Articles

SGLT2 inhibitors in patients with heart failure: @r®Cromvtrk

a comprehensive meta-analysis of five randomised
controlled trials
Muthiah Vaduganathan®, Kieran F Docherty”, Brian L Claggett, Pardeep S Jhund, Rudolf A de Boer, Adrian F Hernandez Silvio E Inzucchi,
Mikhail N Kosiborod, Carolyn S P Lam, Felipe Martinez, Sanjiv Shah, Akshay S Desai,john J V McMurrayt, Scott D Solomon t
Summary
Background SGLT2 inhibitors are strongly recommended in guidelines to treat patients with heart failure with reduced Lancet 2022; 400: 757-67
ejection fraction, but their clinical benefits at higher ejection fractions are less well established. Two large-scale trials, Published Online
DELIVER and EMPEROR-Preserved, in heart failure with mildly reduced or preserved ejection fraction have been August 27,2022
done, providing power to examine therapeutic effects on cardiovascular mortality and in patient subgroups when hitps:/doi org/10.1016/
50140-6736(22)01429-5
combined with the earlier trials in reduced ejection fraction.
This online publication has been
corrected. The corrected version
Methods We did a prespecified meta-analysis of DELIVER and EMPEROR-Preserved, and subsequently included trials frst appeared at thelancet.com
that enrolled patients with reduced ejection fraction (DAPA-HF and EMPEROR-Reduced) and those admitted to hospital onjanvary 12,2023
with worsening heart failure, irrespective of ejection fraction (SOLOIST-WHF). Using trial-level data with harmonised See Comment page 711
endpoint definitions, we did a fixed-effects meta-analysis to estimate the effect of SGLT2 inhibitors on various clinical *Joint first authors
endpoints in heart failure The primary endpoint for this meta-analysis was time from randomisation to the occurrence Hoint senior authors.
of the composite of cardiovascular death or hospitalisation for heart failure. We assessed heterogeneity in treatment Cardiovascular Division,
effects for the primary endpoint across subgroups of interest. This study is registered with PROSPERO, CRD42022327527. Brigham and Women's
Hospital, Harvard Medical
School, Boston, MA, USA
Findings Among 12251 participants from DELIVER and EMPEROR-Preserved, SGLT2 inhibitors reduced composite (M Vaduganathan MD,
cardiovascular death or first hospitalisation for heart failure (hazard ratio 080 [95% CI 0-73-0-87]) with consistent B L Claggett PhD, A S Desai MD,
reductions in both components: cardiovascular death (0-88 [0-77-1-00]) and first hospitalisation for heart failure Prof $ D Solomon MD);British
(0-74[0-67-0-83]). In the broader context of the five trials of 21947 participants, SGLT2 inhibitors reduced the risk of Heart Foundation, University
of Glasgow, Glasgow, UK
composite cardiovascular death or hospitalisation for heart failure (0-77 [0-72-0.82]), cardiovascular death (0-87 (KF DochertyPhD,
[0-79-0-95]), first hospitalisation for heart failure (072 [0-67-0-78]), and all-cause mortality (0-92 [0-86-0-99]). profPs jhund Ph,
These treatment effects for each of the studied endpoints were consistently observed in both the trials of heart failure Prof))V Mchurray MD);
with mildly reduced or preserved ejection fraction and across all five trials. Treatment effects on the primary endpoint University of Groningen,
Groningen, Netherlands
were generally consistent across the 14 subgroups examined, including ejection fraction. (Prof R Ade Boer MD); Duke
University Medical Center,
Interpretation SGLT2 inhibitors reduced the risk of cardiovascular death and hospitalisations for heart failure in a Durham, NG, USA
broad range of patients with heart failure, supporting their role as foundational therapy for heart failure, irrespective (Prof AF HerandezMD,
Prof €S P Lam MBBS);
of ejection fraction or care setting. Yale School of Medicine,
New Haven, CT, USA
Funding None. (PSE lnzucchi MD); Saint Luke's
Mid America Heart Institute,
Kansas City and University of
Copyright © 2022 Published by Elsevier Ltd. All rights reserved. Missouri-Kansas City, MO, USA
(ProfM Kosiborod MD);
Introduction recommendations for SGLT2 inhibitors in heart failure National Heart Centre
SGLT2 inhibitors have been shown to have salutary with mildly reduced and preserved ejection fraction remain Singapore and Duke-National
University of Singapore,
cardioprotective and renoprotective effects in various either absent, because of timing of publications, or weaker Singapore (Prof CS P Lam);
diseases including type 2 diabetes, chronic kidney (class 1) than recommendations for these same therapies Universidad Nacional de
disease, and heart failure. In patients with heart failure, in heart failure with reduced ejection fraction (class 1).** Cérdoba, Cordoba, Argentina
the clinical benefits of SGIT2 inhibitors were first This difference might partly be due to uncertainty around (ProfF Martinez MD);
Northwestern University
established in those with reduced ejection fraction and the consistency of clinical benefits across the class and Feinberg School of Medicine,
are now strongly recommended as a key component of therapeutic effects on individual endpoints that neither Chicago, L, USA
comprehensive disease management. More recently, trial was specifically designed or powered to examine, (Prof Shah MD)
the EMPEROR-Preserved’ and DELIVER trials* showed particularly cardiovascular death. Similarly, whether the Correspondence to
reductions in composite cardiovascular death or heart Prof Scott D Solomon,
clinical benefits of SGLT2 inhibitors in heart failure extend Cardiovascular Medicine,
failure events in patients with heart failure with mildly to all subpopulations including those at the highest end of Brigham and Women's Hospital,
reduced or preserved ejection fraction. the ejection fraction spectrum’ and those already treated Harvard Medical School, Boston,
Clinical practice guidelines were updated after with other therapies commonly used in heart failure® has MA 02115, USA
EMPEROR-Preserved was published, but not been well established.
ssolomon@bwh harvard.edu

www thelancet.com Vol 400 September 3, 2022 757

 Articles

Research in context

Evidence before this study from adjacent populations of heart failure with reduced
Current clinical practice guidelines strongly recommend the use ejection fraction (DAPA-HF and EMPEROR-Reduced) and those
of SGLT2 inhibitors in the treatment of patients with heart hospitalised for worsening heart failure, irrespective of ejection
failure with reduced ejection fraction (class I). However, fraction (SOLOIST-WHF). This comprehensive meta-analysis of
guidelines make a weaker recommendation for their use in over 20 000 participants provides firm evidence that SGLT2
heart failure with mildly reduced or preserved ejection fraction inhibitors reduce the risk of hospitalisation for heart failure,
(class Il, potentially due to residual uncertainties related to extend survival, and improve overall health status in patients
therapeutic effects on mortality. Although two large-scale trials with heart failure. Clinical benefits of SGLT2 inhibitors appeared
of heart failure with mildly reduced or preserved ejection consistent across broad clinical profiles and patient subgroups,
fraction (DELIVER and EMPEROR-Preserved) have been done, and extend to patients with left ventricular ejection fraction of
neither was individually designed nor powered to address these 60% or more and those already treated with other common
issues. heart failure therapies.

Added value of this study Implications of all the available evidence

We did a prespecified meta-analysis examining the effects of This comprehensive meta-analysis supports the role of the
SGLT2 inhibitors on fatal and non-fatal events overall and in SGLT2 inhibitors as foundational therapy in the management
subgroups of interest using data from DELIVER and of heart failure, irrespective of ejection fraction or care setting.
EMPEROR-Preserved. For additional context, we aligned data

