Patient-blood management for COVID19 convalescent plasma therapy: relevance of affinity and donor–recipient differences in concentration of neutralizing antibodies

Abstract

Background

Coronavirus disease 2019 (COVID-19) convalescent plasma (CCP) is being extensively investigated as a treatment, with mixed results to date. Overall, there has been a generalized lack of appropriateness in prescriptions, which, in the field of transfusion medicine, is termed patient-blood management.

Objectives

We aimed to separate study design variables that could affect clinical outcome after CCP therapy. We focus here on variables such as pretransfusion antibody testing in recipients, dose adjustments and antibody affinity measurements.

Sources

We searched PubMed and preprint servers for relevant preclinical and clinical studies discussing each of these variables in the field of CCP therapy.

Content

We show evidence that neglecting those variables has affected the outcomes of the vast majority of CCP clinical trials to date.

Implications

A better understanding of such variables will improve the design of the next generation of CCP clinical trials. This will likely lead to better clinical outcomes and will minimize risks of immune evasion from subneutralizing doses of neutralizing antibodies.

Introduction

The ongoing coronavirus disease 2019 (COVID-19) pandemic had caused more than 150 million cases and 23.2 million deaths as of April 29, 2021. Although immunity largely depends on specific T lymphocytes [1], several therapeutics based on severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) neutralizing antibodies (nAb) are under investigation. COVID-19 convalescent plasma (CCP) is currently under study as a treatment for SARS-CoV-2 infection [2]. To date, clinical benefits have only been formally proved in early disease stages [[3], [4], [5]] (WHO scores 3 to 4 [6]). However, CCP has been extensively used in advanced disease stages (WHO scores 5 to 7), often as a last resort (compassionate) therapy in patients who have been hospitalized for weeks after onset of symptoms and have received multiple lines of treatment. Such practice has been generally approved by local ethics committees at the beginning of the pandemic, but we feel that it is time to test the scientific basis of this practice using the tools of evidence-based medicine. The field of transfusion medicine has pioneered patient-blood management (PBM), and we feel that CCP should also respond with prescription appropriateness. We recently solicited an urgent definition of CCP specifications by regulatory authorities (including, for example, the minimum antibody content, the methods that should be used to quantify such antibody content [7], and the shelf life of the unit) [8]. More points need to be addressed to elevate CCP from alchemy to a modern drug.

How can we define an appropriate dose adjustment?

To date, only a few trials have attempted to adjust the CCP dose, but this has invariably been accomplished by calculating the volume of CCP (mL) per recipient body weight (kg) [5]. Such an approach is legitimate when plasma is transfused to correct deficiencies of clotting factors, but it is completely lacking fundamental pharmacodynamic reasoning if the active ingredient is instead an nAb. It should be kept in mind that viral neutralization follows precise stoichiometry, and that a 200-mL CCP dose is finally diluted about ten-fold in about 2.5 L of plasma within the bloodstream. This means that in order to induce a measurable (two-fold) increase in nAb titres, the reinfused dose of nAb should be at least ten-fold higher than the one measured in the candidate recipient. Although plaque reduction neutralization test (PRNT) remains the reference standard for measuring nAb and can be safely used retrospectively, it takes days to report a measurement: hence, novel PRNT variants or surrogate high-throughput serology with high correlations [7] should be used prospectively when taking decisions on new patients.

How much neutralizing antibody is needed to exert a clinically significant impact on an ongoing SARS-CoV-2 infection?

Several studies have measured the concentration of virions (in genome copies per gram of tissue) in respiratory, digestive and immune system tissues of rhesus macaques, after infection with SARS-CoV-2 [[9], [10], [11]]. An estimate for the total number of virions can be obtained from these measurements by multiplying the viral concentration of each tissue by the total tissue mass [12,13]. The lungs are the largest tissue in terms of mass (~1 kg) and have the highest viral concentration, so they contribute the most to the overall estimate. Other tissues, such as the nasal mucosa, larynx, bronchial tree and adjacent lymph nodes, contribute at most an additional 10% to the estimate based on the lungs, and viraemia appears negligible in most cases. Moving from macaque-to-human mass ratios and loads in human nasopharyngeal swabs [14], Sender et al. estimated that each infected person carries 10⁹–10¹¹ virions during peak infection, and transmission electron microscopy scans indicate that at a given moment there are ~10⁵ viral particles within a single cell, although only about ten of them may be infectious [15]. A higher prevalence of detectable SARS-CoV-2 plasma viral load is associated with worse respiratory disease severity, lower absolute lymphocyte counts and increased markers of inflammation, including C-reactive protein and interleukin-6. SARS-CoV-2 viral loads, especially plasma viraemia, are associated with increased risk of mortality [16].

