key: cord-0767225-7l3axjeh
authors: Gresele, Paolo; Momi, Stefania; Marcucci, Rossella; Ramundo, Francesco; De Stefano, Valerio; Tripodi, Armando
title: Interactions of adenoviruses with platelets and coagulation and the vaccine-induced immune thrombotic thrombocytopenia syndrome
date: 2021-08-19
journal: Haematologica
DOI: 10.3324/haematol.2021.279289
sha: fda6fc9fc5c6be1f1d7886ddb1416f48d0281153
doc_id: 767225
cord_uid: 7l3axjeh

The COVID-19 pandemic has had a heavy impact on global health and economy and vaccination remains the primary way of controlling the infection. During the ongoing vaccination campaign some unexpected thrombotic events have emerged in subjects who had recently received the AstraZeneca (Vaxzevria) vaccine or the Johnson&Johnson (Janssen) vaccine, two adenovirus vector-based vaccines. Epidemiological studies confirm that the observed/expected ratio of these unusual thromboses is abnormally increased, especially in women in fertile age. The characteristics of this complication, with venous thromboses at unusual sites, most frequently in the cerebral vein sinuses but also in splanchnic vessels, often with multiple associated thromboses, thrombocytopenia, and sometimes disseminated intravascular coagulation, are unique and the time course and tumultuous evolution are suggestive of an acute immunological reaction. Indeed, plateletactivating anti-PF4 antibodies have been detected in a large proportion of the affected patients. Several data suggest that adenoviruses may interact with platelets, the endothelium and the blood coagulation system. Here we review interactions between adenoviral vectors and the hemostatic system that are of possible relevance in vaccine-associated thrombotic thrombocytopenia syndrome. We systematically analyze the clinical data on the reported thrombotic complications of adenovirus-based therapeutics and discuss all the current hypotheses on the mechanisms triggering this novel syndrome. Although, considering current evidence, the benefit of vaccination clearly outweighs the potential risks, it is of paramount importance to fully unravel the mechanisms leading to vaccineassociated thrombotic thrombocytopenia syndrome and to identify prognostic factors through further research.

T he COVID-19 pandemic has had a heavy impact on global health and economy and vaccination remains the primary way of controlling the infection. During the ongoing vaccination campaign some unexpected thrombotic events have emerged in subjects who had recently received the AstraZeneca (Vaxzevria) vaccine or the Johnson&Johnson (Janssen) vaccine, two adenovirus vector-based vaccines. Epidemiological studies confirm that the observed/expected ratio of these unusual thromboses is abnormally increased, especially in women in fertile age. The characteristics of this complication, with venous thromboses at unusual sites, most frequently in the cerebral vein sinuses but also in splanchnic vessels, often with multiple associated thromboses, thrombocytopenia, and sometimes disseminated intravascular coagulation, are unique and the time course and tumultuous evolution are suggestive of an acute immunological reaction. Indeed, plateletactivating anti-PF4 antibodies have been detected in a large proportion of the affected patients. Several data suggest that adenoviruses may interact with platelets, the endothelium and the blood coagulation system. Here we review interactions between adenoviral vectors and the hemostatic system that are of possible relevance in vaccine-associated thrombotic thrombocytopenia syndrome. We systematically analyze the clinical data on the reported thrombotic complications of adenovirus-based therapeutics and discuss all the current hypotheses on the mechanisms triggering this novel syndrome. Although, considering current evidence, the benefit of vaccination clearly outweighs the potential risks, it is of paramount importance to fully unravel the mechanisms leading to vaccineassociated thrombotic thrombocytopenia syndrome and to identify prognostic factors through further research.

The coronavirus disease pandemic has prompted an unprecedented effort to develop highly effective vaccines to prevent further spreading of the infection, the associated mortality and the enormous strain on healthcare systems. Indeed, in a previously unimaginable short time, many vaccines have been developed. Several of them underwent controlled randomized phase III clinical trials and, as of 22 June, 2021, 13 have been licensed globally for clinical use. By July 18, 2021 they had been administered to more than 1.9 billion subjects worldwide (923 million of whom are fully vaccinated; 3.66 billion doses have been administered globally; 26.3% of the world's population has received at least one dose of a COVID-19 vaccine). This represents the most massive vaccination campaign ever undertaken (https://www.who.int/emergencies/diseases/novel-coronavirus-2019/covid-19-vaccines; https://ourworldindata.org/covid-vaccinations).

Although careful scrutiny of vaccine safety in controlled randomized phase III clinical trials did not highlight significant thrombotic risks, exceedingly rare events may have been missed and indeed during the vaccination campaign several cases of thrombosis, in particular thrombotic events at unusual sites associated with thrombocytopenia, were reported. Most events occurred in subjects who had received the ChAdOx1 (Vaxzevria) vaccine in the preceding weeks, but more recently several cases have also been reported following the Ad26-CoV2S Johnson&Johnson (Janssen) vaccine. [1] [2] [3] [4] [5] [6] [7] [8] [9] Not only was the observed/expected ratio of these thromboses abnormally high in subjects receiving the Vaxzevria vaccine, but the clinical characteristics of the events were unique, associating unusual site venous thromboses, mainly cerebral vein sinus thrombosis (CVST), with thrombocytopenia and sometimes disseminated intravascular coagulation (DIC). In contrast, no thromboses were reported in about 90 million subjects who had received the messenger RNA (mRNA)-based Pfizer BioNTech and only very few in those who had received the Moderna vaccine (Spikevax), although the latter had characteristics apparently dissimilar from those observed in Vaxzevria recipients, with one exception. 10 These findings suggest that the reported thrombotic complications, which have variously been called vaccineinduced prothrombotic immune thrombocytopenia (VIPIT), vaccine-induced immune thrombotic thrombocytopenia (VITT), thrombotic thrombocytopenia syndrome (TTS) and vaccine-associated thrombotic thrombocytopenia syndrome (VATTS), 2,11-13 are peculiar to adenoviral (Ad) vector-based vaccines and have led to limitations and/or temporary suspensions of the use of such vaccines in several countries.

