Is Incretin-Based Therapy Ready for the Care of Hospitalized Patients With Type 2 Diabetes?: Insulin therapy has proven itself and is considered the mainstay of treatment

In contrast to the solid data supporting the use of insulin in hospitalized patients, there are only a handful of studies investigating the use of native glucagon-like peptide-1 (GLP-1) or GLP-1 receptor agonist (GLP-1RA) infusions in the inpatient setting. Most of these studies are uncontrolled, open-label, and of short-duration with small numbers of subjects with or without a history of diabetes (13–20) (Table 1). The majority of these studies were performed with the primary objective of investigating the safety and efficacy of these agents, with the hypothesis that GLP-1 infusions will achieve the desired glycemic targets without the risk for hypoglycemia. To date, none of these benefits have been demonstrated, and in many GLP-1–treated patients, rescue therapy with insulin was required to achieve and maintain the desired glycemic targets with no difference in the frequency of hypoglycemia when compared with insulin therapy. In addition, there was added risk for gastrointestinal side effects (nausea, vomiting, constipation), which occurred in up to 66% of subjects (15,16,18).

In one small study of 20 subjects (5 with type 2 diabetes) undergoing coronary artery bypass grafting, glycemic control and left ventricular function was compared between 10 subjects (2 with type 2 diabetes) randomized to GLP-1 infusions initiated 12-h preoperatively at a dose of 1.5 pmol/kg/min and continued for 48-h postoperatively, and 10 subjects (3 with type 2 diabetes) assigned to standard insulin therapy (16). Standard therapy was not clearly defined but included IV insulin infusion, which was given to 5 subjects in each group. GLP-1 resulted in better glycemic control in the pre- and perioperative periods compared with insulin therapy, but there were no differences in postoperative blood glucose (BG) levels between treatment groups. In addition, there were no differences in insulin levels, hemodynamic parameters, or the number of hypoglycemic events during the study period. Half of the GLP-1 subjects required exogenous insulin infusions to achieve the desired level of glycemic control during the postoperative period. In a different study, 20 insulin-naïve patients with type 2 diabetes were randomly assigned to 12-h infusions of GLP-1 (3.6 pmol/kg/min) or insulin following elective coronary artery bypass grafting procedures (14). Rescue therapy with insulin was common for BG >7.77 mmol/L in the GLP-1 group. The insulin group received more insulin and required more dose adjustments in the first 6 h following surgery. Mean BG levels were similar in each group (7.9 ± 0.3 vs. 8.1 ± 0.1 mmol) without difference in hypoglycemia.

There are several studies conducted primarily to investigate the effect of GLP-1 infusions on left ventricular function in patients following acute myocardial infarction complicated by severe systolic dysfunction and in patients with congestive heart failure (CHF) (15). Each of these reports provided information on glycemic outcomes in addition to cardiovascular parameters; however, none of these studies included an insulin comparison. In one small nonrandomized pilot study, 10 patients (5 with type 2 diabetes) received a 72-h infusion of GLP-1 (1.2 pmol/kg/min) in addition to standard therapy alone, the latter of which was provided to 11 control subjects (4 with type 2 diabetes) (16). GLP-1 infusion was associated with reductions in glucose and insulin levels, improvements in left ventricular function, but some nondiabetic subjects experienced hypoglycemia, and gastrointestinal side effects (anorexia, nausea, vomiting, and constipation). Similarly, in an open-label nonrandomized study, 6-week infusions of GLP-1 (1.25–2.5 pmol/kg/min) added to standard therapy in 12 subjects with CHF, 8 with type 2 diabetes resulted in improved cardiovascular parameters and lower glucose levels compared with standard therapy alone; however, there was more hypoglycemia and gastrointestinal side effects in the GLP-1–treated group (17). In another double-blind crossover study, 20 patients without diabetes with New York Heart Association II and III heart failure and ischemic heart disease were randomized to receive 48-h infusions of GLP-1 or placebo in random order (18). GLP-1 infusions resulted in more hypoglycemia with 8 patients experiencing 9 episodes of hypoglycemia (glucose <3.5 mmol/L) compared with none with placebo. Five patients were unable to complete the study because of severe nausea and vomiting and were excluded from analysis. Contrary to the results in prior studies, there was no beneficial effect of GLP-1 on left ventricular ejection fraction, diastolic function, or myocardial contractile function.

There are only two small uncontrolled pilot studies investigating the use of GLP-1RA for glycemic management in critically ill patients (19,20). One open-label study compared the efficacy and safety of SC administration of exenatide at doses of 5–10 μg every 12 h (n = 6) with standard intensive insulin therapy (n = 18) in severely burned pediatric patients without diabetes. Similar levels of glycemic control were achieved in both groups (130 ± 28 vs. 138 ± 25 mg/dL) (19); however, the dose of administered insulin was significantly lower in the exenatide group (22 ± 14 vs. 76 ± 11 unit/patients/day, P = 0.01). Three patients receiving exenatide required rescue therapy with SC insulin to maintain glycemic control. The number of BG determinations was identical, as was the incidence hypoglycemia (0.38 events/patient/month). There were no reported gastrointestinal side effects.

In another pilot nonrandomized, uncontrolled, open-label study evaluating the safety and efficacy of IV exenatide in 40 cardiac ICU patients, 75% with type 2 diabetes (20), subjects received an initial 30-min bolus of 0.05 µg/min followed by 0.025 µg/min for 24–48 h. Exenatide infusions resulted in similar mean steady-state BG and hypoglycemic events when compared with historic control subjects treated with IV insulin infusions targeting BG 90–119 mg/dL (n = 94) or 100–140 mg/dL (n = 39). The mean steady-state BG in the group treated with exenatide (139 ± 41 mg/dL) was similar to that achieved with IV insulin therapy (115 ± 36 mg/dL and 147 ± 52 mg/dL, respectively). Hypoglycemia (BG <70 mg/dL) was reported in 10% of patients receiving exenatide compared with 21 and 15% in those treated with IV insulin (P = 0.27). A total of 8 patients (20%) experienced nausea because of exenatide, and 6 patients (15%) requested early termination because of severe nausea.

