Articles

Semaglutide and cardiovascular outcomes in patients with

obesity and prevalent heart failure: a prespecified analysis of
the SELECT trial
John Deanfield, Subodh Verma, Benjamin M Scirica, Steven E Kahn, Scott S Emerson, Donna Ryan, Ildiko Lingvay, Helen M Colhoun, Jorge Plutzky,
Mikhail N Kosiborod, G Kees Hovingh, Søren Hardt-Lindberg, Ofir Frenkel, Peter E Weeke, Søren Rasmussen, Assen Goudev, Chim C Lang,
Miguel Urina-Triana, Mikko Pietilä, A Michael Lincoff, for the SELECT Trial Investigators

Summary
Background Semaglutide, a GLP-1 receptor agonist, reduces the risk of major adverse cardiovascular events (MACE) Lancet 2024; 404: 773–86
in people with overweight or obesity, but the effects of this drug on outcomes in patients with atherosclerotic See Comment page 727
cardiovascular disease and heart failure are unknown. We report a prespecified analysis of the effect of once-weekly Institute of Cardiovascular
subcutaneous semaglutide 2·4 mg on ischaemic and heart failure cardiovascular outcomes. We aimed to investigate Sciences, University College
if semaglutide was beneficial in patients with atherosclerotic cardiovascular disease with a history of heart failure London, London, UK
(Prof J Deanfield FRCP); Division
compared with placebo; if there was a difference in outcome in patients designated as having heart failure with of Cardiac Surgery, Li Ka Shing
preserved ejection fraction compared with heart failure with reduced ejection fraction; and if the efficacy and safety of Knowledge Institute of
semaglutide in patients with heart failure was related to baseline characteristics or subtype of heart failure. St Michael’s Hospital, Unity
Health Toronto, University of
Toronto, Toronto, ON, Canada
Methods The SELECT trial was a randomised, double-blind, multicentre, placebo-controlled, event-driven phase 3 (Prof S Verma MD PhD); Division
trial in 41 countries. Adults aged 45 years and older, with a BMI of 27 kg/m² or greater and established cardiovascular of Cardiovascular Medicine,
disease were eligible for the study. Patients were randomly assigned (1:1) with a block size of four using an interactive Brigham and Women’s
Hospital, Harvard Medical
web response system in a double-blind manner to escalating doses of once-weekly subcutaneous semaglutide over
School, Boston, MA, USA
16 weeks to a target dose of 2·4 mg, or placebo. In a prespecified analysis, we examined the effect of semaglutide (B M Scirica MD, J Plutzky MD);
compared with placebo in patients with and without a history of heart failure at enrolment, subclassified as heart Department of Medicine, VA
failure with preserved ejection fraction, heart failure with reduced ejection fraction, or unclassified heart failure. Puget Sound Health Care
System and University of
Endpoints comprised MACE (a composite of non-fatal myocardial infarction, non-fatal stroke, and cardiovascular
Washington, Seattle, WA, USA
death); a composite heart failure outcome (cardiovascular death or hospitalisation or urgent hospital visit for heart (Prof S E Kahn MB ChB);
failure); cardiovascular death; and all-cause death. The study is registered with ClinicalTrials.gov, NCT03574597. Department of Biostatistics,
University of Washington,
Seattle, WA, USA
Findings Between Oct 31, 2018, and March 31, 2021, 17 604 patients with a mean age of 61·6 years (SD 8·9) and a mean
(S S Emerson MD PhD);
BMI of 33·4 kg/m² (5·0) were randomly assigned to receive semaglutide (8803 [50·0%] patients) or placebo Pennington Biomedical
(8801 [50·0%] patients). 4286 (24·3%) of 17 604 patients had a history of investigator-defined heart failure at enrolment: Research Center, Baton Rouge,
2273 (53·0%) of 4286 patients had heart failure with preserved ejection fraction, 1347 (31·4%) had heart failure with LA, USA (Prof D Ryan MD);
Department of Internal
reduced ejection fraction, and 666 (15·5%) had unclassified heart failure. Baseline characteristics were similar between
Medicine/Endocrinology and
patients with and without heart failure. Patients with heart failure had a higher incidence of clinical events. Semaglutide Peter O’Donnell Jr School of
improved all outcome measures in patients with heart failure at random assignment compared with those without Public Health, UT
heart failure (hazard ratio [HR] 0·72, 95% CI 0·60–0·87 for MACE; 0·79, 0·64–0·98 for the heart failure composite Southwestern Medical Center,
Dallas, TX, USA
endpoint; 0·76, 0·59–0·97 for cardiovascular death; and 0·81, 0·66–1·00 for all-cause death; all pinteraction>0·19). (Prof I Lingvay MD MDSC);
Treatment with semaglutide resulted in improved outcomes in both the heart failure with reduced ejection fraction Institute of Genetics and
(HR 0·65, 95% CI 0·49–0·87 for MACE; 0·79, 0·58–1·08 for the composite heart failure endpoint) and heart failure Molecular Medicine, University
with preserved ejection fraction groups (0·69, 0·51–0·91 for MACE; 0·75, 0·52–1·07 for the composite heart failure of Edinburgh, Edinburgh, UK
(Prof H M Colhoun MD);
endpoint), although patients with heart failure with reduced ejection fraction had higher absolute event rates than Department of Cardiovascular
those with heart failure with preserved ejection fraction. For MACE and the heart failure composite, there were no Disease, Saint Luke’s Mid
significant differences in benefits across baseline age, sex, BMI, New York Heart Association status, and diuretic use. America Heart Institute and
Serious adverse events were less frequent with semaglutide versus placebo, regardless of heart failure subtype. University of Missouri-Kansas
City School of Medicine,
Kansas City, MO, USA
Interpretation In patients with atherosclerotic cardiovascular diease and overweight or obesity, treatment with (Prof M N Kosiborod MD); Novo
semaglutide 2·4 mg reduced MACE and composite heart failure endpoints compared with placebo in those with and Nordisk, Søborg, Denmark
without clinical heart failure, regardless of heart failure subtype. Our findings could facilitate prescribing and result (Prof G K Hovingh MD PhD,
S Hardt-Lindberg MD PhD,
in improved clinical outcomes for this patient group. O Frenkel MD,
P E Weeke MD PhD,
Funding Novo Nordisk. S Rasmussen PhD); Department
of Vascular Medicine,
Amsterdam University Medical
Copyright © 2024 The Author(s). Published by Elsevier Ltd. This is an Open Access article under the CC BY 4.0 Centers, Amsterdam,
license.

www.thelancet.com Vol 404 August 24, 2024 773

 Articles

Netherlands
(Prof G K Hovingh); Department Research in context
of Cardiology, Queen Giovanna
University Hospital, Evidence before this study 4286 patients with a history of heart failure at enrolment and
Sofia, Bulgaria Heart failure is more common in patients with overweight or showed, for the first time to our knowledge, that baseline
(Prof A Goudev MD DSc);
obesity than in the general population and is typically heart status does not alter the benefits of semaglutide on MACE,
Division of Molecular and
Clinical Medicine, School of failure with preserved ejection fraction, which has had few heart failure outcomes, cardiovascular mortality, and all-cause
Medicine, University of treatment options until recently. Semaglutide, a GLP-1 receptor mortality in patients with overweight or obesity without
Dundee, Dundee, UK agonist, has been shown to improve quality of life and exercise diabetes with established cardiovascular disease. We found no
(Prof C C Lang MD); Faculty of
performance in patients with heart failure with preserved ejection difference in the effect of semaglutide in patients with heart
Health Sciences, Universidad
Simón Bolívar, Barranquilla, fraction, with and without diabetes, in the STEP-HFpEF and STEP- failure with preserved ejection fraction or heart failure with
Colombia HFpEF DM trials, in association with weight loss. However, to our reduced ejection fraction, based on baseline characteristics
(M Urina-Triana MD PhD); Heart knowledge, no data have been published on cardiovascular (age, sex, and weight) or glycaemic status. Patients with heart
Center, Turku University
outcomes for major adverse cardiovascular events (MACE), heart failure with reduced ejection fraction, who had higher absolute
Hospital, University of Turku,
Turku, Finland failure complications, cardiovascular death, or all-cause death event rates than those with heart failure with preserved
(M Pietilä MD PhD); Department with semaglutide in patients with heart failure with preserved ejection fraction, had an equivalent reduction in all
of Cardiovascular Medicine, ejection fraction. The outlook for such patients is poor and is prespecified endpoints with semaglutide treatment, without
Cleveland Clinic Lerner College
of Medicine, Case Western
worse with overweight and obesity. In small studies, not limited increased serious adverse events and with reduced all-cause
Reserve University, Cleveland, to patients with heart failure with reduced ejection fraction and mortality.
OH, USA (A M Lincoff MD) obesity, it was suggested that treatment with a GLP-1 receptor
Implications of all the available evidence
Correspondence to: agonist, liraglutide, could result in harm. The safety profile of
Prof John Deanfield, Institute of Semaglutide reduced MACE, heart failure composite,
GLP-1 receptor agonists in patients with different clinical heart
Cardiovascular Sciences, cardiovascular death, and all-cause death in patients with
failure subtypes has not been reported in adequately powered
University College London, atherosclerotic cardiovascular disease and overweight or
London WC1E 6DD, UK studies.
obesity and heart failure, and can be administered safely
john.e.deanfield@gmail.com
Added value of this study regardless of heart failure subtype. Our findings could facilitate
In the SELECT trial, the largest study of GLP-1 receptor agonists prescribing and result in improved clinical outcomes for this
to date, we evaluated the effect of semaglutide in patient group.

