key: cord-1032383-b5yrfllu
authors: Liu, S. T. H.; Lin, H.-M.; Baine, I.; Wajnberg, A.; Gumprecht, J. P.; Rahman, F.; Rodriguez, D.; Tandon, P.; Bassily-Marcus, A.; Bander, J.; Sanky, C.; Dupper, A.; Zheng, A.; Altman, D. R.; Chen, B. K.; Krammer, F.; Mendu, D. R.; Firpo-Betancourt, A.; Levin, M. A.; Bagiella, E.; Casadevall, A.; Cordon-Cardo, C.; Jhang, J. S.; Arinsburg, S. A.; Reich, D. L.; Aberg, J. A.; Bouvier, N. M.
title: Convalescent plasma treatment of severe COVID-19: A matched control study
date: 2020-05-22
journal: nan
DOI: 10.1101/2020.05.20.20102236
sha: e22b9ea6bfe840f94e9e78375904cd4022c4061e
doc_id: 1032383
cord_uid: b5yrfllu

Background Since December 2019, Coronavirus Disease 2019 (COVID-19) has become a global pandemic, causing mass morbidity and mortality. Prior studies in other respiratory infections suggest that convalescent plasma transfusion may offer benefit to some patients. Here, the outcomes of thirty-nine hospitalized patients with severe to life-threatening COVID-19 who received convalescent plasma transfusion were compared against a cohort of retrospectively matched controls. Methods Plasma recipients were selected based on supplemental oxygen needs at the time of enrollment and the time elapsed since the onset of symptoms. Recipients were transfused with convalescent plasma from donors with a SARS-CoV-2 (severe acute respiratory disease coronavirus 2) anti-spike antibody titer of [≥]1:320 dilution. Matched control patients were retrospectively identified within the electronic health record database. Supplemental oxygen requirements and survival were compared between plasma recipients and controls. Results Convalescent plasma recipients were more likely than control patients to remain the same or have improvements in their supplemental oxygen requirements by post-transfusion day 14, with an odds ratio of 0.86 (95% CI: 0.75~0.98; p=0.028). Plasma recipients also demonstrated improved survival, compared to control patients (log-rank test: p=0.039). In a covariates-adjusted Cox model, convalescent plasma transfusion improved survival for non-intubated patients (hazard ratio 0.19 (95% CI: 0.05 ~0.72); p=0.015), but not for intubated patients (1.24 (0.33~4.67); p=0.752). Conclusions Convalescent plasma transfusion is a potentially efficacious treatment option for patients hospitalized with COVID-19; however, these data suggest that non-intubated patients may benefit more than those requiring mechanical ventilation.

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is a positive-sense, single-60 stranded RNA virus belonging to the family Coronaviridae. Humans infected with SARS-CoV-2 61 may develop Coronavirus Disease 2019 , which manifests across a wide spectrum of 62 clinical severity ranging from a mild upper respiratory tract illness to a diffuse viral pneumonia 63 causing acute respiratory failure, with sequelae including acute lung injury, multi-organ 64 dysfunction syndrome, and death. 1-3 Antibody responses to coronavirus infections typically 65 appear 2-3 weeks after the onset of illness and are rarely observed earlier. 4-6 66 Although the relationship between disease severity and antibody response has yet to be firmly 67 established, 7 transfusion with convalescent plasma may provide a therapeutic option in the 68 current treatment-limited environment. [8] [9] [10] Historical evidence supports the efficacy of 69 convalescent plasma transfusions to treat a variety of infectious diseases, including influenza, 70

Junin virus, and severe acute respiratory syndrome (SARS). [11] [12] [13] [14] Initial data supporting 71 convalescent plasma transfusions for COVID-19 include three case series from China of 5, 10, and 72 6 patients. [15] [16] [17] In respiratory infections specifically, the strongest evidence suggests that the 73 benefit of passive antibody transfer is most demonstrable in patients who were treated within 74 days of symptom onset. 12, 13, 18, 19 Therefore, we hypothesized that treatment of patients with 75 convalescent plasma early in the disease course may reduce morbidity and mortality associated 76 . 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.

The copyright holder for this preprint this version posted May 22, 2020 May 22, . . https://doi.org/10.1101 May 22, /2020 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.

