Systematic review and meta-analysis of tocilizumab in persons with coronavirus disease-2019 (COVID-19)

We performed a meta-analysis to determine safety and efficacy of tocilizumab in persons with coronavirus disease-2019 (COVID-19). We searched PubMed, Web of Science and Medline using Boolean operators for studies with the terms coronavirus OR COVID-19 OR 2019-nCoV OR SARS-CoV-2 AND tocilizumab. Review Manager 5.4 was used to analyze data and the modified Newcastle–Ottawa and Jadad scales for quality assessment. We identified 32 studies in 11,487 subjects including three randomized trials and 29 cohort studies with a comparator cohort, including historical controls (N = 5), a matched cohort (N = 12), or concurrent controls (N = 12). Overall, tocilizumab decreased risk of death (Relative Risk [RR] = 0.74; 95% confidence interval [CI], 0.59, 0.93; P = 0.008; I2 = 80%) but not of surrogate endpoints including ICU admission (RR = 1.40 [0.64,3.06]; P = 0.4; I2 = 88%), invasive mechanical ventilation (RR = 0.83 [0.57,1.22]; P = 0.34; I2 = 65%) or secondary infections (RR = 1.30 [0.97,1.74]; P = 0.08; I2 = 65%) and increased interval of hospitalization of subjects discharged alive(mean difference [MD] = 2 days [<1, 4 days]; P = 0.006; I2 = 0). RRs of death in studies with historical controls (RR = 0.28 [0.16,0.49; P < 0.001]; I2 = 62%) or a matched cohort (RR = 0.68 [0.53, 0.87]; P = 0.002; I2 = 42%) were decreased. In contrast, RRs of death in studies with a concurrent control (RR = 1.10 [0.77, 1.56]; P = 0.60; I2 = 85%) or randomized (RR = 1.18 [0.57,2.44]; P = 0.66; I2 = 0) were not decreased. A reduced risk of death was not confirmed in our analyses which questions safety and efficacy of tocilizumab in persons with COVID-19.

We conducted a systematic review and meta-analysis of 32 studies of safety and efficacy of tocilizumab in persons with COVID-19 which had a comparator cohort. Overall, we found tocilizumab decreased risk of death but not rates of intensive care unit(ICU) admission, invasive mechanical ventilation, secondary infections, and increased interval of hospitalization in persons discharged alive. However, a reduced risk of death was not confirmed in our analyses of studies with concurrent controls nor randomized trials. These data question safety and efficacy of tocilizumab in persons with COVID-19.

Search strategy and selection criteria
PubMed, Web of Science and Medline were searched using Boolean operators for studies with the terms coronavirus OR COVID-19 OR 2019-nCoV OR SARS-CoV-2 AND tocilizumab. Start and stop dates were 2020/1/1 and 2020/ 10/27. Two investigators independently reviewed abstracts of identified citations and selected articles for full review. Discordances were resolved by a third reviewer. Results were also manually searched and reviewed. We found 1492 articles excluding 914 duplicates. After further review we focused on 32 studies, 29 non-randomized comparator studies, and three RCTs [1,2,. A flow diagram of the search strategy and article selection is displayed in Fig. 1. Review Manager 5.4 was used to analyze data and the modified Newcastle-Ottawa score (NOS) and Jadad scale for quality assessment.

Inclusion and exclusion criteria
Inclusion criteria included English language reports of clinical trials and observational studies with a comparator cohort and with outcomes reporting, but not limited to, survival. Reviews and case reports were excluded as were studies with a NOS < 6.

Risk of bias assessment
Risk of bias was assessed using the Jadad scale in four domains: (1) random sequence generation, (2) allocation concealment, (3) blinding of participants, and (4) complete reporting of withdrawals and dropouts [53]. Methodological quality of comparator studies was assessed using the modified Newcastle-Ottawa scale (NOS) [54,55] consisting of three domains: (1) subject selection, (2) comparability of the study groups, and (3) outcomes assessment. A score of 0-9 was allocated to each relevant study. Observational studies with a NOS score <6 (N = 1) were excluded [56].

Statistics
We pooled data and utilized Relative Risks (RRs) and Confidence Intervals (CIs) to describe dichotomous outcomes, including risk of death, ICU admission, invasive mechanical ventilation, secondary infection . We used mean difference (MD) and CIs for continuous outcomes including interval of hospitalization. We grouped the cohort studies into unmatched historical controls and concurrent controls, matched and unmatched or subgroup analyses. A fixedeffects model was used when I 2 ≤ 50% and the Cochran Q statistic P > 0.1 and a random-effects model when I 2 > 50% and Q statistic P ≤ 0.1. Funnel plots were used to screen for potential publication bias. Statistical analyses were carried out with Review Manager 5.4 (Cochrane Collaboration)

Surrogate clinical endpoints
To test the efficacy of tocilizumab on rate of ICU admission we included seven studies [25-27, 31, 41, 42, 45] Fig. 3 The impact of tocilizumab on ICU admission. Risk of ICU admission.

Publication bias
Potential for publication bias is shown in Fig. 6. We found potential publication bias in studies of death in subjects receiving or not receiving tocilizumab with some studies falling outside the 95% CI of the funnel plot. There was publication bias in studies included in the meta-analysis.

Discussion
Increased concentrations of inflammatory cytokines (IL-6, granulocyte-macrophage colony stimulating factor (GM-CSF) and tumor necrosis factor-a (TNF-a) are reported in persons with COVID-19 [60,61]. IL-6 is produced by diverse immune cells and implicated in development of acute respiratory distress syndrome (ARDS) and CRS Fig. 5 The correlation between tocilizumab and secondary infection. Risk of secondary infection. [62,63]. Some data suggest increased IL-6 concentrations correlate with risk of death [61,64]. Several metaanalyses claim tocilizumab is safe and effective in COVID-19 [17,21,23,24]. Most studies we include gave tocilizumab to subjects with evidence of inflammation including a CRP concentration >100 mg/L, a ferritin concentration >900 ng/ml and/or a D-dimer concentration >1500 ug/L [1].
Evaluating all 32 studies we found tocilizumab reduced risk of death but not several surrogate endpoints, including ICU admission, invasive mechanical ventilation, and secondary infections. Hospitalization interval was significantly increased. However, in our analysis of RCTs and studies with a concurrent control cohort we could not confirm a decreased risk of death. This conclusion differs from most prior meta-analyses [17][18][19][20][21][22][23][24] which failed to analyze outcomes by study-design (Table 3). A recent meta-analysis concluded an association between tocilizumab and lower mortality by low certainty evidence from cohort studies [22]. Our data contradict this assumption. Two recent analyses which included only studies with a comparator cohort support our conclusion [65,66]. Also, Mao and colleagues reported use of tocilizumab did not decrease risk of death possibly because of an increased risk of secondary bacterial infections [67].
There are important limitations to our study. Firstly, subjects were heterogeneous in COVID-19 severity although most had severe to critical COVID-19. Secondly, many studies were observational and lacked an appropriate control cohort. We tried to overcome potential biases in these studies by analyzing outcomes by study-type.  RR relative risk, OR odds ratios, RD risk differences, RCTs randomized controlled trials.
In conclusion, tocilizumab decreased risk of death but not rates of surrogate endpoints including ICU admission, invasive mechanical ventilation, secondary infections and did significantly alter interval of hospitalization. A reduced risk of death was not confirmed in our meta-analysis of randomized trials or studies with a concurrent control cohort. These data question safety and efficacy of tocilizumab in persons with COVID-19. Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons. org/licenses/by/4.0/.