Analysis of anti-Omicron neutralizing antibody titers in different convalescent plasma sources.

The latest SARS-CoV-2 variant of concern Omicron, with its immune escape from therapeutic anti-Spike monoclonal antibodies and vaccine-elicited sera, demonstrates the continued relevance of COVID19 convalescent plasma (CCP) therapies. Lessons learnt from previous usage of CCP suggests focusing on outpatients and immunocompromised recipients, with high neutralizing antibody (nAb) titer units. In this analysis we systematically reviewed Omicron neutralizing plasma activity data, and found that approximately 50% (426/911) of CCP from unvaccinated donors neutralizes Omicron with a very low geometric mean of geometric mean titers for 50% neutralization (GM(GMT50)) of about 17, representing a more than 24-fold reduction from paired WA-1 neutralization. Two doses of mRNA vaccines in nonconvalescent subjects had a similar 50% percent neutralization with Omicron neutralization GM(GMT(50)) about 24. However, CCP from vaccinees recovered from previous variants of concern or third-dose uninfected vaccinees was nearly 100% neutralizing with Omicron GM(GMT(50)) over 200, a 12-fold Omicron neutralizing antibody increase compared to unvaccinated convalescents from former VOCs. These findings have implications for both CCP stocks collected in prior pandemic periods and plans to restart CCP collections. Thus, CCP from vaccinated donors provides an effective tool to combat variants that defeat therapeutic monoclonal antibodies.

Introduction

The SARS-CoV-2 Omicron variant of concern (VOC) (originally named VUI-21NOV-01 by Public Health England and belonging to GISAID clade GRA(B.1.1.529+BA.*) was first reported on November 8, 2021 in South Africa, and shortly thereafter was also detected all around the world. Omicron mutations impact 27% of T cell epitopes[1] and 31% of B cell epitopes of Spike, while percentages for other VOC were much lower[2]. The Omicron variant has further evolved to several sublineages which are named by PANGO phylogeny using the BA alias: the BA.1 wave of Winter 2021–2022 has been suddenly replaced by BA.2 and BA.2.12.1 in Spring 2022, and by the BA.4 and BA.5 waves in Summer 2022..

The VOC Omicron is reducing the efficacy of all vaccines approved to date (unless 3 doses are delivered) and is initiating an unexpected boost in COVID19 convalescent plasma (CCP) usage, with Omicron being treated as a shifted novel virus instead of a SARS-CoV-2 variant drift. Two years into the pandemics, we are back to the starting line for some therapeutic classes. Specifically, Omicron escapes viral neutralization by most monoclonal antibodies (mAbs) authorized to date with the lone exception of bebtelovimab[3]. Despite the development of promising oral small-chemical antivirals (molnupiravir and nirmatrelvir), the logistical and economical hurdles for deploying these drugs worldwide has prevented their immediate and widespread availability, and concerns remain regarding both molnupiravir (both safety[4] and efficacy[5]) and nirmatrelvir (efficacy), expecially in immunocompromised subjects. COVID19 convalescent plasma (CCP) was used as a frontline treatment from the very beginning of the pandemic. Efficacy outcomes have been mixed to date, with most failures explained by low dose, late usage, or both, but efficacy of high-titer CCP has been definitively proven in outpatients with mild disease stages[6,7]. Neutralizing antibody (nAb) efficacy against VOC remains a prerequisite to support CCP usage, which can now be collected from vaccinated convalescents, including donors recovered from breakthrough infections (so-called “hybrid” or “VaxCCP”)[8]: pre-Omicron evidence suggest that those nAbs have higher titers and are more effective against VOCs than those from unvaccinated convalescents[9, 10]. From a regulatory viewpoint, to date, plasma from vaccinees that have never been convalescent does not fall within the FDA emergency use authorization

