Research

JAMA Internal Medicine | Original Investigation

Effect of Tocilizumab vs Usual Care in Adults Hospitalized With COVID-19

and Moderate or Severe Pneumonia
A Randomized Clinical Trial
Olivier Hermine, MD, PhD; Xavier Mariette, MD, PhD; Pierre-Louis Tharaux, MD, PhD; Matthieu Resche-Rigon, MD, PhD; Raphaël Porcher, PhD;
Philippe Ravaud, MD, PhD; for the CORIMUNO-19 Collaborative Group

Visual Abstract
IMPORTANCE Severe pneumonia with hyperinflammation and elevated interleukin-6 is a Editorial page 12
common presentation of coronavirus disease 2019 (COVID-19).
Related articles pages 24 and
OBJECTIVE To determine whether tocilizumab (TCZ) improves outcomes of patients 41
hospitalized with moderate-to-severe COVID-19 pneumonia.
Supplemental content
DESIGN, SETTING, AND PARTICPANTS This cohort-embedded, investigator-initiated,
multicenter, open-label, bayesian randomized clinical trial investigating patients with
COVID-19 and moderate or severe pneumonia requiring at least 3 L/min of oxygen but
without ventilation or admission to the intensive care unit was conducted between March 31,
2020, to April 18, 2020, with follow-up through 28 days. Patients were recruited from 9
university hospitals in France. Analyses were performed on an intention-to-treat basis with
no correction for multiplicity for secondary outcomes.

INTERVENTIONS Patients were randomly assigned to receive TCZ, 8 mg/kg, intravenously plus
usual care on day 1 and on day 3 if clinically indicated (TCZ group) or to receive usual care
alone (UC group). Usual care included antibiotic agents, antiviral agents, corticosteroids,
vasopressor support, and anticoagulants.

MAIN OUTCOMES AND MEASURES Primary outcomes were scores higher than 5 on the World
Health Organization 10-point Clinical Progression Scale (WHO-CPS) on day 4 and survival
without need of ventilation (including noninvasive ventilation) at day 14. Secondary
outcomes were clinical status assessed with the WHO-CPS scores at day 7 and day 14, overall
survival, time to discharge, time to oxygen supply independency, biological factors such as
C-reactive protein level, and adverse events.

RESULTS Of 131 patients, 64 patients were randomly assigned to the TCZ group and 67 to UC
group; 1 patient in the TCZ group withdrew consent and was not included in the analysis. Of
the 130 patients, 42 were women (32%), and median (interquartile range) age was 64
(57.1-74.3) years. In the TCZ group, 12 patients had a WHO-CPS score greater than 5 at day 4
vs 19 in the UC group (median posterior absolute risk difference [ARD] −9.0%; 90% credible
interval [CrI], −21.0 to 3.1), with a posterior probability of negative ARD of 89.0% not
achieving the 95% predefined efficacy threshold. At day 14, 12% (95% CI −28% to 4%) fewer
patients needed noninvasive ventilation (NIV) or mechanical ventilation (MV) or died in the
TCZ group than in the UC group (24% vs 36%, median posterior hazard ratio [HR] 0.58; 90%
CrI, 0.33-1.00), with a posterior probability of HR less than 1 of 95.0%, achieving the
predefined efficacy threshold. The HR for MV or death was 0.58 (90% CrI, 0.30 to 1.09). At
day 28, 7 patients had died in the TCZ group and 8 in the UC group (adjusted HR, 0.92; 95%
CI 0.33-2.53). Serious adverse events occurred in 20 (32%) patients in the TCZ group and 29 Author Affiliations: Author
(43%) in the UC group (P = .21). affiliations are listed at the end of this
article.
CONCLUSIONS AND RELEVANCE In this randomized clinical trial of patients with COVID-19 and
Group Information: The members of
pneumonia requiring oxygen support but not admitted to the intensive care unit, TCZ did not the CORIMUNO-19 Collaborative
reduce WHO-CPS scores lower than 5 at day 4 but might have reduced the risk of NIV, MV, or Group appear at the end of the
article.
death by day 14. No difference on day 28 mortality was found. Further studies are necessary
for confirming these preliminary results. Corresponding author: Olivier
Hermine, MD, PhD, Université de
TRIAL REGISTRATION ClinicalTrials.gov Identifier: NCT04331808 Paris, Assistance Publique-Hôpitaux
de Paris (AP-HP), Hôpital Necker,
INSERM, Imagine Institute, 149-161
JAMA Intern Med. 2021;181(1):32-40. doi:10.1001/jamainternmed.2020.6820 rue de Sèvres, Paris 75743, France
Published online October 20, 2020. Last corrected on May 3, 2021. ([email protected]).

32 (Reprinted) jamainternalmedicine.com

© 2020 American Medical Association. All rights reserved.

Downloaded From: https://jamanetwork.com/ on 04/02/2023

 Effect of Tocilizumab vs Usual Care in Adults Hospitalized With COVID-19 and Moderate or Severe Pneumonia Original Investigation Research

