Background

Coronavirus disease 2019 (COVID-19) has had a substantial impact on day-to-day living over the last 4 years. Although COVID-19 is asymptomatic or results in mild symptoms in most individuals, some patients still require hospitalisation due to development of severe pneumonia [1].

Severity of COVID-19 was defined initially by respiratory parameters [2, 3]. Now additional markers indicating systemic inflammation, such as high C-reactive protein (CRP) levels, and markers indicating dysregulated coagulation, such as elevated D-dimer and fibrinogen, low platelet counts and prolongation of prothrombin time, have been associated with disease severity [4,5,6,7]. Markers of dysregulated coagulation may indicate hypercoagulability (also called COVID-19-associated coagulopathy) that may lead to intravascular thrombotic complications. Together with various hyperinflammatory immune responses, these mechanisms lead to immunothrombosis, which is thought to be a major contributor to morbidity and mortality in COVID-19 [8, 9].

Given these links with systemic inflammation, immune-modulating therapies have now become part of the therapeutic pathway in patients hospitalised with COVID-19 [3]. Indeed, hospitalised patients with severe or critical COVID-19 have been shown to benefit from treatment with immunomodulatory drugs, some of which have been granted regulatory approval and are included in COVID-19 treatment guidelines (e.g. dexamethasone, tocilizumab and baricitinib) [10, 11]. For these medications, results from different trials provided evidence of benefit in certain patient subpopulations with COVID-19 [12].

However, despite these developments and the declining rates of severe COVID-19, expansion of treatment approaches for hospitalised COVID-19 patients remains desirable. Currently approved medications may not be available universally, vaccines may not elicit an immune response or may be contraindicated, or new, more virulent variants may appear, against which current antiviral therapies may be less effective or effective vaccines may not yet be available.

Trimodulin is a human plasma-derived native polyvalent antibody preparation in clinical development for respiratory tract infections. In contrast to other intravenous immunoglobulin (Ig) preparations (IVIg), which contain ≥ 95% IgG, trimodulin contains ~ 56% IgG plus relevant amounts of IgM (~ 23%) and IgA (~ 21%). In addition to anti-pathogen activity, polyvalent IgM is immune modulating at the complement level [13,14,15], and both polyvalent IgM and IgA are immune modulating at the cytokine level [16,17,18]. Trimodulin is also assumed to contain relevant amounts of natural IgM [19]. Natural IgM is a first-line defence against pathogens but also plays a role in maintenance of tissue homeostasis via the clearance of damaged and apoptotic cells [19,20,21]. Given these multiple modes of action, use of trimodulin represents a new therapeutic strategy for COVID-19 compared with those that suppress the immune system more broadly or target only a single component of an inflammatory pathway.

In a previous phase II clinical trial, trimodulin improved outcomes of patients with severe community-acquired pneumonia (sCAP) on invasive mechanical ventilation (IMV), evidenced by a significantly lower mortality rate in subpopulations with elevated CRP levels, or with reduced IgM serum concentrations, or both [22]. The hypothesis for the present Escape from severe COVID-19 (ESsCOVID) clinical trial was that trimodulin may prevent inflammation-driven progression of severe COVID-19 to critical disease or even death. Accordingly, the efficacy and safety of trimodulin in adults hospitalised with severe COVID-19 was investigated. An additional post hoc analysis was performed to identify those patients that benefited most from treatment with trimodulin to inform the design of future clinical trials.

Methods

Trial design

ESsCOVID was a phase II, randomised, placebo-controlled, double-blind, multicentre clinical trial (NCT04576728) conducted across 16 centres in Brazil, France, Russia, and Spain. Blinding (investigators, patients, and all personnel involved in the conduct and outcome assessments of the trial) was maintained until after database lock. The trial was conducted according to the International Council for Harmonisation, Good Clinical Practice standards and the Declaration of Helsinki, and with independent ethics committee approval. Written informed consent from the patient, or legally authorised representative, was obtained in compliance with all local legal requirements.

