key: cord-0721474-ffaysnmz authors: Avni, Tomer; Leibovici, Leonard; Cohen, Ido; Atamna, Alaa; Guz, Dmitri; Paul, Mical; Gafter-Gvili, Anat; Yahav, Dafna title: Tocilizumab in the treatment of COVID-19 - a meta-analysis date: 2021-05-19 journal: QJM DOI: 10.1093/qjmed/hcab142 sha: d789cec6d0b788f7a61f62aa65919a85896577bc doc_id: 721474 cord_uid: ffaysnmz OBJECTIVE: To evaluate whether IL-6 inhibitor tocilizumab (TCZ) reduces mortality among hospitalized COVID-19 patients. METHODS: Systematic review and meta-analysis of randomized controlled trials (RCTs) comparing TCZ versus placebo/control, for treatment of adults with COVID-19. Primary outcome was 28-30 days all-cause mortality. Search was conducted up to April 1(st) 2021. Two independent reviewers screened citations, extracted data, and assessed risk of bias. Relative risk (RR) with 95% confidence intervals (CI) were pooled. We performed subgroup analysis for patients with critical illness and sensitivity analyses. RESULTS: Eight RCTs were included, assessing 6,481 patients with mostly severe non-critical COVID-19 infection. TCZ was associated with a reduction in all-cause 28–30-day mortality compared to placebo/control (RR = 0.89, 95%CI 0.82-0.96). Among the subgroup of critically ill patients no reduced mortality was demonstrated (RR = 0.94, 95%CI 0.74-1.19). No mortality benefit with TCZ was demonstrated in trials that used steroids for >80% of patients. TCZ was associated with significantly reduced risk for mechanical ventilation (MV); for combined endpoint of death or MV; and for intensive care unit (ICU) admission. No significant difference in adverse events was demonstrated. Risk of serious superinfection was significantly lower with TCZ (RR = 0.57, 95%CI 0.35-0.93). CONCLUSION: The treatment with TCZ reduces 28-30 days all-cause mortality, ICU admission, superinfections, MV and the combined endpoint of death or MV. Among critically ill patients, and when steroids were used for most patients, no mortality benefit was demonstrated. Additional research should further define sub-groups that would benefit most and preferred timing of administration of TCZ in severe COVID-19. Coronavirus disease 2019 (COVID- 19) is an emerging respiratory disease caused by the novel corona virus and causes severe mortality and morbidity world-wide (1, 2) . Patients suffering from severe COVID-19 infection, tend to present with an exaggerated inflammatory response consisting of proinflammatory cytokines and chemokines, predominantly interleukin-6 (IL-6) (3) (4) (5) . The IL-6 receptor monoclonal antibody, Tocilizumab (TCZ), inhibits IL-6 signaling pathway and is currently under use for inflammatory disorders as rheumatoid arthritis and giant cell arteritis. (6) Several observational studies have examined the role of IL-6 blockade with TCZ in patients with moderate and severe COVID-19 infection (7, 8) . A systematic review of observational studies has shown reduced mortality with TCZ by a risk ratio (RR) of 0.27, 95% confidence interval (CI) 0.12 -0.59 (9) , and another review showed both reduced mortality and reduced need for mechanical ventilation with TCZ (10). Since these results are based on observational data alone, higher levels of evidence by examining results from RCTs is necessary. In this systematic review and meta-analysis, we aimed to examine the current evidence from RCTs on the efficacy and safety of TCZ in the treatment of COVID-19 infection. We aimed to define sub-groups who would benefit most from this therapy. Considering the priority of the results for decision-making, and taking into account expected new evidence from several similar ongoing trials, we planned to conduct this review as a living systematic review. (11) This review was conducted and reported in accordance with PRISMA guidelines. (12) Inclusion criteria Randomized controlled trials (RCTs), regardless of publication status, were included. We included trials assessing hospitalized adult patients with COVID-19 infection, diagnosed as defined in individual trials. The intervention assessed was any TCZ versus usual care with or without placebo. We excluded studies that directly compared IL-6 inhibitors with other drugs for COVID-19 infection (corticosteroids, immunomodulators, antivirals). We excluded studies that did not report on mortality. In studies that included more one intervention, we extracted results only for the TCZ and control arms. The primary outcome was all-cause 28-or 