REVIEWS

Effect of glucose-lowering therapies

on heart failure
Michael Nassif and Mikhail Kosiborod
Abstract | Heart failure is one of the most common comorbidities of diabetes mellitus.
Glucose-lowering therapies that can prevent heart failure or improve outcomes in patients with
established heart failure are of critical importance among those with type 2 diabetes. Several types
of glucose-lowering drugs have been assessed in this setting. Metformin has been shown to
modestly improve the outcomes of patients with heart failure, whereas the effect of insulin in those
with established heart failure is less clear. The effect of sulfonylureas on improving heart failure is
controversial; observational reports have suggested that they are harmful in these patients,
but these data have not been confirmed in randomized, controlled trials. Thiazolidinediones are
contraindicated in patients with established heart failure and have also been known to cause heart
failure. Furthermore, certain dipeptidyl peptidase 4 inhibitors seem to increase heart failure
hospitalization. The effects of glucagon-like peptide 1 receptor agonists might differ in patients
with or without established heart failure, particularly those with decompensated heart failure with
a reduced ejection fraction. However, perhaps the most important finding has been that
sodium/glucose cotransporter 2 (SGLT2; also known as SLC5A2) inhibitors reduce heart failure
hospitalizations and, in the case of empagliflozin, markedly reduce the rate of cardiovascular
death. Given the known neutral (or even harmful) effects of other glucose-lowering drugs on heart
failure outcomes, SGLT2 inhibitors might well be considered the drug class of choice in patients
with diabetes and heart failure, or in those at high risk of developing heart failure.

Division of Cardiology,
Saint Luke’s Mid America
Heart Institute, 4401 Wornall
Road, Kansas City,
Missouri 64111, USA.
Correspondence to M.K.
mkosiborod@saint-lukes.org
doi:10.1038/nrcardio.2017.211
Published online 25 Jan 2018
The prevalence of type 2 diabetes mellitus (T2DM)
among patients with heart failure continues to increase
and is now approaching 40% in global clinical trials1–4.
A post-hoc analysis of PARAGIDM‑HF — a contem‑
porary clinical trial of patients with heart failure and
reduced ejection fraction (HFrEF) — showed that even
among patients with no known history of T2DM, 49%
had prediabetes, and 21% had unrecognized T2DM
on the basis of haemoglobin  A1c (HbA1c) levels2,5.
Furthermore, heart failure has emerged as the most
common cardiovascular complication of T2DM, with
its incidence exceeding that of myocardial infarction
or stroke6.
Diabetes is well known to exacerbate all forms of
cardio­vascular disease, particularly heart failure7,8.
T2DM was previously hypothesized to promote athero­
sclerosis, which in turn mediates myocardial ischaemia
and subsequent cardiac dysfunction. However, data
from trials published in the past 2 decades suggest that
the relationship between heart failure and T2DM is far
more complex than previously thought. T2DM can
directly impair the myocardium, leading to increased left
ventricular (LV) hypertrophy and adverse LV remodel­
ling, which results in reduced systolic and diastolic
function9–12. LV mass is significantly greater in patients
with T2DM, even in the absence of coronary disease13–15.
Up to one-third of patients with T2DM have LV hyper‑
trophy and up to two-thirds have echocardiographic
evidence of diastolic dysfunction13–15. The most likely
cause of LV hypertrophy and diastolic dysfunction in
these patients is the formation of ­reactive oxygen ­species
by advanced glycation end products, which can lead to
deposition of collagen in the myocardium and insulin-­
resistance-mediated fibrosis16. No single mechanism can
explain the complex intersection of cardiomyo­pathy and
diabetes: endothelial dysfunction, alterations in calcium
haemostasis, autonomic dysfunction, and shifts in myo‑
cyte metabolic fuel sources almost certainly all have a
contributing role16,17. Subclinical myocardial injury
and necrosis, as measured by serum concentrations of
troponin T measured by high-sensitivity assay, nearly
double or quadruple in patients with prediabetes or
T2DM, respectively, compared with healthy controls
and are associated with marked increases in the risk of

NATURE REVIEWS | CARDIOLOGY ADVANCE ONLINE PUBLICATION | 1

©
2
0
1
8
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
REVIEWS

Key points the types of glucose-lowering therapies used) had no

effect on cardiovascular mortality 23–25 or even increased
• Heart failure is now the most common cardiovascular complication of type 2 diabetes cardio­vascular mortality 26–28. A meta-analysis of four
mellitus (T2DM), with its incidence exceeding that of myocardial infarction or stroke large outcome trials revealed a modest reduction in
• Since 2008, the FDA and the European Medicines Agency require proof of nonfatal myocardial infarction with long-term inten‑
cardiovascular safety for all glucose-lowering agents, resulting in an increase sive control of HbA1c versus conventional therapy but
in trials focusing on the cardiovascular effects of T2DM drugs no differences in all-cause death, cardiovascular death,
• No large randomized, controlled trials have been conducted to examine the effects or heart failure hospitalizations22.
of insulin, metformin, or sulfonylureas in patients with established heart failure
• Use of thiazolidinediones is cautioned in all patients with signs and symptoms Metformin
of heart failure, and they are contraindicated in patients with established heart failureMetformin, administered orally, is currently recom‑
• Different types of dipeptidyl peptidase 4 inhibitors are associated with varying levels mended by practice guidelines as the first-line therapy
of heart failure risk, possibly owing to different selectivity for glucagon-like peptide 1
for the majority of patients with T2DM. However, this
receptor agonists
drug was previously contraindicated in individuals with
• Sodium/glucose cotransporter 2 inhibitors (specifically empagliflozin) are associated heart failure owing to potential concerns regarding the
with significant reductions in heart failure hospitalization and prevention of heart
development of lactic acidosis29. To date, no dedicated,
failure-related and arrhythmia-related deaths
large, randomized, controlled trial has evaluated the
effects of metformin on incident heart failure among
patients with T2DM or in patients with established
cardiovascular death18. Ultimately, these factors contrib‑ heart failure. A pilot study was conducted in 2009 to
ute to a higher risk of developing heart failure, mark‑ assess the feasibility of a large outcomes study that ran‑
edly worsened symptoms, and increased mortality after domly assigned patients with heart failure to either met‑
heart failure ensues. Moreover, incident heart failure is formin or placebo, but the study was terminated owing
emerging as the deadliest cardiovascular complication to futility, which was due to slow enrolment 30. Using
of T2DM, with survival of <25% over 5 years among the Saskatchewan Health Database, which comprised
elderly patients19. 1,833 patients with T2DM and incident heart failure,
Whereas T2DM is clearly a risk factor for the devel‑ metformin (alone or in combination with other drugs)
opment of heart failure, strong evidence indicates that was found to be associated with lower morbidity and
heart failure is also a risk factor for the development of mortality compared with sulfonylurea monotherapy,
T2DM. A Danish cohort study noted a strong associ­ even after controlling for age, sex, heart failure medica‑
ation between diuretic dosage and future development tions, and total physician visits before diagnosis of heart
of T2DM20. Therefore, T2DM and heart failure repre‑ failure31,32 (TABLE 1). In addition, a systematic review
sent both ‘the chicken and the egg’, with both entities of nine observational analyses, which included a total of
exerting synergistic deleterious effects and portending 34,000 patients, found a reduced risk of death that
a poor prognosis19. This complex intersection of T2DM was associ­ated with metformin compared with other
and heart failure is becoming increasingly important, glucose-­lowering drugs, with no increases in adverse
as s­ everal classes of medications developed to reduce events in those with either heart failure or chronic
glucose levels in patients with T2DM have been shown renal insuffi­c iency 32. In response to these positive
potentially to increase the risk of heart failure, whereas findings, the contra­indication warning for metformin
others have neutral or even beneficial effects. These use in patients with heart failure was removed by the
observations have resulted in calls for cardiologists FDA in 2007. Although some clinicians have called for
to begin taking an active role in managing glucose-­ metformin to be considered the glucose-lowering treat‑
lowering medications and for endocrinologists to begin ment of choice for patients with T2DM and heart failure,
tailoring treatments on the basis of individualized given the paucity of data from dedicated studies, such a
cardio­vascular risk. In this Review, we describe different ­recommendation is premature32–34.
classes of glucose-­lowering medications, with a focus on
their effects on heart failure outcomes. Insulin
Reports of insulin-induced oedema were first described
Intensive glycaemic control strategies nearly a century ago35. In both animal models and clin‑
Given that observational studies have revealed a clear ical studies, the effects of insulin on sodium retention
epidemiological association between higher HbA1c levels seemed to occur via amiloride-sensitive sodium chan‑
and a greater risk of developing heart failure, strategies nels in the distal tubules of nephrons36,37. Whether the
to reduce glucose levels were hypothesized to improve antinatriuretic effects of insulin contribute to the devel‑
cardiovascular complications associated with T2DM, opment of heart failure in susceptible individuals with
such as heart failure21. However, numerous large clinical T2DM, or to worsening heart failure among those who
trials that directly compared intensive versus conven‑ already have established disease, is largely unknown.
tional glucose control in patients with T2DM did not Investigators in the ORIGIN trial38 randomly assigned
report any changes in the incidence of heart failure in 12,537 patients with impaired glucose tolerance,
patients receiving intensive glucose-lowering therapy 22. impaired fasting glucose, or diabetes to treatment with
Furthermore, these studies demonstrated that tight insulin glargine or placebo. The majority of patients in
­control of blood glucose (without specific focus on this cohort did not have heart failure at baseline but were

