Customize your JAMA Network experience by selecting one or more topics from the list below.
In its most severe form, severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), the virus that causes coronavirus disease 2019 (COVID-19), leads to a life-threatening pneumonia and acute respiratory distress syndrome (ARDS). The mortality rate from COVID-19 ARDS can approach 40% to 50%.1,2 Although the mechanisms of COVID-19–induced lung injury are still being elucidated, the term cytokine storm has become synonymous with its pathophysiology, both in scientific publications and the media. Absent convincing data of their effectiveness in COVID-19, drugs such as tocilizumab and sarilumab, which are monoclonal antibodies targeting interleukin (IL)-6 activity, are being used to treat patients; trials of these agents typically cite the cytokine storm as their rationale (NCT04306705, NCT04322773). A critical evaluation of the term cytokine storm and its relevance to COVID-19 is warranted.
Cytokine storm has no definition. Broadly speaking, it denotes a hyperactive immune response characterized by the release of interferons, interleukins, tumor-necrosis factors, chemokines, and several other mediators. These mediators are part of a well-conserved innate immune response necessary for efficient clearance of infectious agents. Cytokine storm implies that the levels of released cytokines are injurious to host cells. Distinguishing an appropriate from a dysregulated inflammatory response in the pathophysiology of critical illness, however, has been a major challenge. To add further complexity, most mediators implicated in cytokine storm demonstrate pleotropic downstream effects and are frequently interdependent in their biological activity. The interactions of these mediators and the pathways they inform are neither linear nor uniform. Further, although their quantified levels may suggest severity of responses, they do not necessarily imply pathogenesis. This complex interplay illustrates the limitations of interfering in the acute inflammatory response based on single mediators and at indiscriminate time points.
Why has the “cytokine storm” been so closely associated with COVID-19? During the SARS epidemic caused by SARS-CoV-1, the term cytokine storm was described as a feature and associated with adverse outcomes.3 Several early case series in COVID-19 reported levels of some plasma cytokines elevated above the normal range. In most cases, however, they are lower than plasma levels in previous cohorts of patients with ARDS. Interleukin-6, a proinflammatory cytokine, is a key mediator in the acute inflammatory response and the purported cytokine storm. The Table summarizes reported IL-6 levels in 5 cohorts of patients with COVID-19,1,2,4-6 each with more than 100 patients, and 3 cohorts of patients with ARDS.7-9 Although the median values are above the normal range in many (but not all) cases, they are lower than the median values typically reported in ARDS. The median values in randomized clinical trials conducted by the National Heart, Lung and Blood Institute’s ARDS Network are approximately 10- to 40-fold higher, even when only patients with severe COVID-19 are considered.7-9 The hyperinflammatory phenotype of ARDS is characterized by elevated proinflammatory cytokines, an increased incidence of shock, and adverse clinical outcomes.7-9 The characteristics of this phenotype could be considered as most consistent with those expected with the cytokine storm. However, median IL-6 levels in patients with the hyperinflammatory phenotype of ARDS are 10- to 200-fold higher than levels in patients with severe COVID-19 (Table).
Putting the unsubstantiated theory of the cytokine storm aside, the more intriguing question to ask is why are clinical outcomes in COVID-19 so unfavorable despite relatively low levels of circulating IL-6? One hypothesis is that severe viral pneumonia from COVID-19 produces primarily severe lung injury, without the same magnitude of systemic responses in most patients with COVID-19 as reported in prior studies of the hyperinflammatory phenotype in ARDS.7-9 For example, a recent postmortem report of patients with COVID-19 ARDS identified severe vascular injury, including alveolar microthrombi that were 9 times more prevalent than found in postmortem studies of patients with influenza ARDS.10 Ongoing research may identify more specific mechanisms of COVID-19–mediated lung injury.
