Outpatient COVID-19 convalescent plasma recipient antibody thresholds correlated to reduced hospitalizations within a randomized trial

Abstract

BACKGROUNDCOVID-19 convalescent plasma (CCP) virus-specific antibody levels that translate into recipient posttransfusion antibody levels sufficient to prevent disease progression are not defined.METHODSThis secondary analysis correlated donor and recipient antibody levels to hospitalization risk among unvaccinated, seronegative CCP recipients within the outpatient, double-blind, randomized clinical trial that compared CCP to control plasma. The majority of COVID-19 CCP arm hospitalizations (15/17, 88%) occurred in this unvaccinated, seronegative subgroup. A functional cutoff to delineate recipient high versus low posttransfusion antibody levels was established by 2 methods: (i) analyzing virus neutralization-equivalent anti-Spike receptor-binding domain immunoglobulin G (anti-S-RBD IgG) responses in donors or (ii) receiver operating characteristic (ROC) curve analysis.RESULTSSARS-CoV-2 anti-S-RBD IgG antibody was volume diluted 21.3-fold into posttransfusion seronegative recipients from matched donor units. Virus-specific antibody delivered was approximately 1.2 mg. The high-antibody recipients transfused early (symptom onset within 5 days) had no hospitalizations. A CCP-recipient analysis for antibody thresholds correlated to reduced hospitalizations found a statistical significant association between early transfusion and high antibodies versus all other CCP recipients (or control plasma), with antibody cutoffs established by both methods-donor-based virus neutralization cutoffs in posttransfusion recipients (0/85 [0%] versus 15/276 [5.6%]; P = 0.03) or ROC-based cutoff (0/94 [0%] versus 15/267 [5.4%]; P = 0.01).CONCLUSIONIn unvaccinated, seronegative CCP recipients, early transfusion of plasma units in the upper 30% of study donors' antibody levels reduced outpatient hospitalizations. High antibody level plasma units, given early, should be reserved for therapeutic use.TRIAL REGISTRATIONClinicalTrials.gov NCT04373460.FUNDINGDepartment of Defense (W911QY2090012); Defense Health Agency; Bloomberg Philanthropies; the State of Maryland; NIH (3R01AI152078-01S1, U24TR001609-S3, 1K23HL151826NIH); the Mental Wellness Foundation; the Moriah Fund; Octapharma; the Healthnetwork Foundation; the Shear Family Foundation; the NorthShore Research Institute; and the Rice Foundation.

Keywords: COVID-19; Immunoglobulins; Immunotherapy.

Figure 1. CONSORT diagram depicting enrollment, allocation, and analytical flow of recipients.

Figure 2. CCP donor neutralizing antibody and anti–S-RBD levels establish a functional cutoff associated with hospitalization protection in screened seronegative recipients.

(A) Here, we use the 1:40 dilutional titer for the neutralizing antibody (nAb) to identify the upper limit of donor anti–S-RBD IgG 2,728 AUC associated with protection from hospitalization. A dilutional titer of 1:10 is the limit of detection for the nAb. (B) The ratio of matched donor anti–S-RBD IgG AUC to that of their respective CCP seronegative recipients that was used to infer the functional cutoff in recipients was determined to be 21.3. Red dots correspond to those hospitalized and black dots are those not hospitalized. (C) Anti–S-RBD IgG AUC levels among donors and posttransfusion recipients segregated by screen vaccination status and serostatus compared by Kruskall-Wallis with Dunn’s post hoc correction. **P < 0.002, ***P < 0.001. Unvaccinated subsequently hospitalized (red dots) posttransfusion recipients in screen seronegative (n = 15) and screen seropositive (n = 2). Black dots are donors and blue dots are nonhospitalized participants. (D) Screen seronegative, unvaccinated recipient D0 (posttransfusion) antibody (n = 361) segregated by recipient days from symptom onset to transfusion and high (>128 AUC) or low (≤128 AUC) anti–S-RBD IgG levels. Recipient high and low cutoffs were calculated using a 21.3-fold drop from donor anti–S-RBD AUC (upper value of the 95% confidence interval) at a 1:40 nAb titer associated with protection. Subsequently hospitalized (red dots) and nonhospitalized (blue dots) recipients are shown. The n values and percentages in each quadrant are the proportion hospitalized among quadrant total. (E) Predicted probabilities of hospitalization across early versus late and high (>128 AUC) versus low (≤128 AUC) anti–S-RBD IgG categories of screen seronegative, unvaccinated CCP recipients were compared using Firth’s logistic regression model adjusted for age, sex, BMI, and variant. P values that the predicted probability is greater than 0% (horizontal dashed line) are shown, with P < 0.05 considered significant.

Figure 3. CCP recipient D0 posttransfusion and matched donor antibody levels stratified by duration from symptom onset to transfusion using cutoffs established by the ROC and maximum antibody threshold method.

(A) RCDCs illustrating antibody distribution of early and late CCP recipients and placebo controls, and thresholds (red dashed lines), established by the maximum antibody that best distinguished hospitalized from nonhospitalized cases. Early recipients are delineated at 2.06 log₁₀(anti–S-RBD AUC) (115 AUC), while late recipients are delineated at 2.58 log₁₀(anti–S-RBD AUC) (380.2 AUC). Curves exclude 5 early participants and 1 late control participant whose posttransfusion plasma was not available. The n values shown are CCP recipients (n = 171 early, n = 190 late) plus placebo recipients (n = 161 early, n = 200 late). (B) Screen seronegative, unvaccinated recipient D0 posttransfusion antibody (n = 361) segregated by early versus late administration assessed as days from symptom onset to transfusion and high versus low antibody using early/late stratum-specific cutoffs established by the maximum antibody that best distinguished hospitalized from nonhospitalized cases. Subsequently hospitalized (red) and nonhospitalized (blue) recipients are shown. The n values and percentages in each quadrant are the proportion hospitalized among quadrant total. (C) Predicted probabilities of hospitalization across early versus late and high versus low categories among screen seronegative, unvaccinated CCP recipients estimated using Firth’s logistic regression adjusted for age, sex, BMI, and variant. P values that predicted the probability is greater than 0% (represented by the horizontal dashed line) are shown for each category, with P < 0.05 considered significant.

Figure 4. Antibody levels over 3 months after transfusion by hospitalization status and treatment group for screen seronegative, unvaccinated recipients.

Log₁₀-transformed antibody levels up to 90 days after transfusion were segregated by treatment and hospitalization status of recipients using a linear mixed-effects regression model, adjusted for variant, age, sex, and BMI. CCP recipients have greater AUC levels on D0, but by D14, the hospitalized recipients have greater AUC levels than nonhospitalized. The average time from transfusion to hospitalization was 3.05 days, with all posttransfusion hospitalizations occurring between D0 and D14. The dashed line represents the log-transformed cutoff (1.924 AUC) for seropositivity. This diagnostic threshold is equivalent to the anti–S-RBD IgG log₁₀-transformed 180 titer.