key: cord-0762534-0a8a78sv
authors: Hermine, Olivier; Mariette, Xavier; Porcher, Raphael; Resche-Rigon, Matthieu; Tharaux, Pierre-Louis; Ravaud, Philippe
title: Effect of Interleukin-6 Receptor Antagonists in Critically Ill Adult Patients with COVID-19 Pneumonia: two Randomised Controlled Trials of the CORIMUNO-19 Collaborative Group
date: 2022-02-03
journal: Eur Respir J
DOI: 10.1183/13993003.02523-2021
sha: 0b3d0278b8f201889ce0396c96afcd91a8070064
doc_id: 762534
cord_uid: 0a8a78sv

QUESTION: To determine whether anti-IL-6 Receptors improve outcomes of critically ill patients with COVID-19 pneumonia. PATIENTS AND METHODS: Two cohort-embedded, investigator-initiated, multicenter, open-label, Bayesian randomised controlled clinical trials. Patients were randomly assigned to receive either usual care (UC) or UC+Tocilizumab (TCZ) 8 mg/kg (TOCI-2 trial), or UC or UC+ Sarilumab (SARI) 200 mg (SARI-2 trial), both intravenously on day 1 (D1) and on D3 if clinically indicated. RESULTS: Between 31 March and 20 April 2020, 97 patients were randomised in the TOCI-2 trial, to receive UC (n=46) or UC+TCZ (n=51). At day 14, numbers of patients who did not need NIV, MV and alive with TCZ or UC were similar (47% versus 42%, median posterior hazard ratio [HR] 1.19, 90% CrI, 0.71 to 2.04), with a posterior probability of HR>1 of 71.4%. Between 27 March and 4 April 2020, 91 patients were randomised in the SARI-2 trial, to receive UC (n=41) or UC+SARI (n=50). At day 14, numbers of patients who did not need NIV, MV and alive with SARI or UC were similar (38% versus 33%, median posterior hazard ratio [HR] 1.05, 90% CrI, 0.55 to 2.07), with a posterior probability of HR>1 of 54.9%. Overall, the risk of death up to day 90 was: UC+TCZ versus UC (24% versus 30%, HR 0.67 [0.30 to 1.49]); UC+SARI versus UC (29% versus 39%, (HR 0.74 ([0.35 to 1.58]). Both TCZ and SARI increased serious infectious events. CONCLUSION: In critically ill patients with COVID-19, anti-IL-6 Receptors did not significantly increase the number of patients alive without any NIV, MV by D14. TRIAL REGISTRATION: ClinicalTrials.gov numbers: (NCT04331808 and NCT04324073)

COVID-19 is a respiratory disease induced by a novel coronavirus (SARS-CoV-2), having already caused more than 2.5 million deaths over the world [1] [2] [3] [4] . Most people with COVID- 19 have only mild or uncomplicated symptoms. Still, approximately 10% to 15% have moderate or severe diseases requiring hospitalization and oxygen support, and 3% to 5% require admission to an intensive care unit (ICU) mainly for ventilation assistance [4, 5] .

Patients with COVID-19 pneumonia present non-specific inflammatory responses, including edema and inflammatory cell infiltration in the lungs. Besides the specific pathogenic effect of SARS-CoV-2, host immune response, in addition to its role in controlling virus replication, may result in hyper inflammation leading to worsening pulmonary function. It is, at least in part, related to the production of several pro-inflammatory cytokines and chemokines, including interleukin 6 (IL-6). As demonstrated in the RECOVERY collaborative group, dexamethasone (DXM) 6 mg/day for 10 days decreased the 28-day mortality among patients receiving oxygen, including mechanical ventilation (MV), high flow, and non-invasive ventilation (NIV) [6] . Thus, the benefit from glucocorticoids in moderate-to-severe and critically ill patients suggests that an excessive host inflammatory response is responsible for much of the serious illness and death from COVID-19.