In light of these uncertainties, we undertook a which were unpublished at the time of the analysis, were
prespecified meta-analysis of the two largest trials of heart included with the involvement of the trial's steering
failure with mildly reduced or preserved ejection fraction. committee. Data were extracted using standardised forms
We also extended this meta-analysis to include trials in for outcomes of interest by two authors (MV and KFD) and
patients with reduced ejection fractions (DAPA-HF and any discrepancies were resolved by consensus. If outcomes
EMPEROR-Reduced)’ and those admitted to hospital with published in other trials were not publicly available from
worsening heart failure who were enrolled with any the DELIVER and DAPA-HF trials, then we did individual
ejection fraction (SOLOIST-WHE)' to increase power to participant-data level analyses to derive treatment effect
assess various clinical endpoints, both overall and within estimates for these outcomes. Data from key secondary
subgroups of interest. In this comprehensive meta- papers were used from the EMPEROR program to support
analysis of five placebo-controlled trials, we estimated the data harmonisation.”*
effects of SGLT2 inhibitors on heart failure hospitalisations,
mortality outcomes, and health status overall, and in 14 Outcomes and subgroups
clinically relevant subgroups. The primary endpoint ofthe meta-analysis was a composite
of time to cardiovascular death or first hospitalisation for
Methods heart failure. Secondary endpoints examined included
Search strategy and selection criteria cardiovascular death, first hospitalisation for heart failure,
We did a prespecified meta-analysis of two trials of heart cardiovascular death or any worsening heart failure event
failure with mildly reduced or preserved ejection fraction (hospitalisation for heart failure or urgent heart failure visit
(DELIVER and EMPEROR-Preserved), and further requiring intravenous heart failure therapies), and death
analysed these data together with two trials of heart failure from any cause. Outcomes were adjudicated by masked
with reduced ejection fraction (DAPA-HF and EMPEROR- clinical endpoints committees, except for urgent heart
Reduced) and a trial of patients with recent worsening failure visits, which were not adjudicated in EMPEROR-
heart failure (SOLOIST-WHF). To ensure other important Preserved and SOLOIST-WHF, which relied on
trials were not missed, we did a systematic review of the investigator-reports for all events. Treatment effects on
literature via PubMed and MEDLINE of randomised, health status and quality of life were assessed using the
placebo-controlled trials with cardiovascular and kidney Kansas City Cardiomyopathy Questionnaire (KCCQ, scores
outcomes of SGIT2 inhibitors published between ranging from 0 to 100, with higher scores reflecting better
Jan 1, 2015, and July 1, 2022. To capture trials designed to health status and fewer symptoms and physical limitations)
examine clinical outcomes, we limited our selection to analysed at baseline and 8 months after randomisation.
studies enrolling at least 1000 participants with heart The effect of SGLT2 inhibitors on the primary endpoint
failure. The pre-registered search query, which was run on was examined across 14 subgroups of interest: left
See Online for appendix July 1, 2022, is in the appendix (p 1). Despite systematic ventricular ejection fraction (LVEF), history of diabetes,
search, no additional trials were identified that met criteria age, sex, race, geographical region, KCCQ total symptom
for inclusion (appendix p 4). Data from the DELIVER trial, score, body-mass index, estimated glomerular filtration

758 v thelancet.com Vol 400 September 3, 2022

 Articles

rate (eGFR), New York Heart Association (NYHA) in the intention-to-treat datasets were considered and
functional class, hospitalisation for heart failure within 12 analyses included all randomised participants. We did a
months, N-terminal prohormone of B-type natriuretic fixed-effects meta-analysis ~ with inverse-variance
peptide (NT:proBNP) concentration, baseline use of weighting for each outcome and for individual subgroups
mineralocorticoid receptor antagonists (MRAs), and to generate pooled estimates for the effect of SGLT2
baseline use of angiotensin receptor neprilysin inhibitors inhibitors compared with placebo. Between-trial
(ARNIs). The outcome of cardiovascular death or first heterogeneity of treatment effect was examined using
hospitalisation for heart failure was not available in the Cochran’s Q test. We tested treatment-by-subgroup
SOLOIST-WHF trial for subgroups. heterogeneity of effect using Cochran’s Q test. We
Select adverse events (amputations, diabetic ketoacidosis, calculated the number needed to treat (NNT) using the
hypoglycaemia, and renal events) were collated from method of Altman and Anderson.” We calculated a
DELIVER and EMPEROR-Preserved, but not directly weighted mean of the median follow-up times for NNT
compared or meta-analysed given differential data capture reporting. We considered p values below 0-05 to be
and exact definitions for these safety events (appendix p 2). statistically significant
The protocol for the meta-analysis of the DELIVER and
Statistical analysis EMPEROR-preserved trials was prespecified in the
All effect sizes were extracted as point estimates with DELIVER academic statistical analysis plan and
95% Cls. For the time-to-first event endpoints, the meta- preregistered with PROSPERO (CRD42022327527) before
analysis included data from Cox proportional hazards unmasking of the DELIVER trial results. The addition of
‘models reported as hazard ratios (HRs) and 95% Cls. For the three other trials was done post hoc (CRD42022347574).
health status and quality of life, we did responder All trials were assessed as high quality with a low risk of
analyses to identify participants with clinically Dias across the five trials (appendix p 3). All participants
meaningful improvement (=5 point increase) or provided written consent and the study protocols were
deterioration (=5 point decrease) in each of the KCCQ approved by the institutional review boards at all
scores from baseline to 8 months, analysed by logistic participating sites. Meta-analysis calculations were done
regression. For efficacy endpoints, participants included using STATA (version 16.1).

DAPA-HF (n=4744) DELIVER (n=6263) EMPEROR-Reduced (n=3730) EMPEROR-Preserved (n=5988) SOLOIST-WHF (n=1222)
Investigational drug Dapagli Dapaglifiozin Empagliflozin Empagliflozin Sotagli
Enrollment period 201718 2018-21 201719 2017-20 2018-20
Sites 410sitesin 20 countries 350sitesin 20 countries 520 sites in 20 countries 622 ites in 23 countries 306 sites in 32 countries
Key inclusion citeria LVEF 540%; LVEF >40%and evidence LVEF s40%; LVEF >40%; Type 2 diabetes;
elevated NT-proBNP; of structural heart elevated NT-proBNP; evidence of structural heart admitted to the hospital, or
NYHA functional
class IV~ disease; NYHA functional class IHV disease or history of heart failure urgent heart ailure vsit for
elevated NT-proBNP; hospitalisation within worsening heart failure;
NYHA functional 12 months; previous treatment with loop
dlass II-IV; elevated NT-proBNP; NYHA divretic for >30 days;
ambulatory or functional dlass -V previous diagnosis of heart falure
hospitalised patients (>3 months);
elevated BNP or NT-proBNP;
randomised when
haemodynamically stable, before
hospital discharge or within 3 days
ofdischarge
Key exclusion criteria CGFR <30mL/min/L73m; eGFR €GFR <20 mL/min/1.73 m’; €GFR <20 mL/min/1.73 m’; €GFR <30 mL/min/1.73 m*
SBP <95 mm Hg. <25 mL/min/1.73 m’; SBP <100 mm Hg SBP <100 mm Hg
SBP <95 mm Hg
Median follow-up time 18 months 28 months 16 months 26 months 9 months
Primary outcome Time to first cardiovascular Timeto first Time to first cardiovascular Time to first cardiovascular Total number of cardiovascular
death or heart failure cardiovascular death or death or heart failure death or heart ailure death and heart failure:
hospitalisation or urgent heart failure hospitalisation hospitalisation hospitalisations and urgent visits
visit
Placebo-group event rates
Heart failure 9:8/100 person-years 65100 person-years 15:5/100 person-years 877100 person-years
hospitalisation
Cardiovascular death 7:9/100 person-years 3-8/100 person-years 81/100 person-years 3:8/100 person-years 12:5/100 person-years
All-cause death 95/100 person-years 7:6/100 person-years 10.7/100 person-years 67/100 person-years 1631100 person-years
(Table 1 continues on next page)

www thelancet.com Vol 400 September 3, 2022 759

 Articles

DAPA-HF (n=4744) DELIVER (n=6263) EMPEROR-Reduced (n=3730) EMPEROR-Preserved (n=5988) SOLOIST-WHF (n=1222)
(Continued from previous page)
Baseline characteristics
Mean age, years 663(109) 717(96) 665 (11:2) 719(96) 70(64-76)"
Sex
Women 1109 (23-4%) 2747 (43.9%) 893 (239%) 2676 (44.7%) 412 (337%)
Men 3635 (76:6%) 3516(56:1%) 2837 (76:1%) 3312 (553%) 810 (663%)
NYHA functional class
I 3203(675) 4713(753) 2800 (75:1) 4883 (815%)
v 1541(325) 1549 (247) 930(249) 1101 (184)
Mean LVEF, % 311% (68) 542% (8.8) 272%(61) 54:3% (8:8) 35% (28-45)"
Median NT-proBNP, 1437 (857-2650) 1011(623-1751) 1910 (1115-3480) 974(499-1731) 1800 (843-3582)
pg/mt
Mean eGFR, 658 (19.4) 610(191) 622(215) 606 (19.9) 497 (40.5-64:6)"
mUmin/1.73 m*
Diabetes 2139 (451%) 2806 (44.8%) 1856 (49-8%) 2938 (491%) 1222 (100%)
History of heart failure 2251 (47.4%) 2539 (405%) 115130.9%)t 1369 (22.9%)t 1222 (100%)
hospitalisation
Heart failure medical
therapy
ACE inhibitor 2661 (561%) 2295 (36:6%) 1703 (457%) 4832 (80.7%)% 495 (40:5%)
ARB 1307 (27:6%) 2272 (363%) 908 (243%) 4832 (807%)% 515 (42:1%)
ARNI 508 (107%) 301(48%) 727 (195%) 134 22%) 205 (16:8%)
MRA 3370 (71:0%) 2667 (426%) 2661 (713%) 2244(375%) 788 (64:5%)
B blocker 4558 (96-1%) 5177 (827%) 3533 (947%) 5167 (86:3%) 1125 (921%)
Device therapy
CRTPor CRT-D 354(75%) 100 (1.6%) 442 (11.8%)
ICD or CRT-D 1242 26.2%) 168 (27%) 1171(31.4%)
Data are n (%), unless otherwise indicated. I pooled data of both treatment groups for each trial were not available, then the data forthe placebo group are displayed. ACE=angotensin converting enzyme.
ARB-angiotensin receptor blocker. ARNI=angiotensin eceptor nepriysin inhibitor. CRT-D=cardiac resynchronization therapy-defibrillator. CRT-Pcardiac resynchronisation therapy.-pacemaker. eGFRestimated
glomerular filtration rate. ICD=implantable cardioverter defbrillator. MRA=mineralocortcoid receptor antagonist. NT-proBNP=N-terminal prohormone of B-type natriuretic peptide. NYHA=New York Heart
Assodiation. “Median (1QR). tHeart failure hospitalisation withinthe peceding 12 months #Number of patients taking a renin-angiotensin system inhibitor alone o in combination with a nepriysin inhibitor.
Table 1: Characteristics of included trials and randomly assigned patients