There are some limitations to these observations. Almost all the COVID-19 IL-6 data are from clinical laboratory tests. In most studies, details of the exact methods used are not available; calibration issues could lead to underestimating IL-6 levels compared with measurements based on enzyme-linked immunosorbent assay used in prior ARDS studies.7-9 Furthermore, plasma levels of cytokines may not be representative of lung inflammation. Given the number of COVID-19 cases worldwide, the data on IL-6 levels are from a very small fraction of patients. Nevertheless, the theory of the cytokine storm is based on these data, and the case for its presence in COVID-19 seems weak. A more appropriate conclusion would be that in comparison to other causes of ARDS, COVID-19 is characterized by lower levels of circulating cytokine responses. Perhaps the most valid conclusion, however, is that the current data are insufficient to ascertain the precise role and scope of dysregulated cytokine responses in COVID-19.
Widespread acceptance of the term cytokine storm in COVID-19 has motivated the use of potent immunomodulatory therapies both in the setting of clinical trials and on a compassionate basis. These drugs, such as IL-6 inhibitors and high-dose corticosteroids, block pathways critical to host immune responses. Many monoclonal antibody drugs are being repurposed from treating patients with chronic inflammatory conditions where optimal pharmacokinetics demand prolonged half-lives. Long-lasting and indiscriminate suppression of inflammation in the acute critical care setting raises concerns about impaired clearance of SARS-CoV-2 and increased risk for secondary infections. Enthusiasm for the use of immunomodulatory approaches in COVID-19 seems to derive in large part from clinical experience with cytokine release syndrome (CRS), a term frequently interchanged with cytokine storm. In the 2016 study of CRS by Maude and colleagues, patients who developed CRS following treatment with chimeric antigen receptor T cells were effectively treated with tocilizumab.11 Notably, the peak plasma IL-6 level in patients who developed CRS was approximately 10 000 pg/mL—almost 1000-fold higher than that reported in severe COVID-19. Conceivably, these therapies could be effective in COVID-19, but the likelihood for success would be enhanced by selecting the right patients with predictive enrichment and the right timing for intervention.7
Given reports that dexamethasone may improve survival for patients with COVID-19 and ARDS, it should be determined whether these effects differ between ARDS phenotypes and if they occur despite the absence of a circulating hyperinflammatory cytokine response. If so, the additional information about dexamethasone would further substantiate the importance of studying local inflammatory responses to COVID-19 in the lungs.
For these reasons, the term cytokine storm may be misleading in COVID-19 ARDS. Incorporating a poorly defined pathophysiological entity lacking a firm biological diagnosis may only further increase uncertainty about how best to manage this heterogeneous population of patients. The manifestations of elevated circulating mediators in the purported cytokine storm are likely to be endothelial dysfunction and systemic inflammation leading to fever, tachycardia, tachypnea, and hypotension. This constellation of symptoms already has a long history in critical care, known as systemic inflammatory response syndrome, and was used to define sepsis for decades. Interventions targeting single cytokines in sepsis, unfortunately, also have a long history of failure. Although the term cytokine storm conjures up dramatic imagery and has captured the attention of the mainstream and scientific media, the current data do not support its use. Until new data establish otherwise, the linkage of cytokine storm to COVID-19 may be nothing more than a tempest in a teapot.
Corresponding Author: Pratik Sinha, MB, ChB, PhD, University of California, San Francisco, 505 Parnassus Ave, Box 0111, San Francisco, CA 94143-0111 (email@example.com).
Published Online: June 30, 2020. doi:10.1001/jamainternmed.2020.3313
Conflict of Interest Disclosures: Dr Matthay reports grants from NIH/NHLBI, the Department of Defense, the California Institute of Regenerative Medicine, Bayer Pharmaceuticals, and Roche/Genentec. Dr Calfee reports grants from NIH during the conduct of the study and grants from Roche/Genentech and Bayer and personal fees from Quark Pharmaceuticals, Gen1e Life Sciences, and Vasomune outside the submitted work. No other disclosures were reported.
Funding/Support: The authors were supported by NIH grants GM008440 (Dr Sinha), HL140026 and HL123004 (Dr Mathay), and HL140026 (Dr Calfee).
Role of the Funder/Sponsor: The funders had no role in the creation of this article; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Sinha P, Matthay MA, Calfee CS. Is a “Cytokine Storm” Relevant to COVID-19? JAMA Intern Med. 2020;180(9):1152–1154. doi:10.1001/jamainternmed.2020.3313
Artificial Intelligence Resource Center