30-days mortality, or when lacking these data, allcause mortality as reported by authors at the closest time point. Secondary outcomes included long-term survival, need for MV, combined endpoint of death or MV, intensive care unit (ICU) admission, length of hospital / ICU stay, need for oxygen / oxygen free days, number discharged alive, and adverse events (AEs), including any AEs, serious AEs (SAEs) and superinfections. Outcome measures were collected on an intention-to-treat (ITT) basis. In cases in which such data were not presented, modified ITT (mITT) or per-protocol results were used. For the outcome of MV, if this was not reported separately, we deduced the result from the combined endpoint of death or MV. We searched PubMed, medRxiv.org, LILACS, Cochrane library, clinicaltrials.gov and ECCVID 2020 conference proceedings (ESCMID conference on corona virus disease), up to April 1 st 2021. We also hand-searched all references of included trials, websites of the drugs manufacturers, and previous meta-analyses for additional trials. For the PubMed search the term "IL-6 inhibitors" and specific inhibitors / antibodies names and their MESH terms were crossed with the terms "COVID-19", "corona virus disease 2019", "SARS COV 2" and with the Cochrane highly sensitive filter for RCTs (13). For other data sources, we used the search string "tocilizumab" and "COVID-19". No language restrictions were used. All authors were contacted to complement data by e-mail. Two reviewers independently inspected each reference identified by the search, scanned fulltexts of relevant studies, applied inclusion criteria and extracted the data. Disagreements in data extraction were resolved by discussion with a third reviewer. Risk of bias was assessed in duplicate using domain-based evaluation, classifying studies primarily according to the risk of non-random allocation, i.e. allocation concealment and sequence generation. These were graded as low, unclear and high risk for bias, as recommended in The Cochrane Handbook (13). Additional domains assessed included blinding, incomplete outcome data and selective outcome reporting. We graded the strength of the evidence and built a summary of evidence (SOF) table using the GRADEpro tool (14) . We also inspected adherence to WHO classification of COVID-19 infection severity (15) and reporting of AEs (16) . For dichotomous data, individual study results are expressed as RR with 95% confidence intervals (CI). RRs and mean differences were pooled using a fixed effect model (Mantel-Haenszel method) (17). Heterogeneity was defined by a chi-square test of heterogeneity <0.1 or an I 2 measure of inconsistency >40%. If significant heterogeneity was identified, we used random effect model. We performed sensitivity analyses by the adequacy of allocation concealment and generation, blinding, publication status, and concomitant steroid use. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 values and age. Due to the small number of included trials, we did not use a funnel plot to assess small-studies effects. As a living systematic review, we plan to repeat the search monthly and update the metaanalysis with each new trial. The updates will discontinue when authors of this review will decide that a reasonable level of certainty has been reached for the study's question. (11) The search yielded 747 results of which 36 were potentially relevant. Twenty-eight records were excluded, thus, eight trials (seven published (18) (19) (20) (21) (22) (23) (24) , one pre-print (25) were included in the meta-analysis (Supplemental Figure 1 , PRISMA flow chart). Details regarding the characteristics of trials are presented in Table 1 . The trials were conducted between March 2020 and January 2021 and recruited 6,481 patients, among them, 63.5% were recruited in the RECOVERY trial (25) . The criteria of COVID-19 severity for inclusion in the original trials was: moderate to severe pneumonia in six trials; severe to critical patients in one (22) and critically ill patients (receiving cardiorespiratory support) in one trial (24). Of the seven trials including moderatesevere and severe to critical patients , three trial included patients under MV at randomization as follows: 14% (25), 16.2% (22) and 37% (23) ; two other included 15% (19) , and 4% (21) patients hospitalized in ICU at randomization. (For severity definitions in individual trials see Table 1 and Supplemental Table 1 ). Microbiological