Introduction to improve heart failure outcomes in both heart failure

The worldwide increase in the prevalence of obesity is with preserved ejection fraction and heart failure with
contributing to a rapid rise in cardiovascular disease and reduced ejection fraction in patients with or without
diabetes, with resulting stress on health-care systems.1 diabetes,7–9 there remains a considerable unmet clinical
Semaglutide, a glucagon-like peptide-1 (GLP-1) receptor need, especially in patients with overweight or obesity. In
agonist, first introduced to manage dysglycaemia, results the STEP-HFpEF trials, semaglutide 2·4 mg weekly
in both weight loss2 and reductions in major adverse reduced heart failure symptoms and improved exercise
cardiovascular events (MACE) in patients with diabetes.3 function in patients with obesity-related heart failure
The Semaglutide Effects on Heart Disease and Stroke in with preserved ejection fraction with and without
Patients with Overweight or Obesity (SELECT) trial diabetes, by targeting metabolic drivers associated with
showed that once-weekly, subcutaneous semaglutide obesity rather than myocardial loading or neurohumoral
2·4 mg reduced MACE by 20% compared with placebo mediators.10,11 However, the size of those clinical trials
in patients with pre-existing atherosclerotic cardio­ precluded the investigators from drawing conclusions
vascular disease and overweight or obesity (BMI about the effect of semaglutide on clinical heart failure
≥27 kg/m²), but who did not have diabetes.4 events, MACE, or mortality. The effect of semaglutide on
The increase in obesity in the general population has heart failure outcomes and MACE in patients with heart
also been associated with a rise in heart failure failure with reduced ejection fraction has not been
prevalence. A large proportion of patients with heart studied in dedicated clinical trials, and this is of clinical
failure and obesity have heart failure with preserved importance because there has been concern that some
ejection fraction, which is likely to be causally related to GLP-1 receptor agonists might be ineffective or
the pathophysiological consequences of obesity.5 Heart potentially harmful in this setting.12–16
failure with reduced ejection fraction and heart failure In SELECT, 4286 of the enrolled patients had a history
with preserved ejection fraction share many clinical of investigator-defined heart failure, categorised to be
features, but their cause and response to treatment are heart failure with preserved ejection fraction, heart
different, with benefits of traditional treatments for heart failure with reduced ejection fraction, or unclassified.
failure with reduced ejection fraction being less clear in We report a prespecified analysis of the effect of once-
patients with heart failure with preserved ejection weekly subcutaneous semaglutide 2·4 mg on ischaemic
fraction.5,6 Although SGLT2 inhibitors have been shown and heart failure cardiovascular outcomes. In these

774 www.thelancet.com Vol 404 August 24, 2024

 Articles

patients, we asked the following questions. First, was LVEF into three categories: less than 40%, 40–49%, and
semaglutide compared with placebo beneficial in 50% and greater (appendix p 7). The study is registered See Online for appendix
patients with atherosclerotic cardiovascular disease with with ClinicalTrials.gov, NCT03574597.
a history of heart failure? Second, was there a difference
in outcome in patients designated as having heart failure Randomisation and masking
with preserved ejection fraction or heart failure with Patients were randomly assigned (1:1) with a block
reduced ejection fraction? Finally, was the efficacy and size of four using an interactive web response system in a
safety of semaglutide in patients with heart failure double-blind manner to escalating doses of once-weekly
related to baseline characteristics or subtype of heart subcutaneous semaglutide over 16 weeks to a target dose
failure? of 2·4 mg, or placebo.4,17 The trial product (the pen device)
containing the semaglutide and the placebo was visually
Methods identical and was packed in a manner that maintained
Study design and patients masking. Investigators were allowed to reduce the study
The SELECT trial was a randomised, double-blind, product if there were tolerability issues. It was
multicentre, placebo-controlled, event-driven phase 3 recommended that patients were treated according to the
trial in 41 countries (804 sites), which evaluated whether evidence-based standard of care. The protocol and baseline
once-weekly subcutaneous semaglutide 2·4 mg, when characteristics have been published previously.4,17,18
given as an adjunct to standard of care recommendations
for the time period of intervention, was superior to Outcomes
placebo in reducing the risk of MACE in patients with In this prespecified analysis, our primary outcomes were
established cardiovascular disease and overweight or time from random assignment to first occurrence of
obesity, without a history of diabetes. The protocol for MACE (defined as a composite of cardiovascular death,
SELECT was approved by the institutional review board non-fatal myocardial infarction, or non-fatal stroke); a
and ethics committee at each participating centre. All heart failure composite (consisting of cardiovascular
patients provided written informed consent before any death or hospitalisation or urgent hospital visit for heart
trial-specific activity. Details of study design, population, failure); cardiovascular death; and all-cause death. All
and primary outcome have been reported previously.4,17,18 reported clinical events were adjudicated by an
Adults aged 45 years and older, with a BMI of 27 kg/m² independent committee, who were masked to trial group
or greater and established cardiovascular disease were assignment, in accordance with these prespecified
eligible for the study. Established cardiovascular disease criteria (appendix pp 2–3). Safety was assessed as the
was defined as at least one of: previous myocardial number and nature of serious adverse events in the trial
infarction, previous ischaemic or haemorrhagic stroke, or those leading to discontinuation of the trial product.
or symptomatic peripheral artery disease. Exclusion
criteria included previous myocardial infarction, stroke, Statistical analysis
hospitalisation for unstable angina pectoris, or a transient This event-driven trial was designed to provide 90%
ischaemic attack within 60 days of screening; glycated power to detect a relative risk reduction of 17% for a
haemoglobin (HbA1c) of 6·5% (48 mmol/mol) or greater; primary endpoint event in the semaglutide group
history of any form of diabetes; New York Heart compared with the placebo group (hazard ratio [HR] 0·83)
Association (NYHA) class IV heart failure; presence of at an overall one-sided significance level of 0·025. This
end-stage kidney disease; or need for chronic or design required that a minimum of 1225 primary
intermittent dialysis. Sex was investigator reported as endpoint events be accrued.
being either male or female. Statistical analyses were based on the intention-to-treat
The present, prespecified analysis explored the effects principle and included all randomly assigned patients
of semaglutide versus placebo in patients who were (with and without heart failure), irrespective of adherence
enrolled with an investigator-defined history of heart to semaglutide or placebo or changes to background
failure and stratified by heart failure subtype. Patients medications using in-trial data. All analyses were done
were classified as having heart failure with preserved using time to first event from random assignment. For
ejection fraction, heart failure with reduced ejection composite endpoints, the endpoint was said to occur
fraction, or unclassified heart failure at the time of when the first component of the composite took place.
enrolment. For inclusion in SELECT, investigators were For reporting of individual components of composites
asked to provide an NYHA class, and although (cardiovascular death, myocardial infarction, stroke, and
echocardiographic analyses were not required for hospitalisation or urgent hospital visit for heart failure)
inclusion, investigators were asked to provide key the reported rates were for the first event of the particular
parameters from the most recent echocardiogram (ECG; component, thus providing full accounting of the total
within 18 months), including left ventricular ejection number of cardiovascular deaths, first myocardial
fraction (LVEF). After the first 1783 patients were infarctions, first strokes, or first hospitalisation or urgent
recruited to the study, investigators were asked to define hospital visit for heart failure. In the analysis by heart

