Early, very high-titre convalescent plasma therapy in clinically vulnerable individuals with mild COVID-19: an international, randomised, open-label trial

Abstract

Background: COVID-19 convalescent plasma (CCP) is a treatment option for COVID-19. This study investigated the safety and efficacy of early, very high-titre CCP in immunocompromised individuals with mild COVID-19.

Methods: This randomised, controlled, open-label trial assessed CCP in immunocompromised patients (n = 120) with mild COVID-19 in 10 clinical trial centres across Germany, France, and the Netherlands. Patients were randomised 1:1 to receive either standard of care (SoC) alone (SoC group) or SoC and 2 units of CCP. Most patients (89.7%) had received ≥3 SARS-CoV-2 vaccinations. The primary endpoint was hospitalisation for progressive COVID-19 symptoms or death by day 28 after randomisation, analysed on a modified intention-to-treat basis (117 patients). The safety analysis included the full analysis set. The trial is registered with EudraCT 2021-006621-22, and ClinicalTrials.gov, NCT05271929.

Findings: Between April 11, 2022 and November 27, 2023, 120 patients were enrolled. Patients in the CCP group received a median of 559 ml CCP from convalescent, vaccinated donors with very high levels of SARS-CoV-2 antibodies (median 81,810 IU/ml) at a median 4 days after symptom onset. The primary outcome occurred in 5/58 patients (8.6%) in the SoC group and in 0/59 patients (0%) in the CCP group, difference -8.6% (95% confidence interval of difference -19% to -0.80%; p-value 0.027; Fisher's exact test). The course of SARS-CoV-2 antibodies in the patients demonstrated a passive transfer of antibodies by the CCP, in particular neutralising effects against new SARS-CoV-2 variants. Whole genome sequencing of SARS-CoV-2 in patients during follow-up showed significant intra-host viral evolution, but without differences between groups. CCP was well tolerated.

Interpretation: Early administration of high-titre CCP can prevent hospitalisation or death in immunocompromised patients with mild COVID-19.

Funding: Support-e project (European Union's Horizon 2020 Programme), German Federal Ministry of Education and Research, ZonMw, the Netherlands Organisation for Health Research and Development.

Keywords: COVID-19; Convalescent plasma; Neutralising antibody; Randomised trial; SARS-CoV-2.

Fig. 1

Trial profile. One patient was randomised by error. The patient in the SoC group was already hospitalised for COVID-19. Two patients in the CCP group withdrew consent, one patient before and one after administration of CCP.

Fig. 2

Forest plot of absolute risk differences in primary outcome of the mITT group (a) and subgroups (b).

Fig. 3

Change in serum SARS-CoV-2 antibody concentration and neutralisation capacity of patients through day 28. Change of the anti-SARS-CoV-2 antibody concentration (panels a–d) and change of the neutralising capacity (panels e–h) in serum of patients on follow up visit 1 (FU1, day 3; red symbols), visit 2 (FU2, day 14; blue symbols), and visit 3 (FU3, day 28; green symbols) compared to baseline measured by anti-SARS-CoV-2-QuantiVac-ELISA (IgG) (a), Elecsys Anti-SARS-CoV-2 S (b) or Genscript Surrogate Neutralisation Test against wild-type (c) or Omicron (d) and neutralising titre (50% inhibition of pseudovirus; NT50) measured against BA.2 (e), BA.5 (f), XBB.1.5 (g) or BQ.1.1 (h). Results are shown for SoC group (light symbols) and the CCP group (bold symbols). For analysis of the course of the anti-SARS-CoV-2 antibodies in the patients, the difference between the antibody concentrations at the follow-up examinations (follow-up 1, day 3; follow-up 2, day 14; and follow-up 3, day 28) and the concentration before the start of therapy (baseline) was calculated for each individual patient and each follow-up time point (Δ from baseline values). Positive values indicate an increase of the antibody concentration or NT50, negative values a decrease of antibody concentration or NT50 compared to baseline. Horizontal lines indicate the median and error bars the interquartile range. The median baseline levels of the SoC and CCP groups in the QuantiVac ELISA were 163.6 BAU/ml and 138.3 BAU/ml (n.s.) (see Appendix Figure S5a), in the Elecsys-Assay 771.0 and 540.0 IU/ml (n.s.) (see Appendix Figure S5b), in the GenScript Assay against wild type 75.9% and 69.3% (n.s.) (see Appendix Figure S5c) and in the GenScript Assay against omicron 2.2% and 4.0% (n.s.) (see Appendix Figure S5d). The mean baseline NT50 titres in the SoC and CCP groups against BA.2 were 76.3 and 93.8 (n.s.) (see Appendix Figure S5e), against BA.5 were 43.5 and 34.8 (n.s.) (see Appendix Figure S5f), against XBB.1.5 were 10 and 10 (n.s.) (see Appendix Figure S5g), and against BQ.1.1 were 10 and 10 (n.s.) (see Appendix Figure S5h). Geometric means of the change of SARS-CoV-2 antibody concentrations and neutralisation capacity are summarised in Appendix Table S8. For each follow visit, the Δ from baseline between the SoC group and the CCP group was compared by Kruskal–Wallis test followed by Dunn's test for correction of multiple comparisons. The p-values for the pairwise comparisons are p > 0.05 (ns; not significant), ∗p < 0.05, ∗∗p < 0.01, ∗∗∗∗p < 0.0001.

