|Home | About | Journals | Submit | Contact Us | Français|
Incretin-based therapies are effective glucose-lowering drugs that have an increasing role in the treatment of type 2 diabetes, due to their efficacy, safety, and ease of use. Both glucagon-like peptide-1 receptor agonists (GLP-1Ra) and dipeptidyl peptidase-4 inhibitors (DPP-4i) are commonly used for glycemic control as an adjunct to metformin, other oral antiglycemic agents, or insulin. GLP-1Ra may have additional effects, such as weight loss, that may be advantageous in obese patients. There is a large body of evidence from randomized controlled clinical trials supporting the cardiovascular safety of DPP-4i and some GLP-1Ra, at least in the short term. However, concerns have been raised, particularly regarding their safety in patients with heart failure. In this review, we have provided a brief, but practical evidence-based analysis of the use of incretin-based agents in patients with diabetes, their efficacy, and cardiovascular safety.
Type 2 diabetes mellitus (T2DM) is characterized by progressive decline in β-cell function and increased risk of cardiovascular (CV) complications (1). Although diet, exercise and weight loss are considered central tenets in management of T2DM, pharmacological approaches are almost always required. The ease of use of incretin agents, their glycemic efficacy, low to no risk for hypoglycemia, and ancillary benefits may allow their earlier incorporation with other evidence-based therapies in the treatment of T2DM. However, recent reports of these agents being associated with heart failure (HF) have created considerable confusion (2, 3). This review critically evaluates the evidence supporting the efficacy (glycemic control, weight loss, lipoprotein effects, blood pressure, and CV events) and safety of currently available incretin agents (dipeptidyl peptidase-4 inhibitors [DPP-4i] and glucagon-like peptide-1 receptor agonists [GLP-1Ra]) from randomized controlled trials (RCTs) published from 2008 onwards. The intent is to provide a succinct overview of the use of these agents for the CV practitioner.
Incretins are gut-derived members of the glucagon superfamily, released in response to nutrient ingestion (mainly glucose and fat). Glucagon-like peptide-1 (GLP-1) and gastric inhibitory peptide (GIP) are the major physiological incretins (4). Secreted by the small intestine, these peptides amplify the insulin secretory response to nutrients through their cognate G protein-coupled receptors in the pancreatic β cell (5). Together, GLP-1 and GIP account for 50% to 70% of total insulin secreted following an oral glucose load, and contribute to the control of postprandial hyperglycemia (4). The effects on insulin secretion are maintained only until a “normal” threshold level of plasma glucose is achieved, thus minimizing risk for hypoglycemia. GLP-1 receptor signaling involves activation of adenylyl cyclase and cyclic adenosine monophosphate (cAMP)-dependent activation of protein kinase A, which promote insulin secretion. Recent evidence at physiologically relevant concentrations of GLP-1 (picomolar range) provides evidence of restoration of β cell competence by these hormones through increased membrane excitability mediated by ion channels (6). The incretin effect is markedly diminished in T2DM, due to resistance at the level of the β cell, except in late stages, when there may be a reduction in plasma GLP-1/GIP levels. Incretins, such as GLP-1, also address hyperglucagonemia in T2DM by inhibiting glucagon secretion and reducing hepatic gluconeogenesis. Finally, pharmacological doses of GLP-1 are well known to decrease gastric emptying and appetite through central nervous system mechanisms, thus contributing to weight loss (Figure 1) (7).
Both GLP-1 and GIP have short half-lives in vivo (approximately 1 to 2 min) and are rapidly degraded to their inactive forms (GLP-1[9–36] and GIP[3–42] in humans) by the peptidase DPP-4. DPP-4 is a widely expressed serine peptidase that inactivates peptides with an alanine, proline, or serine residue in the penultimate position from the N-terminus (5). DPP-4 catalytic inhibition elevates GLP-1 and GIP levels, although the extent of elevation (picomolar) is small compared to pharmacological supplementation with GLP-1 analogs (nanomolar). These pharmacokinetic considerations may, in part, explain effects, such as weight loss and gastric slowing, with GLP-1Ra that are not seen with DPP-4i (Table 1). DPP4 is also involved in the catalytic inactivation of a number of other peptides, which may exert effects independent of GLP-1/GIP levels (Central Illustration) (8).