Role of the funding source 30 mL/min/1.73m? in DAPA-HF and SOLOIST-WHF. All

There was no funding source for this study. trials were placebo-controlled and examined oral doses of
the investigational therapy (dapagliflozin 10 mg once daily
Results in DELIVER and DAPA-HF; empagliflozin 10 mg once
Overall, 21947 participants were included across five trials. daily in the EMPEROR trials; and sotagliflozin 200 mg
Median follow-up time ranged from 9 months in once daily [with dose titration to 400 mg once daily
SOLOIST-WHEF to 2-3 years in DELIVER (table 1). Except depending on side-effects] in SOLOIST-WHEF).
for SOLOIST-WHF, which randomly assigned patients The rates of incident hospitalisation for heart failure,
shortly after an episode of worsening heart failure, and cardiovascular death, and all-cause mortality were higher
DELIVER, in which a small proportion of patients (10%) in trials enrolling outpatients with heart failure with
were randomly assigned during or shortly after reduced ejection fraction than in those enrolling patients
hospitalisation for heart failure, most patients included in with heart failure with mildly reduced or preserved
this meta-analysis had chronic ambulatory heart failure. ejection fraction, and the highest event rates were
All trials required evidence of increased concentrations of reported in the SOLOIST-WHF trial, reflecting that
natriuretic peptides, although the minimum threshold for patients were randomly assigned following an episode of
eligibility differed between trials, ranging from 300 pg/mL worsening heart failure (table 1).
in patients in sinus rhythm in DELIVER and EMPEROR- Patients in trials of heart failure with reduced ejection
Preserved to 5000 pg/mL for patients in atrial fibrillation fraction were younger and more frequently men compared
or flutter and an LVEF of 36-40% in EMPEROR-Reduced. with those enrolled in trials of heart failure with mildly
Minimum eGFR for inclusion ranged from at least reduced or preserved ejection fraction (table 1). Most
20 mL/min/1-73 m2in the EMPEROR trials to at least patients in each trial were in NYHA functional class II.

760 v thelancet.com Vol 400 September 3, 2022

 Articles

Baseline median NT-proBNP actoss the trials ranged from

974 pg/mL in EMPEROR-Preserved to 1910 pg/mL in
EMPEROR-Reduced (table 1). Median eGFR was lowest in
SOLOISTWHF (50 mL/min/1-73m?). We found Cardiovascular death or heart failure hospitalisation
differences in background medical treatment according to Number with event/ Hazard ratio
number of patients () (95% )
ejection fraction, with greater use of ARNI and an MRA in
SGUT2inhibitors_ Placebo
patients with reduced ejection fraction (table 1).
HEmUEF HFpEF
Among 12251 participants from DELIVER and DELIVER A7SBBIS2%) 577/3132 (18 4%) 080(071-091)
EMPEROR-Preserved, SGLT2 inhibitors reduced com- EMPEROR-Preserved 4152997 (138%) 51112991 (17:1%) 079(069-090)
Subtotal 080(073-087)
posite cardiovascular death or first hospitalisation for heart Test for overall treatment effect p<0.0001
failure (HR 0-80 [95% CI 0-73-0-87]), without evidence of Testfor heterogeneity of effct p=089
HEEF
heterogeneity by trial (figure 1). The results were consistent DAPAHE 3B2UBTI06%) 495/2371(209%) 075(065-085)
across both components of the composite endpoint, EMPEROR-Reduced 36111863 (19.4%) 46211867 (247%) 075(065-086)
Subtotal 075(068-083)
including cardiovascular death (0-88 [0-77-1-00)) and first Test for overall treatment effect p<0.0001
hospitalisation for heart failure (0-74 [0-67-0-83]) and Test for heterogeneityof effct p=100
AIILVEF (hospitalised patients)
similar if worsening heart failure events (including both SOLOIST-WHE 071(056-089)
hospitalisations and urgent visits for heart failure), instead 077(072:082)
erlltreatment ffect p<0.0001
of hospitalisations alone, were included in the composite Test for heterogeneity of effct p=0 87 —
outcome (0-80 [0-73-0-87); appendix p 5). For the Cardiovascular death
cardiovascular death endpoint, we found consistent meta- HFmIEF HFpEF
analysis results if unknown or undetermined deaths were DELIVER WBBLE4N) 2613132(83%) - 0880074105
instead classified as cardiovascular deaths in both trials EMPEROR Preserved 1862097 (62%) 213/2991(71%) L 088073107
Subtotal 088(077-1.00)
(0-90 [0-80-1-01]). We found no significant effect on all- Test for overall treatment effect p=0.052
cause death (0-97 [0-88-1-06]). Treatment effects for the Test for heterogeneityof effct p-100
HFEF
composite endpoint of cardiovascular death or first DAPACHF 2IIBO6H 2732371 (115%) 082(069-098)
hospitalisation for heart failure were consistent across all
EMPEROR Reduced 18711863 (100%) 20211867 (108%) F— 092075112)
Subtotal 086(076-098)
subgroups of interest (appendix p 6). Test for overall treatment effect p=0.027
Although the incidence of adverse events could not be
Test for hterogeneityof effect p-0.40
AIILVEF (hospitalised patients)
directly compared across trials because of differences in SOLOIST-WHE SU60B(84%) 5BI614(94%) 084(058-122)
Overall 087(079-095)
event ascertainment and reporting, any serious adverse Test for overall treatment ffect p=0.0022
event occurred numerically less frequently in the SGLT2 Test for heterogencityof effect p-0.94. —
inhibitor groups compared with in the placebo groups in Heart failure hospital
both trials (table 2). Rates of select adverse events were HFmIEF HFpEF
infrequent and well-balanced between groups in both DELIVER 32981310105%) 41831320133%) 077(067-089)
trials (table 2). EMPEROR-Preserved 25912997 (86%) 35212991 (11.8%) 071(060-083)
Subtotal 074(067-083)
In incorporating data from all five outcomes trials of Testfor overall treatment effect p<0.0001
SGLT?2 inhibitors, the pooled estimates of the treatment Test for heterogeneity of effect p=0.46
HF(EF
effect of SGIT2 inhibitors compared with that of placebo DAPAHF 212373 (97%) 31802371.(134%) 070(059-083)
on the outcomes of interest are shown in figure 1. Overall,
EMPERORReduced 246/1863 (132%) 34211867 (183%) 069 (059-0-81)
Subtotal 069 (062-0.78)
treatment with an SGLT2 inhibitor reduced the risk of Testfor overall treatment effect p<0.0001
Testfor heterogeneity of effect p-0-90
cardiovascular death or hospitalisation for heart failure Overall 072067078)
(HR 0-77 [95% CI 0-72-0-82), with an NNT of25 (20-31) Testfor overall treatment effect p<0.0001
Testfor heterogeneity of effect p=074 —
over a weighted mean of 23 months’ follow-up (figure 1).
The risk of a first hospitalisation for heart failure was All-cause death
reduced in patients randomly assigned to an SGIT2 HEMUEF/HFpEF
inhibitor (0-72 [0-67-0-78]), with an NNT of 28 (24-35; DELIVER 497/3131 (15:9%) 526/3132 (16:8%) — 0.94(0-83-1.07)
EMPERORPreserved 422/2007 (141%) 272991 143%) W 10008115
figure 1). SGLT2 inhibitors also reduced cardiovascular Subtotal 007(088-106)
death (0-87[0-79-0-95]; NNT 88 [54-229]) and death from Testfor overall reatment effect p-0.48
Testfor heterogeneity of effect p-052
HFTEF
DAPAHF 276/2373 (116%) s20m371139% 083(071:097)
EMPERORReduced 249/1863 (134%) 266/1867 (14-2%) — 092(077-1-10)
Subtotal 087 (077-0.98)
Testfor overall treatment effect p-0.018
Figure 1: Pooled treatment effect estimates of SGLT2 inhibitors compared Testfor heterogeneity of effect p-039
‘with placebo on cardiovascular outcomes in patients with heart failure AIlLVEF (hospitalised patients)
o harmonise with the DELIVER definition, the endpoint of cardiovascular death SOLOIST-WHF. 65/608 (107%) 76/614 (12:4%) +— 082(059-114)
alone in EMPEROR-Preserved excludes unknown or undetermined deaths. Overall 092(086-099)
Testfor overall treatment effect p-0.025
HEmIEF=heart failure with mildly reduced ejection fraction. HFpEF=heart failure Testfor heterogeneity of effect p=0.46
with preserved ejection fraction. HFrEF=heart failure with reduced ejection 050 100— 1
fraction. LVEF-left ventricular ejection fraction.

