Convalescent Plasma for COVID-19. A randomized clinical trial

Background After recovery from COVID-19, most patients have anti-SARS-CoV-2 neutralizing antibodies. Their convalescent plasma could be an inexpensive and widely available treatment for COVID-19. Methods The Convalescent-plasma-for-COVID (ConCOVID) study was a randomized trial comparing convalescent plasma with standard of care therapy in patients hospitalized for COVID-19 in the Netherlands. Patients were randomized 1:1 and received 300ml of plasma with anti-SARS-CoV-2 neutralizing antibody titers of at least 1:80. The primary endpoint was day-60 mortality and key secondary endpoints were hospital stay and WHO 8-point disease severity scale improvement on day 15. Results The trial was halted prematurely after 86 patients were enrolled. Although symptomatic for only 10 days (IQR 6-15) at the time of inclusion, 53 of 66 patients tested had anti-SARS-CoV-2 antibodies at baseline. A SARS-CoV-2 plaque reduction neutralization test showed neutralizing antibodies in 44 of the 56 (79%) patients tested with median titers comparable to the 115 donors (1:160 vs 1:160, p=0.40). These observations caused concerns about the potential benefit of convalescent plasma in the study population and after discussion with the data safety monitoring board, the study was discontinued. No difference in mortality (p=0.95), hospital stay (p=0.68) or day-15 disease severity (p=0.58) was observed between plasma treated patients and patients on standard of care. Conclusion Most COVID-19 patients already have high neutralizing antibody titers at hospital admission. Screening for antibodies and prioritizing convalescent plasma to risk groups with recent symptom onset will be key to identify patients that may benefit from convalescent plasma. Clinicaltrials.gov: NCT04342182


Structured abstract for full paper Background
After recovery from COVID-19, most patients have anti-SARS-CoV-2 neutralizing antibodies.
Their convalescent plasma could be an inexpensive and widely available treatment for COVID- 19.

Methods
The Convalescent-plasma-for-COVID (ConCOVID) study was a randomized trial comparing convalescent plasma with standard of care therapy in patients hospitalized for COVID-19 in the Netherlands. Patients were randomized 1:1 and received 300ml of plasma with anti-SARS-CoV-2 neutralizing antibody titers of at least 1:80. The primary endpoint was day-60 mortality and key secondary endpoints were hospital stay and WHO 8-point disease severity scale improvement on day 15.

Results
The trial was halted prematurely after 86 patients were enrolled. Although symptomatic for only 10 days (IQR 6-15) at the time of inclusion, 53 of 66 patients tested had anti-SARS-CoV-2 antibodies at baseline. A SARS-CoV-2 plaque reduction neutralization test showed neutralizing antibodies in 44 of the 56 (79%) patients tested with median titers comparable to the 115 donors (1:160 vs 1:160, p=0.40). These observations caused concerns about the potential benefit of convalescent plasma in the study population and after discussion with the data safety monitoring board, the study was discontinued. No difference in mortality (p=0.95), hospital stay (p=0.68) or day-15 disease severity (p=0.58) was observed between plasma treated patients and patients on standard of care.

Conclusion
Most COVID-19 patients already have high neutralizing antibody titers at hospital admission.
Screening for antibodies and prioritizing convalescent plasma to risk groups with recent symptom onset will be key to identify patients that may benefit from convalescent plasma.