At the beginning of the epidemic in France, when no standard of care was defined, including the use of corticosteroids and anticoagulation, we decided to set up the publicly supported Given the potential deleterious effect of IL-6 in COVID-19 hyperinflammation [7] [8] [9] [10] [11] , we evaluated the benefit-risk effect of tocilizumab (TCZ) and sarilumab (SARI), two anti-human IL-6 receptor (IL-6R) monoclonal antibodies that inhibit IL-6 signaling. In non-ICU patients, we found an effect of TCZ [12] [13] but not of SARI [14] for preventing evolution to ventilation or death at D14, but no effect on survival at D28 and a trend for a better survival at D90. However, these studies were not designed to evaluate overall survival. In the present two studies, we investigated the effectiveness of TCZ and SARI versus usual care on freeventilation survival in critically ill patients with COVID-19 on non-invasive or mechanical ventilation

At the beginning of the SARS-CoV-2 pandemic, we set up a cohort of COVID-19 patients with moderate, severe, or critical pneumonia (CORIMUNO-19 Cohort, NCT04324047). This cohort was used to perform a series of randomized controlled trials testing different has been previously reported [12] and CORIMUNO-SARI-2 (Sarilumab (SARI) treatment) trials (NCT04324073, a common identifier with the CORIMUNO-SARI 1 trial conducted in patients with moderate or severe pneumonia) and also recently reported [14] . Accrual took place in 12 (TOCI-2) and 8 (SARI-2) different French University hospitals. Each center could include patients only in one protocol. Because of the emergency nature of the trial and feasibility issues, no placebo of TCZ and SARI was prepared.

The CORIMUNO Cohort and all embedded trials (i.e., trials using data collected in the CORIMUNO cohort) were approved by an ethics committee (CPP Île-de-France VI) and relevant authorities. Legal issues and trial procedures are presented in detail in SAP. Written informed consent was obtained from all patients or from the patient's legal representative for entering the CORIMUNO Cohort, and longitudinal data (including clinical status, biological data, and outcomes) were recorded as part of their participation in the cohort. In this consent, patients and/or their families 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 practice, for logistical reasons, only one trial took place at each site at a given time. A specific additional written consent was obtained from eligible patients or their families who were randomly selected to be offered TCZ or SARI. Patients randomized to be offered TCZSARI but who declined to be treated or who could not be treated were analyzed in an intention to treat basis in the arm of randomization (TCZ/SARI). Eligible patients assigned to receive usual care (UC) were not notified about the trial, but their CORIMUNO-cohort data were available for analysis. It is a classical process for Cohort-nested multiple Randomized Controlled Trials [15] . All patients were included in ICU. Thus, numbers of them were not able to give consent. In this situation, according to French law, emergency consent was signed by the physician after approval from the family. When the patients recovered, a pursuit consent had to be signed. If the patient refused to sign it, it was considered a consent withdrawal, and the patient's data could not be analysed. This trial was reported according to CONSORT guidelines. Patients from the CORIMUNO-19 cohort were eligible for CORIMUNO-TOCI-2 or SARI-2 trial if they had a WHO-CPS score > 5, including patients with non-invasive ventilation (NIV) or mechanical ventilation (MV). Exclusion criteria are detailed in the SAP.

Participants were randomly assigned in a 1:1 ratio to receive UC+TCZ or UC for TOCI-2 protocol, UC+SARI, or UC for SARI-2 protocol via a web-based secure centralized system.

An independent statistician provided a computer-generated assignment randomization list stratified by center and blocked with varying block sizes unknown to the investigators.

Centers were eligible to participate either in TOCI-2 or SARI-2 trials but not both. Both trials were performed during the same period.

TCZ was administered intravenously (IV) at 8 mg/kg on day 1 (D1) and SARI was administrated IV at a fixed dose of 400 mg on day 1 (D1). Administration of an additional fixed dose of TCZ 400 mg IV or SARI 400mg on D3 was recommended and left to the treating physician. UC (antibiotic agents, antiviral agents, corticosteroids, vasopressor support, anticoagulants) was provided at the discretion of the clinicians since, at that time, no standard of care (SOC) was defined, including the use of corticosteroids.

The early co-primary outcome is the proportion of patients with a decrease of WHO score of at least 1 point at day 4. Results are presented as the proportion of patients who improved so that effective treatment would be associated with an increase in proportion. The longer-term co-primary outcome is the cumulative incidence of successful tracheal extubation (defined as duration extubation > 48h) at day 14 if patients have been intubated before day 14 or removal of NIV or high flow (for > 48h) if they were included under oxygen by NIV or High flow (score 6) and remained without intubation. Death or new DNR order (if applied after the inclusion of the patient) are considered as a competing event. Both outcomes were consistent with the Core Outcome Set proposed by the WHO [16] (see Statistical Analysis Plan (SAP)).

