European Journal of Heart Failure (2020) 22, 1247–1258 RESEARCH ARTICLE

doi:10.1002/ejhf.1867

Efficacy and safety of sodium–glucose

co-transporter 2 inhibition according to left
ventricular ejection fraction in DAPA-HF
Pooja Dewan1, Scott D. Solomon2, Pardeep S. Jhund1, Silvio E. Inzucchi3,
Lars Køber4, Mikhail N. Kosiborod5, Felipe A. Martinez6, Piotr Ponikowski7,
David L. DeMets8, Marc S. Sabatine9, Olof Bengtsson10, Mikaela Sjöstrand10,
Anna Maria Langkilde10, Inder S. Anand11, Jan Bělohlávek12, Vijay K. Chopra13,
Andrej Dukát14, Masafumi Kitakaze15, Béla Merkely16, Eileen O’Meara17,
Morten Schou18, Pham Nguyen Vinh19, and John J.V. McMurray1*, DAPA-HF
Investigators and Committees
1 BHF Cardiovascular Research Centre, University of Glasgow, Glasgow, UK; 2 Cardiovascular Medicine, Brigham and Women’s Hospital, Boston, MA, USA; 3 Section of
Endocrinology, Yale University School of Medicine, New Haven, CT, USA; 4 Rigshospitalet Copenhagen University Hospital, Copenhagen, Denmark; 5 Saint Luke’s Mid America
Heart Institute and University of Missouri-Kansas City, Kansas City, MO, USA; 6 National University of Cordoba, Cordoba, Argentina; 7 Wroclaw Medical University, Wroclaw,
Poland; 8 Department of Biostatistics & Medical Informatics, University of Wisconsin, Madison, WI, USA; 9 TIMI Study Group, Cardiovascular Division, Brigham and Women’s
Hospital and Harvard Medical School, Boston, MA, USA; 10 AstraZeneca R&D, Gothenburg, Sweden; 11 VA Medical Center, University of Minnesota, MN, USA; 12 2nd Department
of Internal Medicine, Cardiovascular Medicine, General University Hospital, Charles University in Prague, Czech Republic; 13 Department of Cardiology, Max Super Specialty
Hospital, New Delhi, India; 14 Department of Internal Medicine, Comenius University in Bratislava, Bratislava, Slovakia; 15 Cardiovascular Division of Medicine, National Cerebral
and Cardiovascular Center, Osaka, Japan; 16 Heart and Vascular Center, Semmelweis University, Budapest, Hungary; 17 Montreal Heart Institute and Université de Montreal,
Montreal, Canada; 18 Department of Clinical Medicine, Herlev-Gentofte Hospital, Herlev, Denmark; and 19 Department of Internal Medicine, Tan Tao University, Tan Duc city,
Vietnam

Received 31 March 2020; revised 22 April 2020; accepted 7 May 2020 ; online publish-ahead-of-print 15 June 2020

Aims The aim of this study was to examine whether left ventricular ejection fraction (LVEF) modified efficacy and safety
of dapagliflozin 10 mg compared with placebo in the 4744 patients with LVEF ≤40% randomized in the Dapagliflozin
And Prevention of Adverse-outcomes in Heart Failure trial (DAPA-HF).
.....................................................................................................................................................................
Methods We examined whether LVEF, analysed categorically or continuously, modified the effect of dapagliflozin. The primary
and results efficacy outcome was the composite of a worsening heart failure (HF) event (unplanned HF hospitalization/an urgent
HF visit requiring intravenous therapy) or cardiovascular death. Mean LVEF was 31.1% and LVEF categories analysed
were: <26% (n = 1143), 26–30% (n = 1018), 31–35% (n = 1187), and >35% (n = 1396). Each 5% decrease in LVEF
was associated with a higher risk of the primary outcome [hazard ratio (HR) 1.18; 95% confidence interval (CI)
1.13–1.24]. The benefit of dapagliflozin was consistent across the spectrum of LVEF: the dapagliflozin vs. placebo
HR was 0.75 (95% CI 0.59–0.95) for LVEF <26%, 0.75 (0.57–0.98) for LVEF 26–30%, 0.67 (0.51–0.89) for LVEF
31–35%, and 0.83 (0.63–1.09) for LVEF >35% (P for interaction = 0.762). Similarly, the effect of dapagliflozin on
the components of the primary endpoint was not modified by baseline LVEF (P for interaction for cardiovascular
death = 0.974, and for worsening HF = 0.161). Safety of dapagliflozin was also consistent across the range of LVEF
and neither efficacy nor safety were modified by diabetes status.

*Corresponding author. British Heart Foundation Cardiovascular Research Centre, University of Glasgow, 126 University Place, Glasgow, G12 8TA, UK. Tel: +44 141 3303479,
Fax: +44 141 3306955, Email: [email protected]

© 2020 The Authors. European Journal of Heart Failure published by John Wiley & Sons Ltd on behalf of European Society of Cardiology.
This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in
any medium, provided the original work is properly cited.
1248 P. Dewan et al.

Conclusion Left ventricular ejection fraction was a significant predictor of hospitalization and mortality in patients with HF with
reduced ejection fraction but did not modify the beneficial effect of dapagliflozin, overall or separately, in patients
with and without diabetes.
Clinical Trial Registration: ClinicalTrials.gov Identifier NCT03036124
..........................................................................................................
Keywords Heart failure • Dapagliflozin • Left ventricular ejection fraction

the previous 12 months were required to have LVEF measured at the

Introduction

...........................................................................................................................................
time of enrolment.
Left ventricular ejection fraction (LVEF) is the most commonly Key exclusion criteria included symptoms of hypotension or systolic
used measure of left ventricular systolic function. Not only does blood pressure (SBP) <95 mmHg, estimated glomerular filtration rate
it help diagnose heart failure (HF) with reduced ejection fraction (eGFR) <30 mL/min/1.73 m2 and type 1 diabetes.
(HFrEF), and distinguish between patients with HFrEF and HF with Patients were randomized to receive either dapagliflozin (10 mg
once daily) or matching placebo in a 1:1 ratio. Randomization was strat-
preserved ejection fraction (HFpEF), but it is also an important
ified based on either history of diabetes or on a glycated haemoglobin
predictor of morbidity and mortality.1,2 Both the risk of HF
level of ≥6.5% at enrolment (but for analyses, baseline diabetes was
hospitalization and cardiovascular mortality are higher in patients defined as a medical history of diabetes or a glycated haemoglobin level
with lower LVEF.1 of ≥6.5% at both the enrolment and randomization visits).
In the Dapagliflozin And Prevention of Adverse-outcomes in The median duration of follow-up was 18.2 months (minimum of
Heart Failure trial (DAPA-HF), 4744 patients with HF and a LVEF 5 days and maximum of 27.8 months).
≤40% were randomized to receive either the sodium–glucose
co-transporter 2 (SGLT2) inhibitor dapagliflozin or matching Outcomes
placebo.3 Patients allocated to dapagliflozin had a 26% lower risk of The primary outcome was a composite of a worsening HF event
the primary outcome of a worsening HF event (HF hospitalization (an unplanned hospitalization for HF or an urgent HF visit requiring
or an urgent HF visit requiring intravenous therapy) or cardiovas- intravenous therapy) or cardiovascular death.
cular death, compared with placebo. In the present report, we eval- The first secondary outcome was the composite of hospitalization
uated whether LVEF at baseline modified the effects of dapagliflozin for HF or cardiovascular death. Other secondary outcomes included
in the patients enrolled in DAPA-HF, overall and in participants with a composite of the total number of hospitalizations for HF (first and
and without diabetes separately. repeat) and cardiovascular death, and change from baseline to 8 months
in the total symptom score (TSS) of the Kansas City Cardiomyopathy
Questionnaire (KCCQ).5 The KCCQ is scored from 0 to 100 with
higher scores indicating better status and a change of ≥5 points is
Methods regarded as a clinically meaningful change. A further secondary end-
Patients and study design point was a composite of worsening renal function, including: (i) a sus-
tained decline in eGFR of ≥50%; (ii) end-stage renal disease – defined
The design and primary results of the DAPA-HF trial are published.3,4 as a sustained (≥28 day) eGFR of <15 mL/min/1.73 m2 , sustained dial-
The trial was approved by ethics committees at 410 participating cen- ysis or renal transplantation; or (iii) renal death. Lastly, death from
tres in 20 countries and all participants gave written informed consent. any cause was also analysed. Safety outcomes included serious adverse
Patients were eligible at screening if they were at least 18 years events, adverse events leading to treatment discontinuation and other
of age, were in New York Heart Association (NYHA) functional adverse events of special interest (adverse events related to volume
classes II to IV, had a LVEF ≤40%, and an elevated N-terminal pro depletion, renal adverse events, major hypoglycaemic episodes, bone
brain natriuretic peptide (NT-proBNP, ≥600 pg/mL or ≥400 pg/mL if fractures, diabetic ketoacidosis, amputations). Fournier’s gangrene and
hospitalized for HF within the previous 12 months). In patients with laboratory findings of note.
atrial fibrillation or atrial flutter on their baseline electrocardiogram,
NT-proBNP had to be ≥900 pg/mL, regardless of history of hospital-
ization for HF. Patients were required to receive standard HF drug and Statistical analysis
device therapy, including an angiotensin-converting enzyme inhibitor In this analysis, patients were divided into four LVEF categories, similar
(ACEI), an angiotensin receptor blocker (ARB) or an angiotensin to those used in prior analyses and reflective of clinical practice, namely:
receptor–neprilysin inhibitor (ARNI), along with a beta-blocker and (i) <26%; (ii) 26–30%; (iii) 31–35%; and (iv) >35%.1,6,7 Baseline char-
mineralocorticoid receptor antagonist (MRA), unless contraindicated acteristics are reported for each LVEF category as means ± standard
or not tolerated. Glucose-lowering therapy (including insulin) was con- deviation, median with interquartile range and proportions, as appro-
tinued in patients with diabetes, with adjustments made, as required, priate. A non-parametric Wilcoxon-type rank sum test and chi-square
during follow-up. tests were used for continuous and categorical variables, respectively.
Left ventricular ejection fraction was required to have been The effect of dapagliflozin, compared to placebo, on each out-
measured within 12 months of enrolment, by echocardiography, come across the different LVEF categories was examined using Cox
radionuclide ventriculography, contrast angiography, or cardiac mag- regression. Event rates per 100 person-years and hazard ratios (HRs)
netic resonance imaging. Patients without a LVEF measurement within adjusted for previous HF hospitalization (except for all-cause death and

