Xiang 2022 Oi 220915 1663087733.39363

Original Investigation | Cardiology
Optimal Pharmacologic Treatment of Heart Failure

With Preserved and Mildly Reduced Ejection Fraction
A Meta-analysis
Boyang Xiang, BM; Ruiqi Zhang, BM; Xiaoguang Wu, BM; Xiang Zhou, PhD
Abstract Key Points

Question What is the optimal drug
IMPORTANCE In recent years, significant progress has been made in the pharmacologic treatment
combination for treatment of heart
of heart failure (HF) with reduced ejection fraction (HFrEF), but there is still insufficient evidence for
failure (HF) with preserved ejection
drug therapy for HF with preserved ejection fraction (HFpEF) and mildly reduced ejection
fraction (HFpEF) and mildly reduced
fraction (HFmrEF).
ejection fraction (HFmrEF)?
OBJECTIVE To compare the outcomes associated with different drug combinations for the Findings In this network meta-analysis,
treatment of HFpEF and HFmrEF. 19 randomized clinical trials enrolling
20 633 patients with HFpEF or HFmrEF
DATA SOURCES A search of the PubMed, Embase, and Cochrane Central Register of Controlled were included. Compared with placebo,
Trials (CENTRAL) databases was conducted for studies published from inception to October 9, 2021. no included drug classes were
associated with a reduced risk of death,
STUDY SELECTION Randomized clinical trials on the use of angiotensin-converting enzyme (ACE) but sodium-glucose cotransporter 2
inhibitors, angiotensin receptor blockers (ARBs), angiotensin receptor-neprilysin inhibitors (ARNIs), (SGLT2) inhibitors, angiotensin
mineralocorticoid receptor antagonists (MRAs), β-blockers, and sodium-glucose cotransporter 2 receptor-neprilysin inhibitors, and
(SGLT2) inhibitors for patients with HFpEF or HFmrEF. mineralocorticoid receptor antagonists
were associated with a significant
DATA EXTRACTION AND SYNTHESIS Data extraction and bias assessment were independently decrease in hospital admission for HF;
performed by 2 reviewers following the Preferred Reporting Items for Systematic Reviews and Meta- SGLT2 inhibitors were the optimal drug
analyses (PRISMA) guideline. All data for 3 outcomes were pooled with a fixed-effect model. class for decreasing the risk of admission
for HF.
MAIN OUTCOMES AND MEASURES The main outcomes were first hospitalization for HF, all-cause
Meaning The results of this study
mortality, and cardiovascular mortality. Hazard ratios (HRs) and 95% credible intervals (CrIs) were
suggest that the incremental use of
evaluated using a bayesian network meta-analysis model.
combinations of SGLT2 inhibitors,
angiotensin-converting enzyme
RESULTS In this analysis, 19 randomized clinical trials, including 20 633 patients with HF and an
inhibitors–angiotensin receptor
ejection fraction of 40% or more, without a remarkable risk of bias were included. Compared with
blockers, and β-blockers was associated
placebo, no treatments were associated with a significant reduction in the risk of all-cause death or
with accumulative benefits in HF
cardiovascular death. SGLT2 inhibitors, ARNIs, and MRAs were associated with a significant decrease
hospitalization rather than all-cause
in the risk of HF hospitalization compared with placebo (SGLT2 inhibitors: HR, 0.71 [95% CrI, 0.60-
death among patients with HFpEF
0.83]; ARNIs: HR, 0.76 [95% CrI, 0.61-0.95]; MRAs: HR, 0.83 [95% CrI, 0.69-0.99]), and SGLT2
and HFmrEF.
inhibitors were the optimal drug class in terms of reducing the risk for HF admission. Sensitivity
analysis results demonstrated a progressive decrease in the risk of HF admission and an advance in
mean rank associated with the increasing use of drug classes. + Supplemental content
Author affiliations and article information are
CONCLUSIONS AND RELEVANCE The findings of this study suggest that SGLT2 inhibitors were the listed at the end of this article.
optimal drug class for HFpEF and HFmrEF, consistent with the most recent guideline
recommendation. The incremental use of combinations of SGLT2 inhibitors, ACE inhibitors or ARBs,
(continued)
Open Access. This is an open access article distributed under the terms of the CC-BY License.
JAMA Network Open. 2022;5(9):e2231963. doi:10.1001/jamanetworkopen.2022.31963 (Reprinted) September 20, 2022 1/11
Downloaded from jamanetwork.com by guest on 01/17/2024

