormality of taste receptors), the incretin effect is diminished, with loss of GLP-1 response to enteric nutrients. Thus GLP-1 release depends on a functioning T1R3 receptor, further suggesting the importance of understanding of these mechanisms.

Anthony Sclafani (Brooklyn, NY) discussed a further intriguing aspect of gastrointestinal taste receptors, the role of taste versus “postoral” receptors in the appetite for sugar. Feeding rats a “supermarket diet” of cereals and marshmallows, high in fat and sugar, leads to dietary obesity. Sclafani asked, “Why does the rat love to eat sugar?” The tastes of sugar and artificial sweeteners lead to dopaminergic and endorphin brain reward signals, with gastrointestinal satiety mechanisms leading to negative feedback from the gut via hormonal output. To separate the mouth from postingestive organs, models have been developed using a gastric fistula to eliminate the element of satiety. In “real-feeding” animals, intake declines via postoral satiety signals, while sham feeding with the fistula kept open continues food intake in an unabated fashion. Another technique used for study has been the infusion of sugar-containing solutions into the stomach, with water producing little satiety whereas a 16% sugar solution leads to suppression of further nutrient intake. The gut mediates positive feedback, with reward-signals occurring in a delayed fashion involving learning, leading to animals developing preference for oral flavors administered in association with intragastric sugar infusion.

Sclafani showed studies designed to characterize the site of action of enterally administered sugar solutions. Duodenal administration leads to the same degree of preference as gastric administration, while gastric administration with the stomach clamped failed to cause the learned response. Infusion in the hepatic and portal veins failed to cause the response as well, suggesting that the presence of sugars in the intestine is required for this effect, perhaps occurring via taste receptors present in the duodenum. Interestingly, sucralose has an opposite effect to sucrose when administered intragastrically, perhaps explaining the greater desirability of natural sweeteners. The enteric effects of sugars may become a pharmacological target if artificial sweetener–like molecules that lead to the release of satiety-mediating hormones from the gut can be developed. We do not yet know the mechanisms by which the activation of gastrointestinal receptors leads to central nervous system (CNS) signaling. Vagotomy does not interrupt the signal, suggesting the possibility of an endocrine message. Other intragastric conditioning factors include fats, proteins, and ethanol, with Sclafani speculating that gut receptors may be present for all classes of nutrients.

Richard D. Mattes (Lafayette, IN) discussed studies of the sensory properties and health implications of dietary fat in humans. Aristotle proposed in the 4th century B.C.E. that fat was a basic taste, but, in part because of lack of evidence that fat could be transduced into signals, recently this was thought not to be the case. It is now recognized that fatty acids may alter potassium-rectifying channel activities, leading to taste cell hyperpolarization, so that rather than being primary taste stimuli, fats may increase the signals from other taste sensations. Dietary fats are detected by texture, odor, and probably taste. Oral fat exposure initiates cephalic phase responses, including gastric lipase secretion, changes in gastrointestinal transit activity, pancreatic exocrine response, and gut hormonal response.

However, in human studies with 1% linoleic acid plus NaCl, sucrose, caffeine, citric acid, and other pure taste stimulators, sensitivity was lower rather than higher, suggesting that this mechanism does not apply. CD36, the scavenger receptor, may be another pathway, present in taste receptor cells, particularly those on the lateral and posterior portions of the tongue, with some evidence that different portions of the tongue show differential sensitivity to different fatty acids. CD36 binds monounsaturated, saturated, and polyunsaturated fats, and there is evidence in humans that all three fats can separately be identified. Other potential mechanisms include fatty acid transporter proteins such as FATp4, which is present in the enterocyte. Mattes echoed Margolskee’s theme that “the tongue may not be all that different from the intestine.” Another potential mechanism by which fats lead to taste sensation involves simple diffusion, with cellular metabolism involving activity of acylCoA synthase then activating fat recognition. There are many mechanisms for detecting fatty acids. In studies using a range of fatty acid concentrations, with monounsaturated, saturated, and polyunsaturated fatty acids, it certainly is possible to detect fat by texture, viscosity, lubricity, olfaction, vision, irritancy, or oxidation products. To eliminate texture as a factor in studies of the phenomenon, acacia and mineral oil were added in a series of studies. Olfaction and vision were eliminated using nose clips and red light, with capsaicin pretreatment to eliminate irritant mechanisms. After 15 min the tongue is desensitized, and reliable thresholds for all three fatty acid types can be demonstrated at a concentration of ∼0.03%.

