COVID-19, steroids and other immunomodulators: The jigsaw is not complete.

After influenza A1-09 in 2008–2009, the global pandemic caused by the novel coronavirus SARS-CoV-2 infection (COVID-19) is the second pandemic of the 21st century. While it has not ended yet, it has caused a level of morbidity, mortality, social and economic disruption not seen since the 1918 influenza pandemic. It is also the source of an unprecedented research effort which has allowed us to better understand this new human coronavirus in only a few months. Faced with the challenge of a new virus with no specific treatment, and due to the belief that immunomodulatory therapies have been hailed for years as the next revolution in sepsis [1], it came as no surprise that they have rapidly been tested in this setting. However, aided by social media and pre-publication servers, every new proposed therapy has garnered extensive coverage and even political debate before the evidence could be peer-reviewed. Worse, the way in which many clinicians have leapt to include these same agents into their standard of care is a source of concern and goes against the need to carefully evaluate evidence before subjecting our patients to new treatments.

Among possible candidates, corticosteroids were one of the first tested immunomodulatory agents, although research in previous viral outbreaks (influenzae [2], MERS [3] or SRAS [4]) did not advocate for their use in those indications. However, the lack of high quality randomised controlled trials (RCT) on this specific topic keeps the controversy alive.

Conversely, some specific septic indications in which the net immune response tilts in favour of pro-inflammatory pathways suffer no further debate. The most notable example being bacterial meningitis [5] where potent inflammation in the subarachnoid space is linked to unfavourable outcome, which corticosteroids have been shown to reduce. In critical care, equipoise persist about somewhat heavily documented and overlapping topics such as acute respiratory distress syndrome (ARDS), Community Acquired Pneumonia (CAP) or general sepsis.

In severe CAP, corticosteroids use has resurfaced owing specifically to a large recent RCT [6], which found less treatment failure when methylprednisolone was used as an adjunctive therapy in patients with severe CAP and high inflammatory response. In ARDS, the use of steroids was associated with decreased mortality for persistent [7] and early ARDS [8] in rather small RCTs. Some recent meta-analyses headed by the ESICM/SCCM allowed them to recommend their use in ARDS in their 2017 guidelines with a 28-day taper (risk ratio for death 0.75; 95% CI, 0.59–0.95 with steroids) [9]. Just before the global SARS-CoV-2 pandemic hit, the DEXA-ARDS [10] may have finally sealed the deal finding a large benefit to the use of large doses of corticosteroids (up to 20 milligrams a day of dexamethasone) in both ventilator-free-days at day 28 and mortality in early ARDS.

In COVID-19 infection, the largest RCT is the RECOVERY [11] trial whose preliminary results demonstrated a significant 2.7% overall 28-day mortality reduction to dexamethasone (6 mg per day for 10 days) compared with the usual care group. Prespecified subgroup analysis showed this reduction to be larger in patients receiving oxygen supplementation through invasive mechanical ventilation (MV) (RR 0.64, 95% CI [0.51 – 0.81]) or without invasive MV (RR 0.82, 95% CI [0.72 to 0.94]). Nevertheless, and despite the large size of this study (2104 patients in the dexamethasone arm, 4321 in the other arm), there are a number of important concerns and caveats; dexamethasone was only one arm of an open-label platform trial which did not report side effects and the absence of efficacy in important subgroups driving COVID-19 mortality (patients over 70 or 80 years) should mitigate the message. This trial also gave clues about which subgroups may reap the most benefits from steroids; indeed, women did not fare better with them and early administration (in the first 7 days of the disease) did not lead to improvement either. Furthermore, Metcovid, a Brazilian double-blind RCT, [12] comparing the use of 0.5 mg/kg of methylprednisolone to placebo and enrolling 393 patients, found no difference in 28-day mortality (37.1% vs 38.2%, p =  0.629), although the patients were significantly younger, appeared to have a more severe disease at enrolment (33.8% on mechanical ventilation vs 15.5%) and received significantly higher doses of steroids (a 100 kg patient would have received 50 mg of methylprednisolone, equivalent to roughly 9.4 mg of dexamethasone).

Another Brazilian study, the CoDEX trial [13], an open-label RCT enrolling 299 patients, also compared dexamethasone to placebo in patients with COVID-19 and moderate to severe ARDS. The investigators used a much higher dose regimen than in RECOVERY (20 mg for 5 days then 10 mg for 5 days), which resulted in an increase in ventilator-free-days defined as days alive and free of MV at day 28 (2.26 days, 95% CI, 0.2–4.4). There was no difference in 28-day mortality, which was a secondary outcome.