Total IgG antibody levels specific for SARS-CoV-2 Spike protein were measured 3 weeks following symptom onset showing a concentration in the serum in the order of ~10 μg/mL [17]. Only a fraction of ~5% of the total anti-Spike IgG antibodies has the capacity to neutralize the virus [18]. Combining the concentration of IgG nAb with a mean IgG molecular weight of 150 kDa, Sender et al. estimated that each millilitre of serum contains 3 × 10¹² neutralizing IgG molecules [15]. Of course, these estimates neglect differences in the SARS-CoV-2 neutralizing ability of IgG subclasses [19], as well as the fundamental role in neutralization of IgM and monomeric IgA in plasma [20,21], and of dimeric IgA in secretions [22].

Combining this estimate with the measurement of viral concentration within the lung tissue [10] and accounting for 30–40 Spike homotrimers on each SARS-CoV-2 virion [23,24], Sender et al. estimated that there are 10³–10⁵ nAb per Spike protein [15]. Previous work on morphologically similar viruses such as influenza virus [25] and flavivirus [26] found that a ratio of one nAb to two to four receptor-binding proteins was sufficient to neutralize the binding of a virion to its cellular receptor. Hence, during the ongoing SARS-CoV-2 infection there is an apparent huge excess of nAb (Fig. 1 ), raising doubts about the rationale for nAb-based treatments.

Fig. 1.

The apparent neutralizing antibody excess paradox in severe acute respiratory syndrome coronavirus 2 infection.

Again, these estimates ignore other contributing factors, such as decoy receptors (e.g. angiotensin converting enzyme 2 (ACE2) -positive exosomes [27]) or antagonists (e.g. antagonistic exosomes [28], anti-OC43 antibodies [29], or ABO isoagglutinins [30]).

The biophysical parameters that govern the interaction between any antibody and its cognate antigen are its binding affinity and its concentration. Taken at face value, this estimate seems to suggest that CCP can lead to an excess of nAb molecules. This may be true even if the antibody concentrations in the lung tissue are lower than in the blood, and despite the extensive glycosylation patterns found on the Spike protein that shield many of its epitopes from nAb binding [24], and so decrease the efficiency of neutralization [31]. Hence variations in the dissociation constant K _d (the equilibrium constant that measures the propensity of the antigen–antibody complex to dissociate reversibly) could contribute to explaining the imperfect correlation between high-throughput serology measuring anti-Spike receptor-binding domain IgG and viral neutralization test titres [7]: affinity and concentration can be directly measured in plasma samples of seropositive individuals using surface plasmon resonance [32] or microfluidic antibody affinity profiling [33]. In order to estimate the effectiveness of antibody neutralization, we need to estimate the fractional occupancy of the viral epitopes by antibodies. This fractional occupancy is determined by the strength of the binding of the nAb to the viral particles, given by K _d [26]. The K _d between human ACE2 and SARS-CoV-2 Spike protein are mostly in the range of 15 nM (using surface plasmon resonance [[34], [35], [36]]) to 30 nM (using microfluidics [37]), which is 10- to 20-fold higher than that between human ACE2 and the SARS-CoV-1 Spike protein [38]. This implies that COVID-19 plasma needs a K _d ten-fold higher than in SARS-CoV-1, so that the 1:160 protective nAb titre threshold in SARS is hard to extrapolate for SARS-CoV-2. Hence the concentration of nAb in an average COVID-19 patient is 3 nM and the fractional occupancy needed is 25%–75%. The concentration of antibodies is needed to ensure that enough of the epitopes are bound, regardless of the high ratio between the number of nAb and viral particles [15]. Individual patients show K _d of anti-receptor-binding domain antibodies spanning more than two orders of magnitude (from 80 pM to 25 nM) despite having similar antibody concentrations (8–69 nM). Moreover, these patients showed progressively higher antibody concentrations but constant K _d values, suggesting that IgG affinities against most antigens did not mature with time [37,39]. Incomplete avidity maturation during acute infection might explain the observed decline of the humoral response during convalescence and follow up, potentially exposing individuals to SARS-CoV-2 reinfections. Peptides surrounding a key mutationally constrained B-cell epitope are predicted to bind poorly to common major histocompatibility complex class II alleles, suggesting a lack of major histocompatibility complex class II support in T-cell–B-cell cooperation, impacting the generation of high-potency nAb [40]. Higher levels of anti-Spike avidity were associated with older age, male sex and hospitalization, and also correlate with high nAb titres in viral neutralization tests [41].