Mattes addressed potential implications of taste of dietary fat, asking whether taste might be present throughout the gastrointestinal tract. The effects of the cephalic phase of dietary fat in the oral cavity may involve ghrelin release, changes in gastric emptying, pancreatic polypeptide release, and effects of a variety of other mediators. The presence of fat in the stomach and mouth has a variety of effects, with oral fat administration elevating serum triglyceride levels. The triglyceride appearing in serum appears to be derived from fats stored in enterocytes. Ingestion of glucose but not of water clears fat from enterocytes. Oral fat also increases pancreatic polypeptide, contributing to VLDL synthesis and secretion. The principle triglyceride-elevating effect occurs late after food ingestion, suggesting that other factors are involved.

There is evidence that fat taste plays a role in obesity, with different rat strains showing differing responses, some showing increased food ingestion in response to fat, and others appearing to have aversive responses to fat. In humans, there is evidence of a satiety-inducing effect of fat. There is evidence that some but not all persons taste linoleic acid, with the sensitive individuals possibly having a reward response to fat. Mattes pointed out that GLP-1 responds to ingestion of sugars and fats; cholecystokinin is most strongly affected by fat; ghrelin is most suppressed by carbohydrates and protein; there may also be heterogeneity in gut hormonal responses, as well as heterogeneity in response to different macronutrients.

Edmund T. Rolls (Oxford, U.K.) discussed mechanisms of taste-specific satiety. Taste, smell, and texture all have effects, with taste acting at a primary taste cortex area, with distribution to the orbitofrontal cortex, where there is combination with olfaction and with visual and other sensory input to integrate the pleasantness of the stimulus. The lateral hypothalamus in primates responds to the taste of glucose in the mouth, but also to the sight of food. This is a crucial site where hunger interacts with taste. After feeding, these hypothalamic neurons demonstrate a lessened taste response, suggesting the hypothalamus as a site of pleasure integration. Variety overrides food intake and hence may contribute to obesity. The phenomenon of reduction in response does not occur in the primary taste cortex, with the experience of palatability of food appearing to be integrated at the level of the orbitofrontal cortex, where, Rolls said, “Flavor is built.” This can be demonstrated when visual stimuli are conditioned to be associated with aversive or pleasant tastes. An important part of the response to fat, which Rolls termed the “texture response, ” occurs in the orbitofrontal cortex; analysis of single neuron responses in animal models adds to the evidence that taste reward is integrated in the orbitofrontal cortex. Using functional nuclear magnetic resonance (fNMR) brain imaging, umami can be shown to activate the orbitofrontal cortex, particularly when presented with complimentary tastes. Foods eaten to satiety decrease in pleasantness, with a corresponding decrease in the fNMR orbitofrontal cortex signaling. In a study with six different olfactory stimuli, the medial orbitofrontal cortex represents pleasant while the lateral orbitofrontal cortex represents unpleasant stimuli. So this portion of the brain builds associations between different food-related stimuli. Umami is not pleasant when ingested alone, but becomes pleasant when a person is exposed to umami in the setting of a vegetable odor. Fat texture is also represented in these parts of the brain. There are also effects of cognitive factors on perception. For example, when isovaleric acid, which gives the taste to brie cheese, was presented to persons reading the words “cheddar cheese” or “body odor,” fNMR similarly reflected the respective pleasant or unpleasant reactions. Stress and emotional eating represents another factor, whose site of integration in the brain is not known. Rolls hypothesized that the increase in prevalence of obesity involves altered response to satiety inputs from gut hormones, which must reach the orbitofrontal cortex and hypothalamus. Visual, taste, olfactory, and cognitive inputs have been intensively addressed by food manufacturers, overriding evolutionarily developed satiety signals.

