Type 2 diabetes mellitus and heart failure: a position statement from the Heart Failure Association of the European Society of Cardiology

Prevalence of type 2 diabetes mellitus and heart failure in general populations

The prevalence of T2DM, which encompasses 90–95% of diabetic individuals, has globally increased from 4.7% in 1980 to 8.5% in 2014,5 albeit diagnostic criteria have changed over that period.6, 7 Contemporary data suggest a stable overall HF prevalence of 11.8% (range 4.7–13.3%) in the general population.8

Prevalence of heart failure in patients with type 2 diabetes mellitus

In the Reykjavik study in the general population, the prevalence of HF in people with T2DM was 12%.9 In this study, HF was more common in patients with T2DM aged >70 years (i.e. 16% and 22% of men and women, respectively). In the Kaiser Permanente population, patients with T2DM aged <75 years had an approximately three-fold higher prevalence of HF compared to those without T2DM.10 In those aged 75–84 years, T2DM was associated with a doubling of risk for HF. In these relatively old studies, HF phenotype (i.e. HFrEF or HFpEF) or biomarker status was not reported. In clinical trials of T2DM patients, the prevalence of HF at baseline has varied between approximately 10% and 30% (Table 1).11-24

Prevalence of type 2 diabetes mellitus in patients with heart failure

In the general population, HF is associated with a higher prevalence of T2DM compared to patients without HF (Table 2),9, 25-29 but marked regional differences have been observed both in Europe and in rest of the world. In studies conducted in Iceland9 and Italy,27 T2DM prevalence was four and three times higher, respectively, whereas in Italy, T2DM prevalence was almost doubled in HF subjects (Table 2). Approximately 25% of patients with HF in England25 and Denmark28 also had T2DM. Despite younger age and less obesity, a significantly higher prevalence of T2DM (57%) was observed in a population-based cohort of Southeast Asian HF patients compared to Caucasian patients (24%).30 The reasons for the wide regional variation in T2DM prevalence in HF patients warrant further international studies with shared study design and standardized data collection.

In clinical trials of chronic HF patients, the prevalence of T2DM was around 30%, irrespective of HF phenotype (i.e. HFrEF and HFpEF) (Table 3).31-48 The highest prevalence of T2DM was seen in trials of acute HF (around 40%).

In registries of hospitalized HF patients in North America and Europe, the prevalence of T2DM is around 40–45%,49-52 and a slight increase in the prevalence was reported in North America over time.49, 52. In the Swedish HF Registry (68% from hospitals and 32% from primary care), T2DM was more prevalent in HF patients with CAD compared to those without (30% vs. 19%).53

Incidence of new type 2 diabetes mellitus in patients with heart failure

In patients with HF, data from observational and clinical trials demonstrate an increased risk for new-onset T2DM compared to patients without HF. In a Kaiser Permanente study, the incidence of T2DM was significantly higher in patients with than without HF (i.e. 13.6/1000 vs. 9.2/1000) over a 5-year follow-up.10 In a Danish nationwide cohort study, 8% of HF patients developed T2DM over 3 years, and the severity of HF was associated with a stepwise increased risk of developing T2DM.54 Similar incidence of T2DM was reported in clinical trials of HF patients, as demonstrated by the CHARM program, in which 7.8% of patients developed T2DM over 2.8 years.55, 56 In the EMPHASIS-HF trial including HFrEF patients, the incidence of T2DM was 3.7% over a median follow-up of 21 months.57 Notably, HF treatment with angiotensin-converting enzyme (ACE) inhibitors was shown to lower the incidence of T2DM in HFrEF patients; in a substudy of the SOLVD trial, 6% of patients in the enalapril arm developed T2DM over a mean follow-up of 2.9 years as opposed to 22% in the placebo arm.58 Registry data corroborate that the use of renin–angiotensin system inhibitors is associated with attenuated risk for T2DM in HF patients receiving loop diuretics.54 Clinical trials also demonstrated that the severity of HF, as indicated by a higher New York Heart Association (NYHA) class, increases the likelihood of developing T2DM.27, 59

