Diabetes trials should focus more on heart failure
Heart failure: a cardiovascular outcome in diabetes that can no longer be ignored
McMurray JJV, Gerstein HC, Holman RR, Pfeffer MA
The Lancet Diabetes & Endocrinology, online 13 March 2014. doi:10.1016/S2213-8587(14)70031-2
In the case of diabetes drugs, regulatory approval was generally based on the capacity of a glucose-lowering agent to safely and effectively lower HbA1c, since improving glycaemic control has been shown to substantially reduce the risk of microvascular complications, and to modestly decrease the risk of cardiovascular (CV) events with extended follow-up. HbA1c concentration is, however, a surrogate measure for CV risk, based on the prevailing assumption that lowering of glucose would in the long run lead to beneficial patient outcomes.
Recently, CV and diabetes researchers have advocated the importance of explicit measurement of macrovascular events during the evaluation of new drugs for diabetes. Changes in the regulatory process for assessment of new drugs have been announced accordingly. Both the FDA and EMA now require evidence that a new drug will not result in an unacceptable increase in CV risk.
In randomised trials CV safety is generally evaluated with major adverse cardiovascular events (MACE) as the primary outcome. MACE is typically defined as the composite of CV death, myocardial infarction (MI) or stroke, although other relevant CV events may be included according to the FDA and EMA guidance on this topic.
Many randomised trials are currently evaluating new glucose-lowering drugs. According to the authors of this viewpoint, a major concern with these trials is the universal absence of hospital admission for heart failure (HF) as a prespecified component of their primary composite CV outcomes. They believe that hospital admission for HF is one of the most common and prognostically important CV complications of diabetes, and the one CV outcome for which the risk has been shown unequivocally to be increased by certain glucose-lowering therapies. In this viewpoint, the authors outline why they believe that heart failure should be systematically evaluated in cardiovascular outcome trials of new glucose-lowering drugs, either as a component of the primary composite outcome or as a prespecified secondary endpoint.
After elaborating on what heart failure is and how it may develop, the potential but unclear causal role of diabetes in HF is discussed. The prevalence of HF is higher in individuals with diabetes than in those without. Diabetes accelerates the development of coronary atherosclerosis and is also associated with hypertension. It is, however, unclear whether diabetes also causes a specific cardiomyophathy. Diabetes is thought to promote the development of myocardial fibrosis and diastolic dysfunction, while it is also associated with other adverse CV-related processes.
Although observational studies have reported higher rates of HF than of other CV events in type 2 diabetes patients, many trials do not report HF as an outcome. Recent large-scale trials of glucose-lowering therapies reveal however that HF occurred at a frequency similar to that of stroke and MI, despite the fact that most trials exclude patients with HF at baseline, as well as elderly patients. HF appears to be a common complication of advanced diabetes, especially when nephropathy develops.
Not only the high prevalence of HF in patients with diabetes is important to consider; since its occurrence is associated with a poor subsequent prognosis. Both observational and clinical studies have found higher mortality rates in diabetic patients with HF, as compared to diabetics without HF.
Although regulatory bodies now require proof that glucose-lowering drugs are safe or beneficial with respect to MI and stroke, no evidence exists that these agents may cause these events. Two agents (the thiazolidinediones pioglitazone and rosiglitazone) have however been associated with increased risk of HF. It has been postulated that thiazolidinedione-induced HF was merely a reflection of fluid retention. Data from RECORD now suggest otherwise, as risk of hospitalisation for HF was higher for patients on rosiglitazone. In PROACTIVE, both patients on pioglitazone and placebo had poor prognosis after developing HF, as compared to both treatment groups without HF.
In SAVOR-TIMI 53, patients on DPP4-inhibition had a higher risk of hospital admission for HF as compared with patients on placebo. It remains to be elucidated whether this is a play of chance, a drug-specific or a class effect, or even a broader effect of incretin-based therapies, thereby underscoring the importance of HF as an outcome of diabetes. Contradicting the findings of SAVOR-TIMI 53, HF is characterised by high DPP-4 activity, and in animals models for HF it was beneficial to reduce this activity with DPP-4 inhibitors. Another clinical trial evaluating the DPP-4 inhibitor alogliptin (EXAMINE) reported a non-significantly greater number of hospital admissions for HF with alogliptin as compared with placebo.
Since HF has not been a focus of studies in patients with diabetes, little is known about how these agents may cause HF. Fluid retention cannot fully explain the findings of RECORD and PROACTIVE, although it may unmask HF in susceptible individuals with undiagnosed underlying left ventricular dysfunction. Many aspects of anti-diabetic treatment remain unclear. Agents like DPP-4 inhibitors have many substrates, thus many biological pathways may be affected.
Thus, there is a clear need for improved studies on myocardial function in patients with concomitant diabetes and HF, and a better understanding of the cardiac effects of new pharmacological treatments for diabetes.
The authors of this viewpoint are aware that some may argue that composite outcomes should reflect the same disease process. They argue that composites may also reflect all major and common events that represent the burden of the disease in question to patients and society. Furthermore, they add that for any composite outcome, interpretation is difficult when a treatment has divergent effects on individual components, and these may not reflect the same disease process.
Find this article online