www thelancet.com Vol 400 September 3, 2022 761

 Articles

DELIVER EMPEROR-Preserved
and confer clinically meaningful improvements in health-
related quality of life, with benefits seen rapidly within
Dapagliflozin Placebo. Empagliflozin Placebo months of treatment initiation. The clinical benefit of
(n=3126) (n=3127) (n=2996) (n=2989)
SGLT? inhibitors appeared consistent across a broad range
Any serious adverse event 1361 (435%) 123(455%) 1436 (47-9%) 1543 (516%) of patients, and extended to patients with LVEF of 60% or
Amputation 19 (06%) 25(08%) 16 (05%) 23(08%) more as well as those already treated with an MRA or
Diabetic ketoacidosis 2(01%) 0(00%) 4(01%) 5(02%) ARNI. SGIT2 inhibitors were safe and well-tolerated,
Hypoglycaemia 6(02%) 7(02%) 73 (24%) 78 (2:6%) without excess in serious adverse events or key adverse
Renal 73(23%) 79(25%) 363 (12:1%) 384 (12:8%) events of interest.
Adverse events were not directly compared or meta-analysed because of differential data capture and exact defnitions ‘We first examined the pooled treatment effects of SGLT2
forthese safety events in both tias. I both tials, the safty analyses were done in treated patients who received at inhibitor in the two dedicated trials of heart failure with
leasta ingle dose of the study medication. In EMPEROR-Preserved, athough limb amputations were reported through mildly reduced or preserved ejection fraction, a population
the end of the trial, other adverse events were only reported up to 7 days after discontinuation of tudy medication. in which the identification of effective therapeutics has
Similarly, in DELIVER, al reported adverse events were on-treatment or within 30 days of discontinuation of study
medication. In DELIVER, diabetes ketoacidos's includes events that were adjudicated as definite r probable cases, and historically been challenging. Trials have only identified
hypoglycaemic events represent major hypoglycaemia. DELIVER collected adverse event data from serious adverse modest clinical benefits with previously examined
events, adverse events leading to drug discontinuation or interruption, and selected adverse events, except in select therapies, and no trial to date has definitively shown a
countriesthat required reporting of all adverse events. The appendix ( 2) juxtaposesthe relevant definitins for these:
adverse events in both trils reduction in risk of all-cause or cause-specific mortality.
These findings might partly be due to greater phenotypic
Table 2: Adverse events in DELIVER and EMPEROR-Preserved heterogeneity and lower risks of death from cardiovascular
causes in this population compared with patients with
any cause (0-92 [0-86-0-99); NNT 92 [52-733]; figure 1). heart failure with reduced ejection fraction. Clinical
We found no evidence of between-trial heterogeneity of practice guidelines similarly convey this uncertainty, with
treatment effect for any of these outcomes (figure 1). More no class I recommendations offered for heart failure with
participants in the SGLT2 inhibitor groups than in placebo mildly reduced or preserved ejection fraction for any
groups had clinically meaningful improvements and fewer individual therapy (aside from diuretics). This large-scale
participants had dlinically meaningful deterioration in meta-analysis of DELIVER and EMPEROR-Preserved
each of the three KCCQ summary scores by 8 months, with harmonised data capture of patient profiles and
without evidence of heterogeneity by trial (appendix p 7). endpoint definitions increased power to assess various
The effect of SGLT2 inhibitors on the composite of clinical endpoints, including cardiovascular death. Risk
cardiovascular death or first hospitalisation for heart failure reductions in the primary composite endpoint were driven
was consistent across 14 clinically relevant subgroups by substantial and statistically robust treatment effects on
(figure 2), except for NYHA functional classification, in hospitalisations for heart failure, with more modest and
which we found an attenuation of effect in patients with statistically borderline effects on cardiovascular death.
NYHA functional classification 11T or IV (HR 0-86 [95% CI Point estimates for both components were highly
0-77-0-95]) compared with those with NYHA functional concordant between the two trials and were similar across
dlassification I1 (0-72 [0-67-0-79]; p value for heterogeneity variant endpoint definitions. These data complement the
0-015; figure 2). However, the effect of SGLT2 inhibitor clinically important health status benefits seen with
treatment was similar across tertiles of baseline KCCQ- SGLT2 inhibitors in this population in previous dedicated
total symptom score (p value for heterogeneity 0-98; trials.”” Taken together, these data should inform clinical
figure 2). We found consistent benefits across ejection decision making and guidelines.
fraction groups: 40% or less (0-75 [0-68-0-83]), 41-49% The five trials of SGLT2 inhibitors in heart failure
(0-78 [0-67-0-90]), 50-59% (0-79 [0-68-0-93]), and at least enrolled complementary populations and provided
60% (0-81[0-69-0-96); p value for heterogeneity 0-83). broader context to examine therapeutic effects across the
spectrum of disease severity and patient profiles. The
Discussion greatest benefit of the addition of an SGLT2 inhibitor to
This meta-analysis of two large, dedicated outcomes trials standard therapy in patients with heart failure was a
of SGIT2 inhibitors in heart failure with mildly reduced or 28% relative reduction in the risk of hospitalisation for
preserved ejection fraction showed that the SGIT2 heart failure, with an NNT of 28 to prevent one event over
inhibitors dapaglifiozin and empaglifiozin similarly and a follow-up of 23 months. Although smaller, the effect on
robustly reduced cardiovascular death or hospitalisation mortality was significant. These effects should be
for heart failure, without evidence of heterogeneity interpreted in the context of very high background rates of
between trials. In the comprehensive examination of use of guideline-recommended therapy for heart failure in
evidence from five trials enrolling over 20000 participants, all trials included in the meta-analysis. These estimates
SGLT2 inhibitors reduced the risk of mortality and for reductions in cardiovascular death are also highly
worsening heart failure across a broad range of patients concordant with those observed in other patient
with heart failure, irrespective of LVEF or care setting. populations extensively studied with SGLT2 inhibitors,
SGIT2 inhibitors were shown to ameliorate symptoms such as those with type 2 diabetes. Furthermore, patients

762 v thelancet.com Vol 400 September 3, 2022

 Articles

treated with SGLT2 inhibitors were 10-20% more likely to However, because these were not head-to-head
have improvements in health status and, conversely, were comparisons, we cannot exclude the possibility that select
10-20% less likely to face important deterioration in differences in clinical efficacy and safety might still exist.
health status compared with patients in control groups. This meta-analysis also clarifies previous uncertainties
This composite evidence underscores the benefits of in the efficacy of SGLT2 inhibitors in specific patient
SGLT2 inhibitors on meaningful dlinical events, symptom groups. A previous meta-analysis of DAPA-HF and
burden, and overall health status in patients with heart EMPEROR-Reduced reported potential heterogeneity in
failure. We found no statistical heterogeneity across the treatment effects by both race and region;* in the
five trials for any endpoint and thus the clinical benefits of present meta-analysis with over twice the number of
the three tested therapies are assumed to be similar. patients, no evidence of heterogeneity was seen. We