Treatment with GLP-1 and GLP-1RA is associated with a high incidence of gastrointestinal side effects, as was observed in the majority of reported studies (Table 1). Nausea and vomiting can be potentially dangerous in hospitalized patients with altered sensorium, who are maintained in a supine position, or who receive sedating medications, all of which increase the risk for aspiration pneumonia. In addition, the risk for pancreatitis, although rarely reported with the GLP-1 therapy, cautions against the use of these agents in patients with abdominal pain or postsurgical ileus.

The observed increase in heart rate of 2–5 bpm reported in clinical trials with GLP-1 was also reported in several of the inpatient studies (17,18). The mechanism underlying the increase in heart rate has not yet been clarified, but in at least in one report was attributed to possible undetected hypoglycemia (18). Although preliminary cardiovascular safety analyses of GLP-1RA demonstrate trends toward reduced cardiovascular events (21,22), long-term studies are needed to determine the clinical relevance of these chronotropic effects, particularly in critically ill patients.

Current practice guidelines recommend against inpatient use of oral antidiabetic drugs in part because of the absence of efficacy studies as well as safety concerns (6–8). A major limitation to inpatient use of oral antidiabetic agents relates to the delay in and unpredictable onset of action, which prevents rapid attainment of glycemic control or dose adjustments to meet the changing needs of the acutely ill patient. The low risk of hypoglycemia and good tolerability of dipeptidyl peptidase 4 (DPP-4) inhibitors however make them attractive considerations for use in hospitalized patients. At this time however no randomized clinical trial studies have reported on the use of these agents in the hospital setting. A concern with the use of DPP-4 inhibitors is the increase in frequency of upper respiratory infections observed in some preclinical trials (23). There are no reports of an increase in serious infections, but this observation raises concern for the use of these medications in inpatients with compromised immune systems, such as those undergoing solid organ transplantation.

Since approval of exenatide and sitagliptin in 2005–2006, the use of GLP-1RA and DPP-4 inhibitors has been widely incorporated into clinical practice and treatment guidelines from leading diabetes organizations for use in outpatient settings. The question is whether or not these agents are safe for use in the inpatient population. It is important to emphasize that these are new medications for which even the long-term safety of their outpatient use remains to be established. It is therefore premature to make recommendations for their routine use in the inpatient population.

Insulin therapy remains the standard of care for achieving and maintaining glycemic control in hospitalized patients with critical and noncritical illness. Insulin is a known entity that has now been in clinical use for almost a century. Hypoglycemia, the major complication of insulin therapy, is avoidable in the majority of cases with appropriate use of either intravenous or basal-bolus insulin regimens with regular monitoring of bedside BG levels and modification of insulin dosage in response to changes in clinical and nutritional status (1,7,8).

The proposed reasons for using GLP-1 therapy for the management of hyperglycemia in hospitalized patients include a theoretical improvement in glucose control with low frequency of hypoglycemia, and less nursing time to monitor BG levels and adjust insulin doses. In the small numbers of patients studied to date, glycemic control with GLP-1 therapies has been shown to be superior to placebo, but not to insulin therapy (Table 1). In fact, almost 50% of subjects in some studies required rescue therapy with insulin in order to achieve and maintain glycemic control. The frequency of hypoglycemia was not reduced with GLP-1 therapies compared with insulin. In addition, in the studies investigating the frequency of BG monitoring and nurse time, no differences were found between insulin and GLP-1 therapies.

Novel drugs for the management of patients with diabetes and hyperglycemia are welcome when they are proven to be safe in improving glycemic control and in reducing cardiovascular and other complications. The preliminary experience with native GLP-1 is promising and has the potential to improve cardiac function in patients with heart failure and acute ischemic cardiovascular events (15–17). However, this requires further study in a larger number of patients. It is possible that these favorable results may extend to the use of GLP-1RA and DPP-4 inhibitors; however, until the safety and efficacy are addressed with large randomized controlled clinical trials, the routine use of these agents for inpatient glycemic control cannot be recommended.

History has taught us that despite the potential promising effects of some drugs, long-term and widespread use unmasks undesirable and in some cases life-threatening side effects resulting in the removal of these agents from clinical use. Recent examples from diabetes management occurred following the introduction of thiazolidinediones, a class of medications associated with effective glucose-lowering properties as well as potential cardiovascular benefits. However, troglitazone was removed from the market following reports of sometimes fatal idiopathic hepatic failure (24); and the use of rosiglitazone became highly restricted following reports of higher mortality because of ischemic cardiovascular events (25). More recent observational and clinical studies have shown an association between the use of thiazolidinediones and increased bone fractures (26). Until there are well-conducted published studies demonstrating both the efficacy and safety of using the potentially promising GLP-1RA or DPP-4 inhibitors in the inpatient setting, it is best to observe the sound principles of evidence-based medicine and adhere to the well-known saying from the 12th century: All that glitters is not gold.

G.E.U. is supported by the American Diabetes Association (7-03-CR-35) and the National Institutes of Health (UL1 RR025008) (Atlanta Clinical and Translational Science Institute).

G.E.U. has received research support from Sanofi and Merck Pharmaceuticals. M.K. has received research support from Sanofi. No other potential conflicts of interest relevant to this article were reported.