Effect of anakinra in COVID-19

We read with interest the Article by Giulio Cavalli and colleagues¹ that described the effects of the interleukin-1 receptor antagonist anakinra in patients who were critically ill with COVID-19. The authors reported promising effects of high-dose anakinra treatment, including improvement of respiratory function and better survival compared with a historical control group that received standard treatment. Although these results are encouraging, there are notable differences in patient characteristics between the groups that might have confounded the observed effects.

First, the authors indicate that they studied a group of patients with COVID-19 and hyperinflammation. This makes sense, as anakinra was shown to be effective in patients with sepsis and features of macrophage activation syndrome (MAS) in a post-hoc analysis of a randomised clinical trial done more than two decades ago.² MAS is characterised by an excessive inflammatory response, most specifically elevated ferritin concentrations.³ According to a validated diagnostic score of MAS, a ferritin level higher than 2000 ng/mL increases the probability of MAS.⁴ Notably, in the study by Cavalli and colleagues,¹ median ferritin levels were only 1237 ng/mL in the anakinra group compared with 2218 ng/mL in the control group, indicating a much less pronounced inflammatory state (and thus better prognosis) in the anakinra group.

Second, it is established that advanced age is an important risk factor for severe disease course in patients with COVID-19, with a hazard ratio for mortality of 1·7 (95% CI 1·1–2·7) for patients aged 65 years or older.⁵ In the study by Cavalli and colleagues,¹ the median age of patients treated with anakinra was 62 years, 8 years lower than that of the control group, representing another substantial disadvantage for the control group. The less pronounced inflammatory state together with this age discrepancy represent relevant confounders that might explain the improvement in respiratory function and clinical outcome of this group, irrespective of anakinra treatment.

A commonly used method to prevent imbalance in patient characteristics (and thereby bias) is propensity score matching, which is used to make groups more comparable in non-randomised studies. Therefore, it would be interesting to learn if the effects of anakinra observed in the study by Cavalli and colleagues hold up to scrutiny in a propensity score matched design.

In conclusion, there is clear rationale for using anakinra in the treatment of patients who are critically ill with COVID-19 and hyperinflammation. However, data from additional cohorts using propensity score matching or, ideally, a randomised controlled trial are needed to confirm its putative benefits.