2 | ADVANCE ONLINE PUBLICATION www.nature.com/nrcardio

©
2
0
1
8
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
 REVIEWS

at high risk of cardiovascular events. Over a follow-up
period of 5 years, no differences in hospitalization for
heart failure were observed between the treatment
groups38. The BARI‑2D trial39 (n = 2,368) compared
insulin-­provision therapy (insulin with or without
sulfonyl­ureas) with insulin-sensitization therapy (met‑
formin plus thiazolidinediones). Surprisingly, no differ‑
ences in the rate of incident heart failure were observed
between the two treatment groups (16.6% for insulin
provision versus 19.4% for insulin sensitization). Given
the known association between incident heart failure
and thiazolidinedione use, the researchers speculated
that metformin might be protective or that insulin and
sulfonylureas, alone or in combination, might increase
the risk of heart failure.
To date, no large randomized, controlled trials have
been conducted to examine the effects of insulin on
clinical outcomes in patients with established heart fail‑
ure. Data from retrospective analyses have been con‑
flicting. Retrospective interpretation of the effects of
insulin treatment on heart failure outcomes is difficult
as these effects are often a marker of diabetes severity
and duration rather than a direct contributor to adverse
outcomes. In a post-hoc analysis of the CHARM trial40,
which involved patients with chronic heart failure with
or without LV systolic dysfunction, insulin therapy was
associated with an increased risk of all-cause mortality
(risk ratio 1.25, 95% CI 1.03–1.51). An observational
study involving 554 patients with advanced heart failure
noted that even after adjustments in a Cox multi­variable
analysis, insulin-treated diabetes was found to be an
independent predictor of mortality (HR 4.30, 95% CI
1.69–10.94) whereas non-insulin-treated diabetes was
not (HR 0.95, 95% CI 0.31–2.93)41. The effects of insu‑
lin on mortality were also evaluated in a retrospective
study involving Medicare beneficiaries in the USA with
established heart failure42. After adjusting for duration of
disease and other comorbidities, the investigators found
no association between insulin treatment and mortality
(HR 0.96, 95% CI 0.88–1.05).

Table 1 | Effects of glucose-lowering drugs on HF outcomes

Glucose- Comparator groups HF at n Study design Follow‑up HR (95% CI) for HF Refs
lowering baseline (weeks) hospitalization
agent (%)
Metformin Metformin versus oGLDs (mostly SUs) 100 34,000 Meta-analysis NA 0.92 (0.86–0.98) 32
Metformin versus SUs 0 12,272 Retrospective cohort 130 0.83 (0.70–0.99)* 31
Insulin Insulin + SUs versus metformin + TZDs 7 2,368 RCT 276 16.6% versus 19.4%; P = 0.09 39
Insulin glargine versus placebo NR 12,537 RCT 322.4 0.90 (0.77–1.05) 38
Insulin versus oGLDs 100 554 Retrospective cohort 52 4.3 (1.69–10.94)‡ 41
Insulin versus oGLDs NR 16,417 Retrospective cohort 52 0.96 (0.88–1.05) ‡
42
SUs SUs versus metformin NR 91,521 Retrospective cohort 369 1.18 (1.04–1.34) 48
SU versus metformin NR 17,863 Retrospective cohort NR 1.24 (1.09–1.36) 49
SU versus oGLDs 0 4,075 RCT 556 0.91 (0.52–1.52) 23
TZDs Rosiglitazone ± ramipril versus placebo 0 5,269 RCT 208 7.04 (1.60–31.0) 56
Pioglitazone versus placebo 0 §
5,238 RCT 138 1.49 (1.23–1.80) 57
Rosiglitazone versus placebo 100 224 RCT 52 7.09 (1.60–30.96) 58
Pioglitazone versus placebo 0 §
3,895 RCT 250 3.8% versus 3.7%; P = 0.80 ||
62
DPP4 Saxagliptin versus placebo 13 14,735 RCT 109 1.27 (1.07–1.51) 79
inhibitors
Alogliptin versus placebo 28 5,380 RCT 72 1.19 (0.89–1.58) 82
Sitagliptin versus placebo 18 16,492 RCT 156 1.00 (0.83–1.20) 84
Vildagliptin versus placebo 100 253 RCT 52 NR 87
GLP1 Lixisenatide versus placebo 22 6,068 RCT 108 0.96 (0.75–1.23) 116
receptor
agonists Liraglutide versus placebo 17.8 9,340 RCT 198 0.87 (0.73–1.05) 69
Semaglutide versus placebo 23.6 3,297 RCT 109 1.11 (0.77–1.61) 68
Liraglutide versus placebo 100 300 RCT 26 1.30 (0.89–1.83) 72
Liraglutide versus placebo 100 241 RCT 24 10% versus 3%; P = 0.04¶ 73
SGLT2 Empagliflozin versus placebo 10 7,020 RCT 161 0.65 (0.50–0.85) 93
inhibitors
Canagliflozin versus placebo 14 10,142 RCT 126 0.67 (0.52–0.87) 92
SGLT2 versus oGLDs 6 309,056 Retrospective cohort 64 0.61 (0.51–0.73) 97
*Hazard ratio for death or hospitalization. ‡Hazard ratio for all-cause mortality. §Not reported, but trial specifically excluded patients with NYHA class II–IV HF.
||
Percentage of patients who developed new congestive HF. ¶Composite of major adverse cardiac events. DPP4, dipeptidyl peptidase 4; GLP1, glucagon-like
peptide 1; HF, heart failure; NA, not available; NR, not reported; oGLD, other glucose-lowering drug; RCT, randomized, controlled trial; SGLT2, sodium/glucose
cotransporter type 2; SU, sulfonylurea; TZD, thiazolidinedione.

