Articles

Imatinib in patients with severe COVID-19: a randomised,

double-blind, placebo-controlled, clinical trial
Jurjan Aman, Erik Duijvelaar, Liza Botros*, Azar Kianzad*, Job R Schippers*, Patrick J Smeele*, Sara Azhang, Imke H Bartelink, Ahmed A Bayoumy,
Pierre M Bet, Wim Boersma, Peter I Bonta, Karin A T Boomars, Lieuwe D J Bos, Job J M H van Bragt, Gert-Jan Braunstahl, Lucas R Celant,
Katrien A B Eger, J J Miranda Geelhoed, Yurika L E van Glabbeek, Hans P Grotjohan, Laura A Hagens, Chris M Happe, Boaz D Hazes,
Leo M A Heunks, Michel van den Heuvel, Wouter Hoefsloot, Rianne J A Hoek, Romke Hoekstra, Herman M A Hofstee, Nicole P Juffermans,
E Marleen Kemper, Renate Kos, Peter W A Kunst, Ariana Lammers, Ivo van der Lee, E Laurien van der Lee, Anke-Hilse Maitland-van der Zee,
Pearl F M Mau Asam, Adinda Mieras, Mirte Muller, Elisabeth C W Neefjes, Esther J Nossent, Laurien M A Oswald, Maria J Overbeek,
Carolina C Pamplona, Nienke Paternotte, Niels Pronk, Michiel A de Raaf, Bas F M van Raaij, Merlijn Reijrink, Marcus J Schultz, Ary Serpa Neto,
Elise M A Slob, Frank W J M Smeenk, Marry R Smit, A Josien Smits, Janneke E Stalenhoef, Pieter R Tuinman, Arthur L E M Vanhove,
Jeroen N Wessels, Jessie C C van Wezenbeek, Anton Vonk Noordegraaf, Frances S de Man, Harm J Bogaard

Summary
Background The major complication of COVID-19 is hypoxaemic respiratory failure from capillary leak and alveolar Lancet Respir Med 2021
oedema. Experimental and early clinical data suggest that the tyrosine-kinase inhibitor imatinib reverses pulmonary Published Online
capillary leak. June 17, 2021
https://doi.org/10.1016/
S2213-2600(21)00237-X
Methods This randomised, double-blind, placebo-controlled, clinical trial was done at 13 academic and non-academic
See Online/Comment
teaching hospitals in the Netherlands. Hospitalised patients (aged ≥18 years) with COVID-19, as confirmed by an https://doi.org/10.1016/
RT-PCR test for SARS-CoV-2, requiring supplemental oxygen to maintain a peripheral oxygen saturation of greater S2213-2600(21)00266-6
than 94% were eligible. Patients were excluded if they had severe pre-existing pulmonary disease, had pre-existing This online publication has
heart failure, had undergone active treatment of a haematological or non-haematological malignancy in the previous been corrected.
The corrected version first
12 months, had cytopenia, or were receiving concomitant treatment with medication known to strongly interact with
appeared at thelancet.com/
imatinib. Patients were randomly assigned (1:1) to receive either oral imatinib, given as a loading dose of 800 mg on respiratory on June 25, 2021
day 0 followed by 400 mg daily on days 1–9, or placebo. Randomisation was done with a computer-based clinical data *Contributed equally
management platform with variable block sizes (containing two, four, or six patients), stratified by study site. The Department of Pulmonary
primary outcome was time to discontinuation of mechanical ventilation and supplemental oxygen for more than Medicine, Amsterdam
48 consecutive hours, while being alive during a 28-day period. Secondary outcomes included safety, mortality at Cardiovascular Sciences,
28 days, and the need for invasive mechanical ventilation. All efficacy and safety analyses were done in all randomised Amsterdam UMC, VUMC, Vrije
Universiteit Amsterdam,
patients who had received at least one dose of study medication (modified intention-to-treat population). This study is Amsterdam, Netherlands
registered with the EU Clinical Trials Register (EudraCT 2020–001236–10). (J Aman PhD, E Duijvelaar MD,
L Botros MD, A Kianzad MD,
Findings Between March 31, 2020, and Jan 4, 2021, 805 patients were screened, of whom 400 were eligible and randomly J R Schippers MD, P J Smeele MD,
S Azhang MD, A A Bayoumy MD,
assigned to the imatinib group (n=204) or the placebo group (n=196). A total of 385 (96%) patients (median age 64 years L R Celant MD,
[IQR 56–73]) received at least one dose of study medication and were included in the modified intention-to-treat population. Y L E van Glabbeek MD,
Time to discontinuation of ventilation and supplemental oxygen for more than 48 h was not significantly different between C M Happe PhD, B D Hazes BSc,
the two groups (unadjusted hazard ratio [HR] 0·95 [95% CI 0·76–1·20]). At day 28, 15 (8%) of 197 patients had died in the R J A Hoek MSc,
E L van der Lee BSc,
imatinib group compared with 27 (14%) of 188 patients in the placebo group (unadjusted HR 0·51 [0·27–0·95]). After A Mieras PhD, E J Nossent MD,
adjusting for baseline imbalances between the two groups (sex, obesity, diabetes, and cardiovascular disease) the HR for C C Pamplona MSc,
mortality was 0·52 (95% CI 0·26–1·05). The HR for mechanical ventilation in the imatinib group compared with the M A de Raaf PhD, M Reijrink BSc,
A J Smits MD,
placebo group was 1·07 (0·63–1·80; p=0·81). The median duration of invasive mechanical ventilation was 7 days
A L E M Vanhove BSc,
(IQR 3–13) in the imatinib group compared with 12 days (6–20) in the placebo group (p=0·0080). 91 (46%) of 197 patients J N Wessels MD,
in the imatinib group and 82 (44%) of 188 patients in the placebo group had at least one grade 3 or higher adverse event. J C C van Wezenbeek MSc,
The safety evaluation revealed no imatinib-associated adverse events. Prof A Vonk Noordegraaf PhD,
F S de Man PhD,
Prof H J Bogaard PhD);
Interpretation The study failed to meet its primary outcome, as imatinib did not reduce the time to discontinuation of Department of Pulmonology
ventilation and supplemental oxygen for more than 48 consecutive hours in patients with COVID-19 requiring (S Azhang, M J Overbeek PhD)
supplemental oxygen. The observed effects on survival (although attenuated after adjustment for baseline imbalances) and Department of Internal
Medicine (H M A Hofstee PhD),
and duration of mechanical ventilation suggest that imatinib might confer clinical benefit in hospitalised patients with
Haaglanden Medisch Centrum,
COVID-19, but further studies are required to validate these findings. The Hague, Netherlands;
Department of Pharmacy
Funding Amsterdam Medical Center Foundation, Nederlandse Organisatie voor Wetenschappelijk Onderzoek/ (I H Bartelink PhD, P M Bet PhD)
and Department of Intensive
ZonMW, and the European Union Innovative Medicines Initiative 2.
Care (P R Tuinman PhD,
Prof L M A Heunks PhD),
Copyright © 2021 Elsevier Ltd. All rights reserved. Amsterdam UMC, VUMC,

www.thelancet.com/respiratory Published online June 17, 2021 https://doi.org/10.1016/S2213-2600(21)00237-X 1

 Articles

Amsterdam, Netherlands; Introduction interleukin-6 with tocilizumab or sarilumab also reduces