C
oronavirus disease 2019 (COVID-19) is a respiratory dis-
ease induced by a novel coronavirus (severe acute re- Key Points
spiratory syndrome coronavirus 2 [SARS-CoV-2]) caus-
Question What is the effect of tocilizumab, an anti–interleukin-6
ing substantial morbidity and mortality.1-4 Most people with receptor antibody, in patients with COVID-19 and
COVID-19 have only mild or uncomplicated symptoms, but ap- moderate-to-severe pneumonia?
proximately 10% to 15% have moderate or severe disease that
Findings In this randomized clinical trial that included 130
requires hospitalization and oxygen support, and 3% to 5% re-
patients hospitalized with COVID-19 and moderate-to-severe
quire admission to an intensive care unit (ICU).5 In severe cases, pneumonia, tocilizumab did not reduce the World Health
COVID-19 can be complicated by acute respiratory distress syn- Organization 10-point Clinical Progression Scale scores lower than
drome (ARDS). Older age, male sex, and comorbid diseases are 5 at day 4, and the proportion of patients with noninvasive
risk factors for death.6-8 ventilation, intubation, or death at day 14 was 36% with usual care
A previous trial showed that the antiviral agent remdesi- and 24% with tocilizumab. No difference in mortality over 28 days
was found between the 2 groups.
vir reduced the length of recovery by 4 days but did not re-
duce the number of patients needing mechanical ventilation Meaning Tocilizumab may reduce the need for mechanical and
(MV), nor the death rate.9 In addition, the RECOVERY collab- noninvasive ventilation or death by day 14 but not mortality by day
orative group demonstrated that dexamethasone (DXM) 6 mg/d 28; further studies are necessary to confirm these preliminary
results.
for 10 days decreased the 28-day mortality among patients re-
ceiving MV or oxygen.10
At the beginning of the epidemic in France, when no stan- cal trials testing different therapeutic regimens (CORI-
dard of care was defined, we decided to set up the publicly sup- MUNO-19 Cohort). Two separate populations were recruited:
ported CORIMUNO-19 platform dedicated to perform cohort, patients with moderate or severe pneumonia and patients with
open-label, randomized controlled trials of immune modula- critical pneumonia. An institutional review board-approved
tory drugs in hospitalized patients with moderate or severe amendment to the protocol on April 6, 2020, clarified the defi-
COVID-19. The overall objective was to explore several im- nition of these 2 populations as follows (eMethods in Supple-
mune modulatory drugs to design larger trials to confirm the ment 2): (1) patients with moderate or severe pneumonia and
best drug in a defined population of patients with severe with WHO 10-point Clinical Progression Scale (WHO-CPS) score
COVID-19 pneumonia and eventually new standard of care. of 5 receiving at least 3L/min oxygen (O2) but without high-
Patients with severe COVID-19 pneumonia present nonspe- flow oxygen (HFO) (defined by using Optiflow device with more
cific inflammatory responses, including edema and inflamma- than 15 L/min O2), noninvasive ventilation (NIV) or mechani-
tory cell infiltration in the lungs. Besides the specific pathogenic cal ventilation (MV) and (2) patients with critical pneumonia de-
effect of SARS-CoV-2, this deleterious excessive and noneffective fined as WHO-CPS score of 6 or more (ie, with HFO, NIV, or MV).
host immune response plays an important role during the disease This article reports on CORIMUNO-TOCI 1, a CORIMUNO,
course. It is related to a hyperinflammatory status comprising a multicenter, open-label, randomized clinical trial in patients
number of proinflammatory cytokines and chemokines, one of with moderate or severe pneumonia. The trial registration,
the most predominant being interleukin 6 (IL-6).11,12 NCT04331808, describes this study and a second study, CORI-
Given the potential deleterious effect of IL-6 in COVID-19 MUNO-TOCI 2, a trial conducted in patients with critical pneu-
pneumonia,13-15 we first evaluated the benefit-risk effect of to- monia. Although the 2 trials were placed under the same reg-
cilizumab (TCZ), an anti–human IL-6 receptor (IL-6R) mono- istration number, they are distinct studies with different
clonal antibody that inhibits IL-6 signaling by binding soluble participants, and were never intended to be reported together.
IL-6R and membrane IL-6R and is approved for rheumatoid Accrual for this study took place in 9 French university hos-
arthritis, juvenile inflammatory arthritis and refractory giant pitals. Because of the emergency nature of the trial and feasi-
cell arteritis. Tocilizumab is also approved for systemic in- bility issues, no placebo of TCZ was prepared.
flammatory response caused by the massive release of proin- The CORIMUNO Cohort and all embedded trials (ie, trials
flammatory cytokines in response to iatrogenic disease (eg, chi- using data collected in the CORIMUNO cohort) were ap-
meric antigen receptor T-cell therapies). Preliminary data from proved by an ethics committee (CPP Île-de-France VI) and rel-
observational studies suggested the possible efficacy of TCZ evant authorities. Legal issues and trial procedures are pre-
for moderate, severe, or critical COVID-1916-18 but, until now, sented in detail in the eMethods in Supplement 2. Written
no data are available from randomized clinical trials. We set informed consent was obtained from all patients or from the
up this multicenter randomized clinical trial assessing the abil- patient’s legal representative for entering the CORIMUNO co-
ity of TCZ to improve the outcome of patients hospitalized with hort, and longitudinal data (including clinical status, biologi-
moderate-to-severe COVID-19 pneumonia. cal data and outcomes) were recorded as part of their partici-
pation in the cohort. In this consent, patients were made aware
that a number of trials may occur via the cohort, and that they
would likely be offered to participate in some of them. In prac-
Methods tice, for logistical reasons, only 1 trial took place at each site
Trial Design and Study Oversight at a given time. A specific additional written consent was ob-
The trial protocol is available in Supplement 1. We enrolled pa- tained from eligible patients who were randomly selected to
tients with COVID-19 to perform a series of randomized clini- be offered TCZ and agreed to receive this treatment. Eligible

jamainternalmedicine.com (Reprinted) JAMA Internal Medicine January 2021 Volume 181, Number 1 33

© 2020 American Medical Association. All rights reserved.

Downloaded From: https://jamanetwork.com/ on 04/02/2023

 Research Original Investigation Effect of Tocilizumab vs Usual Care in Adults Hospitalized With COVID-19 and Moderate or Severe Pneumonia

