Treatment of Hyperglycemia in Patients with Diabetes and Heart Failure
Patients with T2DM have a marked increase in the prevalence of HF and have a poorer prognosis when they develop HF than non-diabetic individuals. A major consideration is the extent to which hyperglycemia itself contributes to these disparities. Hyperglycemia may impact HF in several ways: (1) chronic hyperglycemia may contribute to the development of HF, (2) hyperglycemia may worsen the outcome of acute HF, (3) chronic hyperglycemic management in the patient with chronic HF may influence clinical outcomes, (4) specific anti-hyperglycemic agents may have direct effects on the development and/or clinical outcomes of HF.
As noted in the section on epidemiology, observational studies have shown a striking relationship between HbA1c and the development of HF in patients with T2DM and no HF at baseline. During a mean follow-up of 2.2 years, each 1 % increase in HbA1c was associated with an 8 % increase in HF and for a mean follow-up of 7 years, HF development increased progressively from <5 % with an HbA1c of 7.0 to 72 % with an HbA1c greater than 10 % [
18].
The level of chronic glycemic control and the stage of HF determine the clinical outcomes of CV death and hospitalization for HF in patients with DM. The large clinical intervention trials in participants with T2DM to determine the effect of intensive (HbA1c 6.5 to 6.8 %) versus ordinary glycemic control (HbA1c 7.5 to 8.0 %) on CV events failed to show any effect on HF outcomes (HR = 1.00, 95 % CI = 0.86–1.16) [
79]. While these trials included 27,049 participants, the baseline incidence of HF was very low (approximately 5 %), and only 905 participants were hospitalized for HF annually. The overall difference in mean HbA1c between the groups in all of those intervention trials was 0.9 %.
In contrast, observational studies in patients with HF and DM show a strong relationship between glycemic control and clinical outcomes. In a population-based study of 16,524 patients seen in emergency departments in Ontario, Canada with acute HF syndromes, the presenting blood glucose level predicted clinical outcomes [
80]. A presenting blood glucose >11.1 mmol/l in patients with DM (
n = 7249) was associated with an increased risk of all-cause death (HR = 1.48,
P = 0.01) or DM-related hospitalizations (HR = 1.39,
P < 0.001). A presenting blood glucose >9.4 mmol/L was associated with an increase in risk of hospitalization for HF (HR = 1.15,
P = 0.002) or any CV cause (HR = 1.09,
P = 0.009) in the entire patient population [
80]. The Worcester WHES study observed the effects of elevated serum glucose levels on survival after acute HF in 5428 non-diabetic patients, 3807 patients with diagnosed DM, and 513 patients with admission hyperglycemia. In-hospital death rate was highest in those with admission hyperglycemia (9.9 %) as compared to individuals without DM (7.5 %) and those with known DM (6.5 %). In contrast, the patients with known DM had the highest death rates 3 months (HR = 1.14 compared to those without DM, 95 % CI = 1.09–1.20) and 1 and 2 years (HR = 1.18, 95 % CI = 1.11–1.24) after discharge [
51•].
The ASTRONAUT trial (aliskren) [
81] and the EVEREST trial (tolvaptan) [
82] were clinical trials in participants hospitalized for symptomatic HF. In both studies, the participants were assessed for post-discharge outcomes, and differences were determined in participants with known DM as compared to participants without DM. The extent of glycemic control was not monitored by glucose levels or HbA1c in either study. ASTRONAUT trial showed no difference in inpatient outcomes between participants with and without DM. However, 12 months after discharge, participants with DM had higher CV mortality or hospitalization for HF (non-diabetic HR = 0.80; diabetic HR = 1.15,
P < 0.03). A mean of 9.9 months after discharge, participants with DM in the EVEREST trial had a 17 % increase in CV mortality or hospital admission for HF. Patients with DM in the EVEREST trial were treated by diet (20 %), oral agents (36 %), or insulin (48 %). The insulin-treated group had an increase in the outcome of CV mortality or HF hospitalization (HR = 1.25, 95 % CI = 1.00–1.57) compared to the other treatments. The increase in outcome events in the short-term post-discharge period of these two studies suggests an influence of glycemic control on outcomes in participants with DM and HFrEF.