© 2020 The Authors. European Journal of Heart Failure published by John Wiley & Sons Ltd on behalf of European Society of Cardiology.
Dapagliflozin effects by left ventricular ejection fraction 1249

replaced by baseline eGFR for the renal outcomes) and stratified by observed with an ACEI or ARB. These patterns were similar in

........................................................................................................................................................................
diabetes status are reported for each LVEF category. The proportional patients with and without diabetes and according to randomization
hazards assumption was fulfilled for all major outcomes. The relation- arm (online supplementary Table S1 and Table S4).
ship between LVEF as a continuous variable, and the risk of each major Among patients with diabetes at baseline, there was no signifi-
clinical outcome, was also examined in restricted cubic spline analyses. cant difference in the use of specific glucose-lowering medications
LVEF was modelled as a fractional polynomial to assess its interaction as
and insulin across the LVEF categories (Table 1 and online supple-
a continuous variable with treatment and displayed as a graph using the
mentary Table S1).
mfpi function in Stata.8 The interaction between LVEF and treatment
on change in KCCQ-TSS at 8 months was tested in a linear regression
model with interaction between LVEF and treatment tested for using Relationship between baseline left
the Wald method. The proportion of patients experiencing a 5-point
increase, and a 5-point decrease, in KCCQ-TSS at 8 months was exam- ventricular ejection fraction
ined in a logistic regression model, with the interaction term between and hospitalization and mortality
LVEF and treatment described using the Wald test. The HR per 5-point outcomes
decrease in baseline LVEF was calculated for the primary outcome and
its components, the composite outcome of cardiovascular death or The rate of the primary outcome in placebo-treated patients in
hospitalization for HF, hospitalization for HF and all-cause death, and the lowest LVEF category was 20.7 [95% confidence interval (CI)
was adjusted for treatment and previous HF hospitalization (except 17.7–24.1] per 100 patient-years, compared with 11.9 (9.9–14.3)
for all-cause death). All models were stratified by diabetes status as per 100 patient-years in patients in the highest LVEF category
specified. (Table 2). The corresponding rates of the primary outcome in
All analyses were conducted using Stata version 16 (Stata Corp., patients with diabetes in the lowest and highest LVEF categories
College Station, TX, USA). A P-value of <0.05 was considered statis- were 26.8 (95% CI 21.8–33.0) and 14.6 (95% CI 11.5–18.6)
tically significant. per 100 patient-years, respectively. In participants without dia-
betes these rates were 16.1 (95% CI 12.8–20.3) and 9.5 (95% CI
Results 7.1–12.6) per 100 patient-years, respectively (online supplemen-
tary Table S2 and Figure S4).
Left ventricular ejection fraction ranged from 2% to 40% (although As illustrated in Figure 1 and online supplementary Figure S3,
one patient had a LVEF of 45%). The mean and median LVEF the risk of the clinical outcomes of interest increased as LVEF
were 31.1 ± 6.8% and 32% (IQR 26–37%), respectively. There decreased. Table 3 shows that each 5-point decrease in LVEF was
were 1143 patients with a LVEF <26%, 1018 patients with a LVEF associated with an 18% higher risk of the primary outcome (HR
between 26% and 30%, 1187 with a LVEF between 31% and 35%, 1.18, 95% CI 1.13–1.24) in the overall cohort. Corresponding HR
and 1396 patients had a LVEF >35%. for a 5-point decrease in LVEF in participants with diabetes was
1.20 (95% CI 1.12–1.27) compared to 1.17 (95% CI 1.10–1.26)
Patient characteristics in patients without diabetes.
In the overall population, the increment in risk of cardiovascular
As shown in Table 1, patients with a lower LVEF were younger death was 20% per 5-point decrease in LVEF (HR 1.20, 95% CI
(mean 64 years in the lowest vs. 68 years in the highest LVEF 1.13–1.28) with a similar increment in risk for an episode of
category), more likely to be male, less likely to be from Europe worsening HF (HR 1.20, 95% CI 1.14–1.27). The HR for all-cause
or of white race, compared to patients with a higher LVEF. Fewer death was 1.13 (95% CI 1.07–1.20). The increase in HR per 5-point
patients with a lower LVEF had hypertension, diabetes, a previous decrease in LVEF for each of the latter three outcomes was similar
myocardial infarction, or atrial fibrillation. A higher proportion of in participants with and without diabetes (Table 3).
patients in the lowest LVEF category had a non-ischaemic aetiology Median time from measurement of LVEF to randomization was
and more had a previous hospitalization for HF. Conversely, there 48 days (Q1–Q3 14–130). A total of 3962 (84%) patients had
was no significant difference in median KCCQ-TSS score, or in the their LVEF measured within 6 months prior to randomization. The
proportion of patients in NYHA class II vs. III/IV, across the LVEF incremental increase in risk of clinical outcomes with decreasing
categories. Patients with a lower LVEF had a higher NT-proBNP LVEF was also consistent in both those who had LVEF measured
level (median 1827 pg/mL in the lowest vs. 1275 pg/mL in the ≤6 months prior to randomization and in those who had LVEF
highest LVEF category) and higher creatinine concentration. When measured >6 months prior to randomization (Table 3).
patients with and without diabetes were examined separately,
those with diabetes more often had a history of hypertension and
myocardial infarction (and an ischaemic aetiology), as well as worse Effect of dapagliflozin, compared
NYHA class, higher NT-proBNP and lower eGFR, compared to with placebo, on hospitalization
participants without diabetes, across the range of LVEF (online and mortality outcomes, according
supplementary Table S1).
to baseline left ventricular ejection
A greater proportion of patients with low LVEF were prescribed
diuretics. Use of sacubitril/valsartan, a MRA, digoxin, cardiac resyn- fraction
chronization therapy and an implantable cardioverter-defibrillator For each of the hospitalization and mortality outcomes examined,
increased with decreasing LVEF, whereas the opposite trend was the event rate was lower in patients receiving dapagliflozin, than in

© 2020 The Authors. European Journal of Heart Failure published by John Wiley & Sons Ltd on behalf of European Society of Cardiology.
1250 P. Dewan et al.