confirmation of COVID-19 infection was required for inclusion in included trials (by molecular methods in five trials or by IgM antibodies in one trial (21) ). However in three trials, 6% (25), 10% (18) and 15.4% (24) of the patients, had no microbiological confirmation. The intervention assessed was one dose of intravenous TCZ 8 mg/kg, reported to be administered within less than 24 hours from randomization (or within 24 hours after the first randomization for the RECVOERY trial (25) ) or need for respiratory support in six trials (See Table 1 ). A second dose was allowed after 8-24 hours in six trials, based on clinical decision, and was mandatory in one trial (20) . Placebo was used in 3/8 trials (19, 21, 23) . Usual care included antivirals, steroid s, anticoagulants, and others (See Supplemental Table 3 for detailed concomitant treatments). The median time from symptoms onset to randomization was reported in all but one trial (24), which reported time from hospital admission, and was between 8 to 10 days. The primary outcome of the trials consisted of death at day 28 (25) ; death or need for MV at day 14 (18, 20, 22) , day 28 (19, 21, 23) and in-hospital mortality (24); and / or clinical improvement as defined by authors, consisting of improvement on oxygenation, freedom of respiratory support, ICU and hospital discharge (or readiness to discharge), days-free organ support and / or improvement on COVID-19 severity scale. The follow up duration was 28-30 days postrandomization. Hermine et al. (18) Risk of bias assessment is detailed in Table 2 . Low risk sequence generation and allocation concealment were reported in all trials. Three of the eight trials were double blinded (19, 21, 23) while the other were unblinded. Results were analyzed by intention-to-treat when available, or for the largest population available otherwise. In one trial (18) , concerns were raised regarding selective outcome reporting, as the primary outcome was changed during the recruitment. Informed consent and ethical committee approval were described in all trials. All trials received industrial sponsorship. Two trials were terminated early due to futility / safety ((20,22) . GRADEpro tool for examining the SOF is detailed in supplemental tables. AEs were reported in concert with the common terminology criteria for reporting of adverse events (CTCTE) or MedDRA hierarchy guidelines in six trials, and a central safety 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 committee (24,25). Definitions for severe AEs were not stated / graded in four trials (18, 20, 24, 25) . (See Table 2) Primary outcome -all-cause mortality at 28 days Mortality at 28-30 days was reported by seven trials, and in-hospital mortality was reported in one (24) and varied significantly from 2% to 32.1%. On meta-analysis, all-cause mortality at 28 days, was significantly reduced by TCZ compared to usual care, RR 0.89, 95% CI 0.82-0.96, I 2 =4%, 8 trials ( Figure 1a ). After exclusion of the RECOVERY trial (25) which consisted 77% of the total weight, from the analysis, the effect estimate remained similar but without statistical significance, RR 0.89, 95% CI 0.75-1.06. Sensitivity analysis restricted to trials that used steroids in over than 80% of patients in both arms (19, 22, 24, 25) showed no mortality benefit with TCZ RR 0.92, 95% CI 0.74-1.14, with substantial heterogeneity (Supplemental Figure 2 ). Sensitivity analysis by blinding showed no difference in mortality compiling data from three double blinded trials (RR 0.97, 95% CI 0.69-1.37, I 2 =0%), as opposed to reduced mortality in open-label trials (RR 0.88, 95% CI 0.81-0.96, I 2 =35%) (Supplemental Figure 3 ). Mortality at 28 days / in-hospital for the subgroup of critically ill patients was reported in three trials (23) (24) (25) and was similar between TCZ and usual care, RR 0.94, 95% CI 0.74- Mortality at 14-15 days was reported in three trials (18, 20, 22) , and was significantly increased with TCZ, RR 2.18, 95% CI 1.01-4.69, I 2 =31%). Salama et al. (19) reported 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 mortality at 60 days which was similar between TCZ and placebo (29/250, 11.6% vs.5/127, 11.8%). The REMA-CAP trail (24) reported 90-day survival as a time to event variable, with an adjusted hazard ratio of 1.60, SD 0.21 of TCZ compared to control. Need for MV was reported at 14 days (18) and at 28 days (21, 23, 25) or both (22) . TCZ was associated with a significantly