The approved agents for the management of T2DM, their pharmacology, dosing, adverse effects, and costs are outlined in Online Table 1. Exenatide is the 39-amino acid synthetic version of exendin-4 (originally isolated from the Gila monster, a species of venomous lizard native to the southwestern United States) that is resistant to DPP4 degradation. Exenatide long-acting release (LAR) is a depot formulation of exendin-4, entrapped noncovalently into biodegradable poly-D,L-lactide-co-glycolide microspheres from which the drug is gradually released, increasing the half-life to 6 days. Liraglutide is a GLP-1 (7–37) analog, containing a fatty acid chain at a lysine residue in position 26 through a γ-glutamyl spacer, which facilitates binding to albumin, increasing its half-life. Albiglutide and dulaglutide are synthetic fusion proteins composed of dimers of degradation-resistant GLP-1 with a linker protein that prolongs half-life. There are currently 4 DPP-4i (“gliptins”) approved by the Food and Drug Administration (FDA) (Central Illustration). All gliptins are orally available, low-nanomolar selective inhibitors and do not interfere with other members of the DPP family. The plasma half-lives of these agents do not reflect tissue half-life, as most DPP-4 is tissue-bound. Online Table 2 summarizes agents under clinical development, including once-weekly DPP-4i.
Both short-acting and long-acting GLP-1Ra as monotherapy significantly reduce glycosylated hemoglobin (HbA1c) by 0.4% to 1.6%, improve fasting and postprandial hyperglycemia, and increase the proportion of participants reaching a HbA1c target of ≤7%. When compared head-to-head against DPP-4i or sulfonylurea (SU), in patients who either failed metformin or when metformin was used as an adjunctive therapy, GLP-1Ra achieved greater HbA1c lowering with additional weight loss (9–13). When compared against metformin or TZD, GLP-1Ra may provide 0.2% to 0.6% incremental HbA1c lowering (10, 14–16). Weight loss with both short-acting and long-acting GLP-1Ra ranges from 1.4 to 4 kg, at the end of 6 months compared to baseline weight (9–13). Weight loss is seen as early as 4 weeks, with a plateau in 4 to 6 months (9). In a large meta-analysis of all GLP-1 trials up to 2011, the weighted mean difference following GLP-1Ra therapy was −2.9 kg (95% CI: –3.6 to –2.2; 21 trials, 6,411 participants). The weight-loss effects of short-acting versus long-acting agents appear to be comparable (17). It may be reasonable to recommend GLP-1Ra in combination with metformin, or as an alternative to other agents, depending on patient preferences, need for weight loss, and potential for side effects (see later discussion).
Patients who cannot manage an insulin regimen and/or needing additional weight loss may be excellent candidates for GLP-1Ra. In studies conducted versus basal insulin (glargine, detemir) in patients receiving 1 or more oral antidiabetic drugs (OADs), GLP-1Ra significantly improved HbA1c and reduced weight compared with insulin (18–22). Adjusted weight loss in the comparator group ranged from 2 to 5 kg.
The rationale for combining GLP-1Ra with insulin is on the basis of complementary pharmacological effects on prandial and fasting glycemia, additional benefits of attenuating weight gain/hypoglycemia, and the potential of replacing complex basal bolus regimens. Addition of GLP-1Ra to insulin (± OAD) significantly improves glycemic control in placebo-controlled studies and attenuates weight gain with insulin (23, 24).
When used as monotherapy, initial combination therapy, or in conjunction with single or ≥2 OADs, DPP-4i provided incremental glycemic lowering of 0.4% to 1.1%. Previous meta-analysis has shown that the degree of glycemia lowering is similar with the various approved agents (25). Compared with metformin monotherapy, DPP-4i were associated with a smaller decline in HbA1c and a lower chance of attaining the HbA1c goal of <7% (26, 27). When compared to SU (on metformin regimens) DPP-4i demonstrated inferior HbA1c lowering in short-term studies (28–31). In studies >52 weeks, DPP-4i, although noninferior, showed more HbA1c lowering (32). One possible explanation is that over longer durations, DPP-4i may allow better glycemic control, due, in part, to secondary failure of SU (33). In studies comparing DPP-4i with SU, DPP-4i have been associated with a modest weight loss and substantially lower risk for hypoglycemia (28–32, 34). Compared to thiazolidinediones (TZD), DPP-4i may provide poorer glycemic control, but are weight neutral compared with weight gain with thiazolidinediones (10, 15). In trials comparing DPP-4i versus sodium glucose cotransporter 2 (SGLT-2) inhibitors, either alone or on top of metformin ± SU, SGLT-2 inhibitors were superior, with more patients attaining the HbA1c goal of <7% with weight loss (35).
When used in conjunction with insulin ± 1 OAD, DPP-4i provide significant incremental glycemic lowering with attenuation of weight gain incurred with insulin (36–39). Thus, these agents can be used in conjunction with insulin, although there is evidence that efficacy diminishes in patients on multiple OADs with long duration of T2DM (40).