Several retrospective studies have examined the effect of the intensity of glycemic control on mortality in patients with DM and chronic HF. Table
2 summarizes the results from three such studies, and all indicate that there is a U-shaped relationship between mortality and HbA1c levels [
83‐
85]. The most comprehensive study examined the mortality during a 2-year follow-up period from the database records of 5815 patients with DM and HF managed in ambulatory clinics in VA Medical Centers [
83]. Ninety-four percent of patients were male with a mean age of 69.2 years; 45 % had reduced LVEF and 55 % had preserved LVEF. The other two studies were each from a single center, included only patients with reduced LVEF, and reported data from a 2-year follow-up [
84,
85]. The patients in the Tomova et al. study had more severe HF with 75 % NYHA class III and IV [
84]. The data from the three studies suggest that the lowest mortality occurs when glycemic control maintains the HbA1c between 7.5 and 8.0 % (Table
2).
Table 2
Glycemic control and mortality in patients with heart failure and diabetes
N = 5815 | Mortality at 2 years |
N = 358 | Mortality or urgent heart transplantation at 2 years |
N = 123 | Mortality at 2 years |
Quintile 1 HbA1c ≤ 6.4 % | 25 % | Quartile 1 HbA1c ≤6.4 % | 52 % | Quartile 1 HbA1c <6.6 % | 9/31 29 % |
Quintile 2 HbA1c 6.4 to ≤7.1 % | 23 % | Quartile 2 HbA1c 6.5 to 7.2 % | 58 % | Quartile 2 HbA1c 6.6 to 7.7 % | 13/33 39 % |
Quintile 3 HbA1c 7.1 to ≤7.8 % | 17.7 % | Quartile 3 HbA1c 7.3 to 8.5 % | 39 % | Quartile 3 HbA1c 7.8 to 8.9 % | 3/29 10 % |
Quintile 4 HbA1c 7.8 to ≤9.0 % | 22.5 % | Quartile 4 HbA1c ≥8.6 % | 35 % | Quartile 4 HbA1c >8.9 % | 7/30 23 % |
Quintile 5 HbA1c > 9.0 % | 23.2 % | | | | |
An additional issue in treating the patient with DM and HF is the choice of anti-hyperglycemic agents used. There are no adequate clinical trials of metformin, sulfonylurea, or ordinary insulin dose on HF pathogenesis or treatment. The only data available are from observational studies. A retrospective analysis of 6185 patients with HF and DM treated in ambulatory clinics in Veterans Affairs medical centers and followed for 2 years showed a propensity score adjusted mortality in metformin-treated patients of 16.1 versus 19.8 % in patients not treated with metformin (HR = 0.76,
P < 0.01). HF hospitalization was no different between metformin treatment and no metformin treatment [
86]. Based on the US and Canadian cardiac failure data, treatment with metformin is not absolutely contraindicated in patients who have isolated HF. The risk of lactic acidosis due to metformin is negligible in these patients and is unrelated to the plasma concentration of metformin. Metformin-associated lactic acidosis may occur when kidney function is decreased in patients with decompensated HF. Metformin provides a greater degree of CV protection and should not be withheld in patients with DM with stable HF who do not have other risk factors for acute decompensated HF or lactic acidosis [
87].
In a single center observational study of 554 consecutive patients with advanced systolic HF, patients were stratified into 3 groups: 43 patients with DM were treated with insulin and 89 were non-insulin-treated. The 1-year survival was 89.7 % in the 422 patients without DM, 85.8 % in the non-insulin-treated patients with DM, and 62.1 % for the insulin-treated patients with DM,
P < 0.00001. After Cox multivariate analysis, insulin-treated DM was found to be an independent predictor of mortality (HR = 4.30, 95 % CI = 1.69–10.94) whereas non-insulin-treated DM was not (HR = 0.95, 95 % CI = 0.31–2.93) [
88]. A large CV outcome trial (ORIGIN) comparing low-dose basal insulin glargine to standard DM care which was carried out in 12,537 participants with pre-diabetes and newly diagnosed T2DM over a median duration of 6.2 years found neither a beneficial nor detrimental effect of insulin glargine on the development of HF [
89]. The study was not designed to specifically study HF, and only 310 of 6264 participants on insulin glargine and 343 of 6273 participants on standard care were hospitalized for HF. The mean HbA1c for the insulin glargine and standard treatment groups for the study were 6.2 and 6.5 %, respectively. Appropriate interventional trials will be necessary to assess the effects of insulin treatment on development and outcomes of HF.
