COVIDOSE: Low-dose tocilizumab in the treatment of Covid-19

Trial Design and Oversight

COVIDOSE was an investigator-initiated, non-randomized, open-label trial performed at a single site (University of Chicago Medicine [UCM]). Details of the low-dose tocilizumab treatment strategy have been described previously.¹⁴ The University of Chicago sponsored the trial but institutionally had no role in the design, execution, or analysis of this trial. There was no industry involvement in COVIDOSE. Disclosures and conflicts of interest have been reported. COVIDOSE was registered with the National Clinical Trials Registry (NCT04331795).

The UCM institutional review board (IRB) approved the trial protocol. Given the rapidly evolving landscape of Covid-19, the IRB authorized an internal trial operating committee to facilitate dose modifications based on incoming data (full details of this operating structure are available in the COVIDOSE trial protocol and Supplementary Appendix). An internal data monitoring committee oversaw the trial’s progress and adjudicated outcomes, safety signals, and potential protocol deviations in weekly meetings. All authors attest to the validity of the data and adherence to the trial protocol, with reporting of violations where indicated.

Study Participants

Eligible patients were those hospitalized at UCM, aged 18 years or older, and who had: 1) positive polymerase chain reaction test, most commonly nasopharyngeal, for SARS-CoV-2 RNA, 2) chest radiograph findings such as bilateral ground glass opacities or hazy bilateral infiltrates consistent with viral or atypical pneumonia (as determined by UCM radiologists providing usual clinical care independent of COVIDOSE), 3) documented fever, defined as temperature ≥ 38·0°C in the 24 hours prior to the time of tocilizumab administration as measured by commonly accepted clinical methods (predominantly oral or axillary), and 4) C-reactive protein (CRP) ≥ 40 mg/L. Key exclusion criteria included invasive mechanical ventilation, vasopressor medications, active therapy with biological immunosuppressive or Janus kinase inhibitor medications, and previous receipt of an investigational antiviral agent or off-protocol anti-IL6R therapy. Full eligibility criteria are available in the trial protocol. Patients or their legally authorized representatives provided written or electronic informed consent.

Trial Procedures

Enrolled patients were subdivided into two groups (Group A and Group B) based on laboratory signs of hyperinflammation and the presence or absence of risk factors for Covid-19-related mortality (based on extant univariate regressions at the time of study and as agreed to by multidisciplinary panel of institutional experts). Risk factors, nearly all of which were subsequently validated,²³ included: any previous intensive care unit admission; previous non-elective intubation; hospitalization for exacerbation of congestive heart failure or chronic obstructive pulmonary disease in the past 12 months; coronary artery disease requiring percutaneous coronary intervention or coronary artery bypass grafting; stroke with residual neurologic deficit; pulmonary hypertension; home supplemental oxygen use; interstitial lung disease; asthma requiring inhaled corticosteroid use; history of pneumonectomy, lobectomy, or radiation therapy to lung; HIV/AIDS; cancer diagnosis (any stage) receiving non-hormonal treatment; immunodeficiency; end-stage renal disease requiring hemodialysis or peritoneal dialysis; body-mass index > 30 kg/m2; and inpatient supplemental oxygen requirement > 6L/min at the time of enrollment. Group A patients had CRP ≥ 75 mg/L and at least one risk factor for mortality. Group B patients had CRP of 40-74 mg/L or lacked risk factors for Covid-19-related mortality.

After group assignment, patients were treated according to the low-dose tocilizumab algorithm, integrated into usual clinical care (Figure S1).¹⁴ Group A patients received tocilizumab 200 mg (cohort 1A) or 120 mg (cohort 2A). Group B patients received 80 mg (cohort 1B) or 40 mg (cohort 2B). All dosage levels were protocol-based and determined on a cohort basis by the trial operating committee. Vital signs and laboratory studies were monitored per usual clinical care. CRP was evaluated immediately prior to tocilizumab administration (baseline) and approximately 24 hours following tocilizumab administration, except where detailed. Patients were eligible for re-dosing with tocilizumab. In the originally developed re-dosing schema, the decision was guided strictly by CRP response. In the final version of the protocol, however, the re-dosing decision was guided by biochemical and clinical parameters. Patients were re-dosed with tocilizumab if signs of clinical worsening (as defined by increased supplemental oxygen requirement or worsening fever curve determined by maximum temperature) were accompanied by CRP decline of < 25% when compared to baseline at the 24-hour assessment (Figure S1). In the COVIDOSE algorithm, treating physicians maintained the option of administering off-protocol tocilizumab 400 mg as indicated per their clinical judgment. Please see the Supplementary Appendix for further information.