Sample size calculation

Data from clinical studies conducted at a similar time during the COVID-19 pandemic, reported that ~ 40% of severe patients on non-invasive ventilation or high-flow oxygen deteriorated and of these, approximately 50% died [2, 23,24,25,26]. Therefore, a deterioration/mortality rate of 40% of patients was assumed in the placebo group. A previous phase II trial with trimodulin (CIGMA trial [22]) was performed in patients with sCAP caused by any pathogen. In a subgroup with elevated CRP levels, the mortality was reduced by 16.7% and progression to septic shock was reduced by 7.8% [22]. Accordingly, a deterioration/mortality rate of 20% was assumed in the trimodulin group. Based on these assumptions the trial was powered at 80% to detect a difference of 20 percentage-points in the composite primary endpoint (placebo: 40%, trimodulin: 20%) with a sample size of 164 patients. Sample size estimation was performed using nQuery Version 8.5.1.0 and the statistical analysis was performed using SAS® version 9.4.

Patient population

Adult patients (≥ 18 years of age) hospitalised with severe COVID-19 were enrolled. At screening, patients were required to have laboratory-confirmed severe acute respiratory syndrome coronavirus type 2 (SARS-CoV-2) infection from a test performed on a respiratory tract sample within the last 5 days and a diagnosis of community-acquired severe COVID-19 within 10 days after hospital admission. Severe COVID-19 was defined as the need for non-invasive ventilation (NIV) and/or high-flow oxygen (HFO; via nasal cannula or mask; score 5 [hospitalised with severe disease] on the 9-category ordinal scale, where 0 is non-hospitalised [discharged/cured] and 8 is death; Additional file 1: Table S1). In addition, patients were required to have at least one of five clinical respiratory parameters (dyspnoea, respiratory frequency ≥ 30 breaths/min, SpO2 ≤ 93%, PaO2/FiO2 100–300 mmHg, lung filtrates > 50% within 24 to 48 h). Patients were also required to have at least one measurement of CRP ≥ 50 mg/L within 36 h prior to the start of treatment.

Patients were excluded if they deteriorated prior to randomisation, as reflected by, for example, the need for IMV (score > 5 on the 9-category ordinal scale; Additional file 1: Table S1) or improved so they were on low-flow oxygen or no oxygen prior to randomisation (score < 5). Patients were also excluded if they had severe neutropenia (neutrophil count < 500/mm3), thrombocytopenia (platelet count < 30,000/mm3) or haemoglobin < 7 g/dL within 24 h of treatment initiation, known haemolysis, or had known thrombosis or thromboembolic events (TEEs) within the previous 3 months. Patients particularly at risk of TEEs for reasons other than COVID-19 were also excluded. In addition, patients on dialysis or with severe renal impairment, estimated glomerular filtration rate < 30 mL/min/1.73 m2 assessed within 24 h of starting treatment, or patients with end-stage renal disease or known focal segmental glomerulosclerosis were excluded, as were those with known severe lung diseases interfering with COVID-19 treatment, decompensated heart failure, pre-existing hepatic cirrhosis or severe hepatic impairment (Child–Pugh score ≥ 9 points) or hepatocellular carcinoma, and those who had received treatment for thorax, head, neck or haematological malignancies in the previous 12 months.

Randomisation and treatment schedule

Eligible patients were randomised 1:1 on Day 1 to either trimodulin or placebo stratified by centre according to a pre-defined randomisation list generated and implemented by interactive response technology. Trimodulin (BT588; Biotest AG, Dreieich, Germany) or an equal volume of placebo (1% human albumin solution; Biotest AG, Dreieich, Germany) were administered as intravenous infusions on 5 consecutive days (Days 1 to 5). The volume of trimodulin or placebo administered was 3.65 mL/kg body weight/day. This corresponded to doses of 182.6 mg trimodulin/kg body weight/day or 36.5 mg albumin/kg/body weight/day (as used in the previous phase II CIGMA trial). Infusion was started at a rate of 0.1 mL/min and increased by 0.1 mL every 10 min if tolerated up to a maximum infusion rate of 0.5 mL/min. Patients were followed up to Day 29 or up to hospital discharge, whichever occurred first. An end-of-trial telephone interview was conducted on Day 29 for patients discharged or transferred.

Clinical assessments

Each patient was tested for SARS-CoV-2 at screening. Local laboratory assessment of clinical chemistry, haematology and coagulation parameters was performed on Days −1, 1–5 (pre-dose), 7, 14 and 21, and on Day 29/discharge. In addition a physical examination was performed on Days 1 and 29/discharge, with vital signs (including blood gas measurements) assessed on Days −1, 1–7, 9, 14 and 21, and Day 29/discharge. Samples for pharmacokinetic (PK) assessment were taken pre-dose on Days 1 and 5, post-dose on Day 5 and on Day 29/discharge. Samples for pharmacodynamic (PD) assessment were taken pre-dose on Days 1, 3 and 5, and on Days 9 and 29/discharge.