Fig. 4

Sequencing of SARS-CoV-2 virus from nasopharyngeal specimens of patients. Sequencing was performed with Oxford Nanopore. Sequences obtained on follow up visit 1 (FU1, day 3), visit 2 (FU2, day 14), and visit 3 (FU3, day 28) were compared to baseline (D1). Sequence coverage of SARS-CoV-2 genome from nasopharyngeal swabs in the SoC group and the CCP group were compared. Horizontal lines indicate the mean and error bars the 95% confidence interval of the mean. Mean values were compared between the SoC and the CCP group by Kruskal–Wallis test followed by Dunn's test for correction of multiple comparisons. The p-values for all pairwise comparisons were p > 0.05 (not significant) (a). Panel (b) indicate the count of new acquired mutations compared to baseline (D1). Horizontal lines indicate the mean and error bars the 95% confidence interval of the mean. Mean values were compared between the SoC and the CCP group by Kruskal–Wallis test followed by Dunn's test for correction of multiple comparisons. The p-values for all pairwise comparisons were p > 0.05 (not significant). (c) The proportion of patients with still detectable SARS-CoV-2 and new mutations in the SoC and CCP group. P-values for comparison of treatment groups were calculated by Fisher's exact test and were not significant (p > 0.05), whereas the difference within treatment group during time was significant (∗p < 0.05). Nucleotide variant evolution of new variants compared to baseline data (D1) for SoC (d) or CCP (e) group is visualised by the frequency of reads with the distinct variant call connected by vertical lines from D1 to FU. (f–h) Proportion of new variants in SARS-CoV-2 genes were calculated for every sample with at least one new variant. (i–k) Frequencies of different mutation types of new variants were calculated for every sample with at least one new variant. (f–k) Means (horizontal line; error bars = 95% confidence interval) were compared between treatment groups SoC and CCP and showed no significant difference calculated by Kruskal–Wallis test followed by Dunn's test for correction of multiple comparisons. (l) For the analysis we only considered samples with at least one new mutation compared to baseline (D1). Among those, we calculated the proportion of mutations localised at sites that already exhibited mutations at baseline (D1) in comparison to the Wuhan reference sequence. Depicted are means with error bars as 95% confidence interval. (m) Variants depicted in L were compared between SoC and CCP treatment and showed no significant difference calculated by Kruskal–Wallis test followed by Dunn's test for correction of multiple comparisons.

Fig. 4

Sequencing of SARS-CoV-2 virus from nasopharyngeal specimens of patients. Sequencing was performed with Oxford Nanopore. Sequences obtained on follow up visit 1 (FU1, day 3), visit 2 (FU2, day 14), and visit 3 (FU3, day 28) were compared to baseline (D1). Sequence coverage of SARS-CoV-2 genome from nasopharyngeal swabs in the SoC group and the CCP group were compared. Horizontal lines indicate the mean and error bars the 95% confidence interval of the mean. Mean values were compared between the SoC and the CCP group by Kruskal–Wallis test followed by Dunn's test for correction of multiple comparisons. The p-values for all pairwise comparisons were p > 0.05 (not significant) (a). Panel (b) indicate the count of new acquired mutations compared to baseline (D1). Horizontal lines indicate the mean and error bars the 95% confidence interval of the mean. Mean values were compared between the SoC and the CCP group by Kruskal–Wallis test followed by Dunn's test for correction of multiple comparisons. The p-values for all pairwise comparisons were p > 0.05 (not significant). (c) The proportion of patients with still detectable SARS-CoV-2 and new mutations in the SoC and CCP group. P-values for comparison of treatment groups were calculated by Fisher's exact test and were not significant (p > 0.05), whereas the difference within treatment group during time was significant (∗p < 0.05). Nucleotide variant evolution of new variants compared to baseline data (D1) for SoC (d) or CCP (e) group is visualised by the frequency of reads with the distinct variant call connected by vertical lines from D1 to FU. (f–h) Proportion of new variants in SARS-CoV-2 genes were calculated for every sample with at least one new variant. (i–k) Frequencies of different mutation types of new variants were calculated for every sample with at least one new variant. (f–k) Means (horizontal line; error bars = 95% confidence interval) were compared between treatment groups SoC and CCP and showed no significant difference calculated by Kruskal–Wallis test followed by Dunn's test for correction of multiple comparisons. (l) For the analysis we only considered samples with at least one new mutation compared to baseline (D1). Among those, we calculated the proportion of mutations localised at sites that already exhibited mutations at baseline (D1) in comparison to the Wuhan reference sequence. Depicted are means with error bars as 95% confidence interval. (m) Variants depicted in L were compared between SoC and CCP treatment and showed no significant difference calculated by Kruskal–Wallis test followed by Dunn's test for correction of multiple comparisons.