JAMA Network Open | Cardiology Treatment of Heart Failure With Preserved or Mildly Reduced Ejection Fraction
Abstract (continued)
and β-blockers may be associated with accumulative benefits in HF hospitalization rather than
all-cause death among patients with HFpEF and HFmrEF.
JAMA Network Open. 2022;5(9):e2231963. doi:10.1001/jamanetworkopen.2022.31963
Introduction
Approximately 64 million people worldwide have heart failure (HF), with considerable related
morbidity and mortality.1,2 Based on the left ventricular ejection fraction (LVEF), recent guidelines
classify HF into HF with reduced ejection fraction (HFrEF), HF with mildly reduced ejection fraction
(HFmrEF), and HF with preserved ejection fraction (HFpEF).3,4
For 30 years, pharmacologic therapy for HFrEF has continued to improve. Sodium-glucose
cotransporter 2 (SGLT2) inhibitors, mineralocorticoid receptor antagonists (MRAs), angiotensin
receptor blockers (ARBs), angiotensin-converting enzyme (ACE) inhibitors, β-blockers, and
angiotensin receptor-neprilysin inhibitors (ARNIs) have been recommended as disease-modifying
treatments for patients with HFrEF by US and European clinical guidelines.3,4 The incremental use of
combinations of these drugs has contributed to progressive benefits for patients with HFrEF.5
Nevertheless, the development of medical therapy for HF with higher ejection fraction has
failed to meet expectations. Many large-scale clinical trials on ARNIs, MRAs, ACE inhibitors, ARBs,
and β-blockers demonstrated neutral results in the primary composite outcome of death or HF
hospitalization for patients with HFpEF and HFmrEF.6 Still, these drug classes have been
recommended to treat HFpEF and HFmrEF by US guidelines because of their marginal benefits
regarding HF hospitalization.4 Recently, a large clinical trial found that, for patients with HFpEF and
HFmrEF, empagliflozin significantly improved the primary composite end point of HF hospitalization
or cardiovascular death compared with placebo, which was attributed to a decrease in HF
hospitalizations without a significant decrease in mortality.7 Therefore, SGLT2 inhibitors have been
recommended for patients with HFpEF or HFmrEF in preference to other drug classes in the latest US
guideline.4 To our knowledge, there have been no head-to-head studies comparing SGLT2 inhibitors
with other HF drug classes for this population.
Network meta-analysis is a reliable method to simultaneously compare multiple different
interventions without head-to-head studies. In this article, we used a bayesian network meta-
analysis to compare the outcomes associated with these HF treatment drugs for patients with HFpEF
or HFmrEF and to explore whether the benefits associated with these drugs for patients with HF with
higher ejection fraction were as accumulative as for patients with HFrEF.5
Methods
Search Strategies and Selection Criteria
This study was conducted following the Preferred Reporting Items for Systematic Reviews and Meta-
analyses (PRISMA) reporting guideline and extension statement for reporting of network
meta-analysis.8 We searched the PubMed, Embase, and Cochrane Central Register of Controlled
Trials (CENTRAL) databases from inception to October 9, 2021, using strategies presented in the
eAppendix in the Supplement. This analysis included randomized clinical trials (RCTs) with follow-up
of more than 3 months of adult patients with symptomatic HF and LVEF of 40% or more, assessing
any of the following interventions: ARNIs, MRAs, ACE inhibitors, ARBs, SGLT2 inhibitors, and
β-blockers.
An RCT was excluded if the whole trial population had the following characteristics that possibly
affected treatment response: (1) same sex, (2) comorbidity (eg, hypertension, diabetes, or atrial
fibrillation), (3) ischemic HF, (4) acute HF, or (5) myocardial infarction. Trials including only a

proportion of patients with these characteristics were included. Cardiac function is known to be
significantly associated with the outcomes of patients with HF; therefore, trials enrolling mostly
patients with severe cardiac function were also removed because of the homogeneity of the
evidence.
Study Selection and Data Extraction

Study selection, bias assessment, and data extraction were conducted independently by 2
researchers (B.X. and R.Z.), and divergences were resolved by discussion. For all eligible studies, data
on study design, population characteristics, interventions, and outcomes were extracted into a
database. The risk of bias for eligible studies was assessed across 7 domains.9
Outcome Measures
Outcomes estimated in this analysis included all-cause death, cardiovascular death, and first hospital
admission for HF during the entire follow-up period. These outcomes were reported as efficacy or
safety end points in the included studies.
Statistical Analysis
Based on the combination of direct and indirect study evidence, bayesian network meta-analysis can
indirectly compare the outcomes associated with different interventions when direct evidence is
absent. In the present study, previously described bayesian network meta-analysis models were
applied.10 These models combined binary and time-dependent data to ensure the integrity of the
data while preserving their time attributes. Time-dependent data were preferentially entered into
the model to reduce the loss of their time attributes. Unless a higher between-study heterogeneity
was found or the random-effect model was more parsimonious than the fixed-effect model, priority
was given to showing results from the fixed-effect model for easier interpretation of analysis results.
Noninformative prior distributions were used: normal distributions (mean, 0; variance, 106). We used
OpenBUGS, version 3.2.3 (University of Cambridge) to conduct network analysis with published
codes based on 200 000 iterations with 2 chains and a 100 000 burn-in period. The posterior
distributions of the model were presented as hazard ratios (HRs) and 95% credible intervals (95%
CrIs). The probability (P score) that the therapy was associated with better outcomes than the
comparator was presented after the transformation of the estimated points and their 95% CrIs.
Heterogeneity was assessed with the I2 statistic, and inconsistency was assessed with the node-split
model using R, version 4.1.1 (R Group for Statistical Computing).11 If there were closed loops in the
network plot, the P value of inconsistency could be calculated; otherwise, the P value was not
available. A 2-sided P < .05 was considered statistically significant. Ranking probabilities were
presented in the form of a rankogram plot.
To weaken the interference of concomitant drugs with analysis results, a prespecified sensitivity
analysis was performed. If more than 50% of the patients in a trial were using the drug of interest at
baseline, intervention classes were regarded as including the concomitant drug. The threshold
defining the concomitant therapy referred to published articles.5,12,13 When more than 50% of
patients in a trial received concomitant drugs, the intervention class was classified as a combined
treatment (primary trial drug class plus concomitant drug class). If necessary, additional sensitivity
analysis was performed.
Results
Study Selection and Characteristics
In this network analysis, 11 855 publications were identified, and 19 RCTs, including a total of 20 633
patients with HF and an ejection fraction of 40% or more, were included after screening (eFigure 1 in
the Supplement); 19 trials reported all-cause death, 16 trials reported cardiovascular death, and 12
trials reported hospital admission for HF. Network diagrams of direct evidence based on RCTs are