www.thelancet.com Vol 404 August 24, 2024 775

 Articles

failure subtype, only patients with a documented clinical investigator’s assessment, 2273 (53·0%) of 4286 patients
subtype of heart failure with preserved ejection fraction had heart failure with preserved ejection fraction,
or heart failure with reduced ejection fraction were 1347 (31·4%) had heart failure with reduced ejection
included. fraction, and 666 (15·5%) had unclassified heart failure.
Demographics and baseline characteristics were ECG data were available for 2159 (95·0%) of 2273 patients
summarised according to heart failure type at baseline (no with heart failure with preserved ejection fraction,
heart failure, heart failure, and subtype within heart 1296 (96·2%) of 1347 patients with heart failure with
failure: heart failure with preserved ejection fraction, heart reduced ejection fraction, and 293 (44·0%) of 666 patients
failure with reduced ejection fraction, and unclassified with unclassified heart failure. Data on LVEF for the
heart failure). Time from random assignment to an different heart failure subtypes are provided in the
outcome was analysed with a Cox proportional hazards appendix (p 7).
model with treatment group (semaglutide or placebo) as a The clinical characteristics of patients with a history of
fixed factor. The HRs with 95% CIs and two-sided p values heart failure (overall and by subtypes) and those without a
are presented. Patients who withdrew from the trial, died history of heart failure were well balanced across the
from causes not included in the endpoint, or were lost to treatment groups (table). In the heart failure subgroups, a
follow-up were censored at the time of withdrawal, death, higher proportion of the heart failure with preserved
or last contact with the treating physician. Subgroup ejection fraction group were female compared with the
analyses for time-to-event endpoints were based on the heart failure with reduced ejection fraction group. As
same Cox proportional hazards model by adding the expected, a higher proportion of the patients with heart
specific subgroup as a factor and an interaction between failure with reduced ejection fraction had previous
treatment group (semaglutide and placebo) and the myocardial infarction compared with those with heart
subgroup. Similarly, time-to-event outcomes were plotted failure with preserved ejection fraction, and the same
by treatment group and subgroup using the Aalen– pattern was observed for those with two or more inclusion
Johansen estimator and presented as cumulative criteria for cardiovascular disease. At enrolment,
incidences, considering non-cardiovascular death or all- 1371 (32·0%) of 4286 patients with heart failure were in
cause death as a competing event dependent on the NYHA class I, 2540 (59·3%) were in class II, and
outcome. 364 (8·5%) were in class III. Bodyweight, waist
Selected types of serious adverse events were circumference, and estimated glomerular filtration rate
summarised by the proportion of patients with an event by were similar in the heart failure with preserved ejection
system organ class and treatment discontinuation, using fraction and heart failure with reduced ejection fraction
in-trial data according to heart failure subtype. CIs were groups. The incidence of heart failure with preserved
not adjusted for multiplicity and should therefore not be ejection fraction (n=2273) was associated with increasing
used to infer definitive treatment effects. A two-sided BMI at enrolment, calculated as the proportion in the
significance level of 5% was considered significant. overall population with reported heart failure status at
Statistical analyses were done with SAS (version 9.4). baseline (n=17 600; BMI <30 kg/m², 546 (10·9%) of
An independent data monitoring committee reviewed 5023 patients; 30 kg/m² to <35 kg/m², 970 (13·0%) of
unblinded efficacy and safety data on an ongoing basis 7474 patients; 35 kg/m² to <40 kg/m², 471 (14·1%)
and at prespecific timepoints. of 3344 patients; 40 kg/m² to <45 kg/m², 190 (16·2%) of
1173 patients; ≥45 kg/m², 96 (16·4%) of 586 patients). This
Role of the funding source pattern was not seen in patients with heart failure with
The funder of the study was responsible, along with an reduced ejection fraction or unclassified heart failure (data
academic steering committee, for the study design, not shown). Mean systolic blood pressure was lower in
contributed to data collection, data analysis, and data patients with heart failure with reduced ejection fraction
interpretation, and participated in the review of the compared with patients with heart failure with preserved
manuscript in collaboration with the authors. ejection fraction.
There were expected differences in medications at
Results baseline, with a higher proportion of patients with heart
The baseline characteristics of the SELECT study failure with reduced ejection fraction receiving loop
population have been previously reported.4 Briefly, diuretics and aldosterone antagonists compared with
between Oct 31, 2018, and March 31, 2021, 17 604 patients those with heart failure with preserved ejection fraction
with a mean age of 61·6 years (SD 8·9) and a mean BMI (table). No patients were receiving an SGLT2 inhibitor at
of 33·4 kg/m² (5·0) were randomly assigned to receive enrolment, but 545 (3·1%) of 17 604 patients started using
semaglutide (8803 [50·0%] patients) or placebo an SGLT2 inhibitor during the study. Plasma lipid levels
(8801 [50·0%] patients). 12 732 (72·3%) of and blood pressure were well treated in all groups at trial
17 604 participants were male and 4872 (27·7%) were entry.
female. 4286 (24·3%) of 17 604 patients had a history of There were some minor differences between patients
heart failure at enrolment. Based on the treating with heart failure with preserved ejection fraction and

776 www.thelancet.com Vol 404 August 24, 2024

 Articles

Patients with heart Patients without heart Patients with heart Patients with heart Patients with
failure (n=4286) failure (n=13 314) failure with failure with unclassified heart
preserved ejection reduced ejection failure (n=666)
fraction (n=2273) fraction (n=1347)
Age, years 61·9 (8·7) 61·5 (8·9) 61·7 (8·7) 61·7 (8·7) 63·0 (8·7)
Sex
Male 3148 (73·4%) 9582 (72·0%) 1567 (68·9%) 1147 (85·2%) 434 (65·2%)
Female 1138 (26·6%) 3732 (28·0%) 706 (31·1%) 200 (14·8%) 232 (34·8%)
Race*
White 3841 (89·6%) 10 948 (82·2%) 2072 (91·2%) 1169 (86·8%) 600 (90·1%)
Asian 206 (4·8%) 1241 (9·3%) 106 (4·7%) 87 (6·5%) 13 (2·0%)
Black or African American 137 (3·2%) 533 (4·0%) 50 (2·2%) 48 (3·6%) 39 (5·9%)
Other† 91 (2·1%) 435 (3·3%) 44 (1·9%) 33 (2·4%) 14 (2·1%)
Ethnicity*
Hispanic or Latino 407 (9·5%) 1414 (10·6%) 195 (8·6%) 138 (10·2%) 74 (11·1%)
Region
North America 661 (15·4%) 3739 (28·1%) 259 (11·4%) 220 (16·3%) 182 (27·3%)
South America 335 (7·8%) 817 (6·1%) 166 (7·3%) 121 (9·0%) 48 (7·2%)
Europe 1256 (29·3%) 5434 (40·8%) 540 (23·8%) 544 (40·4%) 172 (25·8%)
Africa 108 (2·5%) 736 (5·5%) 26 (1·1%) 50 (3·7%) 32 (4·8%)
Asia 317 (7·4%) 1884 (14·2%) 146 (6·4%) 146 (10·8%) 25 (3·8%)
Other 1609 (37·5%) 704 (5·3%) 1136 (50·0%) 266 (19·7%) 207 (31·1%)
Cardiovascular inclusion criteria
Myocardial infarction only 2959/4170 (71·0%) 8946/13 083 (68·4%) 1552/2209 (70·3%) 1044/1319 (79·2%) 363/642 (56·5%)
Stroke only 563/4170 (13·5%) 2571 (19·7%) 347/2209 (15·7%) 63/1319 (4·8%) 153/642 (23·8%)
Peripheral artery disease only 108/4170 (2·6%) 669 (5·1%) 69/2209 (3·1%) 13/1319 (1·0%) 26/642 (4·0%)
≥2 inclusion criteria 540/4170 (12·9%) 897 (6·9%) 241/2209 (10·9%) 199/1319 (15·1%) 100/642 (15·6%)
New York Heart Association class
Class I 1371 (32·0%) ·· 749 (33·0%) 371 (27·5%) 251 (37·7%)
Class II 2540 (59·3%) ·· 1342 (59·0%) 840 (62·4%) 358 (53·8%)
Class III 364 (8·5%) ·· 181 (8·0%) 134 (9·9%) 49 (7·4%)
Unknown 11 (0·3%) ·· 1 (<0·1%) 2 (0·1%) 8 (1·2%)
Smoking status
Current smoker 758 (17·7%) 2192 (16·5%) 404 (17·8%) 265 (19·7%) 89 (13·4%)
Never smoked 1542 (36·0%) 4577 (34·4%) 917 (40·3%) 348 (25·8%) 277 (41·6%)
Previous smoker 1986 (46·3%) 6544 (49·2%) 952 (41·9%) 734 (54·5%) 300 (45·0%)
Concomitant medication
β blockers 3573 (83·4%) 8782 (66·0%) 1868 (82·2%) 1211 (89·9%) 494 (74·2%)
Angiotensin-converting-enzyme 2123 (49·5%) 5805 (43·6%) 1126 (49·5%) 681 (50·6%) 316 (47·4%)
inhibitors
Angiotensin receptor blockers 1408 (32·9%) 3778 (28·4%) 765 (33·7%) 432 (32·1%) 211 (31·7%)
Thiazides 381 (8·9%) 1645 (12·4%) 218 (9·6%) 87 (6·5%) 76 (11·4%)
Loop diuretics 1275 (29·7%) 933 (7·0%) 513 (22·6%) 586 (43·5%) 176 (26·4%)
Aldosterone antagonists 1171 (27·3%) 648 (4·9%) 404 (17·8%) 642 (47·7%) 125 (18·8%)
Thiazide-like diuretics 409 (9·5%) 627 (4·7%) 287 (12·6%) 54 (4·0%) 68 (10·2%)
Other potassium-sparing diuretics 6 (0·1%) 37 (0·3%) 2 (0·1%) 3 (0·2%) 1 (0·2%)
Angiotensin receptor-neprilysin 217 (5·1%) 48 (0·4%) 28 (1·2%) 168 (12·5%) 21 (3·2%)
inhibitor
Bodyweight, kg 98·5 (18·7) 96·1 (17·3) 97·9 (18·5) 98·7 (18·0) 100·3 (20·4)
BMI, kg/m²
<30 1059 (24·7%) 3964 (29·8%) 546 (24·0%) 366 (27·2%) 147 (22·1%)
≥30 to <35 1781 (41·6%) 5693 (42·8%) 970 (42·7%) 571 (42·4%) 240 (36·0%)
≥35 to <40 915 (21·3%) 2429 (18·2%) 471 (20·7%) 282 (20·9%) 162 (24·3%)
≥40 to <45 354 (8·3%) 819 (6·2%) 190 (8·4%) 95 (7·1%) 69 (10·4%)
≥45 177 (4·1%) 409 (3·1%) 96 (4·2%) 33 (2·4%) 48 (7·2%)
(Table continues on next page)