The copyright holder for this preprint this version posted May 22, 2020. . https://doi.org/10.1101/2020.05.20.20102236 doi: medRxiv preprint 5 volume, was infused over 1 to 2 hours. Recipients were monitored every 15 minutes for signs of 99 transfusion-related reactions and then followed post-transfusion for outcomes. 100

To confirm the independent effect of convalescent plasma transfusion on improvement in 102 oxygenation and survival, we conducted a propensity score-matched analysis using The Mount 103

Sinai Hospital's COVID-19 confirmed patient pool from the same calendar period (24 March 2020 104 to 8 April 2020). A logistic regression was fit to predict the potential for plasma therapy based 105 on time series data obtained at baseline upon admission, prior to transfusion, and the day of 106 transfusion. Among the predictors, exact matching was enforced on the administration of 107 hydroxychloroquine and azithromycin, intubation status and duration, length of hospital stay, 108 and oxygen requirement on the day of transfusion. Other medications were administered too 109 infrequently to enforce exact matching. Balance was well achieved between the plasma and 110 control groups, as all predictors had a standardized mean difference less than 0.2. Details of the 111 matching method and results are described in the Supplementary Appendix. A medical data 112 team reviewed charts of the control patients to determine outcomes at 1, 7, and 14 days. The 113 data team was not informed of the recipient to whom each control patient was matched. Because 114 control patients were matched to plasma recipients by length of stay prior to transfusion, "day 115 0" was defined as the day of transfusion for the plasma recipients and as the corresponding day 116 in the hospitalization course of the control patients. 117

Patients were then evaluated for their supplemental oxygen requirements and survival at three 119 time points: days 1, 7, and 14 post-transfusion. Four categories of supplemental oxygen use 120 status were collected for both cases and controls. These include, in order of increasing severity: 121 . 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.

The copyright holder for this preprint this version posted May 22, 2020. . https://doi.org/10.1101/2020.05.20.20102236 doi: medRxiv preprint 6 room air without supplemental oxygen required; low-flow oxygen delivery by standard nasal 122 cannula; high-flow oxygen delivery, including non-rebreather mask; high-flow nasal cannula or 123 bi-level positive airway pressure (BiPAP) non-invasive ventilation; and mechanical ventilation. A 124 patient's oxygenation status at the three time points was considered to have worsened if they 125 changed from a lower-to a higher-severity category compared to Day 0, or if they had died prior 126 to the time point. A generalized estimating equations (GEE) approach with a logit link for binary 127 data was used to model the effect of plasma on the odds of oxygenation improvement on days 128 1, 7, and 14 following transfusion, controlling for oxygen status on day 0. An independent 129 working correlation structure was assumed for the patients within each cluster; however, the p-130 values were calculated based on the empirical standard errors. Since some patients were 131 discharged with continued oxygen supplementation, the oxygen status of discharged patients 132 was assumed to be no worse than low-flow oxygen by standard nasal cannula. 133

Kaplan-Meier survival curves and the log rank test were used to depict the overall post-135 transfusion survival. A Cox model was fit to estimate the hazard ratio for in-hospital mortality 136 for the plasma group, with matched clusters treated as random effects and onset of intubation 137 as a time-varying covariate. In addition, interactions between convalescent plasma 138 administration and intubation duration were tested to see if the plasma effects were the same 139 in subgroups. 140

Both oxygen status and survival models were adjusted for duration of symptoms prior to 141 admission and drugs administered, as these data were only ascertained after the matching was 142 completed. The initial drug list consisted of COVID-19 therapies used during the time of the 143 . 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.

The copyright holder for this preprint this version posted May 22, 2020. . https://doi.org/10.1101/2020.05.20.20102236 doi: medRxiv preprint 7 study that included azithromycin, broad-spectrum antibiotics, hydroxychloroquine, therapeutic 144 anticoagulants, corticosteroids, directly acting antivirals, stem cells, and interleukin 1 and 145 interleukin 6 inhibitors. Only those that had a p-value < 0.5, however, were included in the final 146 model for adjustment. A liberal p-value was used here to be inclusive of any potential 147 confounders. As a sensitivity analysis, the 1:2 matching without replacement data were also 148 analyzed, where the balance between the matched pairs was enhanced but the study power 149 was reduced. Descriptive data are reported as number (percent), mean (± standard deviation) 150 or median [min, max], as appropriate. Analysis was performed using SAS 9.4 (SAS Institute Inc., 151

Cary, NC). All tests were 2-sided and statistical significance was defined as a p value < 0.05, 152 unless otherwise indicated. 153

Results 154

The average age of the recipients of convalescent plasma transfusion was 55 (± 13) years (Table  156 1). The cohort was approximately two-thirds male and one-third female, similar to the 157 proportions of men and women with severe disease in prior studies. 1 Recipients generally had 158 few baseline co-morbidities: 54% were obese (body mass index ≥30) and 18% had a current or 159 former history of tobacco use. One patient had end-stage renal disease requiring peritoneal 160 dialysis. The median duration of symptoms prior to initial presentation was 7 [0, 14] days. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 22, 2020. . https://doi.org/10.1101/2020.05.20.20102236 doi: medRxiv preprint 8 transfusion. In addition to receiving convalescent plasma transfusion, many recipients received 167 a variety of inpatient pharmacotherapies throughout their hospitalizations (Table 2 ). There were 168 no significant differences between plasma recipients and control patients in exposures to 169 measured pharmacotherapies, except for therapeutic anticoagulation. 7, the plasma group also showed a reduction in the proportion of patients with worsened 180 oxygenation status, but the group difference did not reach statistical significance. 181