presented in Figure 1. Multiple direct connections among the intervention classes comprised 52 761
patient-years for death from any cause, 52 090 patient-years for death from cardiovascular causes,
and 50 524 patient-years for HF hospitalization.
Most of the included studies were double-blind, multicenter RCTs, and nearly all of the included
RCTs had no significant risk of bias (eFigure 2 in the Supplement). The median trial duration ranged
from 4 to 40 months, with the median duration of only 1 study being less than 6 months. The sample
sizes of the included trials ranged from 40 to 5988 patients, with 6 trials enrolling fewer than 100
patients and 6 trials enrolling more than 500 patients.
Figure 2 shows the population characteristics at baseline stratified by intervention classes,
demonstrating the transitivity of our network meta-analysis. The age, sex, obesity degree,
hemodynamics, and LVEF of the populations were deemed similar among the 7 different
intervention classes (6 medication classes plus placebo). The cardiac function of the eligible
population was primarily concentrated in New York Heart Association Class II and III. More detailed
information about study characteristics is presented in eTable 1 in the Supplement. The background
therapies of the included patients are presented in eTable 2 in the Supplement. The numbers of
patients taking ARBs and ACE inhibitors at baseline were not reported separately in many studies;
thus, the 2 classes of drugs together were classified as angiotensin system inhibitors (ASIs) in the
sensitivity analysis.
Primary Network Meta-analysis

Based on the combination of direct and indirect evidence from the included RCTs, the comparative
effectiveness of 7 intervention classes for 3 outcomes were assessed with a fixed-effect model, and
the results of the primary analysis are presented in Figure 3 and eFigure 3 in the Supplement. No
significant heterogeneity or inconsistency for the 3 outcomes was found. As shown in Figure 3, 6
treatments were not associated with benefits with respect to the risk of all-cause or cardiovascular
death compared with placebo. However, SGLT2 inhibitors, ARNIs, and MRAs were significantly
associated with a decreased risk of hospital admission for HF compared with placebo (SGLT2
inhibitors: HR, 0.71 [95% CrI, 0.60-0.83]; ARNIs: HR, 0.76 [95% CrI, 0.61-0.95]; MRAs: HR, 0.83
[95% CrI, 0.69-0.99]). In eFigure 4 in the Supplement, the ranking probabilities for all interventions
Figure 1. Network Plots of Direct Evidence
All-cause death
Cardiovascular death
A Primary analysis B Sensitivity analysis HF hospitalization
ACE
inhibitor β-Blocker ASI β-Blocker
ASI + ASI +
ARB Placebo MRA Placebo β-blocker
β-blocker
+ MRA
ASI +
SGLT2 ARNI + β-blocker
ARNI
inhibitor β-blocker + SGLT2
inhibitor
The width of the connecting line was positively correlated with the patient-years of blocker; ARNI, angiotensin receptor-neprilysin inhibitor; ASI, angiotensin system
direct evidence. Different colors of connecting lines corresponded to different inhibitor (ACE inhibitor or ARB); HF, heart failure; MRA, mineralocorticoid receptor
outcomes. ACE indicates angiotensin-converting enzyme; ARB, angiotensin receptor antagonist; and SGLT2, sodium-glucose cotransporter 2.

for 3 outcomes are demonstrated. The rankogram plot for HF hospitalization showed that SGLT2
inhibitors were most likely the optimal treatment, followed by ARNIs and MRAs, which was
consistent with the rank of estimated points.
Figure 2. Study Population Characteristics at Baseline
A Mean age B Mean BMI
SGLT2 inhibitor SGLT2 inhibitor
ARNI ARNI
MRA MRA
Intervention
Intervention
β-Blocker β-Blocker
ARB ARB
ACE inhibitor ACE inhibitor
Placebo Placebo
0 20 40 60 80 0 10 20 30 40
Mean age, y Mean BMI
C Mean LVEF D Mean heart rate
ARNI ARNI
MRA MRA
Intervention
Intervention
ARB ARB
Placebo Placebo
0 20 40 60 80 0 20 40 60 80
Mean LVEF, % Mean HR, beats/min
E NYHA class F Sex
NYHA I NYHA II NYHA III NYHA IV Male Female
ARNI ARNI
MRA MRA
Intervention
Intervention
ARB ARB
Placebo Placebo
0 0.2 0.4 0.6 0.8 1.0 0 0.2 0.4 0.6 0.8 1.0
NYHA class Sex
Population characteristics at baseline were stratified by 7 intervention classes (6 (calculated as weight in kilograms divided by height in meters squared); HR, heart rate;
medication classes plus placebo) and were overall deemed to be similar among the LVEF, left ventricular ejection fraction; MRA, mineralocorticoid receptor antagonist;
intervention classes. ACE indicates angiotensin-converting enzyme; ARB, angiotensin NYHA, New York Heart Association; and SGLT2, sodium-glucose cotransporter 2.
receptor blocker; ARNI, angiotensin receptor-neprilysin inhibitor; BMI, body mass index