Introduction
Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) puts a tremendous strain on healthcare systems. A recently completed clinical trial demonstrated that anti-inflammatory therapy with dexamethasone significantly decreases overall mortality. 1 Antivirals as lopinavirritonavir failed to show survival benefit 2 , but a more recent randomized trial showed a shortened time to clinical recovery in patients treated with remdesivir and comparable results were observed in a trial on interferon beta-1b, lopinavir-ritonavir, and ribavirin. However, it remains to be seen if any of these antiviral therapies will decrease mortality. 3,4 Also, these drugs are not widely available and the rapid distribution to hospitals across the world is extremely challenging. 1 Therefore, other readily available, affordable and effective antiviral therapies are needed.
Convalescent plasma (ConvP), which may contain high levels of virus neutralizing antibodies, could be an alternative treatment option for SARS-CoV-2 patients. A similar strategy has been pursued during the 2003 SARS and later MERS outbreaks. 5 Conclusive evidence for the effectivity of ConvP as a treatment for human coronavirus infections has yet not been documented in large randomized clinical trials. Preclinical research however indicated a protective effect of human ConvP when given to hamsters infected with SARS-CoV-2 early in the disease course. 6 Although large volumes of ConvP can have indirect effects as well, we assume that key to the efficacy of ConvP through direct antiviral effect may be the presence of high titers of virus neutralizing antibodies. Following this rationale, benefit can only be expected if it is administered to viremic patients with little or no autologous neutralizing antibodies. Although ConvP seems to be safe, no overall clinical benefit of ConvP therapy was observed in a prematurely interrupted randomized trial from China. 7,8 In this study, patients had been symptomatic for 30 days on average when they received ConvP which contrasts sharply with the time from symptom onset to hospital admission in cohort studies from China (11 days) and Europe (13 days). 9, 10 The recent observations showing that close to 100% of patients have detectable neutralizing antibodies three weeks after symptom onset may explain the lack of a therapeutic effect observed. 11,12 All other evidence on the possible efficacy of ConvP comes from uncontrolled case series or with historical patients as comparator. [13][14][15] Despite the fact that in the USA already 8932 patients had received ConvP as part of the early access program on the 11 th of May 2020, it remains unknown if ConvP will . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 3, 2020. . https://doi.org/10.1101/2020.07.01.20139857 doi: medRxiv preprint be beneficial if administered early on in the COVID-19 disease course. 8 The ConCOVID study was setup in centers across the Netherlands in order to determine the effect of ConvP on mortality in COVID-19 patients early after hospital admission.

Study design
The ConCOVID study was designed as a nationwide multicenter open-label randomized clinical trial. The trial network includes 14 secondary and academic hospitals in the Netherlands.

Intervention, primary and secondary endpoints
Patients were randomly assigned via a web-based system at a 1:1 ratio to the current standard of care at each hospital with or without the addition of 300ml of ConvP, the standard volume of one plasma unit produced by Sanquin Blood Supply, was administered intravenously on the day of inclusion. Patients without a clinical response and a persistently positive RT-PCR could receive a second plasma unit after five days. Off-label use of EMA-approved drugs (e.g. chloroquine, azithromycin, lopinavir/ritonavir, tocilizumab, anakinra) as a treatment for COVID-19 was allowed in hospitals were this was part of the standard of care. We scored the clinical status with the ordinal 8-point WHO COVID-19 disease severity scale on days 1, 15 and 30. 16 Serum samples and nasopharyngeal swabs were collected at inclusion preceding treatment and thereafter. The primary endpoint of the study was overall mortality until . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 3, 2020. . https://doi.org/10.1101/2020.07.01.20139857 doi: medRxiv preprint discharge from the hospital or a maximum of 60 days after admission whichever came first.
Key secondary clinical endpoint we describe here are the improvement on the 8-point WHO COVID-19 disease severity scale from inclusion to day 15, hospital length of stay and safety.

Assays
We analyzed serum samples of donors and patients for the presence of neutralizing antibodies by performing a PRNT with the SARS-CoV-2 virus (German isolate; GISAID ID EPI_ISL 406862; European Virus Archive Global #026V-03883) as we have described previously. 11 More details are available in the online supplement. Serum was also tested for the presence of anti-SARS-CoV-2 total Ig and IgM with the Wantai Enzyme Linked Immunosorbent Essay (ELISA) test (Wantai Biological, Beijing). We previously showed that a positive total Ig or a IgM with an optical density (OD) ratio >10 (which equals an OD of 2.0), correlates closely with virus neutralizing antibody titers (PRNT50) of at least 1:80. 17