www.thelancet.com Vol 404 August 24, 2024 777

 Articles

Patients with heart Patients without heart Patients with heart Patients with heart Patients with
failure (n=4286) failure (n=13 314) failure with failure with unclassified heart
preserved ejection reduced ejection failure (n=666)
fraction (N=2273) fraction (n=1347)
(Continued from previous page)
BMI, kg/m² 33·9 (5·3) 33·1 (4·9) 34·0 (5·4) 33·4 (4·9) 34·9 (6·0)
BMI, kg/m² 32·7 (30·0–36·6) 32·0 (29·6–35·5) 32·8 (30·1–36·7) 32·3 (29·8–35·8) 33·6 (30·3–37·8)
Waist circumference, cm 112·4 (13·5) 111·0 (13·0) 111·6 (13·5) 112·8 (12·9) 114·3 (14·4)
Systolic blood pressure, mm Hg 129·4 (15·0) 131·5 (15·5) 130·7 (13·6) 126·0 (15·8) 131·6 (16·5)
Diastolic blood pressure, mm Hg 78·7 (9·9) 79·5 (10·0) 79·4 (9·3) 77·3 (10·2) 79·4 (11·0)
Pulse, beats per min 69·0 (10·0) 68·7 (10·9) 69·0 (9·5) 68·6 (10·5) 70·0 (10·6)
Glycated haemoglobin, % 5·80 (0·33) 5·78 (0·34) 5·80 (0·33) 5·82 (0·32) 5·78 (0·34)
Median high-sensitivity C-reactive 2·0 (0·9–4·5) 1·8 (0·9–4·0) 2·0 (0·9–4·4) 1·9 (0·9–4·3) 2·4 (1·1–5·5)
protein, mg/L
Median lipid, mmol/L
Total cholesterol 4·1 (3·5–5·0) 3·9 (3·4–4·7) 4·2 (3·5–5·0) 3·9 (3·4–4·7) 4·3 (3·6–5·2)
HDL cholesterol 1·1 (1·0–1·3) 1·1 (1·0–1·4) 1·1 (1·0–1·3) 1·1 (0·9–1·3) 1·2 (1·0–1·4)
LDL cholesterol 2·1 (1·6–2·9) 2·0 (1·6–2·6) 2·2 (1·7–2·9) 2·0 (1·6–2·6) 2·3 (1·7–3·1)
Triglycerides 1·6 (1·2–2·2) 1·5 (1·1–2·1) 1·6 (1·2–2·2) 1·5 (1·2–2·1) 1·5 (1·1–2·2)
Renal—eGFR and ACR
eGFR, mL/min per 1·73m² 79·6 (18·3) 83·4 (17·0) 80·6 (17·9) 78·2 (18·8) 79·0 (18·7)
ACR, mg/g 92·1 (2772·6) 80·7 (3262·1) 120·4 (3793·1) 53·0 (264·0) 74·2 (424·9)
ACR, mg/g 7·7 (4·5–17·0) 7·3 (4·5–15·0) 7·2 (4·3–15·0) 8·2 (4·6–18·8) 8·6 (4·9–18·5)
Data are mean (SD), n (%), n/N (%), or median (IQR). Four patients had unknown heart failure status at baseline. For the overall population, race was not reported for 95
(1·1%) patients in the semaglutide group and 74 (0·8%) patients in the placebo group. Information on whether patients identified as Hispanic or Latino was not reported for
95 (1·1%) patients in the semaglutide group and 76 (0·9%) patients in the placebo group. ACR=albumin–creatinine ratio. eGFR=estimated glomerular filtration rate. *Race
and ethnicity were self-reported by patients. †The category Other includes patients who reported their race as American Indian or Alaska Native, Native Hawaiian or Pacific
Islander, or other.

Table: Baseline characteristics of the SELECT cohort by heart failure status at enrolment

patients with heart failure with reduced ejection fraction failure at enrolment to a similar degree (figure 1).
and those with unclassified heart failure (table). For Analysis of the effect of semaglutide on the individual
example, patients with unclassified heart failure were components of MACE and the heart failure composite
slightly older and more likely to be female, had higher was not prespecified but the results are provided in the
high-sensitivity C-reactive protein levels, and had lower appendix (pp 8–9). Patients with heart failure at
NYHA class. Fewer patients with unclassified heart failure enrolment had a higher incidence of cardiovascular
were treated with β blockers at baseline compared with death and hospitalisation or urgent hospital visit for
patients with heart failure with preserved ejection fraction heart failure, and both components were reduced by
and patients with heart failure with reduced ejection semaglutide treatment compared with placebo.
fraction. Within the heart failure population at baseline, in those
Patients with heart failure at baseline who were treated treated with placebo, patients with heart failure with
with placebo had a higher rate of MACE, heart failure reduced ejection fraction had a higher rate of MACE,
composite, cardiovascular death, and all-cause death, heart failure composite, cardiovascular death, and all-
than those without heart failure who were treated with cause death than those with heart failure with preserved
placebo (figure 1). Semaglutide improved all outcome ejection fraction (figure 2; appendix pp 10–11). Treatment
measures in patients with heart failure at random with semaglutide resulted in improved outcomes in both
assignment compared with those without heart failure the heart failure with reduced ejection fraction (HR 0·65,
(HR 0·72, 95% CI 0·60–0·87 for MACE; 0·79, 0·64–0·98 95% CI 0·49–0·87 for MACE) and heart failure with
for the heart failure composite endpoint; 0·76, 0·59–0·97 preserved ejection fraction groups (0·69, 0·51–0·91 for
for cardiovascular death; and 0·81, 0·66–1·00 for all- MACE), with no indication of treatment heterogeneity in
cause death). Differences in event rates between those these outcomes (figure 2; appendix p 13). Over half of the
treated with semaglutide versus placebo emerged within observed composite heart failure events in each treatment
6 months and continued to expand over a mean group for each subgroup were ascribed to cardiovascular
follow-up of 39·8 months. Semaglutide also improved death before any hospitalisation or urgent hospital visit
MACE (0·84, 0·74–0·97) and all-cause mortality for heart failure (36 [67·9%] of 53 events for the
(0·81, 0·67–0·97) in patients without a history of heart semaglutide group and 39 [59·1%] of 66 events for the

778 www.thelancet.com Vol 404 August 24, 2024

 Articles

Placebo—heart failure Placebo—no heart failure

Semaglutide 2·4 mg—heart failure Semaglutide 2·4 mg—no heart failure
A Semaglutide 2·4 mg HR Interaction B Semaglutide 2·4 mg HR Interaction
vs placebo (95% CI) p value vs placebo (95% CI) p value
Heart failure 0·72 (0·60–0·87) Heart failure 0·79 (0·64–0·98)
16 0·19 0·64
No heart failure 0·84 (0·74–0·97) No heart failure 0·85 (0·68–1·06)

14

12
Proportion of patients (%)

10

0
0 6 12 18 24 30 36 42 48 0 6 12 18 24 30 36 42 48
Number at risk
Semaglutide 2·4 mg—heart failure 2155 2126 2085 2039 1996 1739 1370 970 421 2155 2133 2100 2068 2028 1772 1402 1008 444
Semaglutide 2·4 mg—no heart failure 6647 6568 6475 6387 6257 5489 4406 3155 1312 6647 6606 6553 6488 6396 5636 4541 3268 1371
Placebo—heart failure 2131 2082 2022 1978 1925 1651 1308 912 371 2131 2084 2038 2002 1968 1707 1361 954 399
Placebo—no heart failure 6667 6568 6463 6346 6237 5448 4351 3102 1301 6667 6625 6561 6481 6411 5632 4522 3243 1367

C Semaglutide 2·4 mg HR Interaction D Semaglutide 2·4 mg HR Interaction

vs placebo (95% CI) p value vs placebo (95% CI) p value
Heart failure 0·76 (0·59–0·97) Heart failure 0·81 (0·66–1·00)
16 0·24 0·98
No heart failure 0·93 (0·72–1·21) No heart failure 0·81 (0·67–0·97)