Incidence of heart failure in patients with type 2 diabetes mellitus

Recently, a population-based study of 1.9 million patients with T2DM without overt CV disease, followed for 5.5 years, demonstrated that incident HF was observed more frequently (14.1%) than vascular events, including myocardial infarction (MI) or stroke.60 T2DM is an independent risk factor for the development of HF.10 In a retrospective cohort followed for up to 72 months, patients with T2DM were more likely to develop HF than patients without T2DM (incidence rate 30.9 vs. 12.4/1000 person-years, rate ratio 2.5).61 In elderly patients with T2DM, the incidence of HF was two-fold higher compared to patients without T2DM (121 vs. 62 cases/1000 patient-years).62 In the UKPDS 35 trial including newly diagnosed diabetic patients, HF incidence steeply increased with the severity of dysglycaemia ranging from 2.3 to 11.9/1000 person-years for patients with glycated haemoglobin (HbA_1c) <6% and HbA_1c >10%, respectively.63 Similarly, in observational studies (NHANES64 and ARIC65), the incidence of HF in patients with T2DM was higher than in those without T2DM, with the corresponding hazard ratios (HRs) of 1.85 and 3.54. Indeed, in the ARIC study, higher HbA_1c levels in T2DM patients were associated with significantly more incident HF cases than in patients with T2DM and lower HbA_1c levels.65 The incidence of HF in T2DM patients compared to those without T2DM is even higher in patients with established CAD, in which each 1% increase in HbA_1c level was associated with a 36% increased risk for HF hospitalization.66, 67 Patients with pre-diabetes in the ARIC study also had more HF than those without pre-diabetes.68

Clinical presentation, quality of life and functional status of patients with type 2 diabetes mellitus and heart failure

Patients with T2DM and both HFrEF1, 34, 69, 70 and HFpEF1 have worse NYHA functional class and more HF-related symptoms and signs than patients without T2DM, despite having similar ejection fraction.69, 70 In the SOLVD-Prevention trial of patients with asymptomatic left ventricular systolic dysfunction, patients with T2DM were more likely to progress to symptomatic HF than those without T2DM, although the increased risk appeared to be confined to patients with HF secondary to CAD.71

Most trials also demonstrated worse quality of life in patients with T2DM and concurrent HF (both HFrEF and HFpEF), as compared to patients without T2DM.40, 69 Patients with T2DM and HFrEF also have shorter 6-minute walk distances and decreased peak oxygen uptake in comparison to non-diabetics.55, 69, 72

Type 2 diabetes mellitus and mortality in patients with heart failure

In all population-based studies, T2DM was associated with increased all-cause mortality in HF patients, albeit substantial regional differences were reported across Europe, and no differentiation between HFrEF and HFpEF was performed (Table 4).26, 29, 51, 73-81 In Sweden, there was a moderately higher risk (HR 1.60)53 and in the Netherlands a significantly higher risk of death (HR 3.19)26 attributed to T2DM. Additionally, in the Rotterdam study, T2DM was associated with an excess risk for CV death (HR 3.25) that was similar to the risk of all-cause mortality.26 Likewise, all studies of the effect of T2DM on mortality in HF outpatients have found a higher mortality risk attributable to T2DM (Table 4).

Concerning patients hospitalized for HF, data on the association between T2DM and in-hospital mortality are divergent. In the OPTIMIZE-HF, ADHERE and Get With the Guidelines-HF registries in the United States, T2DM was not associated with higher in-hospital mortality.82-85 Conversely, in the ALARM registry (six European countries, Mexico and Australia), and in the European Society of Cardiology (ESC) HF Long-Term Registry, T2DM was independently associated with a higher risk of in-hospital mortality.51, 86 There is a suggestion from some cohorts82, 87 that short-term mortality in HF patients post-discharge may be similar or slightly lower in those with T2DM. However, with longer-term follow-up, an association between T2DM and worse outcomes in HF patients becomes evident. For example, in the EVEREST trial in which patients were followed for 9.9 months after a HF hospitalization, T2DM conferred a slightly higher mortality.45 Also, in patients from Scotland, T2DM increased mid-to-long-term mortality following hospitalization for HF.87 Likewise, in the ESC HF Long-Term Registry, the presence of T2DM was independently associated with increased 1-year all-cause mortality.51, 73