Hospitalisation and intensive care unit (ICU) dates, as well as oxygen supply type and dates and the daily clinical status of the patients according to the ordinal scale, were recorded.

Standard of care

Standard of care (SoC) included procedures for acute respiratory distress syndrome (ARDS), IMV and extra corporeal membrane oxygenation (ECMO), including prone positioning and weaning. SoC also covered all prior and concomitant medication given due to the COVID-19 infection and the subject’s clinical situation, including antivirals, antibiotics, corticosteroids and antithrombotic therapy, according to local guidelines and protocols. Use of other Ig preparations, interferons, blood products (including [convalescent] plasma and albumin), passive immunisations, or active vaccinations and extracorporeal cytokine adsorbing therapy was prohibited within 21 days before entering the trial and during the trial. Use of other antibody-containing products, including immune-modulating monoclonal antibodies and antiviral monoclonal antibody products, was allowed prior to, but was prohibited during the trial. In cases where these types of prohibited medications were administered during the trial (e.g. in an emergency, to avoid further aggravation, or by accident), these patients were excluded from the per-protocol set (PPS), PK and/or PD sets.

Efficacy assessment

The primary endpoint was assessed in the full analysis set (FAS, n = 166). The primary efficacy endpoint was a composite endpoint of two parameters assessed by using the 9-category ordinal scale (Additional file 1: Table S1): the clinical deterioration rate assessed during the post-treatment follow-up period (between Days 6 and 29) and the 28-day all-cause mortality rate (score = 8) assessed from the first day of treatment (Day 1) up to the end of the follow-up period (Day 29). Deterioration was defined as worsening to requirement for IMV (score = 6) and/or development of additional organ dysfunctions, organ failures, sepsis and/or septic shock (score = 7). The maximum score reached within 28 days was applied.

Secondary efficacy endpoints included, among others, proportion of patients recovered on Day 29 (score ≤ 2 on the ordinal scale), days of IMV and ICU-free days, hospital-free days, and days without oxygen supply.

Safety assessment

Safety was assessed between signing the informed consent form and Day 29 in the safety analysis set (SAF; n = 166). Adverse events (AE) were classified using the Medical Dictionary for Regulatory Activities (MedDRA) version 17.1 [27]. Treatment-emergent adverse events (TEAEs; defined as an AE that occurred from the time of first dose of study medication until the end of the trial, independent of relation to study medication), vital signs, electrocardiograms and laboratory parameters, including clinical chemistry, haematology, coagulation and urinalysis, were recorded.

Statistical analyses

The composite primary endpoint was evaluated in the FAS (including all patients who received at least one dose of study medication and had at least one primary efficacy post-dose assessment) using a 2-sided chi-square test with a significance level of 5%. As supportive analysis, a logistic regression was performed to adjust for potential risk factors (age, sex, diabetes, history of heart disease, or other comorbidity at time of informed consent). Kaplan–Meier curves were presented for time to deterioration/mortality. A forest plot including odds ratios (OR) and 95% confidence intervals (CI) for the primary composite endpoint (clinical deterioration plus 28-day mortality) was created for different patient populations.

As a sensitivity analysis, efficacy was evaluated in the PPS, which excluded patients from the FAS with major protocol deviations (not meeting eligibility criteria, use of prohibited medication, trial discontinuation prior to Day 6 for any reason other than death). Safety, demographics and baseline characteristics were assessed using the SAF, which included all patients who received at least one dose of study medication.

In the case of missing deterioration/28-day mortality data, the clinical status up to and including Day 29 was imputed (Additional file 1: Handling of missing data).

Pharmacokinetic analysis was performed in the PK set (n = 146), which included patients who did not receive prohibited medication affecting Ig serum concentrations (e.g. plasma, IVIg or monoclonal antibody therapies) during the trial. For all subjects in the PK set, observed serum concentrations of IgM, IgA, and IgG (g/L) were summarised (mean ± standard deviation [SD], median, interquartile range [IQR]) by visit. Statistical summaries were presented per treatment group, and for survivors versus non-survivors in the two treatment groups.

