Successful immunomodulators for treatment of COVID-19 have opened the pathway for comparative trials.

Since COVID-19 became a dominant factor in our daily lives, we have learned a tremendous amount about it through both study and shared experience. Therapies for patients hospitalized for COVID-19 infection have moved from ineffective shots in the dark to mildly effective antivirals to mortality-reducing immunomodulatory therapies.[1] The widely accepted view of COVID-19 as a dual phase illness with a viral component and inflammatory component has led to the successful search for immunomodulator therapies to attenuate host response and improve outcomes.[2,3] As studies have advanced, our standard of care has as well; patients with severe and critical COVID-19 now receive combinations of these corticosteroids with JAK inhibitors or IL-6 inhibitors.[[4], [5], [6]] What has largely not occurred thus far are comparative, head-to-head trials to determine which therapies are best. Two comparison trials exist, only one of which is direct. One is the REMAP-CAP trial, comparing tocilizumab and sarilumab separately to standard of care.[7] REMAP-CAP showed improved outcomes in patients receiving tocilizumab or sarilumab with corticosteroids with progressive critical COVID-19 infection compared to controls, though only 48 patients received sarilumab in the study compared to 353 for tocilizumab.[7] The other, by Karampitsakos and colleagues, is a preprint directly comparing tocilizumab to baricitinib in patients with a PaO2/FiO2 <200.[8] This open-label randomized study of 251 patients showed noninferiority of baricitinib to tocilizumab using their criteria (upper bound of the hazard ratio for death or mechanical ventilation not exceeding 1.5).[8].

The paucity of head-to-head trials comparing established treatment options has led different guideline panels to make different recommendations, even though informed by a similar body of evidence (Table 1 ).[[4], [5], [6]] For example, the World Health Organization (WHO) recommends either tocilizumab, baricitinib, or sarilumab, whereas the Infectious Diseases Society of America (IDSA) and US National Institutes of Health (NIH) both recommend tocilizumab and sarilumab; however, recognize sarilumab as an alternative when tocilizumab is not available instead of recommending it on at the same level.[[4], [5], [6]] These stepped recommendations from IDSA and NIH recognize the lack (at the time) of trials directly comparing these treatments and the limited certainty from indirect comparisons.

Table 1

Guideline recommendations for IL-6 inhibitors and baricitinib for COVID-19[[4], [5], [6]]

	World Health Organization	Infectious Diseases Society of America	National Institutes of Health (US)
Tocilizumab	Recommended for patients with severe or critical COVID-19	Suggested for patients with progressive severe or critical COVID-19 and elevated markers of inflammation	Recommended for patients with progressive severe or critical COVID-19 and elevated markers of inflammation
Sarilumab	Recommended for patients with severe or critical COVID-19	Suggested for patients with progressive severe or critical COVID-19 and elevated markers of inflammation when tocilizumab is not available	Recommended for patients with progressive severe or critical COVID-19 and elevated markers of inflammation when tocilizumab is not available
Baricitinib	Recommended for patients with severe or critical COVID-19	Suggested for patients with severe COVID-19	Recommended for patients with progressive severe or critical COVID-19 and elevated markers of inflammation

In the absence of a robust body of evidence from trials reporting on direct comparisons, in this issue of Clinical Microbiology and Infection Albuquerque and colleagues propose a Bayesian meta-regression to indirectly assess the efficacy of sarilumab and baricitinib in comparison to tocilizumab.[9] Authors initially conducted several pairwise Bayesian meta-analyses studying the effects of tocilizumab, sarilumab, and baricitinib verses an inactive comparison (i.e., placebo or standard of care) when added to corticosteroid therapy for COVID-19 on 28-day mortality. The meta-regression, based on multiple levels of assumptions, aimed to determine probabilities of noninferiority of sarilumab and baricitinib to the reference of tocilizumab. The investigators present four models informed by different priors, including using results from two recent head-to-head trials comparing the active treatments. While recognizing wide variability, they concluded that sarilumab and baricitinib may have high probabilities of noninferiority to tocilizumab for the treatment of COVID-19, all in combination with corticosteroids. Authors recognized the heterogeneity in the primary meta-analyses, as the methods used for this indirect comparison could not completely account for differences in study populations across different trials, such as age, severity of disease, circulating variant, etc. Similarly, it is unclear if the meta-regression considered the certainty of evidence for the pairwise meta-analyses, as this varied across therapies. While intra-study differences were small, the studies had different inclusion criteria and resultingly different proportions of patients requiring intensive respiratory therapy, which is the population where tocilizumab shows greatest benefit.

What is a clinician to do with these data? While it is reassuring that the effectiveness of sarilumab and baricitinib may not be inferior to tocilizumab for 28-day mortality, it is still important to take the findings with caution. The IL-6 inhibitors in particular have had a difficult road to determine where their utility lies for COVID-19. A quick scan of the meta-analysis shows that there has been no shortage of studies with either tocilizumab or sarilumab. Some of these early studies were underpowered due to low enrollment, low event rates, inclusion of patients at low risk of poor outcomes, and/or low numbers of patients receiving concomitant corticosteroid therapy, highlighting the importance of pooling the results in a meta-analysis.[5] Many of the studies of both tocilizumab and sarilumab showed no benefit, and it was not until they were studied in critically ill patients on corticosteroids that an effect on mortality was seen. The eligibility for inclusion of primary studies is appropriately limited to evaluating patients who received concomitant corticosteroids; however, this comes with the potential for loss of randomization from trials that did not stratify based on receipt of corticosteroids, as well as lumping patients from subgroups of earlier studies with entire populations of later studies. Both of these have the potential for introducing heterogeneity into the analyses due to changing study characteristics such as standards of care, recognition of severe disease, and concomitant therapies.

Ultimately, the study is a useful reference but does not supersede the need for direct comparisons and/or fit-for-purpose analyses, such as a network meta-analysis of therapies for patients with severe COVID-19. Separately, there are sufficient data showing efficacy of both baricitinib and tocilizumab compared to standard of care, and in overlapping levels of severity both therapies are good options for patients with severe illness. Though studies published on populations earlier in the pandemic of baricitinib included few patients on mechanical ventilators, an analysis of patients on mechanical ventilators or ECMO in the COV-BARRIER study showed a mortality reduction.[10] The data supporting sarilumab are more sparse, which is why recommendations by the IDSA and NIH guideline panels are measured and recommend sarilumab only when better-studied therapies are not available. In the absence of more robust studies, to recommend it on parity with tocilizumab and baricitinib is premature. As our standards of care for severe COVID-19 have advanced, our standards of evidence should as well.

Transparency

Potential conflicts of interest:

JCG has received royalties for book authorship from Jones and Bartlett Learning; consulting fees from Qpex, Shionogi, Merck, Accelerate Diagnostics, Achaogen, Astellas Pharma, Melinta Therapeutics, Nabriva Therapeutics, Paratek Pharma, scPharmaceuticals, Spero Therapeutics, Tetraphase Pharmaceuticals; speaker honoraria from Astellas Pharma, Melinta Therapeutics, Merck, and Shionogi; travel support from Merck; research support from Merck.

RM has no conflicts of interest.

No external funding was received for this work.

Contributions

JCG is the lead and corresponding author. Writing- Original Draft: JCG. Writing- Review and Editing: JCG and RM.