The GLP‐1 system as a therapeutic target

Abstract

Glucagon‐like peptide‐1 (GLP‐1) was first isolated and sequenced twenty years ago. It has been shown to have many effects in man. GLP‐1 stimulates insulin secretion, delays gastric emptying, decreases glucagon levels and reduces appetite, all resulting in a fall in plasma glucose concentrations. More recently, evidence suggests it can stimulate beta cell neogenesis and improve cardiovascular function, and may even be neuroprotective. It is no surprise that this peptide is of increasing interest as a target for the treatment of diabetes. None of the drugs currently available for diabetes are able to achieve targets in all patients and none of them are without side effects. The multiple modes of action of GLP‐1 together with its low propensity for hypoglycaemia appear to give it advantages over currently available treatment modalities. In this review I shall examine the data suggesting that medications modelled on the GLP‐1 system may provide a new therapeutic option for diabetes in the future.

• Dipeptidyl peptidase IV
• exendin‐4
• gastric emptying
• glucagon
• glucagon‐like peptide‐1
• hypoglycaemia
• incretin
• insulinotropic
• LAF237
• liraglutide

Introduction

It is 90 years since Moore et al. first demonstrated that a substance extracted from the mucosa of the gastrointestinal tract could treat diabetes 1. In 1929 Zunz and La Barre named a substance proposed to be released from the gut in response to nutrient ingestion that stimulates insulin secretion ‘incretin’ 2. It was not until just over 40 years ago that it was demonstrated that oral ingestion of glucose produced higher insulin levels than intravenous infusion of an equivalent dose of glucose 3. It has since been shown that the incretin effect is responsible for at least 50% of meal‐induced insulin release in man 4. The incretin effect was defined by Creutzfeldt, who suggested the presence of more than one factor 5. The sequence of glucagon‐like peptide‐1 [GLP‐1] in man was first published in 1983 6. It is highly conserved in all species implying an important function 7. A number of studies indicated that GLP‐1 is the most potent naturally occurring incretin in man 8–9, and has a physiological role in the regulation of postprandial glucose 10.

Key messages

• Current therapeutic strategies for type‐2 diabetes are inadequate, the disease is rapidly increasing in prevalence and new and better drugs are urgently needed.

• GLP‐1 increases plasma insulin, decreases glucagon, delays gastric emptying and reduces appetite whilst causing little or no hypoglycaemia.

• Long‐acting GLP‐1 analogues resistant to breakdown by dipeptidyl peptidase IV and antagonists of this enzyme are potentially advantageous treatments for diabetes.

Actions of GLP‐1

Regulation of carbohydrate metabolism

GLP‐1 not only stimulates insulin secretion (insulinotropic), but has several other actions which coordinate to reduce plasma glucose. GLP‐1 stimulates insulin synthesis 11, and it appears to increase islet cell mass 12, though the mechanism through which this occurs is poorly characterized 13. It also inhibits islet apoptosis 14. Type‐2 diabetes is associated with a progressive decline in beta cell function 15, thus, up‐regulation of the GLP‐1 axis may even slow the progression of diabetes.

GLP‐1 reduces glucagon 8, 16, which opposes the action of insulin. As the major role of glucagon is in the fasted state 17, and levels in diabetes are high 18, antagonism via the GLP‐1 system would be expected to be particularly important in controlling fasting glucose in diabetes. It also reduces glucose by delaying gastric emptying 19–21, and inhibits gastric acid secretion 19. Administration of GLP‐1 with a meal in healthy humans causes a reduction in meal‐related glucose and insulin excursions, implying that delay in gastric emptying is more important than its insulinotropic effect 20. GLP‐1 reduces food intake and act as a satiety factor in rats 22, it can also cause weight loss 23. Further data indicate that it can reduce food intake in man 21, 24, 25, also likely to help ameliorate diabetes.

