In the present study, the effects of anakinra in critically ill COVID-19 patients with clinical features of MAS/hyperinflammation were compared to critically ill COVID-19 patients that did not display these features. Upon start of treatment, body temperature and plasma levels of ferritin and ASAT were significantly higher in the anakinra group, which supports the indications to initiate this therapy. Anakinra treatment resulted in attenuation of the clinical inflammatory response compared with the control group, including a decrease in body temperature, ferritin, white blood cell counts, and procalcitonin plasma levels. Furthermore, a more pronounced improvement in kidney and liver function was observed in the anakinra group. No significant difference was found for the overall SOFA score, likely due to limited statistical power. These findings on clinical parameters of inflammation support anti-inflammatory effects of anakinra in critically ill COVID-19 patients with features of hyperinflammation. No differences in clinical outcome parameters were observed.
Apart from IL1-RA, no relevant between-group differences in circulating IL-6 or other cytokines over time were observed. It is likely that the significant increase in plasma IL-1RA levels after start of anakinra treatment is due to detection of the drug compound itself, as anakinra is recombinant IL-1RA and was administered intravenously. Although anakinra blocks IL-1β signaling, this cytokine is of little theragnostic use, because of its short half-life which renders it undetectable in the circulation in most cases [14]. Based on our current findings, circulating concentrations of other cytokines appear to be a poor reflection of the immunomodulating effects of anakinra as well, and are therefore not likely to be useful as biomarkers to start treatment and/or monitor the effects of anakinra in critically ill COVID-19 patients.
Our study did not show an obvious clinical benefit of the anti-inflammatory effects of anakinra treatment. However, it is of paramount importance to realize that this was not a randomized controlled trial, and that patients in the anakinra and control group may not have had the same prognosis at the moment of alignment. Patients that were treated with anakinra clearly exhibited signs of hyperinflammation, while this was not the case for patients in the control group. It is well known that hyperinflammation/MAS in sepsis patients is associated with impaired clinical prognosis [15]. In accordance, in the post hoc analysis of the anakinra trial in bacterial sepsis patients, the presence of MAS was associated with a mortality of 65% [8]. Treatment with anakinra in this subgroup of MAS patients was associated with a significantly reduced mortality of 35%, which was similar to the 30% mortality observed in sepsis patients without MAS in the same trial. In line with these results, clinical outcomes of COVID-19 patients treated with anakinra in the present study were comparable with those of the control group, who did not display hyperinflammation and may have had a better a priori prognosis. Nevertheless, our study design does not allow for conclusions of possible clinical benefits of anakinra treatment.
While the use of corticosteroids between the two arms of the study did not reach statistical significance, it was fivefold more prevalent in the group without anakinra before the start of the intervention, and remained twice as high after start of anakinra therapy. Because corticosteroids can affect inflammatory parameters and were recently reported to decrease mortality in COVID-19 [16, 17], we performed a sensitivity analysis in patients who did not receive corticosteroids during the study period. This analysis yielded similar results to our primary analysis, ruling out corticosteroids as an important confounder.
The absence of significant respiratory improvement and/or reduced mortality in the anakinra group contrasts previous findings [9,10,11,12]. This might be due to the fact that we included critically ill mechanically ventilated COVID-19 patients, whereas previous studies were performed in less severe cases with either no ventilatory support or non-invasive ventilation [9,10,11,12]. Possibly, earlier or preventive treatment with anakinra may be more effective and/or disease severity is of relevance for the therapeutic efficacy of anakinra. Also, circulating concentrations of all measured cytokines in our study were markedly lower on alignment day than on ICU admission in both groups. It might be speculated that anakinra is more effective in an earlier, more pro-inflammatory stage of COVID-19 pre-ICU admission, during which higher concentrations of cytokines are present. Furthermore, the previously reported clinical improvements in COVID-19 patients treated with anakinra may be due to notable differences in patient characteristics between treatment groups in earlier studies. For example, anakinra was started in case of mild hyperinflammation (e.g., ferritin ≥ 900 ng/ml or CRP ≥ 100 mg/L, or both), whereas the control group had substantially higher levels of both ferritin and CRP compared to the anakinra group [9] or were compared to historical data of bacterial sepsis patients with MAS in the absence of a COVID-19 control group [10]. These issues might have confounded the clinical outcome results demonstrated in these studies. So, in the absence of a prospective controlled trial randomizing patients that fulfill a hyperinflammatory profile, we feel it is too early to judge any possible clinical effects and to decide what would be the optimal timing to start treatment with anakinra in critically ill COVID-19 patients.
This study has several limitations. First, as inclusion criteria were applied to start treatment with anakinra and these criteria were not present in the control group, bias by indication was clearly present. This difference is likely of importance for the prognosis of the patients, so as a consequence, no direct link can be deducted between the use of anakinra and the clinical results. Nevertheless, we did use a contemporary control group, which was not the case in previous studies [9,10,11,12]. Also, we performed propensity score matching based on patient characteristics which yielded similar results. Additionally, aimed to address this possible bias by indication, an additional sensitivity analysis using a subgroup of control patients who partially met the criteria to start anakinra treatment was performed. Although this matching was not perfect (as the presence of fever was shorter, ferritin was somewhat less elevated and patients had no signs of progressing organ failure), this group was better matched to the anakinra group than the control group used for the main analyses. This additional sensitivity analysis also showed a more pronounced decrease in clinical inflammatory markers in the anakinra group compared to the control group. Of relevance, and confirming the main analyses, no trends toward differences in clinical outcome parameters were found. Second, limited statistical power because of the relatively small number of patients preclude conclusions related to the presence or absence of efficacy of anakinra on clinical outcome parameters. Nevertheless, the inflammation-related clinical endpoints show a clear signal in accordance with the mechanism of action of anakinra and kidney and liver function appear to improve with anakinra therapy.