A Critical Analysis of the Clinical Use of Incretin-Based Therapies: The benefits by far outweigh the potential risks

In addition to the influence of GLP-1–based medications on those outcomes that typically determine the morbidity and mortality of patients with type 2 diabetes, i.e., that affect large proportions of such patients, there may be additional safety concerns of special interest. Some signals have suggested an untoward influence of such treatment on the risk for certain rare conditions. For GLP-1 receptor agonists and for DPP-4 inhibitors, these events of special interest are pancreatitis, pancreatic cancer, and thyroid carcinoma (Table 1). In addition, possible consequences of a rise in pulse rate with GLP-1 receptor agonists need to be discussed.

Cases of pancreatitis have been observed in animals (6,7,13) and patients (14) treated with incretin mimetics and DPP-4 inhibitors (5). The questions are whether pancreatitis occurs more often in association with treatment using GLP-1–based medications, and whether it is causally related to such treatment.

Animal studies describe histological changes compatible with damage to the exocrine pancreas with exenatide (6,7) and sitagliptin (5). A similar study examining liraglutide did not confirm such damages induced by an incretin mimetic (13). Other studies find an amelioration of the course of experimentally induced acute pancreatitis in mice with exenatide (15) or an anti-inflammatory pattern of cytokines induced by liraglutide treatment (16). Another open question is whether these findings are representative of human acute or chronic pancreatitis.

Attempts to quantify the number of pancreatitis events while patients are treated with GLP-1 receptor agonists or DPP-4 inhibitors have found odds ratios (ORs) around 1, however with relatively wide CIs (Fig. 1A and C) (17–21). Such data have been taken from claims databases and correlating the prescription of drugs used to treat diabetes with a diagnosis of acute pancreatitis. These analyses have made it clear that obese, type 2 diabetic subjects are more prone to developing acute pancreatitis than the nondiabetic population. On the other hand, one single study reports a more than tenfold excess of pancreatitis in patients using exenatide or sitagliptin (22). This notable exception is a study based on an analysis of the U.S. Food and Drug Administration (FDA) Adverse Event Reporting System (FAERS [formerly known as AERS]). This study, thus, is in obvious contradiction to other epidemiological data. It summarizes reports to the FAERS from a period when publications and changes to drug labels had alerted the medical community to the fact that cases of acute pancreatitis had occurred in patients treated with exenatide and sitagliptin. This probably has prompted some reporting bias. The quality of the individual reports to the FAERS may also be questioned. The standards for the diagnosis of acute pancreatitis (at least two out of three criteria: 1) typical severe abdominal pain, 2) elevations in pancreas-specific enzymes such as amylase and lipase, and 3) typical findings using appropriate imaging procedures [23]) may not always have been applied.

A recent case-control study reported a higher OR for the risk of hospitalization for a diagnosis of pancreatitis in patients taking “incretin-based medications,” since a separate analysis for the use of exenatide (GLP-1 receptor agonist) or sitagliptin (DPP-4 inhibitor) did not yield significant findings (Fig. 1B and D). It cannot be excluded at present that a potential combination of stomach cramps, representing gastrointestinal adverse events of GLP-1 receptor agonists, and spontaneously elevated serum lipase activities were responsible for the hospitalizations and do not represent true pancreatitis episodes. Nevertheless, this small study analyzing only a few patients with pancreatitis is only the second study describing an elevated risk (24), however, only by combining exenatide and sitagliptin treatments as “incretin-based medications” and after adjusting for potential confounders. Figure 1B and D presents ORs and P values (all nonsignificant) calculated without adjustment for potential risk-modifying factors.

Furthermore, treatment with the GLP-1 receptor agonist liraglutide can lead to elevations in lipase without associated symptoms of pancreatitis (25). Using such enzyme measurements to “screen” for pancreatitis may have resulted in false diagnoses of pancreatitis because elevations in pancreatic enzymes do not have the degree of specificity that would be necessary to make it a helpful screening instrument. Indeed, elevated lipase and amylase activity is found quite frequently in patients with type 2 diabetes with an absence of abdominal pain (26). Under these circumstances, most elevated amylase or lipase levels would be chance findings without any relationship to inflammatory changes within the exocrine pancreas. However, the nature of the elevation in serum lipase induced by liraglutide treatment needs to be explored so that we can understand its mechanism. At least this phenomenon indicates an interaction of GLP-1 receptor agonists with the exocrine pancreas, perhaps indicating the presence of GLP-1 receptors in this compartment. Effects of GLP-1 receptor stimulation on pancreatic enzyme synthesis, potential leakage into the circulation rather than directional secretion into pancreatic digestive juice, and a potential induction of a chronic inflammatory response need to be studied. To date, it certainly cannot be taken as a fact that chronic stimulation of the GLP-1 receptor (as occurs during the treatment with incretin mimetics and DPP-4 inhibitors) induces acute or chronic inflammatory responses in the pancreas, nor that, based on a well-delineated mechanism and supported by convincing epidemiological data, the clinical use of incretin-based glucose-lowering medications would cause pancreatitis. Clinically, the development of typical chronic pancreatitis diagnosed because of typical morphological findings and exocrine insufficiency leading to maldigestion, nutritional deficiencies, and weight loss over and above what is expected from continued stimulation of brain GLP-1 receptors (1,27) in patients treated with GLP-1–based medications has never been described.