Thus the increasing levels of food intake and consequent obesity are driven by palatability and marketing, rather than endocrine and genetic factors. When food is “there all the time” and in great variety, persons destined to become obese may eat more concentrated foods rapidly.

New hormonal targets for diabetes and obesity treatment were the topic of another ADA symposium, with Christopher McIntosh (Vancouver, BC, Canada) discussing potential uses of GLP-1 and glucose-dependent insulinotropic polypeptide (GIP) receptor agonists and antagonists in the treatment of diabetes. GIP and GLP-1 are the major gut hormones with effect on glucose homeostasis. The two peptides have high degrees of homology and are equipotent in increasing insulin secretion in the perfused pancreas model, but persons with type 2 diabetes exhibit extreme resistance to GIP, which appears mainly to be due to reduction in second-phase insulin secretion in response to the agent. GIP is a powerful stimulator of insulin synthesis and secretion, stimulates β-cell generation, and also affects adipocyte and bone metabolism. It has effects on glucose absorption and on the vasculature as well. Resistance to GIP is also found in an animal model of type 2 diabetes, the Vancouver Diabetic Zucker Fatty rat. GIP receptor expression is reduced in the pancreas of these animals. If ZDF are treated with phlorizin to normalize glucose, responsiveness increases, suggesting that normalization of glycemia in diabetes may allow GIP responsiveness to be restored. GIP is cleaved by dipeptidyl peptidase (DPP)-IV, the enzyme that also degrades circulating GLP-1. Several DPP-IV–resistant GIP analogs have been produced, some retaining biological activity, with studies showing that [d-ALA(2)]GIP_1–42 improves glucose tolerance in the VDF rats (1).

Another approach involves modification of GIP with long-chain fatty acids to prolong activity. GIP contains two activation domains, 1–14 and 19–30, both of which stimulate insulin secretion in vitro and improve glucose tolerance in rat models, with GIP_1–30 having similar potency to GIP_1–42. When GIP_1–14 and GIP_19–30 are directly connected, a superactive analog is produced, improving β-cell glucose responsiveness and insulin sensitivity in animal models. McIntosh noted that fat is a stronger stimulator of GIP release than glucose, but that GIP only causes insulin secretion with the combined stimulus of oral nutrient plus an increase in plasma glucose. Furthermore, GIP increases chylomicron triglyceride clearance in dog models. GIP stimulates both lipolysis and insulin-mediated lipogenesis, suggesting that it also has effect on the balance between lipogenesis and lipolysis. There is the potential for adverse effects of GIP in obesity. Ob/ob mice have increased intestinal GIP and hyperplasia of the GIP-secreting K-cells. Furthermore, mice lacking the GIP receptor fed a high-fat diet are protected from both obesity and insulin resistance (2). Thus, excessive nutrient intake may cause hyperplasia of cells producing GIP, synergistically further increasing adipose tissue mass. GIP receptor antagonists, then, may play a role in the treatment of obesity. Administration of the antagonist Pro³-GIP, rather than worsening glucose disposal, appears to improve insulin sensitivity and reduce food intake in ob/ob mice with consequent improvement in glycemia and in insulin sensitivity, with evidence of improvement in islet structure (3).