Department of Pulmonology, A major complication of SARS-CoV-2 infection is mortality in patients in the ICU.9
Chest Unit, Suez Canal
University, Suez, Egypt
hypoxaemic respiratory failure, which is the cause of most Despite the direct contribution of alveolar oedema to
(A A Bayoumy); Department of hospital admissions, intubations, and deaths due to hypoxaemia and adverse outcomes in COVID-19, such as
Pulmonology, Noordwest COVID-19.1 The clinical manifestation of lung involvement intubation or death, treatment strategies attenuating or
Ziekenhuisgroep, Alkmaar, in COVID-19 is pneumonitis, often progressing to acute reversing this type of lung damage have so far not been
Netherlands (W Boersma PhD,
N Paternotte MD); Department
respiratory distress syndrome (ARDS). Radiological evaluated. One of the suggested approaches to attenuate
of Respiratory Medicine, investigations typically reveal extensive bilateral ground or reverse lung damage is treatment with the
Amsterdam UMC, AMC, glass opacities and consolidations suggestive of diffuse antineoplastic drug imatinib.1 The ABL tyrosine-kinase
University of Amsterdam, pulmonary oedema. Post-mortem studies have shown that inhibitor imatinib protects against capillary leak and
Amsterdam, Netherlands
(P I Bonta PhD, L D J Bos PhD,
a key feature of COVID-19 is endothelial injury, which is alveolar oedema caused by inflammatory stimuli. Since
J J M H van Bragt PharmD, caused by direct infection of pulmonary endothelial cells,2,3 the first observation of a patient recovering from
K A B Eger MD, R Kos MSc, disruption of the endothelial barrier, and widespread respiratory failure due to pulmonary veno-occlusive
A Lammers MSc, Prof endothelial cell apoptosis.3 disease following imatinib treatment in 2008,10 in-vitro
A-H Maitland-van der Zee PhD,
P F M Mau Asam BSc,
Therapeutic strategies that have been evaluated for and in-vivo studies have revealed that imatinib protects
E M A Slob PharmD); COVID-19 include antiviral treatment and immuno­ the endothelial barrier under inflammatory conditions.11–15
Department of Pulmonology, modulatory drugs. Remdesivir was found to shorten the Subsequent case reports verified the protective effect of
Erasmus Medisch Centrum,
time to recovery in hospitalised patients with COVID-19 in imatinib in other conditions characterised by alveolar
Rotterdam, Netherlands
(K A T Boomars PhD); the ACTT-1 trial,4 but this finding was not replicated in the oedema,16–18 including COVID-19.19
Department of Intensive Care (L WHO Solidarity trial.5 Neither study showed a survival In this report, we aimed to assess the efficacy and safety
D J Bos, L A Hagens benefit from remdesivir. Other antiviral interventions, of oral imatinib in hospitalised patients with COVID-19
MSc, N P Juffermans PhD,
including (hydroxy)chloroquine,5,6 interferon,5 and requiring supplemental oxygen administration.
Prof M J Schultz PhD,
M R Smit MSc) and Hospital lopinavir–ritonavir,5–7 have also failed to show clinical
Pharmacy (E M Kemper PhD), benefit. As an alternative to antiviral treatment, and based Methods
Amsterdam UMC, AMC, on the recognition of a marked hyperinflammatory pattern Study design and patients
Amsterdam, Netherlands;
in patients with a poor outcome, other investigators This randomised, double-blind, placebo-controlled, clinical
Department of Pulmonology,
Sint Franciscus Ziekenhuis, instead aimed to modulate the immune response. trial was done at 13 large academic and non-academic
Rotterdam, Netherlands Dexamethasone reduced COVID-19 mortality from 24% to teaching hospitals in the Netherlands. Patients were
(G-J Braunstahl PhD, 20% in hospitalised patients, and from 40% to 30% in eligible for inclusion if they were aged 18 years or older, had
L M A Oswald MD); Department
those in the intensive care unit (ICU).8 The most recent been admitted to hospital with SARS-CoV-2 infection
of Pulmonology Leiden
University Medical Center, data from the Randomised, Embedded, Multifactorial (confirmed with an RT-PCR test), and required
Leiden, Netherlands Adaptive Platform Trial for Community-Acquired supplemental oxygen to maintain a peripheral oxygen
(J J M Geelhoed PhD, Pneumonia (REMAP-CAP; published in February, 2021) saturation of greater than 94%. Exclusion criteria included,
B F M van Raaij MD);
Department of Pulmonology,
indicates that monoclonal antibody-mediated inhibition of but were not limited to, pre-existing severe pulmonary
Isala Ziekenhuizen, Zwolle,
Netherlands Research in context
(H P Grotjohan PhD);
Department of Pulmonology, Evidence before this study pulmonary capillary leak and oedema that is observed in
Radboud UMC, Nijmegen, We searched PubMed on March 5, 2021, using the following patients with COVID-19. This study failed to meet its primary
Netherlands
(Prof M van den Heuvel PhD,
search term (“SARS-CoV-2” OR “COVID-19”) AND (“imatinib” outcome, defined as time to discontinuation of ventilation and
W Hoefsloot PhD); Department OR “Gleevec” OR “Glivec”) AND (“RCT” OR “clinical trial” OR supplemental oxygen for more than 48 consecutive hours.
of Pulmonology, Antonius “randomized controlled trial”). We searched for clinical trials However, a clinical benefit was suggested by reductions in
Ziekenhuis, Sneek, Netherlands
assessing the effect of imatinib in patients with COVID-19 mortality and duration of mechanical ventilation in the
(R Hoekstra MD); Department of
Intensive Care (N P Juffermans), published between database inception and March 5, 2021, with imatinib group compared with the placebo group. However, the
Department of Pulmonology no language restrictions. Apart from published trial protocols, sample size of the study precludes robust conclusions from the
(P W A Kunst PhD), and no clinical trials were identified. Preclinical studies have shown results of these secondary outcomes. Imatinib had a favourable
Department of Internal
that imatinib improves endothelial barrier integrity, in safety profile in this trial.
Medicine (J E Stalenhoef PhD),
Onze Lieve Vrouwe Gasthuis, particular under inflammatory conditions, and reverses
Implications of all the available evidence
Amsterdam, Netherlands; pulmonary oedema in animal models of acute lung injury.
Department of Pulmonology,
The observed reductions in mortality and duration of
Spaarne Gasthuis, Haarlem, Added value of this study mechanical ventilation in patients given imatinib compared
Netherlands (I van der Lee PhD); To our knowledge, this is the first randomised clinical trial with placebo suggest that imatinib might benefit patients with
Department of Pulmonology, assessing the effects of imatinib in hospitalised patients with COVID-19, particularly those with a severe disease course.
Catharina Ziekenhuis,
Eindhoven, Netherlands
COVID-19 requiring supplemental oxygen. Although other Future trials will need to substantiate our findings, both in
(M Muller MD, E C W Neefjes PhD, interventions target viral replication (eg, remdesivir) or a larger study populations (to allow stratification by disease
Prof F W J M Smeenk PhD); dysregulated immune response (eg, corticosteroids or severity) and in selected patient populations with
Department of Pulmonology,
tocilizumab), there are no interventions that target the COVID-19-associated acute respiratory distress syndrome.
Gelre Ziekenhuis, Apeldoorn,

2 www.thelancet.com/respiratory Published online June 17, 2021 https://doi.org/10.1016/S2213-2600(21)00237-X

 Articles

disease, pre-existing heart failure (a left ventricular ejection before the loading dose was given; an absolute leukocyte Netherlands (N Pronk MD);
fraction of <40%), active treatment of a haematological or count of less than 2·5 × 10⁹ cells per L; an absolute Department of Critical Care
Medicine and Institute of
non-haematological malignancy within 12 months before thrombocyte count of less than 100 × 10⁹ cells per L;
Education and Research,
enrolment, the presence of cytopenia, and concomitant an increase in aminotransferase concentrations to more Hospital Israelita Albert
treatment with medication known to strongly interact with than three times the upper limit of normal if aspartate Einstein, São Paulo, Brazil
imatinib. Full details of the exclusion criteria are provided aminotransferase (AST) and alanine aminotransferase (A Serpa Neto PhD)
Correspondence to:
in the study protocol (appendix pp 7–37). Women of (ALT) concentrations were within reference values at Prof Harm J Bogaard,
childbearing age were asked to undergo a pregnancy test baseline, or increased AST and ALT concentrations to Department of Pulmonary
before enrolment. more than three times the value at baseline if Medicine, Amsterdam UMC,
The trial was approved by the Medical Ethics concentrations were higher than reference values at VUMC, Vrije Universiteit
Amsterdam, Amsterdam 1081,
Committee of the Amsterdam UMC (VUMC, baseline; an increase in bilirubin concentrations of more Netherlands
Amsterdam, Netherlands), and was done in accordance than 1·5 times the upper limit of normal if bilirubin hj.bogaard@amsterdamumc.nl
with Good Clinical Practice guidelines and the concentrations were within the normal range at baseline, See Online for appendix
Declaration of Helsinki. All patients provided written or an increase in bilirubin concentrations of more than
informed consent before randomisation. 1·5 times the baseline value if concentrations were
higher than reference values at baseline; or an increase
Randomisation and masking in QTc of more than 100 msec relative to baseline. In
Patients were enrolled by study investigators and patients discharged from hospital before day 9, clinical
randomly assigned (1:1) to either the placebo group or the laboratory testing and ECG measurements were
imatinib group. Randomisation was done with the suspended, according to guidelines for imatinib
Castor Electronic Data Capturing System (Castor EDC; treatment of chronic myeloid leukaemia in an outpatient
Amsterdam, Netherlands) using variable block sizes setting.20 Adverse events and severe adverse events,
(two, four, or six patients), stratified by study site. graded by use of the Common Terminology Criteria for
Allocation to study groups was done by Castor, and was Adverse Events version 5.0 (November, 2017), were
not accessible to study investigators. After randomisation, recorded on a daily basis. All grade 3 or higher adverse
patients received oral imatinib (as 400 mg tablets) or events reported at any time during the 28-day study
non-identical placebo tablets, which were dispensed by period by the patient or observed by the investigator or
the hospital pharmacy and distributed in sealed the medical staff were recorded. The investigators
containers. Medical staff and investigators were not assessed whether an adverse event was related to
involved in the dispensing of the study drugs. Patients,
medical staff, and investigators were masked to group
assignment. 805 patients screened