DPP-4i are an acceptable treatment option in the presence of chronic kidney disease (CKD) for glycemic lowering, especially when other OADs are contraindicated. This is a major advantage for this class of drugs. With the exception of linagliptin, however, all other approved DPP-4i require dosing adjustments. Exenatide and exenatide LAR is contraindicated in CKD with glomerular filtration rate (GFR) <30 ml/min (Online Table 1). Liraglutide is hepatically metabolized and can be used in patients with CKD, although there is limited evidence in advanced CKD.
FDA regulatory guidance around CV safety of antidiabetic agents has necessitated CV outcomes trials in T2DM (41). Prior to the FDA guidance, a number of studies assessed surrogate markers, which included lipoprotein measures, blood pressure, heart rate, and biomarkers of relevance, as ancillary measures in glycemia-lowering trials.
Prior reviews have also provided detailed discussions of the lipoprotein effects of GLP-1Ra and DPP-4i (5, 8). Although earlier small, brief studies have shown reductions in very low-density and remnant lipoproteins via GLP-1-mediated reduction in absorption of triglycerides and synthesis of Apo48 lipoproteins, often even with 1 dose, there has been only 1 double-blinded study assessing postprandial lipids conducted for >12 weeks. This showed a 43% reduction in postprandial triglycerides and a 50% reduction in postprandial chylomicrons (42). In the SCALE double-blind RCT, there were reductions in triglycerides, along with lower C-reactive protein and weight loss with liraglutide (43).
Clinical trials assessing the glycemic efficacy of GLP-1Ra have observed a small decrease in office systolic blood pressure (SBP) of 2 to 4 mm Hg when compared with placebo (14, 24, 44). This has been replicated in meta-analyses that have shown a 2 to 2.5 mm Hg reduction in SBP (45, 46). However, these findings have not always been replicable (12, 44, 47). In the only ambulatory blood pressure (BP) trial to test the effect of GLP-1Ra, dulaglutide 1.5 mg significantly reduced mean 24-h SBP by ≈ 3 mm Hg. The decrease in blood pressure (BP) was accompanied by an increase in pulse rate of 2 to 3 beats/min (48). The effects on pulse rate have been noted with both short- and long-acting agents against placebo (14, 24, 43). The mechanisms responsible for BP lowering with GLP-1Ra may relate to increased synthesis of atrial natriuretic peptide (ANP), natriuresis, and vascular relaxation (49). However, evidence of increase in ANP in humans with GLP-1Ra administration is lacking at this time (50). The mechanisms for the increase in heart rate are currently unclear (51, 52).
There have been 4 randomized double-blinded placebo-controlled clinical trials (3 for DPP-4i and 1 for GLP-1Ra) that have evaluated hard CV endpoints with incretin agents (Table 2). All 4 trials were designed, not as glycemia-lowering studies, but rather to establish CV safety. Optimization of the glycemic regimen was allowed in the placebo arm. The SAVOR (Saxagliptin Assessment of Vascular Outcomes Recorded in Patients with Diabetes Mellitus) trial tested saxagliptin (5 mg daily [qd], 2.5 mg if GFR ≤50 ml/min) versus placebo in patients with established CV disease or major risk factors over approximately 25 months (53). The study was designed as a superiority trial, with a pre-specified noninferiority comparison. Saxagliptin was not superior to placebo, but met the noninferiority criterion. The EXAMINE (Exploring the Cardiovascular Safety of Therapies for Type 2 Diabetes) trial compared alogliptin (25 mg qd or 6.25 to 12.5 mg if GFR <60 ml/min) versus placebo in acute coronary syndrome (ACS) (54). Major adverse CV events were similar with alogliptin and placebo (54). TECOS (Trial to Evaluate Cardiovascular Outcomes after Treatment with Sitagliptin) assessed the effect of sitagliptin (100 mg, or 50 mg if GFR = 30 to 50 ml/min, once daily) in established CVD whereas in ELIXA (Evaluation of Lixisenatide in Acute Coronary Syndrome), lixisenatide (10 to 20 µg) was compared versus placebo in patients post-ACS (55, 56). There are 5 ongoing RCTs of incretin therapies that will report in the near future (Online Table 3).
An unexpected finding in SAVOR was an increased incidence of HF hospitalization (a pre-specified endpoint) in saxagliptin compared with placebo (3.5% vs. 2.8%; hazard ration [HR]: 1.27; 95% confidence interval [CI]: 1.07 to 1.51), without excess HF-related mortality (53). Previous HF, estimated glomerular filtration rate (eGFR) <60 ml/min, elevated B-type natriuretic peptide (BNP), and albumin/creatinine ratio were the strongest predictors of HF hospitalization. This excess HF risk was not noted with alogliptin or lixisenatide, both of which were in a high-risk ACS patient population, assuaging concerns that increting agents, as a class, were causative. In EXAMINE, the first occurrence of hospitalization for HF occurred in 3.1% and 2.9% of alogliptin and placebo groups, respectively (1.07; 95% CI: 0.79 to 1.46; p = 0.68) (57). There was no excess of CV death. In TECOS, the hospitalization rate for HF was identical in sitagliptin and placebo (3.1% vs. 3.1%; HR: 1.00, 0.83 to 1.20; p = 0.98) (55). Lixisenatide also showed similar hospitalization for HF compared to placebo (HR = 0.96; 95% CI: 0.75 to 1.23) in ELIXA. On the basis of these trials, it is safe to conclude that the HF risk is small, and mainly with saxagliptin. The risk of HF with other GLP-1Ra remains to be determined.