David A. York (Baton Rouge, LA) presented information on enterostatin as a potential target for obesity treatment. Enterostatin is a peptide that selectively reduces fat intake (4). Procolipase is the precursor of enterostatin, expressed in the exocrine pancreas, with enterostatin and colipase its two cleavage products. The structure is preserved across species. Natural antagonists are the casomorphins released from milk proteins. Procolipase gene expression and enterostatin protein are also present in the liver, the stomach, and CNS regions, including the hypothalamus and amygdala. With administration of enterostatin, either centrally or peripherally in rat models, there is dose-dependent selective reduction in dietary fat intake, without change in ingestion of carbohydrate or protein. Chronic administration centrally reduces body weight, body fat, and fat intake without compensatory increase in carbohydrate intake. Enterostatin is released after ingestion of dietary fat and enters the circulation and lymphatics to decrease fat intake. Enterostatin also decreases insulin secretion, both directly and indirectly by decreasing pancreatic parasympathetic tone. Enterostatin has a direct AMP-activated protein kinase–activating and fat oxidation–stimulating effects on myocytes. Enterostatin activates the hypothalamus-adrenal axis, increasing thermogenesis, and causing weight loss with decreased body fat. Enterostatin is also present in the brain and is produced in the amygdala, the most sensitive CNS site in suppressing dietary fat intake via actions on the hypothalamus. Enterostatin causes release of serotonin, further playing a role in the suppression of dietary fat intake.

The enterostatin receptor has been isolated in pancreatic islets and appears to be a G protein–coupled receptor, possibly present in mitochondria. There may be a high-affinity site for enterostatin and low-affinity binding sites for beta casomorphin on the receptor, causing a positive cooperativity phenomenon. Studies of a variety of different peptides show that reduction in antagonist binding increases food intake. Discussing the evidence of enterostatin binding to a mitochondrial protein, York reviewed recent studies of apoA1 binding to hepatocytes via the F1-ATPase β subunit (5) and noted that in addition to apoA1 binding to the F1Fo ATP synthase complex, which both generates cellular ATP in mitochondria and is present on the cell membrane leading to HDL internalization, enterostatin and β casomorphism also bind to F1Fo ATP synthase, with high-affinity enterostatin receptors present on liver plasma membranes. York suggested that enterostatin binds to the β subunit, inhibiting extracellular ATP hydrolysis and activating adynylate cyclase and other pathways. F1-ATPase β subunits are present on the plasma membrane in the hypothalamic cell line GT1–7, with enterostatin in this site increasing adenylate cyclase, stimulating cAMP rather than inhibiting cAMP production. Enterostatin also activates mitogen-activated protein kinases. The complex data currently available suggests that enterostatin inhibits islet cAMP production while increasing cAMP in neurons, suggesting differing mediators. Agouti-related protein (AgRP) is a hypothalamic peptide involved in regulation of feeding, which selectively increases dietary fat intake, and enterostatin inhibits AgRP in a dose-response fashion, whereas enterostatin antagonists enhance AgRP expression, involving a cAMP pathway.

Few human studies have been performed. Intravenous administration of enterostatin has no effect on food intake in humans in acute studies, although York noted that circulating binding proteins may reduce activity of enterostatin administered intravenously (6). Oral enterostatin has not been shown to reduce food intake (7), but is reported to increase satiety, and given the lack of evidence that enterostatin is transported across the blood-brain barrier, York concluded that one must be uncertain as to its role in obesity treatment. With the multiple ligands for its receptor, its intriguing effects on energy metabolism and on fat intake, and the presence of the enterostatin receptor in mitochondria as well as on plasma membranes, however, it remains an important subject for ongoing study.