patients assigned to receive usual care (UC) were not notified Data Quality Monitoring
about the trial, but their CORIMUNO-cohort data were avail- Data quality monitoring included both remote data monitor-
able for analysis. An amendment to the protocol on April 6, ing and on-site monitoring performed by dedicated staff in-
2020, modifying inclusion criteria and the late primary out- dependent of the site investigators, with 100% source data veri-
come is described in the eMethods 2.5 in Supplement 2. A data fication performed for all patients recruited at every site for
safety monitoring board (DSMB) was set up at the beginning all critical data points.
of the study and resigned on April 30, 2020, because of dif-
ferences between the investigators and sponsors and the DSMB Statistical Analysis
with regard to the management of the protocol and the com- To maximize information from limited data generated while
munication of the results. No issues of participant safety or data allowing for a rapid decision, we used a bayesian monitoring
integrity were raised. A new DSMB was appointed on May 1, and analysis of the trial based on the coprimary outcomes. The
2020, which was approved by the Agence Nationale de Sécu- sample size was set at 120, with interim analyses presented
rité du Médicament et des Produits de Santé on May 3, 2020. weekly to the DSMB and a provision to increase the sample size
This trial was reported according to Consolidated Stan- in case of promising but not conclusive results. We computed
dards of Reporting Trials (CONSORT) reporting guidelines. that the trial would have frequentist power of 97.2% to detect
a decrease in event rate from 0.50 to 0.20, and 73.9% to de-
Patients tect a decrease in event rates from 0.50 to 0.30. For the day 4
Patients were included in the CORIMUNO-19 cohort if they had outcome, we used a beta prior distribution with parameters 1
confirmed SARS-CoV-2 infection (positive on rRT-PCR and/or and 1 for the proportion in each arm (eFigure 1 in Supple-
typical chest computed tomographic [CT] scan) with moder- ment 2). For the day 14 outcome, we used a Gaussian prior dis-
ate, severe, or critical pneumonia (O2 >3 L/min, WHO Clinical tribution with a mean of 0 and variance of 106 for the log haz-
Progression Scale [WHO-CPS] score ≥519 [10-point ordinal scale ard ratio (HR) (eTable 1 in Supplement 2). Sensitivity analyses
described in the eMethods 2.4 in Supplement 2]). using a range of prior distribution were then conducted
Patients from the CORIMUNO-19 cohort were eligible for (eMethods, eFigure 2 in Supplement 2). The treatment effect
CORIMUNO-TOCI-1 trial if they had a WHO-CPS score of 5 with was expressed in terms of absolute risk difference (ARD) for
O2 levels of 3 L/min or higher but without noninvasive venti- the day 4 outcome and HR for the day 14 outcome. Using
lation (NIV) or mechanical ventilation (MV). Exclusion crite- Markov chain Monte Carlo methods, posterior probabilities of
ria are detailed in the eMethods in Supplement 2. negative ARD and HR less than 1 were computed. According
to the protocol, posterior probability greater than 0.99 at the
Randomization and Treatments interim analysis or greater than 0.95 at the final analysis in-
Participants were randomly assigned in a 1:1 ratio to receive dicated efficacy. We also computed posterior probability of ARD
TCZ plus usual care (TCZ group) or usual care alone (UC group) less than −5.5% and HR less than 0.85 (denoting moderate or
via a web-based secure centralized system. An independent greater effect). Because the decision rules are 1-sided, consis-
statistician provided a computer-generated assignment ran- tent credible intervals (CrIs) would be theoretical. However,
domization list stratified by center and blocked with varying we chose to report 2-sided 90% CrIs, which have the same up-
block sizes unknown to the investigators. per bound as 1-sided 95% CrIs. A subgroup analysis according
Tocilizumab was administered intravenously (IV) at to antiviral drug use at baseline was prespecified in the pro-
8 mg/kg on day 1. Administration of an additional fixed dose of tocol. Analyses according to the use of corticosteroids or DXM
TCZ, 400 mg IV, on day 3 was recommended if oxygen require- were added post-hoc in light of recent publications. Second-
ment was not decreased by more than 50%, but decision was left ary outcomes were analyzed in a frequentist framework, ex-
to the treating physician. Usual care (antibiotic agents, antivi- cept the analysis of the WHO-CPS scores as an ordinal vari-
ral agents, corticosteroids, vasopressor support, anticoagu- able. The Statistical Analysis Plan and details of the statistical
lants) was provided at the discretion of the clinicians. analyses are in Supplement 1.
Analyses were performed on an intention-to-treat basis
Outcome Measures with no correction for multiplicity for secondary outcomes.
The 2 primary outcomes were (1) the proportion of patients Thus, these results are exploratory and reported as point es-
dead or needing noninvasive or mechanical ventilation on day timates and 95% confidence intervals (CIs). Statistical analy-
4 (>5 on the WHO-CPS); and (2) survival with no need for non- ses involved using SAS (version 9.4, SAS Institute) and R (ver-
invasive or mechanical ventilation at day 14. The day 4 and 14 sion 3.6.1, R Foundation) statistical software.
outcomes were amended on April 6, 2020, to include high-
flow oxygen in noninvasive ventilation to be consistent with
the WHO-CPS definition. Both outcomes were consistent with
the core outcome set proposed by the WHO19 (eMethods 2.4
Results
in Supplement 2). Prespecified secondary outcomes were clini- Patients
cal status assessed with the WHO-CPS at day 7 and day 14, over- From March 31 to April 18, 2020, 131 patients were random-
all survival, time to discharge, and time to oxygen supply in- ized (64 patients to the TCZ group and 67 to the UC group),
dependency. We also measured biological factors such as and the DSMB did not advise further increasing the sample
C-reactive protein levels and adverse events. size. Among the 64 patients assigned to receive TCZ, 1 with-

34 JAMA Internal Medicine January 2021 Volume 181, Number 1 (Reprinted) jamainternalmedicine.com

© 2020 American Medical Association. All rights reserved.

Downloaded From: https://jamanetwork.com/ on 04/02/2023

 Effect of Tocilizumab vs Usual Care in Adults Hospitalized With COVID-19 and Moderate or Severe Pneumonia Original Investigation Research

drew consent and was not analyzed, and 3 did not receive
Figure 1. Study Flowcharta
TCZ because of death (n = 1), technical problems (n = 1), and
patient refusal (n = 1). Among the 60 with TCZ treatment, 28 131 Patients assessed for eligibility
(47%) received a second injection on day 3 (Figure 1). Demo-
graphic and baseline clinical and biological characteristics of
131 Randomized in 9 centers
patients are described in Table 1. The median age was 64
years (interquartile range, 57.1 to 74.3 years), and 88 (68%)
were men. There were no important between-group differ- 64 Randomized to tocilizumab + usual 67 Assigned to usual care at day 1
care at day 1
ences at enrollment.
During the trial, antiviral drugs, glucocorticoids, and pre-
59 Received injection of tocilizumab
ventive or therapeutic anticoagulants were administered in 7 at day 1
(11%), 21 (33%), and 59 (94%) patients, respectively, in the TCZ 28 Received injection
of tocilizumab at day 3
group, and 16 (24%), 41 (61%), and 61 (91%) in the UC group, 1 Received injection of tocilizumab
respectively. Additional immunomodulators were adminis- at day 3
3 Did not receive any injection
tered to 1 patient in the TCZ group (anakinra) and 4 in the UC of tocilizumab
group (anakinra, n = 3; eculizumab, n = 1). The details of the 1 Technical problem
1 Respiratory deterioration prior
treatments received at the time of and after randomization un- to administration
til day 14 are summarized in eTable 2 in Supplement 2. 1 Patient refusal