A Wk B Age
Number ith event/ Hazard ratio Nomberwith event/ Hazard ratio
nomberof patients () (95% ) number of patients (%) (95%C)
SGlT2inkibitors _Placebo SGLT2inibitors Placebo
WeF40% <Gsyears
DAPAHE IWPBBA6TY 4952371009%) —_ 075065085 DELVER 961668 (144%) 110/677 (16:2%) —— 087(066-114)
EMPEROR Reduced 3611863 (19.4%) 4621867 (247%) — 075(065086) EMPEROR Preserved 13411066 (126%) 15211084 (14.0%) B 088 070-111)
subtotal = 075068083 DAPAHF U5/9520152%) 1BUSBATSN) —e—— 075(061-094)
Test foroveraltreatment effect p<0.0001 EMPEROR Reduced 12B/675(190%) 193/740(261%) ——e—— 071(057-089)
Test for hetrogeneityof effectpe1.00 Subtotal — 079(070-089)
WEF 41-49% Test oroverllretment efect pe0.0001
DELIVER 1931067 (181%) 22011049 (10%) — 084(069-102) Testfor heterogencityof efiect p-0.49
EMPEROR Preserved LS/905 (46%) 193/988(195%) ——e—0 071057089 =6syears
subtotal — 078(067090) DELVER 902463 (154%) 46712455 (190%) — 079(065-090)
Test foroveraltreatment effect p=0.0008 EMPEROR Preserved 281/1931(146%) 3591907 (188%) —e— 075 (064-087)
Test for hetrogeneityof effectp-0.26 DAPAHE WINPT e 074062087
LVEF50-59% EMPEROR Reduced 233/118B(196%) 26071127 239%) — 078(066-093)
DELIVER 1B 429 1961123 (75%) —_— 075(064098) Subtotal < 077071083
EMPEROR Presenved 13811028 (34%) 173100 168%) ——=—] 080(064-098) Testforoveral treatment ffect p0.0001
subtotal _ 079(068-093) Testfor hterogenetyof ffct p-052
Test foroveraltreatment efect p=0.0032 Test fortretment by subgroup nteraction p-0.67
Test for hetrogeneiyof effectp-0.94.
WEF260%
DELVER 2USI(130%) 161960(168%) ——e—— o076(060095) D Race
EMPEROR Preerved 132/974(136%) 1451973 (149%) — 087(069-110)
Subtotal _ 081069-096) Asin
Test foroveraltreatment ffect p=0.015 DELVER 960(146%) 100/644 (155%) —— on0&an
Test for hetrogeneityof effectp-0.42 EMPEROR Presenved 54413 (131%) 77/411(187%) 4—————— 065(046-092)
Test fortreatment by subaroup interacton p-03 _ DAPAHF TSI 116564 (206%) 4——e— 065(045-086)
EMPERORRedced 621337 (18.4%) 99/335(296%) 4 057 (0:41-078)
Subtotal — 070(060-081)
C sex Test for overall treatment effect p<0.0001
Tes fo eterogeneityof fect 014
wale White
DELIVER 207/1767068%) 350/1749 200%) — 081065094 pLVER U3asSH) 4362225 (196%) —_ 077 (067-089)
EMPEROR Preserved 25311659 (153%) 20711653 (18.0%) — 081(063-096) EMPEROR Preserved 31012286 (136%) 370/2256 (16.4%) —_ 081(069-094)
DAPAHE 0MB0967%) 400B (10— 073063085 oPaF 7RGE2A65H) 3441671.206%) —_ 078(067-092)
EMPEROR Reduced 294/1426 206%) 3531411 (250%) —_— 080(068-093) EMPEROR Reduced 264/1325(199%) 28911304 (222%) —t 088(075-104)
Subtotal < 078(073:085) subtotal - 081(075-087)
Test foroveraltreatment effect p<0.0001 Testforaverall reatment efect p<0 0001
Test for heterogenetyof effect p-074. Test for heterogeneityofefect p=0 64
Female Black
DELIVER 781364 30%) 271383 164%) —_ 079(065096) pELVER WAL 77BEIBY fo—> 105054204)
EMPEROR Preserved 1621338 (121%) 214/38(160%) ——e— 075(061092) EUPERORPresenved 24/133(1B0%) BS(24N) ———e——1—— 073042125
DAPAHE 90564040%) 0SS5 TAN) ———e— 080(059-108) DAPAHF 2uAR097%) 2104 G08K) ¢—=——| 057(034-057)
EMPERORReduced 67437 (153%) 109/456239%) 4—=—— 059(044-080) EMPERORReduced 24/123(195%) 48/ (358%) +——— 046(028-075)
Subtotal S 074(066084) subtotal _— 063(048-083)
Test foroveraltreatment effect p<0.0001 Test oroverllretment efectp=0.0009.
Test for hteragenetyofeffectp-0.41 Testforheterogeneiyofefect p-0 24
Test fortreatment by subgroup interacton p-0.45. ] Other
050 075 100 125150 DeLvER BOGAOTH 2485 (130%) S 082(0.46-1.47)
EMPEROR Preserved 27/164(165%) I6/9B(1824) —————at——> 095(058157)
DAPAHE TN RO > 18204671)
EMPERORReduced SISLO®K) 1463229 4 — 041(01514)
Subtotal ———— oss061120)
Testforaverall reatment efect p=0.37
Testforheterogeneiyofefiect p-0:34
Testforseatment by subgroup interaction p-015. —_—
050 075 100 125150
(Figure 2 continues on next page)

www.thelancet.com Vol 400 September3, 2022 763,

 Articles

E Region F NYHA functional lass

Nomberwith event/ Hazard ratio Nomber with event/ Hazard ratio
number of patients () (95%C) number of patients () (955
SGiT2inibitors Placebo SGiT2inhibitors Placebo
Latin America NvHAN
DELVER 65602 108%) 861579 (149 ——————1 07305310 pELVER 002314 (130%) 38912399 (162%) —_ 078067091
EMPEROR Preserved 105/758(139%) 1207757 (159%) e 087(067113) EMPEROR Pesened 2752435 (3% 3OUS2(147%) —e 075(064-087)
DR HE U055 9IS 231 4 065(047-089) DAPAHE 189606 (17%) 28671567 (179%) —e— 063052-076)
EMPEROR Roduced 115/641(79%) 151645 (23.4%) ——e— 073(058-094) EMPERORReduced 22011399 (157%) 299/1401013%) ——e—0 071(059-084)
subtotal — 075(065086) Subtotal <= 072(067-079)
Testforoverall reatment ffectp<0 0001 Test foroveraltreatment efect p<0.0001
Test for heterogeneityof effect p-0.56 Test forheterogeneityof efct p-0.36
Europe NYHA IV
DELVER 239/1404160%) 28901511 19:1%) — 0820060007 DEUVER B 135 1870732055%) — 082(066-100)
EMPEROR Preserved 165/1346 (123%) 20201343 15.0%) —_— 080(065-098) EMPERORreserved 140/562(249%) 1501539 27:8%) —_ 086(068-109)
DAPAHE 192/1094(176%) 215/1060(203%) — 084(063103) DAPANE 194767(253%) 200/774(270%) e 091(075-110)
EMPEROR Reduced 1401676 207%) 1491677 (220%) — 094(074-118) EMPEROR Reduced 1411464 (304%) 163/466 (350%) ] 083(066-104)
subtotal — 084(076093) Subtotal s 086(077-095)
Testforoverall eatment efect p-0 0006 Test foroveralreatment efect p=0.0048
Testorheterogeneity ofefect p=0.76 Test for heterogeneityofeffect p=0.89
North America Testfortrestment by subgoup ntraction p-0015
DELVER BBaosn) losanEasy) ———] 076 (057-101)
EMPEROR Preserved 641360(78%) 831359 (31%) 072(052100)
DAPAHE Ssass PBeEI) ——— o7sooy H KeQ-Tss
EMPERORReduced 48212(226%) 64213 (300%) 4———— 069 (048-101)
subtotal —_ 072 (061085 KCQTSSTertled
Testforoverall restment f ectp=0.0 01 DEUVER BI0SOAT4N) 239099002410 ——e—— 069(057-084)
Test orheterogeneity ofefect p-0.98 ENPEROR Prserved — 085(070-104)
o DAPAHE 1W0768Q08%) 0T8N ——e—— o71(057-087)
prvER i) s 08006719 EMPERORReduced 174/595(292%) 199/621(320%) ] 084(069-104)
EMPEROR Presenved 45343(31%) 69/343Q01%) 4——— oso(o4r0gs Subtotal = 077(070-085)
DAPAHE 5430406 12553 03%) e 066(049-088) Testforovrallteatment effectp<0
EMPEROR Reduced 49/248199%) 801245 (327%) 4——— 0s5(038-078) Testforheterogenityofefiect =0330001
subtotal —_ 068 (0sa080) KCQTSSTertie2
Testforoverall reatment ffectp<0 0001 DEUVER ORI U021 e 077(061-057)
Testor heterogeneiy ofefect p=0.15 EMPEROR Prserved —_— 076(060-096)
Testforrestment by sobgroup mteracton p-013 DAPAHE HETIs 50791090 ——e—| 077(061-099)
T EMPERORReduced 1120623(180%) M6/62(235%) ——e—— 073(057094)
subtotal — 076(067-085)
G NT-proBNP concentration Test for verllrestment fect p<0 0001
Test for heterogenciyofeffect p=0.99
NT-proBNP < Mecian KeCQTssTetie3
DELIVER 16201552 (104%) 19301576 (122%) . 085(065-105) DELVER 25/920(136%) 109/880(124%) —f—— 1008514
EMPEROR Presenved 126/1477 (BS%) 1681508 (1 .1%) ——=—— 076(061-096) EMPEROR-Preserved —_— 071(05503)
DAPAHE Q1N IS (130%) 4—e—— 061(048075) DAPATE 7693105 115699 (165%) 4—e—— 062(046-083)
EMPEROR Reduced B0/699 (114%) 11SI724(159%) ———e— 070053093 EMPERORReduced 7A/6IS(17%) 11S/603(189%) 4—e— 061(045-081)
subtotal - 07406608y Subtotal - 076(066-087)
Testforoverall reatment ffectp<0 0001 Test foroveralreatment efect p<0.0001
Test for hterogeneityofeffect =024 Test forheterogenetyof effct p=0.0067
NT.proBNp = Mecian Test fortrestment by subgroup interacion 098 ———t——
DELIVER 31301579 (19:8%) 38411555 (247%) — 077 (066-089) 050 075 100 125150
EMPEROR Prservd 2881516 (19.0%) 34111476 (231%) — 078067097
DAPAHE BsATIE4 4T BT —_— 080(069-094)
EMPEROR Reduced 1691631 (268%) 249/661G77%) ——o—— 065(053-079)
subtotal = 076(070-082)
Test focveraltrestment fect 00001
Testforheterogeneity of efect p-0.40
Testfortreatment by subgroup nteaction p-070 ———
0% 075 100 125150
(Figure 2 continues on next page)