Procedures 405 excluded

After randomisation, patients in the imatinib group 295 did not provide consent
90 due to screening failure
received a loading dose of 800 mg imatinib on day 0, 13 were transferred to another hospital
followed by 400 mg once daily on days 1–9. Patients in 7 withdrew
the placebo group received placebo tablets in a similar
dosing scheme.
400 enrolled and randomised
For evaluation of efficacy, the clinical status of patients
(ie, whether they had been discharged, admitted to a
hospital ward, admitted to the ICU, or had died) and
whether they were receiving ventilatory or oxygen support 196 assigned placebo 204 assigned imatinib
(ie, receiving oxygen supplementation, high-flow nasal
oxygen, non-invasive ventilation, mechanical ventilation,
or extracorporeal membrane oxygenation) were recorded 8 excluded 7 excluded
5 withdrew before first 6 withdrew before first
daily on days 0–9 and on day 28. If patients were discharged dose dose
from the hospital before day 9, their clinical status was 1 transferred to another 1 transferred to another
hospital* hospital*
ascertained by a telephone call on days 9 and 28. 2 due to medication error†
Safety evaluations included clinical laboratory testing
at baseline (day 0) and on days 1, 2, 3, 5, 7, and 9,
consisting of a blood cell count, and assessment of 188 included in 197 included in
final analysis population final analysis population
electrolytes, renal function, and liver enzymes. An
electrocardiogram (ECG) was used to monitor corrected
Figure 1: Trial profile
QT (QTc) intervals at baseline and on days 1, 3, 5, and 9. *Patients lost to follow-up due to transfer to another hospital before the first dose of study medication could be
Stopping criteria included a reduction in haemoglobin of administered. †Study medication was dispensed to the wrong patients and they were withdrawn from the study.
more than 30% relative to the baseline measurement Errors were detected before the first dose of study medication was given.

www.thelancet.com/respiratory Published online June 17, 2021 https://doi.org/10.1016/S2213-2600(21)00237-X 3

 Articles

treatment, and they consulted medical staff if the as assessed by use of the WHO ordinal scale for clinical
potential association was not clear. improvement (with 1 indicating discharge with full recovery
[with no home oxygen use]; 2 indicating discharge without
Outcomes full recovery [with actual home oxygen use]; 3 indicating
The primary outcome was the time to discontinuation of admission to a non-ICU ward and not receiving
ventilation and supplemental oxygen for more than supplemental oxygen; 4 indicating admission to a non-ICU
48 consecutive hours, while being alive during a 28-day hospital ward and receiving supplemental oxygen;
period after randomisation. Secondary efficacy outcomes 5 indicating admission to a medium care unit [MCU] or an
were: mortality at 28 days, the number of ICU admissions, ICU without mechanical ventilation; 6 indicating
the length of ICU admission, the need for mechanical admission to an MCU or ICU with mechanical ventilation;
ventilation, the duration of mechanical ventilation, the and 7 indicating death). The number of ventilator-free days
duration of oxygen supplementation, the duration of was defined as the number of days alive and free from
hospital admission, and clinical status at day 9 and day 28, mechanical ventilation. Patients were only considered as
free from mechanical ventilation if they had a successful
extubation, defined as being free from mechanical
Imatinib group Placebo group
(n=197) (n=188)
ventilation for at least 48 consecutive hours. Non-survivors
were considered to have no ventilator-free days.
Age, years 64 (57–73) 64 (55–74)
Secondary safety outcomes were incidence
BMI, kg/m² 27·5 (25·3–31·1) 29·7 (25·6–32·9)
of grade 3 and 4 adverse and severe adverse events,
Sex
blood cell count, kidney function, liver enzymes,
Male 146 (74%) 118 (63%) N-terminal-pro-B natriuretic peptide (NTproBNP)
Female 51 (26%) 70 (37%) concentrations (all measured on days 0, 1, 2, 3, 5, 7,
Comorbidities* and 9), and QTc intervals (measured on days 0, 1, 3, 5,
Current or former 77 (39%) 76 (40%) and 9). Pharmacokinetic and biomarker analyses, as
smoker
defined in the study protocol, are currently being done
BMI of >30 kg/m² 53 (29%) 83 (47%)
and the results will be published separately.
Diabetes 41 (21%) 54 (29%)
Cardiovascular disease† 35 (18%) 48 (26%)
Statistical analysis
Hypertension 69 (35%) 76 (40%)
Based on an anticipated hazard ratio (HR) of 1·39 for the
COPD or asthma 38 (19%) 33 (18%)
primary outcome (imatinib group vs the placebo group),
Venous 5 (3%) 5 (3%) a one-sided α level of 0·025 for superiority of the
thromboembolism
intervention and a β level of 0·20, the sample size was set
Renal failure 7 (4%) 7 (4%)
at 193 patients per group. Analyses for primary and
Hepatic disease 1 (1%) 1 (1%)
secondary outcomes were done in all patients who
Rheumatic disease 11 (6%) 18 (10%)
underwent randomisation and had received at least
Heart failure 8 (4%) 4 (2%)
one dose of study medication (the modified intention-to-
Medical treatments‡
treat population). All statistical analyses were predefined
Glucose lowering drugs 40 (20%) 54 (29%)
in the study protocol and the statistical analysis plan
Antihypertensive 91 (46%) 102 (54%)
(appendix pp 38–44). No data were missing for the
treatment
primary and secondary outcome analyses. No data
ACE or ARB 51 (26%) 70 (37%)
imputation was done to account for missing data.
Statins 62 (32%) 65 (35%)
Baseline demographic variables are reported as absolute
Platelet inhibitors 42 (21%) 40 (21%)
numbers (proportions) for categorical data, or as means
Oral anticoagulants 17 (9%) 21 (11%)
(SDs) or medians (IQRs) for continuous variables.
Clinical presentation
Categorical variables were analysed by use of the χ² test or
Atypical 24 (12%) 32 (17%)
Fisher’s exact test. Continuous variables were analysed
Chest pain 34 (17%) 27 (14%)
with a Wilcoxon rank-sum test. Imbalances at baseline
Cough 151 (77%) 141 (75%)
(p<0·10) were carried forward as adjustment factors in
Dyspnoea 160 (81%) 158 (84%)
regression analyses for the primary and secondary
Fever 133 (68%) 112 (60%) outcomes.
Other flu-like 82 (42%) 78 (42%) The primary outcome and the secondary outcomes of
symptoms
mortality at day 28 and the need for mechanical ventilation
Time from symptom 10 (8–12) 10 (8–12)
onset to randomisation, were analysed with Kaplan-Meier curves to plot event
days rates over time; between-group differences were expressed
SpO2/FiO2 ratio 321 (265–380) 323 (238–377) as HRs with 95% CIs, calculated as regression analyses.
(Table 1 continues on next page) Kaplan-Meier and Cox regression analyses were done for
time-to-event analyses of secondary outcomes. The

4 www.thelancet.com/respiratory Published online June 17, 2021 https://doi.org/10.1016/S2213-2600(21)00237-X