From the most recently available UK pharmacovigilance data (July 7, 2021), CVST and other major thromboembolic events with concurrent thrombocytopenia had been reported in 147 (average age, 54 years) and 258 subjects (average age, 54 years), respectively, among an estimated 24.6 million recipients of a first dose and an estimated 22.3 million recipients of a second dose of the Vaxzevria vaccine. Thus, the overall incidence after first or unknown doses was 14 This review aims to discuss the interactions between Ad vectors and Ad-based vaccines and the hemostatic system and the hypotheses on the mechanisms triggering VITT.

Based on available data and given that VITT has been associated with Ad-vector-based vaccines, hypotheses on a direct role of the interaction between Ad and blood components can be made.

Ad are non-enveloped DNA viruses with a nucleoprotein core encapsulated by an icosahedral protein capsid from which proteinaceous fibers protrude. The C-terminal knob domain at the distal end of these fibers is responsible for virus binding to its primary cellular receptor, a 46-kDa transmembrane protein [14] [15] [16] which also functions as a receptor for Coxsackie B virus and is, therefore, called coxsackie and Ad receptor (CAR). [15] [16] [17] The high affinity binding of Ad to CAR starts receptor-mediated endocytosis. 18 Moreover, Ad have evolved other mechanisms to facilitate cell entry via recognition of the arginine-glycine-aspartate (RGD) sequence on cell surface integrins. Molecules expressed on host cell surfaces involved in cell infection include the vitronectin-binding integrins αvb3 and αvb5, 19 the fibronectin-binding integrin α5b1 20 and others, such as αVb1, 21 all characterized by a common RGD peptide sequence which is recognized by the RGD ligand in the HI fiber knob loop of the Ad penton base protein. Although the CAR is expressed in almost all tissues, including the adult nervous system and cerebral vasculature, 22,23 muscle, 24 heart 25 and the hematopoietic system, 26 its presence in platelets is debated. Othman et al. identified CAR (by flow cytometry) and its mRNA (by reverse transcriptase polymerase chain reaction) in human platelets 27 while Shimony et al. did not confirm the presence of the receptor and proposed that binding of Ad to platelets is mediated by an interaction between RGD-binding motifs of Ad and platelet αVb3 28 ( Figure 1 ). Indeed, human megakaryocytes either do not express mRNA for CAR or express it at extremely low levels (J. Rowley and A.S. Weyrich, University of Utah, personal communication). After intravenous inoculation in mice, Ad rapidly bind circulating platelets causing their activation and subsequent entrapment in liver sinusoids where virus-platelet aggregates are taken up by Küpffer cells and degraded. Platelet activation is followed by activation of blood coagulation, leading to DIC. 29 Activated platelets also release cytokines promoting endothelial cell activation with secretion of von Willebrand factor, binding of platelets to endothelial cells and the formation of platelet/leukocyte aggregates, eventually triggering the development of microthrombi in liver sinusoids. 21, 29 There is also a complex interplay between Ad and the coagulation system. In fact, the distribution and activity of Ad in blood is affected by interactions with plasma proteins, including complement and vitamin K-dependent coagulation factors, which act as opsonizing agents. Our knowledge of these interactions derives mainly from in vitro observations and it is unknown whether the interplay of Ad with coagulation proteins affects the activity of the latter. Vitamin K-dependent coagulation factors, including the anticoagulant protein C, interact with Ad-5, the most widely used Ad vector. Activated protein C is generated on endothelial cells via the interaction of protein C with the thrombin-thrombomodulin complex and the endothelial protein C receptor (EPCR). Activated protein C requires protein S to express anticoagulant activity. 30 Protein S circulates either free or associated with C4BP, a Adenovirus vaccines and thrombosis haematologica | 2021; 106 (12) regulatory protein of the complement system. 31 C4BP binds to activated platelets through mechanisms involving chondroitin sulfate expressed on activated platelets 32 and to membrane-associated protein S on platelets. 33 Interestingly, the protein C anticoagulant pathway plays a peculiar pathophysiological role in CVST. 34 Small but measurable amounts of EPCR are also found in plasma. Soluble EPCR binds both protein C and activated protein C with an affinity similar to that of membrane-bound EPCR 35 but, in contrast to the latter, it inhibits activated protein C anticoagulant activity thus limiting its ability to inactivate activated factor V, and also binds protein C impeding its activation by thrombinthrombomodulin complexes. 32, 36 An increase in soluble EPCR was observed in CVST, possibly leading to a procoagulant condition and enhanced risk of thrombosis. 37 Finally, Gas6 (encoded by the growth arrest-specific 6 gene), a vitamin K-dependent protein with 44% sequence homology with protein S but devoid of anticoagulant activity, is widely expressed in the cerebral nervous system where it is found on resting endothelial cells. Gas6 potentiates platelet activation acting on Tyro3, Axl and Mer (TAM) receptors leading to thrombus formation 38 and in vitro studies have shown that Gas6 binds to Ad enhancing their gene expression. 39 The affinity of different Ad for coagulation factors is variable, with a considerable number of Ad types unable to bind them. Ad-5, Ad-2 and Ad-16 bind strongly to factor X. 40 Moreover, the ability of Ad to bind coagulation factors is species-specific, e.g., Ad-5 binds human and mouse factor X with similar affinity, but Ad-2 binds human factor X with 10-fold lower affinity than mouse factor X. 41 Vitamin K-dependent coagulation factors VII, IX, X, and protein C mediate the binding of Ad to hepatocytes. [42] [43] [44] For instance, for Ad-5 hepatotropism is critically dependent on the ability of the Ad-5 hexon to bind factor X. In contrast, non-factor X-binding Ad, such as Ad-48 and Ad-26, do not show hepatocyte tropism. 45 The primary reason why factor X is required for Ad-5 transduction to the liver is that it protects Ad-5 from attack by complement. 46 It has been previously ascertained that components of intramuscularly-injected vaccines, including the Ad vector, are disseminated in the circulation 47 and it is thus conceivable that some of the above described activating interactions between Ad and platelets, endothelium and the blood clotting system can occur in recipients of Advector-based vaccines. However, so far no experimental evidence that this may have a role in VITT is available and actually it seems unlikely that sufficiently high circulating levels of a non-replicating Ad vector may be reached to trigger platelet activation or blood coagulation changes. In fact, it should be considered that around 2,500 billion virions/kg are required to trigger this reaction in mice and non-human primates, 29, 48 and even if all the Vaxzevria viral content were to spill-over into the blood after intramuscular administration, a concentration of 0.7 billion/kg Ad viral vectors would be reached, which is probably insufficient to activate platelets/coagulation. 49 Antibody-dependent enhancement and vaccine-associated adverse events Antibody-dependent enhancement (ADE) is an immunological form of a more general phenomenon called enhanced respiratory disease, leading to the clinical worsening of respiratory viral infections. ADE can occur either through an antibody-mediated increase of virus uptake by Fcg receptor IIa (FcgRIIa)-expressing phagocytic cells, thus facilitating viral infection and replication, or by boosting immune activation through excessive Fc-mediated immunological cell effector functions or immune Figure 1 . Hypothesized interactions between platelets and adenoviruses. Adenoviruses (Ad) induce platelet activation either by binding to platelet coxsackie and adenovirus receptor (CAR) or to plateletsurface integrins, such as αvb3 or α5b1. Moreover, circulating Ad-elicited IgG or immune complexes may directly activate platelets through FcgRIIa. Gas6 exposed by cerebral vein endothelial cells may bind Ad and activate platelets acting on Tyro3, Axl and Mer (TAM) receptors. Ad may also bind CAR expressed by cerebral vein vessels in this way activating endothelial cells which in turn may elicit platelet activation.