Sensitivity Analysis
Owing to the poor fit of the model for cardiovascular death, which was possibly associated with the
relatively sparse incidence of cardiovascular death and the relatively uneven distribution of direct
evidence among each connection, we evaluated the association of 7 intervention classes only with
death from any cause and hospitalization for HF with a fixed-effect model. The results of the
sensitivity analysis are summarized in Figure 4 and eFigure 5 in the Supplement. No significant
heterogeneity was found. In Figure 4, none of the 6 interventions was associated with a significant
Figure 3. Effect Sizes of Interventions vs Placebo for 3 Outcomes in the Primary Network Meta-analysis
Favors Does not favor

Intervention class HR (95% CrI) intervention intervention P score
All-cause death
SGLT2 inhibitor 1.00 (0.87-1.15) 0.03
ARNI 1.00 (0.79-1.25) 0.01
MRA 0.92 (0.77-1.08) 0.68
β-Blocker 0.95 (0.67-1.32) 0.20
ARB 1.03 (0.86-1.23) 0.23
ACE inhibitor 1.06 (0.73-1.50) 0.24
Placebo 1 [Reference] 0
Cardiovascular death
SGLT2 inhibitor 0.91 (0.76-1.09) 0.67
ARNI 0.92 (0.68-1.21) 0.43
MRA 0.91 (0.73-1.12) 0.63
β-Blocker 0.89 (0.55-1.35) 0.39
ARB 0.96 (0.77-1.19) 0.28
ACE inhibitor 0.97 (0.61-1.48) 0.08
Hospitalization for HF
SGLT2 inhibitor 0.71 (0.60-0.83) 1.00
A hazard ratio (HR) less than 1 favors the intervention,
ARNI 0.76 (0.61-0.95) 0.98
and a 95% credible interval (CrI) not including 1
MRA 0.83 (0.69-0.99) 0.95
signifies statistical significance. A larger P score
β-Blocker 0.82 (0.44-1.40) 0.48
suggests that an intervention is associated with a
ARB 0.84 (0.71-1.00) 0.94
better outcome. ACE indicates angiotensin-converting
ACE inhibitor 0.85 (0.61-1.15) 0.69
enzyme; ARB, angiotensin receptor blocker; ARNI,
angiotensin receptor-neprilysin inhibitor; HF, heart
0 0.5 1.0 1.5 2.0 failure; MRA, mineralocorticoid receptor antagonist;
HR (95% CrI) and SGLT2, sodium-glucose cotransporter 2.
Figure 4. Effect Sizes of Interventions vs Placebo for 2 Outcomes in the Sensitivity Analysis
Favors Does not favor

Intervention class HR (95% CrI) intervention intervention P score
All-cause death
ASI + β-blocker + SGLT2 inhibitor 0.31 (0.03-1.11) 0.78
ASI + β-blocker + MRA 0.28 (0.02-1.02) 0.81
ARNI + β-blocker 0.30 (0.03-1.07) 0.79
ASI + β-blocker 0.31 (0.03-1.10) 0.78
β-Blocker 0.30 (0.03-1.08) 0.79
ASI 0.33 (0.03-1.16) 0.76
Hospitalization for HF
ASI + β-blocker + SGLT2 inhibitor 0.46 (0.15-1.12) 0.87
ASI + β-blocker + MRA 0.53 (0.17-1.31) 0.77 A hazard ratio (HR) less than 1 favors the intervention,
ARNI + β-blocker 0.58 (0.19-1.41) 0.71 and a 95% credible interval (CrI) not including 1
ASI + β-blocker 0.64 (0.21-1.55) 0.61 signifies statistical significance. An intervention
β-Blocker 0.76 (0.25-1.85) 0.39 associated with a better outcome has a larger P score.
ASI 0.80 (0.32-1.70) 0.39 ARNI indicates angiotensin receptor-neprilysin
Placebo 1 [Reference] 0 inhibitor; ASI, angiotensin system inhibitor; HF, heart
0 0.5 1.0 1.5 2.0 failure; MRA, mineralocorticoid receptor antagonist;
HR (95% CrI) and SGLT2, sodium-glucose cotransporter 2.