reduced risk for MV, RR 0.79, 95% CI 0.68-0.91, I 2 =0% (25) reported new onset of non-invasive and invasive ventilation, and cessation of MV, which were not statistically significant reduced with TCZ. Other efficacy outcomes were without significant difference and are summarized in Table 3 . These included need for oxygen supply measures, change in severity scale, discharge rates and duration of stay. No results on AEs were reported for the REMAP-CAP (24) and the RECOVERY trials (25) . There was no significant difference in AEs and SAEs between the arms (RR 0.97, 95% CI 0.88-1.07, I 2 =28%, and RR 0.87, 95% CI 0.72-1.06, I 2 =0%, respectively). Superinfections were reported in 6 trials, only Hermine et al. (18) reported separately bacterial, fungal and viral infections. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Overall rates of superinfections and severe superinfections were higher with placebo / usual care, RR 0.64, 95% CI 0.64-0.97, I 2 =44%, and RR 0.57, 95% CI 0.35-0.93, I 2 =42%, respectively (Supplemental Figure 4) . Neutropenia, reported in four trials, was significantly increased in the TCZ arm (RR 8.70, 95% CI 2.34-32.39). We could not perform most of the pre-planned subgroup analyses due to significant variation of the outcomes reported and lack of published data. Stone et al. (21) and Veiga et al. (22) reported no significant difference between TCZ and placebo for the outcome of MV or death among elderly patients (age>=65 and >=60 years, respectively). Rosas et al. (23) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 CI 0.74-1.14, with substantial heterogeneity (derived by the trial of Veiga et al. (22) ) Need for MV was significantly reduced by TCZ, as well as a composite endpoint of death or MV, and ICU admission rates. Overall AEs and SAEs did not differ between groups, although any and serious infections (were significantly less common in the TCZ arm. This was regardless of higher neutropenia rates in the TCZ arm, ranging between 2.9 to 14%. The reduced mortality signal was mainly driven by the RECOVERY trial. The limitations of this trial include its being published only as a preprint at this time, open-label design, use of a second randomization which may be prone to treatment response bias, and use of the drug in specific sites according to site drug availability (25) . Among critically ill patients, no difference in mortality was demonstrated in our meta-analysis. The REMAP-CAP trial which included critically ill patients, reported reduced mortality with TCZ (24). Combining this trial with two other trials reporting mortality for a sub-group of similar patients (23, 25) demonstrated no difference in mortality. This outcome is limited by the difference in timeframe for assessing (7, 8, 26, 27) Thirty-day mortality rates in these observational studies mostly exceeded 20%, as opposed to most included RCTs in this metaanalysis, demonstrating mortality rates ranging between 1.5 and 15%, besides Rosas et al., that reported mortality rates of ~20%, The REMAP-CAP trial. that reported ~36% mortality in the control arm, and the RECVOERY trial that reported overall mortality of 31.3% (23) (24) (25) . This probably reflects inclusion of non-critical patients in most of the RCTs, as well as limited external validity of these studies, excluding immunocompromised patients, a group known to suffer from higher mortality due to COVID-19. (28) Two of the observational studies reported higher rates of superinfections with TCZ vs control, (8,27) however a third study showed no difference. (7) We demonstrated reduced MV and ICU admission in our meta-analysis. It should be considered that these outcomes are not "hard" outcomes as all-cause mortality, and might be influenced by local practices. Wide variability has been described in rates of MV across different countries, ranging from ~2 to 33% of hospitalized COVID-19 patients in general, and ~30 to 90% of those admitted to ICU. Similarly, substantial differences in reasons for admission and severity grade are described between ICUs across the globe. (29) The association between need for MV and death is also variably reported and is probably limited 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 by short follow up periods, though rates of death are consistently high among ventilated patients, suggesting the importance of this outcome despite its limitations. (30) Various groups have tried to set core outcomes for COVID-19 treatment trials. All included respiratory status as an outcome. Definitions and timeframes vary, including proportion without respiratory failure at ≥ 28 days, (31) ventilator free days, (32) time to first receiving ventilation during hospital stay, (33) and need for MV within 28 days and up to 6 months. