Delving beneath the surface of hyperinflammation in COVID-19

The pattern and mechanism of inflammation in COVID-19 remains unclear, although multiple cytokines are elevated in severe infection.¹ An early report found that some patients with COVID-19 deteriorate at around day 10 after symptom onset, despite a decreasing nasopharyngeal viral load.² Observations like these led to the hypothesis that deterioration in this group of patients is driven by a dysregulated immune response, or cytokine storm, rather than by direct viral injury, and this idea formed the theoretical basis for investigating anti-inflammatory and immunosuppressant therapies in COVID-19.³

In The Lancet Rheumatology, Jessica Manson and colleagues⁴ report the results of an important observational study that defines a hyperinflammatory subphenotype (termed COV-HI) in cases of COVID-19, based on measurement of C-reactive protein (≥150 mg/L, or a doubling in 24 h from >50 mg/L) and ferritin (>1500 μg/L). Despite having a younger median age and fewer comorbidities, patients with this phenotype had higher mortality (36 [40%] of 90 patients) than patients without the phenotype (46 [26%] of 179) during the 28-day follow-up period, and meeting the criteria was associated with an increased next-day risk of death or need for escalated respiratory support (combined endpoint; hazard ratio 2·24 [95% CI 1·62–2·87]).

This high-risk subphenotype was identifiable using readily available parameters, demonstrating its value. In clinical practice, it would be easy to flag patients with the COV-HI phenotype who, for example, would require close monitoring for the need to escalate to a critical care environment. It is worth noting, however, that the definition of hyperinflammation was arbitrarily defined by the authors, potentially introducing confirmation bias, and it would be useful to see these findings reproduced in independent cohorts.

Although clinically useful, this study is perhaps more accurately described as suggesting the predictive value of inflammatory markers rather than identifying a novel pathobiological subphenotype of COVID-19. COV-HI was mostly identified by concentrations of C-reactive protein on admission, with a small proportion of patients meeting the criteria based solely on ferritin measurement, and no patients meeting the criteria on the basis of C-reactive protein doubling alone. A biomarker for which measurement needs to be repeated at 24 h has some limitations, as it precludes high-risk patients from being immediately identified at the time of baseline measurement. However, as the doubling of C-reactive protein alone did not define the COV-HI subphenotype, its inclusion in the criteria is unlikely to have detracted from the importance of the findings.

Ferritin data were unavailable in many cases, limiting its use in the definition of COV-HI. The ferritin criterion was probably included in the definition on the basis of reports of high levels being associated with hyperinflammation syndromes such as haemophagocytic lymphohistiocytosis. Ferritin might have been measured in sicker patients in whom clinicians already had suspicion of hyperinflammation and who were therefore more likely to have a poor outcome. The measurement of ferritin is not routine in patients with acute respiratory distress syndrome (ARDS), and there is an urgent need to assess whether hyperinflammatory subphenotypes as measured by C-reactive protein or ferritin are specific to ARDS due to COVID-19 or are reflective of hyperinflammatory ARDS more broadly.

The biological mechanisms underpinning the putative hyperinflammatory subphenotype of COVID-19 require further investigation. Perhaps a mechanism other than cytokine storm is responsible for deterioration in patients with COV-HI, with C-reactive protein and ferritin being markers of this alternative process.

What insight does the finding of COV-HI provide into the role of cytokine storm in the pathogenesis of COVID-19? Many patients with severe COVID-19 develop ARDS, and findings from related research in ARDS could improve our understanding of the biology of inflammation in COVID-19.⁵ A previous report showed that cytokine profiles in COVID-19 do not differ from those in non-COVID-19 ARDS, nor from those in sepsis.⁶ In patients with ARDS, two distinct phenotypes have been defined: a hyperinflammatory phenotype (characterised by elevated proinflammatory cytokines, increased incidence of shock, and higher mortality)⁸ and a hypoinflammatory phenotype.⁷ These subphenotypes can be defined using a parsimonious model that includes a small number of clinical and biomarker parameters.⁹ Surprisingly, applying the same model, it was determined that the ARDS hyperinflammatory subphenotype was less prevalent in patients with ARDS due to COVID-19 than in patients with non-COVID-19 ARDS.¹⁰ Among patients with COVID-19-associated ARDS, mortality was roughly 20% higher in those with the hyperinflammatory subphenotype (63%) than in those with the hypoinflammatory subphenotype (39%).¹⁰ This difference is consistent with previous ARDS data; however, mortality in both subphenotypes was around 20% higher in patients with COVID-19 ARDS than in historical patients with non-COVID-19 ARDS.⁸ This inconsistency with this data suggests that the host inflammatory response might not be the main driver of poor outcome in COVID-19. Although these results do not refute the existence of a hyperinflammatory subphenotype of COVID-19, they suggest that a cytokine storm is unlikely to be specific to COVID-19 compared with ARDS from other causes.

These important new data provide evidence that there is heterogeneity in the host response to severe acute respiratory syndrome coronavirus 2 infection that is associated with highly disparate clinical outcomes for patients. It is vital that this work prompts further investigations to understand the mechanisms underlying this heterogeneity and to inform appropriate therapeutic interventions. The rapid development of a precision medicine approach in COVID-19 could also inform clinical trials in ARDS and sepsis, and thus affect the care of critically ill patients in years to come.

© 2020 Shutterstock

This online publication has been corrected. The corrected version first appeared at thelancet.com/rheumatology on September 11, 2020