Early Treatment of COVID-19 Pneumonia with Anakinra Guided by Urokinase Plasminogen Receptor

Early recognition of risk and start of treatment may improve unfavorable outcome of COVID-19. In the SAVE-MORE double-blind randomized trial, 594 patients with pneumonia without respiratory dysfunction at risk as dened by plasma suPAR (soluble urokinase plasminogen activator receptor) ≥ 6 ng/ml were 1:2 randomized to subcutaneous placebo or 100 mg anakinra once daily for 10 days; 85.9% were co-administered dexamethasone. After 28 days, anakinra-treated patients were distributed to lower strata of the 11-point World Health Organization ordinal Clinical Progression Scale (WHO-CPS) (adjusted odds ratio-OR 0.36; 95%CI 0.26–0.50; P < 0.001); anakinra protected from severe disease or death ( ≥ 6 points of WHO-CPS) (OR: 0.46; P: 0.010). The median WHO-CPS decrease in the placebo and anakinra groups was 3 and 4 points (OR 0.40; P < 0.0001); the median decrease of SOFA score was 0 and 1 points (OR 0.63; P: 0.004). 28-day mortality decreased (hazard ratio: 0.45; P: 0.045) and hospital stay was shorter.


Introduction
COronaVIrus Disease-19 (COVID-19) pneumonia can have an unpredictable clinical course. Patients may suddenly deteriorate into severe respiratory failure (SRF), so that early recognition of the turning point and timely onset of targeted treatment appear of outmost importance.
Calprotectin in turn stimulates the aberrant production of interleukin (IL)-1β by the circulating monocytes 4 whereas knock-outing of IL-1α is protective for the host 3 . These observations frame the hypothesis that early detection of increased suPAR may guide targeted therapeutics against IL-1α and IL-1β. Indeed, in the open-label phase II study SAVE, early administration of anakinra guided by suPAR decreased the relative risk for progression into SRF by 70% compared to standard-of-care (SoC) treatment; signi cant reduction of 28-day mortality was also found. The recombinant IL-1 receptor antagonist anakinra blocks both IL-1α and IL-1β 5 .
Despite the important information provided by the SAVE trial, one prospective randomized clinical trial (RCT) is necessary to prove the effectiveness of this approach. SAVE-MORE (suPAR-guided Anakinra treatment for Validation of the risk and Early Management Of seveRE respiratory failure by COVID-19) is a pivotal, con rmatory, phase III RCT aiming to evaluate the e cacy and safety of early start of anakinra guided by suPAR in patients with COVID-19 pneumonia. The primary objective was to evaluate the e cacy and safety of early targeting of IL-1α/β on the clinical state of patients with COVID-19 pneumonia and elevated suPAR levels, over 28 days, as measured by the 11-point ordinal WHO (World Health Organization) Clinical Progression scale (CPS).

Patients
The protocol (available with the full text of this article) of the SAVE-MORE double-blind RCT was nalized after advice by the Emergency Task Force (ETF) of the European Medicines Agency (EMA) for COVID-19 (document EMA/659928/2020). From December 2020 through March 2021, 1060 patients were screened and 606 were randomized. 12 patients withdrew consent and requested removal of all data, leaving a nal intention-to-treat (ITT) analysis cohort of 594 patients; 189 patients were allocated to the SoC and placebo arm, and 405 patients were allocated to the SoC and anakinra arm. Only one patient was lost to follow-up ( Figure 1). Baseline characteristics and co-administered treatments were similar between the two arms ( Table 1). The main reason for exclusion from the study was suPAR less than 6 ng/ml. The follow-up of these patients with suPAR less than 6 ng/ml showed that only 2.9% progressed into SRF until day 14.