NATURE REVIEWS | CARDIOLOGY ADVANCE ONLINE PUBLICATION | 3

©
2
0
1
8
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
REVIEWS

Sulfonylureas rosiglitazone, and pioglitazone. The last two agents sub‑

Sulfonylureas are insulin secretagogues that bind to the
pancreatic β‑cell subunits of ATP-sensitive p ­ otassium
channels to keep the channels closed, causing an influx
of calcium ions into the cell that results in an increased
release of insulin via exocytosis43. Unlike exogenous insu‑
lin, use of sulfonylureas is not associated with sodium
retention or oedema37. Investigators in the UKPDS trial23
compared treatment with insulin and sulfonylureas to a
conventional dietary-based treatment in patients with
newly diagnosed T2DM and found no differences in the
rate of incident heart failure in this low-risk population
(HR 0.91, 95% CI 0.54–1.52). However, the investigators
specifically excluded patients with established heart fail‑
ure, and the data revealed an extremely low event rate
(3% of patients developed heart failure over a median
follow-up period of 10 years).
To date, no randomized, controlled trials have exam‑
ined the effects of sulfonylurea treatment in patients with
established heart failure. Nevertheless, data from an
observational study suggested that sulfonylureas increase
the risk of cardiovascular events and heart failure hospi‑
talization44, perhaps through a mechanism linked to the
nonspecific binding of sulfonylureas to ATP-sensitive
potassium channels. Although sulfonylureas are known
to bind to pancreatic β‑cells, they might also bind to
channels in cardiac myocytes and vascular smooth
muscle cells45. The adverse effects of ATP-sensitive
potassium channel closure in patients with T2DM have
mostly been demonstrated with glyburide43. On the basis
of animal studies, the effects of other sulfonylureas such
as chlorpropamide and glipizide seem to be similar to
those of glyburide43. Interestingly, glimepiride does not
seem to have off-target cardiac effects46. Use of glyburide
was associated with a reduction in LV ejection fraction
during cardiopulmonary stress testing 45.
In 2017, a comprehensive meta-analysis showed that
sulfonylurea treatment increased both cardio­vascu­
lar mortality (HR 1.46, 95% CI 1.21–1.77) and all-cause
mortality (HR 1.26, 95% CI 1.10–1.44) compared with
other antihyperglycaemic drugs. Interestingly, the dif‑
ferences in cardiovascular mortality were greatest when
use of sulfonylureas was compared with the use of
glucagon-like peptide 1 (GLP1) receptor agonists and
sodium/glucose cotransporter 2 (SGLT2; also known
as SLC5A2) inhibitors, but the use of sulfonylureas was
still associated with significantly elevated risk of cardio­
vascular death compared with the use of dipeptidyl
peptidase 4 (DPP4) inhibitors, thiazolidinediones, or
insulin47. Numerous large, retrospective analyses have
also noted an increased risk of heart failure hospitaliza‑
tion with the use of sulfonylureas compared with met‑
formin use31,32,48,49. A large, randomized, controlled trial
will be complete in 2018 and will hopefully shed more
light on this subject 50.
Thiazolidinediones
In the late 1990s, a promising ‘insulin-sensitizing’
class of drugs known as thiazolidinediones emerged,
which included troglitazone (the development of which
was eventually halted given its toxicity in the liver),
sequently gained FDA and European Medicines Agency
approval for the treatment of T2DM. Thiazolidinediones
increase insulin sensitivity by acting on peroxisome
proliferator-activated receptor-γ (PPARγ) receptors in
adipose tissue, muscle, and the liver to increase glu‑
cose utilization and decrease glucose production51,52.
Thiazolidinedione use can increase the risk of fluid
retention; peripheral oedema occurred in 4–6% of
patients treated with thiazolidinedione (compared with
1–2% in those receiving a placebo) and to an even higher
degree in patients with a history of heart failure51,53.
Thiazolidinedione-induced oedema is associated with
reduced urinary sodium and water excretion, with a
syner­gistic effect for those receiving concomitant insu‑
lin, suggesting that this fluid retention occurs at the renal
level54,55. Early trials that assessed thiazo­lidinedione use
excluded patients with NYHA class III or IV heart fail‑
ure, but the drugs were commonly prescribed in those
with less severe heart failure. Since the approval of
thiazo­lidinedione for the treatment of diabetes in 1999,
three randomized, controlled trials have further shed
light on whether its use is associated with elevated risk
of heart failure. In the DREAM study 56, rosiglitazone
was given to patients with impaired fasting glucose or
impaired glucose tolerance and no cardiovascular dis‑
ease with the goal of preventing development of T2DM.
The incidence of hospitalization for heart failure in the
rosiglitazone-treated patients was 0.5% versus 0.1% in
the placebo group (HR 7.03, 95% CI 1.60–30.90). The
ProACTIVE study 57 investigators randomly assigned
patients with known cardiovascular disease to pioglita‑
zone treatment or placebo. The secondary end point of
a composite of all-cause mortality, nonfatal myocardial
infarction, and stroke was reached by 301 patients in the
pioglitazone group versus 358 patients in the placebo
group (HR 0.84, 95% CI 0.72–0.98, P = 0.027). However,
hospitalization for heart failure was increased by 50%
in the pioglitazone-treated patients compared with
placebo-­treated patients (relative risk (RR) 1.49, 95% CI
1.23–1.80). The GSK211 trial58 investigators randomly
assigned patients with NYHA class I–II heart failure
to rosiglitazone or placebo, and all patients underwent
dedicated serial echocardiograms. Patients assigned to
rosiglitazone had a significant increase in diuretic use
and oedema; however, no changes in LV function were
observed, supporting the hypothesis that increased
heart failure symptoms with thiazolidinediones are
likely to be related to plasma volume expansion. Finally,
a meta-analysis found that thiazolidinedione treat‑
ment was associated with increased risk of heart failure
across patients at varying degrees of cardiovascular risk
compared with placebo (RR 1.72, 95% CI 1.21–2.42,
P = 0.002). No heterogeneity of effects across studies
was apparent, which was consistent with a class effect
for thiazolidinediones59.
Several research groups hypothesized that a balanced
dual PPARα and PPARγ agonist might alleviate some
of the volume expansion characteristics of the earlier
PPARγ‑only agonists. In the AleCardio trial60, 7,226
patients with T2DM and acute coronary syndrome were

4 | ADVANCE ONLINE PUBLICATION www.nature.com/nrcardio

©
2
0
1
8
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
 REVIEWS

randomly assigned to receive the dual PPARα and PPARγ
agonist aleglitazar or a matching placebo. The trial was
stopped early by the data safety monitoring committee
because of an excess of hospitalizations for heart failure
(in addition to other adverse events) with aleglitazar and
a lack of efficacy. Similarly, the development of the dual
PPARα and PPARγ agonist muraglitazar was halted after
a pooled analysis showed an association with a greater
incidence of myocardial infarction, stroke, transient
ischaemic attacks, and heart failure when compared
with placebo or pioglitazone treatment 61. In the pooled
muraglitazar data (which included four phase III trials
and one phase II trial), adjudicated heart failure events
occurred in 13 of 2,374 muraglitazar-treated patients
(0.55%) compared with one in 1,351 placebo-treated
patients (0.07%; P = 0.053)61.
In 2016, investigators in the IRIS trial62 randomly
assigned 3,876 patients without diabetes who had had
a stroke or transient ischaemic attack to pioglitazone
treatment or placebo groups. Pioglitazone treatment
significantly reduced the primary end point of stroke
and myocardial infarction, and no increase in the inci‑
dence of heart failure was observed62. However, the IRIS
investigators excluded patients with heart failure, and
had a strict protocol to monitor patients for oedema.
At present, the use of thiazolidinediones is cautioned
in all patients with signs and symptoms suggestive of
heart failure, and thiazolidinediones are contraindicated
in patients with established heart failure and NYHA
class III and/or IV symptoms63.
GLP1 receptor agonists
Degradation-resistant GLP1 receptor agonists are inject‑
able agents that lower blood glucose levels via several
mechanisms: they increase glucose-dependent insu‑
lin secretion, lower postprandial glucagon levels, slow
gastric emptying, and cause satiety and reduced calori­
fic intake64. However, GLP1 receptor agonists do not
indepen­dently increase the risk of hypoglycaemia unless
co-administered with insulin or sulfonylureas. Other
potential benefits of GLP1 receptor agonists include
weight loss and modest reduction in systolic blood pres‑
sure65–67, and some agents within the class of GLP1 recep‑
tor agonists (such as liraglutide and semaglutide) have
been shown to reduce cardiovascular risk significantly
compared with placebo68,69. In two randomized cardio‑
vascular outcome trials involving patients with T2DM
at high cardiovascular risk, the rate of cardio­vascular
death, nonfatal myocardial infarction, or nonfatal stroke
was significantly lower in patients receiving GLP1 recep‑
tor agonists (liraglutide and semaglutide) than in those
receiving placebo68,69.
The effects of GLP1 agonists on heart failure, how‑
ever, remain unclear. In a canine model of heart failure,
GLP1 agonists increased myocardial glucose uptake70.
Additionally, in animal models of cardiomyopathy,
a 48‑h infusion of a GLP1 agonist increased stroke
volume and LV ejection fraction, while decreasing LV
end-diastolic pressure and systemic vascular resistance71.
However, these positive benefits of GLP1 agonists in pre‑
clinical models of heart failure have so far not translated
into a clinical benefit. The LEADER68, SUSTAIN‑6
(REF. 67), and ELIXA71 randomized trials all reported no
changes in the rate of heart failure hospitalization with
GLP1 receptor agonist use compared with placebo. All
three trials included patients with T2DM and at high
cardiovascular risk, but only a minority had heart fail‑
ure at baseline. Of note, heart failure hospitalization was
only a secondary or exploratory end point in all three
­trials and was prospectively adjudicated. Furthermore,
no data were collected on natriuretic peptides levels,
either at baseline or during follow-up.
Two small trials were designed specifically to evalu­
ate the effects of liraglutide in patients with established,
symptomatic HFrEF. Investigators in the FIGHT trial72
randomly assigned 300 patients with HFrEF and clinical
decompensation (with or without T2DM) to liraglutide
or placebo. Liraglutide treatment did not reduce the rate
of heart failure hospitalization or cardio­vascu­lar death,
improve disease-specific quality of life, or decrease ­levels
of heart failure biomarkers. Among the subgroup of
patients with T2DM, the composite primary outcome
seemed to be numerically worse for those treated with
liraglutide than for those receiving placebo, but the results
were not significant. Similar findings were observed
in the LIVE study 73, which involved 241 patients with
chronic HFrEF randomly assigned to receive liraglu‑
tide or placebo. After 24 weeks of treatment, no signifi­
cant differences were observed between the groups for
the primary end point of a change in the LV ejection
fraction. However, liraglutide-­treated patients had a
greater number of serious adverse cardiac events than
those assigned to placebo (12 versus 3; P = 0.04). One
potential concern with liraglutide use in patients with
severe HFrEF is that the drug increases heart rate to a
greater extent than other shorter-­acting GLP1 agonists74.
Meier and colleagues demonstrated a mean increase in
heart rate from baseline of 9.3 ± 1.2 bpm with 1.2 mg
of liraglutide to 3.3 ± 1.3 bpm with 20 mg lixisena­tide
(P = 0.001)75. By contrast, the average increase in heart
rate with liraglutide in the larger LEADER trial68 was
much more modest. This observation is particularly
concerning for patients with HFrEF, in whom resting
heart rate has been shown to correlate with morbidity
and mortality 76. In both the FIGHT72 and LIVE73 trials,
a significant increase in heart rate was observed with
liraglutide versus placebo treatment, and both showed
a numerical, but not significant, increase in arrhyth‑
mias72,73. Given the trends towards increased rates of
heart failure hospitaliza­tions in patients enrolled in the
FIGHT trial72 (particularly among those with T2DM),
and increased rates of severe cardiovascular events in
patients enrolled in the LIVE study 73, further investiga‑
tion is needed to assess the safety of GLP1 agonists in
those with advanced HF.
Dipeptidyl peptidase 4 inhibitors
Given that DPP4 is involved in the rapid degradation
of GLP1, the effects of the incretin system could be
enhanced by inhibition of DPP4 (REF. 77). DPP4 inhib­
itors are included in practice guidelines as potential
second-­line agents for T2DM after metformin, although