complex formation with consequent increase of inflammation and immunopathology. 50 Both ADE pathways can occur when non-neutralizing antibodies or antibodies at sub-neutralizing levels bind to viral antigens without blocking or clearing the infection. ADE has been reported for vaccines against both severe acute respiratory syndrome corona virus (SARS-CoV) and Middle East respiratory syndrome corona virus (MERS-CoV) in vitro and in animal models. 50 The cytoplasmic tail of FcgRIIa activates the protein-tyrosine kinases Src 51,52 and Syk. [53] [54] [55] Srcdependent signaling has been shown to be crucial for ADE triggered by Ebola virus, enhancing viral uptake into cells and thus worsening the infection. 56 Circulating antibodies activating platelet IgG FcgRIIa may be key determinants of a host response leading to uncontrolled platelet aggregation and thrombosis. Studies in transgenic mice expressing human FcgRIIa on platelets showed that the administration of anti-CD9 antibodies caused thrombosis accompanied by platelet consumption, a response that was absent in mice lacking the receptor. 57 The clinical relevance of this pathway for thrombotic disorders in humans is confirmed by the observation that FcgRIIa expression is higher in patients with stroke 58 and that relatively common FcgRIIa polymorphisms are associated with increased risk of thrombosis in patients with heparin-induced thrombocytopenia (HIT). 59 Immuno-complex formation, complement deposition and local immune activation are likely mechanisms triggered by SARS-CoV-2 vaccine. Furthermore, preexisting antibodies to coronavirus strains endemic in humans could mediate ADE by facilitating cross-reactive recognition of SARS-CoV-2 in the absence of viral neutralization. 60 Interestingly, compared to Ad-5 and Ad-6, chimpanzee adenoviruses (ChAd) are much less frequently neutralized by pre-existing antibodies present in humans. The prevalence of vector-neutralizing antibodies against Y25, now renamed ChAdOx1, the vector of the Vaxzevria vaccine, in human sera from British and Gambian adults was found to be 0% (n=100) and 9% (n=57), respectively. 61 The presence of these antibodies in rare patients in Europe might theoretically represent one potential mechanism triggering ADE, and possibly VITT, in vaccine recipients but no data on this are available yet.

Despite the above hypotheses, preliminary in vitro evidence suggests that serum from convalescent COVID-19 patients does not induce either enhancement of SARS-CoV-2 infection or innate immunity responses in human macrophages, indicating that ADE may not be involved in the immune-pathological processes associated with COVID-19 infection or immunization. 62

Ad vectors have been used therapeutically for their ability to transduce and deliver transgenes to different cell types. However, for these indications the clinical use of Ad vectors has been limited to a few tens of patients and the main concerns have been the development of humoral and cellular immunity occurring upon repeated administration and/or the possible neutralization of the vector by pre-existing immunity against the virus, while little attention had been paid to the possible interactions of Ad vectors with platelets and the blood clotting system.

The first use of Ad vectors for gene therapy of inherited disorders or to treat neoplasia dates back to the 1990s. An analysis of the risks associated with the use of Ad-vectored gene therapies among 90 individuals who received 140 administrations for various diseases (cystic fibrosis, metastatic colorectal cancer, cardiovascular disease), showed that 13 deaths were recorded. The authors concluded that none was linked to the Ad vector. 63 The reported hematologic abnormalities were decreased hemoglobin, leukocytosis, thrombocytopenia, and prolongation of the activated partial thromboplastin time (aPTT), with no cases of DIC. 63 It is, however, puzzling that a recently European Medicines Agency-licensed Ad-vectored gene therapy for spinal muscular atrophy received a warning about the possible risk of thrombotic microangiopathy based on the reporting of five cases in treated infants (https://www.ema.europa.eu/en/medicines/human/EPAR/zolgensma).

Beside SARS-CoV-2, Ad vectors have been used for the preparation of other vaccines, including the ChAdOx1vectored vaccines for MERS-CoV and Chikungunya; with only a few hundred volunteers having received these vaccines up to June 2020, 64 no excess of thrombotic events had been noted. 65 Even for the Ebola vaccination campaign, the largest previous example of large-scale vaccination using an Ad vector, a maximum of around 200,000 volunteers were treated, with only one vena cava thrombosis reported (Table 1) . However, it may be extremely difficult to prove that adverse events following immunization are caused by the vaccine itself when their occurrence is extremely rare (https://www.nature.com/articles/d41586-021-00880-9. Accessed on April 9, 2021).