Randy J. Seeley (Cincinnati, OH) discussed the evidence that granulocyte/macrophage–colony stimulating factor (GM-CSF) is a physiologically relevant catabolic hormone. Leptin plays a key role in fuel economy as a feedback signal of stored calories. In its structure, however, leptin is a cytokine with a number of effects independent of its role in food intake, including proinflammatory actions in immune responses, stimulating proliferation, differentiation, and activation of hematopoietic cells. GM-CSF is a proinflammatory cytokine produced in endothelial cells, granulocytes, and macrophages, with many similarities to leptin. With intracerebroventricular administration, GM-CSF decreases food intake in a dose-related fashion, leading to weight loss, with potency 10-fold greater than leptin. GM-CSF appears to produce food-independent effects on energy expenditure that can potentiate weight loss, as well as resulting in prolonged inhibition of food intake over 48–72 h. The CNS is the principal target of leptin. Similarly, GM-CSF has considerably greater action given centrally than peripherally, appearing to reflect a real effect on food intake rather than a toxic or aversive response to inflammation. Leptin is made peripherally and acts in the CNS, but although GM-CSF is expressed to highest extent in lung, as well as in adipose tissue and liver, circulating and peripheral tissue levels show no change in GM-CSF levels from the fed to the fasted state. With therapeutic administration in high doses for treatment of neutropenia, as during cancer chemotherapy, there is also no effect on food intake, further suggesting that peripherally derived GM-CSF does not play a role in energy balance. Seeley noted that mesenteric adipose tissue has much higher expression than subcutaneous adipose tissue of GM-CSF, as is the case with expression of other adipokines, such as tumor necrosis factor (TNF)-α. Animals that fail to express GM-CSF (GM-CSF^–/–) also have marked reduction in mesenteric adipocyte expression of TNF-α, as well as other proinflammatory cytokines. Obesity is associated with increased adipose tissue macrophage accumulation, which is decreased in the GM-CSF^–/– animals. Expression of adiponectin, however, is not affected in this model.

The GM-CSF receptor is one of a family of receptors, including those for interleukin (IL)-5 and IL-3, with a common receptor α-subunit. Immunohistochemical staining for this subunit shows neuronal location in many but not all hypothalamic neurons, particularly in the arcuate and paraventricular nuclei, with in situ hybridization studies showing the GM-CSF gene in these locations, most likely expressed by resident macrophages or microglia rather than directly from neuronal elements. Animals that fail to express GM-CSF (GM-CSF^–/–) develop late-onset obesity with a tripling of body fat mass and with food intake reduced and energy expenditure increased following central administration. The regulation of central GM-CSF, its relationship to other regulatory factors (particularly leptin), and its roles in obesity-induced inflammation and comorbidities remain to be defined.

A number of studies presented at the ADA meeting allowed understanding of approaches to diabetes therapy involving the incretin system. Candelore et al. (abstract 445) and Brady et al. (abstract 592) administered a small molecule glucagon receptor antagonist in high-fat diet–induced obese diabetic mice expressing the human glucagon receptor, reporting a 30% reduction in glycemia. Moyers et al. (abstract 531) reported work by another group studying a similar compound, finding a 47% decrease in plasma glucose levels with a 4.2-fold increase in plasma glucagon levels compared with controls in ob/ob mice. Knop et al. (abstract 46-LB) demonstrated suppression of glucagon with intravenous glucose, but not during an oral glucose tolerance test with similar glucose levels, in eight lean persons with type 2 diabetes, perhaps related to the 50% reduction in incretin effect of potentiation of insulin secretion by oral glucose administration. Sandoval et al. (abstract 47-LB) administered the GLP-1 antagonist des His1 Glu9 exendin-4 versus saline into the third cerebral ventricle in rats before intraperitoneal glucose administration, showing similar insulin levels but 30–50% increase in glucose, suggesting a central action of GLP-1 in improving glucose homeostasis.

