We conducted a proof-of-concept phase 2, prospective, randomized, open-label study across 9 hospitals in Belgium. The trial was approved by the Ethical Committee of Ghent University Hospital and conducted in accordance with Good Clinical Practice guidelines and the Declaration of Helsinki. Study design, coordination, monitoring and data management was performed under the responsibility of the Health Innovation and Research Institute UZ Gent (HIRUZ). UCB provided study medication and assistance with data analysis and funded the study. Long-term follow-up of the patients was funded with the ClinicalTrials.COV grant (BOFCOV2020000801) from University Hospital Ghent. An independent data safety monitoring board monitored participant safety. Every patient or their legal representative provided informed consent before participation. All authors take responsibility for the integrity of the trial and the publication.

Eligible patients were over the age of 18, had a laboratory diagnosis of COVID-19 with symptoms appearing between 6 and 16 days at inclusion, a ratio of the partial pressure of oxygen (PaO2) to the fraction of inspired oxygen (FiO2; PaO2/FiO2) of less than 350 mmHg and bilateral pulmonary infiltrates on chest computed tomography (CT) within the last two days prior to randomization. Presence of systemic inflammation was defined [38] by a single ferritin over 2000 µg/L at inclusion in patients immediately requiring intensive respiratory support. In those without immediate respiratory failure, a ferritin over 1000 µg/L and rising over 24 h needed to be documented, or alternatively lymphopenia below 800/mL with two of the following criteria, any of those rising over 24 h: (1) a rising ferritin above 700 µg/L, (2) a rising lactate dehydrogenase above 300 IU/L, (3) a rising C-reactive protein above 70 mg/L or (4) rising D-dimers above 1000 ng/mL.

Exclusion criteria included mechanical ventilation for more than 24 h at randomization, a clinical frailty score > 3 prior to SARS-CoV-2 infection [39], an unlikelihood to survive beyond 48 h as judged by the treating physician, an active co-infection defined on clinical grounds, thrombocytopenia below 50,000/µL or neutropenia below 1500/µL, active treatment with complement-inhibiting drugs; weight below 54 kg or above 150 kg; high-dose systemic steroid or immunosuppressive drug use for a COVID-19-unrelated disorder. The full list of in- and exclusion criteria can be found in the study protocol (Additional file 1).

Participants were allocated in a 2:1 ratio to the zilucoplan or control arm using simple randomization, stratified by center. Randomization and subsequent data collection were done in an interactive Web Response System (REDCap) [40].

Patients allocated to the zilucoplan group received daily 32.4 mg zilucoplan subcutaneously for 14 days or until discharge (whichever came first) and 2 g ceftriaxone i.v. for maximum 28 days (i.e. during zilucoplan treatment and an additional 14 days after cessation of zilucoplan) as prophylaxis for meningococcal infections, which occur more frequently in C5 deficient states [41]. On clinical grounds, ceftriaxone could be switched to any other antibiotic covering Neisseria meningitidis. Following hospital discharge, antibiotics were switched to 500 mg ciprofloxacin once daily until 14 days after the last zilucoplan administration. To control for the effect of antibiotics, patients in the control group also received daily i.v. 2 g ceftriaxone for 7 days or until discharge (whichever occurred first).

The primary efficacy endpoint was the change in oxygenation (PaO2/FiO2, P(A-a) O2 gradient and a/A PO2) from baseline to day 6 and to day 15 or discharge (whichever occurred first). The methods to derive the parameters for oxygenation are described in the Statistical Analysis Plan (Additional file 2).

Secondary objectives were to study the effects of zilucoplan on clinical outcomes (defined by duration of hospital stay, 6-point ordinal scale (1: death; 2: invasive mechanical ventilation; 3: non-invasive ventilation or high flow oxygen devices; 4: hospitalized and requiring supplemental oxygen; 5: hospitalized and not requiring supplemental oxygen; 6: not hospitalized), time to defervescence, supplemental oxygen use and severity of organ failure assessment (SOFA) score), on progression to mechanical ventilation, ARDS, on duration of ICU stay, on all-cause mortality rates, on the rate of nosocomial infections, and on ferritin and C-reactive protein (CRP) serum levels. A follow-up visit was scheduled between 12- and 22-weeks post-randomization to study long-term clinical evolution. A detailed list of the secondary endpoints is provided in the study protocol (Additional file 1).

Certain predefined endpoints were not analyzed due to insufficient and/or missing data, i.e. duration of mechanical ventilation in ventilated patients, time since randomization until first use of high-flow oxygen devices or mechanical ventilation in non-ventilated patients and duration of ICU stay (Additional file 3).

Key safety endpoints included incidence of adverse events (AE) and serious adverse events (SAE) from first day of study treatment until day 28.

Serum cytokines (IL-6, IL-8, IL-18, IL-1RA, CXCL-10) were quantified using Mesoscale Discovery. Soluble membrane attack complex (sC5b-9) was quantified in cell-free plasma using the MicroVue Complement sC5b-9 Plus ELISA kit (Quidel).

The target difference was the change from baseline measured in PaO2/FiO2 (at day 6 and day 15) between the control and the treated group. Given a a priori calculated sample size of 81 participants, 54 on the zilucoplan arm and 27 on the control arm, there was > 85% power to show a significant difference from standard of care (SoC) at the 2 sided 5% level if the underlying treatment difference was an 80-mmHg difference (25% of 320 mmHg, being the mean at hospital admission prior to this study) in the PaO2/FiO2. This assumed a standard deviation of 105 mmHg and a dropout rate of less than 10%.

All efficacy analyses were carried out on the full analyses data set consisting of all participants who received at least 1 dose of zilucoplan, when randomized to the zilucoplan group, and who had at least one dose of intravenous prophylactic antibiotics, when randomized to the control group. The primary endpoints were analyzed separately using a Mixed Model Repeated Measures (MMRM) analysis with fixed effects for baseline treatment, nominal day (using day 6 and day 15), baseline*nominal day interaction and treatment*nominal day interaction. Participant was fitted as a random effect and an unstructured covariance was used. A last observation carried forward (LOCF) approach was utilized for the participants who were discharged early. If a patient died or withdrew, no data was imputed. The mean change from baseline to day 6 and day 15 for both treatment arms and differences between treatments in mean change from baseline to day 6 and day 15, their 95% CI and p-values were estimated directly from the model. PaO2/FiO2, and a/A PO2 were analyzed on the natural scale. P(A-a)O2 gradient required a log transformation therefore the differences from baseline and between treatment are expressed as ratios.

Logistic regression models, including treatment as a factor, were fitted for mortality. The estimated odds ratio, 95% CI and p-value were computed.

The above analyses were augmented with a series of Bayesian analyses which provide a useful, additional interpretation of the results (Additional file 4).

Safety data were analyzed descriptively in all patients who received at least one dose of zilucoplan in the treated group and all patients who had at least one dose of prophylactic antibiotics in the control group (safety population).

Baseline was defined as the last measure prior to dosing, with the exception of clinical laboratory parameters and cytokines. Blood samples were not always prioritized prior to starting dosing hence values up to 2 h post-first dose were included for clinical laboratory parameters and 30 min post-dose for cytokines.

No adjustments were made for multiplicity.

Statistical analysis was performed using SAS version 9·4 and R. The full statistical analysis plan is available as an online supplement (Additional file 2).