Patients were followed for the duration of their hospitalization and were contacted 28 days after tocilizumab administration. At the 28-day timepoint, survival, clinical status (including admission to a medical or assisted living facility), new or persistent supplemental oxygen requirement, and any notable diagnoses or treatments for secondary infection were documented. Planned enrollment was fifty patients, with the goal of identifying a pharmacodynamically active, low dose of tocilizumab for the treatment of Covid-19. After enrolling the thirty-second patient, remdesivir received Emergency Use Authorization and quickly became institutional standard of care.²⁴ To avoid collecting heterogeneous data confounded by remdesivir use, the COVIDOSE trial operating committee elected to close the study to further enrollment; it was not closed due to safety concerns.

Retrospective Controls

A retrospective control group was assembled using an institutional Covid-19 clinical data warehouse of de-identified medical records by identifying patients who met all COVIDOSE eligibility criteria (Figure S2). An additional antibody control cohort with blood samples available from the institutional Covid-19 biobank was used to compare antibody responses (Figure S3). Full descriptions of the retrospective control populations’ derivations are available in the Supplementary Appendix.

Outcomes

The primary clinical outcome was resolution of fever in the 24-hour period following tocilizumab, defied as a maximum temperature (T_max24hrs) < 38·0°C. Originally, the primary biochemical outcomes were rate of and time to CRP normalization, guided by earlier tocilizumab-related work.^9,14 During the conduct of the study, it became apparent that a patient could be safely discharged before CRP had normalized, making CRP normalization impractical. We therefore report the percentage of patients who achieved biochemical response, defined as a CRP reduction ≥25% from baseline in the 24-48 hours after tocilizumab administration, consistent with a decline determined by CRP half-life.^9,25

Secondary outcomes included overall survival at 28 days, survival to hospital discharge, rate and duration of non-elective mechanical ventilation, time to mechanical ventilation, rate and duration of vasopressor/inotropic agent utilization, time to vasopressor/inotropic agent utilization, and number of days spent in intensive care unit. One month after enrollment began, duration of supplemental oxygen requirement greater than a patient’s baseline was added. Definitions of outcomes are provided in the protocol. We do not include results in this manuscript for all endpoints.

Post-Hoc Evaluation of Clinical Recovery

Randomized, controlled trials of Covid-19 therapies had not been published at the time of COVIDOSE initiation but were published prior to COVIDOSE data maturity.⁶ To add context to COVIDOSE results, time to clinical recovery was evaluated post-hoc. Using a previously developed seven-point ordinal scale,²⁶ disease severity was assessed daily, with the patient’s worst clinical status and consequent ordinal score recorded for a given day.^6,26 Recovery was defined as the first day on which a patient achieved a clinical status of either “hospitalization without need for supplemental oxygen or ongoing medical care” or “not hospitalized”.^6,26 Time to recovery was defined as the time (in days) from study enrollment to achievement of recovery.⁶ The detailed clinical annotations needed to categorize patients were not available for the retrospective control population. Two independently operating authors (GWS, PDR) adjudicated ratings.

SARS-CoV-2 Spike and RBD Enzyme-Linked Immunosorbent Assays

The SARS-CoV-2 Spike and Receptor Binding Domain (RBD) protein expression constructs were obtained from Florian Krammer and Patrick Wilson,²⁷ and used to generate recombinant protein for an enzyme-linked immunosorbent assay (ELISA) adapted from established protocols.²⁷ Goat anti-human immunoglobulin (Ig)-Horseradish peroxidase (HRP) reactive to heavy and light chains of human IgG, IgM, and IgA was used as the secondary antibody (SouthernBiotech; Birmingham, AL, USA).