Pharmacodynamic analysis was performed in the PD set (n = 142), which included patients who did not receive prohibited medications affecting immune responses (e.g. anti-inflammatory or immunosuppressing treatments) during the trial. Use of corticosteroids was permitted in the PD set. Patients in the PD set with available baseline levels for CRP, D-dimer AND platelets (n = 132) were subdivided into patients with early systemic inflammation (n = 80) or advanced systemic inflammation (n = 52).

Post hoc analyses

Exploratory, post hoc analyses driven by data and clinical/disease pathology were performed after unblinding of all data sets, to identify potential patient subgroups that benefitted most from treatment with trimodulin. During the pandemic, threshold levels for inflammatory markers related to increased risk of critical disease or death have been identified. For CRP, a median level of 125 mg/L [28] and an interquartile range (IQR) of 50–150 mg/L [4] or a median of 100.0 (IQR 60.7–179.4) mg/L [29] have been reported for non-survivors. For D-dimer, concentrations > 1 mg/L were found to be the strongest independent predictor of mortality [6], with an IQR of 3.8–8.0 mg/L associated with significant risk of death in critically ill patients [29], a value that is similar to the threshold of > 3.06 mg/L reported by Pan and colleagues [30]. In addition, mortality in COVID-19 has been associated with thrombocytopenia (defined as platelets ≤ 125 × 109/L [7]). Based on these reported thresholds, the relevance of these markers to disease severity, and on data from the present trial, a data-driven subgroup of patients with early systemic inflammation was defined by excluding patients with CRP > 150 mg/L and/or D-dimer ≥ 3 mg/L and/or platelets < 130 × 109/L. Post hoc analyses in this subgroup were conducted in the FAS and PPS.

Results

Patient disposition

Between 6 October 2020 and 29 June 2021, 185 patients were screened, with 166 being randomised to either trimodulin (n = 84) or placebo (n = 82). All 166 randomised patients were included in the SAF and FAS, irrespective of whether they completed or discontinued treatment. A total of 141 patients were analysed in the PPS (trimodulin, n = 70; placebo, n = 71). For participant flow and details on different analysis sets, see Additional file 2: Fig. S1. The trial ended after the planned recruitment of at least 82 patients per arm.

Baseline demographics and patient characteristics

Baseline demographics and patient characteristics were generally balanced between groups (Table 1), although more patients in the placebo group had a history of heart disease. In both groups, the majority of patients received NIV or HFO at the time of treatment initiation. Three patients in the trimodulin group deteriorated before (n = 1; excluded from PPS) or after (n = 2) randomisation and required IMV before/at the start of treatment.

Table 1 Baseline demographics and patient characteristics (SAF)
Full size table

Most patients (68.1%) received at least one medication that started and stopped prior to the first infusion of trimodulin or placebo. Prior medications specifically given for COVID-19 included corticosteroids for systemic use (including dexamethasone [28.3%], prednisolone [5.4%], others [3.6%]), immunosuppressants (tocilizumab [15.1%], olokizumab [12.0%], hydroxychloroquine [9%], baricitinib [4.2%], levilimab [4.2%], tofacitinib [3.6%] and sarilumab [0.6%]) and antivirals for systemic use (including favipiravir [21.1%], remdesivir [6.0%], and others [7.2%]). All patients had at least one concomitant medication that started before and was ongoing at treatment initiation, or that started on or after treatment initiation but no later than the Day 29 visit (Additional file 2: Table S2). Concomitant medications specifically used to treat COVID-19 in the trimodulin and placebo groups during the trial included corticosteroids for systemic use (76.2% and 72.0%, including dexamethasone [54.8% and 42.7%]), antivirals for systemic use (36.9% and 30.5%, such as remdesivir [7.1% and 3.7%]), and immunosuppressants (16.7% and 13.4%, including tocilizumab [3.6% and 1.2%] and baricitinib [1.2% for both]) (Additional file 2: Table S2).