NATURE REVIEWS | CARDIOLOGY ADVANCE ONLINE PUBLICATION | 5

©
2
0
1
8
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
REVIEWS

the data on their long-term clinical benefits are scarce
and safety might vary between different compounds
within the drug class78. Investigators in the SAVOR
TIMI‑53 trial79 randomly assigned patients with T2DM
at high risk of cardiovascular events to either saxagliptin
or placebo and found no difference between treatment
groups with regard to the primary outcome of a com‑
posite of cardiovascular death, myocardial infarction,
or ischaemic stroke. However, saxagliptin treatment was
associated with a 27% increase in the risk of hospitaliza­
tions for heart failure. This finding was ­unexpected
given that preclinical data suggested that DPP4 inhib­
ition should improve cardiac function80,81. A post-hoc
analy­sis of the SAVOR trial demonstrated that the high‑
est risk of heart failure hospitalization was observed
among those with elevated N‑terminal pro‑B‑type
natriuretic peptide levels79. Furthermore, the EXAMINE
trial82, which involved 5,380 patients with T2DM and
acute coronary syndrome, reported that alogliptin had
no effect on the prespecified end points of all-cause
mortality, nonfatal myocardial infarction, nonfatal
stroke, urgent revascularization due to unstable angina,
and hospital admission for heart failure. However,
an exploratory analysis of EXAMINE in which patients
were stratified by presence or absence of heart failure
showed that alogliptin significantly increased the risk of
heart failure hospitalization in those without a history
of heart failure (HR 1.76, 95% CI 1.07–2.90) but not in
those with a history of heart failure (HR 1.00, 95% CI
0.71–1.42)82. The TECOS trial83,84 is the largest outcomes
trial to date assessing the cardiovascular safety of DPP4
inhibitors. Among nearly 15,000 patients with T2DM
at high risk of cardiovascular events, no differences in
heart failure hospitalizations were observed between
sitagliptin and placebo use. Furthermore, results from
a meta-analysis showed no significant increase in risk
of heart failure hospitalizations with DPP4 inhibition
(overall HR 1.12, 95% CI 0.99–1.25, P = 0.06); however,
substantial hetero­geneity existed between the trials that
were pooled and analysed85.
Although the SAVOR79, EXAMINE82, and TECOS83,84
trials enrolled patients with T2DM and at high risk of
cardiovascular disease, only a minority had estab‑
lished heart failure. Investigators in the VIVIDD trial86
specifi­cally enrolled patients with LV systolic dysfunc‑
tion and T2DM and randomly assigned them to either
vilda­gliptin or placebo. Vildagliptin treatment was not
­inferior to placebo regarding changes in LV ejection
fraction among patients with T2DM and previous heart
failure, but an increase in LV systolic and diastolic vol‑
umes over the 52 weeks of follow-up was observed with
vildagliptin treatment 87. Furthermore, the incidence of
worsening heart failure was similar in the two treatment
groups, but all-cause mortality was numerically higher
in the vildagliptin group.
The reasons underlying the discrepancies in heart
failure risk with different DDP4 inhibitors remain
unclear, despite DPP4 inhibitors being perhaps the most
extensively studied class of glucose-lowering med­icines.
Some researchers have hypothesized that different
DPP4 inhibitors have different selectivity for non‑GLP1
substrates, which accounts for their variations in heart
failure risk88. Further investigation is still needed to
conclude whether an excess heart failure risk exists with
DPP4 inhibition therapy overall as a class, or whether it
pertains only to certain drugs within the class. Results
of the 7,000‑patient CARMELINA trial89, which aims to
investigate the long-term effect of linagliptin treatment
on renal function and cardiovascular morbidity and
mortality, will be presented in early 2018. Furthermore,
the safety of DPP4 inhibitors needs further evaluation
in patients with established heart failure. The ongoing
MEASURE‑HF trial90 is designed to compare the effects
of saxagliptin, sitagliptin, or placebo in patients with
T2DM and established HFrEF and includes a detailed
cardiac MRI assessment of LV size and function.
SGLT2 inhibitors
Although SGLT2 inhibitors only modestly reduce HbA1c
levels compared with other glucose-lowering therapies,
they have an entirely insulin-independent mode of
action through increased urinary excretion of glucose91.
Therefore, SGLT2 inhibitors might be a novel treatment
strategy for patients with comorbid heart failure and
T2DM as they are the first class of glucose-­lowering
agents to demonstrate a robust reduction in heart
failure hospitalizations91–93.
The EMPA-REG OUTCOME trial93 involved 7,020
patients with T2DM and established cardiovascular
disease who were randomly assigned to groups treated
with 10 mg or 25 mg of empagliflozin, or placebo. After
a median treatment period of 3.1 years, significantly
fewer patients who received empagliflozin treatment
compared with placebo experienced a primary outcome
of major adverse cardiac events (10.5% versus 12.1%),
cardiovascular-related death (3.7% versus 5.9%), or all-
cause death (5.7% versus 8.3%)93. No significant differ‑
ences were observed in outcomes between the patients
who received the 10 mg or 25 mg doses of empagliflozin,
and the pooled empagliflozin group had a 38% reduced
risk of cardiovascular death compared with the placebo
group93. Although the trial predominantly involved
patients with T2DM and coronary artery disease (with
only 10% of patients having a known history of heart
failure at baseline), the majority of the benefit seemed to
be related to the highly significant reduction in heart fail‑
ure hospitalization (a 35% RR reduction) and prevention
of heart-failure-related and arrhythmia-related deaths93.
The RR reduction in heart failure hospitalization was
statistically similar in those with or without a history
of heart failure. However, because the overwhelming
majority of patients in the trial did not have heart fail‑
ure at baseline, this reduction seemed to be primarily a
heart failure prevention effect 94. A secondary analysis of
EMPA-REG OUTCOME also found a 39% reduction in
the composite renal end point of progression to macro­
albuminuria, a doubling of the serum creatinine level,
renal replacement therapy, and/or renal death95.
Results from studies involved in the CANVAS pro‑
gramme96 support the beneficial effect of SGLT2 inhibi‑
tors on heart failure hospitalization (FIG. 1). The CANVAS
programme involves both the original canagliflozin