Table 1 Baseline characteristics according to left ventricular ejection fraction

<26% 26–30% 31–35% >35% P-value

(n = 1143) (n = 1018) (n = 1187) (n = 1396) for trend
...............................................................................................................................................................
LVEF (%) 22.4 ± 3.7 28.8 ± 1.4 33.7 ± 1.4 38.4 ± 1.4 <0.001
Age (years) 64.2 ± 11.3 66.0 ± 10.8 66.8 ± 10.6 68.1 ± 10.5 <0.001
Women 230 (20.1) 215 (21.1) 277 (23.3) 387 (27.7) <0.001
Region <0.001
Europe 406 (35.5) 398 (39.1) 556 (46.8) 794 (56.9)
Asia/Pacific 283 (24.8) 262 (25.7) 258 (21.7) 293 (21.0)
North America 241 (21.1) 140 (13.8) 171 (14.4) 125 (9.0)
Latin America 213 (18.6) 218 (21.4) 202 (17.0) 184 (13.2)
Race <0.001
White 728 (63.7) 695 (68.3) 857 (72.2) 1053 (75.4)
Black 104 (9.1) 43 (4.2) 48 (4.0) 31 (2.2)
Asian 288 (25.2) 266 (26.1) 263 (22.2) 299 (21.4)
Other 23 (2.0) 14 (1.4) 19 (1.6) 13 (0.9)
Heart rate (bpm) 72.6 ± 12.4 71.5 ± 11.6 70.9 ± 11.5 71.0 ± 11.3 0.001
SBP (mmHg) 116.5 ± 15.1 120.0 ± 15.3 123.4 ± 16.7 126.1 ± 16.2 <0.001
DBP (mmHg) 71.9 ± 10.1 72.9 ± 10.7 74.0 ± 10.7 74.8 ± 10.2 <0.001
BMI (kg/m2 ) 27.7 ± 6.4 27.8 ± 5.8 28.4 ± 5.9 28.6 ± 5.8 <0.001
Medical history
Hypertension 720 (63.0) 743 (73.0) 907 (76.4) 1153 (82.6) <0.001
Diabetes 453 (39.6) 432 (42.4) 485 (40.9) 613 (43.9) 0.062
Myocardial infarction 455 (39.8) 485 (47.6) 538 (45.3) 614 (44.0) 0.123
Atrial fibrillation 384 (33.6) 352 (34.6) 462 (38.9) 620 (44.4) <0.001
Stroke 103 (9.0) 104 (10.2) 107 (9.0) 152 (10.9) 0.210
COPD 137 (12.0) 111 (10.9) 143 (12.1) 194 (13.9) 0.088
Features of HF
HF aetiology <0.001
Ischaemic 548 (47.9) 575 (56.5) 703 (59.2) 848 (60.7)
Non-Ischaemic 493 (43.1) 373 (36.6) 393 (33.1) 428 (30.7)
Unknown 102 (8.9) 70 (6.9) 91 (7.7) 120 (8.6)
Prior HF hospitalization 577 (50.5) 486 (47.7) 548 (46.2) 640 (45.8) 0.016
KCCQ-TSS 77 [59–92] 79 [58–94] 79 [58–92] 76 [57–92] 0.265
NYHA class 0.995
II 754 (66.0) 712 (69.9) 805 (67.8) 932 (66.8)
III/IV 389 (34.0) 306 (30.1) 382 (32.2) 464 (33.2)
NT-proBNP (pg/mL) 1827 [1055–3385] 1551 [886–2806] 1317 [798–2353] 1275 [790–2232] <0.001
eGFR (mL/min/1.73 m2 ) 67.3 ± 19.9 64.8 ± 19.2 65.9 ± 19.7 65.2 ± 18.9 0.062
Creatinine (μmol/L) 105.1 ± 30.5 106.6 ± 31.8 104.3 ± 30.4 102.5 ± 29.2 0.006
Haemoglobin (g/L) 136.6 ± 15.9 135.7 ± 16.0 135.0 ± 16.2 135.0 ± 16.6 0.005
Treatment
Diuretic 1100 (96.2) 960 (94.3) 1098 (92.5) 1275 (91.3) <0.001
ACEI 590 (51.6) 582 (57.2) 655 (55.2) 834 (59.7) <0.001
ARB 283 (24.8) 269 (26.4) 329 (27.7) 426 (30.5) 0.001
ARNI 188 (16.4) 118 (11.6) 130 (11.0) 72 (5.2) <0.001
Any RAS blockera 1051 (92.0) 958 (94.1) 1109 (93.4) 1324 (94.8) 0.009
Beta-blocker 1100 (96.2) 979 (96.2) 1146 (96.5) 1333 (95.5) 0.403
MRA 855 (74.8) 755 (74.2) 841 (70.9) 919 (65.8) <0.001
Digoxin 265 (23.2) 207 (20.3) 193 (16.3) 222 (15.9) <0.001
Ivabradine 66 (5.8) 51 (5.0) 61 (5.1) 50 (3.6) 0.014
PCI 346 (30.3) 374 (36.7) 404 (34.0) 500 (35.8) 0.020
CABG 178 (15.6) 177 (17.4) 197 (16.6) 247 (17.7) 0.231
CRT 116 (10.1) 86 (8.4) 90 (7.6) 62 (4.4) <0.001
ICD 358 (31.3) 250 (24.6) 216 (18.2) 129 (9.2) <0.001
Diabetes medicationsb
Biguanide 230 (50.8) 221 (51.2) 261 (53.8) 304 (50.0) 0.828
DPP-4 inhibitor 68 (15.0) 67 (15.5) 76 (15.7) 99 (16.2) 0.614
GLP-1 analogues 7 (1.6) 5 (1.2) 4 (0.8) 5 (0.8) 0.225
Sulfonylurea 93 (20.5) 105 (24.3) 107 (22.1) 133 (21.7) 0.919
Insulin 112 (24.7) 122 (28.2) 144 (29.7) 162 (26.4) 0.554

Values are shown as mean ± standard deviation, n (%), or median [interquartile range].
ACEI, angiotensin-converting enzyme inhibitor; ARB, angiotensin receptor blocker; ARNI, angiotensin receptor–neprilysin inhibitor; BMI, body mass index; CABG, coronary artery bypass graft; COPD,
chronic obstructive pulmonary disease; CRT, cardiac resynchronization therapy; DBP, diastolic blood pressure; DPP, dipeptidyl peptidase; eGFR, estimated glomerular filtration rate; GLP, glucagon-like
peptide; HF, heart failure; ICD, implantable cardioverter-defibrillator; KCCQ-TSS, Kansas City Cardiomyopathy Questionnaire total symptom score; LVEF, left ventricular ejection fraction; MRA,
mineralocorticoid receptor antagonist; NT-proBNP, N-terminal pro brain natriuretic peptide; NYHA, New York Heart Association; PCI, percutaneous coronary intervention; RAS, renin–angiotensin
system; SBP, systolic blood pressure.
a Any patient on ACEI/ARB/ARNI.
b Only in patients with a medical history of diabetes (n = 1983).

© 2020 The Authors. European Journal of Heart Failure published by John Wiley & Sons Ltd on behalf of European Society of Cardiology.
 Table 2 Clinical outcomes according to left ventricular ejection fraction

Overall (n = 4744) <26% (n = 1143) 26–30% (n = 1018) 31–35% (n = 1187) >35% (n = 1396) P-value for
interaction
............................................. .......................................... .......................................... ............................................. ...............................................
Placebo Dapagliflozin Placebo Dapagliflozin Placebo Dapagliflozin Placebo Dapagliflozin Placebo Dapagliflozin
(n = 2371) (n = 2373) (n = 601) (n = 542) (n = 498) (n = 520) (n = 581) (n = 606) (n = 691) (n = 705)
.............................................................................................................................................................................................................................................................................
Primary composite outcome
Events, n (%) 502 (21.2) 386 (16.3) 161 (26.8) 110 (20.3) 114 (22.9) 94 (18.1) 113 (19.5) 84 (13.9) 114 (16.5) 98 (13.9)
Event rate per 100 pt-years 15.8 (14.5–17.2) 11.7 (10.6–13.0) 20.7 (17.7–24.1) 15.2 (12.7–18.4) 17.4 (14.5–20.9) 13.2 (10.8–16.2) 14.4 (12.0–17.3) 9.9 (8.0–12.3) 11.9 (9.9–14.3) 9.7 (8.0–11.8)
Unadjusted hazard ratio 0.74 (0.65–0.85) <0.001 0.75 (0.59–0.95) 0.75 (0.57–0.98) 0.67 (0.51–0.89) 0.83 (0.63–1.09) 0.762
Cardiovascular death
Events (%) 273 (11.5) 227 (9.6) 93 (15.5) 69 (12.7) 61 (12.3) 57 (11.0) 59 (10.2) 49 (8.1) 60 (8.7) 52 (7.4)
Event rate per 100 pt-years 8.0 (7.1–9.0) 6.6 (5.8–7.5) 11.0 (9.0–13.5) 9.0 (7.1–11.4) 8.7 (6.7–11.1) 7.7 (5.9–10.0) 7.0 (5.4–9.1) 5.6 (4.2–7.4) 5.9 (4.6–7.6) 4.9 (3.8–6.5)
Dapagliflozin effects by left ventricular ejection fraction