As of 1 May 2020, 12.8% of plasma recipients and 24.4% of the 1:4 matched control patients had 183 died (21.6% in the 1:2 matched dataset), and 71.8% and 66.7% (68.9%) had been discharged alive, 184 respectively. The median follow-up time was 11 [1, 28] days for the plasma group and 9 [0, 31] 185 days for the control group. Overall, we observed improved survival for the plasma group (log-186 rank test: p=0.039) (Figure 2) . In a covariates-adjusted Cox model, convalescent plasma 187 transfusion was significantly associated with improved survival in non-intubated patients (hazard 188 ratios: 0.19 (95% CI: 0.05 ~0.72); p=0.015), but not in intubated patients (1.24 (0.33~4.67); 189 . 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.

The copyright holder for this preprint this version posted May 22, 2020. . https://doi.org/10.1101/2020.05.20.20102236 doi: medRxiv preprint 9 p=0.752) (P-value for the plasma and intubation interaction term was 0.050). There is no 190 evidence that the effect of plasma depended on the duration of symptoms (p=0.19 for the plasma 191 by duration interaction). The results remain robust in the model without covariates adjustment 192 and in the 1:2 matched sample (Figure 3) . 193

The COVID-19 pandemic poses an unprecedented challenge, as physicians and scientists struggle 195 in real time to identify effective interventions against SARS-CoV-2 and its complications. This 196 initial assessment offers evidence in support of convalescent plasma transfusion as an effective 197 intervention in COVID-19. Preliminary data suggest a potential mortality benefit, but greater 198 numbers are needed to draw definitive conclusions. Interestingly, these data suggest that the 199 survival effect of convalescent plasma may begin to manifest more than 1 week after transfusion. 200

If this observation is borne out in subsequent studies, it could indicate that convalescent plasma 201 prevents longer-term complications, such as acute lung injury or multi-organ dysfunction 202 syndrome; however, this speculation awaits confirmation in a larger patient cohort. 203

This study has many unique strengths. It describes the largest cohort of COVID-19 patients 204 treated with convalescent plasma thus far worldwide. Furthermore, New York City has a large 205 and very diverse population, and its metropolitan area was among the earliest and hardest hit by 206 the COVID-19 pandemic in the United States. Over this study's 16-day enrollment period (24 207

March 2020 to 8 April 2020), the Mount Sinai Health System admitted 4,152 confirmed COVID-208 19-positive patients. This large pool from which to draw control patients permitted an aggressive 209 matching algorithm. Data from three different time frames --baseline, prior to transfusion, and 210 day of transfusion -informed the matching of controls to cases to maximize their similarity. 211 . 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.

The copyright holder for this preprint this version posted May 22, 2020 . . https://doi.org/10.1101 In addition, the efficacy of passive antibody transfer relies heavily on the quality of the donor 212 convalescent plasma. Mount Sinai rapidly developed and clinically deployed an ELISA to titrate 213 SARS-CoV-2-specific antibodies in serum, 18 enabling our center to refer for blood collection only 214 those convalescent donors with the highest peripheral serum antibody titers of ≥1:320. 22 Prior 215 smaller studies have reported on a variety of titer cutoffs, 15,16 and at the time of this publication 216 some centers are bypassing donor antibody titer pre-collection completely. 17 Although the total 217 quantity of anti-SARS-CoV-2 spike antibodies were assessed, it must be noted that we have not 218 yet assessed the functionality of these antibodies in neutralizing the virus. Recent studies with 219 SARS CoV-2 have generally found a high correlation between ELISA S protein binding activity and 220 neutralization of SARS CoV-2. 21,23 221

Although controls were retrospectively identified by propensity matching, the conclusions drawn 222 from these data are not as robust as a prospective, randomized, placebo-controlled study. 223

Furthermore, the convalescent plasma recipient cohort is highly heterogeneous in regards to 224 oxygen needs at the time of transfusion and the duration of symptoms prior to admission. Other 225 than intubated versus non-intubated patients, the small size of this cohort lacks sufficient power 226 to permit additional subgroup analyses. We did not observe significant benefit of convalescent 227 plasma in intubated patients, consistent with past literature demonstrating that passive antibody 228 transfer therapies are most efficacious early in disease. 12, 13, 18, 19 However, the number of 229 intubated patients in this study is small, limiting our ability to reach any conclusions about this 230 population. Future studies that include more mechanically ventilated patients will be needed to 231 address this uncertainty. 232

No significant transfusion-related morbidity or mortality were observed in the convalescent 233 plasma recipient cohort; however, potential harms are associated with plasma transfusion. There 234 . 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.