6 | ADVANCE ONLINE PUBLICATION www.nature.com/nrcardio

©
2
0
1
8
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
 REVIEWS

cardiovascular safety trial (which was used to gain FDA
approval for SGLT2 inhibitor use in patients with T2DM
in 2013)96 and a separate CANVAS‑R trial92 designed
specifically to demonstrate the cardiovascular benefit
of SGLT2 inhibitors. Investigators from the CANVAS
programme enrolled a total of 10,142 patients with either
established cardiovascular disease (65%) or a high risk
of cardiovascular events (35%) and randomly assigned
them to canagliflozin (100 mg or 300 mg) or placebo92,96.
The rate of the primary outcome (nonfatal myocardial
infarction or stroke, or cardiovascular-related death) was
lower in the pooled canagliflozin treatment group than
in the placebo group (26.9 versus 31.5 per 1,000 patient-
years)92. Although the reduction in cardiovascular and
all-cause death with canagliflozin versus placebo treat‑
ment did not reach significance, patients in the pooled
canagliflozin treatment group showed a significant 33%
RR reduction in heart failure hospitalization92. However,
adverse effects occurred more frequently with canagli‑
flozin than with placebo, most notably lower extrem‑
ity amputations (6.3 versus 3.4 per 1,000 patient-years)
and fractures (15.4 versus 11.9 per 1,000 patient-years)92.
Furthermore, compared with placebo, canagliflozin
signifi­cantly improved the composite renal end point of
a sustained 40% reduction in the estimated glomerular
filtration rate, the need for renal replacement therapy,
or death from renal causes.
Real-world data from a large multinational, non­
interventional study that combined data from well-­
established registries across six countries also support
the notion of a class-wide benefit of SGLT2 inhibitors on
heart failure outcomes. The CVD-REAL study 97 ­analysed
>300,000 patients with T2DM and compared the heart
failure outcomes of those who were newly i­nitiated
on SGLT2 inhibitors with those who were started on
other glucose-lowering medications. The main analysis
(matched 1:1 with the use of a propensity score method‑
ology) demonstrated a 39% RR reduction in heart failure
hospitalizations that was associated with SGLT2 inhib­
itor use versus the use of other glucose-lowering drugs97.
These reductions were also observed for the outcome of
total heart failure events and were consistent in patients
with or without established heart failure98.
On the basis of results of the EMPA-REG OUTCOME
trial93, in December 2016, the FDA granted empagli­
flozin an indication for reduction in cardiovascular
death, a landmark decision never before seen in the
history of glucose-lowering treatments99. However,
the mechanism of action through which SGLT2 inhib‑
itors might exert a benefit on heart failure remains
unclear and is the subject of intense investigation100.
Notably, the effects of SGLT2 inhibitors on heart fail‑
ure hospitalization, renal outcomes, and cardiovascular
mortality (in the case of empagliflozin) become apparent
within weeks of treatment and are maintained for several
years93. Mechanisms beyond glucose lowering or diur­
esis per se are likely to underlie the dramatic reduction
in heart failure events91. Several animal studies aimed at
evaluating the mechanisms underlying SGLT2 inhibitor
activity have shown that the drug can reduce oxidative
stress, improve endothelial function, and modulate
Trial n Number of events HR (95% CI)
HF hospitalization
EMPA-REG 7,020 221 0.65 (0.50–0.85)
CANVAS 10,142 243 0.67 (0.52–0.87)
CVD-REAL 196,802 423 0.61 (0.51–0.73)
HF hospitalization or death
EMPA-REG 7,020 463 0.66 (0.55–0.79)
CANVAS 10,142 652 0.78 (0.67–0.91)
CVD-REAL* 215,622 1,983 0.54 (0.48–0.60)
0.5 1 2
HR
Favours SGLT2 inhibitors
Figure 1 | HF outcomes with different sSGLT2| inhibitors.
Nature Reviews Cardiology
In randomized, controlled trials and large cohort studies,
sodium/glucose cotransporter 2 (SGLT2) inhibitors were
associated with reduced risks of heart failure (HF)
hospitalization and death. *CVD-REAL was a retrospective
cohort study, not a randomized, controlled trial.
neurohormonal pathways in addition to having anti-­
inflammatory effects91,101. SGLT2 inhibitors have also
been postulated to reduce plasma volume without neuro­
hormonal activation102 or possibly alter metabolic fuel
sources away from glucose oxidation towards metabo‑
lism of free fatty acids and ketone bodies, which might
have a role in improving myocardial efficiency and func‑
tion103. However, others have hypothesized that the main
driver of benefit might be the unique effects of SGLT2
inhibitors on renal sodium and glucose h ­ andling, which
lead to improvements in diabetes-related maladaptive
afferent renal arteriolar vasoconstriction 104. Given
the favourable effects of SGLT2 inhibitors on visceral
adiposity, haematocrit levels, and blood pressure and
lipid profiles, multiple modes of action are likely to be
involved. Results from the biomarker substudy of the
CANVAS trial support a pleiotropic effect of SGLT2
inhibitors105. In older patients with T2DM (who were
predominantly free from prevalent cardiovascular dis‑
ease), canagliflozin was associated with reduced levels of
serum N‑terminal pro‑B‑type natriuretic peptide (a bio‑
marker of increased filling pressures linked to higher risk
of developing heart failure) and troponin I measured by
high-sensitivity assay (a biomarker of subclinical myo‑
cardial necrosis), as compared with placebo, at 26, 52,
and 104 weeks105.
Although the effect of SGLT2 inhibitors on heart fail‑
ure hospitalization is impressive, both the EMPA-REG
OUTCOME study 93 and the CANVAS programme96
enrolled very few patients with established heart fail‑
ure. For this reason, current ESC guidelines have stated
that empagliflozin might be used to delay the onset of
heart failure in patients with T2DM, but the guidelines
include no mention of SGLT2 inhibitors as a potential
treatment for patients with established heart failure106.
Whether the cardioprotective effects of empagliflozin
(and all SGLT2 inhibitors) can translate to benefits for
those with established heart failure remains to be deter‑
mined. Furthermore, little is known about the heart
failure characteristics of those patients with a history of
heart failure enrolled in the EMPA-REG OUTCOME93
and CANVAS96 studies; therefore, ascertaining whether

NATURE REVIEWS | CARDIOLOGY ADVANCE ONLINE PUBLICATION | 7

©
2
0
1
8
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
REVIEWS

the benefit of the drug is likely to extend to patients with
HFrEF or heart failure with preserved ejection fraction
(HFpEF) is also premature. Likewise, whether patients
without diabetes or with prediabetes could derive any
benefits from SGLT2 inhibitors remains to be seen.
However, these uncertainties might be addressed by
three ongoing large trials assessing outcomes of patients
with well-phenotyped heart failure. The EMPEROR-
PRESERVED 107 and REDUCED 108 trials will com‑
pare the effects of empagliflozin with placebo in both
HFpEF and HFrEF populations in separate patient trials
powered for heart failure outcomes. Furthermore, the
ongoing DAPA-HF trial109 will evaluate the effects of
dapagliflozin versus placebo on heart failure outcomes in
4,500 patients with HFrEF. Several smaller randomized
trials also aim to evaluate the effects of SGLT2 inhibi‑
tors on biomarkers, health status, and filling pressures
in patients with various heart failure phenotypes110,111.
Conclusions
Heart failure is one of the most common and morbid
complications of T2DM. The use of glucose-lowering
therapies that can prevent heart failure, and/or improve
outcomes in patients with established heart failure is of
critical importance for those with T2DM. Metformin
seems to be associated with a modest but favourable
effect on heart failure events, whereas insulin seems to
have a neutral effect on the development of heart failure.
Furthermore, sulfonylurea use might be associated with
harmful cardiovascular effects, but these effects have
not been confirmed in randomized, controlled trials.
Unfortunately, data on all three classes of medications
(metformin, sulfonylureas, and insulin) from random­
ized, controlled trials are scarce with regard to heart fail‑
ure outcomes, leaving large gaps in knowledge in this
patient population.
Following the regulatory guidance change for assess‑
ment of novel glucose-lowering agents in 2008 (REF. 112),
both the FDA and the European Medicines Agency now
require proof of cardiovascular safety, resulting in a
drama­tic growth in clinical investigations focusing on the
cardiovascular effects of drugs developed for T2DM113,114.
As a result of these large cardiovascular outcomes trials,
>180,000 patients with established cardiovascular disease
or at high risk of cardiac events have been assessed in
numerous trials of glucose-lowering therapies. Nearly all
these cardiovascular safety trials have had major adverse
cardiovascular events as their primary outcome, with
heart failure hospitalization being treated as a secondary
or exploratory outcome despite heart failure being argu‑
ably the most important cardiovascular complication of
T2DM. Additionally, whereas patients included in these
trials have established cardiovascular disease or are at
high risk of cardiovascular events, very few have estab‑
lished heart failure. Given the potential harm or benefit of
certain classes of glucose-lowering drugs on the develop­
ment of heart failure, heart failure outcomes need to be
systematically evaluated in appropriately powered stud‑
ies. Of note, an ongoing large cardiovascular outcome
trial of dapagliflozin (the DECLARE TIMI‑58 study 115)
has included a co-­primary end point of cardiovascular
death or heart failure hospitalization and major adverse
cardiovascular events.
Even with heart failure as an exploratory end point, the
cardiovascular outcome trials of novel glucose-­lowering
compounds have revealed important and surprising
results with regard to heart failure risk. Although the
increased risk of thiazolidinedione on heart failure has
now been established, the increase in heart failure hospi‑
talization with certain DPP4 inhibitors was u ­ nexpected.
Furthermore, the role of GLP1 receptor agonists among
patients with heart failure remains unclear, and their
effects might differ between patients with or without
established HF, particularly those with decompen­sated
HFrEF. Perhaps the most important finding is that SGLT2
inhibitor use is associated with a reduced rate of heart
failure hospitalizations (and, in the case of empagli‑
flozin, markedly reduces cardiovascular death). Before
the EMPA-REG OUTCOME trial93, no single class of
glucose-lowering compounds had ever been shown to
reduce the risk of heart failure events. However, whether
the benefits of SGLT2 inhibitors in preventing heart
failure will extend to patients with established HFpEF
or HFrEF is still unknown and will be addressed in
­ongoing heart failure outcome trials. Nevertheless, given
the known neutral (or even harmful) effects of other
glucose-­lowering drugs on heart failure outcomes, SGLT2
inhibitors might well be, and perhaps should be, consid‑
ered the drug class of choice in those with heart failure
or at high risk of develop­ing heart failure. Moreover,
although many questions are left unanswered and much
more work is still to be done in this field, the opportu‑
nity not only to demonstrate safety but also to confirm
­heart-failure-related b
­ enefits is a welcome change.