reduction in all-cause death or HF hospitalization compared with placebo, but the probability that
the therapy was associated with greater decrease in HF hospitalization than placebo increased
gradually with the incremental use of drug classes. The point estimates and mean ranks of
interventions for all-cause death and HF hospitalization are summarized in Figure 5, illustrating a
progressive decrease in the risk of HF admission and an advance in the mean ranks of interventions
with the increasing use of drug classes. This trend was not observed for all-cause mortality. As shown
in eFigure 5 in the Supplement, ASIs plus β-blockers plus SGLT2 inhibitors were associated with a
significant reduction in the risk of HF hospitalization compared with ASIs (HR, 0.57 [95% CrI,
0.30-0.98]), β-blockers (HR, 0.60 [95% CrI, 0.48-0.75]), ASIs plus β-blockers (HR, 0.71 [95% CrI,
0.60-0.83]), and ARNIs plus β-blockers (HR, 0.79 [95% CrI, 0.64-0.97]). In addition, ASIs plus
β-blockers (HR, 0.85 [95% CrI, 0.72-0.99]), ARNIs plus β-blockers (HR, 0.77 [95% CrI, 0.62-0.94]),
and ASIs plus β-blockers plus MRAs (HR, 0.71 [95% CrI, 0.55-0.89]) were associated with a
decreased risk of HF admission compared with β-blockers. Moreover, ASIs plus β-blockers plus MRAs
were associated with a significant reduction in HF hospitalization compared with ASIs plus β-blockers
(HR, 0.83 [95% CrI, 0.69-0.99]). The rankogram plot for HF hospitalization showed that ASIs plus
β-blockers plus SGLT2 inhibitors was most likely the optimal therapy, followed by ASIs plus β-blockers
plus MRAs and ARNIs plus β-blockers (eFigure 6 in the Supplement). The results of the sensitivity
analysis did not conflict with the results of the primary analysis.
Our primary analysis demonstrated positive results only in terms of HF hospitalization. Our
network meta-analysis included 6 studies enrolling fewer than 100 patients, of which 4 studies
reported hospital admission for HF. Additional sensitivity analysis was performed to ensure the
stability of the results of the primary analysis for HF hospitalization. The results of the network meta-
analysis for HF hospitalization after removing certain studies with small sample sizes are presented
in eTable 3 in the Supplement, illustrating that these studies with small sample sizes had little impact
on our results and that our results were relatively stable.
Discussion
The results of the primary analysis demonstrated that SGLT2 inhibitors, ARNIs, and MRAs were
associated with a significant reduction in the risk of HF admission compared with placebo for patients
with HFpEF or HFmrEF and that SGLT2 inhibitors were most likely the optimal drug class, in line with
the latest guideline. The results of the sensitivity analysis showed that the incremental use of the
Figure 5. Effect Sizes of Interventions in All-Cause Death and Hospitalization for Heart Failure (HF)
1.5
All-cause death
HF hospitalization 6
1.0
Hazard ratio
Mean rank
The point estimates of interventions compared with

0.5
placebo in all-cause death and HF hospitalization are
2
summarized with bars. The mean ranks estimated
based on the ranking probabilities are summarized
with lines. An intervention associated with a better
outcome has a smaller mean rank. ARNI indicates
0 0
Placebo ASI β-Blocker ASI + ARNI + ASI + ASI + angiotensin receptor-neprilysin inhibitor; ASI,
β-blocker β-blocker β-blocker β-blocker angiotensin system inhibitor; MRA, mineralocorticoid
+ MRA + SGLT2
inhibitor receptor antagonist; and SGLT2, sodium-glucose
Intervention class cotransporter 2.

included drug classes was probably associated with a progressive decrease in the risk of HF
hospitalization among patients with HF and an LVEF of 40% or more.
A previous network meta-analysis demonstrated that, for patients with HF and an LVEF of 40%
or more, ARNIs and ACE inhibitors were associated with a significant decrease in the risk of hospital
admission for HF, but MRAs were not.14 These results are partly inconsistent with the results of our
analysis. In the previous analysis, a positive result of an ACE inhibitor trial during the first year, rather
than a negative result during the entire follow-up, was extracted,15 and a result of cardiac
hospitalization in an MRA trial was extracted as a result of HF hospitalization.16 These differences
may have led to the inconsistencies between our analysis and the previous analysis.
In our sensitivity analysis, we focused on the trend of estimated points rather than the
estimated points themselves. Although 6 treatments were not significantly associated with a
reduction in the risk of hospital admission for HF compared with placebo, which was likely owing to a
weak connection between ASIs and placebo, the downward trend of the estimated points for the 6
treatments vs placebo suggested that the increasing use of drug classes was probably associated
with the incremental benefits associated with HF hospitalization. Moreover, the probability that the
therapy was associated with a greater decrease in HF hospitalization compared with placebo
increased gradually with the increasing use of drug classes, reinforcing the viewpoint that the
incremental benefits associated with treatment were not accidental. However, this downward trend
was not observed in terms of all-cause mortality. The 2021 European Society of Cardiology HF
guidelines proposed the hypothesis that the lack of disease-modifying effects of drugs among
patients with HFpEF was probably caused by the interference of concomitant drugs.3 For practical
reasons, a trial confirming this hypothesis was difficult to implement, but the results of the sensitivity
analysis to some degree explain this hypothesis (namely, the concomitant drug probably interfered
with the disease-modifying effects of drugs in terms of HF hospitalization rather than
all-cause death).
Our results showed the incremental benefits associated with drug therapies for patients with
HF and an LVEF of 40% or more, which did not mean that the benefits exist for all patients with
HFmrEF or HFpEF. Although patients with HFmrEF have outcomes more akin to patients with HFpEF
than patients with HFrEF, the response to medical therapies among patients with HFmrEF appears
to be more similar to that among patients with HFrEF.3,4 Thus, for patients with HF, the classification
of LVEF based on patient prognosis may be inconsistent with that based on treatment response, and
the patients with a relatively lower LVEF who were included in our study may have received more
benefits associated with drug therapies. In 2 large-scale RCTs of HFpEF and HFmrEF, we found that
the benefits associated with ARNIs and MRAs emerged primarily for patients with a baseline LVEF
less than or equal to the median (57% for ARNIs; 56% for MRAs), but these benefits were weakened
or even disappeared for patients with a baseline LVEF greater than the median.17,18 A similar situation
was found for SGLT2 inhibitors.7 Therefore, the LVEF cutoff to classify whether a drug is effective or
not for patients with HFpEF or HFmrEF may be different from the LVEF cutoff for dividing HF into
HFmrEF and HFpEF, and the use of specific drugs should refer to the subgroup analysis of the LVEF in
their major trials. The clinical syndrome of HFpEF is heterogeneous, which probably led to the neutral
results in many clinical trials. Relative to HFrEF, the etiologic factors associated with HFpEF are more
diverse and carry their own distinct pathophysiology and natural history. For the same drug
treatment, people with different etiologic factors may have different treatment responses.
Phenotypic classification of HFpEF is needed owing to the substantial heterogeneity, whereas there
is currently no effective or uniform classification method.6 More studies exploring the phenotypic
classification of HFpEF are still required to identify the patients who do not respond to certain drugs.
According to the baseline characteristics of the included trials, patients with HFpEF or HFmrEF,
most of whom had taken ASIs and/or β-blockers before enrollment, tended to be older and to have
obesity and many comorbid conditions (eg, hypertension, atrial fibrillation, and diabetes). Our
primary analysis demonstrated that ACE inhibitors and ARBs were not significantly associated with a
reduction in the risk of HF hospitalization for patients with HFpEF or HFmrEF. Nevertheless, eTable 2