Clinical trial results are somewhat conflicting regarding the risk of all-cause and CV mortality attributed to T2DM in HF patients, but most clinical trials reported an increased risk of death in patients with concurrent T2DM and HF (Table 5).1, 31-35, 37, 40, 42, 44, 45, 69, 88-92 In HFrEF, five out of eight trials demonstrated an association between T2DM and increased all-cause mortality, with the reported HRs between 1.3 and 2.0 (mostly around 1.5) (Table 5). Also three HFrEF trials reported increased CV death, with HRs between 1.5 and 1.8.1, 31, 33 Concerning HFpEF, all trials reported increased all-cause mortality (HRs 1.5 to 1.8) and two out of four trials also reported an increased risk of CV mortality in patients with T2DM compared to patients without T2DM, with HRs 1.6 to 1.9 (Table 5). In the CHARM trial, T2DM was an independent risk factor for both all-cause mortality and CV mortality even after adjustment for 32 covariates.1 Additionally, in the same study, T2DM had a greater association with higher all-cause and CV mortality in patients with HFpEF than HFrEF.1

A recent meta-analysis of 31 registries and 12 clinical trials with 381 725 patients with acute and chronic HF, with a median follow-up of 3 years confirms that T2DM is independently associated with a higher risk of all-cause death (random-effects HR 1.28), CV death (HR 1.34), hospitalization (HR 1.35), and the combined endpoint of all-cause death or hospitalization (HR 1.41), and the observed long-term risk appears greater in patients with chronic than in those with acute HF.93

Type 2 diabetes mellitus and causes of death in patients with heart failure

In the CHARM trial, patients with T2DM and both HFrEF and HFpEF were more likely to die of all subtypes of CV death [i.e. death due to HF, sudden cardiac death (SCD), death due to MI and death due to stroke].1 The PARADIGM-HF study also reported that patients with T2DM and HFrEF were more likely to die of CV as well as all-cause mortality compared with patients without T2DM.69 In the BEST trial, T2DM was an independent risk factor for death from pump failure.94

Aside from CV death, results from the Emerging Risk Factors Collaboration, including 820 900 people, demonstrate that T2DM is independently associated with increased risk of death from several cancers (i.e. liver, pancreas, ovary, colorectum, lung, bladder, and breast), renal and liver disease, pneumonia and other infectious diseases, mental and nervous system disorders, non-hepatic digestive diseases, external causes, and chronic obstructive pulmonary disease.95 The study found that a 50-year-old with T2DM died, on average, 6 years earlier than an individual without T2DM, with about 40% of the difference in survival attributable to excess non-vascular deaths.95

Is the higher risk of type 2 diabetes mellitus only seen in heart failure secondary to coronary artery disease?

Whether or not the increased risk of mortality with T2DM in HF patients is seen in both those of ischaemic and non-ischaemic aetiology is uncertain. The majority of the available data suggests that T2DM is associated with higher risk of mortality in both patients of ischaemic and non-ischaemic aetiology (Table 6).29, 88, 94, 96-99 In a population-based Danish study, which followed patients for 6.8 years, patients with T2DM and HF had higher mortality whether or not they had CAD.96 The higher risk appeared early and persisted throughout follow-up. In the CHARM trial, patients with both HFrEF and HFpEF had higher mortality attributed to T2DM whether or not they had CAD.1 In the DIAMOND trial, T2DM was associated with a higher risk of mortality in both ischaemic and non-ischaemic HF.100 These consistent findings conflict with two smaller population-based studies in the United States29 and France97 and one Spanish single-centre study101 of patients hospitalized with HF, which suggested that diabetes was only associated with higher mortality in those with non-ischaemic aetiology. In three early clinical trials (SOLVD,98 BEST,94 and DIG102) the risk appeared to be confined to those with an ischaemic aetiology.

Is the higher risk of mortality with type 2 diabetes mellitus and heart failure seen in both women and men?