Primary Outcomes 8 Lost to follow-up at day 28 3 Lost to follow-up at day 28

1 Discharged alive at day 9 3 Discharged alive with last
On day 4, 12 of 63 (19%) patients randomized to receive TCZ 1 Discharged alive at day 13 contact at day 14
had a WHO-CPS score higher than 5 vs 19 of 67 (28%) in the 6 Discharged alive between
day 15 and 28
UC group (median posterior ARD, −9%; 90% CrI, –21 to 3)
(eTable 3 in Supplement 2). The posterior probability of nega-
63 Were analyzed for primary 67 Were analyzed for primary
tive ARD (TCZ better than UC) was 89.0% and ARD less than outcomes outcomes
−5.5% was 68.4%.
On day 14, at least 1 event (NIV, HFO, MV, or death) had oc- a
One patient withdrew consent and asked data to be erased. According to
curred in 15 patients in the TCZ group (24%) (cumulative in- European regulation, no data were analyzed for this patient.
cidence of event 24%; 95% CI, 13% to 35%) and 24 patients in
the UC group (cumulative incidence 36%; 95% CI, 33%-58%)
(Figure 2A; Table 2; eTable 4 in Supplement 2). The posterior ARDS) occurred in the TCZ group and 11 (ARDS, n = 9; multi-
probability of any efficacy of TCZ (HR<1) was 95.0%, and of organ failure, n = 1; pulmonary embolism, n = 1) in the UC
moderate or greater efficacy (HR<0.85) was 87.4% (posterior group. Of note, 3 deaths occurred after day 28 in the UC group.
median HR, 0.58; 90% CrI, 0.33-1.00) (eTable 5 in Supple- Causes of death are shown in Table 3.
ment 2). Results on both primary outcomes were similar in the The cumulative incidence of patients who have been
subgroup of patients with rRT-PCR-confirmed SARS-CoV-2 in- weaned from oxygen at day 28 was 89% (95% CI, 78%-95%)
fection (eTable 6 in Supplement 2). and 75% (95% CI, 62%-83%) in the TCZ and UC group, respec-
The number of patients with MV or death at Day 14 was 11 tively (HR, 1.41; 95% CI, 0.98-2.01) (eTable 9 in Supple-
(17%) and 18 (27%) in the TCZ and UC groups. The posterior ment 2). The cumulative incidence of discharge by day 28 was
probability of HR less than 1 and HR less than 0.85 was 92.5% 83% (95% CI, 70%-90%) and 73% (95% CI, 61%-82%), respec-
and 84.4%, respectively (posterior median HR, 0.58; 90% CrI, tively (HR, 1.52; 95% CI, 1.02-2.27) (eTable 9 in Supple-
0.30-1.09) (Figure 2B). ment 2).
On prespecified (for antiviral drugs) or post-hoc sub-
group analyses (for corticosteroids, including DXM), the Biological Response
effect of TCZ was numerically higher if combined with anti- C-reactive protein level and neutrophil count decrease
viral drugs (HR, 0.28; 90% CrI, 0.07-1.06) or corticosteroids was rapid in the TCZ arm, and lymphocyte count was
(HR, 0.38; 90% CrI, 0.13-1.11) (eFigure 3 in Supplement 2). increased (eFigure 5 in Supplement 2). No patient in the TCZ
group remained with high C-reactive protein level after
Secondary Outcomes day 4.
The evolution of WHO scores during 14-day follow-up is given
eFigure 4 and eTable 7 in Supplement 2. Among patients who Safety
were not in ICU at randomization, 11 of 60 (18%) in the TCZ A total of 28 (44%) and 36 (54%) patients in the TCZ and UC
group and 22 of 64 (36%) in the UC group were subsequently groups reported adverse events between randomization and
admitted to the ICU (risk difference, 18%; 95% CI, 0.4%-31%, day 28 (Table 3). Serious adverse events occurred in 20 (32%)
this analysis was unplanned). At day 28, 7 patients had died in the TCZ group and 29 (43%) in the UC group (P = .21). The
in the TCZ group and 8 in the UC group (adjusted HR, 0.92; 95% number of serious adverse events was lower in the TCZ than
CI, 0.33-2.53) (Figure 2C; Table 3; eTable 8 in Supplement 2). UC group (27 vs 57) with a decreased incidence of serious bac-
Overall, with a median follow-up of 28 days, 7 deaths (all from terial infections (2 vs 11).

jamainternalmedicine.com (Reprinted) JAMA Internal Medicine January 2021 Volume 181, Number 1 35

© 2020 American Medical Association. All rights reserved.

Downloaded From: https://jamanetwork.com/ on 04/02/2023

 Research Original Investigation Effect of Tocilizumab vs Usual Care in Adults Hospitalized With COVID-19 and Moderate or Severe Pneumonia