Unadjusted hazard ratio 0.82 (0.69–0.98) 0.030 0.84 (0.61–1.14) 0.88 (0.62–1.27) 0.77 (0.53–1.13) 0.85 (0.59–1.24) 0.974
HF hospitalization/urgent visita
Events (%) 326 (13.7) 237 (10.0) 104 (17.3) 70 (12.9) 80 (16.1) 51 (9.8) 76 (13.1) 51 (8.4) 66 (9.6) 65 (9.2)
Event rate per 100 pt-years 10.3 (9.2–11.4) 7.2 (6.3–8.2) 13.3 (11.0–16.2) 9.7 (7.7–12.3) 12.2 (9.8–15.2) 7.2 (5.4–9.4) 9.7 (7.7–12.1) 6.0 (4.6–7.9) 6.9 (5.4–8.8) 6.4 (5.0–8.2)
Unadjusted hazard ratio 0.70 (0.59–0.83) <0.001 0.74 (0.54–1.00) 0.57 (0.40–0.81) 0.61 (0.43–0.87) 0.95 (0.67–1.34) 0. 161
Total HF hospitalization/
CV death
Events 742 567 250 175 178 130 160 125 154 137
Event rate per 100 pt-years 21.9 (20.4–23.5) 16.5 (15.2–18.0) 29.8 (26.4–33.8) 23.0 (19.8–26.6) 25.3 (21.8–29.3) 17.5 (14.8–20.8) 19.1 (16.4–22.3) 14.3 (12.0–17.0) 15.3 (13.0–17.9) 13.0 (11.0–15.4)
Unadjusted hazard ratio 0.75 (0.65–0.88) <0.001 0.78 (0.59–1.03) 0.68 (0.51–0.92) 0.72 (0.53–1.00) 0.87 (0.64–1.18) 0.702
Mean ± SD change in
KCCQ-TSS at 8 months
Mean ± SD change at 8 months 3.3 ± 19.2 6.1 ± 18.6 3.2 ± 19.6 6.1 ± 19.8 2.0 ± 18.8 5.9 ± 19.0 3.3 ± 19.7 6.4 ± 17.4 4.3 ± 18.8 6.0 ± 18.6 0.607
Between treatment differenceb 2.8 (1.6–4.0) 2.9 (0.4–5.5) 3.9 (1.3–6.5) 3.1 (0.8–5.5) 1.7 (–0.5–3.8)
Proportion with increase in score ≥5 50.9 58.3 48.1 57.6 51.7 58.8 51.3 60.1 52.5 56.8 0.754
at 8 months
Proportion with decrease in score 32.9 25.3 35.3 29.1 34.5 24.9 32.9 22.2 29.6 25.5 0.734
≥5 at 8 months
All-cause death
Events (%) 329 (13.9) 276 (11.6) 100 (16.6) 77 (14.2) 72 (14.5) 68 (13.1) 75 (12.9) 59 (9.7) 82 (11.9) 72 (10.2)
Event rate per 100 pt-years 9.7 (8.7–10.8) 8.0 (7.1–9.0) 11.8 (9.7–14.4) 10.1 (8.1–12.6) 10.2 (8.1–12.9) 9.1 (7.2–11.6) 8.9 (7.1–11.2) 6.7 (5.2–8.7) 8.1 (6.5–10.0) 6.8 (5.4–8.6)
Unadjusted hazard ratio 0.83 (0.71–0.97) 0.022 0.87 (0.64–1.17) 0.89 (0.64–1.24) 0.73 (0.52–1.03) 0.86 (0.62–1.17) 0.866

CV, cardiovascular; HF, heart failure; KCCQ-TSS, Kansas City Cardiomyopathy Questionnaire total symptom score; pt, patient; SD, standard deviation.
Hazard ratio represents comparison of dapagliflozin against placebo with 95% confidence interval.
Hazard ratios adjusted for previous HF hospitalization at baseline (except all-cause death) and stratified by diabetes status.
a
Requiring intravenous therapy for HF.
b
Expressed as difference with 95% confidence interval in () in dapagliflozin compared to placebo.

© 2020 The Authors. European Journal of Heart Failure published by John Wiley & Sons Ltd on behalf of European Society of Cardiology.
1251
1252 P. Dewan et al.

Figure 1 Clinical outcomes according to baseline left ventricular ejection fraction (LVEF). Restricted cubic spine analyses of clinical outcomes
according to baseline LVEF: primary composite outcome, cardiovascular death, heart failure (HF) hospitalization/urgent visit for HF, and all-cause
death. Figures have been restricted to 10–40% LVEF but the results are derived from models based on the entire spectrum of LVEF in DAPA-HF.
HR, hazard ratio.

those assigned to placebo, across all the LVEF categories (Table 2 cohort, and in participants with and without diabetes analysed
..............................................................

and online supplementary Figure S2). separately).

The effect of dapagliflozin was consistent across the range of Because the absolute risk of events was highest in patients in
LVEF for the primary composite outcome [P-value for interaction: the lowest LVEF category, the absolute benefit of dapagliflozin was
0.762 (categorical) and 0.205 (continuous)], cardiovascular death larger in patients with a lower LVEF. For example, applying the
[P-value for interaction: 0.974 (categorical) and 0.997 (continuous)] overall relative risk reduction of 26% to patients with a LVEF of
and HF hospitalization or urgent visit for HF [P-value for inter- <26% yielded an absolute risk reduction of 54 fewer patients with
action: 0.161 (categorical) and 0.150 (continuous)] and all-cause an event per 1000 person-years of follow-up, compared with 31
death [P-value for interaction: 0.866 (categorical) and 0.962 (con- per 1000 person-years of follow-up in the LVEF >35% category.
tinuous)] (Table 2 and Figure 2).
The benefit of dapagliflozin over placebo for these outcomes
Relationship between baseline left
was also consistent in patients with and without diabetes anal-
ysed separately, across the range of LVEF studied (online sup-
ventricular ejection fraction and change
plementary Table S2 and Figure S5). Similarly, beneficial effects of in Kansas City Cardiomyopathy
dapagliflozin also remained constant across the range of LVEF Questionnaire total symptom score
regardless of the time of measurement of LVEF (online supplemen- The mean change in KCCQ-TSS in the placebo group was similar
tary Figure S6). in each of the LVEF categories analysed (Table 2). The propor-
The favourable effect of dapagliflozin on the composite of HF tion of patients in the placebo group exhibiting a ≥5 point increase
hospitalization (first and repeat) and cardiovascular death was also (improvement) in KCCQ-TSS was similar across LVEF categories.
consistent across the spectrum of LVEF studied (in the overall The same was true for the proportion of patients reporting a

© 2020 The Authors. European Journal of Heart Failure published by John Wiley & Sons Ltd on behalf of European Society of Cardiology.
Dapagliflozin effects by left ventricular ejection fraction 1253

≥5 point decrease (deterioration) in KCCQ-TSS. These findings

........................................................................................................................................................................
Table 3 Change in risk of clinical outcomes per 5-point decrease in baseline left ventricular ejection fraction – overall, by diabetes status and by time of left

.........................................................................................................................................................................................
were similar in patients with and without diabetes (online supple-

LVEF measured >6 months

mentary Table S2).

before randomization

1.14 (1.02–1.27) 0.018

1.15 (1.03–1.28) 0.015

1.21 (1.06–1.39) 0.006

1.19 (1.05–1.35) 0.005
1.20 (1.05–1.37) 0.008

1.03 (0.90–1.18) 0.663

1.09 (0.94–1.27) 0.234
Effect of dapagliflozin, compared
with placebo, on change in Kansas City
(n = 779)a

Cardiomyopathy Questionnaire total

symptom score, according to baseline
left ventricular ejection fraction
The mean increase (improvement) in KCCQ-TSS with
LVEF measured ≤6 months

dapagliflozin, compared with placebo, was similar in each of

1.19 (1.13–1.25) <0.001
1.18 (1.13–1.25) <0.001

1.19 (1.12–1.27) <0.001

1.22 (1.14–1.30) <0.001
1.20 (1.12–1.27) <0.001

1.22 (1.15–1.29) <0.001

1.15 (1.08–1.22) <0.001

before randomization

the LVEF categories (P-value for interaction: 0.607) (Table 2; online

supplementary Figure S1). Compared with placebo, more patients
treated with dapagliflozin showed a ≥5 point improvement, and
fewer a ≥5 point deterioration, in each of the LVEF categories
Hazard ratios (and relative risk) adjusted for randomized treatment, previous HF hospitalization at baseline (except all-cause death) and stratified by diabetes status.
(n = 3962)a

analysed. These findings were similar in patients with and without

diabetes.
Numbers represent hazard ratio with 95% confidence interval for each 5-point decrease in LVEF. Rate ratio for total HF hospitalization/CV death.

Relationship between baseline left

1.18 (1.11–1.26) <0.001

1.20 (1.11–1.31) <0.001

1.20 (1.12–1.27) <0.001

1.24 (1.15–1.34) <0.001

1.22 (1.13–1.32) <0.001

1.24 (1.15–1.33) <0.001

1.12 (1.04–1.21) 0.004

ventricular ejection fraction

Diabetes (n = 2139)

and pre-specified safety outcomes

There was no significant difference in the proportion of those
on placebo who discontinued the study due to any reason across
the LVEF categories (Table 4 and online supplementary Table S3).
Similarly, no difference was seen in the proportion of patients on
placebo with adverse events due to renal causes, fractures, ampu-
tation, or major hypoglycaemic events. However, the proportion of
No diabetes (n = 2605)

1.20 (1.10– 1.32) <0.001

1.19 (1.11–1.27) <0.001
1.17 (1.10–1.26) <0.001

1.19 (1.10–1.28) <0.001

1.17 (1.07–1.28) <0.001
1.16 (1.06–1.26) 0.001

1.15 (1.06–1.24) 0.001

patients with volume depletion was higher in the lower LVEF cate-
gories. A fall in SBP among those on placebo during follow-up was
slightly higher in those with LVEF >35% but no such observation
was made with respect to change in creatinine (Table 4).