in the Supplement showed that in 2 large included trials on ACE inhibitors and ARBs, most of the
enrolled patients had taken β-blockers at baseline; thus, β-blockers were likely to interfere with the
effectiveness of ACE inhibitors and ARBs. In the sensitivity analysis, after weakening the interference
of concomitant β-blockers, ACE inhibitors and ARBs were associated with a significant decrease in
the risk of HF hospitalization for patients with HFpEF or HFmrEF, suggesting that β-blockers were
probably an interferent. We found that ASIs plus β-blockers plus SGLT2 inhibitors vs ASIs were
associated with greater effectiveness than ASIs plus β-blockers plus SGLT2 inhibitors vs ASIs plus
β-blockers in terms of HF hospitalization, suggesting that β-blockers may be a class of beneficial
interferent and that the addition of β-blockers probably led to an incremental benefit for patients
with HFpEF or HFmrEF.
Because of the weak disease-modifying effectiveness of pharmacologic therapy for patients
with HF and an LVEF of 40% or more, high expectations were placed on the effectiveness of surgical
treatment previously. Studies demonstrated that, for patients with HF and an LVEF of 40% or more,
atrial septostomy was associated with improved quality of life and cardiac function.19,20
Nevertheless, a recent clinical trial suggested that atrial septostomy did not significantly improve the
composite outcome or individual outcome of cardiovascular death and HF events.21 In the absence
of disease-modifying effects of surgical treatment, proper use of disease-modifying drugs is
important for the management of HFpEF and HFmrEF. In clinical practice, most patients have taken
ACE inhibitors or ARBs and β-blockers to treat comorbid conditions. For these patients with HF and
an LVEF of 40% or more, ARNIs may be considered to replace ACE inhibitors or ARBs if tolerated,
and SGLT2 inhibitors and MRAs may be recommended to be added when indicated.
Limitations
This study has some limitations. Owing to the internal selection bias of RCTs, their results cannot be
applied to all patients with HF in clinical settings. Our analysis did not consider the doses of the
included drugs, which may have had a small effect on our results. Owing to the absence of the data
on subgroups of patients with HFmrEF or HFpEF in most of the included trials, we did not perform a
subgroup analysis to estimate the outcomes associated with drug therapies for HFmrEF and HFpEF
separately. Owing to the low number of eligible studies, other subgroup analyses were difficult to
perform. Thus, we did not further explore the treatment response of patients with different
phenotypes for interventions and potential heterogeneity. Because the proportion of patients taking
concomitant drugs was less than 100% and may have changed slightly during the follow-up period
in our sensitivity analysis, the estimated point was relatively inaccurate and served merely as a
reference. The regimen containing ARNIs, MRAs, and SGLT2 inhibitors was probably the most
effective therapy, but no trial on this treatment regimen was included. Future studies of SGLT2
inhibitors that include more patients with HFpEF or HFmrEF using MRAs and ARNIs are needed to
explore this treatment combination.
Conclusions
This network meta-analysis found that, for patients with HFpEF or HFmrEF, no included drug classes
were significantly associated with a reduced risk of death, but SGLT2 inhibitors, ARNIs, and MRAs
were associated with a significant decrease in the risk of HF admission; SGLT2 inhibitors were the
optimal drug class. Our results are consistent with the latest guideline recommendations. If
indicated, SGLT2 inhibitors may be preferentially recommended for patients with HFpEF or HFmrEF.
The increasing use of combinations of drug therapies may be associated with accumulative benefits
in terms of HF hospitalization rather than all-cause death for patients with HF and an LVEF of 40% or
more. More studies that explore the phenotypic classification and the LVEF cutoff for the efficacy of
drug therapies for patients with HFpEF or HFmrEF are needed.