Except for common mild/moderate reactome reactions, the most frequently recorded adverse events in clinical trials were hematologic (e.g., mild hemoglobin decrease, thrombocytopenia, leukopenia) the majority of which recovered a few days or weeks after vaccination. The extent and rate of hematologic adverse events associated with Ad-vectored vaccines are summarized in Table 1 . Occasional abnormalities of coagulation were reported, with prolongation of the aPTT, possibly due to the development of transient antiphospholipid antibodies. Thrombotic events were rare both for human and nonhuman Ad-vectored vaccines. One case of phlebitis was observed among 114 volunteers who received a recombinant, replication-defective Ad-5-vectored vaccine expressing human immunodeficiency virus (HIV)-1 antigenic proteins. 66 Another case of deep vein thrombosis was observed among 58 volunteers after administration of a recombinant, replication-defective Ad-35-vectored vaccine expressing HIV-1 antigens. 67 Both events were considered unrelated to the vaccine.

A systematic review identified 200 clinical studies on active immunization against SARS-CoV-2. The second most used vaccine platform, after mRNA-based vaccines, was represented by Ad vectors (24%). 68 Concerning chimpanzee Ad-vectored vaccines (ChAdOx1 nCoV-19) , neutropenia was the most common hematologic abnormality (Table 1) . Across all studies, vaccines had a good safety profile with no difference in severe reactions between study arms. 69, 70 In a phase III trial with a recombinant, replication-incompetent human Ad-26 vector encoding the SARS-CoV-2 spike protein, with 43,783 participants, 11 venous thromboembolic events were observed in the vaccine group compared to three in the placebo group (Table 1) ; however, most subjects had underlying medical conditions that might have contributed to these events. In the vaccine group there were six cases of lower leg deep venous thrombosis and four cases of pulmonary embolism. Interestingly, however, a CVST, with cerebral hemorrhage and thrombocytopenia, occurred 21 days after vaccination in a 25-year-old male who had multiple predisposing factors, including preexisting cerebral sigmoid sinus stenosis and infection from an unknown pathogen. Subsequent testing identified anti-PF4 antibodies at the time of the event. The patient recovered. 71 In a phase III controlled randomized clinical trial with a recombinant Ad-26-vectored and a recombinant Ad-5vectored vaccine (Sputnik V) among 16,501 participants, ten vascular events (0.061%) were observed including: one deep vein thrombosis (0.006%), one transient ischemic attack (0.006%), one cerebral circulation failure (0.006%), one vascular encephalopathy (0.00659%) and two acute myocardial infarctions (0.012%) (4 additional events were non thrombotic) 72 (Table 1 ).

When the anti-SARS-CoV-2 vaccination campaign was well underway worldwide a few cases of spontaneous, severe thromboembolic events in otherwise healthy subjects began to be reported, leading to a pause in the administration of the Vaxzevria vaccine in several European countries (https://www.ema.europa.eu/en/news/emas-safetycommittee-continues-investigation-covid-19-vaccine-astrazenecathromboembolic-events). Soon after several case reports were published, mainly concerning young females, with new ones continuing to accrue, although many were not subject to rigorous central review and with anti-PF4 antibodies measured using disparate methods, not allowing to conclude that all were typical VITT cases Up to July 17, 2021, 105 such cases with two Ad-vectored vaccines had been published ( Table 2) with some common clinical features characterizing a new syndrome, including thrombocytopenia, often severe, venous thrombosis at unusual sites, in particular of the cerebral sinuses but also of the splanchnic veins, frequently associated with thromboses in multiple sites, both venous and arterial, and sometimes DIC combined with hemorrhage. A comparative evaluation of the clinical characteristics of the published ChAdOx1 or Ad26.CoV2.S VITT cases suggests that while clinical symptoms are comparable, Ad26.CoV2.S-associated cases show more thrombosis and intracerebral hemorrhage, lower D-dimer and less altered aPTT, but a similar mortality. 9 In a recent, large nationwide healthcare register-based study in Denmark and Norway involving 281,264 ChAdOx1-S-vaccinated subjects aged 18-65 and as controls the entire agematched populations of the two countries studied in the period 2016-2019, the standardized morbidity ratio for CVST was 20.25 (8.14-41.7), with an excess of 2.5 events per 100,000 vaccinations, particularly evident in women 18-44 years old, 73 confirming the crucial relationship between Vaxzevria administration and occurrence of VITT. The catastrophic syndrome, burdened by a 20-50% mortality rate, has the time course and tumultuous evolution of an acute immunological reaction and indeed three groups of investigators identified, in several of their patients, circulating antibodies to PF4/heparin complexes using an enzyme-linked immunosorbent assay (ELISA) and a heparin-induced platelet activation assay, 2-4 and thus proposed that this disorder is a peculiar form of autoimmune HIT.

HIT is a rare immune-mediated adverse drug reaction that may occur after exposure to heparin. Circulating heparin binds to PF4, a positively charged platelet protein released in plasma upon activation. PF4 normally binds to negatively charged glycosaminoglycans on the endothelium, displacing antithrombin and thus activating coagulation. However, PF4 binds with greater affinity to heparin, forming heparin/PF4 complexes which become neoantigens inducing the formation of autoantibodies. Heparin-PF4-IgG immune complexes in turn bind to platelet FcgRIIA receptors causing activation, aggregation, and additional release of PF4, with ignition of a positive feedback loop leading to further platelet activation and consumption. Moreover, these complexes also activate monocytes, which release tissue factor, thus promoting concomitant activation of coagulation.