An early report from the Framingham study reported that the mortality risk related to T2DM was confined to women and not to men.103 In two population-based studies from Scotland and Sweden, the increased mortality risk of T2DM was seen in both women and men, but the effect was slightly greater in women.87, 96 Likewise, in the recent ESC HF Long-Term Registry and in the CHARM trial, T2DM was a risk factor for mortality in both men and women.1, 73

Does glycated haemoglobin predict mortality in patients with heart failure and type 2 diabetes mellitus?

In the CHARM trial, high HbA_1c was associated with increased all-cause and CV mortality in patients with T2DM and both HFrEF and HFpEF.104 A 1% increase in HbA_1c was associated with an increased HR of 1.1 for CV mortality.104 In patients from a US study of HF clinics, a U-shaped relationship with regard to increased all-cause mortality was found.71 Patients with either very low or very high HbA_1c were at greatest risk. A similar U-shaped curve was found in a single-centre study from Scotland.105 In one single-centre observational study of 123 young patients with advanced HF and T2DM, patients with a HbA_1c of <7% had higher rates of all-cause mortality.106 In the GISSI-HF study, including 6935 chronic HF patients, the presence of T2DM and higher HbA_1c levels was an independent predictor of all-cause mortality (HRs 1.43 and 1.21, respectively) and the composite outcome of mortality and CV hospitalization (HRs 1.21 and 1.14, respectively).107

In summary, high HbA_1c levels in T2DM and HF are consistently associated with higher mortality. Conversely, low HbA_1c levels can be associated with good outcomes (at least in a clinical trial cohort), but can be associated with worse outcomes (in population-based studies and those with very advanced HF).

Pre-diabetes and undiagnosed type 2 diabetes mellitus and risk of mortality in heart failure

In the PARADIGM-HF trial, patients with pre-diabetes were at increased risk of mortality.69 Patients with undiagnosed T2DM were also at higher risk of mortality than subjects without T2DM, but the risk was not as high as in patients with previously known T2DM. In CHARM, pre-diabetes and undiagnosed T2DM were both associated with greater rates of HF hospitalization, CV and all-cause mortality than those without T2DM.108 However, not all studies have reported an increased mortality risk with pre-diabetes. In a study of 970 non-diabetic patients with HF, an increased 1-year mortality risk was found only in patients with HbA_1c >6.7% and reduced left ventricular ejection fraction (≤45%), but not in those with HFpEF.109 Also, in the GISSI-HF study of unselected HF patients, pre-diabetes was not an independent predictor of increased mortality.107 The reasons behind these discrepancies might be attributed to differences in patient characteristics and warrant further assessment.

Type 2 diabetes mellitus and risk for heart failure hospitalization

Several clinical trials documented that patients with T2DM and HFrEF were more likely than patients without T2DM to be hospitalized for HF.1, 37, 69, 70, 94 In the CHARM trial, rates of hospitalization for HF in patients with T2DM were greater for those with HFpEF than HFrEF and patients with HFpEF and T2DM were almost 2.5 times more likely to be hospitalized for HF than those without T2DM.1 In I-Preserve, patients with T2DM and HFpEF were also more likely to be hospitalized with HF.40

Readmission after a hospitalization for heart failure

Registry data indicate that patients with T2DM had more all-cause rehospitalizations than those without T2DM.79, 82, 110 In a population-based study in Scotland, T2DM was a predictor of readmission for HF (with the increased risk greatest in younger women).87 In the ESC HF Long-Term Registry, T2DM was independently associated with rehospitalization for HF.51 Likewise, in the EVEREST trial, T2DM was associated with greater rates of HF rehospitalization (HR 1.19).45

In addition, as demonstrate by the OPTIMIZE-HF and Get With The Guidelines-HF registries in the United States, patients with HF and T2DM experience slightly longer hospitalizations than patients without T2DM.82-84

Type 2 diabetes mellitus, myocardial infarction and stroke in patients with heart failure

The only trial to investigate the association between T2DM and risk of MI and stroke in HF patients was the CHARM trial demonstrating that the presence of T2DM increased the risk for MI and stroke irrespective of HF phenotype (i.e. HFrEF or HFpEF).1

