Are physicians neglecting the risk of heart failure in diabetic patients who are receiving sulfonylureas? Lessons from the TOSCA.IT trial.

Abstract

Many antidiabetic drugs are known to precipitate the onset of heart failure or aggravate the clinical course of the disease once it has become clinically apparent. Insulin has been associated with weight gain, and its use has been accompanied by an increased risk of heart failure.1 Pioglitazone and rosiglitazone are known to enhance sodium retention and have increased the risk of heart failure in randomized controlled clinical trials.2 Glucagon-like peptide-1 receptor agonists have worsened the clinical course of patients with established heart failure,3-5 and certain dipeptidyl peptidase-4 inhibitors (saxagliptin and alogliptin) have been associated with an increased risk of new-onset heart failure;6 this finding has led regulatory agencies to mandate that a warning about the risk of heart failure be included in the labelling of these drugs. Interestingly, all antidiabetic drugs that have been associated with worsening heart failure exert their antihyperglycaemic effect through potentiation of the action of insulin. In contrast, the use of sodium–glucose cotransporter 2 inhibitors, whose glucose-lowering effect is not insulin-dependent, has led to a decrease in the risk of heart failure.7 These observations support the hypothesis that heart failure is an inherent risk of antidiabetic medications that signal through insulin, in part because insulin enhances sodium reabsorption by the kidney.8 Sulfonylureas are the most commonly prescribed class of oral antihyperglycaemic drugs that exert their glucose-lowering effect through the action of insulin and, for many decades, they have served as a cornerstone of treatment of type 2 diabetes. Their importance in clinical practice has been maintained, even though large-scale trials (that evaluated the first-generation members of this drug class) raised concerns about their cardiovascular safety.9 However, worries about the potential for an adverse cardiovascular effect of sulfonylureas have largely focused on the possibility that their use might increase the likelihood of atherosclerotic ischaemic events, particularly myocardial infarction. Comparatively little attention has been paid to the effect of sulfonylureas on the risk of new-onset or worsening heart failure. This neglect is difficult to understand, since many patients with type 2 diabetes have heart failure, and the onset of heart failure contributes substantially to the morbidity and mortality of those with glucose intolerance. Concerns that physicians might be ignoring the possibility that sulfonylureas increase the risk of heart failure have been highlighted by the recent publication of the Thiazolidinediones Or Sulfonylureas Cardiovascular Accidents Intervention Trial (TOSCA.IT).10 TOSCA.IT was a multicentre, randomized, open-label clinical trial, which enrolled middle-aged patients with type 2 diabetes who were taking metformin. Patients were randomly assigned to either treatment with a thiazolidinedione (pioglitazone) or to a sulfonylurea (glibenclamide, glimepiride, or gliclazide), the choice of the sulfonylurea being determined by local practice. The primary endpoint focused on the occurrence of atherosclerotic ischaemic events (cardiovascular death, non-fatal myocardial infarction and non-fatal stroke). Over a median duration of follow-up of 4.8 years, the investigators reported no difference between the treatment groups on this outcome. Yet, the most intriguing finding of the TOSCA.IT trial was not the reported effect on the primary endpoint, but its observations concerning the effects of the randomized treatments on the risk of heart failure. Prior to the start of the trial, the investigators were concerned about the occurrence of heart failure, and thus, they excluded from participation any patient who had a history of any prior occurrence of clinically overt cardiac dysfunction (even those with so-called class I symptoms). However, inexplicably, the researchers did not include heart failure in the definition of the primary endpoint or the key secondary endpoint. Only after the trial was ongoing did the data and safety monitoring board take the unusual step of recommending that heart failure be designated as a stand-alone endpoint, and furthermore, the board requested that its occurrence be adjudicated by the endpoint committee. The TOSCA.IT investigators reported that the incidence of heart failure in the trial was low and was not significantly different in the two treatment groups. A difference in the risk of heart failure between the two groups would have been difficult to discern because of the small number of events. Yet, as reported by the investigators in their supplementary material, a majority of the adjudicated heart failure events in the pioglitazone group led physicians to the decision to discontinue treatment. In contrast, none of the heart failure events in patients taking a sulfonylurea in this open-label trial led practitioners to stop the study medication. The onset of heart failure is a serious complication for any