There are tens of different vaccine schedules theoretically possible according to EMA and FDA approvals, including a number of homologous or heterologous boosts, but the most commonly delivered schedules in the western hemisphere have been: 1) BNT162b2 or mRNA-1273 for 2 doses eventually followed by a homologous boost; 2) ChAdOx1 for 2 doses eventually followed by a BNT162b2 boost; and 3) Ad26.COV2.S for 1 dose eventually followed by a BNT162b2 boost[11]. Many more inactivated vaccines have been in use in low-and-middle income countries (LMIC), which are target regions for CCP therapy: this is feasible given the lower number of patients at risk for disease progression there (lower incidences of obesity, diabetes, and hypertension, and lower median age) and the already widespread occurrence of collection and transfusion facilities. Most blood donors there have already received the vaccine schedule before, after or without having been infected, with a nAb titer generally declining over months[12]. Hence identifying the settings where the nAb titer is highest will definitively increase the efficacy of CCP collections. Variations in nAb titers against a given SARS-CoV-2 strain are usually reported as fold-changes in geometric mean titer of antibodies neutralizing 50% of cytopathic effect or foci (GMT50) compared to wild-type strains: nevertheless, fold-changes for groups that include non-responders can lead to highly artificial results and possibly over-interpretation. Rigorous studies have hence reported the percentage of responders as primary outcome and provided fold-changes of GMT50 where calculation is reasonable (100% responders in both arms)[13].

To date the most rigorous data repository for SARS-CoV-2 sensitivity to antivirals is the Stanford University Coronavirus Antiviral & Resistance Database, but as of July 24, 2022 the tables there summarizing “Convalescent plasma” and “Vaccinee plasma” (https://covdb.stanford.edu/search-drdb/?form_only) do not dissect the different heterologous or homologous vaccination schemes, the simultaneous occurrence of vaccination and convalescence, or the time from infection/vaccine to neutralization assay. Consequently, a more in-depth analysis is needed to better stratify CCP types.

Methods

On July 23, 2022, we searched PubMed, medRxiv and bioRxiv for research investigating the efficacy of CCP (either from vaccinated or unvaccinated donors) against SARS-CoV-2 VOC Omicron for article (pre)published after December 1, 2019, using English language as the only restriction. In PubMed we used the search query “(“convalescent plasma” or “convalescent serum”) AND (“neutralization” or “neutralizing”) AND “SARS-CoV-2””, while in bioRxiv and medRxiv we searched for abstract or title containing “convalescent, SARS-CoV-2, neutralization” (match all words). When a preprint was published, the latter was used for analysis. We also screened the reference lists of reviewed articles for additional studies not captured in our initial literature search. Articles underwent evaluation for inclusion by two assessors (D.F. and D.S.) and disagreements were resolved by a third senior assessor (A.C.). We excluded review articles, meta-analyses, studies reporting antibody levels by serological assays other than neutralization, as well as studies exclusively analyzing nAbs in vaccine-elicited plasma/serum from non-convalescent subjects. In unvaccinated subjects, convalescence was annotated according to infecting sublineage (pre-VOC Alpha, VOC Alpha, VOC Beta, VOC Delta, or VOC Omicron sublineages). Given the heterologous immunity that develops after vaccination in convalescents, the infecting lineage was not annotated in vaccine recipients. In vaccinees, strata were created for 2 homologous doses, 3 homologous doses, or post-COVID-19 and post-vaccination (Vax-CCP). The mean neutralizing titer for WA-1 (pre-Alpha wild-type), Omicron and number out of total that neutralized Omicron was abstracted from studies.

Statistical significance between means was investigated using Tukey’s test.

Results

Our literature search identified 29 studies dealing with the original Omicron lineage (BA.1), that were then manually mined for relevant details : the PRISMA flowchart for study selection is provided in Figure 1. Given the urgency to assess efficacy against the upcoming VOC Omicron, most studies (with a few exceptions[14, 15, 16, 17]) relied on Omicron pseudovirus neutralization assays, which, as opposed to live authentic virus, are scalable, do not require BSL-3 facilities, and provide results in less than 1 week. GMT50 of nAb and fold-reduction (in GMT50 against Omicron compared to wild-type SARS-CoV-2 (e.g., WA-1) were the most common ways of reporting changes, which reduces variability due to difference in neutralization assays used.

Figure 1

PRISMA flowchart for the current study.