Risk for heart failure hospitalization in patients with type 2 diabetes mellitus without a previous history of heart failure

In the ARIC registry, representing a cohort of 14 079 people in the community without known HF, T2DM was the most powerful risk factor for incident HF hospitalization.110 In a large meta-analysis of patients with T2DM but without HF, predictors of incident HF included insulin use, HbA_1c and fasting glucose.111

Mortality in type 2 diabetes mellitus patients with heart failure

In the CV outcomes trials of new therapies for T2DM, the development of HF is associated with markedly higher mortality (especially in RECORD112 and SAVOR-TIMI 5316). Patients with T2DM who developed HF had a 10 to 12 times greater mortality than those who did not develop HF.3, 113 In addition, they are also at a 2.45-fold greater risk of CV death compared with patients with T2DM but without HF.114

Unrecognized heart failure in patients with type 2 diabetes mellitus and unrecognized type 2 diabetes mellitus in patients with heart failure

Observational evidence indicates that a significant proportion of patients aged ≥60 years (27.7%) may have unrecognized HF (22.9% and 4.8%, HFpEF and HFrEF, respectively) based on the ESC diagnostic criteria.115, 116 On the other hand, pre-diabetes and undiagnosed T2DM are common in patients with HF. In the PARADIGM-HF trial, 13% of patients with HFrEF had undiagnosed T2DM and 25% had pre-diabetes.69 Likewise, 11% of ‘non-T2DM’ patients with HFrEF in the RESOLVD trial had undiagnosed T2DM.117 In the CHARM study, undiagnosed T2DM was common in both HFrEF and HFpEF.108 In the ESC HF Long-Term Registry, even higher proportion of HF patients (19.1%) had undiagnosed T2DM.73

Considering prognostic implications of concurrent T2DM and HF, these findings stress the importance of developing screening strategies for unrecognized HF among T2DM patients and vice versa. Since evidence of strategies for HF screening is sparse,116 in T2DM patients, screening for HF might be currently based on clinical characteristics (i.e. age, history of CAD, exercise-related shortness of breath, body mass index, laterally displaced apex beat) that have been shown to reliably identify elderly subjects at risk of HF that may require further assessment (e.g. echocardiography).118 Such a strategy may be used to prevent complications and possibly improve outcomes, particularly in subjects with HFrEF.119 Conversely, since undiagnosed T2DM is common among patients with HF, it is prudent to screen patients without known T2DM in accordance with current recommendations using the 8 h fasting plasma glucose, 2 h glucose tolerance test or HbA_1c levels (equally appropriate).120

Diabetic cardiomyopathy

In 1954, Lundbaek was the first to propose the existence of a specific diabetic heart muscle disease without involvement of CAD or hypertension.129 Two decades later, Rubler et al.130 described diabetic-related post-mortem findings in four patients with T2DM, glomerulosclerosis and HFrEF with normal epicardial coronary arteries. There is no definition of diabetic cardiomyopathy, which makes studies of epidemiology, pathophysiology, natural history and associated clinical outcomes challenging. The most commonly accepted definition refers to a myocardial dysfunction which occurs in the absence of all other CV disease.120, 131

Phenotypes of type 2 diabetes mellitus-related cardiomyopathy

Pharmacological therapy

Devices and surgery

Exercise prescription

Recently, a single large trial (HF-ACTION34) investigated the effects of exercise training in patients with mild to moderately severe HF symptoms. In an adjusted analysis, exercise training led to an 11% (P=0.03) reduction in the primary composite outcome of all-cause mortality or all-cause hospitalization. The trial enrolled 32% of patients with T2DM and there was no interaction between T2DM status and the effect of exercise on clinical outcomes.