found a nominally significant interaction for the from this meta-analysis support the safety and benefits
treatment effect according to NYHA functional class, of commencing SGIT2 inhibitor, irrespective of care
with an attenuated effect in patients in NYHA class 11T setting, although an ongoing trial is investigating a
or IV compared with NYHA class I1. However, subgroup strategy of in-hospital initiation of an SGLT2 inhibitor
analysis of patient-reported symptom burden, assessed (NCT04363697).
using the KCCQ total symptom score, did not show any Previous trials of renin-angiotensin system inhibitors,
evidence of heterogeneity in treatment effects and other MRAs, and ARNIs have identified attenuation of benefits
measures of severity of heart failure, including of these therapies among patients with truly normal
natriuretic peptides, recency of hospitalisation, and LVEF.** A similar pattern of attenuation at higher LVEF
LVEF, did not show any evidence ofheterogeneity. Data was suggested for select endpoints in the EMPEROR-

764 wwthelancet.com Vol 400 September3, 2022

 Articles

| Diabetes status ) Body-massindex

Numberwith event/ Hazard Nomberwith event/ Hazardratio
numberof patients () (ossc) numberof patients (%) (o5%)
SGiT2inhbitors _ Paccbo SGT2intibitors _ Placebo
Diabetes Bs30kgin
oELVER B0 (77%) 29811405 (212%) —_ 081(068.095) DELIVER nEs (56N W05 —e— 075(063.089)
EMPEROR Presenved 239/1466 (163%) 29111472 (1984) —_ 079(067-004) EMPEROR Presenved 192/1343(143%) 21911349 162%) —t 085(070-103)
oAPAHE B (198) 269N0640252%) —e— 075(063-090) DAPAHE V4B (4N VIBBQLAN) ——— 068(054-086)
EMPEROR Redced 2001927 (216%) 265029 (285%) —e—0 072(060-087) EMPERORReduced 135/600(225%) 1401567 (247%) e 085(067-1.08)
Subtotal < 0770070088 Subtotal < 078(070-086)
Test foroveraltrestment efect p<0.0001 est oroverllrestment efect pe0.0001
Test for hetrogeneiyof efect p=0.80 Tes for hterogeneityof ef ectp-0.43
Nodiabetes B <0kgin’
DEVER 20BN 27901727 162%) —_— 080(067095) DELVER 61748 29211736 168%) — 085072-101)
EMPEROR Preseved 76SI(1SK) 2201519(145%) ——e——0 078(064095) EMPERORPresenved 2231654 (135%) 2021642(178%) —e—o 074(062.088)
DAPAHE 1691298130 2707 07.4%) ——e— 073(060-089) DAPAHF 2B/1537(168%) 31601533 206%) —_ 079(067-093)
EMPEROR Redoced 161936 17:2%) 1971938 21.0%) —_ 07B(064097) EMPERORReduced 226/263(179%) 321300 048%) —e— 070(055-083)
Subtotal < 0770070-085) subtotal < 077(071-084)
Test foroveral treatment effect p<0.0001 Testforoverall reatment efect p<0.0001
Tes for heterogeneityof effect p-0.92 estforheterogeneiyofefect p=0-42
Test fortreatment by subgroup interaction p-093 [ Tes for treatment by subgroupinteracion p=0.84.

K Kidney function L Hospitalisation for heart failure in previous12 months

<GFR<60 mL/min per 1 73m’ Hear falur hospitalisation within 12 monthsof andomisation
DEVER 26611516 (175%) 34211554 220%) —_ 077(066090) DEIVER 1ByE0 (22N NOBSEBENY ——e— 073 (060-088)
EMPEROR Preserved 263/1504 (175%) 32111484 (21.6%) —_— 078(066-09) EMPEROR Prsened 157/699 (225%) 19206700876 ——e—— 073(059-090)
DAPAHE WU2099%) %6458 —e— 0730061088 DAPAHE 16883 18UE63(273%) ——e—— 063(050-080)
EMPEROR Redoced 202/893 (22:6%) 2371906 (262%) — 083(069-100) EMPERORReduced 153/S77(265%) 1771574 (30:8%) —_— 079 (064-039)
Subtotal < 078(071-084) Subtotal S 072(065080)
Test foroveral treatment effect p<0 0001 Test oroverlrestment efect pe0.0001
et for heterogencity of efect 081 Testforhetergenelty ofefectp-0.58
€GFR =60 miminper 173m No hear falur hospitalisation witin 12 months f andomisation
DEUVER 20071615 @29%) 23511577 (14.9%) — 086(071104) DELVER 292302026%) 34712327 (149%) — 084(072-098)
EMPEROR Preserved 15211493 (102%) 189/1505 (12:6%) — 0B1(065100) EMPEROR resenved 2582298 (112%) 319/2321 (137%) —_— 081(068-095)
OAPAHE 1UI0(135N) 206 (075 —e— 07506209 DAPAHF 5T (530 31471708 (18.4%) —_— 082(069-096)
EMPERORReduced 159/969(16.4%) 224/960(233K) ———— 067(055:083) EMPERORReduced 208/1286(162%) 2851203(20%) —e— 071060085
Subtotal < 077(070085) Subtotal < 080(073-087)
Test foroveralrestment efect p<0.0001 Testoroverallreatment efect p<0.0001
Test for heterogeneityof effect p-034 Testfor heterogeneiyofefect 052
Test fortreatment by subgroup ineraction p=092 Tes for treatment by subgroup ineraction p=0-15 _

M mRA N ARNI
NoMRA No ARNI
DEUVER MBS IS5 32311805 (179%) — 08607310 DEIVER 44412966 (150%) 54712996 (18:3%) — 080071091
EMPEROR Preseved 2331878 (124%) J06/1866(16.4%) —e— 073062087 DAPAMHF 61N 412113 09%) — 075(065-086)
DAPAHE 0206771520 1400697 201%) 072(056-093) EMPERORReduced 31011523 (20.4%) 36971480 (249%) — 077(066-090)
EMPERORReduced 118557 (12%) 132/512058%) ——e—— 076(059097) Subtotal < 078(072:084)
Subtotal < 078(071:086) Testforoveral tretment effect 0001
Test foroveralreatment efect p<0.0001 estforheterogeneiyofefect =079
et for heterogencty of efect p-0-49 AN IS8 062N 4 076(046-126)
MrA DELVER 40250060%) 54258 (209%) ————e—1— 077051115
DEVER WNOATH) BT 091N e 074(062:089) DAPAHF SUM0(S0N) 93387 Q40%) 4—e—— 064(045-089)
EMPEROR Preserved 182/1119 (163%) 205/1125 18.2%) — 087(071106) EMPEROR Reduced _ 070(056-089)
OAPAHE 280/1696 165%) 355/1674 (212%) — 075(064088) Subtotal
EMPEROR Reduced 243/1306 (186%) 33011355 44%) —=— 075(063:088) Testforoveral treatment effect p=0.0031
subtotl = 077(070-084) Testfor hterogeneityof effct p-075
Test foroveralrestment efect p<0.0001 Testforeatment by subgroup intraction p-0.45
Test for heterogeneityof effect p-0.61
Test fortreatment by subgroup ineracton p=086
050 075 100 125150 050 075 100 125150
Figure 2 Treatment effects of SGLT2 inhibitors on the composite of cardiovascular death or first hospitalisation for heart failure across 14 clinically relevant subgroups
The age-based subgroup for EMPEROR-Preserved was dichotomised at 70 years. In the DELIVER trial, Saud Arabia was included i the Europe region. The NT-proBNP-based median concentration was
calculated on the basis of atral fibrillation or flutter status in EMPEROR- Preserved. ARNI=angiotensin receptor neprilysin inhibitor. eGFR=estimated glomerular filtration rate. KCCQ-TSS=Kansas City
Cardiomyopathy Questionnaire—Total Symptom Score. MR ineralocorticoid receptor antagonists. NT-proBNP- -terminal prohormone of B-type natriuretic peptide. NYHA=New York Heart
Association.