Table 1. Patient Characteristics at Baseline

Tocilizumab, Usual care,
Characteristic No. No./No. (%) No. No./No. (%)
No. 63 67
Age, median (IQR), y 64.0 (57.1-74.3) 63.3 (57.1-72.3)
Male 44/63 (70) 44/67 (66)
Female 19/63 (30) 23/67 (34)
Weight, median (IQR), kg 80.0 (70.0-90.0) 55 78.0 (70.0-90.0)
BMI,a median (IQR) 46 27.9 (23.3-30.8) 46 27.4 (24.5-31.3)
WHO-CPS score (0-10) = 5 63/63 (100) 67/67 (100)
rRT-PCR–confirmed SARS-CoV-2 infection 56/63 (89) 61/67 (90)
Temperature, median (IQR), °C 37.3 (36.8-38.2) 37.9 (37.0-38.6)
Respiratory rate, median (IQR), bpm 56 24.0 (22.0-30.0) 57 26.0 (24.0-30.0)
Flow, median (IQR), L/min 5.0 (3.0-8.0) 5.0 (3.0-6.0)
SpO2, median (IQR), % 95.0 (93.0-96.0) 95.0 (93.0-97.0)
Time from symptoms onset to randomization, median 62 10 (7-13) 66 10 (8-13)
(IQR), d
Time from hospital admission to randomization, 63 1 (1-4) 67 1 (1-2)
median (IQR), d
Coexisting conditions
Chronic cardiac disease 20/61 (33) 20/67 (30.0)
Diabetes 20/61 (33) 23/67 (34)
Chronic kidney disease (stage 1 to 3) or dialysis 5/61 (8) 13/67 (19)
Asthma 5/61 (8) 3/67 (5)
Chronic pulmonary disease (not asthma) 3/61 (5) 3/67 (5)
Active malignant neoplasm 4/61 (7) 5/67 (8)
Smoking
No 55/61 (90) 62/67 (93)
Current 1/61 (2) 2/67 (3)
Former 5/61 (8) 3/67 (4)
Laboratory values, median (IQR)
CRP, mg/L 56 119.5 (74.5-219.5) 63 127.0
(84.0-171.0)
D-Dimer, μg/L 50 869 (524-1380) 50 1250 (780-1812)
Neutrophil count, G/L 60 4.9 (3.9-7.5) 63 5.1 (3.4-6.6)
Lymphocyte count, G/L 60 1.0 (0.7-1.4) 60 1.1 (0.6-1.2)
Abbreviations: ALD, alanine
Lymphocytes to neutrophils ratio 48 0.2 (0.1-0.3) 40 0.2 (0.1-0.3) aminotransferase; ALT/SGOT, alanine
Hemoglobin, g/dL 62 12.8 (11.9-13.8) 65 12.3 (10.9-13.4) transaminase; AST, aspartate
Platelet count, g/L 62 230 (187-324) 65 226 (163-286) aminotransferase; BPM, breaths per
min; CPK, creatine phosphokinase;
ALT/SGPT, IU/L 57 40.0 (30.0-67.0) 62 35.0 (22.0-55.0) CRP, C-reactive protein;
AST/SGOT, IU/L 58 50.0 (34.0-66.0) 62 55.0 (36.0-74.0) IQR, interquartile range; LDH, lactate
dehydrogenase, rRT-PCR, real-time
Albumin, g/L 43 30.0 (27.0-36.0) 42 32.2 (28.0-36.0)
reverse transcription polymerase
Creatinine, μmol/L 61 71.0 (56.0-87.0) 64 75.0 (59.5-119.5) chain reaction; SpO2, oxygen
Blood urea, mmol/L 62 5.8 (4.4-7.7) 65 5.1 (4.2-8.6) saturation; WHO-CPS World Health
Organization 10-point Clinical
Ferritin, mg/L 43 1292 (424-2484) 46 1070 (563-1790)
Progression Scale.
LDH, IU/L 46 401 (313-582) 51 434 (351-558) a
Body mass index, calculated as
CPK, IU/L 42 136.0 (48.0-284.0) 41 105.0 weight in kilograms divided by
(67.0-236.0) height in meters squared

who did not on WHO-CPS scores greater than 5 at day 4. The

Discussion proportion of patients with NIV, HFO, MV or death at day 14
was reduced with TCZ. The probability that TCZ reduced the
In this trial embedded in the CORIMUNO-19 cohort, we have risk of with NIV, HFO, MV, or death at day 14 was 95%, and the
tested the role of TCZ given to patients with COVID-19 and mod- probability that this risk reduction would be higher than 15%
erate-to-severe pneumonia who were neither admitted to an (HR<0.85), indicating moderate or greater benefit, was 87%.
ICU nor required high-flow, NIV, or MV. There were no signifi- However, the posterior median HR was 0.58 with 90% CrI of
cant differences between those who received TCZ and those 0.33-1.00. We did not observe any effect on mortality at day

36 JAMA Internal Medicine January 2021 Volume 181, Number 1 (Reprinted) jamainternalmedicine.com

© 2020 American Medical Association. All rights reserved.

Downloaded From: https://jamanetwork.com/ on 04/02/2023

 Effect of Tocilizumab vs Usual Care in Adults Hospitalized With COVID-19 and Moderate or Severe Pneumonia Original Investigation Research

Figure 2. Occurrence of Primary Outcome Events During Follow-up

A Probability of primary outcome, death or MV, HFO, or NIV, at day 14

0.6 Parameter Value

Median HR 0.58
Proportion with primary event
0.5
90% CrI 0.33-1.00
0.4 95% CrI 0.30-1.11
Usual care
P (HR <1) 0.95
0.3 P (HR <0.95) 0.93
Tocilizumab
P (HR <0.85) 0.87
0.2
P (HR <0.8) 0.83
0.1

0
1 2 3 4 5 6 7 8 9 10 11 12 13 14
Day
No. at risk
Tocilizumab 63 62 58 55 51 50 48 48 48 48 48 48 48 48
Usual care 67 64 53 49 48 46 45 44 44 44 44 43 43 43

B Probability of death or MV at day 14

0.6 Parameter Value
Median HR 0.58
Proportion with primary event

0.5
90% CrI 0.30-1.09
0.4 95% CrI 0.26-1.23
P (HR <1) 0.925
0.3 Usual care P (HR <0.95) 0.903
P (HR <0.85) 0.844
0.2 Tocilizumab
P (HR <0.8) 0.804
0.1

0 Kaplan-Meier cumulative estimates of

1 2 3 4 5 6 7 8 9 10 11 12 13 14
Day
probability of (A) the primary
No. at risk
outcome, death or ventilation
Tocilizumab 63 63 60 58 55 55 54 54 52 52 52 52 52 52 support (mechanical ventilation,
Usual care 67 67 60 58 56 54 51 50 50 50 50 49 49 49 high-flow or noninvasive ventilation);
B, death or mechanical ventilation;
C Probability of overall survival at day 28 (C) overall survival in the tocilizumab
arm compared with the usual care
1.0 arm. In panel A, events occurring on
Tocilizumab
day 1 occurred on the same day as but
Proportion with primary event

0.8 Usual care after randomization. For the primary

event and death or mechanical
0.6 ventilation, data are analyzed in a
bayesian framework, and median
0.4 posterior HR and 90% credible
intervals (CrIs) are presented,
together with posterior probabilities
0.2
of achieving specified effects, in the
tables on the right. Overall survival
0 was analyzed in a frequentist
1 4 7 10 14 21 28
framework, so these probabilities
Day
No. at risk
were not relevant. HFO indicates
Tocilizumab 63 62 60 55 54 50 46 high-flow oxygen; MV, mechanical
Usual care 67 66 64 61 61 56 50 ventilation; NIV, noninvasive
ventilation.

