Trial design and oversight
CARDEA was a phase 2, randomized, double-blind, placebo-controlled trial that tested the addition of Auxora to corticosteroids and standard of care in patients with severe COVID-19 pneumonia (ClinicalTrials.gov identifier, NCT04345614). The study of Auxora in patients with severe COVID-19 pneumonia was initially conducted as an open-label study that started enrollment on April 8, 2020. The FDA provided guidance on May 12, 2020, to limit further enrollment under the open-label design and transition to a randomized, double-blind, placebo-controlled trial. As such, the open-label study was terminated on May 13, 2020 and the results were published [22].
CARDEA was initially designed to enroll 400 patients with two specified groups that were to be stratified equally across the treatment arms: 80 patients with a baseline imputed PaO2/FiO2 ratio of > 200 and 320 patients with a baseline imputed PaO2/FiO2 ratio of ≤ 200. The PaO2/FiO2 was imputed from a SpO2/FiO2 using a non-linear equation. The SpO2 was obtained using pulse oximetry. The FiO2 was read from the controlled oxygen source in patients requiring high flow nasal cannula. For patients on an uncontrolled oxygen source, a conversion table was provided to all sites to estimate the FiO2 based on the method of oxygen delivery and oxygen flow rate [23]. The baseline imputed PaO2/FiO2 was the worst value in the 24 h prior to screening.
It had been noted in the open-label study that patients with a baseline imputed PaO2/FiO2 > 200 had neither required invasive mechanical ventilation nor died so their enrollment in CARDEA was to confirm this observation [22]. After the first 23 patients with a baseline imputed PaO2/FiO2 > 200 were randomized, a blinded analysis confirmed this observation. As a result, further enrollment of this patient subgroup was halted following the first IDMC review to avoid impacting efficacy signals from the study. From that point forward, only patients with a baseline imputed PaO2/FiO2 ≤ 200 were randomized into the study with the enrollment goal of 320 patients in this group being unchanged. The study was terminated, however, after 261 patients with a baseline imputed PaO2/FiO2 ratio ≤ 200 were randomized based on declining rates of US COVID-19 hospitalizations in the spring of 2021 and the more frequent use of tocilizumab in CARDEA candidate patients at many trial sites following recommendations by the National Institutes of Health’s COVID-19 Treatment Guidelines Panel [24]. The use of tocilizumab in combination with Auxora had been prohibited by regulatory guidance.
In CARDEA, patients were randomized 1:1 to Auxora plus standard of care or placebo plus standard of care. Participants, investigators, study teams, and the sponsor were all blinded to study drug assignment. Randomization was stratified by the baseline imputed PaO2/FiO2 ratio of > 200 vs ≤ 200 through a central, concealed, web-based, automated system. An independent statistician created the randomization schedule with stratified block randomization method using SAS proc plan procedure. Within each stratum, the treatment codes were assigned at a 1:1 ratio of Auxora and placebo with the block size of 4.
Auxora was administered by a 4-h IV infusion at 2.0 mg/kg (1.25 mL/kg) at 0-h and 1.6 mg/kg (1 mL/kg) at 24 and 48 h. Placebo was a matching formulation without the active pharmaceutical ingredient and was also dosed as a 4-h IV infusion at 1.25 mL/kg at 0-h and 1 mL/kg at 24 and 48 h. Patients were assessed for recovery and mortality using an ordinal scale in a standardized manner as described in the electronic case report form. The initial assessments occurred immediately before each infusion. Seventy-two hours after the start of the first infusion, patient assessment occurred every 24 h (± 4 h) until 240 h and then continued every 48 h until Day 30 or discharge. Patients discharged before Day 25 were contacted at Day 30 (± 5 days). All patients were again assessed Day 60 (± 5 days); patients who remained in the hospital after Day 30 were assessed by review of hospital records and those who had been discharged were contacted by telephone. Public information (e.g., death reports, governmental information) was used by sites to ascertain Day 60 mortality status in patients who refused direct contact after discharge or had withdrawn from the trial.
All patients were required to receive dexamethasone or equivalent dose of another corticosteroid as well as pharmacological prophylaxis against development of venous thromboembolic disease. Remdesivir use was recommended for all patients, and convalescent plasma administration was allowed according to local standard of care. Other immunomodulators for the treatment of COVID-19 pneumonia, including tocilizumab and JAK inhibitors, were prohibited due to regulatory guidance.