Primary and secondary outcomes
The unadjusted odds ratio (OR) of the WHO-CPS by day 28 was 0.36 (95% con dence intervals-CI 0.26-0.49; P<0.001) ( Figure 2A and Table 2) corresponding to 2.78-fold higher odds of having a better outcome on the 11-point scale at day 28. The testing of the assumptions of the ordinal regression analysis i.e. the Goodness-of-t test and the parallel lines test were not statistically signi cant denoting an even distribution of the treatment effect size for all 11-points of the WHO-CPS.
According to the advice of the COVID-ETF of the EMA the e cacy of anakinra should be validates by multivariate ordinal regression analysis using as co-variates the variables used for strati ed randomization, i.e., disease severity, intake of dexamethasone, BMI more than 30, and country. At the univariate analysis treatment with anakinra and dexamethasone and disease severity were signi cantly associated with the nal outcome. However, in the multivariate analysis, treatment with anakinra was the only variable that was signi cantly associated with nal outcome (OR 0.36; 95%CI 0.26-0.50; P<0.001) ( Figure 2B).
According to the advice from COVID-ETF, the analysis of the primary endpoint should have been supported by three con rmatory analyses: comparison of the WHO-CPS by day 14; logistic regression analysis separately for patients at the two spectra of WHO-CPS at day 28; and time progression to respiratory failure by day 14. The rst spectrum of the WHO-CPS was de ned as patients fully recovered with negative viral load (WHO-CPS 0 points) contrary to patients with persistent disease (WHO-CPS between points 1 to 10). The second spectrum was de ned as patients pointed 6 or more in the WHO-CPS (severe hospitalized and dead) contrary to patients pointed 5 or less. All three con rmatory analyses fully supported the clinical bene t of anakinra treatment. More precisely, the unadjusted OR of the ordinal regression analysis of the WHO-CPS by day 14 was 0.57 (95% CI 0.42-0.77; P<0.001) ( Figure S1); after multivariate adjustment this was 0.58 (0.42-0.79; P: 0.001) ( Table S1) showing also that anakinra treatment was the only variable affecting the outcome of patients by day 14.
Regarding the second con rmatory analysis the multivariate logistic regression model for the rst spectrum of the WHO-CPS showed that anakinra treatment and baseline severity were associated with persistence of the disease by day 28; anakinra was protective from disease persistence (OR: 0.36; 95% CI 0.25-0.53; P<0.001) ( Table S2). The multivariate logistic regression model for the second spectrum of the WHO-CPS showed that anakinra treatment was protective from critical disease or death (OR: 0.46; 95% CI 0.26-0.83; P: 0.010) (Table S2). 28-day mortality was also lower among patients allocated to SoC and anakinra treatment (6.9% versus 3.2% respectively) ( Figure S2). The third con rmatory analysis validated the results of the phase 2 SAVE trial. In this analysis, anakinra treatment prevented the progression to respiratory failure by day 14 ( Figure S3 and Table S3) (31.7% in the SoC and placebo arm versus 20.7% in the SoC and anakinra arm).
The rate of protocol deviations from the SoC treatment was signi cantly greater among patients allocated to the placebo arm than patients allocated to the anakinra arm (14.3% versus 3.2% respectively; P<0.001). These protocol deviations in the SoC and placebo arm were mainly related to increasing the dose and/or duration of dexamethasone administration (Table S4). All ve sensitivity analyses con rmed further the analysis of the primary endpoint (Table S5).
Analysis of the ve clinical secondary endpoints showed a signi cant bene t from anakinra treatment on all these endpoints. More precisely, the decreases of the WHO-CPS score from baseline by days 28 and 14 and of the SOFA (sequential organ failure assessment) score from baseline by day 7 were signi cantly greater in the SoC and anakinra arm ( Table 2 and Tables S6 to S8). Moreover, in the anakinra group, the average time until hospital discharge was one 1 day shorter and the time until ICU discharge was 4 days shorter ( Table 2 and Figures S4 and S5).
Over-time follow-up of laboratory values showed that among patients treated with anakinra: a) the absolute lymphocyte count was increased by day 7; b) circulating IL-6 was decreased by days 4 and 7; and c) plasma C-reactive protein (CRP) was decreased by day 7 (Figure 3).

Adverse events
Overall, the incidence of serious treatment-emergent adverse events (TEAEs) through day 28 was lower in patients in the anakinra and SoC group (16.3%) compared to the placebo and SoC group (21.2%). The non-serious TEAEs were similar in both treatment groups (Table 3 and Tables S9 and S10).