ARTICLE INFORMATION
Accepted for Publication: July 30, 2022.
Published: September 20, 2022. doi:10.1001/jamanetworkopen.2022.31963
Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2022 Xiang B
et al. JAMA Network Open.
Corresponding Author: Xiang Zhou, PhD, Department of Cardiology, The Second Affiliated Hospital of Soochow
University, No. 1055 Sanxiang Road, Suzhou 215004, China (zhou-xiang@suda.edu.cn).
Author Affiliations: Department of Cardiology, The Second Affiliated Hospital of Soochow University, Suzhou,
China.
Author Contributions: Drs Xiang and Zhou had full access to all of the data in the study and takes responsibility for
the integrity of the data and the accuracy of the data analysis.
Concept and design: Zhou.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Xiang, Zhang, Wu.
Critical revision of the manuscript for important intellectual content: Zhou.
Statistical analysis: All authors.
Obtained funding: Zhou.
Supervision: Zhou.
Conflict of Interest Disclosures: None reported.
REFERENCES
1. GBD 2016 Disease and Injury Incidence and Prevalence Collaborators. Global, regional, and national incidence,
prevalence, and years lived with disability for 328 diseases and injuries for 195 countries, 1990-2016: a systematic
analysis for the Global Burden of Disease Study 2016. Lancet. 2017;390(10100):1211-1259. doi:10.1016/S0140-
6736(17)32154-2
2. Mozaffarian D, Benjamin EJ, Go AS, et al; Writing Group Members; American Heart Association Statistics
Committee; Stroke Statistics Subcommittee. Heart disease and stroke statistics—2016 update: a report from the
American Heart Association. Circulation. 2016;133(4):e38-e360. doi:10.1161/CIR.0000000000000350
3. McDonagh TA, Metra M, Adamo M, et al; ESC Scientific Document Group. 2021 ESC guidelines for the diagnosis
and treatment of acute and chronic heart failure. Eur Heart J. 2021;42(36):3599-3726. doi:10.1093/eurheartj/
ehab368
4. Heidenreich PA, Bozkurt B, Aguilar D, et al. 2022 AHA/ACC/HFSA guideline for the management of heart failure:
a report of the American College of Cardiology/American Heart Association Joint Committee on Clinical Practice
Guidelines. J Am Coll Cardiol. 2022;79(17):e263-e421. doi:10.1016/j.jacc.2021.12.012
5. Xiang B, Yu Z, Zhou X. Comparative efficacy of medical treatments for chronic heart failure: a network meta-
analysis. Front Cardiovasc Med. 2022;8:787810. doi:10.3389/fcvm.2021.787810
6. Borlaug BA. Evaluation and management of heart failure with preserved ejection fraction. Nat Rev Cardiol.
2020;17(9):559-573. doi:10.1038/s41569-020-0363-2
7. Anker SD, Butler J, Filippatos G, et al; EMPEROR-Preserved Trial Investigators. Empagliflozin in heart failure with
a preserved ejection fraction. N Engl J Med. 2021;385(16):1451-1461. doi:10.1056/NEJMoa2107038
8. Hutton B, Salanti G, Caldwell DM, et al. The PRISMA extension statement for reporting of systematic reviews
incorporating network meta-analyses of health care interventions: checklist and explanations. Ann Intern Med.
2015;162(11):777-784. doi:10.7326/M14-2385
9. Higgins JP, Thomas J, Chandler J, Cumpston M, Li T. Cochrane Handbook for Systematic Reviews of
Interventions. John Wiley & Sons; 2019. doi:10.1002/9781119536604
10. Woods BS, Hawkins N, Scott DA. Network meta-analysis on the log-hazard scale, combining count and hazard
ratio statistics accounting for multi-arm trials: a tutorial. BMC Med Res Methodol. 2010;10:54. doi:10.1186/1471-
2288-10-54
11. Shim SR, Kim SJ, Lee J, Rücker G. Network meta-analysis: application and practice using R software. Epidemiol
Health. 2019;41:e2019013. doi:10.4178/epih.e2019013
12. Burnett H, Earley A, Voors AA, et al. Thirty years of evidence on the efficacy of drug treatments for chronic
heart failure with reduced ejection fraction: a network meta-analysis. Circ Heart Fail. 2017;10(1):e003529. doi:10.
1161/CIRCHEARTFAILURE.116.003529