Effect of dapagliflozin, compared

CV, cardiovascular; HF, heart failure; LVEF, left ventricular ejection fraction.

with placebo, on pre-specified safety

1.20 (1.14–1.27) <.0.001
1.18 (1.13–1.24) <0.001
1.18 (1.13–1.24) <0.001

1.20 (1.13–1.28) <0.001

1.20 (1.13–1.27) <0.001

1.22 (1.16–1.28) <0.001

1.13 (1.07–1.20) <0.001

outcomes, according to baseline left

a Date of LVEF measurement was set to missing for three patients.
Overall (n = 4744)

ventricular ejection fraction

ventricular ejection fraction measurement

There was no significant difference in the proportion of patients

who discontinued for any reason or those who discontinued due to
an adverse event between the treatment groups in any of the LVEF
categories, including due to volume depletion (P-value for interac-
tion: 0.548 and 0.544, respectively) (Table 4). Similarly, no difference
b Requiring intravenous therapy for HF.

in the magnitude of change in SBP or creatinine during follow-up

Total HF hospitalization/CV

was seen between the treatment groups in each LVEF category

HF hospitalization/urgent

(P-value for interaction: 0.529 and 0.258, respectively) (Table 4).

HF hospitalization
Primary outcome

hospitalization
CV death or HF

All-cause death

Discussion
CV death

death
visitb

In this analysis of 4744 patients with HFrEF in DAPA-HF, the

baseline characteristics of patients varied across the spectrum
of LVEF, as expected. Patients with and without diabetes also

© 2020 The Authors. European Journal of Heart Failure published by John Wiley & Sons Ltd on behalf of European Society of Cardiology.
1254 P. Dewan et al.

Figure 2 Effect of dapagliflozin on clinical outcomes according to baseline left ventricular ejection fraction (LVEF). Fractional polynomial
analyses showing the effect of dapagliflozin treatment on clinical outcomes across the range of LVEF: primary composite outcome, cardiovascular
death, heart failure (HF) hospitalization/urgent visit for HF, and all-cause death. Figures have been restricted to 10-40% LVEF but the results
are derived from models based on the entire spectrum of LVEF in DAPA-HF. HR, hazard ratio.

differed as expected, but these differences were consistent across death. These findings are very similar to what was reported in
...........................................................

the range of LVEF studied. LVEF was a powerful predictor of the Prospective comparison of ARNI with ACEI to Determine
the risk of hospitalization and death overall and in patients with Impact on Global Mortality and morbidity in Heart Failure trial
and without diabetes separately. The benefit of dapagliflozin on (PARADIGM-HF), where the corresponding increments in risk
mortality/morbidity outcomes was not modified by baseline LVEF, for each 5-point reduction in LVEF were 17%, 17% and 14%,
irrespective of diabetes status. By contrast, symptom severity at respectively.9 These findings are also consistent with earlier studies
baseline did not vary according to LVEF. Symptoms improved to a assessing the relationship between LVEF and outcomes in HFrEF.1
similar extent with dapagliflozin across the range of LVEF studied. The relative increase in risk of death and hospitalization for a
The benefit of dapagliflozin on symptoms, in relation to LVEF, was 5-point decrement in baseline LVEF was similar in participants with
consistent in patients with and without diabetes. and without diabetes, although the absolute risk for a given LVEF
As in previous studies, patients with lower LVEF were younger, was higher in individuals with diabetes.
more likely to be male, had fewer comorbidities and less likely While the benefit of effective therapies for HFrEF has gener-
to have an ischaemic aetiology. Although there was no difference ally been found to be similar across the LVEF spectrum, the range
in NYHA class across the LVEF categories, NT-proBNP was of LVEF in such analyses has been limited as few landmark trials
substantially higher in patients in the lowest, compared with the included patients with a LVEF >35%.10–24 Furthermore, several
highest, LVEF category (despite a much higher prevalence of atrial earlier studies suggested greater benefit of therapy at the lower
fibrillation in the latter). end of the LVEF spectrum.6 However, we found that, compared
We found that each 5-point decrement in baseline LVEF was with placebo, the benefit of dapagliflozin on the primary and sec-
associated with a 20% higher risk of cardiovascular death, a 20% ondary mortality/morbidity outcomes was consistent across the
higher risk of HF hospitalization and a 13% higher risk of all-cause range of LVEF studied. This benefit according to LVEF was also

© 2020 The Authors. European Journal of Heart Failure published by John Wiley & Sons Ltd on behalf of European Society of Cardiology.
 Table 4 Discontinuation, safety outcomes and laboratory measures according to left ventricular ejection fraction

Overall (n = 4744) <26% (n = 1143) 26–30% (n = 1018) 31–35% (n = 1187) >35% (n = 1396) P-value for
interaction
............................... ............................. ............................. ............................. ..............................
Placebo Dapagliflozin Placebo Dapagliflozin Placebo Dapagliflozin Placebo Dapagliflozin Placebo Dapagliflozin
(n = 2371) (n = 2373) (n = 601) (n = 542) (n = 498) (n = 520) (n = 581) (n = 606) (n = 691) (n = 705)
.........................................................................................................................................................................................
Any discontinuation
Events, n (%) 258 (10.9) 249 (10.5) 65 (10.8) 66 (12.2) 57 (11.5) 62 (11.9) 71 (12.2) 60 (9.9) 65 (9.4) 61 (8.7)
Odds ratio 0.96 (0.80–1.15) 0.665 1.15 (0.80–1.65) 1.05 (0.71–1.53) 0.79 (0.55–1.13) 0.90 (0.63–1.30) 0.548
Discontinuation due to
AEa
Events, n (%) 116/2368 (4.9) 111/2368 (4.7) 30/600 (5.0) 31/540 (5.7) 19/498 (3.8) 26/518 (5.0) 33/579 (5.7) 27/606 (4.5) 34/691 (4.9) 27/704 (3.8)
Odds ratio 0.95 (0.73–1.25) 0.734 1.16 (0.69–1.95) 1.32 (0.72–2.43) 0.76 (0.45–1.28) 0.76 (0.45–1.28) 0.544
Dapagliflozin effects by left ventricular ejection fraction

Volume depletiona
Events, n (%) 162/2368 (6.8) 178/2368 (7.5) 49/600 (8.2) 54/540 (10.0) 42/498 (8.4) 37/518 (7.1) 29/579 (5.0) 39/606 (6.4) 42/691 (6.1) 48/704 (6.8)
Odds ratio 1.11 (0.89–1.38) 0.368 1.26 (0.84–1.89) 0.83 (0.53–1.32) 1.29 (0.79–2.12) 1.14 (0.74–1.75) 0.400
Renala
Events, n (%) 170/2368 (7.2) 153/2368 (6.5) 47/600 (7.8) 39/540 (7.2) 37/498 (7.4) 33/518 (6.4) 45/579 (7.8) 37/606 (6.1) 41/691 (5.9) 44/704 (6.3)
Odds ratio 0.89 (0.71–1.12) 0.326 0.94 (0.60–1.46) 0.84 (0.52–1.37) 0.75 (0.48–1.18) 1.07 (0.69–1.67) 0.899
Fracturea
Events, n (%) 50/2368 (2.1) 49/2368 (2.1) 13/600 (2.2) 9/540 (1.7) 12/498 (2.4) 14/518 (2.7) 11/579 (1.9) 11/606 (1.8) 14/691 (2.0) 15/704 (2.1)
Odds ratio 0.98 (0.66–1.46) 0.919 0.78 (0.33–1.85) 1.13 (0.52–2.47) 0.95 (0.41–2.21) 1.07 (0.51–2.23) 0.857
Amputationa
Events, n (%) 12/2368 (0.5) 13/2368 (0.5) 3/600 (0.5) 1/540 (0.2) 1/498 (0.2) 4/518 (0.8) 3/579 (0.5) 6/606 (1.0) 5/691 (0.7) 2/704 (0.3)
Odds ratio 1.08 (0.49–2.38) 0.842 0.41 (0.04–3.97) 3.82 (0.43–34.35) 1.72 (0.43–6.95) 0.41 (0.08–2.12) 0.336
Major hypoglycaemic
episodea
Events, n (%) 4/2368 (0.2) 4/2368 (0.2) 0 (0.0) 0 (0.0) 0 (0.0) 1/518 (0.2) 0 (0.0) 0 (0.0) 4/691 (0.6) 3/704 (0.4)
Systolic BP, mmHg
Change from baseline −0.38 ± 15.3 −1.92 ± 14.9 0.52 ± 15.6 −0.88 ± 15.2 −0.11 ± 14.8 −2.37 ± 13.8 −0.57 ± 15.9 −1.20 ± 14.9 −1.18 ± 14.7 −2.97 ± 15.4 0.529
at 8 months
Differenceb −1.40 (−2.27 to −0.52) 0.002 −1.13 (−2.87 to 0.61) 0.202 −2.09 (−3.96 to −0.22) 0.028 −0.63 (−2.40 to 1.14) 0.486 −1.69 (−3.32 to −0.05) 0.044
Creatinine, mg/dl
Change from baseline 0.04 ± 0.25 0.07 ± 0.24 0.04 ± 0.2 0.08 ± 0.2 0.06 ± 0.3 0.07 ± 0.2 0.05 ± 0.2 0.05 ± 0.2 0.03 ± 0.2 0.06 ± 70.3 0.258
at 8 months
Differenceb 0.02 (0.01 to 0.04) 0.009 0.04 (0.004 to 0.07) 0.029 0.01 (−0.02 to 0.05) 0.543 0.01 (−0.03 to 0.04) 0.675 −0.03 (0.001 to 0.06) 0.044

BP, blood pressure.

a Only in safety analysis set.