HIT is a potentially fatal condition, associated with the development of arterial or venous thrombosis. 74 Thrombocytopenia occurs in more than 85% of HIT patients and is usually of moderate severity, with median platelet counts of approximately 50-60×10 9 /L, although values <20×10 9 /L can be found in approximately 10% of cases. Typically, the platelet count starts to decrease 5-10 days after initiation of heparin, but early-onset thrombocytopenia (rapid-onset), within 24 h of exposure, can develop in 25-30% of cases if patients have been treated with heparin in the preceding 3 months. 75 Thromboembolic complications occur in 35-75% of HIT patients and are usually severe. They can be venous (i.e., deep vein thrombosis and pulmonary embolism, but rarely also CVST or splanchnic thrombosis), arterial (ischemic stroke, myocardial infarction, acute occlusion of limb arteries) or microvascular (digital infarction). 76 Recently, another clinical picture not triggered by exposure to heparin has been recognized and defined as autoimmune HIT. 77 The main characteristic of this condition is the presence of circulating antibodies able to activate platelets also in the absence of heparin. Polyanion molecules potentially involved in the development of autoimmune HIT are typically bacteria and virus components, hypersulfated chondroitin sulfate, DNA and RNA and polyphosphates. 78 Patients with this syndrome show slightly different clinical features from those with classical HIT, including severe thrombocytopenia (<20x10 9 /L), sometimes in combination with DIC, microvascular thrombosis and CVST in up to 40% of cases. 78 From a therapeutic standpoint, besides the indication for an alternative anticoagulant, valid also for HIT, the intravenous Adenovirus vaccines and thrombosis haematologica | 2021; 106 (12) administration of high doses of IgG in combination with steroids has been proposed for autoimmune HIT. 77 Very early after the first reports of unusual types of thrombosis associated with the Vaxzevria vaccine, a German group suggested a tentative pathogenic mechanism underlying these rare events, which they named VIPIT, based on findings in nine cases of previously vac-cinated subjects. 11 Indeed, in sera from four of these subjects the investigators detected antibodies to PF4/heparin complexes using an enzyme-immunoassay; these antibodies were inhibited by the addition of high concentrations of heparin (i.e., 100 U/mL), and the sera were able to activate washed control platelets when either PF4 or the Vaxzevria vaccine was added in vitro to the samples. 11 The reference numbers preceded by an S refer to references listed in the Online Supplementary Material. a One patient had only thrombocytopenia. b The patient was reported to be thrombocytopenic. c Bilateral superior ophthalmic vein thrombosis. d An eighth case included in this series was reported in reference S10. e No thromboses were detected (possibly not a VITT). f One of the cases did not have thrombosis. g One case was initially treated with full dose low molecular weight heparin, with no apparent worsening, and was then changed to a non-heparin anticoagulant upon identification of positive anti PF4 antibodies. h The platelet count was stated to be normal. i Ages are reported as ranges, not individual. NR: not reported; PF4: platelet factor 4; VITT: vaccine-induced immune thrombotic thrombocytopenia.

The investigators showed that platelet activation was triggered by FcgRIIA stimulation, because an FcgRIIAblocking antibody prevented this phenomenon. Given the absence of previous exposure to heparin, the authors suggested a condition resembling autoimmune HIT. More recently, in a preliminary report published in a non-peerreviewed internet repository, the German group went on to suggest that the Ad vector and/or some protein components of the Vaxzevria vaccine activate platelets to release PF4 which then forms complexes with virus proteins and other anionic constituents of the vaccine, generating neoantigens against which antibodies develop and induce strong platelet activation via FcgRIIa stimulating granulocyte activation with NETosis and ultimately catastrophic thrombosis. 47 The presence of EDTA in the vaccine would favor vascular leakage at the inoculation site, facilitating dissemination of the vaccine components in blood ( Figure 2 ). Very recently a study using alanine scanning mutagenesis explored the binding sites on PF4 of antibodies isolated from patients with VITT or with classical HIT. While the binding of VITT anti-PF4 antibodies was restricted to eight surface amino acids, all located within the heparinbinding site of PF4, HIT anti-PF4 antibodies bound amino acids corresponding to two different sites on PF4; moreover, VITT antibodies had a stronger binding response than HIT antibodies. The authors concluded that VITT antibodies mimic the effect of heparin by binding to a similar site on PF4, allowing PF4 tetramers to cluster with the formation of immuno-complexes which, in turn, cause FcγRIIa-dependent platelet activation. 79 These peculiar characteristics explain why the identifi-cation of VITT requires different tests from those needed for the identification of classical HIT. 80 In suspected VITT anti-PF4 antibodies can be identified by an ELISA, but not by other rapid immunological assays typically positive in HIT, such as the STic ® Expert HIT kit, latex immunoassays and chemiluminescence-based assays, 4, 80 and it is better characterized by a heparin-induced platelet aggregation (HIPA) or PF4-induced platelet activation (PIPA) test. 2, 4, 81 It should be noted that no single ELISA method detected all possible/probable VITT cases. 82 While the autoimmune HIT hypothesis is important and provides the basis for understanding this novel catastrophic autoimmune thrombotic syndrome, several aspects do not fit completely with the clinical presentation of VITT and various issues remain unclarified.

First, among the reported VITT cases in which anti-PF4 antibodies were measured, in a few they were negative (Table 2) . 4, 5 Second, it is expected that treatment with heparin would worsen the clinical evolution of these patients, and indeed it is generally cautiously recommended not to use this drug. 2, 4, 13 However, in around one fourth of the heparin-treated cases, the use of this anticoagulant was successful (Table 2) . Third, while VITT resembles autoimmune HIT in several respects, the latter is not as frequently associated with CVST and rarely with DIC, and the above summarized mechanistic hypothesis does not explain the preferential localization of the venous thrombotic events in the cerebral and splanchnic circulations.