patient, particularly a patient with diabetes. In the TOSCA.IT trial, physicians were aware of each patient's treatment assignment, but only heart failure events that occurred in the pioglitazone group triggered a response by the clinician. Is that because the heart failure events were more serious in the pioglitazone group? Or is it because physicians did not consider the possibility that sulfonylureas can contribute to the occurrence of heart failure in diabetic patients? The likelihood of observer bias is strengthened by the interpretations of the data by the investigators in their published report. The investigators concluded that there was no difference in the risk of heart failure in the two treatment groups, seemingly to assuage concerns about the known risk of heart failure in patients treated with thiazolidinediones.2 Yet, any conclusion that the two treatment groups had a similar risk of heart failure in the trial should not be reassuring. Since thiazolidinediones are known to increase the risk of heart failure, any inference by the TOSCA.IT investigators that the risks of heart failure were similar in the two groups not only fails to assuage concerns about pioglitazone, but it should also cause physicians to worry that sulfonylureas increase the risk of heart failure when prescribed to diabetic patients in the doses typically used in clinical practice. Certainly, based on the decisions and conclusions of the TOSAC.IT trial, neither the investigators who designed and reported the results of the study nor the physicians who recruited patients appeared to believe that a causal relationship between sulfonylureas and heart failure exists. The possibility that sulfonylureas adversely affect the evolution and progression of heart failure is consistent with prior experimental and clinical evidence. The ATP-sensitive potassium channel normally acts to reduce calcium overload during periods of stress.11 Sulfonylureas bind to an atypical ATP-binding cassette protein that inhibits the activity of the channel, and thus, exert actions that would be expected to be detrimental to myocardium under stress.11 Interestingly, mutations in the genes encoding for the sulfonylurea cassette protein increase the susceptibility to heart failure, and expression of the sulfonylurea subunit is increased in the failing heart.11 In addition, sulfonylureas have been consistently associated with an increased body weight,10 which is likely related (in part) to an action of insulin to cause sodium retention by the kidney.8 The results of both clinical trials and observational studies are consistent with the finding that sulfonylureas increase the risk of heart failure. Glucose-lowering with the use of oral antihyperglycaemic drugs that are insulin-dependent has been accompanied by a significant increased risk of weight gain and heart failure; the risk of heart failure increases by 7.1% for each 1.0 kg increase in body weight.12 In the UK General Practice Database, the use of sulfonylureas was associated with an 18–30% increased risk of heart failure.13 In an analysis of the National Veterans Health Administration database, the prescribing of sulfonylureas was associated with a 32% increase in the risk of heart failure.14 The strengths of this latter report are its focus on new users of antidiabetic medications and its reliance on propensity score matching to exclude a potential effect of unmeasured confounders. Its findings have been supported in a meta-analysis of observational studies, which noted an increased risk of heart failure of 17% in any users and 22% in new users of sulfonylureas.15 Furthermore, the risk of heart failure appears to be dependent on the dose of the sulfonylurea, with larger doses leading to higher risks.16 Finally, in patients with established heart failure, use of a sulfonylurea has been associated with an increased risk of death.17 The totality of available evidence suggests that sulfonylureas increase the risk of heart failure and its consequences, when prescribed to patients with type 2 diabetes. This risk has been identified in both randomized trials and observational studies, and experimental data have identified potential mechanisms that might mediate such a deleterious effect. Despite the consistency of the findings in the medical literature, practitioners in the community seem to be unaware of the link between sulfonylurea use and heart failure, and they generally do not associate the onset of heart failure with an adverse effect of treatment. All antidiabetic drugs that act through insulin can cause worsening heart failure, and sulfonylureas (which are insulin secretagogues) do not appear to be an exception. Given the widespread use of these drugs and the enormous importance of cardiac dysfunction in diabetes, recognition by clinicians of the link between sulfonylureas and heart failure is urgently needed. Conflict of interest: M.P. has recently consulted for Amgen, AstraZeneca, Bayer, Boehringer Ingelheim, Cardiorentis, Celyad, Daiichi Sankyo, Gilead, NovoNordisk, Novartis, Relypsa, Sanofi, Takeda, and ZS Pharma. None of these relationships have any connection with the topic of this paper.