© 2020 The Authors. European Journal of Heart Failure published by John Wiley & Sons Ltd on behalf of European Society of Cardiology.
b Dapagliflozin–placebo.
1255
1256 P. Dewan et al.

consistent in patients with and without diabetes. Consequently, dapagliflozin was consistent across the range of LVEF examined

........................................................................................................................................................................
patients with a low LVEF obtained a particularly large absolute bene- overall, and in patients with and without diabetes.
fit from dapagliflozin because individuals with a low LVEF, especially As has been shown previously, patients in DAPA-HF with lower
if diabetic, were at much greater absolute risk than patients with a LVEF were more likely to have adverse events related to volume
higher LVEF. Whether the benefit of SGLT2 inhibition will extend depletion but no difference was seen between the treatment
to patients with HF and mid-range/mildly reduced and frankly pre- groups even in those with the lowest LVEF, allaying concerns of a
served ejection fraction remains to be determined, especially in potentially greater risk of volume depletion in HF patients in whom
patients without type 2 diabetes. Retrospective subgroup analy- diuretic use is almost universal.36,37 The only other significant
ses of prior trials with SGLT2 inhibitors in individuals with type observation among the safety outcomes was the larger fall in SBP
2 diabetes and predominantly atherosclerotic cardiovascular dis- in the highest LVEF category but this was most likely a function of
ease (or cardiovascular risk factors) have suggested that these their higher baseline SBP.
drugs may be beneficial in patients with HFpEF, but these find-
ings are far from conclusive.25,26 This question will be answered,
definitively, by the ongoing EMPagliflozin outcomE tRial in Patients Study limitations
With chrOnic heaRt Failure With Preserved Ejection Fraction Our study had several limitations. This was a post hoc analysis in
(EMPEROR-Preserved; NCT03057951) and Dapagliflozin Evalua- which patients were divided into arbitrary, clinically relevant LVEF
tion to Improve the LIVEs of Patients With PReserved Ejection categories.1,6,7 Additionally, LVEF was measured using different
Fraction Heart Failure (DELIVER; NCT03619213) trials, both of methods at different sites and there was no core laboratory.
which have enrolled patients with LVEF >40%.27 Time of measurement of LVEF before randomization also varied,
Two novel aspects of this study were the analysis of symptoms but this variation did not affect outcomes. We did not have
and the analysis of recurrent events, in relation to baseline LVEF, information on the method used to measure LVEF. There was also
and according to diabetes status, and the effect of treatment on digit preference in the reporting of LVEF measurements, as often
these outcomes. The two large pharmacological therapy trials that found.1,9 SBP below 95 mmHg and eGFR below 30 mL/min/1.73 m2
have reported the effect of treatment on KCCQ in HFrEF have not were exclusion criteria in DAPA-HF and this may have skewed the
described the relationship between KCCQ score and LVEF or the characteristics of our patients in the lowest LVEF category, more
effect of therapy according to LVEF.28,29 However, in the CHARM of which might have been expected to have lower SBP and worse
program, there was no clear association between LVEF and a renal function.
different patient-reported outcome, a finding that is consistent
with the current observations in DAPA-HF.30 In addition, KCCQ
scores are similar in patients with HFrEF and HFpEF, which also Conclusion
suggests little correlation between this patient-reported outcome
Left ventricular ejection fraction at baseline was a significant
and LVEF.31 Interestingly, change in KCCQ-TSS from baseline was
predictor of hospitalization and mortality (but not symptoms) in
also independent of LVEF, and similar in patients with and without
patients with HFrEF enrolled in DAPA-HF. LVEF did not modify the
diabetes. The reason why symptoms and health-related quality
beneficial effect of dapagliflozin on mortality/morbidity outcomes,
of life correlate poorly with LVEF is uncertain but, importantly,
or symptoms, in patients with HFrEF overall, and in those with and
dapagliflozin improved symptoms as well as other outcomes. This
without diabetes separately.
beneficial effect of dapagliflozin, whether assessed as mean change
in KCCQ-TSS, or the proportion of patients with a clinically
meaningful change (≥5 points), was similar across LVEF categories,
both overall, and in patients with and without diabetes.
Supplementary Information
Analysis of recurrent non-fatal, along with fatal, events may pro- Additional supporting information may be found online in the
vide a better quantification of the full burden of HF, compared Supporting Information section at the end of the article.
with conventional time-to-first event analysis.32–34 Repeat admis- Figure S1. Placebo corrected mean change in Kansas City Car-
sions are distressing for patients, a marker of disease progression, diomyopathy Questionnaire total summary score at 8 months
represent an adverse prognostic change, and are expensive. Like- according to left ventricular ejection fraction – overall and accord-
wise, analysis of recurrent events is a rigorous test of the effect ing to diabetic status.
of treatment, as it measures persistence of pharmacological effect Figure S2. Effect of randomized treatment on clinical outcomes,
and adherence (e.g. treatment discontinuation after a first event according to left ventricular ejection fraction: primary composite
will reduce any effect of therapy on subsequent events).35 That this outcome, cardiovascular death, heart failure hospitalization/urgent
type of analysis reflects disease burden is clearly shown by the very visit, and all-cause death.
high event rates compared with time-to-first event analysis in the Figure S3. Restricted cubic spine analyses of clinical outcomes
present analysis, e.g. reaching almost 40 per 1000 person-years according to baseline left ventricular ejection fraction and diabetes
of follow-up for HF hospitalization and cardiovascular death in status: primary composite outcome, cardiovascular death, heart
patients with diabetes in the lowest LVEF category. However, the failure hospitalization/urgent visit, and all-cause death.
benefit of dapagliflozin was almost identical in the recurrent events Figure S4. Effect of randomized treatment on clinical outcomes,
and time-to-first analyses, and the relative risk reduction with according to left ventricular ejection fraction and diabetes status:

© 2020 The Authors. European Journal of Heart Failure published by John Wiley & Sons Ltd on behalf of European Society of Cardiology.
Dapagliflozin effects by left ventricular ejection fraction 1257

primary composite outcome, cardiovascular death, heart failure personal fees and other from D.L. DeMets Consulting. M.S.S.