A further advantage of GLP‐1 as a potential agent to treat diabetes would be if it enhanced glucose disposal independent of insulin. This was suggested 17, 26, but most studies have failed to reproduce it 27, 28, and the effect remains controversial. GLP‐1 is glucose‐dependent 29, thus, it reduces glucose to a greater extent in the hyper‐ than the normoglycaemic state 30. This fact was used to infer that it does not cause fasting hypoglycaemia. In some studies this is the case 31, 32, whereas in others hypoglycaemia can occur 30, 33–35, however, apparently only in normal volunteers, not people with type 2 diabetes. Overall it seems that therapeutic agents for type‐2 diabetes based on GLP‐1 will cause little hypoglycaemia, giving it major advantage over insulin or sulphonylureas 36.

Cardiovascular action

GLP‐1 has cardiovascular actions though these vary between species. In the rat it increased heart rate and blood pressure 37. In a rat model of hypertension, GLP‐1 reduced blood pressure, seemingly via a natriuretic and diuretic action 38. GLP‐1 receptor knockout mice have reduced heart rate and increased left ventricular wall thickness 39, consistent with a physiological role in cardiovascular metabolism. In the calf, it is chronotropic without effect on blood pressure 40. In the dog it appeared to enhance recovery from ischaemic myocardial stunning 41.

Cardiovascular effects of GLP‐1 in man also vary. Subcutaneous (s/c) injections in healthy volunteers increased blood pressure 34, whereas continuous s/c infusion tended to lower it 33. GLP‐1 infusion improved endothelial function in patients with stable coronary artery disease whilst having no effect in normal subjects 42. Intravenous (iv) GLP‐1 improved left ventricular function in patients with severe heart failure after large myocardial infarction and successful primary angioplasty 43. Given the frequency of endothelial dysfunction in diabetes, it may have beneficial effects long‐term, and may have a short‐term role in certain high risk patients.

Action on sodium and water balance

Peripheral administration of GLP‐1 inhibited water intake in the rat 44. This effect was abolished by destruction of the arcuate nucleus and circumventricular organs 45. The area postrema forms part of the circumventricular organs and peripheral administration activates this area 46, implying peripheral GLP‐1 may be acting centrally. It also acts as a natriuretic and diuretic in the hypertensive rat 38.

As well as a reduction in food intake with GLP‐1 administration, a decrease in water intake in healthy volunteers 47 and people with type‐2 diabetes 25 has been found. It is natriuretic in healthy human subjects and patients with insulin resistance, suggesting a possible renoprotective effect 48.

Neurological function

GLP‐1 may also be neuroprotective. It reduced apoptosis in models of neurodegeneration, 49 and reduced levels of amyloid 50. It improved spatial learning in rats and restored a learning deficit in GLP‐1 knockout mice: indeed receptor overexpression improved learning and memory 51. There are no human data reported, however, a beneficial role in Alzheimer's disease has been proposed 50.

Therapeutic principle of GLP‐1

GLP‐1 is rapidly broken down by dipeptidyl peptidase IV (DPP‐IV). The plasma half‐life in man is 1–3 minutes 52. Nevertheless, exogenous GLP‐1 can reduce plasma glucose in type‐2 diabetes 16, 53, 9. This has been confirmed using s/c injections 54, buccal tablets 55, and s/c infusion 56–59, with additional effect over metformin 56 and glitazones 59. All these studies verified that the GLP‐1 system is a target for therapy of type‐2 diabetes, alone or in combination providing impetus to the search for a long‐acting analogue.

GLP‐1 analogues

Exendin‐4

Exendin‐4, isolated from the salivary gland of the Gila monster, is a long acting GLP‐1 receptor agonist 60. The half‐life of this peptide in man is about 30 minutes 61, making it a superior prospect to GLP‐1 as an anti‐diabetic agent. Exendin‐4 has a longer lasting and more potent glucose‐lowering effect than GLP‐1 in diabetic mice 62, rats and monkeys 63. It stimulates beta cell replication and islet neogenesis 64, 65 improving glucose tolerance 65, and reduces weight gain in rats 66. Exendin‐4 also mimics the neuroprotective effects of GLP‐1 49.