Preserved trial, although the interaction test for the in both DELIVER and EMPEROR-Preserved, the clinical
primary endpoint across LVEF subgroups was not benefits of SGLT2 inhibitors clearly extend to patients with
significant.” However, when pooling data from subgroups heart failure and LVEF of 60% or more, with an

www.thelancet.com Vol 400 September3, 2022 765

 Articles

approximate 20% risk reduction in the primary composite definitions of most other efficacy endpoints, safety event
endpoint. As such, SGIT2 inhibitors should not be definitions could not be reconciled because of differential
withheld from patients with heart failure who would timeframes of assessment and data ascertainment. A renal
otherwise be appropriate candidates for the therapy and composite endpoint was not a prespecified secondary trial
yet have an ejection fraction of 60% or more. Indeed, the endpoint in DELIVER. Consequently, DELIVER did not
benefits of SGLT2 inhibitor are complementary and systematically collect these data and renal events were only
additive to those of an ARNI and MRA across the range of available as serious adverse events or adverse events
ejection fraction. In addition to their established role as leading to drug discontinuation. As such, renal endpoints
treatments for heart failure with reduced ejection fraction, could not be compared across trials and thus were omitted
both of these medications might be considered at higher from the meta-analysis. No correction was made for
ejection fractions in the guidelines based on the post hoc multiplicity of testing for subgroup analyses.
analyses indicating benefit in patients with an LVEF ofless In conclusion, this meta-analysis of 21947 patients with
than normal (approximately 55-60%). A nominally heart failure across the full spectrum of ejection fraction,
significant interaction was found in EMPEROR-Preserved including both outpatients and hospitalised patients,
for baseline MRA use for the endpoint of first and showed that SGLT? inhibitors significantly reduce the risk
recurrent hospitalisations for heart failure, with the of mortality and worsening heart failure and improve
suggestion of less benefit among those treated with an patient symptoms and overall health status when added to
MRA compared with those who were not* This meta- standard therapy for heart failure. SGLT2 inhibitors
analysis, however, showed consistency of benefits should be considered foundational therapy in all patients
irrespective of background use of ARNI or MRA. with heart failure, irrespective of LVEF or care setting.
Therefore, these data support the use of SGLT2 inhibitors Contributors
across the spectrum of ejection fraction, regardless of MV, BLC, PSJ, JJVM, and SDS conceived of and designed the study.
background therapies.® The superior strength of evidence MV, KFD, BLC, PSJ, JJVM, and SDS did the analysis. MV and KED
for SGIT2 inhibitors, compared with an MRA and ARNIT
drafted the manuscript. All authors contributed to data interpretation
and writing of the final version of the manuscript, and all authors were
in heart failure with mildly reduced or preserved ejection responsible for the decision to submit the manuscript. MV, KFD, BLC,
fraction, along with their favourable safety profile, the and SDS accessed and verified the data and all authors had full access to
minimal requirement for monitoring, rapid onset of the study data.
benefit, and beneficial effects on kidney function, supports Declaration of interests
prioritising initiation of SGLT2 inhibitor use in all patients MV has received research grant support o served on advisory boards for
with heart failure.
American Regent, Amgen, AstraZeneca, Bayer Baster Healthcare,
Bochringer Ingelheim, Cytokinetics, Lexicon Pharmaceuticals, Novrtis,
Although the meta-analysis of DELIVER and EMPEROR- Pharmacosmos, Relypsa, Roche Diagnostics, and Sanofi; received speaker
Preserved was prespecified and preregistered, the fees from AstraZeneca, Novartis, and Roche Diagnostics; and participates
supportive five-trial meta-analysis was done post hoc. on dlinical trial committees for studies sponsored by Galmed, Novartis,
Bayer, Occlutech, and Impulse Dynamics. KFD's emploger has been
Furthermore, no alpha was ascribed to this meta-analysis remunerated by AstraZeneca for clinical trial work. KED also reports
and as such, these results cannot be considered hypothesis speakers fees from AstraZeneca and research funding from Boehringer
testing. We did not have access to individual participant- Ingelheim. KFD, PS], and JJVM are funded by a British Heart Foundation
level data from the EMPEROR trials or SOLOIST-WHF Centre of Research Excellence Grant. BLC has received consulting fees
from Bochringer Ingelheim. PS]'s employer has been remunerated by
and thus relied on published data available in the public AstraZeneca, Bayer, and Novo Nordisk for clinical trial work. PS] also
domain; certain subgroup variables might be better reports consulting and speakers'fees from Novartis, AstraZeneca, and
modelled as continuous measures rather than at the Bochringer Ingelheim; and research funding Boehringer Ingelheim.
reported cut-points. Although the meta-analysis improved RadB has received research grant support from AstraZeneca, Abbott,
Bochringer Ingelheim, Cardior Pharmaceuticals, lonis Pharmaceuticals,
precision around pooled treatment estimates in Novo Nordisk, and Roche; and received speaker fees from Abbot,
subpopulations of interest, interaction testing might still AstraZeneca, Bayer, Novartis, and Roche. AFH reports research grant
be underpowered. Subgroup data for the outcome of support from American Regent, Amgen, AstraZeneca, Bayer, Boehringer
interest were not available for the SOLOIST-WHF trial.
Ingelheim, Merck, Novartis, Somologic, and Verily; and consulting Fees
from Amgen, AstraZeneca, Bayer, Biofourmis, Boston Scientific,
The findings from the meta-analysis are most generalisable Cytokinetics, Merck, Novartis and NovoNordisk. SEI has served on clinical
to patients seen in clinical practice settings similar to those trial committees or as a consultant to AstraZeneca, Bochringer Ingelheim,
of enrolled trial participants. All trials enrolled fewer than Novo Nordisk, Lexicon, Merck, Pfizer, VIV Therapeutics, Abbott, and
5-10% Black patients, partly reflective of the global racial Esperion; and has given lectures sponsored by AstraZeneca and
Boehringer Ingelheim. MNK reports research grant support from
representation of populations served by participating sites AstraZeneca and Boehringer Ingelheim; and consulting fees from
around the world. None of the included trials enrolled Alnylam, AstraZeneca, Amgen, Bayer, Bochringer Ingelheim,
patients with severe kidney dysfunction (eGFR Cytokinetics, Esperion, Eli Lilly, Janssen, Lexicon, Merck (Diabetes and
<20 mL/min/1-73 m2) or on dialysis; therefore, no Cardiovascular), Pharmacosmos, Novo Nordisk, Sanofi, and Vifor. CSPLis
supported by a Clinician Scientist Award from the National Medical
conclusions regarding the efficacy or safety of SGLT2 Research Council of Singapore; has received research support from Bayer
inhibitors in these patients can be inferred. Urgent heart and Roche Diagnostics; has served as consultant or on the advisory board,
failure visits were not centrally adjudicated in the steering committee, or executive committee for Abbott, Actelion, Alleviant
EMPEROR-Preserved trial. Although we were able to align Medical, Allysta Pharma, Amgen, AnaCardio, Applied Therapeutics,

