Infect Dis Ther https://doi.org/10.1007/s40121-020-00373-8 ORIGINAL RESEARCH Combination of Tocilizumab and Steroids to Improve Mortality in Patients with Severe COVID-19 Infection: A Spanish, Multicenter, Cohort Study Belén Ruiz-Antorán . Aránzazu Sancho-López . Ferrán Torres . Vı́ctor Moreno-Torres . Itziar de Pablo-López . Paulina Garcı́a-López . Francisco Abad-Santos . Clara M. Rosso-Fernández . Ana Aldea-Perona . Eva Montané . Ruth M. Aparicio-Hernández . Roser Llop-Rius . Consuelo Pedrós . Paloma Gijón . Carolina Hernández-Carballo . Marı́a J. Pedrosa-Martı́nez . Consuelo Rodrı́guez-Jiménez . Guillermo Prada-Ramallal . Lourdes Cabrera-Garcı́a . Josefa A. Aguilar-Garcı́a . Rocı́o Sanjuan-Jimenez . Evelyn I. Ortiz-Barraza . Enrique Sánchez-Chica . Ana Fernández-Cruz on behalf of the TOCICOV-study group Received: October 7, 2020 / Accepted: November 16, 2020 Ó The Author(s) 2020 ABSTRACT Methods: We performed a multicentre retro- spective cohort study in 18 tertiary hospitals in Background: We aimed to determine the Spain from March to April 2020. Consecutive impact of tocilizumab use on severe COVID-19 patients admitted with severe COVID-19 treated (coronavirus disease 19) pneumonia mortality. with tocilizumab were compared to patients not treated with tocilizumab, adjusting by inverse probability of the treatment weights (IPTW). Belén Ruiz-Antorán and Aránzazu Sancho-López Tocilizumab’s effect in patients receiving ster- contributed equally to this work. oids during the 48 h following inclusion was Electronic supplementary material The online analysed. version of this article (https://doi.org/10.1007/s40121- Results: During the study period, 506 patients 020-00373-8) contains supplementary material, which is available to authorized users. with severe COVID-19 fulfilled the inclusion B. Ruiz-Antorán (&)  A. Sancho-López I. de Pablo-López Clinical Pharmacology Department, Hospital Clinical Pharmacology Unit, Hospital Universitario Universitario Puerta de Hierro-Majadahonda, Ramón y Cajal (IRYCIS), Madrid, Spain Instituto de Investigación Sanitaria Puerta de Hierro-Segovia de Arana, Madrid, Spain P. Garcı́a-López e-mail: bruizantoran@gmail.com Pneumology Department, Hospital Universitario Torrecardenas, Almerı́a, Spain F. Torres Medical Statistics Core Facility, August Pi i Sunyer F. Abad-Santos Biomedical Research Institute, Hospital Clinic, Clinical Pharmacology Department, Hospital Barcelona, Spain Universitario de la Princesa, Instituto Teófilo Hernando, Faculty of Medicine, Universidad F. Torres Autónoma de Madrid (UAM), Instituto de Biostatistics Unit, Faculty of Medicine, Autonomous Investigación Sanitaria la Princesa (IP), Madrid, University of Barcelona, Barcelona, Spain Spain V. Moreno-Torres  E. Sánchez-Chica C. M. Rosso-Fernández Internal Medicine Department, Hospital Unit of Clinical Pharmacology, Unit of Clinical Universitario Puerta de Hierro-Majadahonda, Investigation and Clinical Trials, Hospital Instituto de Investigación Sanitaria Puerta de Universitario Virgen del Rocı́o, Sevilla, Spain Hierro-Segovia de Arana, Madrid, Spain Infect Dis Ther criteria. Among them, 268 were treated with 0.449–0.804), p \ 0.001] (interaction tocilizumab and 238 patients were not. Median p = 0.094). time to tocilizumab treatment from onset of Conclusions: These results show that survival symptoms was 11 days [interquartile range of patients with severe COVID-19 is higher in (IQR) 8–14]. Global mortality was 23.7%. Mor- those treated with tocilizumab than in those tality was lower in patients treated with tocili- not treated and that tocilizumab’s effect adds to zumab than in controls: 16.8% versus 31.5%, that of steroids administered to non-intubated hazard ratio (HR) 0.514 [95% confidence inter- patients with COVID-19 during the first 48 h of val (95% CI) 0.355–0.744], p \ 0.001; weighted presenting with respiratory failure despite oxy- HR 0.741 (95% CI 0.619–0.887), p = 0.001. gen therapy. Randomised controlled studies are Tocilizumab treatment reduced mortality by needed to confirm these results. 14.7% relative to no tocilizumab treatment Trial registration: European Union electronic [relative risk reduction (RRR) 46.7%]. We cal- Register of Post-Authorization Studies (EU PAS culated a number necessary to treat of 7. Among Register) identifier, EUPAS34415 patients treated with steroids, mortality was lower in those treated with tocilizumab than in Keywords: COVID-19; Mortality; SARS-CoV-2; those treated with steroids alone [10.9% versus Steroids; Tocilizumab 40.2%, HR 0.511 (95% CI 0.352–0.741), p = 0.036; weighted HR 0.6 (95% CI A. Aldea-Perona C. Hernández-Carballo Clinical Pharmacology Department, Consorcio Parc Internal Medicine Department, Hospital Salut MarInstituto Hospital del Mar de Universitario Nuestra Señora Candelaria, Tenerife, Investigaciones Médicas (IMIM)Universidad Spain Autónoma de Barcelona (UAB), Barcelona, Spain M. J. Pedrosa-Martı́nez E. Montané Clinical Pharmacology Department, Hospital Clinical Pharmacology Department, Hospital Universitario Puerto Real, Cádiz, Spain Universitario Germans Trias i Pujol, Barcelona, Spain C. Rodrı́guez-Jiménez Clinical Pharmacology Department, Hospital E. Montané Universitario de Canarias, Tenerife, Spain Department of Pharmacology, Therapeutics and Toxicology, Barcelona, Spain C. Rodrı́guez-Jiménez Departamento de Medicina Fı́sica y Farmacologı́a, R. M. Aparicio-Hernández Facultad de Medicina, Universidad de La Laguna Clinical Pharmacology Department, Hospital (ULL), Tenerife, Spain Universitario Central de La Defensa Gómez Ulla, Madrid, Spain G. Prada-Ramallal Epidemiology, Statistics and Research Methodology R. Llop-Rius Unit, Institute for Health Research Foundation Clinical Pharmacology Department, Hospital (FIDIS), Santiago de Compostela, Spain Universitari de Bellvitge, Barcelona, Spain