Regarding the related question of chronic changes in the exocrine pancreas leading to pancreatic duct proliferation and the formation of preneoplastic lesions (like pancreatic intraepithelial neoplasms or pancreatic duct glands [28]), data from animal studies are similarly controversial with studies showing alterations of the exocrine pancreatic histology indicative of chronic pancreatitis with exenatide treatment (5–7), while another recent study using liraglutide did describe occasional pancreatitis as a rare finding—but not at all related to the dose of liraglutide—with similar numbers in placebo-treated rats, mice, and monkeys (13). It appears highly unlikely that there should be a difference intrinsic to the two GLP-1 receptor agonists used (exenatide vs. liraglutide). A recent finding reported that pancreas specimens from organ donors with type 2 diabetes, who had received treatment with the DPP-4 inhibitor sitagliptin (n = 7) or exenatide (n = 1), relative to patients with type 2 diabetes treated with other agents, had marked β-cell hyperplasia, β-cells coexpressing insulin and glucagon, hyperplasia of α-cells expressing glucagon, increased expression of proliferation markers, and an increased prevalence of preneoplastic lesions (29). This finding needs to be confirmed in a larger, representative sample of pancreas specimens obtained without preceding long-term critical illness, which alone may be responsible for some proliferative responses (30).

To put the state of this present discussion into perspective, it should be made clear that at most early proliferative or preneoplastic changes have been observed, which as such are not proof that eventually the process described will give rise to pancreatic cancer. Thus we have to discuss a potential risk (and certainly want to learn more about the long-term consequences of stimulating GLP-1 receptors for the exocrine pancreas), but not an actual threat to patients treated with incretin-based medications, based on a well-characterized mechanism with a risk clearly elevated based on sound epidemiological analyses. It is reassuring that no case of clinically evident chronic pancreatitis has been described after initiating treatment with incretin-based medications. Certainly, there is also no case report of pancreatic cancer diagnosed after exposing a patient to GLP-1 receptor agonists or DPP-4 inhibitors in a patient in whom there had previously been a morphologically tumor-free pancreas. Since pancreatic carcinomas develop slowly (31), one would probably not expect to see such a case after at most a few years of treatment, considering the recent introduction of the incretin-based medications, even if there were such a long-term risk.

GLP-1 receptor agonists have the potential to induce proliferative changes in rodent thyroid C cells. Liraglutide increased the number of cases with C-cell hyperplasia, adenomas, and medullary thyroid carcinomas in mice and rats (9). In these species such abnormalities are also found spontaneously, i.e., in the absence of GLP-1 receptor stimulation, especially in male rats, in which medullary thyroid carcinoma developed in some animals treated with placebo (9). Accordingly, rodent C-cell lines in cell culture responded to GLP-1, exenatide, and liraglutide with acutely producing cyclic AMP and secreting calcitonin (9). Similar cell lines of human origin do not show such acute responses when GLP-1 receptors are stimulated (9). Whereas rodent C-cell lines are equipped with GLP-1 receptors at a high level of expression, this is not the case in their human counterparts (9). Along the same lines, long-term treatment in obese human subjects with high liraglutide doses up to 3-mg per day does not lead to elevations in plasma calcitonin (32). Based on these results, the ability of GLP-1 receptor stimulation to induce proliferative responses in human C cells has been judged as probably absent. Medullary thyroid carcinomas are an extremely rare form of thyroid carcinomas in humans (33). No case report has been published describing a medullary thyroid carcinoma in a patient receiving a treatment with a GLP-1 receptor agonist who prior to such treatment had a morphologically normal thyroid gland and low calcitonin concentrations. Given the rare incidence of medullary thyroid carcinoma, 1) the consequences of a potential elevation in the risk induced by incretin mimetics would still remain small, and 2) to prove or exclude such a relationship, efficient surveillance of extremely large numbers of patients would be needed.