Drugs that increase heart failure hospitalizations

Over the last 15 years there has been concern that some of T2DM drugs might increase the risk for HF (Table 7).16, 17, 179-182 Drugs that are now known to increase the risk for HF are thiazolidinediones (TZDs) and a dipeptidyl peptidase-4 (DPP4) inhibitor, saxagliptin.16, 17 In the RECORD112 and the PROactive trials,183 patients randomized to TZDs, rosiglitazone and pioglitazone, respectively, had more HF events than those on placebo. In the SAVOR-TIMI 53 trial (saxagliptin vs. placebo), saxagliptin significantly increased the risk for HF hospitalizations (HR 1.27, P=0.007).16 Patients at greatest risk were those with a history of HF, an estimated glomerular filtration rate (eGFR) ≤60 mL/min, or elevated baseline levels of N-terminal pro B-type natriuretic peptide (NT-proBNP).16 In both RECORD and SAVOR-TIMI trials, patients who developed HF had a high rate of subsequent death. On that basis, pioglitazone, rosiglitazone and saxagliptin are contraindicated in patients with HF or at risk of HF.

Not all DPP4 inhibitors are associated with higher rates of HF (Table 8).16-24, 184-186 In the EXAMINE trial of alogliptin vs. placebo in patients who had had an acute coronary syndrome, there was not a statistically significant increase in the risk of HF hospitalizations in patients randomized to alogliptin.19, 184 Likewise, sitagliptin in the TECOS trial had no signal of excess rates of HF.18, 185 Two ongoing trials, CAROLINA (Cardiovascular Outcome Study of Linagliptin Versus Glimepiride in Patients With Type 2 Diabetes; NCT01243424), and CARMELINA (Cardiovascular and Renal Microvascular Outcome Study With Linagliptin in Patients with Type 2 Diabetes Mellitus; NCT01897532), will allow further clarification on the role DPP4 inhibitors in patients with T2DM and HF.

Drugs that might increase the risk for heart failure

Over many years there has been suspicion that insulin, which causes sodium and water retention, may increase the risk for the development of HF. In large observational studies, insulin is associated with higher mortality rates than metformin.2 There have been similar concerns with sulphonylureas which, as insulin secretagogues, have also been associated with higher death rates than metformin.2 These studies, although large, are non-randomized and therefore inconclusive. In the only RCT of insulin vs. placebo [ORIGIN, 12 537 people with CV risk factors plus impaired fasting glucose, impaired glucose tolerance, or T2DM (i.e. not in patients with HF)], insulin was not associated with higher rates of HF hospitalization than placebo.187 Remarkably, despite the use of insulin and sulphonylureas for decades, there are no other placebo-controlled randomized trials.

Currently, sulphonylureas and insulin could be used in T2DM patients with HF (usually as a second- or third-line treatment), although their safety in HF is still inconclusive.

Antidiabetic drugs that might be safe in heart failure

It has been proposed that metformin might be safe and efficacious in patients with T2DM and HF. This was based on large observational studies where metformin was associated with lower mortality and HF hospitalization rates than other T2DM drugs (primarily insulin and sulphonylureas).2 There are no RCTs of metformin in patients with T2DM and HF. Whether or not metformin is efficacious or safe is inconclusive. Previous concerns that metformin may cause metabolic acidosis are no longer justified.2 Accordingly, metformin could be recommended as first-line treatment for patients with T2DM and HF who have preserved or moderately reduced renal function (i.e. eGFR >30 mL/min).

Glucagon-like peptide 1 (GLP-1) receptor agonists have been the subject of many large placebo-controlled trials in patients with T2DM and CV disease or at high risk of CV disease (Table 8).22-24, 186 In these trials, GLP-1 receptor agonists had a neutral effect on the risk for HF hospitalization. Similarly, no signal for a higher risk for HF hospitalization was seen with acarbose (vs. placebo) in patients with insulin resistance and CAD.188 Bromocriptine has not been studied with respect to its effect on HF outcomes.