Sitagliptin and vildagliptin are two inhibitors of DPP-IV approaching Food and Drug Administration approval. Mu et al. (abstract 588) compared the DPP-IV inhibitor des-fluoro-sitagliptin with rosiglitazone and glipizide in the streptozotocin-treated high fat–fed mouse type 2 diabetic model, showing increased β-cell mass and β-cell to α-cell ratio and improved glucose-stimulated insulin secretion with reduction in glucagon levels in the former group. Rosenstock et al. (abstract 556) randomized 353 pioglitazone-treated persons to sitagliptin versus placebo for 24 weeks, with HbA_1c (A1C) decreasing from 8% to 7.2% vs. 7.8%, with an 18-mg/dl placebo-adjusted fall in fasting glucose from baseline of 167 mg/dl, without difference in body weight between the two groups. Karasik et al. (abstract 501) treated 701 persons receiving metformin ≥1,500 mg/day with sitagliptin 100 mg daily versus placebo for 24 weeks, resulting in placebo-subtracted A1C fall from 8% by 0.7% with fasting and 2-h post-meal glucose levels decreasing 25 and 51 mg/dl, increased fasting and postmeal insulin and C-peptide levels, and similar 0.7 vs. 0.6 kg weight loss. Nonaka et a (abstract 537) reported a 12-week study of sitagliptin versus placebo in 151 Japanese type 2 diabetic patients, finding A1C to decrease from 7.5% to 6.8% versus increasing from 7.7% to 8.1%, with 2-h postmeal glucose levels decreasing 69 mg/dl vs. increasing 12 mg/dl, although body weight fell 0.1 vs. 0.7 kg, respectively.

The mechanism of action of vildagliptin was the topic of a variety of studies. Azuma et al. (abstract 5-LB) compared vildagliptin 50 mg twice daily versus placebo in a 6-week crossover study of 16 type 2 diabetic persons, with insulin clamp during deuterated glucose and ¹³C palmitate turnover studies, finding similar fasting glucose production rates but 12% greater glucose clearance during fasting, 13% greater glucose disposal during hyperinsulinemic clamp at an insulin level similar to that in the placebo group, and 17% greater insulin sensitivity during a meal study, suggesting that there may be a small effect of vildagliptin in improving insulin sensitivity. Balas et al. (abstract 122) studied 16 type 2 diabetic persons following an evening meal, showing 38% greater suppression of endogenous glucose production, 21% greater insulin secretion, and 93% greater glucagon suppression following a single 100-mg dose of vildagliptin versus placebo. D’Alessio et al. (abstract 454) measured minimal model insulin sensitivity and insulin response to glucose in 12 drug-naïve type 2 diabetic persons treated with vildagliptin 50 mg twice daily versus placebo for 12 weeks. A1C decreased from 6.7% by 0.4% with vildagliptin and was unchanged with placebo. Despite there being no increase in plasma intact GLP-1 during the studies, which were performed in the fasting state, the acute insulin response increased 2.4-fold and insulin sensitivity increased 57%. Part of the improvement was maintained 2–4 weeks after discontinuation of treatment. He et al. (abstract 483) showed that an increase in postprandial insulin levels is only demonstrated after a high glucose load, and not with a mixed meal. Matikainen et al. (abstract 524) treated 31 type 2 diabetic persons with vildagliptin versus placebo for 4 weeks, finding that the incremental post–lipid-rich meal total triglyceride decreased 85% and that for chylomicron triglyceride decreased 91% with increased GLP-1 and decreased glucagon levels, suggesting these non–insulin-mediated effects to be important in clearance of ingested nutrients.