Sample size and statistical analysis plan
With an anticipated 50% overall mortality reduction from 20% in the control arm, which was the reported mortality in hospitalized patients in the Netherlands when the protocol was designed and with a control to intervention ratio of 1:1, 426 patients were needed for the study to have 80% power with a global alpha level of 0.05 and adjusted alpha level for the primary endpoint of 0.0480, accounting for 1 interim analysis. The full statistical analysis plan is available in the online supplement and the full protocol. Due to the premature interruption of the trial and resulting lower event rates we present both the results of the multivariable (adjusted) logistic regression analysis as originally planned and the unadjusted analysis (table 2 and 3 of online supplement).
The study was reviewed and approved by the institutional review board of the Erasmus University Medical Center. Written informed consent was obtained from every patient or a legal patient representative. The DSMB reviewed the safety of the participants on a regular basis and recommended the study team regarding the further conduct of the study at predefined time points. The study was registered as NCT04342182 at clinicaltrials.gov.

Results
. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 3, 2020. To confirm the observation that the majority of patients already had high neutralizing antibody titers at hospital admission, we tested an additional 37 serum samples that had been . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 3, 2020. . https://doi.org/10.1101/2020.07.01.20139857 doi: medRxiv preprint collected within three days after hospital admission from COVID-19 patients admitted on a general ward at Erasmus MC in the weeks preceding the start of the ConCOVID study. With a median age of 65 years (IQR 56 -74), 60% males and symptom duration of 9 days (IQR 4 -13) these patients were comparable to the study population. Here as well we found that 26/37 (70%) of patients had anti-SARS-CoV-2 Ig antibodies and in 23/37 (62%) at a ratio >10, indicating high neutralization capacity.
In a post-hoc analysis we evaluated if ConvP administration accelerated the increase in neutralizing antibody titers over time. For this, the PRNT50 titers on day 7 were compared with the titers on day 1 in the subgroup of patients with a titer <1:160 at baseline. In all 9 patients, including 5 randomized to standard of care, a fourfold increase in titers was observed on day 7 (supplemental figure 1c ).

Efficacy and safety
The adjusted OR for overall mortality for patients treated with ConvP was 0.95 (CI 0.20 -4.67., Altogether, the observations we made on antibody titers in patients and donors convinced us that a complete redesign of the study was needed and could not be resolved with a substantial study amendment. Indeed, we do not anticipate clinical benefit from ConvP for patients with high titers of autologous neutralizing antibodies present at baseline. The results were discussed with the DSMB on the 10 th of June and the decision was made to end the study under its current design.
. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