Secondary outcomes were clinical status assessed with the WHO-CPS at D4, D7, and D14, overall survival at D90, time to discharge, time to oxygen supply independence, and change in biological factors such as C-reactive protein level, lymphocytes, and neutrophil counts.

Data quality monitoring included both remote data monitoring and on-site monitoring performed by dedicated staff independent of the site investigators, with 100% source data verification performed for all patients recruited at every site for all critical data points.

To maximize information from limited data generated while allowing for a rapid decision, we used Bayesian monitoring and analysis of the trial based on the co-primary outcomes. The sample size was set at 120, with interim analyses presented weekly to the data safety monitoring board (DSMB) and a provision to increase the sample size in case of promising but not conclusive results. We computed that the trial would have frequentist power 97.2% to detect an increase in primary outcome rate from 0.50 to 0.80 and 73.9% to detect a decrease in event rate from 0.50 to 0.70. For the D4 outcome, we used a beta prior distribution with parameters 1 and 1 for the proportion in each arm. For the D14 outcome, we used a Gaussian prior distribution with mean 0 and variance 10 6 for the log hazard ratio. Sensitivity analyses using a range of prior distribution were then conducted (see SAP). The treatment effect was expressed in terms of absolute risk difference (ARD) for the D4 outcome and sub-distribution Those analyses were carried out in a frequentist framework.

Analyses were performed on an intention-to-treat basis with no correction for multiplicity for secondary outcomes. Thus, results on secondary outcomes should be regarded as exploratory and are reported as point estimates and 95% confidence intervals (CIs). Statistical analyses involved using SAS v9.4 (SAS Institute) and R v3.6.1.

From March 31 to April 18, and from March 27 to April 4, 2020, 97 and 91 patients were randomized in trial TOCI-2 (51 patients to UC+TCZ and 46 UC alone) and trial SARI-2 (50 patients to UC+SARI and 41 UC alone), respectively. The Data Monitoring Committee DMC did not advise further increasing the sample size and trials were stopped because the number of COVID-19 cases dropped dramatically after mild of April (end of the first epidemic wave in France). Among the 51 and 50 patients assigned to receive TCZ or SARI, two in each protocol withdrew consent and were not analyzed. Among the 46 and 41 patients assigned to receive UC, three in the TOCI protocol and eight in the SARI protocol withdrew consent and were not analyzed. Among the 49 or 45 with TCZ or SARI treatment, 38 (47%) and 26 (58%) received a second injection on D3 ( Figure 1 ). Demographic and baseline clinical and biological characteristics of patients are described in Table 1 . The median age was 64.6 or 61.2 years (interquartile range, 58.7 to 70.6 or 53.9 to 66.9 years), and 72% or 77% were men, in TOCI-2 and SARI-2 protocols, respectively. There were no major between-group differences at enrollment in TOCI-2 and SARI-2 protocols.

At randomization, very few patients received antiviral therapy or glucocorticoids, and notably, no patients received dexamethasone (eTable1). Neutrophil count (G/L) 6.8 (5.3-8.6, n=48) 9.0 (4.8-12.5, n=37) 6.5 (4.9-8.6, n=47) 6.7 (4.6-10.3, n=31) Pre-specified (for antiviral drugs) or post-hoc subgroup analyses (for corticosteroids, including DXM) were not performed because the proportion of patients on antiviral drugs or corticosteroids was too low (<5%) in both protocols (eTable 5 (TOCI-2) and eTable 6 (SARI-2)). Additionally, we performed additional post-hoc subgroups analysis according to the All treatment effects are estimates adjusted on age and center, with 95% confidence intervals except otherwise stated.

* Non-invasive ventilation or high-flow oxygen ** Median posterior absolute risk difference with 90% Credible interval (90% CrI). † Median posterior hazard ratio adjusted for age and center. ‡ Median posterior odds ratio in a proportional odds model, adjusted for baseline WHO-CPS score, age and center. † † Only for patients in the ICU at randomization (TOCI-2 trial: n=40 in the tocilizumab arm and n=37 in the usual care arm, SARI-2 trial: n=38 in the sarilumab arm and n=28 in the usual care arm). RD: risk difference; HR: hazard ratio; OR: odds ratio MD: mean difference; 95% CI: 95% confidence interval; IQR: interquartile range.

The evolutions of WHO scores during 14-day follow-up are given eFigure 3 and eTable 7.