A French study by Dequin and colleagues [14] was terminated early due to the results of RECOVERY after enrolling 149 patients into placebo or hydrocortisone 200 mg/day (e.g. 7.5 mg of dexamethasone) for 7 days followed by 7 days of tapering. They reported a trend to a lower death rate with steroids (14.7% vs 27.4%) which did not reach statistical significance. Over 80% of the patients required mechanical ventilation at baseline with a mean of around 9.5 days of symptoms prior to enrolment. No separate analysis of age or gender have been published, however the more even gender distribution in this study may be important if there is less benefit to steroids in women. The corticosteroids domain of the REMAP-CAP study [15] was also terminated early for the same reason. This trial recruited 143 patients into either 50 or 100 mg of hydrocortisone every 6 h, 152 patients who received 50 mg only if shock was present and 108 patients into a no steroid group. There was also a trend to lower mortality in the steroids group but it did not reach statistical significance (30% vs 33%). No information was provided in this study on age group or gender. Most studies on the topic did not report more serious adverse events in the steroids groups; however, the Metcovid trial did show it led to significantly higher insulin doses. While this may not be a problem in most ICU settings, hyperglycaemia is associated with worse outcomes in COVID-19 [16].

Based on available evidence, the NIH recently recommended the use of dexamethasone at a daily dose of 6 mg (oral or intravenous) or hydrocortisone 160 mg/day, for up to 10 days, in hospitalised patients with COVID-19 who require delivery of oxygen through a high-flow device or non-invasive ventilation and in those who require invasive mechanical ventilation or extracorporeal membrane oxygenation. For hospitalised patients who require low-flow supplemental oxygen, the use of steroids is optional and should be associated with an antiviral (remdesivir) because of a theoretical risk of deceleration of viral clearance.

Inhibitors of interleukin (IL)-6 have also been rapidly tested to dampen the inflammatory response associated with COVID-19 pneumonia because high blood levels were associated with systemic inflammation and hypoxic respiratory. Two classes of EMA-approved IL-6 inhibitors exist: anti-IL-6 receptor monoclonal antibodies (e.g., sarilumab, tocilizumab) and anti-IL-6 monoclonal antibodies (siltuximab). Mostly used for autoimmune disorders, tocilizumab was first proposed more than 10 years ago as an effective treatment in refractory rheumatoid arthritis and from there spread to other inflammatory diseases such as giant cell arteritis and juvenile arthritis. It also has the perk to be known to ICU physicians, albeit for a shorter time. Indeed, it was approved by the FDA in 2018 as an effective therapy for “Cytokine Release Syndrome” (CRS), a known complication of the targeted therapies (mostly CAR-T-Cells) used primarily for a growing corpus of haemopathies. The pathophysiology of CRS is complex but heavily relying on the IL-6 – IL-6-R classical pathway to which the inflammatory response to COVID-19 has quickly been compared [17] and observational data suggested that both Tocilizumab and Sarilumab [18], [19] may benefit patients with COVID-19.

Most recently, results from two RCTs have been published [20], [21]. First, the CORIMUNO-TOCI-1 trial randomised 131 patients with COVID-19 requiring oxygen in conventional wards to either Tocilizumab or Usual Care. Of the two thresholds for treatment efficacy predefined in the Bayesian analysis, the authors reported better survival without mechanical ventilation in the intervention group but no difference in the clinical progression scale. Salvarini et al. also enrolled 126 hospitalised patients with moderate to severe COVID-19 pneumonia but had to terminate their study prematurely for futility when interim analysis found no difference on their primary composite outcome (defined as death from all cause, ICU admission or clinical worsening). This negative result may in part be attributed to the study protocol allowing compassionate use of tocilizumab in the usual care group (14 out of 60 patients received it) or to the selection of lower-risk patients. Indeed, the higher mortality reported in the CORIMUNO-TOCI-1 trial (13% vs 2.4%) despite using comparable doses of Tocilizumab underscores the fact that their population was older and presented with more comorbidities at baseline. However, with only one favourable outcome barely achieving statistical significance, early recourse to Tocilizumab does not appear as a game changer. More than 10 ongoing RCT should hopefully help shed more light on its place in the management in COVID-19 shortly even though their sheer number tells a tale of poorly coordinated research effort with significant room for improvement in the coming years.

All in all, the only immunomodulatory option available to clinicians should remain corticosteroids at this point. Few if any interventions are universally helpful in all patients in all circumstances. Understanding in which subgroups of patients the benefit of therapy exceeds the risk is critical to achieving best outcomes. The willingness of oversight committees to cease trials based on a single publication, despite lingering questions, compromises our ability to precisely individualise the risks and benefits. We must however keep in mind than the work is not yet done and more research is required on the interaction with antivirals, anticoagulation and specific subsets of patients to be more selective in reducing potential harm and optimise benefits. Finally, interactions between IL-6R blockade should it be proven effective and steroids will need to be elucidated.

Conflicts of interest

The authors declare no conflicts of interest.