L. Cabrera-Garcı́a C. Pedrós Clinical Pharmacology Department, Hospital Clinical Pharmacology Unit, Consorci Hospital Universitario Clı́nico San Carlos, Madrid, Spain General Universitari de València, Valencia, Spain J. A. Aguilar-Garcı́a P. Gijón Internal Medicine Department, Hospital Costa del Clinical Microbiology and Infectious Diseases Sol Marbella, Málaga, Spain Department, Hospital General Universitario Gregorio Marañón, Instituto de Investigación R. Sanjuan-Jimenez Sanitaria Gregorio Marañón, Madrid, Spain UICEC IBIMA, Plataforma SCReN, Málaga, Spain Infect Dis Ther DIGITAL FEATURES Key Summary Points This article is published with digital features, Over-exuberant cytokine release occurred including a summary slide, to facilitate under- in some patients infected by SARS-CoV-2. standing of the article. To view digital features Treatment strategies aimed to break down for this article go to https://doi.org/10.6084/ this hyper-inflammatory response were m9.figshare.13228859. considered. This study describes the experience with a INTRODUCTION series of consecutive COVID-19 patients treated with TCZ in 18 tertiary hospitals in In the past few decades, newly evolved coron- Spain and compares the outcomes of this aviruses have posed a global threat to public cohort with those observed in a similar health. cohort of SARS-CoV-2 infected patients The most recent one, SARS-CoV-2 coron- who did not receive tocilizumab. avirus, has produced a pandemic infecting more Tocilizumab treatment reduced mortality than 10 million people and causing more than by 14.7% relative to no tocilizumab 500,000 deaths worldwide, challenging not treatment (RRR 46. 7%). Tocilizumab with only healthcare systems but also the culture and steroid-48 h treatment reduced mortality economy of the population [1]. by 29.1% relative to no tocilizumab SARS-CoV-2 causes COVID-19 (coronavirus treatment (RRR 72.8%). disease 19), which in most cases is an asymp- tomatic or mildly symptomatic condition that Mortality is improved when tocilizumab is resolves without therapy or with minor sup- used concomitantly or shortly followed by portive treatment. In some cases, SARS-CoV-2 steroids (first 48 h). These results can cause pneumonia, which in 20% may be contribute to the body of evidence moderate in severity. However, a subgroup of supporting the use of tocilizumab in SARS- patients with COVID-19 pneumonia develops CoV-2 infection. rapidly progressing respiratory failure that may Randomised controlled studies are needed necessitate mechanical ventilation and support to confirm these results and establish the in an intensive care unit (ICU) and/or result in potential place of tocilizumab in the multi-organ and systemic manifestations in treatment of COVID-19. terms of sepsis, septic shock, and multiple organ dysfunction syndromes [2]. Available evidence suggests that a hyper-in- flammatory syndrome (HIS) that resembles secondary hemophagocytic lymphohistiocyto- sis (sHLH) may have a pathogenic role [3, 4]. R. Sanjuan-Jimenez This is consistent with the general knowledge Servicio de Farmacologı́a Clı́nica, Hospital about coronavirus (CoV) infections, where the Universitario Virgen de la Victoria, Málaga, Spain immune response has been shown to play an E. I. Ortiz-Barraza essential role in controlling and eliminating Internal Medicine Department, Hospital CoV infections, and maladjusted immune Universitario Ramón y Cajal (IRYCIS), Madrid, Spain responses may result in immunopathology and A. Fernández-Cruz (&) impaired pulmonary gas exchange [5]. Infectious Diseases Unit, Internal Medicine At the time of conducting the study, treat- Department, Hospital Universitario Puerta de ment was mainly symptomatic, with oxygen Hierro-Majadahonda, Instituto de Investigación therapy representing the major treatment Sanitaria Puerta de Hierro-Segovia de Arana, Madrid, Spain intervention for patients with severe infection. e-mail: anafcruz999@gmail.com Several antiviral treatments have been used Infect Dis Ther such as lopinavir/ritonavir, chloroquine, METHODS hydroxychloroquine, remdesivir, and alpha-in- terferon. No vaccine is currently available. Design, Study Period, and Subjects Based on the observation that acute, life- threatening respiratory injury associated with This was a retrospective, observational cohort an over-exuberant cytokine release occurred in study performed in 18 tertiary Spanish hospitals some patients infected by SARS-CoV-2, treat- including patients with severe COVID-19. ment strategies aiming to break down this The study population were adult patients hyper-inflammatory response were considered. (C 18 years) with COVID-19, confirmed by PCR Elevated levels of blood interleukin-6 (IL-6) on nasopharyngeal swab, who were consecu- have been identified as a one of the risk factors tively admitted to the participating hospitals associated with severe COVID-19 [5]. between 3 March and 20 April 2020. In this context, tocilizumab, a humanized Patients were included consecutively IgG1 monoclonal antibody targeting the IL-6 according to the date of admission to hospital receptor, was postulated as a suitable treatment until the planned sample size was met in a option [6]. Preliminary experiences with tocili- competitive manner. Eighteen centres con- zumab showed promising results with tributed data to both cohorts (Online Appendix improvements seen in clinical symptoms, oxy- Table 1). Eligible patients were hospitalised genation status, and inflammatory laboratory patients outside the intensive care units (ICU) parameters [7]. This limited evidence together with documented pneumonia (by either imag- with a strong mechanistic rationale, plus cur- ing and/or the presence of rales/crackles on rent knowledge on the role of anti-IL6 inhibi- physical examination) with severe