Pharmacokinetics

In the PK set, mean levels of IgM (1.3 g/L vs 1.2 g/L), IgA (2.9 g/L vs 2.8 g/L) and IgG (10.1 g/L vs 9.9 g/L) were similar between trimodulin (n = 73) and placebo (n = 73) at baseline and were all well within the normal range (Fig. 1). For patients receiving placebo, the mean concentrations of all three Igs remained close to baseline levels until Day 29. Treatment with trimodulin resulted in a significant increase in all three Igs up to Day 5 end of infusion compared with baseline: for IgM, a mean ± SD concentration of 2.5 ± 0.98 g/L (median: 2.4 g/L; IQR 1.9–2.9 g/L) was achieved (P < 0.001), and this value was marginally above the upper limit of the normal (ULN, Fig. 1A). For IgA, a mean concentration of 5.1 ± 1.4 g/L (median: 5.0; IQR 4.1–5.7 g/L) was achieved (P < 0.001), and this value was above the ULN (Fig. 1B). For IgG, a mean concentration of 15.9 ± 3.1 g/L (median: 15.3; IQR 14.1–17.6 g/L) was achieved (P < 0.001), and this value was close to the ULN (Fig. 1C). For all three Igs, levels had returned to near baseline by Day 29. No difference in PK was observed between survivors and non-survivors (Additional file 2: Table S3).

Fig. 1
figure 1

Pharmacokinetics of IgM, IgA and IgG in COVID-19 patients (PK set). Serum concentrations (mean ± SD) for IgM (A), IgA (B) and IgG (C) were assessed in patients in the PK set. Graphs show PK assessments from samples taken pre-dose on Day 1 (trimodulin, n = 73; placebo, n = 73) and post-dose on Day 5 (trimodulin, n = 53; placebo, n = 63) and Day 29 (trimodulin, n = 7; placebo, n = 8). Dotted line: normal reference ranges [31]. COVID-19 coronavirus disease 2019, Ig immunoglobulin, PK pharmacokinetics, SD standard deviation

Full size image

Efficacy

In the FAS, clinical deterioration/28-day all-cause mortality (composite primary endpoint) was reported in 33.3% of patients in the trimodulin group and 34.1% of patients in the placebo group (OR 0.96; 95% CI: 0.51, 1.84; P = 0.912, Fig. 2A). Supplementary logistic regression analysis to correct for risk factors (covariates: age, sex, diabetes, history of heart disease or other comorbidity) at time of informed consent, determined an OR of 1.07 (95% CI: 0.55, 2.09; P = 0.836) (not shown). In line with these results, no difference (P = 0.84, log-rank test) in clinical deterioration/28-day all-cause mortality was observed in Kaplan−Meier analysis between the trimodulin and placebo groups in the overall population (Fig. 2B). In the PPS (not shown), clinical deterioration/mortality was 31.4% in the trimodulin group and 35.2% in the placebo group (OR 0.84; 95% CI: 042, 1.70; P = 0.634). No difference was observed in any of the evaluated secondary efficacy endpoints for trimodulin vs placebo in the FAS or the PPS (Table 2 and additional file 2: Table S4). For other secondary efficacy endpoints (such as time to clinical deterioration, time to mortality, time to clinical improvement to score = 3 or score = 4), no conclusions could be derived as median time was not reached due to high level of censoring for > 70% of the subjects for various reasons (e.g. no clinical deterioration occurred in > 50% of subjects, worsening occurred before first infusion, death, or other discontinuation before event). Consequently, these data are not shown.

Fig. 2
figure 2

Impact on clinical deterioration and 28-day mortality (overall population [FAS]). A Bar graph represents the proportion of patients achieving the composite endpoint (patients who clinically deteriorated [between Days 6 and 29] plus patients who died [between Days 1 and 29]) and the individual components of this composite endpoint. P values calculated by chi-square test. Error bars denote 95% CIs. B Kaplan–Meier of probability of survival without an event (defined as deterioration or mortality between Days 1 and 29) in the FAS. P value was calculated by log-rank test. CI confidence interval, FAS full analysis set

Full size image
Table 2 Primary and secondary* efficacy endpoints (overall trial population [FAS and PPS])
Full size table

Safety

At least one TEAE occurred in 78.6% and 78.0% of patients in the trimodulin and placebo groups, respectively. TEAEs were most commonly (overall incidence > 10% of patients) reported in the following System Organ Classes: investigations (43 [51.2%] and 38 [46.3%] patients for the trimodulin and placebo groups, respectively) and respiratory, thoracic and mediastinal disorders (28 [33.3%] and 30 [36.6%] patients, respectively). The most commonly reported (> 5% of the patients in either group) TEAEs by the MedDRA preferred term are presented in Table 3. The most commonly reported TEAE by preferred term was electrocardiogram QT prolonged. There were no significant differences in the rate of any TEAEs between the two treatment groups.