Figure 2 and Table 1 summarize that neutralizing activity to WA-1 from CCP collected from subjects infected with pre-Alpha SARS-CoV-2 (Supplementary Table 1), Alpha VOC (Supplementary Table 2), Beta VOC (Supplementary Table 3), Delta VOC (Supplementary Table 4) or plasma from nonconvalescent subjects vaccinated with 2 mRNA vaccine doses (Supplementary Tables 5 and 6)The same plasma types computed a geometric mean of multiple GMT50 from many studies with about a 21-fold reduction against BA.1 geomeans compared to wild-type SARS-CoV-2 geomeans. CCP from uninfected vaccinees receiving a third vaccine dose registered geomean of the GMT(50) of 2,723 (or 10- fold higher nAb geomean of the GMT50) to wild-type viral assays: in this group the nAb geomean of the GMT50 fold-reduction against BA.1 was 9, but importantly the geomean of the GMT(50) was close to 291 again. The approximately 21-fold reduction in nAb geomean of the GMT(50) from wild-type to BA.1 was reversed by the 10–15-fold increase in nAb geomean of the GMT(50) from either boosted vaccination or VaxCCP.

Figure 2

Geometric mean neutralizing titers (GMT50) against WA-1 versus Omicron BA.1 by study for A) unvaccinated convalescent plasma and B) vaccinated plasma with or without COVID-19. Geomeans for entire study groups with neutralization of WA-1 in filled circles with Omicron in empty circles with geomeans and fold reduction (FR) above data and number of studies above x-axis. All geomeans are not statistically significant in difference by multiple comparison in Tukey’s test.

Table 1

Comparison of WA-1 to Omicron BA.1 nAb and percent with any Omicron BA.1 nAb amongst VOC CCP and vaccination status.

plasma type	number of studies	WA-1 nAb GMT50	Omicron BA1 nAb GMT50	fold reduction in nAb GMT50 vs. Omicron BA.l	total number individuals in all studies	total Omicron BA.l neutralizingnumber	Omicron BA.l neutralizing percent
pre-Alpha	27	326	15	21	679	300	44
Alpha	6	227	5	45	101	38	38
Beta	5	91	8	11	37	19	51
Delta	7	462	42	11	94	69	73
2 dose BNT162b2 plasma	22	639	26	25	434	204	47
2 dose mRNA-1273 plasma	9	644	21	31	134	81	60
post-COVID-19/full vacc plasma	19	2977	211	14	305	269	88
3 dose BNT162b2 plasma	17	2,723	291	9	307	293	95

In addition to the nAb GMT50 levels showing potency, the percentage of individuals within a study cohort positive for any level of BA.1 neutralization shows the likelihood of a possible donation having anti-BA.1 activity. All studies but one tested a limited number of 20 to 40 individuals. The pre-Alpha CCP showed that most (18 of 27 studies) had less than 50% of individuals tested within a study with measurable BA.1 neutralizing activity: only 2 out of 27 studies indicated 100% of individuals tested showed BA.1 neutralization (Figure 3). Likewise, most of the studies investigating Alpha and Beta CCP showed similar percent with nAb. Delta CCP had 6 of 7 studies with more than 50% BA.1 neutralization. The plasma from studies of the 2-dose mRNA vaccines indicated a more uniform distributive increase in percent of individual patients with measurable Omicron nAb’s. The stark contrast is Vax-CCP, where 16 of 19 studies had 100% of individuals tested with anti-BA.1 nAb. The 3-dose vaccinee studies similarly had 12 of 17 studies with 100% measurable nAb.

Figure 3

Percent of individual plasma samples in each study showing any titer of Omicron BA.1 neutralization. The percent of samples within a study condition which neutralized Omicron graphed in increasing percentages with the number of samples tested on the right y axis. A) pre-Alpha CCP neutralization of Omicron; B) Alpha, Beta and Delta CCP neutralization of Omicron C) 2 dose mRNA vaccines neutralization of Omicron D) post-COVID-19/post-vaccine (VaxCCP) and uninfected 3-dose vaccine neutralization of Omicron.