The success of nAb-based treatment is dependent on the ratio between the concentration of antibodies and the dissociation constant of the neutralizing fraction of antibodies. During CCP treatment, the donor plasma (about 200 mL) is diluted by roughly a factor of 10 upon transfusion into the patient (2.5 L of plasma). For the donor antibodies to still bind viral proteins effectively following this dilution, in the case where the patient has produced no antibodies, the antibody binding site concentration should exceed the K _d by at least a factor of 10. Therefore, only the patients harbouring anti-SARS-CoV-2 Spike antibodies with very strong K _d values should be considered as CCP donors, a parameter that cannot be sufficiently estimated with conventional methods. Whereas Fc effector functions are not needed when nAb are administered as prophylaxis, they are necessary for optimal therapy [42]. That said, it should be remembered that if any of the systems downstream of neutralization (e.g. clearance of virus–nAb complexes by phagocytes and Fc receptors) becomes saturated, the virus–nAb complex is likely to dissociate.

Do neutralizing antibody levels correlate with disease severity?

It is now well established that individuals with mild/moderate disease display a slower rise and lower peak in anti-nucleocapsid (N) and anti-Spike S1 subunit IgG levels compared with individuals with severe/critical disease (although anti receptor-binding domain IgG and neutralization responses, but not anti-N, reached similar levels at 2–4 months [43]) [44]. Also, individuals with severe disease mount nAb with a higher frequency and a higher PRNT titre than those with moderate and mild disease [45]. According to the danger model of immunity [46], mild clinical presentations may be associated with a poor immune response. However, the fact that an individual with severe disease with high nAb is not able to neutralize the ongoing infection may sound counterintuitive. Although in these cases antibody-dependent enhancement of SARS-CoV-2 infection by pre-existing cross-reacting antibodies facilitating cell entry via Fc receptors cannot be formally excluded [47], antibodies with insufficient K _d, as explained above, remain the most likely explanation.

Garcia-Beltran et al. reported that COVID-19 nAb predict disease severity and survival [48]. Accordingly, Chen et al. reported the highest nAb capacity in sera from patients with severe disease, whereas there was a lack of ability of asymptomatic patients to mount competent nAbs. Furthermore, the compositions of nAb subtypes were also different between recovered patients with severe symptoms and with mild-to-moderate symptoms [49]. Recently, Cervia et al. reported that systemic IgA and IgG production against SARS-CoV-2 develops mainly in severe COVID-19, with very high IgA levels seen in patients with severe acute respiratory distress syndrome, whereas mild disease may be associated with transient serum titres of SARS-CoV-2-specific antibodies but may stimulate mucosal SARS-CoV-2-specific IgA secretion [50].

As the baseline nAb titres are higher in these advanced COVID-19 stages, the nAb titre needed to achieve a clinically measurable benefit is likely to be much higher than the one required by individuals with mild-to-moderate disease.

Has neutralizing antibody-based therapy a chance to work in COVID-19?

So far, most randomized controlled trials have shown no statistically significant clinical benefit (defined as either requirement for mechanical ventilation or mortality) in patients treated after 72 hours from onset of symptoms [[3], [4], [5],51] and in the critically ill COVID-19 stages (WHO score 6–7). This has also proven true when high-titre CCP units have been used (defined as either PRNT >1:160 or with equivalent signal to cut-off ratio in high-throughput serology). This leaves room for exclusive usage of extremely high nAb titre CCP units in advanced stages. Of course, the landscape is different in patients with humoral immune deficiencies (e.g. B-cell alymphocytosis [52], common variable immune deficiency [53], or agammaglobulinaemia [[54], [55], [56]]), in whom lower nAb doses can be clinically useful. Unfortunately, this is exactly the converse of what happens at several transfusion centres, where CCP units not suitable for clinical trials (i.e. with nAb titres below the minimum threshold) are reserved for compassionate uses in advanced COVID-19 stages.