respiratory tors in the treatment of cytokine release failure. Severe respiratory failure was defined by syndromes, supported the progressively exten- presence of a Brescia-COVID Scale score of 2 (i.e. ded use of tocilizumab in clinical practice. patients on oxygen therapy plus one of the Simultaneously numerous clinical studies, following criteria: (1) the patient had dyspnoea including clinical trials, were initiated world- or staccato speech at rest or after minimum wide with the aim to assess the potential effi- activity; (2) respiratory rate [ 22 with [ 6 l/min cacy of tocilizumab in preventing the fatal O2; (3) significant worsening of the chest x-ray) consequences of acute respiratory and multi- or an score of 3 (i.e. the patient required high- organ failure associated with acute respiratory flow nasal ventilation, continuous positive air- syndrome due to severe COVID-19. way pressure (CPAP), or non-invasive ventila- At the time of launching our study (3 March tion (NIV) because of rapid deterioration of 2020), many uncertainties existed about the respiratory exchanges without immediate pos- actual benefits and risks of tocilizumab in the sibility of invasive ventilation (Fig. 1) [8]; and at treatment of SARS-CoV-2 pneumonia. To least one of these parameters: IL6 [ 40 pg/ml, increase knowledge on the actual role of tocili- or increasing lactate dehydrogenase (LDH) or zumab in SARS-CoV-2 pneumonia, we con- LDH [ twice the upper limit of normal, ducted a retrospective multicentre cohort study. increasing C reactive protein (CRP), D-dimer Our study aimed to describe our experience (DD) [ 1500 ng/ml, lymphocytes \ 1200/ll, or with a series of consecutive COVID-19 patients ferritin [ 500 ng/ml. We excluded patients \ treated with tocilizumab in 18 tertiary hospitals 18 years old and those who died within 24 h in Spain and compare the outcomes of this after admission to hospital or after developing cohort with those observed in a similar cohort inclusion criteria. Of these, patients who of SARS-CoV-2 infected patients who did not received tocilizumab therapy according to clin- receive tocilizumab to identify patients who ical practice were assigned to the tocilizumab will most benefit from its use. cohort, whilst patients who did not were assigned to the control cohort. Infect Dis Ther Fig. 1 Brescia-COVID respiratory severity scale Data Collection Definition of the Outcome Epidemiological, clinical, pharmacological, The primary outcome of the study was in-hos- laboratory, and radiological data were extracted pital mortality. The outcome of patients treated from medical records using a standardised data with tocilizumab was compared to that of those collection form. The patients were followed who did not receive tocilizumab. In patients according to clinical practice. Data were col- from the non-treated cohort, baseline day (day lected from days 1, 3, 7, 15, and 28 post-inclu- 0) was defined as the first day that the patient sion. Adverse events possibly or probably fulfilled the inclusion criteria established in the related to tocilizumab treatment were collected protocol. In patients from the tocilizumab for treated patients. All data were included by a cohort, baseline day was considered the day of primary reviewer and subsequently checked by initiation of treatment with tocilizumab. two senior physicians. Definition of the Exposure Laboratory Procedures Exposure to tocilizumab was defined as the use Routine blood examinations included a com- of intravenous tocilizumab at any time during plete blood count, coagulation profile, serum the hospital admission. The decision to pre- biochemical tests (including LDH), CRP, DD, IL- scribe tocilizumab was at the discretion of the 6, and serum ferritin. Chest radiographs or treating physician. computed tomography (CT) scans were also Details of tocilizumab use (including the done for all inpatients. timing of initiation, dosing, and number of doses) were recorded. Likewise, the choice of COVID-19 treatments other than tocilizumab was at the discretion of the treating physician, Infect Dis Ther although based on national and local recom- define insignificant differences in potential mendations for COVID-19 management. confounders, which was achieved by most, but For the main analysis, we generated a vari- not by all, baseline covariates. However, for able with the following mutually exclusive cat- some authors \ 0.20 might also be accept- egories: ‘‘non-use of tocilizumab drug’’ (control able [12, 13] and only IL-6 remained unbal- cohort) and ‘‘use of tocilizumab drug’’ (treat- anced (STD = 0.42) with no possible PS ment cohort). Subsequently, for the treatment correction given the high number of missing cohort, we disaggregated the latter into two data (& 51%). different subgroups: the ’with concomitant Since there were missing data for some key glucocorticoids or corticoids (steroids)’ and ’no covariates needed for the PS calculations, the concomitant steroids’ groups. Concomitant use following variables (all with \ 10% missing of steroids was defined as initial use within the data) were imputed using the EM algorithm, first 48 h after tocilizumab administration which relies on the flexible and reasonable (steroid-48 h). missing at random assumption: WHO ordinal scale, time from symptoms, lymphocytes, neu- Statistical Analysis trophils, platelets, haemoglobin, CRP, DD, temperature, LDH, and prothrombin activation. Categorical variables are described with fre- Baseline categorical data were compared quencies and percentages and continuous vari- using the chi-square test and continuous vari- ables with mean (standard deviation, SD) and ables using ANOVA with rank-transformed data median [interquartile range (IQR): 25th–75th for raw and IPTW adjusted analyses. Raw and percentiles], and the survival function