1. Cohen-Solal, A., Beauvais, F. & Logeart, D. Heart
failure and diabetes mellitus: epidemiology and
management of an alarming association. J. Card. Fail.
14, 615–625 (2008).
2. McMurray, J. J. et al. Angiotensin-neprilysin inhibition
versus enalapril in heart failure. N. Engl. J. Med. 371,
993–1004 (2014).
3. Kannel, W. B., Hjortland, M. & Castelli, W. P. Role of
diabetes in congestive heart failure: the Framingham
study. Am. J. Cardiol. 34, 29–34 (1974).
4. Adams, K. F. Jr. et al. Characteristics and outcomes
of patients hospitalized for heart failure in the
United States: rationale, design, and preliminary
observations from the first 100,000 cases in the
Acute Decompensated Heart Failure National Registry
(ADHERE). Am. Heart J. 149, 209–216 (2005).
5. Kristensen, S. L. et al. Risk related to pre–
diabetes mellitus and diabetes mellitus in heart
failure with reduced ejection fraction. Insights
from prospective comparison of ARNI with ACEI
to determine impact on global mortality and
morbidity in heart failure trial. Heart Fail. 9, e002560
(2016).
6. Parving, H.‑H. et al. Cardiorenal end points in a trial
of aliskiren for type 2 diabetes. N. Engl. J. Med. 367,
2204–2213 (2012).
7. Kannel, W. B. & McGee, D. L. Diabetes and
cardiovascular disease. The Framingham study. JAMA
241, 2035–2038 (1979).
8. Mak, K. H. & Topol, E. J. Emerging concepts in
the management of acute myocardial infarction in
patients with diabetes mellitus. J. Am. Coll. Cardiol.
35, 563–568 (2000).
9. Devereux, R. B. et al. Impact of diabetes on cardiac
structure and function: the strong heart study.
Circulation 101, 2271–2276 (2000).
10. Abbott, R. D., Donahue, R. P., Kannel, W. B.
& Wilson, P. W. The impact of diabetes on survival
following myocardial infarction in men versus women.
The Framingham Study. JAMA 260, 3456–3460
(1988).
11. Bella, J. N. et al. Separate and joint effects of systemic
hypertension and diabetes mellitus on left ventricular
structure and function in American Indians (the Strong
Heart Study). Am. J. Cardiol. 87, 1260–1265 (2001).
12. Palmieri, V. et al. Effect of type 2 diabetes mellitus
on left ventricular geometry and systolic function
in hypertensive subjects: Hypertension Genetic
Epidemiology Network (HyperGEN) study. Circulation
103, 102–107 (2001).
13. Struthers, A. D. & Morris, A. D. Screening for and
treating left-ventricular abnormalities in diabetes
mellitus: a new way of reducing cardiac deaths. Lancet
359, 1430–1432 (2002).