Intravenous infusion of exendin‐4 in healthy humans reduced fasting and postprandial glucose, as well as reducing food intake by 19% 61. The dose used in this study may have been close to the maximal tolerated as double this dose caused nausea in some subjects 61. A hyperglycaemic clamp was used to show that iv exendin‐4 is a potent insulinotropic agent in type‐2 diabetes, with a long duration of action 67. Subcutaneous exendin‐4 caused complete abolition of postprandial glycaemia in people with type‐2 diabetes, plasma glucose dropping or at least failing to rise after a meal in the active group, associated with a potent slowing of gastric emptying 68. However, a number of subjects suffered nausea and/or vomiting. A reduction of fasting glucose was also noted, though no one was hypoglycaemic 68. Subcutaneous exendin‐4 also reduced postprandial hyperglycaemia in type‐1 diabetes by 90%, via inhibition of gastric emptying and glucagon suppression 69.

Twice‐daily s/c exendin‐4 injections for a month in people with type‐2 diabetes caused long‐term glycaemic control, as measured by glycosylated haemoglobin (HbA1c), to improve by 0.8% 70. There were no significant side‐effects, aside from short‐lasting nausea at the start of the study with two subjects. There was no hypoglycaemia with the lowest blood glucose being 3.3 mmol/l, but there was no weight loss. Improvement in glycaemic control did not occur for 24 hours, and the authors concluded that exendin‐4 in this dose/form would be unlikely to be a treatment for type‐2 diabetes 70. Data from Amylin Pharmaceuticals illustrated its effects combined with metformin and/or sulphonylureas 71. Over half the patients were on combined medication. ‘Triple’ therapy with two or three times a day s/c exendin‐4 reduced HbA1c 0.9% in one month, compared with 0.3% in the placebo group. HbA1c indicates average plasma glucose over three months, thus longer trials should have a greater effect. Mild hypoglycaemia did occur, although only in combination with a sulphonylurea: nausea also occurred causing a dropout rate of 3.7% 71. A recent report showed that the nausea caused by exendin‐4 can be reduced by dose titration 72.

The first study of exendin‐4 for longer than one month has recently been reported by Amylin 73. Patients taking sulphonylureas were randomized to one of two doses of s/c exendin‐4 twice a day or placebo for six months. HbA1c decreased by 0.98% with 10 mcg and 0.57% with 5 mcg compared to placebo, the effect was greater in more poorly controlled patients (1.35 and 0.71 respectively if HbA1c>9%). Study withdrawal was 20%–30% in the treatment arms, and 40% in the placebo arm. The latter was partly due to worse glycaemic control as per the protocol, which is likely to have caused a reduction in the measured treatment effect compared with the actual effect. Adverse events causing withdrawal were 3.3% for placebo, 7.2% for low dose and 10.1% for high dose exendin‐4. There was a small weight loss of 1.6 kg over the study period in the high dose group; the placebo group lost 0.6 kg; and reported nausea and hypoglycaemia were more common in the treatment groups 73. The improvement in glycaemic control is not unlike that for agents currently used in type‐2 diabetes. Data looking at its effects in combination with metformin or with sulphonylureas and metformin have been presented in abstract form.

In 2002 Eli Lilly went into partnership with Amylin to develop exendin‐4, also known as exenatide. The peptide currently has more published human data than any other targeting the GLP‐1 receptor. A long acting formulation designed for use once a week to once a month is in clinical trials, but no more information is currently available.

Liraglutide

The other GLP‐1 agonist with considerable published data is the Novo Nordisk product liraglutide. This is an acylated derivative of GLP‐1 that binds to albumin and is resistant to DPP‐IV degradation 74. Subcutaneous injection of liraglutide in rats caused weight loss for ten days 75. It attenuated diabetes in two models of pre‐diabetes, at least partly via reduction in food intake 76, and induced a short term increase in beta cell mass in normal rats 77. Liraglutide reduced glucose in diabetic ob/ob and db/db mice, also increasing beta cell mass in the latter 78. In pigs, it improved glycaemic control chronically, it stimulated insulin, decreased glucagon levels and delayed gastric emptying 79.