The risk of hypoglycemia is low with both GLP-1Ra and DPP-4i and comparable to placebo when used as monotherapy or in combination with metformin/TZD. In the 3 large randomized controlled clinical trials on a background of multiple oral hypoglycemic agents, rates of serious hypoglycemia were comparable to placebo (53–55). The risk of hypoglycemia with DPP-4i/GLP-1Ra is nearly always with concomitant SU therapy and is higher in elderly patients. Thus, dose reduction of SU must be considered when initiating a GLP-1Ra/DPP-4i regimen in elderly patients (58, 59). The risk of severe hypoglycemia when GLP1Ra are combined with insulin is uncommon (20, 36).
Nausea and vomiting, presumably related to effects of GLP-1Ra on gastrointestinal (GI) slowing, are common, decrease over time, and occur more frequently with the shorter-acting preparations (exenatide > liraglutide > exenatide LA = dulaglutide > albiglutide) (52, 60–62). The mechanism appears to be related to GLP-1Ra pharmacokinetics, with abatement when steady-state levels are attained. GI side effects are uncommon with DPP-4i (Online Table 1).
Reactions at the injection site (nodules) are common with the longer-acting agents and typically subside by 3 to 4 weeks (10% to 30% of patients with longer-acting agents, such as exenatide LA versus <5% with short-acting analogs (Online Table 1).
There is no indication of excess risk with pancreatitis or pancreatic cancer in the 3 large trials with DPP-4i (53–55) and the single trial with GLP-1Ra, although the duration of follow-up might not be long enough to rule out a signal. According to a joint FDA-European Medicines Agency statement, there was no evidence that incretin therapy in humans is associated with C medullary thyroid cancer in humans (63).
Incretin-based therapies have an increasing role in the management of T2DM. The American Diabetes Association/European Association for Study of Diabetes, the American Association for Clinical Endocrinology guidelines, and International Diabetes Federation Guidelines recommend that GLP1a or DPP-4i be considered as alternatives to metformin or as combination therapy with metformin when glycemic targets are not reached (Online Table 4). An excellent indication for these agents is when either metformin, sulfonylurea or pioglitazone is contraindicated, not tolerated, or when avoidance of hypoglycemia and/or weight gain is a priority. A GLP-1Ra may be recommended over a DPP-4i, when weight loss is a priority (i.e., body mass index >30) and if greater reductions in HbA1c are desired. A general approach that may work well in many patients requiring a combination strategy is to include a DPP-4i or GLP-1 agonist as part of the regimen on the basis of patient-specific factors. In the elderly at risk for hypoglycemia or in patients with CKD, a DPP-4i may be a good choice, either alone or in combination with other agents. A DPP-4i or GLP-1a may also be used effectively in conjunction with TZD, SU, and SGLT-2 inhibitor. In addition, DPP-4i or GLP-1Ra may be considered as adjunctive therapy with basal insulin regimens in lieu of other, more complicated regimens.
Although an important consideration for antidiabetic agents is their CV benefit, this has not been seen within the duration (2 to 4 years) of currently designed studies (64). It is, however, possible that more delayed benefits may still occur (64). However, when considering treatments with potential delayed benefits, the costs, treatment burden, and adverse effects become especially important. Treatment burden and adverse effects can have an appreciable negative impact on patient quality of life, with prior studies showing that interventions with a high treatment burden for treating glucose (“glycemic disutility”) may substantially attenuate benefit, as measured by quality-adjusted life-years (65). At the current time, the clinician should focus the treatment approach around the patient, with individual preferences, patient disease variables and other factors (i.e., cost) also being considered, once glucose management targets are set.
Dr. Zhong was supported by grants from the NIH (DK105108), AHA (15SDG25700381 and 13POST17210033), and Mid-Atlantic Nutrition Obesity Research Center (NORC Pilot & Feasibility Program).
Disclosures: Dr. Rajagopalan reports personal fees from Takeda Pharmaceuticals, grants and personal fees from Boehringer Ingelheim, during the conduct of the study; other from Glaxo Smith Kline, outside the submitted work;.