Although some trends were observed, no significant difference was observed in both TOCI-2 (Figure 2 C and D) , eTable 12). Causes of deaths were similar in all groups, mainly due to ARDS, and are summarized in Table 3 .

A post hoc pooled analysis of both anti-IL-6 receptor antibodies and UCs are shown for the D14 primary outcome (supplementary Figure 5A ) and D90 overall survival (supplementary Figure 5B ).

Mean C-reactive protein levels decrease rapidly in the TCZ and SARI arms, and lymphocyte count was increased (eFigure 4) mainly in the TCZ but not in SARI arms.

A total of 33 (67%) or 32 (68%) and 30 (70%) or 22 (68%) patients in the TCZ or SARI and UC groups reported adverse events between randomization and D90, respectively (Table 3) .

Serious adverse events occurred in 20 (32%) and 29 (43%), respectively (p=0.21) (eTable 13). The number of bacterial and fungal serious infections was higher in the TCZ and SARI groups than in UC groups (27 vs. 13 or 19 vs. 4). In patients in intensive care unit (ICU) requiring high flow, NIV, or mechanical ventilation (WHO-CPS >6), the only standard of care was DEX [6] , which in the RECOVERY trial has been shown to increase D28 overall survival, particularly in patients treated shortly after admission to ICU. The time of IL-6R antagonism introduction seems to be important, as also suggested by a large emulated multicenter trial emulated trial having involved 433 patients that found that in spite of an increased risk of secondary infections, the risk of in-hospital mortality in this study was lower in patients treated with TCZ in the first 2 days of ICU admission compared with patients whose treatment did not include early use of TCZ [17] .

TCZ and SARI, two anti-IL6 receptors registered for the treatment of rheumatoid arthritis and cytokines release syndromes (only TCZ), have also been tested extensively in the treatment of COVID-19 patients, and their effects are still a matter of discussion. In non-critically ill patients requiring oxygen (WHO-CPS=5), several studies suggest that TCZ reduces mortality at D28 associated with a reduction of severe adverse events, particularly infections [18] . For

SARI, fewer studies have tested its effect, and its efficacy is even uncertain [19] . [18] . In the TOCI-2 and SARI-2 protocols, dexamethasone was very barely used, and a majority of patients (84/146, 56%) were included and treated after day one of ICU admission, which might be too late, pulmonary lesions being already established.

Even if the primary end-point was not reached, it is interesting and somewhat surprising to observe a non-significant trend in day 90 better overall survival with both anti-IL6 receptor antibodies. It might be due to a slight decrease of ventilator-free days until D28 with TCZ and SARI vs. usual care (-2.5 (-6.9 to +1.7) and -1.5 (-6.1 to +3.9), respectively.

Interestingly, the difference of overall survival up to 90 days in our 2 trials between UC and UC + anti-IL-6 receptor antibodies is not far from that observed in the REMAP-CAP study. Regarding safety, the main difference between these 2 trials in ICU patients and the previous trials of anti-IL6-R antibodies in patients on oxygen in medical wards is a numerical increase of serious infections in SARI and TCZ arms (n=46) vs. usual care (n=17). This increased risk of serious infections (which was only statistically significant with TCZ) with anti-IL6R antibodies is easily explained by the higher fragility of these patients in ICU compared with patients in medical wards. However, in our study, bacterial sepsis was rarely the cause of death in both arms and thus, the overall benefit of anti-IL6R antibodies in critically ill patients [18] with COVID-19 may be a compromise between reducing deleterious pulmonary hyper inflammation and not increasing too much the risk of bacterial and fungal complications.

At the beginning of the first COVID-19 pandemic wave, the CORIMUNO-19 cohort was designed to perform several exploratory clinical trials swiftly and provide information on drug candidates of potential interest quickly. The trial was not blinded since it was logistically impossible to set up a double-blind study quickly enough at the time of the pandemics. However, it is unlikely that in this group of critically ill patients, unblinding could lead to measurement bias since the decision of extubation was based on objective parameters related to Pa02/ Fi02 ratio. Another limitation is that UC could differ among centers and over time. However, the short period of accrual and the stratification of randomization may have limited the impact of such a lack of standardization. Because of the design of the CORIMUNO platform, the sample size, which could not be increased because of the end of the first epidemic wave, was small and not designed to show a difference in survival, credibility intervals were wide, and the treatment effect may be underestimated [22] .