Table 3 TEAEs by MedDRA preferred term (frequency of > 5% in either treatment groupa) (SAF)
Full size table

Post hoc analysis

As knowledge on COVID-19 and its disease stages increased during the pandemic and trial conduct, and due to the good safety profile of trimodulin, exploratory post hoc analyses were performed. A relevant proportion of the ESsCOVID trial population was found to have baseline levels of CRP, D-dimer and/or platelets in line with those reported in the literature to be related to progression to critical disease and associated with mortality. The hypothesis was that trimodulin cannot prevent deterioration or mortality in patients with advanced systemic inflammation (Additional file 2: Fig. S2) but may be of benefit earlier in the disease course. Therefore a subpopulation of patients (early systemic inflammation subgroup) was identified to test this hypothesis. These patients had elevated CRP, elevated D-dimer and low platelets, but had not yet reached thresholds indicating advanced inflammation (defined by CRP > 150 mg/L, and/or D-dimer ≥ 3 mg/L and/or platelet counts < 130 × 109/L at baseline).

A total of 47 patients in the trimodulin group and 49 patients in the placebo group met these criteria for early systemic inflammation. Demographics and baseline characteristics of the subgroup were in line with those of the overall population and were balanced between the two treatment groups (Additional file 2: Table S5). However, in the trimodulin group a smaller proportion of patients had a history of heart disease, a smaller proportion were aged > 60 years (although median age was comparable [59.0 years in the trimodulin group vs 61.0 in the placebo group]) and a higher proportion were male. A total of 40 (85.1%) patients in the trimodulin group and 35 (71.4%) in the placebo group received corticosteroids (not shown). In addition to CRP, D-dimer and platelet counts, baseline levels of other inflammatory markers among patients in this subgroup were generally consistent with early systemic inflammation (Additional file 2: Table S6).

Post hoc efficacy analysis

In the subgroup with early systemic inflammation analysed in the FAS, the deterioration/mortality rate was 21.3% (10/47) in the trimodulin group and 36.7% (18/49) in the placebo group (Table 4), a difference of 15.5 percentage points (95% CI: − 4.46, 34.78; P = 0.096) in favour of trimodulin (Fig. 3A). A treatment difference was seen for both the clinical deterioration and mortality components, and probability of event-free survival was higher with trimodulin (P = 0.017) (Fig. 3B).

Table 4 Primary and secondary efficacy endpoints (early systemic inflammation subgroup [FAS and PPS])
Full size table
Fig. 3
figure 3

Impact on clinical deterioration and 28-day mortality (early systemic inflammation subgroup [FAS]). Post hoc analysis of the primary endpoint in the subgroup of patients with early systemic inflammation who had not yet reached the advanced-stage thresholds (C-reactive protein > 150 mg/L and/or D-dimer ≥ 3 mg/L and/or platelet count < 130 × 109/L). A Bar graph represents the proportion of patients that clinically deteriorated (between Days 6 and 29) plus those that died (between Days 1 and 29) and the individual components of this endpoint. P values calculated by chi-square test. Error bars denote 95% CI. B Kaplan–Meier plot showing the probability of survival with an event (defined as deterioration or mortality between Days 1 and 29). P value was calculated by log-rank test. CI confidence interval, FAS full analysis set, n number of patients

Full size image

For analysed secondary efficacy endpoints, no significant difference was observed in the early systemic inflammation subgroup (FAS) with respect to the mean number of days on IMV, the mean number of ICU-free days, or the proportion of patients recovered on Day 29 (Table 4). Results of primary and secondary efficacy endpoints were aligned but were more pronounced in favour of trimodulin in the PPS (Table 4).

A stacked probability analysis demonstrated that, compared with placebo, more patients in the trimodulin group were discharged and fewer patients were in hospital at Day 29. Furthermore, fewer patients deteriorated or died, most markedly between Days 5 and 10, with trimodulin compared with placebo (Additional file 2: Fig. S3). The differences between trimodulin and placebo were more pronounced in the PPS compared with the FAS.