 Articles

assumption of proportional hazards was tested by

examination of Schoenfeld residuals. Two-sided p values Imatinib group Placebo group
(n=197) (n=188)
are provided for all tests.
All data were recorded in Castor EDC. The Clinical (Continued from previous page)
Research Office, an independent contract research Laboratory values at admission
organisation of the Amsterdam UMC, was responsible Haemoglobin, g/dL 13·5 (12·6–14·7) 13·7 (12·6–14·7)
for data monitoring. The safety of patients was monitored C-reactive protein, mg/L 102 (48–158) 95 (45-149)
by a Data Safety Monitoring Board (DSMB), which NTproBNP, ng/L 147 (49–411) 132 (50–352)
convened after enrolment of 30 patients (April 23, 2020) Lactate dehydrogenase, 365 (279–445) 366 (293–496)
and 60 patients (May 7, 2020). During these meetings, U/L
patient safety was assessed on the basis of an unblinded Lymphocytes, 0·90 (0·60–1·19) 0·94 (0·68–1·28)
× 10⁹ cells per L
line listing of severe adverse events. A severe adverse
Neutrophils, 6·00 (4·10–8·56) 5·94 (4·40–8·29)
event was defined as any untoward medical event that × 10⁹ cells per L
occurred during the study period that (1) resulted in
Leukocytes, 7·60 (5·60–10·40) 7·60 (5·77–10·00)
death; (2) was life-threatening; (3) required inpatient × 10⁹ cells per L
hospitalisation or led to an increased duration of Thrombocytes, 248 (184–323) 236 (190–310)
hospitalisation; (4) resulted in persistent or significant × 10⁹ cells per L
disability; or (5) required intervention to prevent Medication initiated at admission
permanent impairment or damage. The DSMB Low-molecular-weight 167 (85%) 150 (80%)
subsequently convened after enrolment of 100 patients heparin
(Oct 22, 2020) and 200 patients (Dec 14, 2020), and Oral anticoagulants 6 (3%) 8 (4%)
patient safety was assessed on the basis of an unblinded Antibiotics 85 (43%) 77 (41%)
line listing of severe adverse events, and futility was Dexamethasone 143 (73%) 133 (71%)
assessed on the basis of predefined analyses of efficacy Remdesivir 40 (20%) 40 (21%)
outcomes. (Hydroxy)chloroquine 15 (8%) 17 (9%)
All statistical analyses were done in RStudio, Data are median (IQR) or n (%). BMI=body-mass index. COPD=chronic obstructive
version 1.3.1073. This study is registered with the EU pulmonary disease. ACE=angiotensin-converting enzyme. ARB=angiotensin
Clinical Trials Register (EudraCT 2020–001236–10). receptor blocker. SpO2=oxygen saturation. FiO2=fractional concentration of
oxygen in inspired air. NTproBNP=N-terminal-pro-B natriuretic peptide.
*Comorbidities as reported at admission or present in the patient’s medical
Role of the funding source record. †Cardiovascular diseases included arrythmias (predominantly atrial
The funders of the study had no role in study design, fibrillation), valvular disease, coronary artery disease, and conduction disorders.
‡Medical treatment (or home medication) as reported at admission or present in
data collection, data analysis, data interpretation, or the patient’s medical record.
writing of the report.
Table 1: Baseline characteristics and demographic variables
Results
Between March 31, 2020, and Jan 4, 2021, 805 patients at Common risk factors for a poor outcome and comorbidities
13 participating hospitals in the Netherlands were included smoking (153 [40%] of 385 patients), obesity
screened (appendix p 2). Of these patients, 400 (50%) were (136 [38%]), diabetes (95 [25%]), cardiovascular disease
enrolled and randomly assigned to the placebo (83 [22%]), and chronic obstructive pulmonary disease or
group (n=196) or the imatinib group (n=204). A total of asthma (71 [18%]). Despite randomisation, sex, obesity,
11 (3%) patients withdrew consent before receiving the diabetes, and cardiovascular disease were slightly
first dose of study medication (figure 1). During the unbalanced between the two groups. No differences in
peaks of the pandemic, two (1%) patients withdrew from baseline laboratory measurements were observed between
the study before receiving their first dose of study the two groups, suggesting similar disease severity at
medication, as they were relocated to non-participating presentation. Medications started at admission included
hospitals with fewer COVID-19 admissions. The final low-molecular-weight heparin, oral anticoagulants, and
analysis population included 197 patients in the imatinib antibiotics (table 1). Dexamethasone was started at
group and 188 patients in the placebo group (figure 1). admission in 276 (72%) patients and remdesivir was
The baseline characteristics of patients are shown in started in 80 (21%) patients. Most patients (276 [72%] of 385)
table 1. The median age of patients was 64 years received dexamethasone on admission to emergency care
(IQR 56–73; range 28–93). 264 (69%) of 385 patients were units before receiving study medication. As participation
men, and the median body-mass index of patients at in other interventional studies was not allowed at
enrolment was 28·5 kg/m² (IQR 25·5–32·4). Clinical enrolment, and because tocilizumab, anakinra, lopinavir–
presentations at baseline included chest pain, dyspnoea, ritonavir, and convalescent plasma were not routinely
cough, and fever. The median duration from onset of administered to patients with COVID-19 in the Netherlands
symptoms to randomisation was 10 days (IQR 8–12) in the at the time the study was done, no patients received these
imatinib group and 10 days (8–12) in the placebo group. treatments during the trial.

www.thelancet.com/respiratory Published online June 17, 2021 https://doi.org/10.1016/S2213-2600(21)00237-X 5

 Articles

During the 28-day study period, 303 (79%) patients

A Discontinuation of ventilation and supplemental oxygen for 48 h
1·00 discontinued supplemental oxygen and mechanical
Treatment period ventilation for more than 48 consecutive hours, including
+ + + +

0·75 + + + + + + + +
160 (81%) of 197 patients in the imatinib group and
Cumulative incidence

+
+ + + 143 (76%) of 188 patients in the placebo group. For
+ the primary outcome, time to discontinuation of
0·50 +
+ Unadjusted HR 0·95 (95% CI 0·76–1·20), p=0·69 supplemental oxygen and mechanical ventilation was not
+
+ significantly different between the two groups
0·25 + Treatment period
+
Placebo (unadjusted HR 0·95 [95% CI 0·76–1·20], p=0·69;
+
+
Imatinib figure 2A). This difference remained non-significant
0·00 +
after adjusting for each baseline imbalance (sex, obesity,
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28
Time (days) diabetes, and cardio­ vascular disease) separately and
Number at risk
(number censored)
combined (table 2).
Placebo 188 187 155 116 90 71 52 44 36 33 32 27 24 22 20 The Kaplan-Meier curves for mortality are shown in
(0) (3) (12) (13) (16) (19) (19) (19) (21) (21) (23) (25) (26) (26) (45) figure 2B. At day 28, 15 (8%) of 197 patients had died in
Imatinib 197 196 171 140 108 89 65 56 48 41 34 32 28 27 24 the imatinib group compared with 27 (14%) of 188 patients
(0) (1) (4) (6) (7) (7) (9) (10) (10) (11) (11) (11) (13) (14) (37)
in the placebo group. The unadjusted HR for mortality in
B Mortality
the imatinib group versus the placebo group was 0·51
1·00 (95% CI 0·27–0·95; p=0·034). The HRs for mortality
Treatment period
100 were attenuated, and for some of the adjusted analyses
0·75 0·15 +
became non-significant after adjusting for im­balances in
Cumulative incidence

0·10 + the baseline characteristics (table 2), as the number of

0·05 men outnumbered the number of women in the imatinib
0·50
0·00 group, and a higher proportion of patients in the placebo
0 4 8 12 16 20 24 28
0·25
group had diabetes, cardiovascular disease, or both than
Unadjusted HR 0·51 (95% CI 0·27–0·95), p=0·034 in the imatinib group. After adjusting for all imbalances,
+ +