14

12
Proportion of patients (%)

10

0
0 6 12 18 24 30 36 42 48 0 6 12 18 24 30 36 42 48
Time since randomisation (months) Time since randomisation (months)
Number at risk
Semaglutide 2·4 mg—heart failure 2155 2141 2114 2083 2047 1796 1427 1029 451 2155 2141 2114 2083 2047 1796 1427 1029 451
Semaglutide 2·4 mg—no heart failure 6647 6606 6558 6500 6417 5655 4560 3285 1380 6647 6606 6558 6500 6417 5655 4560 3285 1380
Placebo—heart failure 2131 2101 2061 2029 1998 1736 1390 983 413 2131 2101 2061 2029 1998 1736 1390 983 413
Placebo—no heart failure 6667 6630 6571 6497 6430 5657 4546 3266 1380 6667 6630 6571 6497 6430 5657 4546 3266 1380

Figure 1: The effect of semaglutide versus placebo according to presence or absence of heart failure at enrolment
Cumulative incidence curves comparing the risk of major adverse cardiovascular events (A), heart failure composite (B), cardiovascular death (C), and all-cause
death (D) comparing semaglutide with placebo according to presence or absence of heart failure. The cumulative incidence rate is calculated using the Aalen–
Johansen method. HR=hazard ratio.

placebo group in patients with heart failure with 40% were compared with those with an LVEF of 40% or
preserved ejection fraction, and 37 [50·7%] of 73 events more (appendix p 16).
for the semaglutide group and 61 [63·5%] of 96 events Treatment outcomes for MACE (figure 3) and the heart
for the placebo group in patients with heart failure with failure composite (figure 4) are shown by baseline patient
reduced ejection fraction; appendix pp 8–9). Sensitivity characteristics. For MACE and the heart failure
analyses were done using LVEF in patients with ECGs. composite, there were no significant differences in
When patients with heart failure with an LVEF of less benefits across baseline age, sex, BMI, NYHA status, and
than 50% were compared with those with an LVEF of diuretic use. There were similar findings for the other
50% or more, the results were consistent with those adiposity measures (data not shown). There was no
based on prespecified investigator-defined heart failure difference in the treatment effect of semaglutide versus
subtype (appendix p 14). Similar results were found when placebo among patients with HbA1c less than 5·7%
patients with heart failure with an LVEF of less than (normoglycaemia) and those with levels of 5·7–6·5%

www.thelancet.com Vol 404 August 24, 2024 779

 Articles

Placebo—heart failure with reduced ejection fraction Placebo—heart failure with preserved ejection fraction
Semaglutide 2·4 mg—heart failure with reduced ejection Semaglutide 2·4 mg—heart failure with preserved ejection
fraction fraction
A Semaglutide 2·4 mg HR Interaction B Semaglutide 2·4 mg HR Interaction
vs placebo (95% CI) p value vs placebo (95% CI) p value
Heart failure with reduced 0·65 (0·49–0·87) Heart failure with reduced 0·79 (0·58–1·08)
ejection fraction 0·82 ejection fraction 0·79
Heart failure with preserved 0·69 (0·51–0·91) Heart failure with preserved 0·75 (0·52–1·07)
ejection fraction ejection fraction

20
18
16
Proportion of patients (%)

14
12
10
8
6
4
2
0
0 6 12 18 24 30 36 42 48 0 6 12 18 24 30 36 42 48
Number at risk
Semaglutide 2·4 mg—heart failure with preserved ejection fraction 1174 1161 1146 1130 1107 962 779 547 244 1174 1166 1157 1148 1129 987 798 570 258
Semaglutide 2·4 mg—heart failure with reduced ejection fraction 654 641 625 603 592 516 383 273 109 654 644 626 609 597 522 390 282 114
Placebo—heart failure with preserved ejection fraction 1099 1082 1060 1048 1024 887 716 497 214 1099 1080 1066 1056 1041 915 749 528 233
Placebo—heart failure with reduced ejection fraction 693 671 641 619 598 504 395 292 105 693 676 651 635 621 527 412 300 111

C Semaglutide 2·4 mg HR Interaction D Semaglutide 2·4 mg HR Interaction

vs placebo (95% CI) p value vs placebo (95% CI) p value
Heart failure with reduced 0·63 (0·43–0·91) Heart failure with reduced 0·72 (0·53–0·99)
ejection fraction 0·27 ejection fraction 0·56
20
Heart failure with preserved 0·87 (0·56–1·34) Heart failure with preserved 0·83 (0·59–1·16)
18 ejection fraction ejection fraction

16
Proportion of patients (%)

14
12
10
8
6
4
2
0
0 6 12 18 24 30 36 42 48 0 6 12 18 24 30 36 42 48
Time since randomisation (months) Time since randomisation (months)
Number at risk
Semaglutide 2·4 mg—heart failure with preserved ejection fraction 1174 1168 1160 1151 1133 992 806 575 261 1174 1168 1160 1151 1133 992 806 575 261
Semaglutide 2·4 mg—heart failure with reduced ejection fraction 654 648 635 620 612 537 404 296 117 654 648 635 620 612 537 404 296 117
Placebo—heart failure with preserved ejection fraction 1099 1089 1075 1068 1056 927 758 538 240 1099 1089 1075 1068 1056 927 758 538 240
Placebo–—heart failure with reduced ejection fraction 693 682 662 647 634 542 430 317 118 693 682 662 647 634 542 430 317 118

Figure 2: The effect of semaglutide versus placebo according to heart failure subtype
Cumulative incidence curves comparing the risk of major adverse cardiovascular events (A), heart failure composite (B), cardiovascular death (C), and all-cause death (D) comparing semaglutide with
placebo according to heart failure subtype. The cumulative incidence rate is calculated using the Aalen–Johansen method. HR=hazard ratio.

(prediabetes).19 There was concordance of treatment group and 101 [14·6%] of 693 in the placebo group) and
effects across the predefined subgroups of patients. 79 (3·5%) of 2273 patients with heart failure with
The use of SGLT2 inhibitors as standard of care was preserved ejection fraction (32 [2·7%] of 1174 patients in
not indicated when SELECT was initiated in 2018; this the semaglutide group and 47 [4·3%] of 1099 patients in
accounts for the absence of patients taking an SGLT2 the placebo group) initiated SGLT2 inhibition therapy
inhibitor at the time of enrolment. However, 179 (13·3%) during the study.
of 1347 patients with heart failure with reduced ejection Serious adverse events occurred less frequently in the
fraction (78 [11·9%] of 654 patients in the semaglutide semaglutide groups compared with the placebo groups

780 www.thelancet.com Vol 404 August 24, 2024

 Articles

HR (95% CI) Interaction

p value

Sex
Heart failure
Male 0·71 (0·58–0·88) 0·69
Female 0·78 (0·52–1·19)
Heart failure with preserved ejection fraction
Male 0·74 (0·54–1·03) 0·35
Female 0·54 (0·30–0·98)
Heart failure with reduced ejection fraction
Male 0·62 (0·46–0·84) 0·27
Female 0·98 (0·45–2·11)
Age, years
Heart failure
<55 0·72 (0·47–1·10) 0·96
55 to <65 0·69 (0·51–0·93)
65 to <75 0·78 (0·56–1·07)
≥75 0·70 (0·40–1·23)
Heart failure with preserved ejection fraction
<55 0·79 (0·39–1·60) 0·43
55 to <65 0·72 (0·45–1·16)
65 to <75 0·76 (0·47–1·22)
≥75 0·34 (0·14–0·83)
Heart failure with reduced ejection fraction
<55 0·46 (0·23–0·89) 0·63
55 to <65 0·68 (0·44–1·06)
65 to <75 0·69 (0·41–1·15)
≥75 0·91 (0·38–2·18)
BMI, kg/m2
Heart failure
<30 0·79 (0·54–1·14) 0·49
30 to <35 0·59 (0·44–0·79)
35 to <40 0·86 (0·59–1·26)
0 to <45 0·79 (0·40–1·56)
≥45 0·94 (0·42–2·13)
Heart failure with preserved ejection fraction
<30 0·70 (0·40–1·25) 0·78
30 to <35 0·71 (0·47–1·08)
35 to <40 0·80 (0·41–1·58)
40 to <45 0·38 (0·10–1·43)
≥45 0·35 (0·07–1·74)
Heart failure with reduced ejection fraction
<30 0·75 (0·43–1·32) 0·14
30 to <35 0·42 (0·25–0·69)
35 to <40 0·73 (0·43–1·23)
40 to <45 1·56 (0·55–4·46)
≥45 1·06 (0·27–4·25)
Glycated haemoglobin
Heart failure
<5·7 0·87 (0·62–1·21) 0·21
≥5·7 0·67 (0·54–0·84)
Heart failure with preserved ejection fraction
<5·7 0·80 (0·49–1·31) 0·46
≥5·7 0·64 (0·45–0·91)
Heart failure with reduced ejection fraction
<5·7 0·76 (0·43–1·32) 0·55
≥5·7 0·62 (0·45–0·86)
New York Heart Association
Heart failure
Class I 0·73 (0·51–1·04) 0·90
Class II 0·74 (0·58–0·94)
Class III 0·84 (0·49–1·42)
Heart failure with preserved ejection fraction
Class I 0·57 (0·35–0·92) 0·24
Class II 0·86 (0·58–1·28)
Class III 0·42 (0·15–1·14)
Heart failure with reduced ejection fraction
Class I 0·99 (0·50–1·95) 0·10
Class II 0·55 (0·39–0·77)
Class III 1·13 (0·53–2·41)