28. According to the protocol, patients were followed until day Some observational studies have suggested a possible ef-
90, and long-term results will be reported separately. Even if ficacy of TCZ for patients with moderate, severe, or critical SARS-
TCZ had no impact at day 28 on mortality, it may have an in- CoV2 infection in China16 and France.17 In a larger retrospec-
terest both at the individual level by reducing the need for in- tive study from Italy, TCZ treatment was associated with a 39%
tensive care, which is known to increase the risk of long-term reduction in the need for invasive MV or death, despite an in-
complications, eventually death, and decrease health- creased frequency of new severe infections (13% vs 4%).18 The
related quality of life. In addition, at the collective level, TCZ overall safety of TCZ was good in our study, with no increase in
use may limit the burden on ICUs, an important issue be- adverse or serious adverse events. The TCZ group showed a sur-
cause a shortage of ICU beds was observed in several geo- prisingly lower rate of serious infections despite decreased neu-
graphical areas at the peak of the pandemic. trophil count and increased rate of neutropenia. These results

jamainternalmedicine.com (Reprinted) JAMA Internal Medicine January 2021 Volume 181, Number 1 37

© 2020 American Medical Association. All rights reserved.

Downloaded From: https://jamanetwork.com/ on 04/02/2023

 Research Original Investigation Effect of Tocilizumab vs Usual Care in Adults Hospitalized With COVID-19 and Moderate or Severe Pneumonia

Table 2. Number of Patients With Noninvasive Ventilation or High-Flow Oxygen, Mechanical Ventilation,
or Death
Tocilizumab Usual care
Variable (n = 63) (n = 67) Difference (95% CI)
Primary outcome by day 14, No. 15 24
Cumulative incidence, % (95% CI) 24 (13 to 34) 36 (23 to 46) −12 (−28 to 4)
First event, No.
NIV/HFO 8 13
MV 3 8
Death/DNR order 4 3
MV or death by day 14, No.
% (95% CI) 17 (8 to 26) 27 (15 to 37) −9 (−24 to 5)
First event, No.
MV 5 14
Death/DNR order 6 4
Deaths
Day 14, No. 7 6
Survival, % (95% CI) 89 (81 to 97) 91 (84 to 98) Abbreviations: DNR, do not
Day 28, No. 7 8 resuscitate; HFO, high-flow oxygen;
MV, mechanical ventilation;
Survival, % (95% CI) 89 (81 to 97) 88 (80 to 96)
NIV, noninvasive ventilation.

Table 3. Serious Adverse Events and Causes of Deaths

Adverse events, No. (%)

Tocilizumab Usual care
Event (n = 63) (n = 67) P value
Patients with at least 1 AE 28 (44) 36 (54) 30a
Patients with multiple AEs 16 (25) 19 (28)
No. of events 66 86 .21b
Serious adverse events
Patients with at least 1 SAE 20 (32) 29 (43) .21a
Patients with multiple SAEs 5 (8) 10 (15)
No. of events 26 57
Atrial fibrillation 0 1
Anemia 1 4
Hyperlipasemia 0 1
Cholestasis 0 2
Hepatic cytolysis 4 4
Multiple organ failure (death) 0 1 (1)
Pulmonary embolism (death) 0 3 (1)
Fever 2 0
Hyperkaliemia 0 1
Hypoglycemia 0 1
Hypertension 1 0
.003b
Acute renal failure 1 2
Arterial ischemia 0 2
Lymphopenia 1 0
Neutropenia 4 0
Pneumothorax 0 1 Abbreviations: AE, adverse event;
ARDS, acute respiratory distress
ARDS (death) 9 (7) 19 (9)c syndrome; SAE, serious adverse
Bacterial sepsis 2 11 event.
a
Fungal sepsis 0 2 Fisher exact test.
b
Viral sepsis 0 1 Poisson model.
c
Tetraparesis 0 1 For 1 patient, the cause of death was
reported as both ARDS and
Cough 1 0
hemorrhagic shock.

38 JAMA Internal Medicine January 2021 Volume 181, Number 1 (Reprinted) jamainternalmedicine.com

© 2020 American Medical Association. All rights reserved.

Downloaded From: https://jamanetwork.com/ on 04/02/2023

 Effect of Tocilizumab vs Usual Care in Adults Hospitalized With COVID-19 and Moderate or Severe Pneumonia Original Investigation Research

