Clinical Implications of Cardiovascular Outcome Trials in Type 2 Diabetes: From DCCT to EMPA-REG

Rydén L., Shahim B., Mellbin L
Clin Ther. 2016 Apr 20. doi: 10.1016/j.clinthera.2016.03.035. [Epub ahead of print]

Glucose lowering and vascular benefits
The idea that hyperglycaemia links diabetes and coronary artery disease was already hypothesised in 1929. Since then, this association has been confirmed in several populations. The decrease in microvascular complications (37%), myocardial infarction (MI, 14%), stroke (12%) and heart failure (16%) over 10 years of follow-up, observed with each 1% reduction of mean HbA1c in the UK Prospective Diabetes Study (UKPDS) is a frequently cited example.
Many trials have tested the hypothesis that glucose normalisation should prevent vascular injury, often focussing on the glucose-lowering capacity of novel compounds, rather than on CV outcomes as well. The Diabetes Control and Complications (DCCT) and its prolonged follow-up Epidemiology of Diabetes Interventions and Complications (EDIC) was the first trial to demonstrate vascular benefits of glucose lowering with insulin in type 1 diabetes (1993).
UKPDS was the first study to show a reduction in microvascular complications and MI after HbA1c reduction on insulin-providing therapy in patients with type 2 diabetes (T2DM). But not all subsequent studies could confirm the hypothesis that tight glycaemic control positively affects macrovascular disease. For instance, ACCORD showed a somewhat higher CV mortality in the intensively managed arm, and was prematurely stopped. A meta-analysis of five major glucose-lowering trials (UKPDS, PROACTIVE, ADVANCE, ACCORD and VADT) did not find an effect on all-cause mortality of a mean reduction of Hb1Ac by 0.9% by intensive vs. conventional glucose-lowering treatment. Non-fatal MI was reduced by 17%.
A possible explanation for the initially beneficial and later neutral or partly negative outcome may be that at the time of UKPDS, no effective lipid-lowering and blood pressure lowering were available, as they were at the time of subsequent glucose-lowering trials. Alternatively, side-effects of glucose-lowering compounds used in high dosages may have played a part in the disappointing results.

Safety of glucose-lowering drugs
A meta-analysis on CV events on rosiglitazone treatment showed an increased risk of MI (43%) and CV death (64%), in comparison with several other glucose-lowering drugs. This meta-analysis has been criticised, for instance due to inclusion of several small studies that were not considered representative for studying CV outcomes. It can be seen as a reason why the US FDA, followed by EMA, requested developers of new drugs for T2DM to provide evidence that CV event risk is not increased.

Contemporary glucose-lowering trials
(important characteristics of mentioned trials are summarised in a table in the article)
Four studies of glucose-lowering agents in patients with T2DM and CV disease or at high CV risk did not find an increased risk of CV endpoints, as compared with control treatment. (ORIGIN: insulin glargine vs. conventional glucose-lowering treatment; SAVOR: saxagliptin vs. placebo in addition to ongoing therapy; EXAMINE: alogliptin vs. placebo added to ongoing therapy; TECOS: sitagliptin vs. placebo added to ongoing therapy).
SAVOR found an increase in hospitalisation for HF. The interpretation of this observation has been debated, since two other studies evaluating this drug class (EXAMINE and TECOS) did not reveal such risk. It should be noted that the molecules of the gliptins differ.
An FDA safety review concluded that T2DM medications saxagliptin and alogliptin may increase risk of HF, especially in those with existing heart or kidney disease.
The GLP-1 agonist lixisenatide was not associated with increased CV outcomes or HF hospitalisation in T2DM patients in the ELIXA trial.

Reason for the lack of impact of glucose lowering
Obviously, observations in epidemiological studies do not necessarily translate a benefit on outcomes when lowering glucose. And the prospective, randomised trials that should have confirmed a proposed association, had some short-comings.
For instance, T2DM is a much more complex disease than hyperglycaemia; and the metabolic disease also involves insulin resistance, dyslipidaemia, oxidative stress, inflammatory activation, endothelial dysfunction and hypercoagulability.
Also, it has been proposed that CV disease starts years before a T2DM diagnosis is made based on a blood glucose threshold. Trials that are initiated once a diagnosis is reached, may be too late to impact macrovascular injury.
Furthermore, it can be argued that commonly used endpoints such as MI and stroke are less relevant these days, due to their improved prognosis. Also, the first manifestation of CV disease in diabetics was shown to be more often peripheral artery disease, HF, unstable angina pectoris, before MI and stroke.
All five mentioned studies added the tested drug to existing treatment and permitted adjustment of glucose-lowering therapy in the comparator arm. HbA1c as a measure of glycaemic control was fairly similar between treatment arms. Thus, these studies should be considered CV safety studies, and showing CV superiority would have been too challenging given the short follow-up and lack of head-to-head comparisons between agents.

New ideas and hopes: the EMPA-REG OUTCOME trial
It is clear that patients with diabetes have a worse prognosis than those without, especially if diabetes is complicated by CV disease. The first new hope to impact CV outcomes in patients with T2DM came from the randomised, prospective EMPA-REG OUTCOME trial that evaluated the SGLT2-inhibitor empagliflozin. In patients with T2DM and established CV disease, treatment with empagliflozin reduced the composite outcome of CV death or nonfatal MI or stroke by 24%.
CV death appeared mostly reduced by a substantial decrease of HF hospitalisation (by 35%). A modest lowering of HbA1c (-0.24%) was obtained with empagliflozin at the end of the study.
When specifically focussing on HF, consistent effects of empagliflozin were seen across various subgroups, including those with or without pre-existing HF and those on different diabetes or HF treatments. Number needed to treat was 35 over 3 years to prevent 1 HF hospitalisation.
It was noted that the benefit of empagliflozin on HF hospitalisation was seen already within 15 weeks. The mechanisms underlying these favourable effects are unknown, but unlikely to result from the modest glucose-, blood pressure and weight reductions seen in empagliflozin-treated patients. It is speculated that a decrease in volume load related to osmotic diuresis and increased sodium excretion may play a role, possibly in combination with reduced arterial stiffness.

Clinical implications
Ongoing trials will provide further insight on whether SGLT2-inhibitors can be adopted as an established class of glucose-lowering drugs. Currently SGLT2-inhibitors are already available for prescription. After consideration of contra-indications, empagliflozin may be prescribed, while keeping in mind that those on diuretics might be sensitive to volume depletion. And patients need to be warned about signs of ketoacidosis as a rare complication.
No CV benefits have as yet been reported in diabetic patients at high risk of, but without established CV disease. Some of the ongoing trials address this question.
The most interesting effect of empagliflozin appears to be mediated via mechanisms other than its glucose-lowering effect. Therefore, it is reasonable to consider price differences when glycaemic control is the main reason for a prescription.

Future research
Mechanistic trials are warranted to better understand the benefits.
The EMPA-REG results also provide the rationale for outcome trials with empagliflozin in HF patients with know and new glucose perturbations.

Find this article online at Clin Ther