An institutional review board at each site approved the trial protocol. Informed consent was obtained from the patient or the patient’s legally authorized representative if the patient was unable to provide consent. The trial was conducted in accordance with Good Clinical Practice guidelines and the principles of the Declaration of Helsinki, and was sponsored by CalciMedica, Inc (La Jolla, CA). An independent data monitoring committee (IDMC) provided trial oversight. Operational support was provided by Bionical-Emas (Paulsboro, NJ) and Princeton Pharmatech (San Francisco, CA) performed the statistical analyses. All authors vouch for the accuracy and completeness of the data and for the fidelity of the trial adherence to the protocol.
The IDMC first reviewed unblinded safety data once 57 patients were randomized, then again after 70 patients with a baseline imputed PaO2/FiO2 ≤ 200 completed 60 days of the trial, and finally after randomization of 209 patients with a baseline imputed PaO2/FiO2 ≤ 200. The IDMC also performed an interim sample size re-estimation based on the recovery rate ratio after 70 patients with a baseline imputed PaO2/FiO2 ≤ 200 reached Day 60.
Patient population
Eligible patients were adults with ≥ 1 symptom consistent with COVID-19 infection, a diagnosis of COVID-19 confirmed by laboratory testing using polymerase chain reaction or other assay, and pneumonia documented by chest imaging. Patients were also required to be receiving oxygen therapy using either a high flow (HFNC) or low flow nasal cannula and have at the time of enrolment a baseline imputed PaO2/FiO2 ratio > 75 and ≤ 300. Patients could not be receiving either non-invasive or invasive mechanical ventilation at the time of enrolment. Full inclusion and exclusion criteria are available in the Additional file 1: Appendix.
Outcomes
The primary endpoint was time to recovery through Day 60, defined as meeting the criteria for category 6 (Hospitalized, not requiring supplemental oxygen or ongoing medical care), category 7 (Discharged, requiring supplemental oxygen), or category 8 (Discharged, not requiring supplemental oxygen) using an 8-point ordinal scale. The key secondary endpoint of all-cause mortality at Day 60 was requested by regulatory guidance. Additional secondary endpoints evaluated in the efficacy set included all-cause mortality at Day 30, the proportion of patients requiring invasive mechanical ventilation or death through Day 60, the proportion of patients requiring invasive mechanical ventilation through Day 60, and differences in outcomes measured by the 8-point ordinal scale through Day 60. Safety endpoints included the occurrence and severity of treatment-emergent adverse events (TEAEs) and serious AEs (SAEs).
The primary and key secondary endpoints were also evaluated in pre-specified subgroups of patients who required oxygen therapy via either HFNC or low flow nasal cannula at baseline or patients having a baseline imputed PaO2/FiO2 ≤ 100 or 101–200 at baseline, and in all randomized patients. The safety endpoints were evaluated in all patients who received study drug, including those with a baseline imputed PaO2/FiO2 > 200.
Statistical analysis
The efficacy set was pre-specified, consisting of those patients with a baseline imputed PaO2/FiO2 ≤ 200. A two-group log-rank test with a 0.05 two-sided significance level would have 90% power to detect a difference in the recovery rate ratio of approximately 1.49 in the 320 patients with a baseline imputed PaO2/FiO2 ≤ 200 who were randomized 1:1 to Auxora or placebo. The Sponsor elected to not change the sample size after the IDMC performed the sample size re-estimation. All supplemental analyses of the primary and first secondary endpoints were also performed in a set of all randomized patients.
Time to recovery through Day 60 in the efficacy set was compared between the Auxora and placebo groups using log-rank test stratified by baseline imputed PaO2/FiO2 ≤ 100 vs. 101–200 and displayed using a Kaplan–Meier estimate. Patients were censored at the last ordinal scale assessment if no recovery event was observed during the study and if they had recovered, been discharged, but Day 60 recovery status was not obtained.
All-cause mortality at Day 60 in the efficacy set was compared between the Auxora and placebo groups using a Cochran-Mantel- Haenszel test stratified by the baseline imputed PaO2/FiO2 ≤ 100 vs. 101–200. In addition, a sensitivity analysis was performed that estimated the 60-day death rate with hypothesis testing based on the Kaplan–Meier estimates and standard errors estimated by Greenwood formula using the log–log transformation of the survival function stratified by the baseline imputed PaO2/FiO2 of ≤ 100 vs. 101–200.
To protect the trial level type 1 error rate at α = 5% (two sided) between the primary endpoint analysis and the key secondary endpoint analysis, the Benjamini and Hochberg testing strategy was used as test statistics of time to recovery and all-cause mortality at Day 60 were positively correlated.
Role of funding source
The funder of the study had primary responsibility for the study design, protocol development, study monitoring, data management and interpretation, and statistical analyses. The funder also contributed to the drafting of the manuscript and decision to submit.