Other unclear aspects are its relatively precocious onset, as early as 4 days after vaccination, which seems too soon to generate high-titer, class-switched, high-affin-Adenovirus vaccines and thrombosis haematologica | 2021; 106(12) 3041 ity anti-PF4 antibodies. Furthermore, there is no evidence that the anti-PF4 antibodies isolated from patients with VITT cause thrombosis and thrombocytopenia in animal models. 49, 83 Finally, recent observations show that 1.2% to 8.0% of subjects receiving a first dose of Vaxzevria develop circulating anti-PF4 antibodies while the prevalence of VITT ranges from 0.0006% to 0.00125%. 84, 85 Other possible pathogenic mechanisms of vaccine-induced immune thrombotic thrombocytopenia

Very recently a preliminary report, published in a nonpeer-reviewed repository, provided an interesting alternative potential pathogenic mechanism of VITT. 86 COVID-19 is caused by SARS-CoV-2 which is a single-strand RNA virus that is translated and replicates only in the cytosol of infected cells in the absence of processes which are necessary when nuclear-encoded genes are transcribed, and in particular of mRNA splicing. Nuclear encoded genes have intronic sequences, thus their transcripts require splice reactions at consensus RNA sequences to eliminate them. When an Ad-vectored viral RNA sequence is administered the vector infects host cells, adenoviral DNA enters the nucleus and is then transcribed by the host transcription machinery. However, the viral piece of DNA deriving from the SARS-CoV-2 virus is not optimized to be transcribed into the nucleus and its open reading frame may thus be disrupted by arbitrary splice events. These splice events would produce shorter spike protein variants, including forms missing the C-terminal membrane anchor, thus leading to soluble circulating spike protein molecules. The soluble spike protein may cause a strong activation of endothelial cells expressing ACE2. 87 Moreover, when the host immune system starts to produce antibodies against the spike protein, endothelial cells binding soluble spike would also be decorated by these antibodies, triggering a strong inflammatory reaction through antibody-dependent or complement-dependent cytotoxicity, thus eliciting VITT. With this hypothesis, the preferential involvement of cerebral haematologica | 2021; 106(12) Figure 3 . A suggested clinical surveillance and diagnostic approach to suspected vaccine-induced immune thrombotic thrombocytopenia. Subjects receiving the Vaxzevria and the Janssen vaccines who develop new onset headache, especially if severe or with unprecedented characteristics, and/or associated with other clinical manifestations (blurred vision, gait disturbance, focal neurological symptoms and/or abdominal pain, vomiting, bloody stool, shortness of breath, petechiae or ecchymoses) should be referred for immediate laboratory evaluation (platelet count and D-dimer measurement). If thrombocytopenia is detected they should undergo anti-PF4 antibody testing, imaging and evaluation for cerebral vein sinus thrombosis, splanchnic vein thrombosis or pulmonary embolism. If confirmed, therapy for vaccine-induced immune thrombocytopenic thrombosis (VITT) should be immediately started according to the statement from the Italian Society for the Study of Haemostasis and Thrombosis (SISET) 13 Although almost one quarter of the reported patients with VITT in whom unfractionated or low-molecular weight heparin was used apparently responded well to treatment (Table 2) , subjects positive for anti-PF4 antibodies, as determined by a heparin-induced platelet aggregation (HIPA) test and/or enzyme-linked immunosorbent assay, or who have not been tested should, for prudence, be treated with alternative anticoagulants, until new information becomes available. In subjects in whom these anti-PF4 antibodies do not cross-react with heparin, as shown by a HIPA test in the presence of a low concentration of heparin, the use of heparin as anticoagulant may be allowed. ELISA: enzyme-linked immunosorbent assay; MRI: magnetic resonance imaging; CT: computed tomography; US: ultrasound. veins could be explained by the non-unidirectional blood flow in these vessels due to the lack of venous valves, with prolonged residence time of the soluble spike protein in this district depending on body posture or when sleeping. The immunological part of this hypothesis is also in agreement with the apparent higher prevalence of VITT in young women, because they have stronger immune reactions than men and older people. To explain the rarity of VITT the authors hypothesized that only some individuals, due to specific major histocompatibility complex combinations, are not able to produce neutralizing anti-spike antibodies which would instead prevent the binding of soluble spike to endothelial ACE2 and its ominous consequences in most vaccine recipients. This hypothesis was partly validated by the identification through in silico analysis of potential splice sites in the AstraZeneca and Johnson&Johnson codon-optimized spike opening frames and by in vitro studies with HeLa cells showing that vaccine-transduced cells generate transcripts smaller than the full spike protein. It would also explain why VITT has not been reported with mRNA vaccines, which release their cargo mRNA directly into the host cells' cytosol where it is translated into spike protein without undergoing splicing reactions. Finally, it would account for why the incidence of VITT seems to be lower with the Johnson&Johnson vaccine than with the AstraZeneca one, given that the latter carries more splice donor sequences than the former. 86 Additional hypotheses on the mechanisms triggering VITT include a genetically determined enhanced expression of FcgRIIa in susceptible subjects, an altered glycosylation state of IgG produced in response to vaccination in some individuals making these antibodies more reactive to platelet FcgRIIa, 49 the leakage of the Ad vector into the circulation and/or the prior presence of cross-reactive antibodies to other coronaviruses forming immune complexes activating platelets. 88 However, the hypothesis that VITT develops in subjects with previous, not apparent SARS-CoV-2 infection with prior circulating IgG antibodies against the spike protein able to activate platelet FcgRIIa 89 should be excluded by the observation that most VITT subjects tested for previous or recent COVID-19 infections were negative. Excessive transcription of the spike protein, which would then activate platelets binding to ACE2, 90 and vaccine-induced expression of the spike by megakaryocytes and platelets, leading to a thrombo-inflammatory storm, 49, 91 have also been proposed. Another hypothesis starts from the observation that both the ChAdOx1 and the Ad26.CoV2.S vaccines use polysorbate 80 as an excipient. Polysorbate 80 is a non-anionic surfactant that crosses the blood-brain barrier and enhances microparticle uptake by endothelial cells. Therefore leakage of Ad vector and polysorbate into the circulation and the spike protein produced by vaccination could preferentially localize in brain vessels triggering endothelial activation. 92 However, considering that VITT usually develops at least 1 week after vaccination, it is very unlikely that circulating Ad vector or vaccine excipients would still be present in blood, making the alterna-tive explanations, and in particular an immunological reaction, more likely.