Therapeutic use of vildagliptin was assessed in a number of studies. Mimori et al. (abstract 527) studied 219 drug-naïve type 2 diabetic Japanese persons treated with placebo or vildagliptin 10, 25, or 50 mg twice daily for 12 weeks, showing A1C to decrease from baseline 7.4% by 0.3%, 0.5%, 0.7%, and 0.9%. Fasting glucose increased 2 mg/dl versus decreasing 11, 14, and 25 mg/dl. Dejager et al. (abstract 120) treated 526 drug-naïve type 2 diabetic persons with metformin 1,000 mg twice daily versus vildagliptin 50 mg twice daily for 52 weeks, finding A1C fall from 8.7% by 1.4% vs. 1.0%, a significantly greater effect of the biguanide. Weight decreased 1.9 kg versus increasing 0.3 kg. Diarrhea occurred in 26 vs. 6%, nausea in 10 vs. 3%, abdominal pain in 7 vs. 2%, and vomiting in 4 vs. 2%. One versus three patients developed mild hypoglycemia. Garber et al. (abstract 121) randomized 416 patients receiving metformin ≥1,500 mg daily to vildagliptin 50 mg daily or twice daily or placebo, finding placebo-adjusted A1C fall from baseline of 8.4% by 0.7% and 1.1% with once- and twice-daily vildagliptin, and fasting glucose decreasing 14 and 31 mg/dl, without weight gain, and with 10 and 15% of vildagliptin versus 18% of the placebo group experiencing gastrointestinal side effects. Rosenstock et al. (abstract 557) treated 459 type 2 diabetic persons with vildagliptin 50 mg twice daily versus 238 with rosiglitazone 8 mg daily for 24 weeks, finding similar 1.1% fall in A1C from baseline of 8.7% in both groups. Body weight was unchanged with vildagliptin, while increasing 1.6 kg with rosiglitazone, with edema in 2.5 vs. 4.9% of patients. Fonseca et al. (abstract 467) reported a 24-week study of 256 type 2 diabetic persons receiving >30 units insulin daily (mean 82 units) randomized to vildagliptin 50 mg twice daily versus placebo, showing A1C decrease by 0.5 vs. 0.2% from a baseline level of 8.5%. Hypoglycemia occurred in 33 vs. 45 patients, with 113 vs. 185 events, none vs. six classified as severe. Interestingly, Nathwani et al. (abstract 474) analyzed 605 patients receiving vildagliptin versus placebo monotherapy and 525 receiving vildagliptin versus placebo in addition to metformin, showing 1–4/1–3 mmHg reductions in blood pressure, suggesting a modest benefit, the mechanism of which was not characterized.

Nair et al. (abstract 532) reported properties of another DPP-IV inhibitor, GRC 8200, which has begun clinical trials. Christopher et al. (abstract 452) demonstrated efficacy of a DPP-IV inhibitor, SYR-322, in Sprague Dawley rats, beagle dogs, and cynomolgus monkeys, with 45–87% oral bioavailability and with maximal inhibition of DPP-IV at 30 min, but with 69% inhibition through 12 h leading to improvement in glucose tolerance and increases in plasma insulin levels in the animal models.

Linnebjerg et al. (abstract 116) studied effects of exenatide in 17 persons with type 2 diabetes using scintigraphic studies, finding a dose-dependent increase in the half-times for solid and liquid gastric emptying from 60 and 34 min with placebo to 111 and 87 min with 5 μg and to 169 and 114 min with 10 μg exenatide doses, respectively, with corresponding lessening of postprandial glucose from 234 to 198 and 180 mg/dl, and with reduction in food consumption of a standard breakfast meal. Wajcberg et al. (abstract 118) assessed the role of glucagon suppression with exenatide by comparing effects of exenatide infusion with and without glucagon during a meal tolerance test in six type 2 diabetic persons. Glucagon levels decreased 20% with exenatide, an effect blocked by concurrent glucagon infusion, with insulin levels increasing from 18 to 66 and to 80 pmol/l, respectively, with the increment in mean plasma glucose during the meal 60 mg/dl in a control group receiving neither agent, 12 mg/dl with administration of exenatide alone, and 25 mg/dl with exenatide and simultaneous glucagon replacement, with turnover studies showing that approximately half of the decrease in hepatic glucose production was due to inhibition of glucagon secretion and half due to increased insulin secretion. Edgerton et al. (abstract 119) administered exenatide versus saline via the portal vein in catheterized dogs during hyperglycemia, with trends to greater hepatic glucose uptake and peripheral glucose uptake.

Kim et al. (abstract 487) reported results of weekly administration of a long-acting release (LAR) formulation of exenatide 0.8 and 2.0 mg versus placebo in 45 type 2 diabetic patients over 15 weeks, observing reductions from baseline of 1.4 and 1.7% versus a 0.4% increase, while fasting glucose decreased 43 and 39 mg/dl versus increasing 18 mg/dl. Thibaudeau et al. (abstract 434) modified exenatide to form a covalent bond with human albumin, showing continued ability to bind to the human GLP-1 receptor, with 2- to 7-day duration of action in murine diabetes models.