There were 5 studies which directly compared anti-WA-1 versus BA.1 nAb titers in nonvaccinated pre-Alpha, Alpha, Beta, and Delta CCP, and vaccinated plasma with the same nAb assay (Figure 4). nAb GMT50 against WA-1 was higher for Alpha and Delta CCP but lower for Beta CCP. nAb geomean of the GMT(50) against BA.1 was actually highest for Beta CCP 13 geomean with geomean levels of 9, 8, 10 for pre-Alpha, Alpha and Delta (Figure 4, panel A). In these 5 studies, nAb geomean of the GMT(50) rose from 2-dose vaccinations to VaxCCP to the 3-dose boosted vaccination. Importantly, for nAb geomean of the GMT(50) against BA.1 were 13 to 103 to 223, respectively representing a 8 to 17-fold rise (Figure 4, panel B).

Figure 4

Geometric mean neutralizing titers (GMT50) of anti-WA.1 or anti-Omicron BA.1 neutralizing antibodies in plasma samples from 5 studies investigating diverse SARS-CoV-2 infecting lineage or vaccination status. 5 studies characterized A) pre-Alpha, Alpha, Beta and Delta CCP for Omicron nAb compared to WA-1, and also B) 2 or 3 doses BNT162b plasma, as well as post-COVID-19 plus BNT162b vaccine (VaxCCP). C) 9 additional studies looked at the same vaccine conditions in the first 5 comparing WA-1 nAb to Omicron nAb.

Another set of 9 matched vaccination studies inclusive of plasma collected after 2- and 3-dose schedules, as well as Vax-CCP depicted a 23-fold rise in geomean of the GMT(50) of anti-BA.1 nAb from the 2-dose vaccine to post COVID-19 vaccinees, and a 21-fold increase after the third vaccine dose. The pattern was similar for nAb geomean of the GMT(50) against WA-1 (Figure 4, panel C).

The AZD1222, 3-dose mRNA-1273 and Ad26.COV2 vaccines were understudied, with 3 or less independent studies at different time points, reported in Table 10. The GMT50 nAb to BA.1 after 3- mRNA-1273 doses ranged 60 to 2000, with a 5 to 15 fold reduction compared with WA-1. GMT50 of anti-BA.1 nAbs after AZD1222 vaccine was modest (~10 to 20), as with Ad26.COV2 vaccine (~20 to 40). Two studies reported on post-COVID-19/post-mRNA-1273 with nAb GMT50 against BA.1 of 38 and 272. Studies with 100% of individual patient samples neutralizing BA.1 included 2 3-dose mRNA-1273 studies, one AZD1222 study, and one post-COVID-19/post-mRNA-1273 study.

Few data exist for comparisons among different vaccine boosts. For CoronaVac® (SinoVac), three doses led to 5.1 fold reduction in anti-BA.1 nAb GMT50 compared to wild-type[18], while for Sputnik V nAb titer moved from a 12-fold reduction at 6–12 months up to a 7-fold reduction at 2–3 months after a boost with Sputnik Light[19, 20]. These in vitro findings have been largely confirmed in vivo, where prior heterologous SARS-CoV-2 infection, with and without mRNA vaccination, protects against BA.1 re-infection[21].

Seventeen studies analyzed the efficacy of CCP and VaxCCP against Omicron sublineages other than BA.1 (summarized in Table 2). Those studieslargely confirmed that Omicron CCP per se is poorly effective against the cognate or other Omicron sublineages[22] (with the lone exception of cross-reactions among lineages sharing L452 mutations[23] and broad-spectrum nAbs elicited by BA.5[24]). On the contrary, both homologous and heterologous efficacy of Omicron VaxCCP is again universally preserved[15, 25]. Despite evidences that concentrated pooled human IgG from convalescent and vaccinated donors has 5-fold reduced potency against BA.5 compared to wild-type SARS-CoV-2[26], such VaxCCP derivative is devoid of IgA and IgM nAbs. These findings have important implications if a VaxCCP program is going to be re-launched at the time of BA.2 and BA.4/5 waves.

Table 2

Efficacy of CCP, vaccinee plasma and VaxCCP expressed as GMT50 against Omicron sublineages.