Antibody titers were determined by measuring total immunoglobulin (IgG, IgM, IgA) reactive against the SARS-CoV-2 Spike glycoprotein and its receptor-binding domain (RBD). The Spike glycoprotein, a structural protein of SARS-CoV-2, is a crucial component in the attachment and entry of the virus into host cells. ELISA titers against the spike protein have been reported to correlate significantly with virus neutralization.²⁷ Specifically, the RBD protein binds the angiotensin-converting enzyme 2 (ACE2) receptor on human cells, and likely serves as a target for neutralizing antibodies that facilitate virus clearance. Full experimental details available in the supplemental methods.

Statistical Analysis

We assessed differences in the clinical and biochemical response rates of patients treated with low-dose tocilizumab compared with the retrospective control population using a two-sided Fisher’s exact test. We also assessed differences in the clinical and biochemical response rates between different tocilizumab doses using a two-sided Fisher’s exact test, unadjusted for cohort or other risk factors. A p-value of < 0·05 was considered to indicate statistical significance. All analyses were performed using Stata (Stata Corp., College Station, TX, USA). For post hoc analysis, descriptive data summaries are reported as median (interquartile range) or percentages. Log₁₀ titers of anti-SARS-CoV-2 antibodies in the tocilizumab-treated and untreated populations were compared using Welch’s t-test in R (R Foundation for Statistical Computing; Vienna, Austria).

Findings

Trial Population

From April 1 through May 13, 2020, 32 patients consented to participate in COVIDOSE. Twelve were assigned to Group A, 8 of whom received 200mg and 4 of whom received 120mg. Twenty were assigned to Group B, 15 of whom received 80 mg and 5 of whom received 40 mg (Figure 1). Median time from hospital admission to study enrollment was 1 day (IQR, 1-2 days). All patients were included in the statistical analysis of fever resolution, and 29 patients were included in the statistical analysis of CRP response. There were 41 patients in the retrospective control population (Figure S2). The COVIDOSE subpopulations and retrospective control population differed significantly in terms of epidemiologic risk factors for Covid-19-related decompensation and hydroxychloroquine use (Table 1). Re-dosing of low-dose tocilizumab is discussed in the Supplementary Materials.

• Download figure
• Open in new tab

Figure 1. Flow of patients in the COVIDOSE trial.

Thirty-two eligible patients consented to participate in the COVIDOSE trial, of which 12 were assigned to Group A and 20 were assigned to Group B on the bases of magnitude of C-reactive protein elevation and epidemiologic risk factors for Covid-19 related mortality. Group A patients received either 200 or 120 mg of tocilizumab and were followed for 28 days following drug administration. Group B patients received either 80 or 40 mg of tocilizumab and were followed for 28 days following drug administration. All 32 patients were evaluable for the purposes of primary clinical outcome and 29 were evaluable for biochemical outcome. No patients were lost to follow-up.

View this table:

• View inline
• View popup
• Download powerpoint

Table 1. Demographic, clinical, and laboratory characteristics of patients enrolled on COVIDOSE and the retrospective control population.

Epidemiologic risk factors include:

Clinical Outcomes

Curves of maximum daily temperature (mean ± SE) for the COVIDOSE and retrospective control populations are shown in Figure 2A. At 24 hours following tocilizumab administration, 24 COVIDOSE patients (75·0%) and 14 retrospective control patients (34·1%) achieved T_max24hrs < 38·0°C (P = 0·001 for difference compared to control). Eight of 12 (66·7%) Group A patients achieved T_max24hrs < 38·0°C (P = 0·055, compared to control) and 16 of 20 (80·0%) Group B patients achieved T_max24hrs < 38·0°C (P = 0·001, compared to control) (Figure 2B). Eleven of 15 (73·3%) patients in the 80 mg group achieved T_max24hrs < 38·0°C, compared to 5 of 5 (100%) patients in the 40 mg group (P = 0·53) (Figure 2C). T_max24hrs < 38·0°C was observed in 5 of 8 (62·5%) patients in the 200 mg group and 3 of 4 (75·0%) patients in the 120 mg group (P = 1·0) (Figure 2C).

