Give the Receptor a Brake: Slowing Gastric Emptying by GLP-1

The conventional view of endocrinologists is that glucose regulation after a meal depends on the interplay of insulin and glucagon secretion, hepatic glucose production, and glucose disposal. However, in recent years the role of gastric motility and its effect on glucose appearance from the gut have been revived as a determinant of glucose tolerance (1,2). This is due in great part to the availability of diabetes therapeutics that act, in part, by delaying gastric emptying (GE). Yet it is important to note that much of the variance in oral glucose tolerance is accounted for by differences in GE rate, an observation made over two decades ago (1). Even a minor perturbation in GE carries a substantial impact on postprandial glycemia in healthy individuals (1) and those with diabetes (3), such that more rapid GE results in a greater initial glycemic response and slower gastric delivery of meal contents to the intestine leads to smaller glucose excursion (4).

Glucagon-like peptide 1 (GLP-1) normalizes glycemia by promoting glucose-dependent insulin secretion and inhibiting glucagon secretion in the fasting state (5). A deceleration of GE is also characteristic of GLP-1 action and occurs both in patients with type 2 diabetes (T2D) and healthy individuals (6–8). In fact, glucose lowering by GLP-1 was significantly attenuated when the actions on GE were overridden by treatment with erythromycin (9). This has led to questions as to whether islet or gastrointestinal effects of GLP-1 predominate in its effects on glycemia (10). Long-acting GLP-1 receptor (GLP-1R) agonists, recently developed and now commonly used in the treatment of T2D, stimulate insulin secretion and reduce GE (5), but the relative impact of these two physiologic actions on acute and chronic control of blood glucose is not clear.

Recent studies have raised the possibility that the effects of GLP-1R stimulation to regulate GE vary with the duration of exposure. For example, GLP-1R agonists with protracted plasma residence time and duration of action seem to have lesser effects on gastric motility and greater effects on islet hormone release compared with shorter-acting GLP-1–based drugs (11). Moreover, in studies with native GLP-1, a waning of the effect on GE has been noted. Nauck et al. (12) infused GLP-1 to healthy subjects over 8–9 h and measured GE and glycemia following sequential meals. They observed that the delay in GE and reduction in postprandial glucose seen with GLP-1 administration during the first meal were less pronounced after the second meal. While this result was established with an appropriate and conservative analysis, the effect size was small and the dye dilution method used to measure GE in this study is not considered to be as precise as other methodologies such as scintigraphy. Therefore, while raising the question that the GE effect of GLP-1 is subject to tachyphylaxis with prolonged elevation of plasma concentrations, confirmation of this finding is awaited.

In this issue, Umapathysivam et al. (13) examined the effects of prolonged, intermittent, and acute GLP-1 administration on GE and glycemia in 10 healthy men. They found that acute infusions of GLP-1, either singly (acute) or separated by 20 h (intermittent), delayed GE to a greater extent than a 24-h continuous infusion. The dose of GLP-1 used in the study was pharmacological and GE was measured by scintigraphy, using detection of ^99mTc calcium-phytate colloid–labeled meal retention rate in the stomach. Both intermittent and prolonged GLP-1 delayed GE when compared with placebo, but the 20-h period of no infusion in the intermittent paradigm had a significantly greater delay in GE when compared with continuous administration. Moreover, there was no difference in the effect of acute administrations of GLP-1 either at the beginning or the end of the intermittent protocol. Postprandial glucose and insulin tracked with GE and were lower during acute/intermittent GLP-1 infusion than during the extended administration. The clever study design used by the authors allowed for direct comparison of repeated short and prolonged GLP-1 infusion on GE in the same individual, increasing the statistical power of these observations. The other major strength of the study is the use of scintigraphy to measure GE, a method that has been well validated and produces reproducible results (14,15).