The copyright holder for this preprint this version posted May 22, 2020 . . https://doi.org/10.1101 11 is a risk of fluid volume overload, particularly in patients with end-stage renal disease or advanced 235 heart failure. Allergic reactions to plasma are typically mild and self-limited. Plasma naturally 236 contains procoagulants, whose additive effects are unknown in this disease, which is 237 independently associated with hypercoagulability; 24 thus, pending more data, additional caution is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 22, 2020 . . https://doi.org/10.1101 12 Catamero, Gillian Sanchez, Suzan Aird, Manpreet Mann, Tarashon Broome, Sonia Kleiner-Arje, 258

Louise Wolf, Angela Lee, Lisa Gaynes, and Karyn Goodman. We dedicate this work to the New 259

Yorkers who have lost their lives to COVID-19 with a special dedication to the health care 260 workers who will always be remembered for their selflessness during this pandemic. 

Dr. Krammer reports that patent applications have been filed for the assay used to select 274 plasma donors, and Mount Sinai has licensed its use to several companies. Dr. Aberg reports 275 grants and personal fees from Gilead, grants and personal fees from Merck, grants and personal 276 fees from Janssen, personal fees from Theratech, personal fees from Medicure, grants from 277

Regeneron, grants and personal fees from Viiv, outside the submitted work. All other authors 278 have nothing to disclose. 279 . 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 May 22, 2020. 

. 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 May 22, 2020 . . https://doi.org/10.1101 

Presenting Characteristics, Comorbidities

Outcomes Among 5700 Patients Hospitalized With COVID-19 in the New York City Area

Characteristics of and Important Lessons From the Coronavirus Disease 284 2019 (COVID-19) Outbreak in China: Summary of a Report of 72314 Cases From the Chinese Center

Disease Control and Prevention

Clinical course and risk factors for mortality of adult inpatients with 287

The time course of the immune response to 289 experimental coronavirus infection of man

Serologic responses of 42 MERS-coronavirus-infected patients 291 according to the disease severity

Chronological evolution of IgM, IgA, IgG and 293 neutralisation antibodies after infection with SARS-associated coronavirus

Antibody responses to SARS-CoV-2 in patients of novel 296 coronavirus disease 2019

The convalescent sera option for containing COVID-19

Convalescent plasma as a potential therapy for COVID-19

Convalescent Plasma to Treat COVID-19: Possibilities and Challenges

Meta-analysis: convalescent blood products for 305 Spanish influenza pneumonia: a future H5N1 treatment?

The effectiveness of convalescent plasma 307 and hyperimmune immunoglobulin for the treatment of severe acute respiratory infections of viral 308 etiology: a systematic review and exploratory meta-analysis

Hyperimmune IV immunoglobulin treatment: a multicenter 310 double-blind randomized controlled trial for patients with severe 2009 influenza A(H1N1) infection

Importance of dose of neutralising 313 antibodies in treatment of Argentine haemorrhagic fever with immune plasma

Treatment of 5 Critically Ill Patients With COVID-19 With 315 Convalescent Plasma

Effectiveness of convalescent plasma therapy in severe COVID-19 317 patients

Treatment with convalescent plasma for COVID-19 patients in Wuhan

The serum treatment of lobar pneumonia

Remarks on THE SERUM TREATMENT OF PNEUMONIA

SARS-CoV-2 Seroconversion in Humans: A Detailed 323

Antigen Production, and Test Setup

A serological assay to detect SARS-CoV-2

22.An EU programme of COVID-19 convalescent plasma collection and transfusion: Guidance on 328 collection, testing, processing, storage, distribution and monitored use. Brussels: European Commission 329 . 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.

The copyright holder for this preprint this version posted May 22, 2020 . . https://doi.org/10.1101 15 Directorate-General for Health and Food Safety, April 4 2020. 

338 . 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.

The copyright holder for this preprint this version posted May 22, 2020. Sputum production 3 (8). 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.

The copyright holder for this preprint this version posted May 22, 2020. (33) Heart rate >100 beats per min 22 (56) Respiratory rate ≥ 20 breaths per min 28 (72) Imaging -no. (%)Chest radiography 38 (97) Chest computed tomography 3 (8)

Laboratory data prior to transfusion . 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. 

. 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.

The copyright holder for this preprint this version posted May 22, 2020. 

. 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.

The copyright holder for this preprint this version posted May 22, 2020 May 22, . . https://doi.org/10.1101 May 22, /2020 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 May 22, 2020 . . https://doi.org/10.1101