Antibodies to PF4 in severe COVID-19: same but different

In this issue of Blood, Zhu et al¹ find evidence of B-cell clones with the hallmarks of heparin-induced thrombocytopenia (HIT) specificity in a subset of patients with severe COVID-19 and platelet-activating antibodies. Given the shared pathogenesis of immunothrombosis, they looked for platelet factor 4 (PF4)/heparin antibodies in 130 patients hospitalized with severe COVID-19. As many as 80% of these patients demonstrated PF4/heparin immunoglobulin G (IgG) antibodies (optical density at 450 nm [OD450] >0.5), but only half of these displayed a functional capacity to activate platelets as measured by increased platelet surface P-selectin expression (PEA) after sensitization with a synthetic Toll-like receptor 9 (TLR-9) agonist motif cytosine triphosphate deoxynucleotide followed by a guanine triphosphate deoxynucleotide (PEA^CpG assay).^2,3 

To explore the functional impact of PEA^CpG+ antibodies, multiplex Luminex analyses of plasma proteins identified evidence of increased platelet activation and inflammation with increased platelet α-granule proteins, ferritin, C-reactive protein, and fibrinogen compared with patients whose PEA^CpG functional analysis was negative.

They found a familiar inhibition of this platelet activation after the addition of high-dose heparin or the low-affinity type 3 receptor for Fc portion of IgG blocking IV.3 as commonly seen in HIT. Exploring this antigenic response in the context of COVID-19 infection, the investigators found a modest positive correlation (r = 0.444) between the presence of PF4/heparin antibodies and antibodies to the receptor-binding domain (RBD) of the spike protein. Adding recombinant RBD does not interfere with PF4/heparin recognition but specifically impairs the functional PEA^CpG. Absorbing RBD-specific antibodies with magnetic beads reduced both platelet activation and PF4/heparin binding of purified IgG from patients with severe COVID-19, suggesting these antibodies have overlapping reactivity and, in some individuals, arise from the same B-cell clone.

To demonstrate the clonal ability to produce both platelet-activating and RBD-specific antibodies, the investigators then isolated RBD-binding IgG1⁺ B cells from 2 patients with COVID-19 with prior heparin exposure but enzyme-linked immunosorbent assay positive for PF4/heparin reactivity. Zhu et al expressed recombinant antibodies in vitro from them and identified 4 from 21 recombinant antibodies with platelet-activating ability. Two of these clones were sequenced and share variable diversity joining heavy-chain complementarity determining region (HCDR3) motifs (“RKH” or “Y5”) also reported in HIT.⁴ 

Analyzing a previously published analysis of COVID-19 convalescent samples,⁵ the investigators then identified 15 RBD-specific B-cell clones with HIT “RKH” or “Y5” HCDR3 motifs from which 4 recombinantly recreated clones were able to produce platelet-activating antibodies. Exploring publicly available sequences from patients with COVID-19 and control subjects prior to the development of COVID-19 vaccines, they also found a higher frequency of HIT HCDR3 motifs in COVID-19 convalescent patients than healthy control subjects, consistent with the development of an RBD-specific B-cell response to COVID-19 infection generating platelet-activating antibodies in a subset of cases (see figure).

But circumstantial evidence is not quite as compelling as direct evidence of increased thrombosis, Sequential Organ Failure Assessment (SOFA) scores, mortality, or morbidity in patients with platelet-activating (PEA^CpG+) RBD-specific antibodies. These antibodies are not as functionally active as HIT antibodies in vitro, requiring CpG sensitization to augment the response sufficiently for detection, in contrast to HIT antibodies that do not require CpG. Perhaps because of their attenuated potency, there was no clinical correlation with platelet counts, or P-selectin expression? Although these RBD/HIT HCDR3 motifs are more frequent after COVID-19 infection (7.14% ± 2.18% vs 4.49% ± 0.92%), they were present in some healthy control subjects prior to the pandemic, so their development is not an event exclusive to COVID-19.

This article raises many intriguing questions. What determinants predispose some patients to generate RBD-specific antibodies with the potential for platelet activation and the hallmarks of HIT HCDR3 motifs? Does the development of these antibodies lead to severe COVID-19, or are severe COVID-19 cases more likely to develop these antibodies? Inflammatory markers and platelet degranulation seem to occur more frequently in patients that were PEA^CpG+, but we do not know which came first.

Were these B-cell clones already present at lower levels and simply enriched by antigenic stimulation by the severity of their COVID-19? What is the clinical impact of platelet antibodies in vivo that require exogenous CpG oligodeoxynucleotides to stimulate TLR-9 to promote P-selectin expression in vitro?

Clearly, any downstream application of these antibodies as biomarkers for severity or adverse clinical outcomes such as thrombosis requires clinical validation. The mechanism of immunothrombosis in COVID-19 is clearly complex. It may be that PEA^CpG is underreporting the deleterious functionality of these antibodies on regional endothelial injury,⁶ monocyte activation,⁷ neutrophil extracellular trap formation,⁸ complement activation,⁹ and fibrinogen-virus interactions.¹⁰ More research by investigators like Zhu et al is needed to interrogate other functional abilities of these same RBD and PF4/heparin cross-reactive antibodies as they may yield results that correlate better with clinical outcomes.

Conflict-of-interest disclosure: The authors declare no competing financial interests.