The elevated risk for thyroid carcinoma in more general terms described in the study exploring the FAERS database (22) is difficult to reconcile. Similar reservations apply regarding reporting bias as mentioned for the pancreatitis/pancreatic carcinoma issue raised earlier (vide supra). Certainly, this would not be compatible with an explanation through a higher number of medullary carcinomas alone, which would need to increase by more than 30-fold in order to explain such numbers. However, whether follicular cells express GLP-1 receptors (9) or whether malignant cells from thyroid tumors of different histological varieties (e.g., papillary thyroid carcinomas) express the GLP-1 receptors (34) is controversial and may be related to the specificity of the antibody or the radioligand used for immunohistochemistry (35). The fact alone that some papillary thyroid carcinomas may show evidence of GLP-1 receptor expression (34) does not prove that such receptors and their stimulation by drugs may contribute to the genesis or proliferation of such tumors. Again, even a convincing case report is missing. Regarding the thyroid issues, certainly more investigations are required, but one hardly can conclude that, based on current knowledge, there is a definitely increased risk for medullary (or other types of) thyroid carcinoma with the use of GLP-1 receptor agonists. Nevertheless, patients with an individually elevate genetic risk should not be treated with such agents. An elevated risk when using DPP-4 inhibitors does not have to considered at all since no such findings have been reported (22).

In the absence of large-scale cardiovascular outcome trials, summaries of cardiovascular events reported as adverse events in clinical trials with incretin-based glucose-lowering medications and meta-analyses based thereon (36) are the best available source of information for an overall judgment at present. Phase 3 studies have accrued a number of cardiovascular events sufficient for a preliminary judgment based on trends. These trends observed for the incretin mimetics exenatide (37) and liraglutide (38) as well the DPP-4 inhibitors sitagliptin (39), vildagliptin (40), saxagliptin (41), linagliptin (42), and alogliptin (43) are surprisingly similar. As shown in Fig. 2, in all these analyses the relative risk for a combined end point composed of acute myocardial infarction, stroke, and cardiovascular death is reduced with any of the GLP-1–based medications relative to placebo or comparator treatment to a value below 1 (Fig. 2). However, the 95% CIs ranged to above 1.0 with most compounds, indicating that the number of events available for this analysis was too small to allow the definite conclusion of a significant improvement in cardiovascular prognosis with incretin-based glucose-lowering treatment.

A potential reduction in cardiovascular event rates with linagliptin treatment is further supported by a recent study comparing linagliptin with the sulfonylurea glimepiride (44).

There is some plausibility based on the influences of GLP-1–based drugs on cardiovascular risk factors (45). GLP-1 receptor agonists reduce body weight by reducing appetite and food intake. They also reduce systolic blood pressure by 2–5 mmHg, mechanistically explained by improved endothelial function and vasodilation, enhanced natriuresis, and fluid excretion. There is a potential for a reduction in postprandial triglyceride-rich lipoproteins, especially with those agents that have and preserve over time a prominent effect on gastric emptying. Effects on “nonclassical” cardiovascular risk factors point in the same direction. Furthermore, GLP-1 receptor stimulation has reduced the extent of myocardial necroses in animal experiments inducing acute myocardial infarction by coronary artery ligation. The results have been surprisingly uniform using different agents (GLP-1, exenatide, liraglutide, sitagliptin) in various species (45). In addition, in animal models of left ventricular failure, GLP-1 and incretin mimetics may increase cardiac output by stimulating glucose and oxygen uptake into the myocardium. Clinical pilot trials support the notion that GLP-1 receptor stimulation may be beneficial in patients with acute coronary syndrome and chronic congestive heart failure (45). Therefore, one may be optimistic that cardiovascular outcome trials being performed to date, which will report after the year 2015 (Table 2), will at least confirm cardiovascular safety with a potential to substantiate the beneficial effects in this important respect.