The design of our study was also not designed to compare the respective effects of TCZ and SARI that showed here similar effects.

In summary, this study does not bring evidence that TCZ or SARI alone are effective in shortening the time of ventilation support in the group of critically ill patients. However, the recent WHO meta-analysis results are consistent with our results on mortality at day 90, suggesting that anti-IL6 receptor antibodies plus dexamethasone could be an option in patients with critical COVID. 

Analyses according to antivirals at baseline were pre-specified in the protocol, but only 8 patients (5 tocilizumab, 3 usual care) were on antivirals at randomization. Additional analyses according to corticosteroids and dexamethasone were added post-hoc to the SAP hoc in the light of publications or press releases. No patient was on dexamethasone at randomization, and only 12 (8 tocilizumab, 4 usual care) were receiving corticosteroids. Accordingly, no subgroup analysis was performed. Post-hoc subgroup analyses according to the WHO-CPS score and the time from ICU admission to randomization (≤ 1 day vs. > 1 day) have been performed, as requested by the trial Scientific Committee. 11 /20 (55%) 0.36 (0.15 to 0.83) * Excluded 7 and 4 patients not in the ICU at randomization in the tocilizumab and usual care arm, respectively, and 1 patient with unknown date of ICU admission. ** 4 and 10 missing data in the tocilizumab and usual care arm, respectively.

Analyses according to antivirals at baseline were pre-specified in the protocol, but only 3 patients (2 sarilumab, 1 usual care) were on antivirals at randomization. Additional analyses according to corticosteroids and dexamethasone were added post-hoc to the SAP in the light of publications. No patient was on dexamethasone at randomization, and only 2 (0 sarilumab, 2 usual care) were receiving corticosteroids. Accordingly, no subgroup analysis stratified on these variables was performed. Post-hoc subgroup analyses according to the WHO-CPS score and the time from ICU admission to randomization (≤ 1 day vs. > 1 day) have been performed, as requested by the trial Scientific Committee. 

Posterior density of the risk difference and adjusted odds ratio for the day 4 outcome (golden line). The dark blue line represents the minimally informative priors. The solid gray lines indicates an RD of 0 or an OR of 1, representing no treatment effect, and the dashed gray lines indicate a moderate benefit (RD = 5.5%, OR=0.85).

Difference in Probability of WHO Score Not Improved

Posterior Distribution 

Clinical Characteristics of Imported Cases of Coronavirus Disease 2019 (COVID-19) in Jiangsu Province: A Multicenter Descriptive Study

Epidemiological and clinical characteristics of 99 cases of 2019 novel coronavirus pneumonia in Wuhan, China: a descriptive study

China Medical Treatment Expert Group for C. Clinical Characteristics of Coronavirus Disease 2019 in China

Clinical course and risk factors for mortality of adult inpatients with COVID-19 in Wuhan, China: a retrospective cohort study

Dexamethasone in Hospitalized Patients with Covid-19

An aberrant inflammatory response in severe COVID-19

Elevated interleukin-6 is associated with severity of COVID-19: a meta-analysis

An inflammatory cytokine signature predicts COVID-19 severity and survival

Interleukin-6 as prognosticator in patients with COVID-19

Immune intervention in COVID-19: a matter of time ?

Effect of Tocilizumab vs Usual Care in Adults Hospitalized With COVID-19 and Moderate or Severe Pneumonia: A Randomized Clinical Trial

CORIMMUNO-19 Collaborative group. Sarilumab in adults hospitalised with moderate-to-severe COVID-19 pneumonia (CORIMUNO-SARI-1): An open-label randomised controlled trial

Rethinking pragmatic randomised controlled trials: introducing the "cohort multiple randomised controlled trial" design

Characterisation WHOWGotC, Management of C-i. A minimal common outcome measure set for COVID-19 clinical research

Association Between Early Treatment With Tocilizumab and Mortality Among Critically Ill Patients With COVID-19

Association Between Administration of IL-6 Antagonists and Mortality Among Patients

Sarilumab COVID-19 Global Study Group. Sarilumab in patients admitted to hospital with severe or critical COVID-19: a randomised, double-blind

Tocilizumab in patients admitted to hospital with COVID-19 (RECOVERY): a randomised, controlled, open-label, platform trial

Influence of trial sample size on treatment effect estimates: meta-epidemiological study

Hôpital Necker, INSERM, Imagine Institute