At least two Ad vector-based vaccines against SARS-CoV-2 have been associated with an excess rate of a special form of catastrophic thrombotic syndrome associated with thrombocytopenia of likely autoimmune origin, not observed so far with mRNA-based vaccines, suggesting that the vectors may play a role in eliciting it. Several elements of the Ad vectors and/or of vaccine composition may theoretically interact with platelets, the endothelium and the blood clotting system precipitating this rare complication. However, the exact sequence of events leading to the development of this syndrome and, most importantly, the reason why it evolves in only very few subjects without apparent predisposing factors remain to be clarified.

It is clear that our understanding of the pathogenesis of VITT is far from complete and that more mechanistic studies are required to clarify it and, it is hoped, to identify risk factors predictive of its development. What is quite likely is that the Ad vector-based vaccine triggers an immunological reaction which, for unknown reasons, in some rare subjects involves especially blood platelets, and possibly some peculiar vascular endothelial districts such as those of the cerebral and splanchnic veins, precipitating the catastrophic vaccine-induced autoimmune thrombocytopenia thrombosis syndrome. Awareness of this condition and the prompt identification/evaluation of affected patients may lead to successful treatment and recovery ( Figure 3) . 13, 84 COVID-19 continues to be a serious global health problem and vaccination against SARS-CoV-2 is the most effective way of limiting illness and death due to the pandemic. Based on the current available information, and in light of the relative rarity of VITT, the benefits of vaccination clearly outweigh the potential risks (https://www.ema.europa.eu/en/documents/dhpc/direct-healthcare-professional-communication-dhpc-zolgensma-onasemnogene-abeparvovec-risk-thrombotic_en.pdf). However, once the global pandemic begins to retreat, the relative importance of even small risks will increase, 65 making it critically important to understand the mechanisms leading to this ominous thrombotic syndrome, to identify the prognostic factors for its development and to define the best management strategies. 13

No conflicts of interest to disclose.

PG, VDS, AT, RM, SM and FR wrote the review article.

Thrombotic thrombocytopenia after Ad26.COV2.S vaccination

Thrombotic thrombocytopenia after ChAdOx1 nCov-19 vaccination

Thrombosis and thrombocytopenia after ChAdOx1 nCoV-19 vaccination

Pathologic antibodies to platelet factor 4 after ChAdOx1 nCoV-19 vaccination

Bilateral superior ophthalmic vein thrombosis, ischaemic stroke, and immune thrombocytopenia after ChAdOx1 nCoV-19 vaccination

Pulmonary embolism and thrombocytopenia following ChAdOx1 vaccination

Safety monitoring of the Janssen (Johnson & Johnson) COVID-19 vaccine -United States

A new enemy is emerging in the fight against the SARS-CoV-2 pandemic

Comparison of vaccine-induced thrombotic events between ChAdOx1 nCoV-19 and Ad26.COV.2.S vaccines

Thrombosis with thrombocytopenia after the messenger RNA-1273 vaccine

A prothrombotic thrombocytopenic disorder resembling heparin-induced thrombocytopenia following coronavirus vaccination

US case reports of cerebral venous sinus thrombosis with thrombocytopenia after Ad26.COV2.S vaccination

Management of cerebral and splanchnic vein thrombosis associated with thrombocytopenia in subjects previously vaccinated with Vaxzevria (AstraZeneca): a position statement from the Italian Society for the Study of Haemostasis and Thrombosis (SISET)

A monoclonal antibody specific for the cellular receptor for the group B coxsackieviruses

Isolation of a common receptor for coxsackie B viruses and adenoviruses 2 and 5

Structural analysis of the mechanism of adenovirus binding to its human cellular receptor CAR

Identification of a conserved receptor-binding site on the fiber proteins of CAR-recognizing Adenoviridae

Stepwise dismantling of adenovirus 2 during entry into cells

Integrins alpha v beta 3 and alpha v beta 5 promote adenovirus internalization but not virus attachment

Integrin a5b1-mediated adenovirus infection is enhanced by the integrin-activating antibody TS2/16

AD receptors

The coxsackievirus-adenovirus receptor protein as a cell adhesion molecule in the developing mouse brain

The coxsackievirus-adenovirus receptor reveals complex homophilic and heterophilic interactions on neural cells

Expression of the primary coxsackie and adenovirus receptor is downregulated during skeletal muscle maturation and limits the efficacy of adenovirus-mediated gene delivery to muscle cells

Human coxsackie-adenovirus receptor is colocalized with integrins alpha(v)beta(3) and alpha(v)beta(5) on the cardiomyocyte sarcolemma and upregulated in dilated cardiomyopathy: implications for cardiotropic viral infections

Maturation and lineage-specific expression of the coxsackie and adenovirus receptor in hematopoietic cells

Adenovirusinduced thrombocytopenia: the role of von Willebrand factor and P-selectin in mediating accelerated platelet clearance

Analysis of adenoviral attachment to human platelets

Adenovirus-platelet interaction in blood causes virus sequestration to the reticuloendothelial system of the liver

Activated protein C: biased for translation

Contribution of chondroitin sulfate A to the binding of complement proteins to activated platelets

Binding of anticoagulant vitamin K-dependent protein S to platelet-derived microparticles

Prognosis of cerebral vein and dural sinus thrombosis: results of the International Study on Cerebral Vein and Dural Sinus Thrombosis (ISCVT)

Identification of functional endothelial protein C receptor in human plasma

Endothelial cell protein C receptor and the risk of venous thrombosis

Soluble thrombomodulin and endothelial cell protein C receptor levels in patients with cerebral venous and sinus thrombosis

Gas6 receptors Axl, Sky and Mer enhance platelet activation and regulate thrombotic responses

Inhibition of type I interferon responses by adenovirus serotype-dependent Gas6 binding