Few CCP trials correlated pretransfusion nAb titres in recipients with nAb titres in CCP units and clinical outcome: in one study, both nAb titres were associated with higher odds for this clinical improvement: pretransfusion nAb was associated with a significant reduction in length of intensive care unit stay and duration of mechanical ventilation [57]. Individuals with severe or life-threatening disease did not show benefit from high-nAb CCP units because they already had high pretransfusion nAb titres [58]. Accordingly, seronegative COVID-19 patients report greater drops in viral load when treated with the monoclonal antibody cocktail REGN-COV2 [59].

Another critical point is the rationale for multiple low-volume (200–300 mL) doses, which are the standard practice in most clinical trials, rather than a single loading dose. While division into sub-units would minimize the discard in case of in-transfusion adverse events, the vast majority of patients in the early stages of disease will be sufficiently haemodynamically stable to sustain transfusion of one or more plasmapheresis units.

As a further point, there is a rationale for transfusing CCP aliquots from different ABO-matched donors to maximize the polyclonality of the nAb. In countries where pathogen reduction technologies and nucleic acid testing on single CCP donations are mandatory, the additional risk for transfusion-transmitted infections associated with transfusion from different donors seems negligible.

Is SARS-CoV-2 replication still driving the pathology in advanced stages of COVID-19?

This is probably the most difficult question to answer. Post-mortem virology studies detected the presence of SARS-CoV-2 genome [[60], [61], [62], [63]], and antigens [63,64]. Several risk factors suggest autoimmunity largely contributes to pathology in advanced stages. The leading cause of death seems angiocentric. Casciola-Rosen et al. showed IgM autoantibodies against ACE2 in 27% of 66 individuals with severe COVID-19 compared with 3.8% of 52 non-hospitalized patients [65]. Accordingly, Gupta et al. reported an increased prevalence of pre-existing anti-interferon-α autoantibodies in systemic lupus erythematosus patients with COVID-19 compared with the reported prevalence in individuals with systemic lupus erythematosus but without COVID-19 and the general population with severe COVID-19 [66]. Franke et al. identified autoantibodies against Yo or NMDA receptor, vessel endothelium, astrocytic proteins and neuropil of basal ganglia, hippocampus or olfactory bulb in 11 severely ill COVID-19 patients presenting with unexplained neurological symptoms [67]. Woodruf et al. identified anti-nuclear autontibodies (ANA) in 44% of 31 critically ill patients with COVID-19 and no known history of autoimmunity [68]: specifically, Gomes et al. showed that anti-DNA antibodies determined upon hospital admission were high in 16% of patients and correlated strongly with later development of severe disease, showing a positive predictive value of 89.5% and accounting for 22% of total severe cases [69]. The same predictive value was true for anti-annexin A2 antibodies in 86 hospitalized patients [70].

Similarly, Bastard et al. identified nAb against type I interferon-α₂ and interferon-ω in about 10% of patients with severe COVID-19 pneumonia [71]. The positive rate of anti-MDA5 antibody in 274 patients with COVID-19 was 48.2% (132/274) and the anti-MDA5 antibody-positive patients tended to present with severe disease (88.6% versus 66.9%, p < 0.0001) [72]. Althaus et al. reported that severe COVID-19 is associated with antibody-mediated up-regulation of platelet apoptosis [73]. Severe illness significantly correlated with elevated anti-cardiolipin IgA, anti-cardiolipin IgM, and anti-β₂ glycoprotein-1 IgA [74,75]. Nevertheless, Borghi et al. reported that anti-phospholipid antibodies show a low prevalence in COVID-19 patients and are not associated with major thrombotic events. Anti-phospholipid antibodies in COVID-19 patients are mainly directed against β₂-glycoprotein-1 but display an epitope specificity different from antibodies in anti-phospholipid syndrome [76].