is IPTW-adjusted regression models were used to described using the Kaplan-Meier function. estimates risks [hazard ratio (HR) with 95%CI We used standardised differences, defined as (95%CI)] for time-to-event variables. differences between groups divided by pooled In all statistical analyses, we applied a two- standard deviation to assess heterogeneity sided type I error of 5%. SPSS v.25 (IBM Corp., between both cohorts for baseline covariates. Armonk, NY) and SAS v9.4 (Cary, NC, USA) The inverse probability of the treatment weights software was used. (IPTW) approach [9] was used to create a pseudo-population in which the two groups Ethics were balanced across baseline covariates. The stabilised weights were calculated using The study was approved by the Spanish Regu- propensity scores (PS) [10] obtained from a latory Authority (Spanish Agency of Medicines logistic regression model aimed to minimise the and Medical Devices, AEMPS) and by the between arms standardised differences [11]. The Research Ethics Committee (REC) at Hospital baseline covariates included in the final model Universitario Puerta de Hierro-Majadahonda were gender, age, hypertension, neurological (FIB-TOC-2020-01), and a waiver for informed exploration, diabetes mellitus, WHO ordinal consent was granted. The study protocol was scale, time from symptoms, confirmed infec- submitted to the local RECs at each study site tion, lymphocytes, neutrophils, platelets, pro- and in line with Spanish legislation; some pro- thrombin activation, temperature, LDH, and vided their own additional approval while oth- baseline medication use of angiotensin con- ers did not need to undergo the full review and verting enzyme (ACEs) inhibitors, lopinavir-ri- approval process, i.e. they recognised the initial tonavir, hydroxychloroquine, corticosteroids, REC approval. The study complied with the interferon, non-steroidal anti-inflammatory provisions in European Union (EU) and Spanish drugs, moxifloxacin, remdesivir, and azi- legislation on data protection and the Declara- thromycin. Covariate balance was assessed tion of Helsinki 2013. using the standardised differences with the goal to achieve values \ 0.10 using the IPTW to Table 1 Baseline characteristic of patients with SARS-CoV-2 infection according to tocilizumab exposure Raw analysis IPTW analysis Tocilizumab Control group p value Standardised Tocilizumab Control group Standardised Infect Dis Ther group (N = 268) (N = 238) difference (%) group (N = 254) (N = 235) difference (%) Gender (men), n (%) 184 (68.7) 140 (58.8) 0.021 - 20.6 165 (65.0) 150 (63.8) 0.9 Age, mean (SD) 65.0 (11.7) 71.3 (14.2) \ 0.0001 - 47.9 66.6 (10.7) 67.3 (14.8) - 1.8 Underlying medical conditions, n (%) High blood pressure 130 (48.5) 145 (60.9) 0.001 - 25.1 132 (52.0) 126 (53.5) - 0.4 Cardiovascular disease 65 (24.3) 74 (31.1) 0.084 - 15.3 68 (26.9) 61 (26.1) 3.5 Diabetes 78 (29.1) 68 (28.6) 0.894 1.2 75 (29.6) 66 (27.8) 5.2 Chronic kidney disease 23 (8.6) 30 (12.6) 0.140 - 13.1 23 (9.2) 23 (9.9) - 0.9 Onco-hematologic 12 (4.5) 12 (5.0) 0.765 - 2.7 10 (3.8) 9 (4.0) - 0.1 Chronic lung disease 46 (17.2) 49 (20.6) 0.324 - 8.8 47 (18.4) 43 (18.4) - 0.3 Transplant (SOT/SCT) 6 (2.2) 4 (1.7) 0.652 4.0 6 (2.2) 7 (3.0) - 4.0 Neurologic 15 (5.6) 42 (17.6) \ 0.0001 - 38.3 21 (8.2) 25 (10.4) - 6.1 Liver disease 10 (3.7) 12 (5.0) 0.470 - 6.4 15 (5.9) 13 (5.5) 2.7 HIV 2 (0.7) 3 (1.3) 0.559 - 5.2 1 (0.5) 2 (0.7) - 1.9 NSAIDs 19 (7.1) 18 (7.6) 0.838 - 1.8 22 (8.6) 16 (6.7) 8.5 ACE inhibitors 53 (19.8) 40 (16.8) 0.389 7.7 47 (18.4) 41 (17.4) 3.4 ARBs 49 (18.3) 56 (23.5) 0.146 - 12.9 61 (23.8) 50 (21.0) 8.9 Treatment, n (%) Hydroxychloroquine 262 (97.8) 220 (92.4) 0.004 24.9 247 (97.1) 223 (94.8) 11.0 Lopinavir/Ritonavir 240 (89.6) 164 (68.9) \ 0.0001 52.6 206 (81.0) 180 (76.3) 10.0 Azithromycin 163 (60.8) 132 (55.5) 0.222 10.9 150 (59.0) 138 (58.4) - 2.5 Remdesivir 1 (0.4) 1 (0.4) 0.932 - 0.8 1 (0.4) 1 (0.4) 0.5 Interferon 124 (46.3) 72 (30.3) \ 0.0001 33.4 105 (41.2) 92 (39.0) 1.3 Table 1 continued Raw analysis IPTW analysis Tocilizumab Control group p value Standardised Tocilizumab Control group Standardised group (N = 268) (N = 238) difference (%) group (N = 254) (N = 235) difference (%) Steroids Steroids prior to D0 87 (32.5) 26 (10.9) \ 0.0001 54.1 61 (23.8) 51 (21.5) - 1.2 WHO classification D0 Admitted, no oxygen therapy 5 (1.9) 5 (2.1) 0.981 0.6 4 (1.6) 4 (1.6) 1.9 Admitted with oxygen 244 (91.0) 216 (90.8) 0.981 234 (91.9) 217 (92.3) therapy High flow 19 (7.1) 17 (7.1) 0.982 16 (6.5) 14 (6.1) Laboratory values, mean (SD) Lymphocytes (9 10E3/mm3) 834.3 (469.7) 902.9 (580.9) 0.235 - 13.0 843.3 (420.5) 854.1 (501.8) - 0.6 Lactate dehydrogenase (UI/l) 451.4 (186.2) 459.8 (339.8) 0.361 - 3.1 448.8 (179.6) 463.16 (288.2) - 9.1 D-dimer (ng/ml) 2200.2 (5216.8) 1950.9 (2588.7) 0.334 6.1 2040.8 (4412.3) 1869.9 (2553.4) - 1.5 C-reactive protein (mg/l) 149.5 (105.7) 148.0 (111.1) 0.584 1.4 148.54 (95.64) 146.0 (105.9) 4.6 Ferritin (ng/ml) 1682.6 (1427.9) 1363.7 (2953.5) \ 0.0001 13.7 1697.1 (1388.3) 1532.0 (2794.2) - 0.6 Interleukin 6 (IL-6) (pg/ml) 206.5 (468.5) 78.8 (94.9) \ 0.0001 37.8 195.4 (375.2) 80.3 (93.7) 42.4 Chest CT (at hospital admission), n (%) Normal imaging 4 (1.6) 9 (4.1) 0.192 3 (1.3) 7 (3.2) Unilateral pneumonia 18 (7.3) 26 (11.9) 17 (7.2) 25 (12.3) Bilateral interstitial 66 (26.9) 58 (26.6) 65 (27.7) 54 (24.7) pneumonia Patchy bilateral pneumonia 100 (40.8) 77 (35.3) 91 (38.7) 80 (36.5) Confluent bilateral 57 (23.3) 48 (22.0) 59 (38.7) 61 (23.3) pneumonia Infect Dis Ther Infect Dis Ther RESULTS difference (%) Standardised During the study period, 506 consecutive patients with COVID-19, respiratory insuffi- - 0.2 ciency, and increased inflammatory parameters were included. Among them, 268 were treated with tocilizumab and 238 patients were not. Control group 10.12 (5.42) Twenty-four patients who died in the first 24 h (N = 235) after developing inclusion criteria were excluded. Baseline Clinical Characteristics group (N = 254) IPTW analysis 538.8 (110.1) 827.9 (366.1) 10.41 (4.71) Characteristics of both cohorts are shown in Tocilizumab Table 1. Median time to tocilizumab from the onset of symptoms was 11 days (IQR 8–14 days). Among 268 patients treated with tocilizu- mab, 22 (8.2%) received 3 doses, 92 (34.3%) received 2 doses, and the remaining 154 difference (%) Standardised (57.4%) received only one dose. Median initial tocilizumab dose was 600 mg (IQR 400–600). 