might be explained by the decreased frequency of transfer to partment than the ward clinician having included the pa-
the ICU, and the more frequent use of steroid treatment. Like- tient. Furthermore, even using a placebo, the credibility of
wise, we found no increase in frequency of other adverse events blinding caregivers could have been debated in some cases
such as hepatitis, cardiovascular events or kidney failure. given the early evident effect of TCZ on C-reactive protein lev-
Strengths of this trial include the multicenter design, a thor- els for most patients. Lack of blinding may also lead to perfor-
ough monitoring to ensure data quality and a homogeneous mance bias and patients assigned to the UC group received cor-
target population of patients with moderate-to-severe pneu- ticosteroids—and especially DXM—twice as often as those in
monia requiring at least 3 L/min oxygen support. This later the TCZ group. However, if any related bias exists, it would
point is crucial since drugs like corticosteroids and IL-6 in- probably favor the UC arm rather than TCZ because DXM im-
hibitors might hurt if employed too early, or be ineffective if proved outcomes of similar patients in the RECOVERY trial. An-
employed too late. Finding the sweet spot, if one exists, will other limitation is that UC could differ among centers and over
probably require multiple trials. time. However, the short period of accrual and the stratifica-
In addition, the promising results of DXM in the RECOVERY tion of randomization may have limited the effect of such lack
trial highlight the need to address the relative and combined ef- of standardization. The sample size was small, credibility
fects of TCZ and DXM. Finding the optimal timing and combina- intervals were wide, and the treatment effect may be
tion, if one exists, will probably require multiple trials. overestimated.22 Also, this study did not allow evaluation of
what would be the effect of TCZ used earlier in the disease
Limitations course. Lastly, we targeted in this trial a narrow segment of the
At the beginning of the first COVID-19 pandemic wave, the COVID-19 patient population (patients with a WHO-CPS score
CORIMUNO-19 cohort was designed to allow for performing of 5 exactly and requiring at least 3 L/min oxygen), and these
quickly several exploratory clinical trials to test some of the results are not generalizable to other populations.
many drug candidates of potential interest. Setting up such
trials necessitated an adaptation to the crisis context while pre-
serving the major ethical and methodological principles.20,21
Consequently, this trial has a number of limitations. The trial
Conclusions
was not blinded because it was logistically impossible at the In this randomized clinical trial of patients with COVID-19 and
time of the pandemic to set up a double-blind study quickly. pneumonia requiring oxygen support but not admitted to the
This probably explains why all the academic-sponsored ran- intensive care unit, TCZ did not reduce WHO-CPS scores less
domized clinical trials published until now in COVID-19 infec- than 5 at day 4 but might have reduced the risk of NIV, MV, or
tion were open randomized clinical trials. Unblinding could death by day 14. No difference on day 28 mortality was found.
lead to measurement bias. However, it is unlikely that inves- These findings should be confirmed with a larger random-
tigators knowing which patients have been allocated to a par- ized clinical trial with longer follow-up. Future studies are
ticular therapy would have based their subsequent therapeu- needed to help determine which group of patients derive the
tic decisions (eg, use MV, HFO, NIV, or turn around ventilation) greatest benefit from the drug and whether combined therapy
on such knowledge. Moreover, in most cases, the clinician tak- with corticosteroids or antiviral agents may further improve
ing the decision of ventilation assistance was from another de- outcomes.

ARTICLE INFORMATION URC Saint-Louis, AP-HP, Hôpital Saint-Louis, 75010, Conflict of Interest Disclosures: Dr Tharaux has
Accepted for Publication: October 1, 2020. France (Resche-Rigon); Université de Paris, Centre received honorarium fees for participation on
of Research Epidemiology and Statistics (CRESS), advisory boards for Retrophin Inc not related to this
Published Online: October 20, 2020. INSERM U1153, Paris, France (Porcher, Ravaud); work. No other disclosures are reported.
doi:10.1001/jamainternmed.2020.6820 Centre d’Epidémiologie Clinique, AP-HP (Assistance Data Sharing Statement: See Supplement 3.
Correction: This article was corrected on January 4, Publique des Hôpitaux de Paris), Hôpital Hôtel
2021, to fix errors in the supplemental group Dieu, Paris, France (Ravaud). Funding/Support: This trial was publicly funded
member list and to add the collaborative group in (Ministry of Health, Programme Hospitalier de
Author Contributions: Drs Resche-Rigon and Recherche Clinique [PHRC COVID-19-20-0143,
the end matter. These collaborators were Porcher had full access to all the data in the study
previously not indexed in PubMed, but the error PHRC COVID-19-20-0029], Foundation for Medical
and take responsibility for the integrity of the data Research (FRM), AP-HP Foundation and the
has been resolved. It was corrected again on May 3, and the accuracy of the data analysis. The authors
2021, to correct the spelling of 2 names in the Reacting program).
contributed equally to this study.
nonauthor collaborators list in the Supplement. Concept and design: All authors. The CORIMUNO-19 Collaborative Group: The
Author Affiliations: Department of Hematology, Acquisition, analysis, or interpretation of data: CORIMUNO-19 Collaborative Group members are
Hôpital Necker, Assistance Publique - Hôpitaux de All authors. listed in Supplement 4.
Paris, France (Hermine); Institut Imagine, Université Drafting of the manuscript: All authors. Role of the Funder/Sponsor: The funding agencies
de Paris, INSERM UMR1183, Paris France (Hermine); Critical revision of the manuscript for important had no access to the trial data and had no role in the
Department of Rheumatology, Hôpital Bicêtre, intellectual content: Hermine, Mariette, Tharaux, design, conduct or reporting of the trial. Roche
Assistance Publique-Hôpitaux de Paris, France Resche-Rigon. donated TCZ in unrestricted grant, and had no role
(Mariette); Université Paris-Saclay, INSERM Statistical analysis: Porcher, Ravaud. in the trial design or conduct; the collection,
UMR1184, Le Kremlin Bicêtre, France (Mariette); Obtained funding: Hermine, Mariette, management, analysis, interpretation of the data;
Paris Cardiovascular Center–PARCC, Université de Resche-Rigon, Ravaud. or in the preparation, review of the manuscript or
Paris, INSERM, F-75015 Paris, France (Tharaux); Administrative, technical, or material support: the approval of the manuscript for submission.
Université de Paris, ECSTRRA Team-CRESS-UMR Hermine, Mariette, Resche-Rigon, Ravaud. Additional Contributions: We are grateful to all
1153, INSERM, 75010, Paris, France (Resche-Rigon); Supervision: Hermine, Mariette, Tharaux, Ravaud. patients who participated in the CORIMUNO study,

jamainternalmedicine.com (Reprinted) JAMA Internal Medicine January 2021 Volume 181, Number 1 39

© 2020 American Medical Association. All rights reserved.