8 | ADVANCE ONLINE PUBLICATION www.nature.com/nrcardio

©
2
0
1
8
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
 REVIEWS

14. Di Bonito, P. et al. Early detection of diabetic
cardiomyopathy: usefulness of tissue Doppler imaging.
Diabet Med. 22, 1720–1725 (2005).
15. Liu, J. E. et al. The impact of diabetes on left
ventricular filling pattern in normotensive and
hypertensive adults: the Strong Heart Study. J. Am.
Coll. Cardiol. 37, 1943–1949 (2001).
16. Kasznicki, J. & Drzewoski, J. Heart failure in the
diabetic population — pathophysiology, diagnosis and
management. Arch. Med. Sci. 10, 546–556 (2014).
17. Fang, Z. Y., Prins, J. B. & Marwick, T. H. Diabetic
cardiomyopathy: evidence, mechanisms, and
therapeutic implications. Endocr. Rev. 25, 543–567
(2004).
18. Selvin, E. et al. Diabetes mellitus, prediabetes,
and incidence of subclinical myocardial damage.
Circulation 130, 1374–1382 (2014).
19. Bertoni, A. G. et al. Heart failure prevalence,
incidence, and mortality in the elderly with diabetes.
Diabetes Care 27, 699–703 (2004).
20. Demant, M. N. et al. Association of heart failure
severity with risk of diabetes: a Danish nationwide
cohort study. Diabetologia 57, 1595–1600 (2014).
21. Aguilar, D., Bozkurt, B., Ramasubbu, K. & Deswal, A.
Relationship of hemoglobin A1C and mortality in heart
failure patients with diabetes. J. Am. Coll. Cardiol. 54,
422–428 (2009).
22. Turnbull, F. M. et al. Intensive glucose control
and macrovascular outcomes in type 2 diabetes.
Diabetologia 52, 2288–2298 (2009).
23. Turner, R. Intensive blood-glucose control with
sulphonylureas or insulin compared with conventional
treatment and risk of complications in patients with
type 2 diabetes (UKPDS 33). Lancet 352, 837–853
(1998).
24. Group, T. A. C. Intensive blood glucose control and
vascular outcomes in patients with type 2 diabetes.
N. Engl. J. Med. 358, 2560–2572 (2008).
25. Duckworth, W. et al. Glucose control and vascular
complications in veterans with type 2 diabetes.
N. Engl. J. Med. 360, 129–139 (2009).
26. Group, T. A. S. Effects of intensive blood-pressure
control in type 2 diabetes mellitus. N. Engl. J. Med.
362, 1575–1585 (2010).
27. Eshaghian, S., Horwich, T. B. & Fonarow, G. C.
An unexpected inverse relationship between HbA1c
levels and mortality in patients with diabetes and
advanced systolic heart failure. Am. Heart J. 151,
91.e1–91.e6 (2006).
28. Action to Control Cardiovascular Risk in Diabetes
Study Group. Effects of intensive glucose lowering in
type 2 diabetes. N. Engl. J. Med. 358, 2545–2559
(2008).
29. Stades, A. M., Heikens, J. T., Erkelens, D. W.,
Holleman, F. & Hoekstra, J. B. Metformin and lactic
acidosis: cause or coincidence? A review of case
reports. J. Intern. Med. 255, 179–187 (2004).
30. Eurich, D. T. et al. Metformin treatment in diabetes
and heart failure: when academic equipoise meets
clinical reality. Trials 10, 12 (2009).
31. Eurich, D. T., Majumdar, S. R., McAlister, F. A.,
Tsuyuki, R. T. & Johnson, J. A. Improved clinical
outcomes associated with metformin in patients
with diabetes and heart failure. Diabetes Care 28,
2345–2351 (2005).
32. Eurich, D. T. et al. Comparative safety and
effectiveness of metformin in patients with diabetes
mellitus and heart failure: systematic review of
observational studies involving 34,000 patients.
Circ. Heart Fail. 6, 395–402 (2013).
33. Bailey, C. J. & Turner, R. C. Metformin. N. Engl.
J. Med. 334, 574–579 (1996).
34. Giamouzis, G., Triposkiadis, F. & Butler, J. Metformin
use in patients with diabetes mellitus and heart failure:
friend or foe? J. Card. Fail. 16, 207–210 (2010).
35. Leifer, A. A case of insulin edema. J. Am. Med. Assoc.
90, 610–611 (1928).
36. Skøtt, P. et al. Effects of insulin on kidney function and
sodium excretion in healthy subjects. Diabetologia 32,
694–699 (1989).
37. DeFronzo, R. A., Cooke, C. R., Andres, R.,
Faloona, G. R. & Davis, P. J. The effect of insulin on
renal handling of sodium, potassium, calcium, and
phosphate in man. J. Clin. Invest. 55, 845–855
(1975).
38. Investigators, T. O. T. Basal insulin and cardiovascular
and other outcomes in dysglycemia. N. Engl. J. Med.
367, 319–328 (2012).
39. Group, T. B. D. S. A. Randomized trial of therapies for
type 2 diabetes and coronary artery disease. N. Engl.
J. Med. 360, 2503–2515 (2009).
40. Pocock, S. J. et al. Predictors of mortality and
morbidity in patients with chronic heart failure.
Eur. Heart J. 27, 65–75 (2006).
41. Smooke, S., Horwich, T. B. & Fonarow, G. C. Insulin-
treated diabetes is associated with a marked increase
in mortality in patients with advanced heart failure.
Am. Heart J. 149, 168–174 (2005).
42. Masoudi, F. A. et al. Thiazolidinediones, metformin,
and outcomes in older patients with diabetes and
heart failure: an observational study. Circulation 111,
583–590 (2005).
43. Bell, D. S. H. Do sulfonylurea drugs increase the risk
of cardiac events? CMAJ 174, 185–186 (2006).
44. Simpson, S. H., Majumdar, S. R., Tsuyuki, R. T.,
Eurich, D. T. & Johnson, J. A. Dose-response relation
between sulfonylurea drugs and mortality in type 2
diabetes mellitus: a population-based cohort study.
CMAJ 174, 169–174 (2006).
45. Scognamiglio, R. et al. Effects of treatment with
sulfonylurea drugs or insulin on ischemia-induced
myocardial dysfunction in type 2 diabetes. Diabetes
51, 808–812 (2002).
46. Bijlstra, P. J., Lutterman, J. A., Russel, F. G., Thien, T.
& Smits, P. Interaction of sulphonylurea derivatives
with vascular ATP-sensitive potassium channels in
humans. Diabetologia 39, 1083–1090 (1996).
47. Bain, S. et al. Cardiovascular events and all-cause
mortality associated with sulphonylureas compared
with other antihyperglycaemic drugs: a Bayesian
meta-analysis of survival data. Diabetes Obes. Metab.
19, 329–335 (2017).
48. Tzoulaki, I. et al. Risk of cardiovascular disease and all
cause mortality among patients with type 2 diabetes
prescribed oral antidiabetes drugs: retrospective
cohort study using UK general practice research
database. BMJ 339, b4731 (2009).
49. Pantalone, K. M. et al. The risk of developing coronary
artery disease or congestive heart failure, and overall
mortality, in type 2 diabetic patients receiving
rosiglitazone, pioglitazone, metformin, or
sulfonylureas: a retrospective analysis. Acta Diabetol.
46, 145–154 (2009).
50. US National Library of Medicine. ClinicalTrials.gov
https://clinicaltrials.gov/ct2/show/NCT01243424
(2017).
51. Yki-Jarvinen, H. Thiazolidinediones. N. Engl. J. Med.
351, 1106–1118 (2004).
52. Nolan, J. J., Ludvik, B., Beerdsen, P., Joyce, M.
& Olefsky, J. Improvement in glucose tolerance and
insulin resistance in obese subjects treated with
troglitazone. N. Engl. J. Med. 331, 1188–1193
(1994).
53. Seki, G. et al. Role of renal proximal tubule transport
in thiazolidinedione-induced volume expansion.
World J. Nephrol. 1, 146–150 (2012).
54. Yang, T. & Soodvilai, S. Renal and vascular
mechanisms of thiazolidinedione-induced fluid
retention. PPAR Res. 2008, 943614 (2008).
55. Raskin, P. et al. A randomized trial of rosiglitazone
therapy in patients with inadequately controlled
insulin-treated type 2 diabetes. Diabetes Care 24,
1226–1232 (2001).
56. Investigators, D. T. et al. Effects of ramipril and
rosiglitazone on cardiovascular and renal outcomes
in people with impaired glucose tolerance or
impaired fasting glucose: results of the Diabetes
REduction Assessment with ramipril and rosiglitazone
Medication (DREAM) trial. Diabetes Care 31,
1007–1014 (2008).
57. Dormandy, J. A. et al. Secondary prevention of
macrovascular events in patients with type 2
diabetes in the PROactive Study (PROspective
pioglitAzone Clinical Trial in macroVascular Events):
A randomised controlled trial. Lancet 366,
1279–1289 (2005).
58. Dargie, H. J. et al. A randomized, placebo-controlled
trial assessing the effects of rosiglitazone on
echocardiographic function and cardiac status
in type 2 diabetic patients with new york heart
association functional class I or II heart failure.
J. Am. Coll. Cardiol. 49, 1696–1704 (2007).
59. Lago, R. M., Singh, P. P. & Nesto, R. W. Congestive
heart failure and cardiovascular death in patients
with prediabetes and type 2 diabetes given
thiazolidinediones: a meta-analysis of randomised
clinical trials. Lancet 370, 1129–1136 (2007).
60. Lincoff, A. et al. Effect of aleglitazar on cardiovascular
outcomes after acute coronary syndrome in patients
with type 2 diabetes mellitus: the alecardio
randomized clinical trial. JAMA 311, 1515–1525
(2014).
61. Nissen, S. E., Wolski, K. & Topol, E. J. Effect
of muraglitazar on death and major adverse
cardiovascular events in patients with type 2 diabetes
mellitus. JAMA 294, 2581–2586 (2005).
62. Kernan, W. N. et al. Pioglitazone after ischemic stroke
or transient ischemic attack. N. Engl. J. Med. 374,
1321–1331 (2016).
63. Nesto, R. W. et al. Thiazolidinedione use, fluid
retention, and congestive heart failure: a consensus
statement from the American Heart Association
and American Diabetes Association. Circulation 108,
2941–2948 (2003).
64. Thrainsdottir, I., Malmberg, K., Olsson, A.,
Gutniak, M. & Rydén, L. Initial experience with GLP‑1
treatment on metabolic control and myocardial
function in patients with type 2 diabetes mellitus
and heart failure. Diabetes Vasc. Dis. Res. 1, 40–43
(2004).
65. Liu, F. P. et al. Glucagon-like peptide 1 receptor
agonist therapy is more efficacious than insulin
glargine for poorly controlled type 2 diabetes:
a systematic review and meta-analysis. J. Diabetes
7, 322–328 (2015).
66. Monami, M., Dicembrini, I., Nardini, C., Fiordelli, I.
& Mannucci, E. Effects of glucagon-like peptide‑1
receptor agonists on cardiovascular risk: a meta-
analysis of randomized clinical trials. Diabetes,
Obes. Metabolism 16, 38–47 (2014).
67. Vilsbøll, T., Christensen, M., Junker, A. E., Knop, F. K.
& Gluud, L. L. Effects of glucagon-like peptide‑1
receptor agonists on weight loss: systematic review
and meta-analyses of randomised controlled trials.
BMJ 344, d7771 (2012).
68. Marso, S. P. et al. Semaglutide and cardiovascular
outcomes in patients with type 2 diabetes. N. Engl.
J. Med. 375, 1834–1844 (2016).
69. Marso, S. P. et al. Liraglutide and cardiovascular
outcomes in type 2 diabetes. N. Engl. J. Med. 375,
311–322 (2016).
70. Bhashyam, S. et al. Glucagon-like peptide‑1 increases
myocardial glucose uptake via p38alpha MAP
kinase‑mediated, nitric oxide-dependent mechanisms
in conscious dogs with dilated cardiomyopathy.
Circ. Heart Fail. 3, 512–521 (2010).
71. Nikolaidis, L. A. et al. Recombinant glucagon-like
peptide‑1 increases myocardial glucose uptake and
improves left ventricular performance in conscious
dogs with pacing-induced dilated cardiomyopathy.
Circulation 110, 955–961 (2004).
72. Margulies, K. B. et al. Effects of liraglutide on clinical
stability among patients with advanced heart failure
and reduced ejection fraction: a randomized clinical
trial. JAMA 316, 500–508 (2016).
73. Jorsal, A. et al. Effect of liraglutide, a glucagon-like
peptide‑1 analogue, on left ventricular function in
stable chronic heart failure patients with and without
diabetes (LIVE) — a multicentre, double-blind,
randomised, placebo-controlled trial. Eur. J. Heart
Fail. 19, 69–77 (2017).
74. Lorenz, M. et al. Differential effects of glucagon-
like peptide‑1 receptor agonists on heart rate.
Cardiovasc. Diabetol. 16, 6 (2017).
75. Meier, J. J. et al. Contrasting effects of lixisenatide
and liraglutide on postprandial glycemic control,
gastric emptying, and safety parameters in patients
with type 2 diabetes on optimized insulin glargine with
or without metformin: a randomized, open-label trial.
Diabetes Care 38, 1263–1273 (2015).
76. Bohm, M. et al. Heart rate as a risk factor in chronic
heart failure (SHIFT): the association between heart
rate and outcomes in a randomised placebo-controlled
trial. Lancet 376, 886–894 (2010).
77. Deacon, C. F. et al. Both subcutaneously and
intravenously administered glucagon-like peptide I are
rapidly degraded from the NH2‑terminus in type II
diabetic patients and in healthy subjects. Diabetes 44,
1126–1131 (1995).
78. Inzucchi, S. E. et al. Management of hyperglycaemia in
type 2 diabetes: a patient-centered approach. Position
statement of the American Diabetes Association (ADA)
and the European Association for the Study of
Diabetes (EASD). Diabetologia 55, 1577–1596
(2012).
79. Scirica, B. M. et al. Heart failure, saxagliptin, and
diabetes mellitus: Observations from the SAVOR-TIMI
53 randomized trial. Circulation 130, 1579–1588
(2014).
80. dos Santos, L. et al. Circulating dipeptidyl peptidase IV
activity correlates with cardiac dysfunction in human
and experimental heart failure. Circ. Heart Fail. 6,
1029–1038 (2013).