The half‐life of liraglutide in man is 11–15 hours 80. Injection into people with type‐2 diabetes reduced fasting glucose, and postprandial glucose via stimulation of insulin, reduction of glucagon and delay in gastric emptying 81. Subcutaneous liraglutide once daily for a week in people with type‐2 diabetes caused a decrease in average plasma glucose of 2 mmol/L and postprandial glucose after meals of 20 per cent 82. Insulin levels were unchanged, though in the presence of reduced plasma glucose this suggests a relative insulinotropic effect, as has been described with GLP‐1 83. It decreased plasma glucagon and glycogenolysis, improved islet cell function and first phase insulin response, but did not affect gastric emptying 82. There was no hypoglycaemia but nausea and abdominal pain were experienced by some.

An eight‐week study of once daily s/c liraglutide reduced HbA1c by 0.8% compared with placebo, with no change in gastric emptying, body weight or energy expenditure 84. Side‐effects of medication were minor. These latter two studies 82, 84 used a lower dose of liraglutide than that of the previous 81 and it seems likely at higher doses that liraglutide does delay gastric emptying.

Administration of liraglutide for 12 weeks caused a decrease in HbA1c of 0.75% with the highest dose 85. There was a tendency for weight loss, with one dose causing significant weight loss of 1.2 kg compared with placebo. There was minor nausea with increase in dose, but no patients withdrew from the study 85. Acute studies in man using GLP‐1 and its agonists have generally shown a decrease in food intake and even weight loss 21, 24, 25, 47, 61, however, the chronic effects are much less impressive. Weight loss of 1 kg versus placebo was found with exenatide administered for six months 73 and 1.2 kg with one dose of liraglutide for three months 85. I suspect, given the number of studies showing decreased food intake without nausea, that the weight loss is a direct effect, at least at certain doses; however, the metabolic significance may be minor, and dosage is obviously critical. Nevertheless, for prescribers and patients alike the fact that a GLP‐1‐based drug is likely to cause weight loss, however minimal, compared with the well recognized weight gain of insulin and sulphonylureas and the weight stabilization of metformin 86 will be crucial in the ability of these drugs to compete in the pharmaceutical market.

Other GLP‐1 agonists

Various strategies have been used to protect GLP‐1 from breakdown by DPP‐IV and thus increase the half‐life. Acylation or modification of one of the first three amino acids of GLP‐1 (7‐36 amide) has been most commonly investigated 87. Conjuchem have patented a modified GLP‐1, which binds to albumin called CJC‐1131, this is currently in phase II clinical trials 88. Human Genome Sciences have a similar albumin‐bound peptide, Albugon, which has been shown to have a prolonged action in mice 89.

DPP‐IV antagonists

An alternative strategy to target the GLP‐1 system, rather than creating DPP‐IV resistant analogues, is to inhibit the enzyme itself. DPP‐IV knockout mice have reduced glycaemic excursion after a glucose challenge 90 and are protected from both obesity and insulin resistance, implying DPP‐IV antagonists are a therapeutic option 91. The enzyme not only inactivates GLP‐1, but also glucose‐dependent insulinotropic peptide (GIP) 92. Knockout of both the GLP‐1 and GIP receptors in a mouse model caused increased glycaemic excursion in response to oral glucose, it also did not respond to DPP‐IV inhibition implying that GLP‐1 and GIP account for most of the incretin action regulated by DPP‐IV 93.

However, a note of caution must be sounded in that DPP‐IV, also known as CD26, has many actions. Not only does it have many peptide hormones as substrates, it also stimulates T‐cell proliferation as well as inactivating cytokines such as tumour necrosis factor‐alpha (TNF‐α) 94. Activity of the enzyme alters in a number of disease states, in particular dropping in depression 95. DPP‐IV structure is highly conserved in evolution like GLP‐1, implying an important physiological role 96 and long‐term blockade may have predictable or hitherto unpredictable side‐effects 97.