Wang et al., using a high-throughput autoantibody discovery technique called Rapid Extracellular Antigen Profiling to screen a cohort of 194 SARS-CoV-2-infected patients and health-care workers for autoantibodies against 2770 extracellular and secreted proteins (the ‘exoproteome’), found that COVID-19 patients exhibit dramatic increases in autoantibody reactivities compared with uninfected controls, with a high prevalence of autoantibodies against immunomodulatory proteins including cytokines, chemokines, complement components and cell surface proteins [77].

It is unlikely that nAb can benefit this kind of autoimmunity, which could instead be the target for non-convalescent polyclonal immunoglobulins, which are currently under study [78].

Are there risks from subneutralizing antibody doses?

It has been recently reported that treatment of immunocompromised patients with nAb-based therapeutics can develop accelerated SARS-CoV-2 genome evolution [[79], [80], [81]]: under these circumstances, emergence of viral strains with unprecedented numbers of mutations in the Spike protein is a risk [82] with obvious implications for public health (e.g. the emergence of new strains can impair the efficacy of vaccine campaigns). The phenomenon does not seem common or fast, because none of eight onco-haematological patients (recipients of haematopoietic stem cell transplants or chimeric antigen receptor T lymphocytes) treated with CCP who remained SARS-CoV-2-positive for 2 months showed significant mutations compared with wild-type strain [83]. Avanzato et al. reported within-host genomic evolution in a patient affected by chronic lymphocytic leukaemia and iatrogenic hypogammaglobulinaemia treated with CCP and shedding infectious SARS-CoV-2 for 70 days, and subgenomic RNA for 105 days [79]. Kemp et al. reported an immune suppressed individual who showed little evolutionary change in the first 65 days while on remdesivir, but who developed D796H and ΔH69/ΔV70 mutations twice after two unsuccessful courses of CCP. In vitro, the mutant showed similar infectivity to wild-type strain but resistance to many CCP donors [81]. Hence it seems advisable to avoid treatment with subneutralizing CCP doses that could exert selective pressures, and to monitor patients showing a lack of response for Spike variants.

Conclusion

Several SARS-CoV-2 variants are emerging [82,84,85], and the efficacy of anti-Spike monoclonal antibodies against them seems lower than CCP [86]. Even the currently marketed anti-Spike vaccines are likely to offer reduced protection against some of these variants [87], so the interest in CCP is likely to remain high for months. It should be mandatory to test pretransfusion nAb titres (employing fast viral neutralization test variants or surrogate high-throughput serology) in candidate CCP recipients. Although there is still rationale for using high nAb titre CCP units in advanced disease stages of COVID-19, we feel that best practices (summarized in Table 1 ) should be implemented in randomized controlled trials. Those practices will largely facilitate the creation of mathematical models aimed at predicting the corrected nAb increase for a given therapeutic dose, and finally develop customized therapeutic doses.

Table 1.

Best practices for running CCP trials

Transfuse CCP in the framework of randomized controlled trial

Measure pre-transfusion nAb titre employing fast methods (pseudotype viral neutralization tests or anti-Spike IgG high-throughput serology having high correlation coefficients)

Exclude from enrolment patients with high pretransfusion nAb titre

Use equal volume of fresh frozen non-convalescent plasma in the control arm to assess the relative contribution of non-nAb factors in plasma

Adjust dose among CCP-treated patients according to body weight, nAb titre and eventually affinity

Use a CCP loading dose (ideally from different donors or pooled donations) rather than multiple low-volume CCP doses at different time points

Repeat nAb titre in recipient serum at short-range time points to assess kinetics

Perform genomic studies in refractory/relapsing patients treated with CCP to assess emergence of mutated SARS-CoV-2 strains

Abbreviations: CCP, COVID-19 convalescent plasma; COVID-19, coronavirus disease 2019; nAb, neutralizing antibody; SARS-CoV-2, severe acute respiratory syndrome coronavirus 2.

Transparency declaration

We declare that we have no conflict of interest related to this manuscript. The authors received no funding related to this manuscript.

Author contributions

DF designed the paper, analysed the data and wrote the first draft. FM and ML designed the figure; and MF, AA, AM and FM revised the final version.

Editor: J. Rodriguez-Baño