........................................................................................................................................................................
hospitalization/urgent visit, and all-cause death. reports grants from Bayer, Daiichi-Sankyo, Eisai, GlaxoSmithKline,
Figure S5. Fractional polynomial analyses showing the effect of Pfizer, Poxel, Quark Pharmaceuticals, and Takeda; grants and per-
dapagliflozin treatment on clinical outcomes across the range of sonal fees from Amgen, AstraZeneca, Intarcia, Janssen Research
left ventricular ejection fraction, according to diabetes status: and Development, The Medicines Company, MedImmune, Merck,
primary composite outcome, cardiovascular death, heart failure and Novartis; and personal fees from Anthos Therapeutics,
hospitalization/urgent visit, and all-cause death. Bristol-Myers Squibb, CVS Caremark, DalCor, Dyrnamix, Espe-
Figure S6. Forest plot showing the hazard ratios (95% confidence rion, IFM Therapeutics, and Ionis. He is a member of the TIMI
interval) for the major clinical outcomes in DAPA-HF according to Study Group, which has also received institutional research grant
left ventricular ejection fraction (LVEF) and the time of measure- support through Brigham and Women’s Hospital from Abbott,
ment of LVEF (≤6 months vs. >6 months) Aralez, Roche, and Zora Biosciences. O.B., M.S. and A.M.L. are
Table S1. Baseline characteristics according to left ventricular full-time employees of AstraZeneca. I.S.A. reports fees for serving
ejection fraction and diabetes status at baseline. as United States national leader of a trial from AstraZeneca, fees
Table S2. Clinical outcomes according to left ventricular ejection for serving on a steering committee from ARCA Biopharma,
fraction and diabetes status at baseline. Amgen, LivaNova, and Novartis, fees for serving on an end-point
Table S3. Safety outcomes according to left ventricular ejection committee from Boehringer Ingelheim, fees for serving as chair
fraction and diabetes status at baseline. of a data and safety monitoring board from Boston Scientific,
Table S4. Baseline characteristics according to left ventricular and advisory board fees from Zensun. J.B. reports receiving
ejection fraction and randomization arm. advisory board fees from Novartis and Pfizer and lecture fees
from Getginge. M.K. reports grant support and lecture fees from
Funding Astellas Pharma, Sanofi, Pfizer, Ono Pharmaceutical, Novartis, and
The DAPA-HF trial was funded by AstraZeneca. Prof McMurray Mitsubishi Tanabe Pharma, lecture fees from Daiichi Sankyo, Bayer,
is supported by British Heart Foundation Centre of Research Boehringer Ingelheim, Kowa Pharmaceutical, Sawai Pharmaceuti-
Excellence Grant RE/18/6/34217. cal, MSD, Shionogi, Kureha, Taisho Toyama Pharmaceutical, Takeda
Conflict of interest: S.D.S. reports grants from AstraZeneca, Pharmaceutical, and Toa Eiyo, and manuscript fees from Japan
Bellerophon, Celladon, Ionis, Lone Star Heart, Mesoblast, National Medical Data Center. B.M. reports receiving lecture fees from
Institutes of Health/National Heart, Lung, and Blood Institute, AstraZeneca, Sanofi Aventis, Servier, and Biotronik and grant sup-
Sanofi Pasteur, and Eidos; grants and personal fees from Alnylam, port and lecture fees from Abbott and Medtronic. E.O’M. reports
Amgen, AstraZeneca, BMS, Gilead, GSK, MyoKardia, Novar- receiving fees for serving on a clinical trial (paid to her institution),
tis, Theracos, Bayer, and Cytokinetics; and personal fees from consulting fees, and lecture fees from AstraZeneca, Bayer, Amgen,
Akros, Corvia, Ironwood, Merck, Roche, Takeda, Quantum and Novartis, consulting fees from Merck, fees for serving on
Genomics, AoBiome, Janssen, Cardiac Dimensions, Tenaya, and a clinical trial (paid to her institution) from American Regent,
Daichi-Sankyo. P.S.J. reports other from AstraZeneca, personal and consulting fees and lecture fees from Pfizer and Boehringer
fees from Novartis and Cytokinetics, and grants from Boehringer Ingelheim. J.V.V.McM. reports nonfinancial support and other
Ingelheim. S.E.I. reports personal fees and non-financial support from AstraZeneca, Cardiorentis, Amgen, Oxford University/Bayer,
from AstraZeneca, Boehringer Ingelheim, Sanofi/Lexicon, Merck, Theracos, Abbvie, Novartis, Glaxo Smith Kline, Vifor-Fresenius,
VTV Therapeutics, and Abbott/Alere, as well as personal fees Kidney Research UK, and Novartis, as well as other support from
from AstraZeneca and Zafgen. L.K. reports other support from Bayer, DalCor, Pfizer, Merck, Bristol Myers, and Squibb. All other
AstraZeneca and personal fees from Novartis and Bristol-Myers authors have nothing to disclose.
Squibb as a speaker. M.N.K. is consultant for Vifor Pharma and
reports personal fees from AstraZeneca; grants, personal fees,
and other from AstraZeneca; grants and personal fees from References
Boehringer Ingelheim; and personal fees from Sanofi, Amgen, 1. Solomon SD, Anavekar N, Skali H, McMurray JJ, Swedberg K, Yusuf S, Granger
NovoNordisk, Merck (Diabetes), Janssen, Bayer, GlaxoSmithKline, CB, Michelson EL, Wang D, Pocock S, Pfeffer MA; Candesartan in Heart Failure
Reduction in Mortality (CHARM) Investigators. Influence of ejection fraction on
Glytec, Novartis, Applied Therapeutics, Amarin, and Eli Lilly. cardiovascular outcomes in a broad spectrum of heart failure patients. Circulation
F.A.M. reports personal fees from AstraZeneca. P.P. reports per- 2005;112:3738–3744.
sonal fees and other from AstraZeneca, Boehringer Ingelheim, 2. Pocock SJ, Ariti CA, McMurray JJ, Maggioni A, Køber L, Squire IB, Swed-
berg K, Dobson J, Poppe KK, Whalley GA, Doughty RN; Meta-Analysis
Bayer, BMS, Cibiem, Novartis, and RenalGuard; personal fees
Global Group in Chronic Heart Failure. Predicting survival in heart failure: a
from Pfizer, Servier, Respicardia, and Berlin-Chemie; other from risk score based on 39 372 patients from 30 studies. Eur Heart J 2013;34:
Amgen; and grants, personal fees, and other from Vifor Pharma. 1404–1413.
3. McMurray JJ, Solomon SD, Inzucchi SE, Køber L, Kosiborod MN, Martinez FA,
D.L.DeM. reports personal fees from Frontier Science, Actelion,
Ponikowski P, Sabatine MS, Anand IS, Bělohlávek J, Böhm M, Chiang CE, Chopra
Population Health Research Institute, Duke Clinical Research VK, de Boer RA, Desai AS, Diez M, Drozdz J, Dukát A, Ge J, Howlett JG, Katova T,
Institute, Bristol-Meyers Squibb, Medtronic, Boston Scientific, Kitakaze M, Ljungman CE, Merkely B, Nicolau JC, O’Meara E, Petrie MC, Vinh
Glaxo Smith Kline, Merck, National Institutes of Health/National PN, Schou M, Tereshchenko S, Verma S, Held C, DeMets DL, Docherty KF, Jhund
PS, Bengtsson O, Sjöstrand M, Langkilde AM; DAPA-HF Trial Committees and
Institute of Allergy and Infectious Diseases, and National Institutes Investigators. Dapagliflozin in patients with heart failure and reduced ejection
of Health/National Heart, Lung, and Blood Institute, as well as fraction. N Engl J Med 2019;381:1995–2008.

© 2020 The Authors. European Journal of Heart Failure published by John Wiley & Sons Ltd on behalf of European Society of Cardiology.
1258 P. Dewan et al.

4. McMurray JJ, DeMets DL, Inzucchi SE, Køber L, Kosiborod MN, Langkilde AM, Zareba W; MADIT-CRT Trial Investigators. Cardiac-resynchronization ther-