Discussion
The majority of the patients in the ConCOVID study already had high virus neutralizing antibodies titers on the day of study inclusion with titers comparable to the 115 recovered donors we screened for ConvP plasma donation. When the study was designed, the timing of neutralizing antibody development after SARS-CoV-2 infection was uncertain and we considered it unlikely that the most patients would have autologous neutralizing antibodies on the day of hospital admission. However, the data that became available to us after the inclusion of 86 patients made it very unlikely that the overall study population would benefit from ConvP therapy without a change in the patient recruitment strategy. Indeed, without a redesign of the study it would be substantially underpowered also after the enrollment of 426 patients and after a meeting with the DSMB the decision was made to end the study. No statistically significant differences in mortality (aOR 0.95, CI 0.20 -4.67, p=0.95) or improvement in the day-15 disease severity (aOR 1.30, CI 0.52 -3.32, p=0.58) was observed when the study was suspended.
We think that the observations we made are relevant for almost all ongoing studies on ConvP as a treatment for COVID-19. Indeed, all but few of these trials are focusing on hospitalized patients and the time from disease onset to admission was repeatedly shown to be comparable to the 10 days in our study. To the best of our knowledge, none of the ongoing studies is screening patients in real-time for the presence of antibodies before inclusion. 9,10 In relation to the plasma donors that we selected, virtually all had anti-SARS-CoV-2 antibodies but only 41% had high neutralizing titers of at least 1:320. This could be related to generally milder disease course of the donors we recruited 12,18 ) Although, the level of antibodies required to ascertain an antiviral effect remains to be established a certain minimum level will be needed because after administration, the antibodies will be diluted at least 10-fold when administered to an adult patient. However, many of the ongoing trials are not directly testing the neutralizing capacity of donor plasma (which is the gold standard in coronavirus serology), but rather rely on a positive anti-SARS-CoV-2 ELISA or do not test for antibodies at all. Recent observations from our laboratory indicate that a positive Wantai SARS-CoV-2 Ig ELISA with an OD ratio ≥10 correlates well with a PRNT50 titer of at least 1:80. 17 This could already help to reliably exclude donors with low neutralizing antibody when PRNT is unavailable, although it . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 3, 2020. . https://doi.org/10.1101/2020.07.01.20139857 doi: medRxiv preprint will not guarantee high PRNT50 titers. A solution may be to actively recruit recovered patients who had more severe COVID-19 disease as donor, as they have been shown to have higher antibody levels. 11,17 Also, the use hyperimmune Ig preparations produced from pooled convalescent plasma (now called COVIg) or the use of specific highly neutralizing antibodies may solve this issue. 19,20 This study has limitations. First, the premature ending of the study prevents definite conclusions regarding the clinical benefit of ConvP. Fortunately, a collaboration between several research groups evaluating ConvP is being formed and will allow for a pre-planned meta-analysis on pooled data from clinical trials. Our data do however show that ConvP should be studied earlier in the disease course. This could mean in the outpatient setting where ConvP could be evaluated in patients with a higher likelihood of disease progression based on clinical (e.g. age, comorbidities) or other (e.g. CRP, LDH) characteristics. Our data also show that in hospitalized patients testing for the presence of antibodies prior to ConvP should be part of the protocol, and stratifying or even excluding patients based on a positive antibody test will be needed. With the large variety of serological assays that have come available, it is important to carefully validate the assays prior to use and ideally correlate the assay to gold standard virus neutralization assays. Second, ConvP may have an effect that is unrelated to the neutralizing antibodies because therapy with intravenous immunoglobulines or plasma can have diverse anti-inflammatory and immunomodulatory effects as well. However, we considered any such effect unlikely because doses of Ig therapy when used as an immunomodulatory agent are typically at least 10-fold higher than the quantity of immunoglobulins in 300ml plasma. Finally, the recently completed dexamethasone arm of the recovery trial demonstrated an improved overall survival with dexamethasone therapy in patients requiring supplemental oxygen therapy. 1 This will almost certainly change the standard of care therapy of COVID-19 and needs to be incorporated in the design of ongoing and future ConvP studies as well.
In conclusion, the majority of patients in the ConCOVID study already had high titers of virus neutralizing antibodies upon enrollment in the study. This made the design of the study unsuitable for its purpose of evaluating the clinical value of ConvP. This observation should . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 3, 2020. . https://doi.org/10.1101/2020.07.01.20139857 doi: medRxiv preprint trigger investigators to reconsider the design of current studies on ConvP for the treatment of patients with COVID-19. Median and inter quartile ranges are given. NA=Not applicable (1) Due to the fact that HLA and HNA-antibody typing which is required for female donors takes time, only plasma from male donors was used during the first 6 weeks of the study.
(2) WHO 8 point COVID-19 disease severity score (at study inclusion for patients and highest score ever during disease course for donors) in which 0 is no clinical or virological evidence of infection, 1 is no limitation of activities, 2 is limitation of activities, 3 is hospitalized, no oxygen, 4 is oxygen by mask or nasal prongs, 5 is noninvasive ventilation or high-flow oxygen, 6 is intubation and mechanical ventilation, 7 is ventilation and additional organ support (vasopressors, renal replacement therapy, ECMO) and 8 is death.
(3) Titer at inclusion for patients (determined in 56 patients) and at the time of plasma donation in donors. (4) 100 out of 115 donors completed the questionnaire. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 3, 2020. . https://doi.org/10.1101/2020.07.01.20139857 doi: medRxiv preprint Figure 1a. PRNT50 in -donors of whom ConvP was used in the study (grey) and in -study patients at baseline (black). . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 3, 2020. . https://doi.org/10.1101/2020.07.01.20139857 doi: medRxiv preprint CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 3, 2020. . https://doi.org/10.1101/2020.07.01.20139857 doi: medRxiv preprint