Downloaded From: https://jamanetwork.com/ on 04/02/2023

 Research Original Investigation Effect of Tocilizumab vs Usual Care in Adults Hospitalized With COVID-19 and Moderate or Severe Pneumonia

and their families. The authors also thank doi:10.46234/ccdcw2020.032 doi:10.46234/ 14. Del Valle DM, Kim-Schulze S, Huang HH, et al.
Pr Maxime Dougados, MD, in charge of the logistics, ccdcw2020.032 An inflammatory cytokine signature predicts
as well as the investigators who collaborated in this 6. Yang X, Yu Y, Xu J, et al. Clinical course and COVID-19 severity and survival. Nat Med. 2020.
study (Supplement 4) and Universities of Paris, outcomes of critically ill patients with SARS-CoV-2 doi:10.1038/s41591-020-1051-9
Paris-Saclay, Paris-Sorbonne, Paris-Nord Sorbonne, pneumonia in Wuhan, China: a single-centered, 15. Grifoni E, Valoriani A, Cei F, et al. Interleukin-6
Paris-Est Créteil, Versailles-Saint Quentin, retrospective, observational study. Lancet Respir Med. as prognosticator in patients with COVID-19. J Infect.
Strasbourg and Lille (Medical Students support), 2020;8(5):475-481. doi:10.1016/S2213-2600(20) 2020;81(3):452-482. doi:10.1016/j.jinf.2020.06.008
INSERM, and Reacting. 30079-5 16. Xu X, Han M, Li T, et al. Effective treatment of
Additional Information: A coordination committee 7. Wang D, Hu B, Hu C, et al. Clinical characteristics severe COVID-19 patients with tocilizumab. Proc
was responsible for the design, conduct, and of 138 hospitalized patients with 2019 novel Natl Acad Sci U S A. 2020;117(20):10970-10975.
reporting of the trial, and an independent data and coronavirus-infected pneumonia in Wuhan, China. doi:10.1073/pnas.2005615117
safety monitoring board (DSMB) oversees all JAMA. 2020;323(11):1061-1069. doi:10.1001/jama.
CORIMUNO trials (Supplement 4). 17. Roumier M, Paule R, Groh M, Vallee A,
2020.1585 Ackermann F. Interleukin-6 blockade for severe
REFERENCES 8. Wu J, Liu J, Zhao X, et al. Clinical characteristics COVID-19. medRxiv. Published online April 22, 2020.
of imported cases of coronavirus disease 2019 doi:10.1101/2020.04.20.20061861.
1. Wu Z, McGoogan JM. Characteristics of and (COVID-19) in Jiangsu province: a multicenter
important lessons from the coronavirus disease 18. Guaraldi G, Meschiari M, Cozzi-Lepri A, et al.
descriptive study. Clin Infect Dis. 2020;71(15):706- Tocilizumab in patients with severe COVID-19:
2019 (COVID-19) outbreak in China: summary of a 712. doi:10.1093/cid/ciaa199
report of 72 314 cases from the Chinese Center for a retrospective cohort study. Lancet Rheumatol.
Disease Control and Prevention. JAMA. 2020;323 9. Beigel JH, Tomashek KM, Dodd LE, et al. 2020;2(8):e474-e484. doi:10.1016/S2665-9913(20)
(13):1239-1242. doi:10.1001/jama.2020.2648 Remdesivir for the treatment of Covid-19: 30173-9
preliminary report. N Engl J Med. 2020; 19. Working WHO; WHO Working Group on the
2. Huang C, Wang Y, Li X, et al. Clinical features of NEJMoa2007764. doi:10.1056/NEJMoa2007764
patients infected with 2019 novel coronavirus in Clinical Characterisation and Management of
Wuhan, China. Lancet. 2020;395(10223):497-506. 10. Group RC, Horby P, Lim WS, et al. COVID-19 infection. A minimal common outcome
doi:10.1016/S0140-6736(20)30183-5 Dexamethasone in hospitalized patients with measure set for COVID-19 clinical research. Lancet
Covid-19: preliminary report. N Engl J Med. 2020; Infect Dis. 2020;20(8):e192-e197. doi:10.1016/
3. Chen N, Zhou M, Dong X, et al. Epidemiological NEJMoa2021436. S1473-3099(20)30483-7
and clinical characteristics of 99 cases of 2019
novel coronavirus pneumonia in Wuhan, China: 11. Mehta P, McAuley DF, Brown M, Sanchez E, 20. Cook DJ, Marshall JC, Fowler RA. Critical illness
a descriptive study. Lancet. 2020;395(10223):507- Tattersall RS, Manson JJ; HLH Across Speciality in patients with COVID-19: mounting an effective
513. doi:10.1016/S0140-6736(20)30211-7 Collaboration, UK. COVID-19: consider cytokine clinical and research response. JAMA. 2020;323
storm syndromes and immunosuppression. Lancet. (16):1559-1560. doi:10.1001/jama.2020.5775
4. Guan WJ, Ni ZY, Hu Y, et al; China Medical 2020;395(10229):1033-1034. doi:10.1016/S0140-
Treatment Expert Group for Covid-19. Clinical 21. Angus DC. Optimizing the trade-off between
6736(20)30628-0 learning and doing in a pandemic. JAMA. 2020;323
characteristics of coronavirus disease 2019 in
China. N Engl J Med. 2020;382(18):1708-1720. 12. Ruan Q, Yang K, Wang W, Jiang L, Song J. (19):1895-1896. doi:10.1001/jama.2020.4984
doi:10.1056/NEJMoa2002032 Clinical predictors of mortality due to COVID-19 22. Dechartres A, Trinquart L, Boutron I, Ravaud P.
based on an analysis of data of 150 patients from Influence of trial sample size on treatment effect
5. Team TNCPERE: The Novel Coronavirus Wuhan, China. Intensive Care Med. 2020;46(5):
Pneumonia Emergency Response Epidemiology estimates: meta-epidemiological study. BMJ. 2013;
846-848. doi:10.1007/s00134-020-05991-x 346:f2304. doi:10.1136/bmj.f2304
Team. The epidemiological characteristics of an
outbreak of 2019 novel coronavirus diseases 13. Zhu J, Pang J, Ji P, et al. Elevated interleukin-6 is
(COVID-19). China CDC Weekly. 2020;2(8):113-122. associated with severity of COVID-19:
a meta-analysis. J med virol. Published online May
29, 2020. doi:10.1002/jmv.26085

40 JAMA Internal Medicine January 2021 Volume 181, Number 1 (Reprinted) jamainternalmedicine.com

© 2020 American Medical Association. All rights reserved.

Downloaded From: https://jamanetwork.com/ on 04/02/2023