.........................................................................................................................................................................
Martinez FA, Bengtsson O, Ponikowski P, Sabatine MS, Sjöstrand M, Solomon apy for the prevention of heart-failure events. N Engl J Med 2009;361:
SD; DAPA-HF Committees and Investigators. The Dapagliflozin And Prevention 1329–1338.
of Adverse-outcomes in Heart Failure (DAPA-HF) trial: baseline characteristics. 22. Zannad F, McMurray JJ, Krum H, van Veldhuisen DJ, Swedberg K, Shi H, Vincent J,
Eur J Heart Fail 2019;21:1402–1411. Pocock SJ, Pitt B; EMPHASIS-HF Study Group. Eplerenone in patients with systolic
5. Green CP, Porter CB, Bresnahan DR, Spertus JA. Development and evaluation heart failure and mild symptoms. N Engl J Med 2011;364:11–21.
of the Kansas City Cardiomyopathy Questionnaire: a new health status measure 23. Swedberg K, Komajda M, Böhm M, Borer JS, Ford I, Dubost-Brama A,
for heart failure. J Am Coll Cardiol 2000;35:1245–1255. Lerebours G, Tavazzi L; SHIFT Investigators. Ivabradine and outcomes in
6. Cleland JG, Bunting KV, Flather MD, Altman DG, Holmes J, Coats AJ, Manzano L, chronic heart failure (SHIFT): a randomised placebo-controlled study. Lancet
McMurray JJ, Ruschitzka F, van Veldhuisen DJ, von Lueder TG, Böhm M, Ander- 2010;376:875–885.
sson B, Kjekshus J, Packer M, Rigby AS, Rosano G, Wedel H, Hjalmarson Å, 24. McMurray JJ, Packer M, Desai AS, Gong J, Lefkowitz MP, Rizkala AR, Rouleau JL,
Wikstrand J, Kotecha D; Beta-blockers in Heart Failure Collaborative Group. Shi VC, Solomon SD, Swedberg K, Zile MR; PARADIGM-HF Investigators and
Beta-blockers for heart failure with reduced, mid-range, and preserved ejection Committees. Angiotensin-neprilysin inhibition versus enalapril in heart failure.
fraction: an individual patient-level analysis of double-blind randomized trials. Eur N Engl J Med 2014;371:993–1004.
Heart J 2018;39:26–35. 25. Rådholm K, Figtree G, Perkovic V, Solomon SD, Mahaffey KW, de Zeeuw D,
7. Ponikowski P, Voors AA, Anker SD, Bueno H, Cleland JG, Coats AJ, Falk V, Fulcher G, Barrett TD, Shaw W, Desai M, Matthews DR, Neal B. Canagliflozin
González-Juanatey JR, Harjola VP, Jankowska EA, Jessup M, Linde C, Nihoy- and heart failure in type 2 diabetes mellitus: results from the CANVAS Program.
annopoulos P, Parissis JT, Pieske B, Riley JP, Rosano GM, Ruilope LM, Rus- Circulation 2018;138:458–468.
chitzka F, Rutten FH, van der Meer P. 2016 ESC Guidelines for the diagnosis 26. Kato ET, Silverman MG, Mosenzon O, Zelniker TA, Cahn A, Furtado RH, Kuder J,
and treatment of acute and chronic heart failure: The Task Force for the diagno- Murphy SA, Bhatt DL, Leiter LA, McGuire DK, Wilding JP, Bonaca MP, Ruff CT,
sis and treatment of acute and chronic heart failure of the European Society of Desai AS, Goto S, Johansson PA, Gause-Nilsson I, Johanson P, Langkilde AM, Raz I,
Cardiology (ESC). Developed with the special contribution of the Heart Failure Sabatine MS, Wiviott SD. Effect of dapagliflozin on heart failure and mortality in
Association (HFA) of the ESC. Eur J Heart Fail 2016;18:891–975. type 2 diabetes mellitus. Circulation 2019;139:2528–2536.
8. Royston P, Sauerbrei W. Two techniques for investigating interactions between 27. Anker SD, Butler J, Filippatos GS, Jamal W, Salsali A, Schnee J, Kimura K,
treatment and continuous covariates in clinical trials. Stata J 2009;9:230–251. Zeller C, George J, Brueckmann M, Zannad F, Packer M; EMPEROR-Preserved
9. Solomon SD, Claggett B, Desai AS, Packer M, Zile M, Swedberg K, Rouleau JL, Trial Committees and Investigators. Evaluation of the effects of sodium–glucose
Shi VC, Starling RC, Kozan Ö, Dukat A, Lefkowitz MP, McMurray JJ. Influence co-transporter 2 inhibition with empagliflozin on morbidity and mortality in
of ejection fraction on outcomes and efficacy of sacubitril/valsartan (LCZ696) patients with chronic heart failure and a preserved ejection fraction: rationale for
in heart failure with reduced ejection fraction: the Prospective Comparison of and design of the EMPEROR-Preserved trial. Eur J Heart Fail 2019;21:1279–1287.
ARNI with ACEI to Determine Impact on Global Mortality and Morbidity in Heart 28. Ekman I, Chassany O, Komajda M, Bhm M, Borer JS, Ford I, Tavazzi L, Swed-
Failure (PARADIGM-HF) trial. Circ Heart Fail 2016;9:002744. berg K. Heart rate reduction with ivabradine and health related quality of life
10. Yusuf S, Pitt B, Davis CE, Hood WB, Cohn JN; SOLVD Investigators. Effect of in patients with chronic heart failure: results from the SHIFT study. Eur Heart J
enalapril on survival in patients with reduced left ventricular ejection fractions 2011;32:2395–2404.
and congestive heart failure. N Engl J Med 1991;325:293–302. 29. Lewis EF, Claggett BL, McMurray JJ, Packer M, Lefkowitz MP, Rouleau JL, Liu J,
11. Pitt B, Zannad F, Remme WJ, Cody R, Castaigne A, Perez A, Palensky J, Wittes J. Shi VC, Zile MR, Desai AS, Solomon SD, Swedberg K. Health-related quality of
The effect of spironolactone on morbidity and mortality in patients with severe life outcomes in PARADIGM-HF. Circ Heart Fail 2017;10:e003430.
heart failure. Randomized Aldactone Evaluation Study Investigators. N Engl J Med 30. Lewis EF, Lamas GA, O’Meara E, Granger CB, Dunlap ME, McKelvie RS, Prob-
1999;341:709–717. stfield JL, Young JB, Michelson EL, Halling K, Carlsson J, Olofsson B, McMurray
12. Dargie HJ, Lechat P. The Cardiac Insufficiency Bisoprolol Study II (CIBIS-II): a JJ, Yusuf S, Swedberg K, Pfeffer MA; CHARM Investigators. Characterization of
randomised trial. Lancet 1999;353:9–13. health-related quality of life in heart failure patients with preserved versus low
13. Eichhorn EJ, Bristow MR. The Carvedilol Prospective Randomized Cumulative ejection fraction in CHARM. Eur J Heart Fail 2007;9:83–91.
Survival (COPERNICUS) trial. Curr Control Trials Cardiovasc Med 2001;2:20–23. 31. Chandra A, Vaduganathan M, Lewis EF, Claggett BL, Rizkala AR, Wang W,
14. Merit-HF Study Group. Effect of metoprolol CR/XL in chronic heart failure: Lefkowitz MP, Shi VC, Anand IS, Ge J, Lam CS, Maggioni AP, Martinez F, Packer M,
Metoprolol CR/XL Randomised Intervention Trial in Congestive Heart Failure Pfeffer MA, Pieske B, Redfield MM, Rouleau JL, Van Veldhuisen DJ, Zannad F,
(MERIT-HF). Lancet 1999;353:2001–2007. Zile MR, McMurray JJ, Solomon SD; PARAGON-HF Investigators. Health-related
15. Granger CB, McMurray JJ, Yusuf S, Held P, Michelson EL, Olofsson B, Östergren J, quality of life in heart failure with preserved ejection fraction: the PARAGON-HF
Pfeffer MA, Swedberg K; CHARM Investigators and Committees. Effects of trial. JACC Heart Fail 2019;7:862–874.
candesartan in patients with chronic heart failure and reduced left-ventricular 32. Rogers JK, Pocock SJ, McMurray JJ, Granger CB, Michelson EL, Ostergren J,
systolic function intolerant to angiotensin-converting-enzyme inhibitors: the Pfeffer MA, Solomon SD, Swedberg K, Yusuf S. Analysing recurrent hospital-
CHARM-Alternative trial. Lancet 2003;362:772–776. izations in heart failure: a review of statistical methodology, with application to
16. McMurray JJ, Östergren J, Swedberg K, Granger CB, Held P, Michelson EL, Olofs- CHARM-Preserved. Eur J Heart Fail 2014;16:33–40.
son B, Yusuf S, Pfeffer MA; CHARM Investigators and Committees. Effects of can- 33. Mogensen UM, Gong J, Jhund PS, Shen L, Køber L, Desai AS, Lefkowitz
desartan in patients with chronic heart failure and reduced left-ventricular systolic MP, Packer M, Rouleau JL, Solomon SD, Claggett BL, Swedberg K, Zile MR,
function taking angiotensin-converting-enzyme inhibitors: the CHARM-Added Mueller-Velten G, McMurray JJ. Effect of sacubitril/valsartan on recurrent events in
trial. Lancet 2003;362:767–771. the Prospective Comparison of ARNI with ACEI to Determine Impact on Global
17. Cohn JN, Tognoni G; Valsartan Heart Failure Trial Investigators. A randomized Mortality and Morbidity in Heart Failure trial (PARADIGM-HF). Eur J Heart Fail
trial of the angiotensin-receptor blocker valsartan in chronic heart failure. N Engl 2018;20:760–768.
J Med 2001;345:1667–1675. 34. Anker SD, Schroeder S, Atar D, Bax JJ, Ceconi C, Cowie MR, Crisp A,
18. Bristow MR, Saxon LA, Boehmer J, Krueger S, Kass DA, De Marco T, Car- Dominjon F, Ford I, Ghofrani HA, Gropper S, Hindricks G, Hlatky MA, Hol-
son P, DiCarlo L, De Mets D, White BG, De Vries DW, Feldman AM; Com- comb R, Honarpour N, Jukema JW, Kim AM, Kunz M, Lefkowitz M, Le Floch C,
parison of Medical Therapy, Pacing, and Defibrillation in Heart Failure (COM- Landmesser U, McDonagh TA, McMurray JJ, Merkely B, Packer M, Prasad K,
PANION) Investigators. Cardiac-resynchronization therapy with or without an Revkin J, Rosano GM, Somaratne R, Stough WG, Voors AA, Ruschitzka F. Tra-
implantable defibrillator in advanced chronic heart failure. N Engl J Med 2004;350: ditional and new composite endpoints in heart failure clinical trials: facilitating
2140–2150. comprehensive efficacy assessments and improving trial efficiency. Eur J Heart Fail
19. Cleland JG, Daubert JC, Erdmann E, Freemantle N, Gras D, Kappenberger L, 2016;18:482–489.
Tavazzi L; Cardiac Resynchronization-Heart Failure (CARE-HF) Study Investiga- 35. Claggett B, Pocock S, Wei LJ, Pfeffer MA, McMurray JJ, Solomon SD. Comparison
tors. The effect of cardiac resynchronization on morbidity and mortality in heart of time-to-first event and recurrent-event methods in randomized clinical trials.
failure. N Engl J Med 2005;352:1539–1549. Circulation 2018;138:570–577.
20. Tang AS, Wells GA, Talajic M, Arnold MO, Sheldon R, Connolly S, Hohn- 36. Solomon SD, Vaduganathan M, Claggett BL, Packer M, Zile M, Swedberg K,
loser SH, Nichol G, Birnie DH, Sapp JL, Yee R, Healey JS, Rouleau JL; Rouleau J, Pfeffer MA, Desai A, Lund LH, Koeber L, Anand I, Sweitzer NK,
Resynchronization-Defibrillation for Ambulatory Heart Failure Trial Investigators. Linssen G, Merkely B, Arango JL, Vinereanu D, Chen CH, Senni M, Sibulo A,
Cardiac-resynchronization therapy for mild-to-moderate heart failure. N Engl Boytsov S, Shi V, Rizkala A, Lefkowitz M, McMurray JJ. Sacubitril/valsartan across
J Med 2010;363:2385–2395. the spectrum of ejection fraction in heart failure. Circulation 2019;141:352–361.
21. Moss AJ, Hall WJ, Cannom DS, Klein H, Brown MW, Daubert JP, Estes 37. Fitchett D. A safety update on sodium glucose co-transporter 2 inhibitors.
NA, Foster E, Greenberg H, Higgins SL, Pfeffer MA, Solomon SD, Wilber D, Diabetes Obes Metab 2019;21 Suppl 2:34–42.

© 2020 The Authors. European Journal of Heart Failure published by John Wiley & Sons Ltd on behalf of European Society of Cardiology.