NATURE REVIEWS | CARDIOLOGY ADVANCE ONLINE PUBLICATION | 9

©
2
0
1
8
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
REVIEWS

81. Shigeta, T. et al. Dipeptidyl peptidase‑4
modulates left ventricular dysfunction in chronic
heart failure via angiogenesis-dependent and
-independent actions. Circulation 126, 1838–1851
(2012).
82. Zannad, F., Cannon, C. P. & Cushman, W. C. Heart
failure and mortality outcomes in patients with type 2
diabetes taking alogliptin versus placebo in EXAMINE:
a multicentre randomised double-blind trial. Lancet
385, 2067–2076 (2015).
83. McGuire, D. K. et al. Association between sitagliptin
use and heart failure hospitalization and related
outcomes in type 2 diabetes mellitus: secondary
analysis of a randomized clinical trial. JAMA Cardiol.
1, 126–135 (2016).
84. Green, J. B. et al. Effect of sitagliptin on cardiovascular
outcomes in type 2 diabetes. N. Engl. J. Med. 373,
232–242 (2015).
85. Kankanala, S. R. et al. Cardiovascular safety of
dipeptidyl peptidase‑4 inhibitors: recent evidence on
heart failure. Am. J. Transl Res. 8, 2450–2458
(2016).
86. McMurray, J. J. V. et al. Effects of vildagliptin on
ventricular function in patients with type 2 diabetes
mellitus and heart failure: a randomized placebo-
controlled trial. JACC Heart Fail. http://dx.doi.org/
10.1016/j.jchf.2017.08.004 (2017).
87. McMurray, J. The Vildagliptin in Ventricular
Dysfunction Diabetes trial (VIVIDD) in European Heart
Failure Congress (Lisbon, 2013).
88. Scirica, B. M. The safety of dipeptidyl peptidase 4
inhibitors and the risk for heart failure. JAMA Cardiol.
1, 123–125 (2016).
89. US National Library of Medicine. ClinicalTrials.gov
https://clinicaltrials.gov/ct2/show/NCT01897532
(2017).
90. US National Library of Medicine. ClinicalTrials.gov
https://clinicaltrials.gov/ct2/show/NCT02917031
(2017).
91. Inzucchi, S. E. et al. SGLT‑2 inhibitors and
cardiovascular risk: proposed pathways and review
of ongoing outcome trials. Diab. Vasc. Dis. Res. 12,
90–100 (2015).
92. Neal, B. et al. Canagliflozin and cardiovascular and
renal events in type 2 diabetes. N. Engl. J. Med. 377,
644–657 (2017).
93. Zinman, B. et al. Empagliflozin, cardiovascular
outcomes, and mortality in type 2 diabetes. N. Engl.
J. Med. 373, 2117–2128 (2015).
94. Fitchett, D. et al. Heart failure outcomes with
empagliflozin in patients with type 2 diabetes at
high cardiovascular risk: results of the EMPA-REG
OUTCOME® trial. Eur. Heart J. 37, 1526–1534
(2016).
95. Wanner, C. et al. Empagliflozin and progression of
kidney disease in type 2 diabetes. N. Engl. J. Med.
375, 323–334 (2016).
96. Neal, B. et al. Rationale, design, and baseline
characteristics of the Canagliflozin Cardiovascular
Assessment Study (CANVAS) — a randomized
placebo-controlled trial. Am. Heart J. 166, 217–223.
e11 (2013).
97. Kosiborod, M. et al. Lower risk of heart failure and
death in patients initiated on sglt‑2 inhibitors versus
other glucose-lowering drugs: the CVD-REAL study.
Circulation 136, 249–259 (2017).
98. Scheerer, M. et al. Gesamtereignisraten für die
hospitalisierung bei herzinsuffizienz (HHI) bei
neueinstellung auf SGLT‑2‑hemmer im vergleich zu
anderen antidiabetika–daten aus dem klinischen
alltag aus Deutschland, Schweden und Norwegen mit
mehr als 45.000 typ 2 diabetes patienten (CVD-Real).
Diabetologie und Stoffwechsel 12, 222–228 (2017).
99. [No authors listed.] FDA approves Jardiance to reduce
cardiovascular death in adults with type 2 diabetes.
U.S. Food and Drug Administration http://www.fda.gov/
NewsEvents/Newsroom/PressAnnouncements/
ucm531517.htm (2016).
100. Husten, L. CardioBrief: Specialty rift brewing
over empagliflozin. MedPage Today http://
www.medpagetoday.com/cardiology/cardiobrief/
58907 (2016).
101. Oelze, M. et al. The sodium-glucose co‑transporter 2
inhibitor empagliflozin improves diabetes-induced
vascular dysfunction in the streptozotocin diabetes rat
model by interfering with oxidative stress and
glucotoxicity. PLoS ONE 9, e112394 (2014).
102. Lambers Heerspink, H. J., de Zeeuw, D., Wie, L.,
Leslie, B. & List, J. Dapagliflozin a glucose-regulating
drug with diuretic properties in subjects with type 2
diabetes. Diabetes Obes. Metab. 15, 853–862
(2013).
103. Mudaliar, S., Alloju, S. & Henry, R. R. Can a shift in
fuel energetics explain the beneficial cardiorenal
outcomes in the EMPA-REG OUTCOME study?
A unifying hypothesis. Diabetes Care 39, 1115–1122
(2016).
104. Sattar, N., McLaren, J., Kristensen, S. L., Preiss, D.
& McMurray, J. J. SGLT2 Inhibition and cardiovascular
events: why did EMPA-REG Outcomes surprise and
what were the likely mechanisms? Diabetologia 59,
1333–1339 (2016).
105. Januzzi, J. L. et al. Effects of Canagliflozin on
cardiovascular biomarkers in older adults with type 2
diabetes. J. Am. Coll. Cardiol. 70, 704–712 (2017).
106. Ponikowski, P. et al. 2016 ESC guidelines for the
diagnosis and treatment of acute and chronic heart
failure. Eur. J. Heart Fail. 18, 891–975 (2016).
107. US National Library of Medicine. ClinicalTrials.gov
https://clinicaltrials.gov/ct2/show/NCT03057951
(2017).
108. US National Library of Medicine. ClinicalTrials.gov
https://clinicaltrials.gov/ct2/show/NCT03057977
(2017).
109. US National Library of Medicine. ClinicalTrials.gov
https://clinicaltrials.gov/ct2/show/NCT03036124
(2017).
110. US National Library of Medicine. ClinicalTrials.gov
https://clinicaltrials.gov/ct2/show/NCT02653482
(2017).
111. US National Library of Medicine. ClinicalTrials.gov
https://clinicaltrials.gov/ct2/show/NCT03030222
(2017).
112. Nissen, S. E. & Wolski, K. Effect of rosiglitazone
on the risk of myocardial infarction and death
from cardiovascular causes. N. Engl. J. Med. 356,
2457–2471 (2007).
113. [No authors listed.] Guideline on clinical investigation
of medicinal products in the treatment or prevention
of diabetes mellitus. U.S. Food and Drug
Administration http://www.fda.gov/downloads/Drugs/
GuidanceComplianceRegulatoryInformation/
Guidances/ucm071627.pdf (2008).
114. [No authors listed.] Guideline on clinical
investigation of medicinal products in the treatment
or prevention of diabetes mellitus. European Medicines
Agency http://www.ema.europa.eu/docs/en_GB/
document_library/Scientific_guideline/2012/06/
WC500129256.pdf (2012).
115. US National Library of Medicine. ClinicalTrials.gov
https://clinicaltrials.gov/ct2/show/NCT01730534
(2017).
116. Pfeffer, M. A. et al. Lixisenatide in patients with type 2
diabetes and acute coronary syndrome. N. Engl.
J. Med. 373, 2247–2257 (2015).
Author contributions
Both authors researched data for the article, discussed the
content, wrote the manuscript, and reviewed and edited it
before submission.
Competing interests
M.K. declares that he is on the advisory boards for Amgen,
AstraZeneca, Boehringer Ingelheim, Eisai, Glytec, GSK,
Merck, Novo Nordisk, Sanofi, and ZS Pharma; is a consultant
for AstraZeneca, Sanofi, and ZS Pharma; and has received
research grants from AstraZeneca and Boehringer Ingelheim.
M.N. declares no competing interests.
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional
claims in published maps and institutional affiliations.

10 | ADVANCE ONLINE PUBLICATION www.nature.com/nrcardio

©
2
0
1
8
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.