766 v thelancet.com Vol 400 September3, 2022

 Articles

AstraZeneca, Bayer, Boehringer Ingelheim, Boston Scientific, Ferreira JP, Zannad F, Pocock SJ, et al. Interplay of
Cytokinetics, Darma, EchoNous, Impulse Dynamics, lonis ‘mineralocorticoid receptor antagonists and empaglifiozin in heart
Pharmaceutical, Janssen Research & Development, Medscape/WebMD failure: EMPEROR-Reduced. | Am Coll Cardiol 2021; 77: 1397-407.
Global, Merck, Novartis, Novo Nordisk, Prosciento, Radliffe Group, Roche 7 McMurray ]V, Solomon SD, Inzucchi SE, et al. Dapaglifiozin in
Diagnostics, Sanofi, Siemens Healthcare Diagnostics, and Us2.ai; and patients with heart failure and reduced ejection fraction.
serves as co-founder and non-executive director of Us2.ai. FM has received N Engl ] Med 2019; 381: 1995-2008.
personal fees from AstraZeneca. SJS reports research grants from the US Packer M, Anker SD, Butler |, et al. Cardiovascular and renal
National Institutes of Health, Actelion, AstraZeneca, Corvia, Novartis, outcomes with empagliflozin in heart failure. N Engl ] Med 2020;
and Pfizer; and consulting fees from Abbott, Actelion, AstraZeneca, 383: 1413-24,
Amgen, Aria CV, Axon Therapies, Bayer, Boehringer Ingelheim, Boston Bhatt DL, Szarek M, Steg PG, et al. Sotagliflozin in patients with
Scientific, Bristol Myers Squibb, Cardiora, Coridea, CVRx, Cyclerion, diabetes and recent worsening heart failure. N Engl ] Med 201
384:117-28.
Cytokinetics, Edwards Lifesciences, Eidos, Eisai, Imara, Impulse
Dynamics, Intellia, Ionis, Ironwood, Lilly, Merck, MyoKardia, Novartis, 10 Anker SD, Siddiqi T], Filippatos G, et al. Outcomes with
Novo Nordisk, Pfizer, Prothena, Regeneron, Rivus, Sanofi, Shifamed, empaglifiozin in heart failure with preserved ejection fraction using
‘Tenax, Tenaya, and United Therapeutics. ASD reports institutional grant DELIVERlike endpoint definitions. Fur ] Heart Fail 2022;
24: 1400-05.
support from Abbott, Alnylam, AstraZeneca, Bayer, Novartis; 1 Butler ], Filippatos G, Jamal Siddiqi T, et al. Empaglifiozin, health
and consulting fees from Abbott, Alnylam, AstraZeneca, Avidity, Axon status, and quality of life in patients with heart failure and
‘Therapeutics, Bayer, Biofourmis, Boston Scientific, Cytokinetics, preserved ejection fraction: the EMPEROR-Preserved Trial.
GlaxoSmithKline, Merck, Novartis, Parel, Regeneron, Roche, and Verily. Circulation 2022; 145: 184-93.
JJVM has received funding to his institution, Glasgow University, for his 2 Butler ], Anker SD, Filippatos G, et al. Empaglifiozin and health-
work on dlinical trials, consulting, and other activities from Alnylam, related quality of life outcomes in patients with heart failure with
Amgen, AstraZeneca, Bayer, Bristol Myers Squibb, Cardurion, reduced ejection fraction: the EMPEROR-Reduced trial. Eur Heart |
Cytokinetics, GlaxoSmithKline, Novartis, Pfizer, and Theracos; and has 2021; 42:1203-12.
received personal lecture fees from the Corpus, Abbott, Hickma, Sun B Zannad F, Ferreira JP, Pocock SJ, et al. SGLT2 inhibitors in patients
Pharmaceuticals, and Medsca. SDS has received research grants from with heart failure with reduced efection fraction: a meta-analysis of
Actelion, Alnylam, Amgen, AstraZeneca, Bellerophon, Bayer, Bristol the EMPEROR-Reduced and DAPA-HF trials. Lancet 2020;
Myers Squibb, Celladon, Cytokinetics, Eidos, Gilead, GlaxoSmithKline, 396: 819-29.
Tonis, Lilly, Mesoblast, MyoKardia, National Institutes of Health/National 1 Packer M, Anker SD, Butler ], et al. Effect of empaglifiozin on the
Heart Lung and Blood Institute, Neurotronik, Novartis, NovoNordisk, clinical stability of patients with heart failure and a reduced ejection
Respicardia, Sanofi Pasteur, Theracos, and US2.AL; and has consulted for fraction: the EMPEROR-Reduced trial. Circulation 2021;
Abbott, Action, Akros, Alnylam, Amgen, Arena, AstraZeneca, Bayer, 143: 326-36.
Boehringer Ingelheim, Bristol Myers Squibb, Cardior, Cardurion, Corvia, 15 Packer M, Butler |, Zannad F, et al. Effect of empagliflozin on
Cytokinetics, Daiichi-Sankyo, GlaxoSmithKline, Lilly, Merck, Myokardia, worsening heart failure events in patients with heart failure and
Novartis, Roche, Theracos, Quantum Genomics, Cardurion, Janssen, preserved ejection fraction: EMPEROR-Preserved trial. Circulation
2021; 144: 1284-94,
Cardiac Dimensions, Tenaya, Sanofi-Pasteur, Dinagor, Tremeau, 16
CellProThera, Moderna, American Regent, and Sarepta Packer M, Zannad F, Anker SD. Heart failure and a preserved
ejection fraction: a side-by-side examination of the PARAGON-HF
Data sharing and EMPEROR-Preserved trials. Circulation 2021; 144: 1193-95.
‘The trial sponsor is committed to responsible data sharing principles, Altman DG, Andersen PK. Calculating the number needed to treat for
including sharing of anonymised individual patient-level data and trials where the outcome is time to an event. BMJ 1999; 319: 1492-95.
relevant clinical documents with qualified researchers. The trial data 19 Spertus JA, Birmingham MC, Nassif M, et al. The SGLT2 inhibitor
availability is according to the criteria and processes described at https:// canagliflozin in heart failure: the CHIEF-HF remote, patient-
astrazenecagrouptrials pharmacm.com/ST/Submission,Disclosure. centered randomized trial. Nat Med 2022; 28: 809-13.
20 Nassif ME, Windsor SL, Borlaug BA, et al. The SGLT2 inhibitor
Acknowledgments dapaglifiozin in heart failure with preserved ejection fraction:
‘The DELIVER and DAPA-HF were funded by AstraZeneca, the a multicenter randomized trial. Nat Med 2021; 27: 1954-60.
EMPEROR trials were funded by Boehringer Ingelheim and Eli Lilly, 2 McGuire DK, Shih WJ, Cosentino F, et al. Association of SGLT2
and SOLOIST-WHF was funded by Sanofi and Lexicon Pharmaceuticals. inhibitors with cardiovascular and kidney outcomes in patients with
References type 2 diabetes: a meta-analysis. JAMA Cardiol 2021; 6: 148-58.
1 McDonagh TA, Metra M, Adamo M, et al. 2021 ESC Guidelines for 2 Zannad F, Ferreira JP, Pocock S, et al. SGLT2 inhibitors in patients
the diagnosis and treatment of acute and chronic heart failure. with heart failure with reduced ejection fraction: a meta-analysis ofthe
Eur Heart ] 2021; 42: 3599-726 EMPEROR-Reduced and DAPA-HF trials. Lancet 2020; 396: 819-29.
2 Heidenreich PA, Bozkurt B, Aguilar D, et al. 2022 AHA/ACC/ 2 Solomon SD, Anavekar N, Skali H, et al. Influence of ejection
HFSA Guideline for the management ofheart failure: a report of fraction on cardiovascular outcomes in a broad spectrum of heart
the American College of Cardiology /American Heart Association failure patients. Circulation 2005; 112: 373844,
Joint Committee on Clinical Practice Guidelines. ] Am Coll Cardiol % Solomon SD, Claggett B, Lewis EF, et al. Influence ofejection fraction
2022; 79: 1757-80. on outcomes and efficacy of spironolactone in patients with heart
3 Anker SD, Butler J, Filippatos G, et al. Empaglifiozin in heart failure failure with preserved ejection fraction. Fur Heart ] 2016; 37: 455-62.
with a preserved cjection fraction. N Engl ] Med 2021; 385: 1451-61. 2 Solomon SD, Vaduganathan M, L Claggett B, et al. Sacubitril/
4 Solomon SD, McMurray JJV, Claggett BL, et al. Dapagliflozin in valsattan across the spectrum of ejection fraction in heart failure.
heart failure with mildly reduced or preserved ejection fraction. Circulation 2020; 141: 352-61.
N Engl ] Med 2022; published online Aug 27 https://doi. 2 Vaduganathan M, Claggett BL, Inciardi RM, Fonarow GC,
0rg/10.1056/NEJMoa2206286. McMurray JJV, Solomon SD. Estimating the benefits of
5 Butler ], Packer M, Filippatos G, et al. Effect of empaglifiozin in combination medical therapy in heart failure with mildly reduced
patients with heart failure across the spectrum of left ventricular and preserved ejection fraction. Circulation 2022; 145: 1741-43.
ejection fraction. Eur Heart ] 2022; 43: 416-26.

wwww thelancet.com Vol 400 September3, 2022 767