Interaction of adenovirus with antibodies, complement, and coagulation factors

Coagulation factor IX mediates serotypespecific binding of species A adenoviruses to host cells

Influence of coagulation factor zymogens on the infectivity of adenoviruses pseudotyped with fibers from subgroup D

Multiple vitamin K-dependent coagulation zymogens promote adenovirus-mediated gene delivery to hepatocytes

Adenovirus binding to blood factors results in liver cell infection and hepatotoxicity

Adenovirus serotype 5 hexon is critical for virus infection of hepatocytes in vivo

The relevance of coagulation factor X protection of adenoviruses in human sera

Towards understanding ChAdOx1 nCov-19 vaccine-induced immune thrombotic thrombocytopenia (VITT)

Acute toxicity after high-dose systemic injection of helper-dependent adenoviral vectors into nonhuman primates

Post-adenoviral-based COVID-19 vaccines thrombosis: a proposed mechanism

Antibody-dependent enhancement and SARS-CoV-2 vaccines and therapies

Association of immunoglobulin G Fc receptor II with Src like protein-tyrosine kinase Fgr in neutrophils

Physical and functional association of Src-related protein tyrosine kinases with Fc gamma RII in monocytic THP-1 cells

Cross-linking of Fc gamma receptor I (Fc gamma RI) and receptor II (Fc gamma RII) on monocytic cells activates a signal transduction pathway common to both Fc receptors that involves the stimulation of p72 Syk protein tyrosine kinase

Fcγ receptors as regulators of immune responses

Fcg receptor signaling in phagocytes

Fcg-receptor IIa-mediated Src signaling pathway is essential for the antibodydependent enhancement of Ebola virus infection

Thrombosis and shock induced by activating antiplatelet antibodies in human Fc gamma RIIA transgenic mice: the interplay among antibody, spleen, and Fc receptor

Potential role of platelet FcgammaRIIA in collagen-mediated platelet activation associated with atherothrombosis

Heparin-induced thrombocytopenia: the role of platelets genetic polymorphisms

Is COVID-19 receiving ADE from other coronaviruses? Microbes Infect

Vaccine vectors derived from a large collection of simian adenoviruses induce potent cellular immunity across multiple species

No evidence for human monocyte-derived macrophage infection and antibody-mediated enhancement of SARS-CoV-2 Infection

Analysis of risk factors for local delivery of low-and intermediate-dose adenovirus gene transfer vectors to individuals with a spectrum of comorbid conditions

Safety and immunogenicity of a candidate Middle East respiratory syndrome coronavirus viral-vectored vaccine: a dose-escalation, open-label, non-randomised, uncontrolled, phase 1 trial

What's the future of vaccines linked to rare clotting disorders? Science breaks down the latest

Safety and immunogenicity study of multiclade HIV-1 Adenoviral vector vaccine alone or as boost following a multiclade HIV-1 DNA vaccine in Africa

A phase I double blind, placebo-controlled, randomized study of a multigenic HIV-1 adenovirus subtype 35 vector vaccine in healthy uninfected adults

Current clinical trials protocols and the global effort for immunization against SARS-CoV-2. Vaccines (Basel)

ChAdOx1 nCoV-19 vaccination prevents SARS-CoV-2 pneumonia in rhesus macaques

Oxford COVID Vaccine Trial Group. Safety and immunogenicity of the ChAdOx1 nCoV-19 vaccine against SARS-CoV-2: a preliminary report of a phase 1/2, single-blind, randomised controlled trial

Safety and efficacy of single-dose Ad26.COV2.S vaccine against Covid-19

Safety and efficacy of an rAd26 and rAd5 vector-based heterologous primeboost COVID-19 vaccine: an interim analysis of a randomized controlled phase 3 trial in Russia

Arterial events, venous thromboembolism, thrombocytopenia, and bleeding after vaccination with Oxford-AstraZeneca ChAdOx1-S in Denmark and Norway: population based cohort study

Heparin induced thrombocytopenia: position paper from the Italian Society on Thrombosis and Haemostasis (SISET)

American Society of Hematology 2018 guidelines for management of venous thromboembolism: heparin-induced thrombocytopenia

Clinical features of heparin-induced thrombocytopenia including risk factors for thrombosis. A retrospective of 408 patients

Autoimmune heparin-induced thrombocytopenia

Spontaneous HIT syndrome: knee replacement, infection, and parallels with vaccine-induced immune thrombotic thrombocytopenia

Antibody epitopes in vaccineinduced immune thrombotic thrombocytopenia

Laboratory testing for suspected COVID-19 vaccine-induced (immune) thrombotic thrombocytopenia

Beneficial effect of exogenous platelet factor 4 for detecting pathogenic heparininduced thrombocytopenia antibodies

Evaluation of laboratory assays for antiplatelet factor 4 antibodies after ChAdOx1 nCOV-19 vaccination

SARS-CoV-2 Vaccineinduced immune thrombotic thrombocytopenia

An observational study to identify the prevalence of thrombocytopenia and anti-PF4/polyanion antibodies in Norwegian health care workers after COVID-19 vaccination

Frequency of positive anti-PF4/polyanion antibody tests after COVID-19 vaccination with ChAdOx1 nCoV-19 and BNT162b2

Vaccineinduced Covid-19 spike open reading frame results in spike protein variant that may cause thromboembolic events in patients immunized with vector-based vaccine. Research Square

SARS-CoV-2 spike protein impairs endothelial function via downregulation of ACE 2

Proinflammatory IgG Fc structures in patients with severe COVID-19

Hypotheses behind the very rare cases of thrombosis with thrombocytopenia syndrome after SARS-CoV-2 vaccination

SARS-CoV-2 interacts with platelets and megakaryocytes via ACE2-independent mechanism

A double-edged sword: antibody-mediated procoagulant platelets in COVID-19

Thrombotic thrombocytopenia after ChAdOx1 nCoV-19 vaccination

This work was supported in part by grants from Fondazione Cassa di Risparmio di Perugia (#19663 (2020.0508) and FISR