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Both IDSA and CDC COVID-19 treatment guidelines (39, 40) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 In summary, compiling data from eight RCTs, we found a significant benefit in 28-day allcause mortality in patients with moderate to severe COVID-19 receiving TCZ versus placebo/control. Need for MV, ICU admission and the outcome of death or MV also significantly favored TCZ. However, for critically ill patients, and in trials using steroids for most patients, no significant difference in mortality was demonstrated. Additional trials should further delineate which patients are most likely to benefit from TCZ based on their severity scale, respiratory support, receipt of other therapy and time form symptoms onset. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Characteristics of and Important Lessons From the Coronavirus Disease 2019 (COVID-19) Outbreak in China: Summary of a Report of 72 314 Cases From the Chinese Center for Disease Control and Prevention Clinical Characteristics of Coronavirus Disease 2019 in China COVID-19: consider cytokine storm syndromes and immunosuppression Elevated interleukin-6 is associated with severity of COVID-19: A meta-analysis Interleukin-6 as prognosticator in patients with COVID-19 A Review of Recent Advances Using Tocilizumab in the Treatment of Rheumatic Diseases Association Between Early Treatment With Tocilizumab and Mortality Among Critically Ill Patients With COVID-19 Tocilizumab in patients with severe COVID-19: a retrospective cohort study Decreased mortality in COVID-19 patients treated with Tocilizumab: a rapid systematic review and meta-analysis of observational studies Efficacy of tocilizumab in COVID-19: A systematic review and meta-analysis Living systematic reviews Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement Guideline Development Tool WHO Working Group on the Clinical Characterisation and Management of COVID-19 infection. A minimal common outcome measure set for COVID-19 clinical research Better reporting of harms in randomized trials: an extension of the CONSORT statement The Cochrane Collaboration Effect of Tocilizumab vs Usual Care in Adults Hospitalized With COVID-19 and Moderate or Severe Pneumonia: A Randomized Clinical Trial Tocilizumab in Patients Hospitalized with Covid-19 Pneumonia Effect of Tocilizumab vs Standard Care on Clinical Worsening in Patients Hospitalized With COVID-19 Pneumonia: A Randomized Clinical Trial Efficacy of Tocilizumab in Patients Hospitalized with Covid-19 Effect of tocilizumab on clinical outcomes at 15 days in patients with severe or critical coronavirus disease 2019: randomised controlled trial Tocilizumab in Hospitalized Patients with Severe Covid-19 Pneumonia Interleukin-6 Receptor Antagonists in Critically Ill Patients with Covid-19 Tocilizumab in patients admitted to hospital with COVID-19 (RECOVERY): preliminary results of a randomised, controlled, open-label, platform trial | medRxiv Effects of tocilizumab on mortality in hospitalized patients with COVID-19: a multicentre cohort study Tocilizumab among patients with COVID-19 in the intensive care unit: a multicentre observational study COVID-19 in Immunocompromised Hosts: What We Know So Far Clinical review: International comparisons in critical carelessons learned Mechanical Ventilation in COVID-19: Interpreting the Current Epidemiology COVID-19: Developing Drugs and Biological Products for Treatment or Prevention Guidance for Industry ICMRA COVID-19 Treatments and Clinical Trials Workshop #2 -Summary | International Coalition of Medicines Regulatory Authorities (ICMRA) Repurposed antiviral drugs for COVID-19 -interim WHO SOLIDARITY trial results | medRxiv Core Outcomes Set for Trials in People With Coronavirus Disease Integrated safety in tocilizumab clinical trials Epidemiology of invasive pulmonary aspergillosis among COVID-19 intubated patients: a prospective study Reporting infections in clinical trials of patients with haematological malignancies COVID-19 Guideline, Part 1: Treatment and Management We thank authors of individual trials for their attention to our requests Funding: None