66.7 Median time from first dose to second dose was 1 day (IQR 1–1) and from first dose to third dose was 2 days (IQR 1–3). Median cumulative dose \ 0.0001 of tocilizumab was 600 mg (IQR 600–1000). p value A PS was developed to estimate each patient’s probability of receiving tocilizumab given their baseline characteristics and reduce Control group selection bias. Characteristics of the treatment 8.43 (4.67) and control cohort after adjustment by IPTW (N = 238) are shown in Table 1. In-Hospital Mortality of Patients Treated with Tocilizumab Compared to Patients group (N = 268) Not Treated with Tocilizumab 532.4 (109.5) 789.2 (354.8) 11.73 (5.20) Raw analysis Tocilizumab Global hospital mortality was 23.7%. Charac- teristics of survivors and non-survivors are shown in Online Appendix Table 2. Mortality was lower in patients treated with Time from onset of symptoms Initial tocilizumab dose (mg) Cumulative tocilizumab dose Tocilizumab treatment, n (%) tocilizumab than in controls [16.8% versus 31.5%, HR 0.514 (95% CI 0.355–0.744), p \ 0.001] (Table 2). Tocilizumab treatment to D0, mean (SD) Table 1 continued reduced mortality by 14.7% relative to no toci- lizumab treatment (RRR 46.7%). We calculated a number needed to treat of 7. Differences in mortality persisted after applying the PS adjus- (mg) ted for tocilizumab treatment [weighted HR 0.741 (95% CI 0.619–0.887), p = 0.001] Infect Dis Ther Table 2 Association between tocilizumab treatment and mortality in patients with SARS-COV-2 infection, according to tocilizumab and steroid-48 h exposure All Number of events Ratio IPTW ratio Tocilizumab, No tocilizumab, HR (95% CI) p value wHR (95% CI) p value N = 268 N = 238 Deaths 45 (16.8%) 75 (31.5%) 0.514 (0.355–0.744) < 0.001 0.741 (0.619–0.887) 0.001 Steroid-48 h N = 119 N = 87 Deaths 13 (10.9%) 35 (40.2%) 0.511 (0.352–0.741) < 0.001 0.600 (0.449–0.804) 0.006 No steroid- N = 149 N = 151 48 h Deaths 32 (21.5%) 40 (26.5%) 0.713 (0.447–1.137) 0.345 0.830 (0.650–1.048) 0.115 HR hazard ratio, IPTW inverse probability of treatment weighting, wHR weighted hazard ratio, steroid-48 h treated with steroids in the 48 h following inclusion *Weighted hazard ratios and 95% confidence intervals were obtained by inverse probability treatment weighting (Table 2). Figure 2 demonstrates differences in An interaction was found between tocilizu- the probability for survival for patients with mab and steroid-48 h use (interaction SARS-CoV2 infection, according to tocilizumab p = 0.094). Among patients treated with steroid- treatment (log-rank p \ 0.001). 48 h, mortality was lower in those treated with tocilizumab than in those treated with steroid- Tocilizumab Effect in Subgroups 48 h alone [10.9% vs. 40.2%, HR 0.511 (95% CI According to Baseline Characteristics 0.352–0.741), p = 0.036; weighted HR 0.6 (95% CI 0.449–0.804), p \ 0.001] (Table 2). The effect of tocilizumab treatment on mortal- Tocilizumab with steroid-48 h treatment ity in different subsets of patients was consis- reduced mortality by 29.1% relative to no toci- tent with a protective effect. Figure 3 shows lizumab treatment (RRR 72.8%). We calculated different effects of tocilizumab according to a number needed to treat of 4. Figure 4 shows different values of variables that displayed sig- differences in probability of survival according nificant interaction. Patients[65 years old with to steroid exposure in patients treated with lymphocyte counts \ 1000 cells/ll, hyperten- tocilizumab. sion, and cardiovascular disease seem to benefit the most from tocilizumab treatment. Adverse Events Effect of Tocilizumab in Patients Adverse events were monitored during the study with Post-Baseline Steroid Treatment period in the tocilizumab group. Although up to 32.6% in the tocilizumab group compared to One-hundred nineteen (44.4%) patients in the 30.3% in the control group had an increase in tocilizumab arm and 87 (36.6%) in the control SGOT/AST above the upper limit of normal, arm were treated with steroids in the 48 h fol- there were no significant differences among the lowing inclusion (steroid-48). Among them study groups. A total of 11 (4.1%) patients in the 71.3% patients received steroids in pulses, and tocilizumab group had serious adverse reactions, 28.7% non-pulsed steroids. Characteristics of including hepatotoxicity with increased liver patients according to tocilizumab and steroid- enzymes (3 patients) or bilirubin (2 patients), 48 h exposure are shown in Online Appendix thrombocytopenia (1 patient), catheter-related Table 3. superficial thrombophlebitis (2 patients), Infect Dis Ther Fig. 2 Probability of survival of patients with SARS-COV-2 infection according to tocilizumab exposure. Descriptive raw analysis Fig. 3 Forest plot of stratified analyses for mortality showing the weighted hazard ratio from IPTW analysis of tocilizumab treatment diarrhoea (1 patient), and headache and ocular DISCUSSION phosphenes (1 patient). Bacteraemia without a source after tocilizumab administration was Our results show that survival of patients with documented in one patient (0.4%). Median fol- severe COVID-19 is significantly higher in low-up time was 12 days (7–18 days). patients treated with tocilizumab than in