0·0625 0·125 0·25 0·5 1·0 2·0 4·0 8·0

Favours semaglutide Favours placebo

Figure 3: MACE outcome according to heart failure and heart failure subtypes
Forest plot showing HRs for MACE by baseline characteristics in patients with heart failure and heart failure subtypes. HR=hazard ratio. MACE=major adverse
cardiovascular events.

www.thelancet.com Vol 404 August 24, 2024 781

 Articles

HR (95% CI) Interaction

p value

Sex
Heart failure
Male 0·83 (0·65–1·05) 0·41
Female 0·66 (0·42–1·06)
Heart failure with preserved ejection fraction
Male 0·89 (0.58–1·38) 0·14
Female 0·49 (0.25–0·97)
Heart failure with reduced ejection fraction
Male 0·77 (0·56–1·07) 0·60
Female 0·98 (0·43–2·27)
Age, years
Heart failure
<55 0·80 (0·46–1·37) 0·81
55 to <65 0·74 (0·52–1·06)
65 to <75 0·89 (0·62–1·27)
≥75 0·65 (0·37–1·16)
Heart failure with preserved ejection fraction
<55 0·66 (0·25–1·77) 0·41
55 to <65 0·78 (0·42–1·45)
65 to <75 0·94 (0·53–1·65)
≥75 0·32 (0·10–1·02)
Heart failure with reduced ejection fraction
<55 0·57 (0·26–1·27) 0·79
55 to <65 0·78 (0·47–1·30)
65 to <75 0·93 (0·55–1·57)
≥75 0·73 (0·33–1·64)
BMI, kg/m2
Heart failure
<30 0·94 (0·61–1·44) 0·89
30 to <35 0·71 (0·49–1·03)
35 to <40 0·82 (0·54–1·25)
40 to <45 0·67 (0·33–1·37)
≥45 0·78 (0·34–1·78)
Heart failure with preserved ejection fraction
<30 0·88 (0·42–1·85) 0·53
30 to <35 1·03 (0·57–1·85)
35 to <40 0·50 (0·21–1·17)
40 to <45 0·45 (0·14–1·45)
≥45 0·52 (0·13–2·09)
Heart failure with reduced ejection fraction
<30 0·88 (0·49–1·58) 0·59
0 to <35 0·57 (0·33–0·98)
35 to <40 0·86 (0·48–1·52)
40 to <45 1·57 (0·45–5·41)
≥45 0·78 (0·18–3·50)
Glycated haemoglobin
Heart failure
<5·7 0·87 (0·60–1·27) 0·54
≥5·7 0·75 (0·58–0·98)
Heart failure with preserved ejection fraction
<5·7 0·70 (0·38–1·28) 0·77
≥5·7 0·78 (0·50–1·22)
Heart failure with reduced ejection fraction
<5·7 0·96 (0·54–1·73) 0·45
≥5·7 0·74 (0·52–1·06)
New York Heart Association
Heart failure
Class I 0·83 (0·52–1·32) 0·31
Class II 0·90 (0·69–1·17)
Class III 0·54 (0·29–0·99)
Heart failure with preserved ejection fraction
Class I 0·59 (0·30–1·17) 0·27
Class II 1·03 (0·63–1·69)
Class III 0·51 (0·20–1·31)
Heart failure with reduced ejection fraction
Class I 1·55 (0·66–3·61) 0·21
Class II 0·76 (0·53–1·08)
Class III 0·54 (0·21–1·42)

0·0625 0·125 0·25 0·5 1·0 2·0 4·0 8·0 16·0

Favours semaglutide Favours placebo

Figure 4: Heart failure composite outcome according to heart failure and heart failure subtypes
Forest plot showing HRs for heart failure composite by baseline characteristics in patients with heart failure and heart failure subtypes. HR=hazard ratio.

782 www.thelancet.com Vol 404 August 24, 2024

 Articles

for patients with and without heart failure (appendix p 12). and overweight or obesity but no diabetes, in terms of
A similar pattern was observed in patients with heart higher event rates versus patients without heart failure.
failure with reduced ejection fraction and heart failure The benefit we observed from semaglutide was seen in
with preserved ejection fraction. Cardiac disorders were addition to usual care recommendations during the
the most common serious adverse events and the period of the trial and could represent a new clinical
difference in the overall serious adverse event rates opportunity. In many countries, patients with clinical
between patients with and without heart failure was atherosclerotic cardiovascular disease often have either
primarily due to a difference in reported cardiac serious overweight or obesity, and the incidence is increasing.1
adverse events. There were no other consistent patterns Since the SELECT trial targeted a population with a BMI
of differences in adverse events of special interest of 27 kg/m² or greater and established cardiovascular
between treatment groups or in the heart failure disease without diabetes, the equal clinical benefit in
subgroups. In the semaglutide group, permanent those with heart failure and without heart failure,
discontinuation of treatment due to adverse events was independent of BMI at study entry, is reassuring.
primarily driven by gastrointestinal disorders and was Obesity has also changed the clinical spectrum of heart
higher than in placebo-treated patients (316 [14·7%] of failure, and has been linked to the rising proportion of
2155 patients vs 191 [9·0%] of 2131 patients in those with patients with heart failure with preserved ejection
heart failure; 1145 [17·2%] of 6647 patients vs 527 [7·9%] fraction, which is now the dominant clinical presentation
of 6667 patients in those without heart failure). The of heart failure.5 The treatment options for heart failure
discontinuation rate in patients receiving semaglutide with preserved ejection fraction have, until recently,
was lowest in patients with heart failure with preserved been few. Our study adds to the findings of the
ejection fraction (149 [12·7%] of 1174 patients) compared STEP-HFpEF trials, in which semaglutide improved
with patients with heart failure with reduced ejection heart failure-related symptoms, physical limitations, and
fraction (114 [17·4%] of 654 patients) or unclassified heart exercise function in patients with obesity-related heart
failure (53 [16·2%] of 327 patients). failure with preserved ejection fraction, with and without
diabetes, compared with placebo.10,11 However, these trials
Discussion were not powered to assess hard outcome endpoints of
In this prespecified analysis of the SELECT trial, we morbidity and mortality. In SELECT, we showed the
showed that semaglutide treatment reduced MACE and a benefits of semaglutide on MACE, a heart failure
heart failure composite endpoint, comprising cardio­ composite, and cardiovascular death in patients with
vascular death and hospitalisation or urgent hospital visit investigator-defined heart failure with preserved ejection
for heart failure, in over 4000 patients with atherosclerotic fraction.
cardiovascular disease and overweight or obesity but no The prognosis of patients with heart failure with
diabetes, who had a history of heart failure at enrolment. reduced ejection fraction is also worsened by coexisting
The improved outcomes were seen early after initiation of obesity.20 We showed, for the first time to our knowledge,
therapy and persisted over the trial period. The benefits that patients with atherosclerotic cardiovascular disease,
of semaglutide for MACE, heart failure composite, overweight or obesity, and heart failure with reduced
cardiovascular death, and all-cause death did not differ in ejection fraction (who had a greatest absolute risk) had
those with heart failure with preserved ejection fraction significant reductions in MACE with semaglutide, and
compared with those with heart failure with reduced the treatment effect on the heart failure composite was
ejection fraction. Patients with heart failure with reduced observed to be similar to that seen in the population with
ejection fraction have a higher cardiovascular absolute heart failure with preserved ejection fraction. This
risk than those with heart failure with preserved ejection finding contrasts with previous smaller studies, LIVE
fraction and previous reports suggested GLP-1 receptor and FIGHT, involving the GLP-1 receptor agonist
agonists could be ineffective or even harmful in such liraglutide, which were not limited to patients with
patients.12–14 Furthermore, clinical benefit with semaglutide obesity and were not adequately powered to assess
was independent of age, sex, baseline BMI, and clinical clinical outcomes.12–14
status. These efficacy findings, together with an acceptable Semaglutide treatment improved our prespecified
safety profile, support the use of semaglutide, in addition outcomes in patients with heart failure across a spectrum
to usual care, to reduce the risk of MACE in a broad of baseline patient characteristics including age, sex,
population of patients with established atherosclerotic adiposity measures, glycaemic control, lipids, and blood
cardiovascular disease and overweight or obesity, pressure. The mechanisms underlying MACE reduction
irrespective of their type of heart failure. and other possible cardiovascular benefits of GLP-1
Patients with obesity are at increased risk for receptor agnosists are likely to be complex. GLP-1
cardiovascular morbidity and mortality. The prevalence receptor agonists modulate multiple metabolic and
of heart failure in this population has been inflammatory pathways and have direct myocardial and
underestimated,5 and our study indicates a poor outlook vascular effects.21 Behavioural changes that occur with
for patients with atherosclerotic cardiovascular disease treatment, including exercise, diet composition, and