13. Burnett H, Cope S, Vieira MC, Sagkriotis A, Senni M, Deschaseaux C. The importance of treatment
classifications that account for concomitant treatments in the context of a network meta-analysis comparing
pharmacological treatments for chronic heart failure. Value Health. 2014;17(7):A327. doi:10.1016/j.jval.2014.
08.595
14. Lin Y, Wu M, Liao B, et al. Comparison of pharmacological treatment effects on long-time outcomes in heart
failure with preserved ejection fraction: a network meta-analysis of randomized controlled trials. Front Pharmacol.
2021;12:707777. doi:10.3389/fphar.2021.707777
15. Cleland JG, Tendera M, Adamus J, Freemantle N, Polonski L, Taylor J; PEP-CHF Investigators. The Perindopril
in Elderly People With Chronic Heart Failure (PEP-CHF) study. Eur Heart J. 2006;27(19):2338-2345. doi:10.
1093/eurheartj/ehl250
16. Edelmann F, Wachter R, Schmidt AG, et al; Aldo-DHF Investigators. Effect of spironolactone on diastolic
function and exercise capacity in patients with heart failure with preserved ejection fraction: the Aldo-DHF
randomized controlled trial. JAMA. 2013;309(8):781-791. doi:10.1001/jama.2013.905
17. Solomon SD, McMurray JJV, Anand IS, et al; PARAGON-HF Investigators and Committees. Angiotensin-
neprilysin inhibition in heart failure with preserved ejection fraction. N Engl J Med. 2019;381(17):1609-1620. doi:
10.1056/NEJMoa1908655
18. Pitt B, Pfeffer MA, Assmann SF, et al; TOPCAT Investigators. Spironolactone for heart failure with preserved
ejection fraction. N Engl J Med. 2014;370(15):1383-1392. doi:10.1056/NEJMoa1313731
19. Hasenfuß G, Hayward C, Burkhoff D, et al; REDUCE LAP-HF study investigators. A transcatheter intracardiac
shunt device for heart failure with preserved ejection fraction (REDUCE LAP-HF): a multicentre, open-label, single-
arm, phase 1 trial. Lancet. 2016;387(10025):1298-1304. doi:10.1016/S0140-6736(16)00704-2
20. Shah SJ, Feldman T, Ricciardi MJ, et al. One-year safety and clinical outcomes of a transcatheter interatrial
shunt device for the treatment of heart failure with preserved ejection fraction in the Reduce Elevated Left Atrial
Pressure in Patients With Heart Failure (REDUCE LAP-HF I) Trial: a randomized clinical trial. JAMA Cardiol. 2018;
3(10):968-977. doi:10.1001/jamacardio.2018.2936
21. Shah SJ, Borlaug BA, Chung ES, et al; REDUCE LAP-HF II investigators. Atrial shunt device for heart failure with
preserved and mildly reduced ejection fraction (REDUCE LAP-HF II): a randomised, multicentre, blinded, sham-
controlled trial. Lancet. 2022;399(10330):1130-1140. doi:10.1016/S0140-6736(22)00016-2
SUPPLEMENT.
eAppendix. Search Strategies
eFigure 1. Flowchart of Study Selection Process
eFigure 2. Bias Assessment of Eligible Studies
eFigure 3. Primary Network Meta-analysis Results for All-Cause Death (A), Cardiovascular Death (B) and
Hospitalization for Heart Failure (C)
eFigure 4. Rankogram Plot for All-Cause Mortality (A), Cardiovascular Mortality (B), and Hospitalization for Heart
Failure (C) in the Primary Analysis
eFigure 5. Sensitivity Analysis Results for All-Cause Death (A) and Hospitalization for Heart Failure (B)
eFigure 6. Rankogram Plot for All-Cause Mortality (A) and Hospitalization for Heart Failure (B) in Sensitivity
Analysis
eTable 1. Study Characteristics of Included RCTs
eTable 2. Baseline Characteristics of Included Intervention and Concomitant Drug Classes Reported
eTable 3. Additional Sensitivity Analysis
eReferences

Xiang 2022 Oi 220915 1663087733.39363

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Xiang 2022 Oi 220915 1663087733.39363

Uploaded by

Copyright:

Available Formats

Original Investigation | Cardiology

Optimal Pharmacologic Treatment of Heart Failure

Abstract Key Points

Downloaded from jamanetwork.com by guest on 01/17/2024

JAMA Network Open. 2022;5(9):e2231963. doi:10.1001/jamanetworkopen.2022.31963

Downloaded from jamanetwork.com by guest on 01/17/2024

Study Selection and Data Extraction

Downloaded from jamanetwork.com by guest on 01/17/2024

Primary Network Meta-analysis

Figure 1. Network Plots of Direct Evidence

Downloaded from jamanetwork.com by guest on 01/17/2024

Figure 2. Study Population Characteristics at Baseline

A Mean age B Mean BMI

SGLT2 inhibitor SGLT2 inhibitor

ACE inhibitor ACE inhibitor

C Mean LVEF D Mean heart rate

SGLT2 inhibitor SGLT2 inhibitor

ACE inhibitor ACE inhibitor

E NYHA class F Sex

NYHA I NYHA II NYHA III NYHA IV Male Female

SGLT2 inhibitor SGLT2 inhibitor

ACE inhibitor ACE inhibitor

Downloaded from jamanetwork.com by guest on 01/17/2024

Favors Does not favor

Favors Does not favor

Downloaded from jamanetwork.com by guest on 01/17/2024

The point estimates of interventions compared with

Downloaded from jamanetwork.com by guest on 01/17/2024

Downloaded from jamanetwork.com by guest on 01/17/2024

Downloaded from jamanetwork.com by guest on 01/17/2024

Downloaded from jamanetwork.com by guest on 01/17/2024

Downloaded from jamanetwork.com by guest on 01/17/2024

You might also like