those Infect Dis Ther Fig. 4 Probability of survival of patients with SARS-COV-2 infection treated with tocilizumab according to steroid-48 h exposure. Descriptive raw analysis not treated. In particular, mortality is reduced tocilizumab versus standard of care have been when tocilizumab is used concomitantly or published [17]. The study was prematurely shortly followed (first 48 h) by steroids. These interrupted because of futility (and recruitment results come from a large cohort of 506 con- difficulties) with one third of the initially secutive patients with COVID-19 pneumonia, planned study population. As recognised by the respiratory insufficiency, and increased inflam- authors, this small open-label clinical trial still matory parameters recruited across 18 hospitals leaves important questions unanswered on the from different regions in Spain. Of these, 268 potential role of tocilizumab in the treatment of were treated with tocilizumab whilst 238 COVID-19. patients were not. These results are considered Recently, steroid treatment has shown pro- clinically relevant as they contribute to the mise as a life-saving therapy in patients with growing body of evidence supporting the use of COVID-19 [18, 19], and, indeed, dexametha- tocilizumab in SARS-CoV-2 infection. sone has been granted EU regulatory approval Our data are consistent with previously for the treatment of COVID-19 patients on published series that report a reduction in the oxygen or mechanical ventilation, becoming risk of death in patients with severe COVID-19 the standard of care in clinical practice. There- treated with tocilizumab compared to standard fore, when analysing the effect of new treat- of care. Most of these are observational studies, ments, it is important to consider the combined some of them comparing tocilizumab use with a use of steroid therapy. control group [14, 15, 16]. Prior studies do not report specifically on the At the time of planning our study, results combined effect of tocilizumab and post-base- from randomised controlled trials had not yet line steroids, although in some of them steroids been published. Recently, the results of the first have been used concomitantly with tocilizumab randomised, open label, controlled trial of in a considerable proportion of cases [17% Infect Dis Ther standard of care (SoC) and 30% tocilizumab] However, this is an observational study and, [14], close to 100% in some cases [20, 21]. as such, has the inherent limitations of this Contrarily, in the only randomised clinical trial kind of study. Indication bias for steroid treat- concomitant use was not allowed unless ment was not adjusted for. It is possible that patients were on chronic treatment. A single- steroid treatment was not evenly administered centre Italian study jointly analysed data across the study population. This could possibly regarding the administration of both tocilizu- explain the high mortality in the subset of mab and steroids, together or separately, and patients who received steroid-48 h, but not suggested the need for evaluating combined use tocilizumab. of these therapies [22]. More recently, Narain The short follow-up time prevented us from et al. published the results of a retrospective detecting important adverse events such as cohort study and concluded that the combina- reactivation of latent infections in patients tion of corticosteroids with tocilizumab showed treated with tocilizumab, particularly in those superior survival outcome compared with SoC treated with tocilizumab and steroids. Thus, treatment and treatment with corticosteroids safety concerns might have been underesti- alone or in combination with anakinra [23]. mated. To overcome this problem, ongoing These results are consistent with our findings. registries have been put in place to identify The timing of steroid administration in medium-term complications. relation to tocilizumab treatment might be Although our cohort was limited to non-ICU decisive. A small non-controlled Spanish cohort patients, tocilizumab’s effect on critical patients study suggests that patients treated with prior or who already require mechanical ventilation or concomitant steroid therapy had better survival ICU admission deserves further investigation. results than those with steroids added at a later One recent article by Somers et al. [25] suggests stage [20]. Another short observational single- that tocilizumab in ICU patients is also benefi- centre study shows an improvement in mortal- cial, although the risk of infection is high. ity in patients treated with tocilizumab when salvage steroids are added in a median of 2.3 days [24]. In our study, tocilizumab adds to CONCLUSION steroid administered to non-intubated cases with severe COVID-19 during the first 48 h of Our study results show that survival of patients presenting with respiratory failure despite oxy- with severe COVID-19 is higher in patients that gen therapy (Brescia-COVID respiratory severity received tocilizumab as part of the standard of scale 2 or worse). care than in those who did not. The benefits The severity of the study population that appear higher when tocilizumab is given con- could most benefit from treatment remains to comitantly with steroids, mostly when given be determined. Some authors suggest that the within the first 48 h of presenting with respira- administration of tocilizumab is more effective tory failure despite oxygen therapy. Neverthe- in severe cases (PaO2/FiO2 \ 150) [14], while less, relevant questions remain unanswered. others report better results in cases with less Large, randomised, controlled studies are nee- oxygen requirement [21]. ded to convincingly establish the efficacy and In our study, tocilizumab showed an added safety of tocilizumab in the treatment of benefit to that of steroids in