Despite the above, a number of oral DPP‐IV inhibitors have been tested in animals, particularly glucose intolerant models, and in man. Several acute studies have been reported: less data are available on the chronic use of DPP‐IV antagonists. The Novartis compound DPP728 given long‐term to mice improved glucose tolerance with increase in insulin levels 98. The Ferring compound FE 999011 improved glycaemia and delayed the onset of diabetes in Zucker fatty rats 99. Chronic administration of the Probiodrug P32/98 improved glycaemia in Zucker rats 100.

DPP728 has also been shown to be active in people with diet‐controlled type‐2 diabetes 101. Two or three times a day oral therapy improved average plasma glucose by 1 mmol/L after four weeks compared with placebo, there was no change in body weight. Tablets were taken ten minutes before food, and were generally well tolerated, the incidence of hypoglycaemia and gastrointestinal side‐effects was low, though one patient did develop transient nephrotic syndrome 101.

LAF237 (Vildagliptin)

Novartis have published no further data on DPP728, and have concentrated on the longer acting DPP‐IV inhibitor LAF237, also known as vildagliptin. Administration once a day thirty minutes before breakfast for four weeks in people with type‐2 diabetes decreased plasma glucose 1 mmol/L. There was little effect on insulin, but glucagon was inhibited. No significant side effects were noted, and there was no change in body weight 102. This group has gone on to perform a longer‐term study of vildagliptin in patients treated with metformin 103. After 12 weeks HbA1c dropped 0.7% compared with placebo, after 1 year it was 1.1% in the 58 out of the original 107 patients who continued. Diabetic control at the start of the study was moderate (HbA1c 7.7 or 7.8). There was reduced fasting glucose and glucose area under the curve (AUC) in response to a test meal with vildagliptin, insulin AUC was unchanged indicating a relative insulinotropic effect. No effect on body weight was found and there was a low incidence of hypoglycaemia 103. Three of 42 patients treated with vildagliptin for a year required an increase in their anti‐hypertensives; the cause of this is unknown, though this would be consistent with the increase in blood pressure previously found with GLP‐1 34.

Conclusion

Much data has been published on GLP‐1 and its agonists in rats and other species. Potentially important theories have been proposed regarding increase in islet cell mass induced by GLP‐1, slowing of the progression of diabetes or delay in its onset with GLP‐1 and its agonists and even a neuroprotective role. However, these have all just been demonstrated in animals, and are only inferred and have not been reported in man. Nevertheless, several reports have been published in man regarding three first‐generation agents targeting the GLP‐1 system. Exenatide reduces HbA1c and probably causes some weight loss, it produces little hypoglycaemia, though it has to be injected twice a day, and nausea is likely to limit its usefulness in some. There is a hint of a much longer‐acting exendin‐like molecule. Less data are available for liraglutide, which seems to have similar efficacy and also may cause some weight loss. It only has to be injected once a day. Vildagliptin is the only oral compound available. It has similar anti‐hyperglycaemic effect to the other two but without causing weight loss. We await evidence to suggest whether the less specific DPP‐IV antagonism is of long‐term advantage or harm.

Currently, the major compounds used in the treatment of type‐2 diabetes are metformin, which is not tolerated by many and contra‐indicated in many more, sulphonylureas, which cause weight gain and hypoglycaemia, and thiazolidinediones, which appear less efficacious than the others, cause weight gain and have less long‐term data. The alternative is insulin; however, the necessary multiple injections are disliked by patients, and weight gain and hypoglycaemia remain problematic. The prevalence of obesity and thence type‐2 diabetes is increasing rapidly. There is an urgent need for newer drugs that have the activity profile of agents acting on the GLP‐1 system: soon we will see whether any of the agents above fit the bill. Excitingly, we have early evidence suggesting the possibility of advantage in the long term, ranging from cardiovascular to neurological protection, features that would make such drugs even more attractive for the long‐term treatment of type‐2 diabetes.