Zinman et al. (abstract 117) studied the effect of a 16-week period of combination treatment using exenatide with thiazolidinediones in 233 persons with type 2 diabetes, 80% of whom also received metformin. A1C decreased from 7.9 to 7.1%, with fasting glucose 27 mg/dl lower than in patients treated with placebo. Weight decreased 1.5 vs. 0.2 kg, suggesting this to be an effective combination treatment approach.

Four studies assessed concomitant administration of exenatide with insulin in type 2 diabetic individuals. Bhatia et al. (abstract 442) treated 41 persons receiving insulin, observing suppression of appetite and a reduction in food intake, with 32 persons losing weight (mean 3.2 kg) and a fall in A1C from 8.1 to 7.2%. Prandial insulin decreased from 52 to 28 units/day, while basal insulin decreased from 60 to 55 units/day and mixed insulin doses decreased from 81 to 57 units/day. Davis et al. used a different approach, endeavoring to substitute exenatide for insulin versus continuing insulin treatment in 45 type 2 diabetic persons. Of those switching to exenatide, 62% maintained control but 38% showed fasting hyperglycemia within 2 weeks; all patients who continued insulin maintained glycemic control. Among those receiving exenatide, C-peptide increased 20%, and weight decreased 4.2 kg, independent of the glycemic response.

Hood (abstract 488) treated 63 type 2 diabetic persons with mean A1C of 6.2% at baseline with exenatide, showing a 0.1% decrease in A1C and a 4% reduction in weight. All eight persons receiving an insulin secretagogue discontinued this treatment. Of the 42 persons receiving insulin, the dose decreased from 67 to 43 units daily, with a reduction in the daily number of injections from 2.8 to 1.7. King et al. (abstract 507) reported clinical experience with 100 persons treated with exenatide, of whom 59 received a thiazolidinedione, 47 a sulfonylurea, 65 metformin, and 36 insulin. Treatment was discontinued because of side effects of nausea in seven patients, injection site urticaria in 3, and abdominal pain in 1. 68 patients completed at least 12 weeks of follow-up, with a fall in A1C of 0.4%. Of 23 evaluable insulin-treated patients, seven were able to reduce the insulin dosage. Comparing the studies, it appears on balance safer to initiate treatment with exenatide and then gradually withdraw insulin, with meal bolus insulin perhaps particularly suitable for withdrawal in these patients.

Carter et al. (abstract 446) administered exenatide with the antiinflammatory compound lisofylline to nonobese diabetic mice, a type 1 model, showing reversal of diabetes with evidence of insulin-positive islet-like cluster cells showing immunohistochemical evidence of the β-cell neogenesis factors neurogenin 3 and pancreatic duodenal homeobox 1, suggesting a potential approach to type 1 diabetes treatment in humans.

Vilsboll et al. (abstract 115) treated 165 type 2 diabetic persons with liraglutide, 0.65 mg, 1.25 mg, or 1.9 mg, or with placebo, once daily for 14 weeks, reporting a placebo-adjusted 1.7% fall in A1C at the highest dose, with 3-kg weight loss in the high dose group, without risk of hypoglycemia. Mari et al. (abstract 522) treated 13 type 2 diabetic persons for 7 days with liraglutide, decreasing mean glucose during one day of monitoring from 175 to 140 mg/dl, without change in insulin and C-peptide concentrations, suggesting increased β-cell glucose responsiveness.

Kapitza et al. (abstract 500) treated 18 metformin-treated type 2 diabetic persons with a GLP-1 analog, BIM51077, administered by continuous subcutaenous infusion for 28 days, finding 49 vs. 17 mg/dl fall in fasting glucose with a similar fall in 24-h mean glucose levels. In what may be a promising approach for long-term treatment, Ramis et al. (abstract 547) studied a zinc-containing slow release formulation of BIM51077 in dogs, showing a 6.8-day elimination half-life.