• Download figure
• Open in new tab

Figure 2. Temperature and C-reactive protein decrease rapidly following administration of low-dose tocilizumab compared to retrospective controls.

(A) Temperature values for COVIDOSE patients (closed circle, solid line) and retrospective control patients (open circle, dotted line) following eligibility (mean ± standard error of mean). (B) Percentage of retrospective control (black), COVIDOSE (dark gray), COVIDOSE Group A (light gray), and COVIDOSE Group B (white) patients meeting the primary outcome of T_max24hrs < 38·0°C. Bars denote comparisons between groups. (C) Relationship between tocilizumab dose and probability of achieving fever resolution. Percentage of COVIDOSE patients at the noted tocilizumab dose levels achieving the outcome of T_max24hrs < 38·0°C. (D) Representative C-reactive protein values for COVIDOSE patients (closed circle, solid line) and retrospective control patients (open circle, dotted line) following eligibility (mean ± standard error of mean). (E) Percentage of retrospective control (black), COVIDOSE (dark gray), COVIDOSE Group A (light gray), and COVIDOSE Group B (white) patients meeting the primary biochemical outcome of CRP decline of ≥ 25% at 24-48 hours. 15 historical control patients with no C-reactive protein readings between 24 and 48 hours were excluded from analysis. Bars denote comparisons between groups. (F) Relationship between tocilizumab dose and probability of achieving CRP decline of ≥ 25%. Percentage of COVIDOSE patients at the noted tocilizumab dose levels achieving CRP decline of ≥ 25%. Bars denote one-sided Fisher’s exact test for difference in proportions. CRP, C-reactive protein. P values are as shown in the figure.

Tocilizumab does not inhibit the antibody response to SARS-CoV-2

To evaluate whether IL-6 axis blockade impairs the generation of antibody responses to SARS-CoV-2, we analyzed longitudinal blood samples from patients enrolled in a Covid-19 biobanking protocol. Twenty-three patients treated with tocilizumab (n = 19 on COVIDOSE, n = 4 off-protocol 400 mg) and 37 matched control patients were retrospectively identified from the biobank (Figure S3; Table S1). All analyzed patients had pulmonary infiltrate on chest radiograph, temperature ≥ 100·4°F, and CRP ≥ 40 mg/L within 24 hours of tocilizumab administration (Figure S3). Matched control patients were eligible if they met the above criteria (Figure S3). Patients with bacteremia or requiring invasive ventilation were excluded (Figure S3). Patients were not excluded if they had received investigational remdesivir (n = 26; 70·3%).

All patients treated with tocilizumab generated antibodies to the SARS-CoV-2 Spike protein and RBD during infection (Figure 3). Anti-Spike (Figure 3A, 3B) and anti-RBD (Figure 3C, 3D) antibodies from samples 10+ days after symptom onset reached comparable maximum log₁₀ titers in tocilizumab-treated and matched control patients (Spike, 4·911 vs 5·000, p = 0·726; RBD, 4·365 vs 4·748, p = 0·149). The rates of titer change from first positive sample to maximum titer within 21 days were not significantly different between tocilizumab-treated and matched control patients (Spike, 1·818 vs 1·298, p = 0·221; RBD, 1·418 vs 1·152, p = 0·412). Although the number of samples in smaller subgroups limits the power of comparisons, there were no differences between COVIDOSE patients and those who received 400 mg (Spike, 4·953 vs 4·745, p = 0·721; RBD, 4·429 vs 4·106, p = 0·48), or between remdesivir-treated or -untreated control patients (Spike, 5·068 vs 4·811, p = 0·63; RBD, 4·867 vs 4·422, p = 0·369) (Figure S4). Taken together, these data suggest that tocilizumab as administered in this study does not impair the induction of anti-SARS-CoV-2 antibodies during infection.

• Download figure
• Open in new tab

Figure 3. Tocilizumab treated patients develop similar peak antibody levels as control patients.

Spike (A,B) and RBD (C,D) antibody titers in tocilizumab-treated (A,C) vs control (B,D) patients over time. Open circles represent time points prior to tocilizumab treatment, and filled circles are after tocilizumab treatment. Each line/color represents a separate patient. Dotted horizontal line represents level of detection in normal donor controls.