+
+
the HR for mortality was 0·52 (95% CI 0·26–1·05;
0·00 +
p=0·068; table 2). Logistic regression analysis yielded an
Number at risk
(number censored)
odds ratio for mortality at 28 days in the imatinib group
Placebo 188 179 172 169 168 166 161 161 versus the placebo group of 0·49 (95% CI 0·25–0·96).
(0) (0) (0) (0) (0) (0) (0) (0) During the 28-day study period, 39 (20%) of 197 patients
Imatinib 197 195 190 189 187 186 184 182 in the imatinib group and 33 (18%) of 188 patients in the
(0) (0) (0) (0) (0) (0) (0) (0)
placebo group were admitted to the ICU. Patients in the
C Need for mechanical ventilation ICU in the two treatment groups were balanced in terms
1·00
Treatment period
of baseline characteristics (data not shown). Of these
100 patients, 30 (77%) in the imatinib group and 26 (79%) in
0·15 the placebo group were intubated and mechanically
+

0·75
++
Cumulative incidence

0·10 + ventilated (figure 2C). The unadjusted HR for mechanical

0·05
0·50 ventilation in the imatinib group versus the placebo group
0·00
0 4 8 12 16 20 24 28 was 1·07 (95% CI 0·63–1·80; p=0·814; figure 2C, table 2),
0·25 Unadjusted HR 1·07 (95% CI 0·63–1·80), p=0·81
and this difference between the groups remained
+ ++ + +
+ +
+ + + + + + + + ++
+ insignificant when adjusting for sex, obesity, diabetes, and
+ +
+
+ + cardiovascular disease (figure 2C, table 2). The median
0·00 +

0 4 8 12 16 20 24 28 duration of mechanical ventilation was 7 days (IQR 3–13)

Time (days) in the imatinib group and 12 days (6–20) in the placebo
Number at risk group (p=0·0080; table 3). In post-hoc analyses, we
(number censored)
Placebo 188 157 148 145 144 144 143 143
restricted the analysis to survivors only, and observed that
(0) (12) (15) (17) (18) (18) (19) (0) the median duration of mechanical ventilation in the
Imatinib 197 178 169 167 162 160 158 158 imatinib group was 7 days (3–12) compared with 12 days
(0) (4) (6) (8) (8) (9) (9) (0) (6–25) in the placebo group (p=0·023). In patients
admitted to the ICU, the number of ventilator-free days
Figure 2: Kaplan-Meier analysis of primary and secondary outcomes
Kaplan-Meier curves showing time-to-event analyses for time to discontinuation of ventilation and supplemental was 22 (14–26) in the imatinib group versus 9 days (0–23)
oxygen for more than 48 consecutive hours, while being alive during the 28-day study period as the primary in the placebo group (p=0·018; table 3).
outcome (A), mortality at 28 days as a secondary outcome (B), and time to invasive mechanical ventilation during The median duration of hospital admission was 7 days
the 28-day study period as a secondary outcome (C). Unadjusted HRs with 95% CIs were calculated by
(IQR 4–11) in the imatinib group and 6 days (3–11) in the
Cox regression analyses. HR=hazard ratio.
placebo group (p=0·51). The median duration of oxygen
supplementation was 7 days (3–12) in the imatinib group

6 www.thelancet.com/respiratory Published online June 17, 2021 https://doi.org/10.1016/S2213-2600(21)00237-X

 Articles

and 5 days (3–11) in the placebo group (p=0·23; table 3).

HR (95% CI) p value
The median duration of ICU stay was 8 days (5–13) in the
imatinib group and 15 days (7–21) in the placebo group Primary endpoint
(p=0·025). The proportions of patients at different stages Unadjusted 0·95 (0·76–1·20) 0·69
of clinical recovery, classified as per the WHO ordinal Adjusted for sex 1·01 (0·81–1·28) 0·90
scale for clinical recovery, at day 9 and day 28 of the study Adjusted for obesity 0·99 (0·78–1·26) 0·92
period are shown in figure 3. Compared with placebo, Adjusted for diabetes 0·96 (0·76–1·20) 0·69
imatinib use was associated with fewer patients classified Adjusted for cardiovascular disease 0·94 (0·75–1·18) 0·59
as having high disease severity on day 9 and day 28 Adjusted for all the above 1·07 (0·62–1·84) 0·82
(figure 3). Mortality (secondary endpoint)
Of all 385 patients, 231 (60%) received the complete Unadjusted 0·51 (0·27–0·95) 0·034
10-day course of study medication; 77 (38%) of 197 patients Adjusted for sex 0·47(0·25–0·89) 0·019
in the imatinib group and 77 (41%) of 188 patients in the Adjusted for obesity 0·46 (0·23–0·92) 0·029
placebo group did not receive the complete course of study Adjusted for diabetes 0·54 (0·29–1·02) 0·057
medication. The mean duration of study drug use was Adjusted for cardiovascular disease 0·54 (0·29–1·01) 0·055
7·6 days (SD 3·4) in the imatinib group and 7·7 days (3·2) Adjusted for all the above 0·52 (0·26–1·05) 0·068
in the placebo group. Frequent reasons for not completing Need for mechanical ventilation (secondary endpoint)
the full course included requesting to stop treatment Unadjusted 1·07 (0·63–1·80) 0·81
(44 [11%] of 385 patients), discontinuation due to discharge Adjusted for sex 0·98 (0·58–1·67) 0·95
or transfer to another hospital (37 [10%]), and dis­ Adjusted for obesity 1·08 (0·63–1·86) 0·77
continuation based on stopping criteria (34 [9%]). Adjusted for diabetes 1·13 (0·66–1·91) 0·66
30 (15%) patients in the imatinib group and 14 (7%) patients Adjusted for cardiovascular disease 1·06 (0·62–1·79) 0·84
in the placebo group requested to discontinue the study Adjusted for all the above 1·02 (0·80–1·30) 0·87
medication (appendix p 5). The most frequent reasons for
All HRs are for the imatinib group versus the placebo group. The final analysis
dis­continuation of study medication reported by population consisted of 188 patients in the placebo group and 197 patients in the
patients included grade 1 or 2 gastrointestinal discomfort imatinib group. HRs and 95% CIs were calculated by use of Cox regression analysis
(appendix p 6). 11 (6%) patients in the imatinib group and and adjusted for sex and the indicated comorbidities. HR=hazard ratio.

23 (12%) patients in the placebo group discontinued Table 2: Clinical outcomes in the final analysis population
treatment because they met predefined stopping criteria
(appendix p 5).
Test results for kidney function, liver and cardiac
Imatinib group Placebo group p value
enzymes, and QTc intervals on days 0–5 are provided in (n=197) (n=188)
the appendix (pp 3–4). Creatinine concentrations on day
Duration of oxygen supplementation, days 7 (3–12) 5 (3–11) 0·23
1 were slightly higher in the imatinib group than in the
Duration of mechanical ventilation, days
placebo group, but no differences were observed on
Final analysis population* 7 (3–13) 12 (6–20) 0·0080
subsequent days. Additionally, no differences in liver
Survivors† 7 (3–12) 12 (6–25) 0·023
enzymes, NTproBNP concentrations, or QTc intervals
Duration of intensive care unit admission, days‡ 8 (5–13) 15 (7–21) 0·025
were observed between the two groups.
Number of ventilator-free days in the ICU population§ 22 (14–26) 9 (0–23) 0·018
During the study, grade 3, 4, and 5 adverse events were
Duration of hospital admission, days 7 (4–11) 6 (3–11) 0·51
recorded. A total of 188 grade 3 or 4 adverse events
were reported in patients in the imatinib group and Data are median (IQR). *Measured in 30 (15%) participants in the imatinib group and 26 (14%) patients in the placebo
group. †Measured in 24 (12%) patients in the imatinib group and 18 (10%) patients in the placebo group. ‡Measured
260 events were reported in the placebo group. 91 (46%) of in 39 (20%) patients in the imatinib group and 33 (18%) patients in the placebo group. §Measured in 39 (20%)
197 patients in the imatinib group and 82 (44%) of patients in the imatinib group and 33 (18%) patients in the placebo group.
188 patients in the placebo group had at least one grade 3
Table 3: Duration of respiratory support and care in the final analysis population
or higher adverse event. The most frequent grade 3 adverse
events included thromboembolic events, decreased
lymphocyte counts, alkalosis, and hyperglycaemia (table 4). occurred after extended anticoagulation treatment for
The most frequent grade 4 adverse event was ARDS. No extracorporeal membrane oxygenation.
specific adverse events were attributed to imatinib
treatment. Two (1%) patients in the placebo group had Discussion
acute renal failure compared with none of the patients in Treatment of hospitalised, hypoxaemic patients with
the imatinib group. Most grade 5 adverse events were COVID-19 pneumonitis with imatinib did not change the
ARDS, which occurred in 12 (6%) patients in the imatinib time to discontinuation of supplemental oxygen and
group and in 35 (12%) patients in the placebo group ventilation when compared with placebo (primary
(appendix p 6). Haemorrhagic stroke was observed in outcome). However, the analysis of secondary outcomes
one (1%) patient in the imatinib group and in indicated that mortality at 28 days was lower in the
one (1%) patient in the placebo group, both of which imatinib group than in the placebo group, which

www.thelancet.com/respiratory Published online June 17, 2021 https://doi.org/10.1016/S2213-2600(21)00237-X 7