The impact of additional parameters (e.g. corticosteroid use, duration of hospitalisation ≤ 10 days before treatment start, HFO at treatment start) on the composite primary endpoint in the different subgroups is shown as a forest plot (Additional file 2: Fig. S4). Results for all subgroups confirm the impact of trimodulin in the ‘early’ FAS and PPS populations.

Post hoc safety analysis

TEAEs occurred numerically more frequently in patients with early systemic inflammation receiving placebo (83.7%) than in those receiving trimodulin (68.1%) (Additional file 2: Table S7).

Discussion

To prepare for future pandemics, efforts should still be made to identify and understand the mechanisms of action of new effective treatments that target symptoms in hospitalised COVID-19 patients. This includes identifying the characteristics of patients who responded or did not respond to a specific treatment. In the ESsCOVID phase II clinical trial, there was no difference between trimodulin and placebo for the proportion of patients meeting the composite primary endpoint of clinical deterioration and 28-day all-cause mortality and no differences with respect to pre-planned secondary endpoints. However, rates of deterioration and mortality were markedly lower with trimodulin in a subgroup of patients with early systemic inflammation in which patients were excluded with high CRP (> 150 mg/L) and/or D-dimer (≥ 3 mg/L) and/or low platelet counts (< 130 × 109/L) at baseline.

The ESsCOVID patient cohort received oxygen supplementation and showed signs of inflammation indicated by elevated CRP and the rationale for investigating trimodulin in these patients was based on its postulated modes of action, including the modulation of dysregulated inflammatory processes to prevent further tissue damage [32, 33]. Trimodulin has multiple immunomodulatory activities through IgM, IgA and IgG, including balancing the complement system, modulating cytokine secretion and modulating monocyte and lymphocyte responses [13, 16,17,18, 32, 33]. By targeting multiple pathways, trimodulin may have broader host − immune supporting functions against COVID-19 than currently approved treatments targeting a single deregulated pathway. This is supported by preclinical studies demonstrating stronger immune modulation on immune cells and complement by trimodulin compared with IVIg, suggesting that the interplay between IgM, IgA, IgG and the immune system may promote more extensive beneficial effects in the body [13, 17, 18].

In addition, trimodulin could be beneficial in patients with COVID-19 who develop secondary infections. Previous studies in severe bacterial infectious diseases have shown that targeting the pathogen with antibiotics alone is not always sufficient (particularly in the case of multi-drug resistant germs or immune compromised patients) and that immunoglobulins provide an additive effect [34,35,36,37,38]. Thus, the additional targeting of pathogens with the IgM/IgA-enriched Ig preparation trimodulin may improve outcomes in patients with such infections [22, 39,40,41].

As discussed above, identifying which patients did not benefit, and understanding why, is important for further development of trimodulin and to identify new effective treatments. In this trial, COVID-19 was largely defined as ‘severe’ based on need for oxygen supplementation via NIV or HFO. The aim was to exclude patients in whom infection- and inflammation-related lung damage had progressed too far. Based on its modes of action, trimodulin may prevent tissue damage but is not expected to be of benefit if disease is too far advanced. It was assumed that such an advanced disease stage would primarily apply to patients already requiring IMV. However, despite exclusion of patients on IMV, approximately one-third of patients had advanced systemic inflammation at baseline (CRP > 150 mg/L, and/or D-dimer ≥ 3 mg/L, and/or platelet count < 130 × 109/L) and elevated levels of other immunological parameters (e.g. ferritin, interleukin [IL]-6 and IL-10). The neutrophil to lymphocyte ratio (NLR) was high, consistent with that shown previously to be associated with poor outcomes in COVID-19 patients [42,43,44,45,46,47]. Patients were thus likely to have progressed beyond the stage of disease where Igs could be expected to have a positive effect. This could indicate that systemic processes like inflammation, vasculitis and coagulopathy had progressed, although patients were not (yet) receiving IMV (Additional file 2: Fig. S2).