www.thelancet.com Vol 404 August 24, 2024 783

 Articles

eating patterns, might have also contributed.22 Our designed to examine the response by ethnicity and sex. In
findings, particularly the improved outcomes SELECT, the diagnosis of heart failure and its clinical
independent of heart failure subtype, should encourage subtype was made by investigators and based on medical
future research, with more detailed phenotyping to health records, without an explicit requirement for
characterise the complex pathophysiology of obesity- echocardiography and heart failure biomarker
related heart failure. The observation of a reduction in measurement. This approach reflects clinical practice but
all-cause mortality in all heart failure groups suggests the results in the characterisation of groups of patients with a
potential for other, as yet unknown, benefits. history of previous heart failure being less precise than in
The patient population in SELECT differs from those dedicated heart failure trials. Nevertheless, the
normally recruited to dedicated heart failure trials. investigator-defined heart failure group had substantially
Patients were enrolled based on atherosclerotic higher rates of MACE, heart failure composite, and all-
cardiovascular disease and overweight or obesity and were cause mortality compared with those without heart
not required to have symptomatic heart failure for failure. Furthermore, those classified as having a history
participation. Therefore, patients were less symptomatic, of heart failure with reduced ejection fraction had nearly
were taking fewer heart failure medications, and were at twice the rate of these events compared with those
lower risk than those in dedicated heart failure trials. Most classified as having heart failure with preserved ejection
patients were in NYHA class I or II (<10% in class III at fraction. ECGs were available for most patients with heart
enrolment) and patients with NYHA class IV were failure, and a time-to-event outcome analysis, based on
specifically excluded. These differences, plus other ejection fraction categories, was consistent with the
differences in trial methodology, hamper cross-trial results for the prespecified investigator-defined heart
comparison. failure groups. Patients recruited to SELECT had
In previous trials, SGLT2 inhibitors have been shown to established astherosclerotic cardiovascular disease with
result in a significant reduction in cardiovascular death one or more of previous myocardial infarction, stroke,
and heart failure events7–9 in patients with established heart and peripheral vascular disease. We showed that this
failure, with and without diabetes, independent of ejection population benefited from semaglutide, irrespective of
fraction. In SELECT, the improvement in the heart failure whether they had investigator-designated heart failure at
composite measure with semaglutide was mainly driven enrolment. Our study population differs from those
by reduced cardiovascular mortality. However, the observed recruited into dedicated heart failure trials and the
reduction in mortality by semaglutide attenuates our benefits should not be extrapolated to other populations
ability to discern a treatment effect on the endpoint of with heart failure. Further trials are warranted to evaluate
hospitalisation or urgent hospital visit for heart failure, in in more detail the effect on heart failure-related outcomes
part because mortality acts as a competing risk. We cannot and physiological and biomarker endpoints. Additionally,
be precise about what is driving the reduction in patients future research must define whether earlier preventive
designated as having cardiovascular death, which included treatment can favourably affect the trajectory to clinical
sudden cardiovascular death, as well as death from acute disease and heart failure. The number of incident
myocardial infarction and heart failure. Therefore, caution hospitalisation or urgent hospital visit for heart failure
should be applied when interpreting our findings, and the events in the patients without heart failure at enrolment
analyses of the components of the heart failure composite in our study was small. Therefore, although the current
were not prespecified. At the onset of recruitment into data provide evidence for cardiovascular efficacy of
SELECT, SGLT2 inhibitors were not yet part of standard of semaglutide in patients with atherosclerotic
care for heart failure. During the trial, SGLT2 inhibitors cardiovascular disease and prevalent heart failure, we
were initiated in only a small proportion of patients with were not able to determine whether semaglutide can
heart failure. Future studies will be needed to explore the reduce incident heart failure; this remains an important
effect of GLP-1 receptor agonists in combination with and unanswered question.
SGLT2 inhibitors, which appear to have different and In conclusion, our findings suggest that a large
potentially complementary modes of action and therefore population of patients with atherosclerotic cardiovascular
benefits, as suggested in the STEP HFpEF DM trial.11,23 disease who have overweight or obesity and heart failure
Additionally, combinations with other emerging drug could benefit from semaglutide, without the need for
therapies for heart failure might be beneficial.23,24 previous detailed cardiovascular risk stratification. The
Our study has both strengths and limitations. The large safety profile and discontinuation rates for semaglutide
number of patients with heart failure and long observation were similar in those with and without a history of heart
time allowed for robust assessment of the effect of failure and between heart failure subtypes, providing
semaglutide in terms of clinical outcomes, adverse additional reassurance regarding the favourable risk–
events, and durability of effects. SELECT was done in benefit balance of semaglutide in these patient groups.
41 countries and thus represents a diverse global In patients with atherosclerotic cardiovascular disease
population. Nevertheless, most participants were White with overweight or obesity but without diabetes,
men, and future GLP-1 receptor agonist trials should be treatment with once-weekly subcutaneous semaglutide