 Articles

from critically ill to a less severe state of disease, with a

Death Admitted to a non-ICU Discharged without full recovery long-term need for oxygen supplementation.
Admitted to an MCU or ICU hospital ward and receiving (home oxygen use)
with mechanical ventilation supplemental oxygen Discharged with full recovery
Independent validation of this clinical benefit from
Admitted to an MCU or ICU Admitted to a non-ICU (no home oxygen use) imatinib in patients with severe forms of COVID-19 is
without mechanical hospital ward without
ventilation supplemental oxygen
required. Trials of oral imatinib in hospitalised patients
A Day 9
with COVID-19 are currently recruiting patients in Spain
(NCT04346147) and in the USA (NCT04394416).
Placebo Although the rationale behind imatinib treatment was
to mitigate pulmonary capillary leak, the question of
Imatinib
why clinical benefit was predominantly observed in
patients with COVID-19-associated ARDS remains.
One explanation might be that the relative contribution
B Day 28 of pulmonary capillary leak is larger in patients with
COVID-19-associated ARDS than in less severe forms of
Placebo COVID-19. This hypothesis is supported by plasma
studies, which have linked circulating markers of
Imatinib vascular injury, such as angiopoietin-2, von Willebrand
factor, and E-selectin, to clinical severity in patients with
0 25 50 75 100 COVID-19.21 Concentrations of circulating angiopoietin-2
Proportion of patients (%) and E-selectin were higher in patients requiring
ICU admission,22 and high concentrations at
Figure 3: Clinical status at day 9 (A) and day 28 (B)
Classification of patients in the imatinib group versus the placebo group according to the WHO seven-point ordinal
presentation23 or increasing concentrations during
scale for clinical improvement at day 9 and day 28 of the study period. MCU=medium care unit. ICU=intensive care admission were predictors of mortality.24 Similar findings
unit. were observed in patients with ARDS unrelated to
COVID-19, in which increasing concentrations of
paralleled reductions in the duration of mechanical angiopoietin-2 and von Willebrand factor were associated
ventilation and length of ICU stay. Of note, the statistical with pulmonary vascular leak and increased ARDS
significance of the change in mortality was somewhat severity.25 The results of these biomarker studies together
attenuated after adjustment for baseline imbalances. In with our clinical observations could indicate that
terms of safety, no imatinib-related adverse events (grade pulmonary capillary leak underlies the transition to a
3 or higher) were observed. clinical ARDS phenotype, which can be targeted by
To our knowledge, this is the first clinical trial to imatinib. Alternatively, imatinib could target pathways
evaluate imatinib in patients with COVID-19. The other than alveolocapillary integrity. Imatinib exerts a
rationale behind the use of imatinib treatment was to vasculoprotective effect via inhibition of the ABL2
mitigate pulmonary capillary leak and alveolar oedema in tyrosine kinase.12,26 ABL2 is an important regulator of the
these patients, as experimental and early clinical evidence cytoskeleton, and, as such, has also been shown to
suggests that imatinib protects the integrity of the mediate endosomal trafficking of SARS-CoV and
vascular barrier in a wide array of inflammatory MERS-CoV viral particles.27 Imatinib has shown antiviral
conditions.11–15 Contrary to our expectations, and despite activity against SARS-CoV27 and SARS-CoV-2 in vitro.28
reductions in mortality and duration of mechanical However, another study found no evidence that viral
ventilation, imatinib did not reduce the time to replication or cell entry of SARS-CoV-2 was inhibited at
discontinuation of respiratory support, as measured by clinically achievable concentrations.29 Whether there are
the combined endpoint of use of supplemental oxygen direct antiviral effects of ABL tyrosine-kinase inhibitors
and mechanical ventilation. This primary outcome was against SARS-CoV-2 therefore remains controversial.
chosen on the premise that imatinib would accelerate This trial has some limitations. First, 14 (7%) of
recovery in all patients. Of all 385 patients included in 188 patients in the placebo group and nine (5%) of
the final analysis, 303 (79%) discontinued supplemental 197 patients in the imatinib group were lost to follow-up
oxygen and mechanical ventilation for more than or withdrew consent between randomisation and the
48 consecutive hours, indicating a mild disease course in first dose of study medication. This attrition was partly
most patients. Although this study does not provide explained by the fact that, during the pandemic, high
evidence of clinical benefit in patients with a mild disease numbers of admissions in particular regions in
course, our data do suggest that imatinib might confer the Netherlands forced hospitals to relocate their
benefit specifically in patients with severe COVID-19. patients to hospitals in other regions. Second, despite
Imatinib did not prevent the need for mechanical randomisation, there were imbalances in several
ventilation, but did reduce its duration. These findings baseline variables between the two groups (including sex
suggest that imatinib might have altered the disease and comorbidities). We corrected for these imbalances
course, specifically in patients with severe COVID-19; by calculating adjusted HRs. Third, hospital discharge

8 www.thelancet.com/respiratory Published online June 17, 2021 https://doi.org/10.1016/S2213-2600(21)00237-X

 Articles

Grade 3 Grade 4
Imatinib group Placebo group Total Imatinib group Placebo group Total
(n=197) (n=188) (n=385) (n=197) (n=188) (n=385)
Any adverse event 156 (79%) 218 (116%) 374 (97%) 32 (16%) 42 (22%) 74 (19%)
Anaemia 3 (2%) 7 (4%) 10 (3%) 0 0 0
Cardiac disorders
Atrioventricular block complete 0 1 (1%) 1 (<1%) 0 0 0
Myocardial infarction 0 1 (1%) 1 (<1%) 0 0 0
Rhythm disorder (atrial fibrillation or flutter) 4 (2%) 4 (2%) 8 (2%) 0 2 (1%) 2 (1%)
Sinus bradycardia 0 1 (1%) 1 (<1%) 0 1 (1%) 1 (<1%)
Myringitis with bloody otorrhoea 0 1 (1%) 1 (<1%) 0 0 0
Adrenal insufficiency 0 2 (1%) 2 (1%) 0 0 0
Eye disorders
Conjunctivitis 0 1 (1%) 1 (<1%) 0 0 0
Other 0 0 0 1 (1%)* 0 1 (<1%)
Gastrointestinal disorders
Diarrhoea 1 (1%) 0 1 (<1%) 0 0 0
Ileus 0 1 (1%) 1 (<1%) 0 0 0
Rectal haemorrhage 0 1 (1%) 1 (<1%) 0 1 (1%) 1 (<1%)
Vomiting 0 1 (1%) 1 (<1%) 0 0 0
Gastric ulcer 0 0 0 1 (1%) 0 1 (<1%)
General disorders and administration site conditions
Fever 1 (1%) 3 (2%) 4 (1%)
Multiorgan failure 0 0 0 0 1 (1%) 1 (<1%)
Infections and infestations
Lip infection 0 2 (1%) 2 (1%) 0 0 0
Lung infection (other than COVID-19) 7 (4%) 10 (5%) 17 (4%) 0 0 0
Other† 5 (3%) 4 (2%) 9 (2%) 0 0 0
Sepsis 1 (1%) 2 (1%) 3 (1%) 0 0 0
Investigations
Increased blood ALT or AST 6 (3%) 5 (3%) 11 (3%) 0 0 0
Increased blood bilirubin 1 (1%) 2 (1%) 3 (1%) 0 0 0
Increased blood creatinine 2 (1%) 1 (1%) 3 (1%) 0 0 0
Increased blood γ-glutamyl transferase 0 3 (2%) 3 (1%) 0 0 0
Prolonged QT corrected interval 8 (4%) 14 (7%) 22 (6%) 0 0 0
Decreased lymphocyte count 20 (10%) 15 (8%) 35 (9%) 0 0 0
Decreased white blood cell count 0 (0) 1 (1) 1 (<1%) 0 0 0
Increased creatinine 0 0 0 0 1 (1%) 1 (<1%)
Decreased platelet count 0 0 0 1 (1%) 1 (1%) 2 (1%)
Metabolism and nutrition disorders
Acidosis 10 (5%) 8 (4%) 18 (5%) 1 (1%) 5 (3%) 6 (2%)
Alkalosis 21 (11%) 20 (11%) 41 (11%) 0 0 0
Anorexia 1 (1%) 0 1 (<1%) 0 0 0
Hyperglycaemia 22 (11%) 37 (20%) 59 (15%) 0 0 0
Hyperkalaemia 1 (1%) 6 (3%) 7 (2%) 0 0 0
Hypernatremia 1 (1%) 4 (2%) 5 (1%) 0 1 (1%) 1 (<1%)
Hypoalbuminemia 2 (1%) 5 (3%) 7 (2%) 0 0 0
Hypokalaemia 2 (1%) 3 (2%) 5 (1%) 1 (1%) 0 1 (<1%)
Hyponatremia 1 (1%) 3 (2%) 4 (1%) 0 0 0
Benign, malignant, or unspecified neoplasm 0 0 0 0 1 (1%)‡ 1 (<1%)
Nervous system disorders
Peripheral motor neuropathy 0 2 (1%) 2 (1%) 0 0 0
Ischaemic stroke 0 2 (1%) 2 (1%) 0 0 0
(Table 4 continues on next page)