Table 1 summarizes the beneficial effects of incretin-based glucose-lowering agents and their advantages over other antidiabetic pharmaceutical agents, but also the open issues discussed earlier in this article in order to define the balance of benefits on the one hand and the risks and harms on the other. Regarding the properties of incretin-based medications as antidiabetic drugs, they are effective in lowering glucose and avoid the problems of some other classes of glucose-lowering medications that are related to the induction of hypoglycemia and weight gain. Surrogate parameters indicate an improvement in the cardiovascular risk profile, and preliminary analyses of cardiovascular outcomes suggest the potential for benefit in this respect. Critical issues exist, but in many respects they are discussed in a controversial manner with only some data in support of an elevated risk (Table 1). Nausea and vomiting may be intolerable and lead to the discontinuation of treatment with GLP-1 receptor agonists. Putative interference of DPP-4 inhibitors with immune function does not lead to increased rates of common infections (39,46). Regarding the issues related to the potential short-term induction of acute and the putative long-term risk for chronic pancreatitis and eventually pancreatic cancer, data at hand today do not convincingly prove such risks. Thyroid issues related to GLP-1 receptors on C cells appear to mainly apply to rodents with a paucity of convincing human data that show a definite risk. This applies even more so to other forms of thyroid cancer. The fact that heart rate may increase with GLP-1 receptor agonists needs to be understood mechanistically. Potential explanations could be a reflex compensating for vasodilation (47) and lower blood pressure (10,11,48), a direct effect on the sinus node, or an increased relationship of sympathetic versus parasympathetic autonomous nervous system tone. Epidemiological findings relating higher heart rates to premature cardiovascular morbidity and mortality probably use pulse rate as a surrogate parameter for physical fitness (49). There is no reason to assume that incretin-based medications would lead to a reduced cardiorespiratory fitness. A lower body weight speaks against this hypothesis.

Thus, while the benefits—expected or proven—from using incretin-based medications seem to be substantial and address risks central to patients with type 2 diabetes, the potential harms and risks typically refer to rare events and are discussed in a controversial manner, e.g., without certainty regarding a potential role of incretin-based medications to cause substantial harm. Obviously more needs to be learned regarding the open questions, but based on today’s available knowledge, incretin-based medications can be considered effective and safe. Safety concerns related to the exocrine pancreas and the thyroid are not substantiated enough. Such considerations should not currently influence our treatment decisions regarding the potential prescription of GLP-1 receptor agonists or DPP-4 inhibitors within a treatment regimen for type 2 diabetes.

M.A.N. has received research grants (to his institution, the Diabeteszentrum Bad Lauterberg) from Berlin-Chemie AG/Menarini, Berlin, Germany; Eli Lilly & Co., Indianapolis, Indiana; Merck Sharp & Dohme, München, Germany; and Novartis Pharma AG, Basel, Switzerland (mono- or oligocentric studies); and from AstraZeneca, Södertälje, Sweden; Boehringer Ingelheim, Ingelheim, Germany; GlaxoSmithKline, King of Prussia, Pennsylvania; Lilly Deutschland GmbH, Bad Homburg, Germany; MetaCure Inc., Orangeburg, New York; Roche Pharma AG, Grenzach-Wyhlen, Germany; Novo Nordisk Pharma GmbH, Mainz, Germany; and Tolerx Inc., a Delaware Corporation, Cambridge, Massachusetts, for participation in multicentric clinical trials.

He has received consulting fees or/and honoraria for membership in advisory boards or/and honoraria for speaking from Amylin Pharmaceuticals, Inc., San Diego, California; AstraZeneca, Mjölndal, Sweden; Berlin-Chemie AG/Menarini, Berlin, Germany; Boehringer Ingelheim, Ingelheim, Germany; Bristol-Myers Squibb EMEA, Rueil-Malmaison, France; Diartis Pharmaceuticals, Inc., Redwood City, California; Eli Lilly & Co., Indianapolis, Indiana; F. Hoffmann-LaRoche Ltd., Basel, Switzerland; GlaxoSmithKline LLC, King of Prussia, Pennsylvania; Intarcia Therapeutics, Inc., Hayward, Calfornia; Lilly Deutschland GmbH, Bad Homburg, Germany; MannKind Corp., Danbury, Connecticut; Merck Sharp & Dohme GmbH, München, Germany; Merck Sharp & Dohme Corp., New Jersey; Novartis Pharma AG, Basel, Switzerland; Novo Nordisk A/S, Bagsværd, Denmark; Novo Nordisk Pharma GmbH, Mainz, Germany; Sanofi Pharma, Bad Soden/Taunus, Germany; Takeda, Deerfield, Illinois; Versartis, Sunnyvale, California; and Wyeth Research, Collegeville, Pennsylvania; including reimbursement for travel expenses in connection with the above-mentioned activities. He owns no stock and is employed by Diabeteszentrum Bad Lauterberg, Bad Lauterberg im Harz, Germany. No other potential conflicts of interest relevant to this article were reported.

The author thanks Ute Buss for help with retrieving literature and Marion Männel and Marion Masekowitz (all from Diabeteszentrum Bad Lauterberg) for secretarial assistance. The author also thanks Juris Meier (Division of Diabetology and Gastrointestinal Endocrinology, Medizinische Klinik I, St. Josef-Hospital, Klinikum der Ruhr-Universität Bochum, Bochum, Germany) for helpful discussions.