Prevention of heart failure by type 2 antidiabetic drugs

A significant breakthrough in contemporary cardiology was the finding that some T2DM drugs are associated with a lower risk of HF hospitalization in patients with CV disease or at high risk of CV disease (Table 8). Two large RCTs that assessed CV safety of the sodium–glucose co-transporter type 2 (SGLT2) inhibitors, empagliflozin and canagliflozin, have shown a significant reduction in HF hospitalization with both drugs.20, 189 The primary outcome in both trials was the three-point major adverse CV event (i.e. CV death, non-fatal MI or non-fatal stroke) and HF hospitalization was a secondary outcome. In the EMPA-REG OUTCOME trial (n=7020), including patients with T2DM, established CV disease and eGFR >30 mL/min/1.73 m2, there was a major reduction in HF hospitalization (HR 0.65) with empagliflozin compared with placebo.20 The observed beneficial effect of empagliflozin became evident early (i.e. 2–3 months of treatment) and was observed across a range of pre-specified subgroups, including patients with (10%) and without investigator-reported HF at baseline, that had a similar reduction in HF hospitalizations with empagliflozin compared with placebo. No echocardiograms or natriuretic peptide measurements are available from this trial, so the detail of the beneficial effect on HF hospitalization is not available. Patients hospitalized for HF during the study had a high mortality, which was lower in patients receiving empagliflozin than placebo (13.5% vs. 24.2%).20 In the CANVAS trial, patients with T2DM (n=10 143) either with established CV disease or at high risk of CV disease, randomized to canagliflozin or placebo had a significantly lower risk of HF hospitalization (HR 0.67).21, 189 Empagliflozin in EMPA-REG OUTCOME, but not canagliflozin in CANVAS, reduced all-cause and CV mortality as well as HF hospitalization. In the EMPA-REG OUTCOME trial, the only major adverse event was an increased risk of genital tract infections, which were treatable, and infrequently recurred.20 In the CANVAS trial, treatment with canagliflozin was associated with a significantly higher risk of lower-limb amputations (6.3 vs. 3.4 per 1000 patient-years; HR 1.97) and possibly a higher risk of fractures compared with placebo.21 Large RCTs of other new T2DM drugs have not shown a reduction in incident HF (Table 8).

Randomized clinical trials with SGLT2 inhibitors

While two drugs (i.e. empagliflozin and canagliflozin) have a favourable effect on HF hospitalization, no T2DM drug has yet been investigated as a treatment for HF. In 2017, three large RCTs with SGLT2 inhibitors (i.e. empagliflozin and dapagliflozin) have started, which will enrol HF patients either with or without T2DM (i.e. T2DM is not a mandatory inclusion criteria). Two trials will assess safety and efficacy of empagliflozin vs. placebo on top of guideline-based medical therapy for the reduction in primary outcome (CV death or HF hospitalization) both in patients with HFrEF (EMPEROR-Reduced, NCT03057977) and HFpEF (EMPEROR-Preserved, NCT03057951) (Table 9). Among secondary outcomes, the two trials will assess all-cause mortality, and renal effects of empagliflozin vs. placebo in patients with HF. The third trial (Dapa-HF, NCT03036124) will assess safety and efficacy of dapagliflozin vs. placebo for the reduction in CV death or HF hospitalization (or urgent HF visit) in patients with HFrEF. Secondary outcomes will include all-cause mortality and effects on renal function. The results of these trials will shed more light on potential beneficial CV and renal effects of SGLT2 inhibitors in HF patients, including those without T2DM.

In addition, a number of ongoing smaller RCTs are assessing the effect of SGLT2 inhibitors on CV outcomes, including various aspects of HF in patients with and without T2DM, as summarized in Table 9.

Randomized clinical trials with GLP-1 receptor agonists

In the LIVE trial, in patients with stable HFrEF, with and without T2DM, there were no significant changes in left ventricular ejection fraction between patients randomized on liraglutide or placebo.180 However, there was a significant increase in heart rate (P<0.0001) and more serious cardiac adverse events with liraglutide (P=0.04). In a placebo-controlled FIGHT trial, of patients with HFrEF, with and without T2DM (41%), liraglutide was not associated with an improvement in the composite primary endpoint of death, rehospitalization and NT-proBNP change.181 Pre-specified subgroup analyses in patients with T2DM did not reveal any significant between-group differences. A small randomized, placebo-controlled trial of albiglutide in HFrEF showed no effect on left ventricular function and 6-minute walk distance.190 These observations have raised some concern regarding the safety of liraglutide in HFrEF patients that warrants further research.