784 www.thelancet.com Vol 404 August 24, 2024

 Articles

2·4 mg compared with placebo reduced MACE, heart Amylyx, AstraZeneca, Avillion, Ayala, Bayer, BeiGene, Boehringer
failure composite, cardiovascular mortality, and all-cause Ingelheim, 89 Bio, BioAge, BioAtla, Bristol Myers Squibb, BridgeBio,
Daiichi Sankyo, Denovo, Fore Therapeutics, GlaxoSmithKline, Inovio,
mortality in patients with and without clinical heart Insmed, Ipsen, Karuna, Lilly, Lundbeck, Mirati, Moderna, Novartis,
failure, regardless of clinical heart failure subtype. Novavax, Novo Nordisk, National Surgical Adjuvant Breast and Bowel
Contributors Project, Pfizer, Principia, Reata, Rebiotx, Roche, Sanofi, Solvd, Sutro
JD designed the study, researched data, contributed to the discussion, Biopharma, and TG Therapeutics. AG declares having received
and wrote the manuscript. SR undertook the statistical analyses, honoraria from Novo Nordisk, AstraZeneca, Novartis, Boehringer
contributed to the discussion, and reviewed and edited the manuscript. Ingelheim, and Bayer. CCL declares having received consulting and
JD and SR directly accessed and verified the study data. BMS, SV, SEK, research honoraria from Amarin, Applied Therapeutics, AstraZeneca,
SSE, DR, IL, HMC, JP, MNK, GKH, OF, PEW, SH-L, and AML designed Boehringer Ingelheim, Bristol Myers Squibb, Moderna, Novartis, Novo
the study, researched data, and reviewed and edited the manuscript. Nordisk, Pfizer, and Roche Diagnostics. MU-T declares having received
SV, AG, CCL, MU-T, and MP recruited patients, contributed to the consulting and research honoraria from Abbott, AstraZeneca, Bayer,
discussion, and reviewed and edited the manuscript. All authors Boehringer Ingelheim, Bristol Myers Squibb, Frosst Laboratories,
approved submission of the final version of the manuscript and had Johnson and Johnson, Menarini, Novartis, Novo Nordisk, Pfizer,
access to the data. JD is the guarantor and takes responsibility for the Procaps, Sanofi-Aventis, Servier, and Tecnofarma. MP declares having
integrity of the data and the accuracy of the data analysis. received honoraria from Novo Nordisk. AML declares having received
honoraria from Novo Nordisk, Eli Lilly, Akebia, Amgen, Ardelyx, Becton
Declaration of interests Dickson, Endologix, Fibrogen, GlaxoSmithkline, Medtronic, Neovasc,
JD declares having received consulting honoraria from Amgen, Provention Bio, ReCor, Brainstorm Cell, Alnylam, and Intarcia for
Boehringer Ingelheim, Merck, Pfizer, Aegerion, Novartis, Sanofi, Takeda, consulting activities and research funding to his institution from
Novo Nordisk, and Bayer, and research grants from British Heart AbbVie, Esperion, AstraZeneca, CSL Behring, Novartis, and Eli Lilly.
Foundation, Medical Research Council (UK), National Institute for All other authors declare no competing interests.
Health and Care Research, Public Health England, MSD, Pfizer,
Aegerion, Colgate, and Roche. BMS declares having received Data sharing
institutional research grants to Brigham and Women’s Hospital from Data will be shared with bona fide researchers who submit a research
Better Therapeutics, Merck, Novo Nordisk, and Pfizer, and consulting proposal approved by an independent review board. Individual patient
fees from Allergan, Boehringer Ingelheim, Better Therapeutics, Elsevier data will be shared in datasets in a de-identified and anonymised format.
PracticeUpdate Cardiology, Esperion, Hanmi, Lexicon, Novo Nordisk, Information about data access request proposals can be found at
and equity in health at Scale and Doximity. SV reports speaking novonordisk-trials.com.
honoraria or consulting fees from Abbott, Amarin, AstraZeneca, Bayer, Acknowledgments
Boehringer Ingelheim, Canadian Medical and Surgical Knowledge The SELECT Trial Investigators recognise the contributions of all the
Translation Research Group, Eli Lilly, HLS Therapeutics, Janssen, study patients and clinical staff, without whose effort this research would
Merck, Novartis, Novo Nordisk, Pfizer, PhaseBio, and TIMI. DR declares not have been possible. Editorial support, including redrawing figures
having received consulting honoraria from Altimmune, Amgen, and formatting, was provided by Richard Ogilvy-Stewart of Apollo,
Biohaven, Boehringer Ingelheim, Calibrate, Carmot Therapeutics, OPEN Health Communications, and funded by Novo Nordisk,
CinRx, Eli Lilly, Epitomee, Gila Therapeutics, Ifa Celtic, Novo Nordisk, in accordance with Good Publication Practice guidelines.
Pfizer, Rhythm, Scientific Intake, Wondr Health, and Zealand. DR also
declares having received stock options from Calibrate, Epitomee, References
Scientific Intake, and Xeno Bioscience. IL declares having received 1 Powell-Wiley TM, Poirier P, Burke LE, et al. Obesity and
cardiovascular disease: a scientific statement from the American
research funding (paid to institution) from Novo Nordisk, Sanofi, Mylan,
Heart Association. Circulation 2021; 143: e984–1010.
and Boehringer Ingelheim. IL received advisory or consulting fees
2 Wilding JPH, Batterham RL, Calanna S, et al. Once-weekly
or other support from Altimmune, AstraZeneca, Bayer, Biomea,
semaglutide in adults with overweight or obesity. N Engl J Med 2021;
Boehringer Ingelheim, Carmot, Cytoki Pharma, Eli Lilly, Intercept, 384: 989–1002.
Janssen/Johnson & Johnson, Mannkind, Mediflix, Merck, Metsera, Novo
3 Marso SP, Bain SC, Consoli A, et al. Semaglutide and
Nordisk, Pharmaventures, Pfizer, Regeneron, Sanofi, Shionogi, cardiovascular outcomes in patients with type 2 diabetes.
Structure Therapeutics, Target RWE, Terns Pharma, The Comm Group, N Engl J Med 2016; 375: 1834–44.
Valeritas, WebMD, and Zealand Pharma. HMC declares being a 4 Lincoff AM, Brown-Frandsen K, Colhoun HM, et al. Semaglutide
stockholder and serving on an advisory panel for Bayer; receiving and cardiovascular outcomes in obesity without diabetes.
research grants from Chief Scientist Office, Diabetes UK, European N Engl J Med 2023; 389: 2221–32.
Commission, IQVIA, Juvenile Diabetes Research Foundation, and 5 Borlaug BA, Jensen MD, Kitzman DW, Lam CSP, Obokata M,
Medical Research Council; serving on an advisory board and speakers Rider OJ. Obesity and heart failure with preserved ejection fraction:
bureau for Novo Nordisk; and holding stock in Roche Pharmaceuticals. new insights and pathophysiological targets. Cardiovasc Res 2023;
SEK declares having received consulting honoraria from Anii 118: 3434–50.
Pharmaceuticals, Boehringer Ingelheim, Eli Lilly, Merck, Novo Nordisk, 6 Redfield MM, Borlaug BA. Heart failure with preserved ejection
and Oramed and stock options from Altpep. JP declares having received fraction: a review. JAMA 2023; 329: 827–38.
consulting honoraria from Altimmune, Amgen, Esperion Therapeutics, 7 Anker SD, Butler J, Filippatos G, et al. Empagliflozin in heart
Merck, MJH Life Sciences, Novartis, and Novo Nordisk; he has received failure with a preserved ejection fraction. N Engl J Med 2021;
a grant, paid to his institution, from Boehringer Ingelheim, and holds 385: 1451–61.
the position of Director, Preventive Cardiology, at Brigham and Women’s 8 McMurray JJV, Solomon SD, Inzucchi SE, et al. Dapagliflozin in
Hospital. MNK declares having served as a consultant or on an advisory patients with heart failure and reduced ejection fraction. N Engl J Med
board for 35Pharma, Alnylam, Amgen, Applied Therapeutics, 2019; 381: 1995–2008.
AstraZeneca, Bayer, Boehringer Ingelheim, Cytokinetics, Dexcom, Eli 9 Zannad F, Ferreira JP, Pocock SJ, et al. SGLT2 inhibitors in patients
Lilly, Esperion Therapeutics, Imbria Pharmaceuticals, Janssen, Lexicon with heart failure with reduced ejection fraction: a meta-analysis of
Pharmaceuticals, Merck (Diabetes and Cardiovascular), Novo Nordisk, the EMPEROR-Reduced and DAPA-HF trials. Lancet 2020;
396: 819–29.
Pharmacosmos, Pfizer, Sanofi, scPharmaceuticals, Structure
Therapeutics, Vifor Pharma, and Youngene Therapeutics; has received 10 Kosiborod MN, Abildstrøm SZ, Borlaug BA, et al. Semaglutide in
patients with heart failure with preserved ejection fraction and
research grants from AstraZeneca, Boehringer Ingelheim, and Pfizer;
obesity. N Engl J Med 2023; 389: 1069–84.
holds stocks in Artera Health and Saghmos Therapeutics; and has
11 Kosiborod MN, Petrie MC, Borlaug BA, et al. Semaglutide in
received honoraria from AstraZeneca, Boehringer Ingelheim, and Novo
patients with obesity-related heart failure and type 2 diabetes.
Nordisk. MNK has also received other research support from N Engl J Med 2024; 390: 1394–407.
AstraZeneca. SSE declares having received consulting honoraria from

www.thelancet.com Vol 404 August 24, 2024 785

 Articles

12 Jorsal A, Kistorp C, Holmager P, et al. Effect of liraglutide, 18 Ryan DH, Lingvay I, Colhoun HM, et al. Semaglutide effects on
a glucagon-like peptide-1 analogue, on left ventricular function in cardiovascular outcomes in people with overweight or obesity
stable chronic heart failure patients with and without diabetes (SELECT) rationale and design. Am Heart J 2020; 229: 61–69.
(LIVE)-a multicentre, double-blind, randomised, placebo-controlled 19 American Diabetes Association Professional Practice Committee. 2.
trial. Eur J Heart Fail 2017; 19: 69–77. Diagnosis and classification of diabetes: standards of care in
13 Margulies KB, Hernandez AF, Redfield MM, et al. Effects of diabetes—2024. Diabetes Care 2024; 47 (suppl 1): S20–42.
liraglutide on clinical stability among patients with advanced heart 20 Amdahl MB, Sundaram V, Reddy YNV. Obesity in heart failure with
failure and reduced ejection fraction: a randomized clinical trial. reduced ejection fraction: time to address the elephant in the room.
JAMA 2016; 316: 500–08. Cardiol Clin 2023; 41: 537–44.
14 Sharma A, Ambrosy AP, DeVore AD, et al. Liraglutide and weight 21 Drucker DJ. The cardiovascular biology of glucagon-like peptide-1.
loss among patients with advanced heart failure and a reduced Cell Metab 2016; 24: 15–30.
ejection fraction: insights from the FIGHT trial. ESC Heart Fail 22 Morgan Stanley. Could obesity drugs take a bite out of the food
2018; 5: 1035–43. industry? Sept 5, 2023. https://www.morganstanley.com/ideas/
15 Fudim M, White J, Pagidipati NJ, et al. Effect of once-weekly obesity-drugs-food-industry (accessed June 20, 2024).
exenatide in patients with type 2 diabetes mellitus with and without 23 Cheng AYY. Why choose between SGLT2 inhibitors and GLP1-RA
heart failure and heart failure-related outcomes: insights from the when you can use both?: the time to act is now. Circulation 2021;
EXSCEL trial. Circulation 2019; 140: 1613–22. 143: 780–82.
16 Khan MS, Fonarow GC, McGuire DK, et al. Glucagon-like peptide 1 24 Neuen BL, Heerspink HJL, Vart P, et al. Estimated lifetime
receptor agonists and heart failure: the need for further evidence cardiovascular, kidney, and mortality benefits of combination
generation and practice guidelines optimization. Circulation 2020; treatment with SGLT2 inhibitors, GLP-1 receptor agonists, and
142: 1205–18. nonsteroidal MRA compared with conventional care in patients
17 Lingvay I, Brown-Frandsen K, Colhoun HM, et al. Semaglutide for with type 2 diabetes and albuminuria. Circulation 2024; 149: 450–62.
cardiovascular event reduction in people with overweight or obesity:
SELECT study baseline characteristics. Obesity (Silver Spring) 2023;
31: 111–22.

786 www.thelancet.com Vol 404 August 24, 2024