reducing mortality COVID-19. of severe COVID-19. The major strength of this study was its multicentric nature, with a large number of patients included from different ACKNOWLEDGEMENTS regions in Spain. The study also included a concurrent comparison with a non-treated We thank the members of the TOCICOV study cohort. The IPTW to adjust for indication bias group for their contribution to the work. adds to the strength of the results. Infect Dis Ther TOCICOV Study Group Collaborators: HOS- LA VICTORIA Enrique Nuño Álvarez, MI PITAL DEL MAR-PARC DE SALUT MAR. Lucena. HOSPITAL COSTA DEL SOL. Josefa X Fernández Sala, P. Dı́az Pellicer, S. Grau Cer- Andrea Aguilar Garcı́a, José Javier Garcı́a rato. HOSPITAL UNIVERSITARIO DE PUERTO Alegrı́a. REAL. José Manuel Dodero Anillo, Alberto Romero Palacios. HOSPITAL UNIVERSITARIO Funding. No funding or sponsorship was VIRGEN DEL ROCÍO. Elisa Cubiles, Ma Ángeles received for this study or publication of this Lobo Acosta, Reyes Fresneda Gutierrez, Alicia article. The Rapid Service Fee was funded by the Marı́an Candón, Sonsoles Salto Alejandre, Jose authors. Miguel Cisneros Herreros, Elisa Cordero Matı́a, Carmen Infante Dominguez, Juan Carlos Cre- Authorship. All named authors meet the spo Rivas, Macarena López Verdugo. HOSPITAL International Committee of Medical Journal UNIVERSITARIO DE CANARIAS. Isabel Suarez Editors (ICMJE) criteria for authorship for this Toste. Paloma Dı́az Pérez. HOSPITAL UNI- article, take responsibility for the integrity of VERSITARI DE BELLVITGE. Francesca Mitjavila the work as a whole, and have given their Villero, Guillermo Suárez Cuartin, Carlota approval for this version to be published. Gudiol Gonzalez, Adriana Iriarte Fuster, Mercè Gasa Galmes, Sandra Pérez, Dolores Rodrı́guez Authorship Contribution. Conceptualiza- Cumplido. HOSPITAL UNIVERSITARIO Nra Sra. tion and study design: ASL, BRA, AFC. CANDELARIA. Arı́stides de León Gil, Emilio J. Methodology: BRA, AFC, ASL, F-T. Data collec- Sanz. HOSPITAL GENERAL UNIVERSITARIO DE tion: all authors. Data interpretation: ASL, BRA, VALENCIA. Francisco Sanz-Herrero, Francesc AFC, FT. Writing first draft: ASL, BRA, AFC. Puchades, Pilar Ortega-Garcı́a. C.H. UNIV. DE Critical revision for important intellectual con- SANTIAGO DE COMPOSTELA. José Antonio tent: all authors. Final approval: All authors. All Dı́az-Peromingo, Carlos Rodrı́guez-Moreno. authors agree to be accountable for all aspects of HOSPITAL UNIVERSITARIO CLÍNICO SAN the work by ensuring that questions related to CARLOS Daniel Lozano, Ana Terleira. HOSPI- the accuracy or integrity of any part of the work TAL UNIVERSITARIO TORRECÁRDENAS. Marı́a will be appropriately investigated and resolved. Estela González Castro, Sergio Ferra Murcia, AFC and BRA had full access to all the data in Elena Marı́a Gázquez Aguilera. HOSPITAL UNI- this study and take complete responsibility for VERSITARIO DE LA PRINCESA. Elena Pintos- the integrity of the data and the accuracy of the Sanchez, Pablo Zubiaur, Elena Santos-Molina, data analysis. Marcos Navares-Gómez, Gina Mejia Abril. HOSPITAL CENTRAL DE LA DEFENSA GÓMEZ Prior Presentation. This manuscript is based ULLA. Amelia Garcı́a-Luque, Miguel Puerro-Vi- on work that has been previously presented at cente, Marı́a Jesús Sánchez Carrillo. HOSPITAL medRxiv, posted 10 September 2020 (https:// UNIVERSITARIO RAMÓN Y CAJAL. M Ángeles www.medrxiv.org/content/10.1101/2020.09. Gálvez Múgica, Mónica Aguilar Jiménez, Cris- 07.20189357v2). tina Sánchez Dı́az. HOSPITAL UNIVERSITARIO PUERTA DE HIERRO. Cristina Avendaño Solá, Disclosures. Belén Ruiz-Antorán, Aránzazu Patricia Mills-Sánchez, Manuel Valle, Gustavo Sancho-López, Ferrán Torres, Vı́ctor Moreno- Centeno, Concepción Payares, Antonio F. Torres, Itziar de Pablo-López, Paulina Garcı́a- Caballero Bermejo, Elena Diago, Rosa Malo de López, Francisco Abad-Santos, Clara M. Rosso- Molina, Juan Antonio Vargas Nuñez, Elena Fernández, Ana Aldea-Perona, Eva Montané, Muñez, Antonio Ramos. HOSPITAL UNIV. Ruth M. Aparicio-Hernández, Roser Llop-Rius, GERMANS TRIAS I PUJOL Ana Marı́a Barrio- Consuelo Pedrós, Paloma Gijón, Carolina Her- canal, Ana Pilar Pérez-Acevedoc, Melani Núñez- nández-Carballo, Marı́a J Pedrosa-Martı́nez, Montero. HOSPITAL UNIVERSITARIO GRE- Consuelo Rodrı́guez-Jiménez, Guillermo Prada- GORIO MARAÑON. Marı́a Olmedo, Sofia De la Ramallal, Lourdes Cabrera-Garcı́a, Josefa A. Villa. HOSPITAL UNIVERSITARIO VIRGEN DE Aguilar-Garcı́a, Rocı́o Sanjuan-Jimenez, Evelyn Infect Dis Ther I. Ortiz-Barraza, Enrique Sánchez-Chica and Ana REFERENCES Fernández-Cruz declare no conflicts of interest. 1. WHO. Coronavirus disease (COVID-19) Situation Compliance with Ethics Guidelines. The Report—164 [Internet]. 2020. https://www.who. study was approved by the Spanish Regulatory int/docs/default-source/coronaviruse/situation-rep Authority (Spanish Agency of Medicines and orts/20200702-covid-19-sitrep-164.pdf?sfvrsn=ac07 4f58_2. Accessed 4 Jul 2020. Medical Devices, AEMPS) and by the Research Ethics Committee (REC) at Hospital Universi- 2. Cascella M, Rajnik M, Cuomo A, Dulebohn SC, Di tario Puerta de Hierro-Majadahonda (FIB-TOC- Napoli R. Features, evaluation and treatment coro- 2020-01), and a waiver for the informed consent navirus (COVID-19). In: StatPearls. StatPearls Pub- was granted. The study protocol was submitted lishing; 2020. to the local RECs at each study site and in line 3. Mehta P, McAuley DF, Brown M, Sanchez E, Tat- with Spanish legislation, some provided their tersall RS, Manson JJ. COVID-19: consider cytokine own additional approval while others did not storm syndromes and immunosuppression. Lancet need to undergo the full review and approval (London, England). 