Interpretation

Low-dose tocilizumab was associated with rapid clinical and biochemical improvement in patients with Covid-19 and related hyperinflammation who did not require invasive ventilation, with no apparent relationship between tocilizumab dose and clinical or biochemical improvement. Moreover, there was no apparent diminution in the development of anti-SARS-CoV-2 antibodies in patients who received tocilizumab. Although the minimum effective dose of tocilizumab was not identified, the COVIDOSE study demonstrated that a tocilizumab dose of 40 mg may be sufficient to blunt the clinical and biochemical signs of Covid-19-related hyperinflammation. Referencing clinical pharmacology arguments,¹⁴ the absence of a dose-response relationship suggests that a dose lower than those utilized in other Covid-19 studies (400 mg) or approved by regulatory bodies for the treatment of CAR-T-related CRS (4-8 mg/kg) may be sufficient. The COVIDOSE trial provides proof-of-concept for a Covid-19 therapeutic strategy centered on the administration of low-dose tocilizumab, with re-dosing guided by readily assessable clinical and biochemical responses.

To our knowledge, COVIDOSE is the first trial to provide prospectively collected evidence of low-dose tocilizumab’s clinical benefit. Marked declines in maximum body temperature and CRP were noted in the 24-48 hours following administration of low-dose tocilizumab – responses that mimic those observed in patients with severe and critical Covid-19 who received tocilizumab 400 mg.⁹ Post hoc analysis of COVIDOSE patients’ clinical trajectories provides encouraging evidence of low-dose tocilizumab’s benefit: Patients requiring supplemental oxygen at enrollment experienced median time to recovery of 4 days, three days shorter than the median time to recovery of a similarly risked subpopulation receiving remdesivir.⁶ The COVIDOSE data will add context to the interpretations of the COVACTA, CORIMUNO-TOCI, and RECOVERY tocilizumab trials when published, as well as the emerging real-world evidence for tocilizumab’s benefit.^10-12,16

The data presented in COVIDOSE suggest that a three- to ten-fold increase in the effective supply of tocilizumab may be possible. In this trial, we administered a total of 4,120 mg of tocilizumab, corresponding to 37 doses for 32 patients; administering 400 mg to each patient would have required 12,800 mg.^9,11 Rapid, global scaling is required for tocilizumab to become a safe, effective, and equitable therapy for hospitalized patients with Covid-19 – an interventional pharmacoeconomics-based approach is one possible strategy for extending supplies.²⁰

Beyond sustainability, low-dose tocilizumab may be clinically advantageous over higher dosages. The rate of secondary bacterial infection in COVIDOSE was approximately one-fourth that of patients who received tocilizumab 8 mg/kg.^10,16 Biological studies conducted in parallel with COVIDOSE did not find statistically significant differences in anti-SARS-CoV-2 antibody production between patients treated with tocilizumab (standard or low dose), remdesivir, or supportive care alone. How the prospectively-collected rates of secondary infections compare between low-dose tocilizumab-, tocilizumab 8mg/kg-, and dexamethasone-treated patients, as well as the combination of dexamethasone and tocilizumab, remains to be seen.² Carefully conducted prospective trials with appropriate biological correlative studies are needed.

The COVIDOSE trial is not without limitations. It was a small, single-center study conducted at a tertiary care center with active research efforts in Covid-19-related supportive care. Our trial lacked a contemporaneous, randomized comparator arm, and the higher rate of epidemiologic risk factors in the retrospective control population may have contributed to its slow recovery in terms of both fever resolution and CRP decline. The absence of a randomized control arm limits formal conclusions about low-dose tocilizumab’s safety and efficacy. Finally, the rapidly changing understanding of Covid-19 and local practice prevented trending CRP to normalization, leading to the need for reporting of a different biochemical outcome than originally expected. The COVIDOSE study, however, provides proof-of-concept as well as valuable lessons that will be incorporated into a randomized, controlled trial evaluating the clinical efficacy of low-dose tocilizumab in Covid-19.