The first clear evidence that anti-hyperglycemic drugs could cause and worsen HF in DM patients was shown with the thiazolidinediones (PPARγ agonists). Pioglitazone and rosiglitazone increase sodium retention by the kidney and cause increased fluid retention which leads to edema and an increase in both the incidence of HF and an increased rate of hospital admissions for HF [
90,
91]. These drugs should not be used in patients with T2DM and HF. As noted previously, despite insufficient data about the effect of insulin treatment on outcomes in patients with T2DM and HF, the potential dangers of hypoglycemia are sufficiently serious in this population to limit insulin use.
The effect of DPP-4 inhibitors in patients with T2DM and HF is complicated. The CV outcome trials show that saxagliptin increases hospitalizations for HF in patients with T2DM [
92]. The alogliptin trial suggests, but is not definitive, that it too may increase HF hospitalizations [
93], while the sitagliptin trial shows no effect on HF morbidity in patients with T2DM [
94]. The unresolved question is whether the increase in HF is a class effect and the differences in results due to differences in clinical design of the trials or whether the effect has little to do with DPP-4 inhibition. At the present time, these data suggest that if a DPP-4 inhibitor is to be used in patients with T2DM and HF, sitagliptin would be the one to use.
The LEADER trial with the GLP-1 receptor agonist liraglutide showed no specific effect on HF outcome but did show a decrease in CV mortality in participants with T2DM [
95]. Thus, it and perhaps other GLP-1 receptor agonists may be useful in treating the patient with T2DM and HF.
The EMPA-REG CV outcome trial unexpectedly showed that the SGLT-2 inhibitor, empagliflozin, has a unique benefit in decreasing both HF hospitalization and death in participants with T2DM at high risk for CV events [
96•]. The mechanism for this effect is unknown but may be related to its effects in causing weight loss, reducing plasma volume, lowering systolic blood pressure, and/or creating a more ketogenic metabolism. A specific outcome trial of empagliflozin in patients with T2DM and HF is needed to verify that the benefits sufficiently outweigh potential side effects before this can be recommended as a preferred treatment. CV safety trials with other SGLT-2 inhibitors such as canagliflozin and dapagliflozin, which will be completed in the next year or two, will show whether the effects on HF outcomes are class effects.
The US FDA has mandated that all new DM drugs must be evaluated for CV safety. This has spawned a large number of completed and ongoing clinical CV trials (Table
3). The trials are primarily designed with the primary endpoint being development of first major CV events defined as CV death, non-fatal myocardial infarction, and non-fatal stroke. HF has not been a major end point, and the studies are not powered for HF outcomes. Nonetheless, data have emerged from some of these trials which have generated significant interest in the indication or contraindication of some of these drugs for the treatment of T2DM patients with HF (Table
3).
Table 3
Randomized clinical intervention trials that evaluated the effects of specific anti-glycemic agents on either the development or treatment of heart failure in patients with type 2 diabetes
PROactive | Pioglitazone | 5238 | 281/198 P = 0.0001 | 149/108 P = 0.007 | Death 25/22
P = 0.634 | Not recommended |
RECORD | Rosiglitazone | 4447 | | 61/29 P = 0.001 | | Not recommended |
SAVOR-TIMI-53 | Saxagliptin | 16,492 | | 289/228 P = 0.007 | | Not preferred DPP-4 Inhibitor |
EXAMINE | Alogliptin | 5380 | | 106/89 P = 0.15 | | Not preferred DPP-4 Inhibitor |
TECOS | Sitagliptin | 14,671 | | 1.07/1.09 P = 0.98 | 2.54/2.50 P = 0.74 | Preferred DPP-4 Inhibitor |
ELIXA | Lixisenatide | 6068 | | 1.8/1.9 P = 0.75 | | No detrimental effects |
EMPA-REG | Empagliflozin | 7020 | | 9.4/14.5 P = 0.002 | 5.7/8.5 P < 0.001 | Beneficial effects on HF |
LEADER | Liraglutide | 9340 | | 1.2/1.4 P = 0.14 | | Beneficial in reducing CV deaths but not HF |
ORIGIN | Glargine Insulin | 12,612 | | 0.85/0.95 P = 0.16 | | Hypoglycemia is a significant risk for HF patients |