www.thelancet.com/respiratory Published online June 17, 2021 https://doi.org/10.1016/S2213-2600(21)00237-X 9

 Articles

Grade 3 Grade 4
Imatinib group Placebo group Total Imatinib group Placebo group Total
(n=197) (n=188) (n=385) (n=197) (n=188) (n=385)
(Continued from previous page)
Delirium 2 (1%) 12 (6%) 14 (4%) 0 1 (1%) 1 (<1%)
Renal and urinary disorders
Haematuria 1 (1%) 2 (1%) 3 (1%) 0 0 0
Acute kidney injury 0 0 0 0 2 (1%) 2 (1%)
Respiratory, thoracic, and mediastinal disorders
Acute respiratory distress syndrome 9 (5%) 6 (3%) 15 (4%) 26 (13%) 20 (11%) 46 (12%)
Pneumothorax 1 (1%) 0 1 (<1%) 1 (1%) 0 1 (<1%)
Pulmonary fibrosis 1 (1%) 0 1 (<1%) 0 0 0
Other§ 4 (2%) 2 (1%) 6 (2%) 1 (1%) 0 1 (<1%)
Aspiration 0 0 0 0 1 (1%) 1 (<1%)
Skin and subcutaneous tissue disorders
Maculopapular rash 1 (1%) 0 1 (<1%) 0 0 0
Other 1 (1%)¶ 0 1 (<1%) 0 0 0
Vascular disorders
Hypotension 1 (1%) 7 (4%) 8 (2%) 0 1 (1%) 1 (<1%)
Thromboembolic event 18 (9%) 14 (7%) 32 (8%) 1 (1%) 0 1 (<1%)
The severity of adverse events was graded according to the Common Terminology Criteria for Adverse Events, version 5.0. All patients who received at least one dose of study
medication were included in the safety population. ALT=alanine aminotransferase. AST=aspartate aminotransferase. *Patient developed a conjunctivitis with a haematoma
in both eyes. The patient was unmasked and treatment was discontinued; symptoms resolved shortly thereafter. †Defined as a positive throat swab culture that resulted in
starting intravenous antibiotic, antifungal, or antiviral treatment. ‡Patient was diagnosed with a malignancy. §Six patients were readmitted to the hospital ward: four due to
complaints of dyspnoea, one due to pneumonia unrelated to COVID-19, and one due to exacerbation of chronic obstructive pulmonary disease. ¶Patient was diagnosed with
a leukocytoclastic vasculitis, confirmed by skin biopsy.

Table 4: Grade 3 or 4 adverse events in the modified intention-to-treat population

policies changed over the course of the pandemic, with received dexamethasone as a co-treatment, indicating
early home or nursing home discharge on continued that the possible benefits of imatinib treatment have an
oxygen treatment30 only possible from the second half of additive effect to standard care. Finally, although imatinib
the year. These policy changes could have affected was originally developed as an outpatient therapy for
measurement of the primary outcome. Fourth, although chronic myeloid leukaemia,31 this study provides extensive
a significant difference in mortality at 28 days was safety data on imatinib in a severe, critically ill population.
observed between the imatinib group and the placebo An extensive safety analysis, which included laboratory
group, the study was not powered to detect significant and ECG tests, as well extensive adverse event reporting,
differences in mortality, precluding robust conclusions revealed that imatinib was well tolerated and safe in
from these analyses. Adjustment for baseline imbalances patients with moderate to severe COVID-19. In fact, only
also yielded a non-significant result; however, the HR for a few adverse events were observed, and no renal, hepatic,
mortality at 28 days remained mostly unchanged, or cardiac toxicity was reported.
indicating that it shows little residual confounding. The observed reductions in mortality and duration of
Finally, the study period of 28 days might be short mechanical ventilation in patients given imatinib are of
relative to other trials of treatments for COVID-19, given direct relevance. First, as our study suggests that imatinib
the high frequency of readmissions due to COVID-19 predominantly benefits patients with severe course of
and that several patients have an extended recovery COVID-19, future trials are needed in selected patients
period. Follow-up for 3 or 6 months, as has been done in with COVID-19-associated ARDS or in larger study
most studies of long-term COVID-19 sequelae, could populations, thus allowing stratification by disease
provide relevant information about the effects of severity. Of note, one trial of intravenous imatinib in
imatinib on the long-term outcomes and functional patients with COVID-19 on mechanical ventilation started
status of patients. patient recruitment in March, 2021 (NCT04794088).
The major strengths of our trial are its randomised, Second, a treatment period of 10 days, as used in our
placebo-controlled, double-blind design, and the fact that study, might need to be reconsidered. We based our
it was done in a day-to-day health-care setting. All choice of treatment duration on early observations that
hospitalised patients aged 18 years or older were eligible, respiratory deterioration in COVID-19 occurs within the
including patients aged older than 85 years and patients first 10 days after onset of symptoms.32 Time-to-event
with a do-not-resuscitate order. In addition, 72% of patients analyses, however, showed that the Kaplan-Meier curves

10 www.thelancet.com/respiratory Published online June 17, 2021 https://doi.org/10.1016/S2213-2600(21)00237-X