Discussion
The SAVE-MORE trial is introducing an entirely novel approach for the management of COVID-19 pneumonia which relies on early intervention: use suPAR to predict who among patients with pneumonia are at greatest likelihood for unfavorable outcome and provide treatment early. Results showed considerable e cacy of 10 days subcutaneous administration of anakinra in patients with COVID-19 and plasma suPAR 6 ng/ml or more. Anakinra treatment led to 2.78-fold higher odds of having a better outcome on the 11-point scale at day 28. The bene t was already apparent from day 14 and this is of major clinical importance since the rst 14 days is the period during which a patient is expected to worsen; anakinra bene t was expanded until day 28. The magnitude of the e cacy of anakinra was shown in all multivariate analyses where in the presence of anakinra treatment the effect of disease severity on the nal outcome was lost. The proportion of patients fully recovered exceeded 50% and those who remained under severe disease were reduced by 54%; the signi cant relative decrease of 28day mortality was 55%. The large majority of the study population had severe COVID-19 and 85.9% were receiving SoC treatment containing dexamethasone. The remarkable improvement of patients under anakinra is also indirectly evidenced by the lack of changes in the SoC regimen. On the contrary, in 17% of patients receiving placebo, treating physicians changed the dexamethasone regimen and they administered either higher doses or even anti-cytokine biologicals.
The results fully validate the ndings of the previous SAVE open-label phase II trial. In SAVE, the incidence of respiratory failure after 14 days with anakinra treatment was 22.3% 5 ; in the SAVE-MORE trial it was 20.7%. For those who were eventually admitted to the ICU, time until discharge was signi cantly shorter in the anakinra and SoC treated group than in the placebo and SoC group; this was also observed in the previous SAVE trial 5 .
Early since the beginning of the COVID-19 pandemic, immunomodulators were suggested as one main strategy to attenuate the exaggerated immune response of the host. The most common administered drugs are anakinra and tocilizumab targeting the IL-1 and the IL-6 pathways respectively. However, the results of RCTs were heterogeneous and provided varying clinical bene t. There are nine published studies on the clinical e cacy of anakinra 5,7-14 ; four have retrospective design, another four have prospective design using parallel comparators and only one is an RCT. Although most of these studies report mortality bene t, it is di cult to compare the ndings to the results of the SAVE-MORE trial. The studies differ with regard to selection of patients, severity of illness and stage of the disease. Also duration of treatment, dose and route of administration was variable. So far, WHO-CPS was not studied as primary endpoint. Indeed, four of the studies were done in patients with critical illness with plasma levels of CRP and ferritin strongly exceeding the levels of the SAVE-MORE study population [7][8][9][10] .
What makes the difference between SAVE-MORE and the rest of RCTs and can explain the overwhelming e cacy of treatment? The reason is very likely patient strati cation using suPAR as a biomarker of in ammation and diseases severity to select the patients most likely to bene t from anakinra treatment.
Based on experimental studies and clinical evidence, early increase of suPAR is pointing towards excess release of DAMPs 3 . Predominant DAMPs are IL-1α that is released from the lung epithelium and calprotectin, which is subsequently exerting systemic effects through the production of IL-1β, designating suPAR as an important biomarker for excessive IL-1 bioactivity. Anakinra blocks both IL-1α and IL-1β by blocking their common receptor. The attenuation of the in ammatory responses by anakinra was shown by the decrease of IL-6 and of CRP circulating concentrations and by the increase of the absolute lymphocyte counts.
The clinical bene t of tocilizumab has been studied in six RCTs. In four of these RCTs, the patient populations were much similar to the population of the SAVE-MORE trial [15][16][17][18] . Clinical bene t from tocilizumab treatment was shown in only one of these four trials. On the opposite, most of clinical bene t from tocilizumab treatment was found in the other two trials, namely RECOVERY 19 and REMAP-CAP 20 , with participants suffering from critical illness. Mortality was decreased from 35% with usual care to 31% in the RECOVERY trial 19 whilst the median number of organ support-free days were increased from 0 days with usual care to 10 days with tocilizumab treatment in the REMAP-CAP trial 20 . The bene t of the more severe patients by tocilizumab may be explained by the biology of the critical illness. We have previously shown that circulating monocytes in critical COVID-19 present with complex immune dysregulation characterized by decreased e ciency for antigen-presentation and inappropriate maintenance of the potential for excess cytokine production: this dysregulation was restored upon exposure to tocilizumab 21 .
In conclusion, the SAVE-MORE trial showed that early start of treatment with anakinra and SoC guided by the biomarker suPAR in patients hospitalized with moderate and severe COVID-19 is leading to 2.78 times better improvement of the overall clinical status as expressed by the WHO-CPS. The frequency for full recovery is increased and the incidence of respiratory failure or death is decreased. This leads to shorter hospital stay. This nding is of outmost clinical importance and carries a major public health dimension, given the ICU overload during the COVID-19 pandemic, especially in countries with limited ICU capacity.

Patients
Enrolled patients were adults of either gender; with molecular diagnosis of infection by SARS-CoV-2; with involvement of the lower respiratory tract as con rmed by chest computed tomography or X-ray; in need for hospitalization; and with plasma suPAR 6 ng/ml or more. Main exclusion criteria were: ratio or partial oxygen pressure to fraction of inspired oxygen less than 150; need of non-invasive ventilation (CPAP or BPAP) or mechanical ventilation; neutropenia; stage IV malignancy; end-stage renal disease; severe hepatic failure; immunode ciencies; chronic intake of corticosteroids; any intake of biological anticytokine drugs including JAK inhibitors the last month. All patients or their legal representatives provided written informed consent before enrollment.