CCP source	target Omicron sublineage
	BA.1	BA.2	BA.2.12.1	BA.4/5
wild-type CCP (unvaccinated)	↓[50] (including BA.1.1)	↓[50]	no data	no data
uninfected 3-dose mRNA vaccinee plasma	↓[50] (including BA.1.1)[15,25]	↓[50]	no data	stronger escape than BA.2[23,51,52]
any pre-Omicron VOC VaxCCP	no data	=[53]	no data	24
Delta VaxCCP	no data	no data	23	23
BA.l CCP	↓[22]	no data	no data	7.5–7.6-fold lower than against BA.1[23,51,52,54,55]
BA.l VaxCCP	1:2929 at 9–12 days[15,25,48,56]	1.3 to 1.8-fold lower[50,57,58] 4.2-fold lower[59] than against the parental BA.l sublineage; no neutralization[60][48]	1.8-fold lower than against BA 2[23,51 61,62] > 5-fold lower compared to wild-type[56][48]	2.6–3.2-fold lower than against BA I[54,55,61,63] 4.5-fold lower than against BA.2[55] > 5-fold lower compared to wild-type[56][48]
BA.2 CCP	no data	no data	no data	poor[55]
BA.2 VaxCCP	1.2-fold lower compared to wild-type[56]	1.5-fold lower compared to wild-type[56]	2.5-fold lower compared to wild-type[56]	[63]2.5-fold lower compared to wild-type[56]
BA.2.12.1 CCP	no data	no data	no data	no data
BA.2.12.1 VaxCCP	no data	no data	no data	no data
BA.4/5 CCP	557 (2-FR)[24]	884 (1-FR)[24]	no data	1,047[24]
BA.4/5 VaxCCP	2,785 (2-FR)[24]	4244 (1-FR)[24]	no data	3,779[24]

Discussion

Since nAbs are by definition antiviral, CCP with a high nAb GMT50 is preferable,, and there is now strong clinical evidence that nAb titers correlate with clinical benefit in randomized clinical trials[6, 7]. Although nAb titers correlate with vaccine efficacy[27, 28], it is important to keep in mind that SARS-CoV-2-binding non-neutralizing antibodies can similarly provide protection via Fc-mediated functions[29, 30]. However, such functions are harder to measure and no automated assay exist for use in clinical laboratories. Hence, whereas the presence of a high nAb GMT50 in CCP is evidence for antibody effectiveness in vitro, the absence of nAb titer does not imply lack of protection in vivo where Fc effects mediate protection by other mechanisms such as antibody-dependent cell-mediated cytotoxicity, complement activation and phagocytosis.

The mechanism by which CCP from vaccinated COVID-19 convalescent individuals better neutralizes Omicron lineagesis probably a combination of higher amounts of nAb and broader antibody specificity. Higher amounts of antibody could neutralize antigenically different variants through the law of mass action[31] whereby even lower affinity antibodies elicited to earlier variants would bind to the Omicron variant as mass compensates for reduced binding strength to drive the reaction forward. In addition, vaccinated COVID-19 convalescent individuals would have experienced SARS-CoV-2 protein in two antigenically different forms: as part of intact infective virions generated in vivo during an infectious process and as antigens in vaccine preparations. As the immune system processes the same antigen in different forms, there are numerous opportunities for processing the protein in different manners that can diversity the specificity of the immune response and thus increase the likelihood of eliciting antibodies that react with variant proteins. Structurally, it has been shown that third dose mRNA vaccination induces mostly class 1/2 antibodies encoded by IGHV1–58;IGHJ3–1 and IGHV1–69;IGHJ4–1 germlines, but not the IGHV2–5;IGHJ3–1 germline, broadly cross-reactive Class 3 antibodies seen after infection[32].

Our analysis provides strong evidence that, unlike what has been observed in Syrian hamster models[33], CCP from unvaccinated donors is unlikely (less than 50%) to have any measurable Omicron neutralization. Although the nAb GMT50 threshold for clinical utility remains poorly defined, it is noticeable that low BA.1 nAb GMT50 were generally detected in CCP after infection from pre-Omicron VOCs.