While the findings by Umapathysivam et al. seem incontrovertible, there are several limitations worth considering. First, the study used doses of GLP-1 that achieved plasma levels well above those occurring in the postprandial state (16), and better reflect pharmacology than physiology; the realistic extension of this study is to short- and long-acting GLP-1R agonists. Second, the study was carried out in healthy individuals so it is unclear whether the findings can be extrapolated to the population of diabetic patients who use GLP-1–based therapeutics. Third, despite a waning effect of 24 h of GLP-1 on GE, it is unclear whether this effect would be abolished if exposure of the GLP-1R agonist continued for longer, as occurs with some of the newer drugs that are dosed weekly. Finally, the interval between the intermittent GLP-1 infusions, 20 h, is significantly longer than average meal intervals or the administration of short-acting GLP-1R agonists such as exenatide. In practice, lengthening the interval of GLP-1R agonist administration could preserve its effect on GE but may lead to worsening glycemic control in subjects who ingest multiple meals per day. Despite these remaining questions, this study is consistent with the previous findings of Nauck et al. (12) and supports tachyphylaxis of the GLP-1 effect on GE. These findings have implications both for understanding the workings of the GLP-1R system and for clinical therapeutics.

The mechanism by which GLP-1 affects gastrointestinal motility is not fully understood but seems to be neurally mediated (17). For example, GLP-1 has been shown to inhibit gastropancreatic function by inhibiting central parasympathetic outflow (18); the effect of GLP-1 on GE is lost in subjects who underwent truncal vagotomy (19). Furthermore, plasma concentrations of pancreatic polypeptide, a surrogate measure of vagal nervous activity (20), show a similar pattern to what has been described here for GE with levels markedly suppressed by GLP-1 during a first meal, but the suppression was significantly less after the second test meal (12), suggesting an adaption of the autonomic nervous system to continuous GLP-1 administration. In contrast to the data supporting neural mediation of GLP-1 effects on the gut, there is little evidence to support direct actions on gastric GLP-1R. Recent studies of GLP-1R distribution have noted expression in the gastric mucosa (21), compatible with regulation of secretion but not motility. Given a current consensus that delayed GE caused by GLP-1R activity is neurally mediated, the waning of this effect with continuous exposure requires a central nervous system mechanism. In this light, it is notable that other drugs with actions that have some component of vagal mediation (e.g., nitroglycerine, antidepressants, and β2-adrenergic receptor agonists) are prone to tachyphylaxis (22,23), raising a precedent for the effects demonstrated in this article.

In the context of the findings on GE, it is of worth considering that other effects of GLP-1 do not seem to be susceptible to attenuation by continuous exposure. Effects of GLP-1 on insulin secretion have not been compared in a variable dosing paradigm such as that applied by Umapathysivam et al. (13). However, in subjects with T2D a 3-h infusion of GLP-1 stimulated insulin secretion to a greater degree than a 30-min exposure (24). In a seminal study, Zander et al. (25) gave GLP-1 to diabetic subjects for 6 weeks through a continuous subcutaneous infusion and noted significant improvement in plasma glucose, HbA_1c, insulin secretion, and insulin sensitivity that was persistent for 6 weeks. Moreover, experience with GLP-1R agonists that have extended durations of action (26) has demonstrated clinical efficacy that persists for months despite chronic elevations of plasma drug levels. These observations suggest that other key actions of GLP-1, presumably on islet hormone secretion, are not subject to tachyphylaxis. Thus, the current available data supports a model of centrally mediated effects of GLP-1, some of which, like GE, are blunted by continuous exposure and peripherally mediated actions, like insulin secretion, that are not (Fig. 1). Of course, this model requires considerable refinement. It is important to understand whether other actions of GLP-1 that are neurally mediated, such as satiety or nausea, also diminish with continuous exposure to GLP-1R agonists. It would also be useful to compare central and peripheral effects in the same subjects at the same time. In either case, the article by Umapathysivam et al. (13) provides a useful approach to these kinds of questions.