 Articles

for mechanical ventilation in the imatinib and placebo 8 RECOVERY Collaborative Group, Horby P, Lim WS, et al.
groups crossed after day 9 (figure 2C), which could Dexamethasone in hospitalized patients with COVID-19.
N Engl J Med 2020; 384: 693–704.
indicate a catch-up phenomenon after the final day of 9 REMAP-CAP Investigators, Gordon AC, Mouncey PR, et al.
study medication. The proportional hazards assumption Interleukin-6 receptor antagonists in critically ill patients with
was violated for the secondary outcome of the need for COVID-19. N Engl J Med 2021; 384: 1491–502.
10 Overbeek MJ, van Nieuw Amerongen GP, Boonstra A, Smit EF,
mechanical ventilation. This violation could have resulted Vonk-Noordegraaf A. Possible role of imatinib in clinical
in an underestimation of the treatment effect and pulmonary veno-occlusive disease. Eur Respir J 2008; 32: 232–35.
advocates for increasing the treatment duration. 11 Su EJ, Fredriksson L, Geyer M, et al. Activation of PDGF-CC by
tissue plasminogen activator impairs blood-brain barrier integrity
In conclusion, this randomised, placebo-controlled trial during ischemic stroke. Nat Med 2008; 14: 731–37.
showed that imatinib does not reduce the time to 12 Aman J, van Bezu J, Damanafshan A, et al. Effective treatment of
discontinuation of supplemental oxygen and mechanical edema and endothelial barrier dysfunction with imatinib.
ventilation in hospitalised patients with COVID-19. Circulation 2012; 126: 2728–38.
13 Chislock EM, Pendergast AM. Abl family kinases regulate
However, the reduction in mortality (even if attenuated endothelial barrier function in vitro and in mice. PLoS One 2013;
after correction for baseline imbalances) and duration of 8: e85231.
mechanical ventilation indicates that imatinib might 14 Letsiou E, Rizzo AN, Sammani S, et al. Differential and opposing
effects of imatinib on LPS- and ventilator-induced lung injury.
confer clinical benefit in patients with COVID-19, and Am J Physiol Lung Cell Mol Physiol 2015; 308: L259–69.
provide a rationale for further studies. 15 Langer V, Vivi E, Regensburger D, et al. IFN-γ drives inflammatory
Contributors bowel disease pathogenesis through VE-cadherin-directed vascular
barrier disruption. J Clin Invest 2019; 129: 4691–07.
JA, ED, FSdM, and HJB had full access to and verified the study data, and
take responsibility for the integrity of the data. JA, LB, AK, LAH, LDJB, 16 Carnevale-Schianca F, Gallo S, Rota-Scalabrini D, et al. Complete
resolution of life-threatening bleomycin-induced pneumonitis after
ASN, PIB, AVN, and HJB provided input on trial design. All authors were
treatment with imatinib mesylate in a patient with Hodgkin’s
involved in conducting the trial. ED and RJAH were involved in data lymphoma: hope for severe chemotherapy-induced toxicity?
management. FSdM statistically analysed the data. JA, ED, FSdM, AVN, J Clin Oncol 2011; 29: e691–93.
and HJB were involved in data interpretation and drafting the 17 Aman J, Peters MJ, Weenink C, van Nieuw Amerongen GP,
manuscript. All authors critically revised the manuscript, and JA and Vonk Noordegraaf A. Reversal of vascular leak with imatinib.
HJB had final responsibility for the decision to submit for publication. Am J Respir Crit Care Med 2013; 188: 1171–73.
Declaration of interests 18 Langberg MK, Berglund-Nord C, Cohn-Cedermark G, Haugnes HS,
JA and AVN are inventors on a patent (WO2012150857A1; 2011) covering Tandstad T, Langberg CW. Imatinib may reduce chemotherapy-
protection against endothelial barrier dysfunction through inhibition of induced pneumonitis. A report on four cases from the
SWENOTECA. Acta Oncol 2018; 57: 1401–06.
the tyrosine kinase abl-related gene (arg). JA reports serving as a
non-compensated scientific advisor for Exvastat. All other authors 19 Morales-Ortega A, Bernal-Bello D, Llarena-Barroso C, et al.
Imatinib for COVID-19: a case report. Clin Immunol 2020;
declare no competing interests.
218: 108518.
Data sharing 20 Hochhaus A, Saussele S, Rosti G, et al. Chronic myeloid leukaemia:
For sharing purposes, reuse conditions will be respected. The ESMO Clinical Practice Guidelines for diagnosis, treatment and
deidentified patient data will be accessible after publication of the Article, follow-up. Ann Oncol 2017; 28 (suppl 4): iv41–51.
and can be requested from the corresponding author (hj.bogaard@ 21 Abou-Arab O, Bennis Y, Gauthier P, et al. Association between
amsterdamumc.nl) by other researchers if reuse conditions are met. inflammation, angiopoietins, and disease severity in critically ill
COVID-19 patients: a prospective study. Br J Anaesth 2020;
Acknowledgments 126: e127–30.
This project was funded by an unrestricted grant from the Amsterdam 22 Smadja DM, Guerin CL, Chocron R, et al. Angiopoietin-2 as a
Medical Center Foundation and a bottom-up grant from Nederlandse marker of endothelial activation is a good predictor factor for
Organisatie voor Wetenschappelijk Onderzoek/ZonMW (10430 01 201 0007). intensive care unit admission of COVID-19 patients. Angiogenesis
In addition, this project received funding from the Innovative Medicines 2020; 23: 611–20.
Initiative 2 Joint Undertaking (grant agreement number 101005142). 23 Vassiliou AG, Keskinidou C, Jahaj E, et al. ICU admission levels of
endothelial biomarkers as predictors of mortality in critically ill
References
COVID-19 patients. Cells 2021; 10: 186.
1 Wiersinga WJ, Rhodes A, Cheng AC, Peacock SJ, Prescott HC.
Pathophysiology, transmission, diagnosis, and treatment of 24 Villa E, Critelli R, Lasagni S, et al. Dynamic angiopoietin-2
coronavirus disease 2019 (COVID-19): a review. JAMA 2020; assessment predicts survival and chronic course in hospitalized
324: 782–93. patients with COVID-19. Blood Adv 2021; 5: 662–73.
2 Varga Z, Flammer AJ, Steiger P, et al. Endothelial cell infection 25 van der Heijden M, van Nieuw Amerongen GP, Koolwijk P,
and endotheliitis in COVID-19. Lancet 2020; 395: 1417–18. van Hinsbergh VW, Groeneveld AB. Angiopoietin-2, permeability
oedema, occurrence and severity of ALI/ARDS in septic and non-
3 Ackermann M, Verleden SE, Kuehnel M, et al. Pulmonary vascular
septic critically ill patients. Thorax 2008; 63: 903–09.
endothelialitis, thrombosis, and angiogenesis in COVID-19.
N Engl J Med 2020; 383: 120–28. 26 Amado-Azevedo J, van Stalborch AD, Valent ET, et al. Depletion of
Arg/Abl2 improves endothelial cell adhesion and prevents vascular
4 Beigel JH, Tomashek KM, Dodd LE, et al. Remdesivir for the
leak during inflammation. Angiogenesis 2021; published online
treatment of COVID-19 - final report. N Engl J Med 2020;
March 26. https://doi.org/10.1007/s10456-021-09781-x.
383: 1813–26.
27 Coleman CM, Sisk JM, Mingo RM, et al. Abelson kinase inhibitors
5 WHO Solidarity Trial Consortium, Pan H, Peto R, et al.
are potent inhibitors of severe acute respiratory syndrome
Repurposed antiviral drugs for COVID-19 - interim WHO
coronavirus and Middle East respiratory syndrome coronavirus
Solidarity Trial results. N Engl J Med 2021; 384: 497–511.
fusion. J Virol 2016; 90: 8924–33.
6 RECOVERY Collaborative Group, Horby P, Mafham M, et al. Effect
28 Han Y, Duan X, Yang L, et al. Identification of SARS-CoV-2
of hydroxychloroquine in hospitalized patients with COVID-19.
inhibitors using lung and colonic organoids. Nature 2021;
N Engl J Med 2020; 383: 2030–40.
589: 270–75.
7 Cao B, Wang Y, Wen D, et al. A trial of lopinavir-ritonavir in adults
29 Zhao H, Mendenhall M, Deininger MW. Imatinib is not a potent
hospitalized with severe COVID-19. N Engl J Med 2020;
anti-SARS-CoV-2 drug. Leukemia 2020; 34: 3085–87.
382: 1787–99.

www.thelancet.com/respiratory Published online June 17, 2021 https://doi.org/10.1016/S2213-2600(21)00237-X 11

 Articles

30 Nederlands Huisartsen Genootschap. Leidraad zuurstofgebruik 32 Huang C, Wang Y, Li X, et al. Clinical features of patients infected
THUIS bij (verdenking op / bewezen) COVID-19. April 22, 2020. with 2019 novel coronavirus in Wuhan, China. Lancet 2020;
https://www.nhg.org/sites/default/files/content/nhg_org/ 395: 497–506.
uploads/050620_zuurstof_thuis_ten_tijde_van_corona_definitieve_
versie_1.1_nvalt-nhg-pznl.pdf (accessed Feb 16, 2021).
31 Druker BJ, Talpaz M, Resta DJ, et al. Efficacy and safety of a specific
inhibitor of the BCR-ABL tyrosine kinase in chronic myeloid
leukemia. N Engl J Med 2001; 344: 1031–37.

12 www.thelancet.com/respiratory Published online June 17, 2021 https://doi.org/10.1016/S2213-2600(21)00237-X