Trial interventions
Patients meeting all inclusion criteria and not meeting any exclusion criterion were subject to blood draw. suPAR was measured in plasma using the suPARnostic ® Quick Triage kit (Virogates S/A, Birkerød, Denmark) and a point-of care reader. Patients with suPAR 6 ng/ml or more were electronically 1:2 randomized into treatment with placebo or anakinra using four randomization strata: classi cation into moderate or severe disease using the WHO de nition 6 ; need for dexamethasone intake; body mass index (BMI) more than 30 kg/m 2 ; and country. The study drug was administered subcutaneously once daily in the thigh or in the abdomen for seven to 10 days. Patients allocated to placebo treatment were daily injected 0.67 ml of 0.9% sodium chloride; and those allocated to active drug 100 mg of anakinra at a nal volume of 0.67 ml. Study drug was prepared by an unblinded pharmacist with access to the electronic study system using a separate username and a password. Administration was done by a blind study nurse. All patients were receiving pre-de ned SoC which consisted of regular monitoring of physical signs, oximetry and anti-coagulation. Patients with severe disease by the WHO de nition 6 were also receiving intravenous 6 mg daily dexamethasone for 10 days. Remdesivir treatment was left at the discretion of the attending physicians; other biologicals targeting cytokines and kinase inhibitors were not allowed.
Study visits were done daily for 10 days; on day 14; and on day 28. At each study-visit the following were recorded: non-serious and serious TEAEs; WHO-CPS; sequential organ failure assessment (SOFA) score; and co-administered treatment. Visits were done by phone for patients discharged by day 7. Data were captured after review of all medical and nursing charts by a physicians' team blinded to the allocation group. Blood samples and nasopharyngeal swabs were collected before start of the study drug and at days 4 and 7 for the measurements of biomarkers.
All serious and non-serious TEAEs were graded according to the Common Terminology Criteria for Adverse Events (version 5.0).

Outcomes
The primary study endpoint was the overall comparison of the distribution of frequencies of the scores from the 11-point WHO Clinical Progression ordinal Scale (CPS) between the two arms of treatment at Day 28. Secondary endpoints included the changes of WHO-CPS by days 14 and 28 from the baseline (before start of the study drug); the change of SOFA score by day 7 from baseline; the time until hospital discharge; the time of stay in the intensive care unit (ICU) for patients eventually admitted to the ICU; and the comparison of biomarkers.

Statistical analysis
The sample size was calculated based on the nding from the phase II SAVE trial 5 that 42% of comparators and 16.3% of anakinra-treated patients by day 28 were presented with 6 or more points of the WHO-CPS. To achieve such a difference in the WHO-CPS scores with 90% power at the 5% level of signi cance, allocation of 200 patients to SoC and placebo treatment and 400 patients to SoC and anakinra treatment were planned. Data were analyzed for the intention-to-treat (ITT) population. Missing data were imputed by last observation carried forward (LOCF). WHO-CPS is an ordinal 11-point variable ranging from 0 to 10 and comparisons were done by univariate and multivariate ordinal regression analysis using logit function. Results were expressed as the odds ratio (OR) and 95% con dence intervals (CI). The two basic assumptions of the model, i.e., proportional odds and the goodness-of-t test were checked. According to EMA's COVID-ETF advice the variables used for strati ed randomization entered as co-variates in the multivariate model, i.e., disease severity, intake of dexamethasone, BMI more than 30, and country. According to the same advice, the analysis of the primary endpoint should have been supported by three analyses: comparison of the WHO-CPS by day 14; logistic regression analysis separately for patients at the two spectra of WHO-CPS at day 28; and time progression to respiratory failure by day 14. The rst spectrum of the WHO-CPS was de ned as patients fully recovered with negative viral load (WHO-CPS 0 points) contrary to patients with persistent disease (WHO-CPS between points 1 to 10). The second spectrum was de ned as patients pointed 6 or more in the WHO-CPS (severe hospitalized and dead) contrary to patients pointed 5 or less. Five sensitivity analyses were conducted to assess robustness: exclusion of population deviating from the SoC; population receiving at least 7 doses of the study drug; complete analysis set; responder analysis treating missing values as non-responders; and comparison of the unadjusted and the adjusted treatment effects. Analysis was conducted using IBM SPSS Statistics v. 26.0. All P values were two-sided and any P value <0.05 was considered as statistically signi cant. The complete statistical analysis plan is provided in the Supplementary Appendix.

Declarations Data Availability
Requests for deidenti ed patient data by researchers with proposed use of the data can be made to corresponding author with speci c data needs, analysis plans and dissemination plans. Those requests will be reviewed by a study steering committee and the study sponsor for release upon publication. Contact: egiamarel@med.uoa.gr   Figure 1 Study ow chart. Abbreviations: FiO2: fraction of inspired oxygen; ITT: intention-to-treat; IV: intravenous; pO2: partial oxygen pressure; suPAR: soluble urokinase plasminogen activator receptor Levels of lymphocytes, interleukin (IL)-6 and C-reactive protein (CRP) over days of follow-up. Day 1 sampling was done before start of administration of the study drug. The P-values of comparisons for each day of follow-up are provided. Abbreviations: CI: con dence interval; SoC: standard-of-care