2020;395:1033–4. process, i.e. they recognised the initial REC 4. McGonagle D, Sharif K, O’Regan A, Bridgewood C. approval. The study complied with the provi- The role of cytokines including interleukin-6 in sions in European Union (EU) and Spanish leg- COVID-19 induced pneumonia and macrophage islation on data protection and the Declaration activation syndrome-like disease. Autoimmun Rev. 2020;19(6):102537. of Helsinki 2013. 5. Li G, Fan Y, Lai Y, Han T, Li Z, Zhou P, et al. Data Availability. The datasets generated Coronavirus infections and immune responses. during and/or analysed during the current J Med Virol. 2020;92(4):424–32. study are available from the corresponding 6. Fu B, Xu X, Wei H. Why tocilizumab could be an author on reasonable request. effective treatment for severe COVID-19? J Transl Med. 2020;18:164. Open Access. This article is licensed under a Creative Commons Attribution-Non- 7. Xu X, Han M, Li T, Sun W, Wang D, Fu B, et al. Effective treatment of severe COVID-19 patients Commercial 4.0 International License, which with tocilizumab. Proc Natl Acad Sci USA. permits any non-commercial use, sharing, 2020;117(20):10970–5. adaptation, distribution and reproduction in any medium or format, as long as you give 8. Foca E. Linee guida sulla gestione terapeutica e di supporto per pazienti con infezione da coronavirus appropriate credit to the original author(s) and COVID-19. 2020; Marzo. http://www.fvcalabria. the source, provide a link to the Creative unicz.it/COVID-19/LINEE-GUIDA/linee-guida-SIMIT Commons licence, and indicate if changes were -marzo-2020.pdf. Accessed 1 Oct 2020. made. The images or other third party material 9. D’Agostino RBJ. Propensity score methods for bias in this article are included in the article’s reduction in the comparison of a treatment to a Creative Commons licence, unless indicated non-randomized control group. Stat Med. otherwise in a credit line to the material. If 1998;17(19):2265–81. material is not included in the article’s Creative 10. Rosenbaum PR, RDTcro. The central role of the Commons licence and your intended use is not propensity score in observational studies for causal permitted by statutory regulation or exceeds the effects. Biometrika. 1983;70:41–55. permitted use, you will need to obtain permis- sion directly from the copyright holder. To view 11. Austin PC. An introduction to propensity score methods for reducing the effects of confounding in a copy of this licence, visit http:// observational studies. Multivariate Behav Res. creativecommons.org/licenses/by-nc/4.0/. 2011;46(3):399–424. 12. Cohen DJ. Statistical power analysis for the behav- ioral sciences. New York: Routledge; 2013. https:// doi.org/10.4324/9780203771587. Infect Dis Ther 13. Austin PC. Balance diagnostics for comparing the Chemother. 2020;64(9):e01168. https://doi.org/10. distribution of baseline covariates between treat- 1128/AAC.01168-20. ment groups in propensity-score matched samples. Stat Med. 2009;28(25):3083–107. 20. Campins L, Boixeda R, Perez-Cordon L, Aranega R, Lopera C, Force L. Early tocilizumab treatment 14. Guaraldi G, Meschiari M, Cozzi-Lepri A, Milic J, could improve survival among COVID-19 patients. Tonelli R, Menozzi M, et al. Tocilizumab in patients Clin Exp Rheumatol. 2020;38:578. with severe COVID-19: a retrospective cohort study. Lancet Rheumatol. 2020;2(8):e474–84. 21. Gorgolas M, Cabello A, Prieto Perez L, Villar Alvarez F, Alvarez Alvarez B, Rodriguez Nieto MJ, et al. 15. Martı́nez-Sanz J, Muriel A, Ron R, Herrera S, Pérez- Compassionate use of tocilizumab in severe SARS- Molina JA, Moreno S, et al. Effects of tocilizumab on CoV2 pneumonia. When late administration is too mortality in hospitalized patients with COVID-19: a late. medRxiv [Internet]. 2020;2020.06.13. multicentre cohort study. Clin Microbiol Infect. 20130088. http://medrxiv.org/content/early/2020/ 2020. https://doi.org/10.1016/j.cmi.2020.09.021. 06/16/2020.06.13.20130088.abstract. Accessed 5 Oct 2020. 16. Rodrı́guez-Baño J, Pachón J, Carratalá J, Ryan P, Jarrı́n I, Yllescas M, et al. Treatment with tocilizu- 22. Mikulska M, Nicolini LA, Signori A, Di Biagio A, mab or corticosteroids for COVID-19 patients with Sepulcri C, Russo C, et al. Tocilizumab and steroid hyperinflammatory state: a multicentre cohort treatment in patients with COVID-19 pneumonia. study (SAM-COVID-19). Clin Microbiol Infect. PLoS ONE. 2020;15(8):e0237831. 2020. https://doi.org/10.1016/j.cmi.2020.08.010. 23. Narain S, Stefanov DG, Chau AS,Weber AG, Marder 17. Salvarani C, Dolci G, Massari M, Merlo DF, Cavuto G, Kaplan B, et al. Comparative survival analysis of S, Savoldi L, et al. Effect of tocilizumab vs standard immunomodulatory therapy for coronavirus dis- care on clinical worsening in patients hospitalized ease 2019 cytokine storm. Chest. 2020. https://doi. with COVID-19 pneumonia: a randomized clinical org/10.1016/j.chest.2020.09.275. trial. JAMA Intern Med. 2020. https://doi.org/10. 1001/jamainternmed.2020.6615. 24. Sanz Herrero F, Puchades Gimeno F, Ortega Garcı́a P, Ferrer Gómez C, Ocete Mochón MD, Garcı́a 18. Fadel R, Morrison AR, Vahia A, Smith ZR, Chaudhry Deltoro M. Methylprednisolone added to tocilizu- Z, Bhargava P, et al. Early short course corticos- mab reduces mortality in SARS-CoV-2 pneumonia: teroids in hospitalized patients with COVID-19. an observational study. J Internal Med. 2020. Clin Infect Dis. 2020;71(16):2114–20. https://doi. https://doi.org/10.1111/joim.13145. org/10.1093/cid/ciaa601. 25. Somers EC, Eschenauer GA, Troost JP, Golob JL, 19. Fernández-Cruz A, Ruiz-Antorán B, Muñoz-Gómez Gandhi TN, Wang L, et al. Tocilizumab for treat- A, Sancho-López A, Mills-Sánchez P, Centeno-Soto ment of mechanically ventilated patients with GA, et al. A retrospective controlled cohort study of COVID-19. MedRxiv. 2020. https://doi.org/10. the impact of glucocorticoid treatment in SARS- 1101/2020.05.29.20117358. Update in: Clin Infect CoV-2 infection mortality. Antimicrob Agents Dis. 2020.