From our data, the following sequence of events appears to mediate VITT (Fig. 6).
In Step 1, a neo-antigen is generated: following intramuscular injection, vaccine components and platelets come into contact, resulting in platelet activation. ChAdOx1 nCov-19 vaccine activates platelet by multiple mechanisms including platelet interaction with adenovirus, cell-culture derived proteins (currently, it is unknown which of the > 1,000 proteins identified in the vaccine are involved in platelet activation), and EDTA. Activated platelets release PF4. As shown by TEM, released PF4 binds to constituents of the vaccine forming multimolecular aggregates, which also include virus proteins, resulting in particles formation of ≥ 120 nm size.
Step 2 generates an inflammatory co-signal that further stimulates the immune response: EDTA in the vaccine increases capillary leakage at the inoculation site, likely by endothelial (VE)-cadherin disassembly.19 Proteins found in the vaccine include virus proteins, but also proteins originating from the human kidney-derived production cell line T-REx HEK-293. Increased vascular permeability facilitates dissemination of these proteins into the blood. Blood dissemination of vaccine components is not unique to ChAdOx1 nCov-19. A ChAdOx1 vector variant (with a hepatitis B vector insert) was detectable by PCR in multiple organs, including liver, heart, and lymph nodes at days 2 and 29 after intramuscular injection in mice.22
Within the circulation, vaccine constituents including its complexes with PF4 are recognized by preformed natural immunoglobulin G antibodies,23 presumably resulting in immune complexes. This contributes to clinical symptoms within 8 to 24 hours following inoculation that are reminiscent of systemic inflammation (fever, chills, large joint arthralgia, occasionally skin lesions, probably reflecting a similar process as known in serum sickness or serum sickness-like illness24). Such symptoms have also been observed as acute vasculitis like reaction when a column used for immunoadsorption leaked protein A with bound antibodies.25 This inflammatory response likely provides an important co-signal that stimulates antibody production by preformed B-cells capable of producing anti-PF4 antibodies, as is known to occur in the pathogenesis of “classical” HIT 26.12 Multimolecular complexes containing PF4 also activate the complement system.27,28 Complement bound to the aggregates subsequently allows binding of the complexes to B-cells via their complement receptor.27
Step 3 leads to prothrombotic reactions: high avidity anti-PF4 antibodies among the anti-PF4 antibodies in VITT patient blood bind and cluster PF4 on the platelet surface, likely involving polyanions such as cell surface chondroitin sulfate or exposed polyphosphate.29 Clustering of PF4 by high-avidity autoantibodies is also crucial for platelet activation in autoimmune heparin-induced thrombocytopenia.15 The resulting PF4/IgG immune complexes activate platelets, which release additional PF4 and polyphosphate. Crosstalk of PF4, activated platelets and antibodies with neutrophils subsequently leads to NETosis. Extracellular DNA in NETs binds PF4 and resulting DNA/PF4 complexes further recruit anti-PF4 antibodies with lower avidity,30,31 which require the polyanion cofactor (DNA).26 This culminates in massive Fcγ receptor-dependent activation of neutrophils, platelets and, most likely (by analogy with HIT), monocytes and endothelial cells (not shown in the present study; for review17).
An important potential natural regulator of this process are extracellular DNases, which degrade NETs. DNase activity in VITT patients with thrombosis was markedly reduced, likely facilitating accumulation of NETs and DNA. Ultimately, ChAdOx1 nCov-19 vaccine-triggered VITT culminates in marked activation of the coagulation system.4,5,32
Broadened reactivity of antibodies in a boosted immune response is a hallmark of certain disorders besides VITT. For example, autoimmune heparin-induced thrombocytopenia features heparin-dependent reactivity that extends to include heparin-independent reactivity.13 Similarly, post-transfusion purpura (PTP)33 reflects a strong alloimmune response that progresses to include autoreactive properties.34 In this regard, we identified an array of cell-culture derived human proteins in the vaccine, potentially predisposing to immune reactions against these antigens. If such proteins express a structures, e.g. by a genetic polymorphism not found in the corresponding endogenous protein of the vaccinated individual, a possible strong alloimmune response (with potential for autoreactivity) in a susceptible vaccine recipient should be considered.
Our study has limitations. The detailed specifications of the ChAdOx1 nCov-19 vaccine are not publicly available and potential impact of about 35–40 µg human cell culture proteins per vaccination dose remain to be assessed by the responsible regulatory agencies. Furthermore, we did not analyze the constituents of other adenovirus-based Covid-19 vaccines such as the Covid-19 Vaccine Janssen and the Sputnik V vaccine (these were not available to us). More importantly, quality control of vaccines requires the comprehensive methodological expertise of regulatory agencies.
Currently, we have not investigated the roles of B-cells or T-cells underlying the VITT immune response, nor the role of complement in initiating the immune response or in contributing to subsequent antibody-mediated immunothrombosis.35 Our PF4 preparation contains a small amount of PF4variant-1 (which is also released by platelets in vivo); we cannot define its role in platelet or granulocyte activation.
We have provided evidence that VITT is not a consequence of antibodies directed against the SARS-CoV-2 spike protein (produced by all vaccines) cross-reacting with PF4 (preprint).36 Those findings, together with our current study, indicate it is the adenovirus vector-based vaccines that are at risk of inducing VITT through adenovirus and/or other PF4-DNA interactions. The degree of acute inflammatory response induced by the vaccine components appears as an important—potentially remediable—cofactor that could be diminished by reducing impurities and omitting EDTA.
In summary, our study provides a mechanism by which an adenoviral vector vaccine can trigger an immune response leading to highly reactive anti-PF4 antibodies with downstream FcγIIa receptor-dependent amplification recruiting neutrophils and triggering NETosis with prothrombotic consequences.