Despite lack of efficacy in the overall population, the overall safety profile of trimodulin in this trial was good and consistent with the previously known potential and identified risks of trimodulin in other trials [22] and with other IVIgs, as reported in the summaries of product characteristics. It is interesting to note the higher incidence of pulmonary embolism and acute kidney injury in patients treated with placebo vs trimodulin, given the vigilance around these effects in patients treated with Igs [48]. The development of lymphopenia in both groups was not surprising and can be explained by the underlying COVID-19 infection, presence of concomitant diseases, or by the use of certain comedications. Rates were not significantly different between groups and were not considered by the investigators to be treatment related. All events (except one moderate event in the placebo group) were mild in severity. The degree of lymphopenia has previously been shown to correlate with clinical severity of COVID-19 [49, 50]. In line with this, in the present study, fewer patients (12.5% [12/96]) in the early systemic inflammation group than in the more advanced inflammation group (18.6% [13/70]) had lymphopenia.

The good safety profile, together with the observation that a large proportion of patients were already in an advanced disease stage, provided the rationale for the post hoc analysis. The observed reduction in rates of deterioration/mortality is in accordance with data from studies with another IgM/IgA-enriched preparation, Pentaglobin (12% IgM, 12% IgA and 76% IgG), in patients with severe COVID-19 [51,52,53]. For example, in a retrospective cohort study in severe and critically ill COVID-19 patients receiving  ≥ 15 g Pentaglobin for at least 3 days, a subgroup of less advanced patients, not yet receiving IMV, showed most benefit from this treatment [53]. Although data regarding use of IVIg in COVID-19 have been controversial, this has aided identification of the critical parameters for IVIg treatment in COVID-19: timing (selecting disease stage) and dose. From the different analyses it became increasingly clear that patients benefited most if treatment was started early, before IMV, and a high dose was administered [54,55,56,57,58,59]. One meta-analysis performed in non-severe, severe, and critical subgroups based on WHO definitions as used in our trial, have supported the idea that the efficacy of IVIg seems to be associated with the severity of COVID-19 [60]. However, this result was not confirmed in other more recent meta-analyses including more studies [61, 62], suggesting that the dose and type of Ig may well play a role.

Thus, if administered in a timely manner, trimodulin may interfere with several pathological processes that could otherwise lead to respiratory failure, sepsis, multi-organ failure and death. The lower rate of TEAEs observed with trimodulin compared with placebo in patients with early systemic inflammation seems to support this idea. The higher rate of TEAEs observed in the placebo group could largely be the result of disease progression or respiratory sequelae, which was prevented in patients receiving trimodulin.

This trial had some limitations. Firstly, the inflammatory markers defining early systemic inflammation and their corresponding cut-off levels in the post hoc analysis were not pre-specified in the clinical trial protocol, and measurements were not conducted in a central laboratory. Measurement of D-dimer is known to differ between institutions [63] and this could have led to differences in designating patients as having early systemic inflammation, although deterioration/mortality rates in the early systemic inflammation subgroup did not differ much if a threshold of 2, 3 or 4 mg/L D-dimer was used. Secondly, SoC for severe COVID-19 differed between participating sites in the various countries. Since SoC was required to be in line with local guidelines and recommendations, and guidelines changed as the therapeutic landscape for the management of COVID-19 evolved, patients received a range of different therapeutic agents throughout the trial (between October 2020 and June 2021). These medications could have had different effects on patient outcomes and thus may have had an impact on the outcome of this trial. Nevertheless, the aim of the trial was to investigate the use of trimodulin as an adjunct to SoC, and as SoC also differs between countries in routine practice, these results are a close representation of the real-world situation.

Conclusions

In this trial, treatment with trimodulin plus SoC did not result in a significantly lower rate of deterioration/mortality compared with placebo plus SoC in the overall population of patients with severe COVID-19 receiving NIV or HFO. However, the favourable effects observed for trimodulin in a subgroup of hospitalised patients with early systemic inflammation warrant further investigation. Indeed, these findings have informed the design of the ongoing phase III trial of trimodulin in patients with community-acquired pneumonia including COVID-19 pneumonia (TRICOVID trial, NCT05531149). Whereas trimodulin used in this trial was prepared from healthy donors with no exposure to SARS-CoV-2, batches for the phase III trial were developed from donors with an increased anti-SARS-CoV-2 titre. It would therefore be reasonable to predict that treatment with trimodulin would result in lower rates of clinical deterioration/mortality than reported in the current study due to additional anti-SARS-CoV-2 activity. If corroborated, targeted therapy with trimodulin for hospitalised COVID-19 patients based on defined thresholds for markers of inflammation and coagulation may offer a new treatment option.