On the contrary, despite the huge heterogeneity of vaccine schedules, CCP from vaccinated and COVID-19 convalescent individuals (Vax-CCP) consistently harbors high nAb titers against BA.1 and novel sublineages if collected up to 6 months since last event (either vaccine dose or infection). These Omicron neutralizing levels are comparable in dilutional titers to that of WA-1 CCP neutralizing WA-1, but their prevalence is much higher at this time, facilitating recruitment of suitable donors. Pre-Omicron CCP boosted with WA-1-type vaccines induces heterologous immunity that effectively neutralizes Omicron in the same assays which rule in or out therapeutic anti-Spike monoclonal antibodies. Consequently, prescreening of Vax-CCP donors for nAb titers is not necessary, and qualification of Vax-CCP units remains advisable only within clinical trials. A more objective way to assess previous infection (convalescence) would be measuring anti-nucleocapsid (N) antibodies, but unfortunately these vanish quickly[34, 35]. Previous symptomatic infection and vaccination can be established by collecting past medical history (PMH) during the donor selection visit, which is cheaper, faster, and more reliable than measuring rapidly declining anti-N antibodies. Although there is no formal evidence for this, it is likely that asymptomatic infection (leading to lower nAb levels in pre-Omicron studies) also leads to lower nAb levels after vaccination compared to symptomatic infection, given that disease severity correlates with antibody titer[36, 37]: hence those asymptomatically infected donors missed by investigating PMH are also less likely to be useful.

The same reasoning applies to uninfected vaccinees receiving third dose boosts, but several authorities, including the FDA, do not currently allow collection from such donors for CCP therapy on the basis that the convalescent polyclonal and poly-target response is a prerequisite for efficacy and superior to the polyclonal anti-Spike-only response induced by vaccinees. This may be a false premise for recipients of inactivated whole-virus vaccines (e.g., BBIBP-CorV or VLA2001): for BBIBP-CorV, the efficacy against Omicron is largely reduced[18, 20, 38], but the impact of boost doses is still unreported at the time of writing. Table 1 and Table 9 clearly show that 3-doses of BNT162b2 are enough to restore nAb levels against Omicron in the absence of SARS-CoV-2 infection.

Another point to consider is that information on nAb levels after the third vaccine dose has been almost exclusively investigated for only 1 month of follow-up, while studies on convalescents extend to more than 6 months: to date it seems hence advisable to start from convalescent vaccinees rather than uninfected 3-dose vaccinees. This is also confirmed by immune escape reported in vivo after usage of vaccine (non-convalescent) plasma[39] despite very high nAb titres, likely due to restricted antigen specificity. Vaccine schedules with a delayed boost seem to elicit higher and broader nAb levels than the approved, short schedules[40, 41, 42, 43], but this remain to be confirmed in larger series. The same is true for breakthrough infections from Alpha or Delta VOC in fully BNT162b2 vaccinated subjects[44], although variation in time from infection due to successive waves is a major confounder.

With the increase of Omicron seroprevalence in time, polyclonal intravenous immunoglobulins collected from regular donors could become a more standardized alternative to CCP, but their efficacy to date (at the peak of the vaccinations campaign) is still 16-fold reduced against Omicron compared to wild-type SARS-CoV-2[45], and such preparations include only IgG and not IgM and IgA, which have powerful SARS-CoV-2 activity[46, 47]. Nevertheless, FDA recently reported efficacy of hyperimmune serum against BA.1, BA.2, BA.3, BA.2.12.1, and BA.4/5[48].

CCP collection from vaccinated convalescents (regardless of infecting sublineage, vaccine type and number of doses) is likely to achieve high nAb titer against VOC Omicron, and, on the basis of lessons learnt with CCP usage during the first 2 years of the pandemic. Although in ideal situations one would prefer RCT evidence of efficacy against Omicron before deployment, there is concern that variants are generated so rapidly that by the time such trials commenced this variant could be replaced for another. Given the success of CCP in 2 outpatient RCTs reducing hospitalization[6, 7] and the loss of major mAb therapies due to Omicron antigenic changes, the high titers in CCP collected from vaccinated convalescents provides an immediate option for COVID-19, especially in LMIC. Given the reduced hospitalization rate with Omicron compared to Delta[49], it is even more relevant to identify patient subsets at risk of progression in order to minimize the number needed to treat to prevent a single hospitalization: moving from the same criteria used for mAb therapies while using the same (now unused) in-hospital facilities seems a logical approach.