Keywords SARS-CoV-2 - COVID-19 - platelets - thrombosis - vaccine
Despite huge efforts underway globally for preventing or limiting the spread of severe
acute respiratory coronavirus disease 2 (SARS-CoV-2), the ongoing coronavirus disease
2019 (COVID-19) pandemic outbreak still appears virtually unstoppable. As already
seen for many other infectious diseases,[1 ] COVID-19 vaccination is a cornerstone and a mainstay for limiting viral spread,
especially for averting hospitalizations, need for intensive care, and deaths worldwide.[2 ] Nonetheless, as for most other therapies, vaccination is not completely free from
side effects and even adverse events, which are typically classified in terms of location
(local, systemic) and severity (mild, modest, and severe).[3 ]
The sudden emergence of cases of atypical thromboses in close proximity to COVID-19
vaccination was first reported almost simultaneously by Greinacher et al[4 ] and Schultz et al.[5 ] Both publications reported an unexpectedly high burden of cerebral venous thrombosis
(CVT) in prevalently young women who received the AstraZeneca ChAdOx1 (Vaxzevria)
vaccine. These case series (11 and 5, respectively) represented both high mortality
(55 and 60%, respectively) and morbidity. A review of cases as of the end of May,
2021, identified a case fatality of nearly 40% (31/79) in subsequent publications.[6 ]
A plethora of overlapping definitions for this syndrome has since been provided by
different organizations including TTS (thrombotic thrombocytopenia syndrome) and VATT
(vaccine associated thrombosis with thrombocytopenia),[6 ] although the term “vaccine-induced immune thrombotic thrombocytopenia” (VITT) seems
to have been primarily selected, given that the key elements include acute thrombosis,
thrombocytopenia (usually with platelet count <150 × 109 /L), elevated D-dimer values along with a positive ELISA (enzyme-linked immunosorbent
assay) test for platelet factor 4 antibodies, all developing soon after COVID-19 vaccination
(typically within 15–20 days)[7 ] ([Table 1 ]).
Table 1
Leading clinical and laboratory features of vaccine-induced thrombocytopenia and thrombosis
(VITT) in recipients of adenovirus-based COVID-19 (coronavirus disease 2019) vaccines
Characteristic
Description
Patients
Mostly young individuals (aged ≤60 y); potential predominance of women
Onset
1–3 wk after receiving adenovirus-based COVID-19 vaccines
Thrombosis
Mostly cerebral venous (sinus) thrombosis, occasionally accompanied by systemic venous
(especially splanchnic and pulmonary) and arterial thrombosis
Pro-thrombotic conditions
Not evident in the majority of cases
Laboratory data
Low platelet count (<150 × 109 /L)
Elevated plasma D-dimer level (>0.5 mg/L)
Positive anti-PF4 antibodies ELISA
Abbreviations: COVID-19, coronavirus disease 2019; PF4, platelet factor 4.
Once potential causes of pseudothrombocytopenia have been ruled out,[8 ] the combination of thrombosis, thrombocytopenia, and raised D-dimers, occurring
soon after COVID-19 vaccination, is the key for identifying these patients as potentially
having an episode of VITT. The clinical signs and symptoms of CVT mostly include the
onset of new and severe headache, which cannot be alleviated by usual analgesics,
often accompanied by blurred vision, nausea or vomiting, dysarthria, weakness, drowsiness
or even seizures.[9 ] Some patients may also have additional symptoms, such as shortness of breath, chest
or abdominal pain and leg swelling, potentially associated with additional thromboses
such as pulmonary thrombosis, splanchnic vein thrombosis, and deep vein thrombosis.
The potentially catastrophic picture of VITT has been well-described in a post-mortem
analysis of two VITT cases by Pomara et al.[10 ] The main features in both cases were massive thrombosis that especially involved
the portal, splenic, and superior mesenteric veins in one case and the superior sagittal
sinus in the other. Microscopic findings revealed the presence of vascular thrombotic
occlusions within the microcirculation of multiple organs, with concomitant presence
of inflammatory infiltrates. Platelet aggregates were also found to diffusely lining
the endothelial layer of small and medium size vessels, with signs of platelet phagocytosis
by myeloid elements in the vascular spaces. The autopsies of two additional patients
who died with VITT, and autopsies performed by Althaus et al[11 ] demonstrated the presence of both arterial and venous thromboses in various organs,
with complete thrombotic obstruction of cerebral sinuses, bilateral pulmonary embolism,
occlusion of glomerular capillaries by hyaline thrombi containing fibrin and platelets.
Epidemiology of VITT and Other Vaccine Associated Thrombocytopenias
Epidemiology of VITT and Other Vaccine Associated Thrombocytopenias
Although the epidemiological picture of VITT remains largely undefined, the frequency
of both vaccine-associated thrombocytopenia and vaccine-associated idiopathic thrombocytopenic
purpura (ITP) has been assessed in a Scottish population-based study, including people
who received both adenovirus-based AstraZeneca and mRNA-based Pfizer vaccines.[12 ] The cumulative risk of developing thrombocytopenia was found to be significantly
higher (relative risk, 2.8; 95% CI, 1.39–5.67) in recipients of AstraZeneca (but not
Pfizer) vaccine than in non-vaccinated individuals 0 to 6 days after vaccination,
while that of developing ITP was between 4.6 and 14.1, higher again in recipients
of AstraZeneca (but not Pfizer) vaccine than in non-vaccinated individuals between
7 and 27 days after vaccination. Overall, the incidence rate of ITP was 1.13 (95%
CI; 0.62–1.63) per 100,000 AstraZeneca vaccine doses. With regard to the risk of thrombosis,
an excess of both venous with CVT (between +19 and +71) and arterial (between +11
and +232) thrombotic events was recorded in subjects aged 16 to 59 years who received
the AstraZeneca vaccine, while such excess risk was not evident in older subjects
as well as in all those receiving the Pfizer vaccine. No substantial variation in
the burden of hemorrhagic events could be found in all recipients of both vaccines.
A recent analysis of European data concluded that the risk of developing thrombocytopenia
in recipients of AstraZeneca vaccine was 151 per million doses (compared with 33 per
million doses in those receiving Pfizer vaccine), while that of developing cerebral/splanchnic
thrombotic episodes or deaths was respectively 30 and 5 per million doses (compared
with 4 and 0.4 per million doses in those receiving Pfizer vaccine).[13 ] Additional important evidence on VITT has been provided in another study, which
examined a series of 62 vascular cerebrovascular adverse events diagnosed in Germany
in close proximity with COVID-19 vaccination, 45 of which were CVT.[14 ] Some important elements emerged. First, 53 events were recorded in patients who
received AstraZeneca vaccine, nine in those who received Pfizer vaccine and none in
those who received Moderna vaccine. Considering that the overall prevalence of CVT
in the general population is typically comprised between 0.02 and 0.15 per 100,000
person-months, the incidence rate of CVT within the first month from administration
of the first vaccine dose was estimated as being 0.55 (95% CI, 0.38–0.78) per 100,000
person-months for all vaccines, increasing to 1.52 (95% CI, 1.00–2.21) per 100,000
person-months for AstraZeneca, but being as low as 0.11 (95% CI, 0.03–0.29) per 100,000
person-months for Pfizer and 0.00 (95% CI, 0.00–1.48) per 100,000 person-months for
Moderna vaccines. The adjusted incidence rate ratio of AstraZeneca vaccine was nearly
10-fold higher (9.68; 95% CI, 3.46–34.98) compared with mRNA-based vaccines, and approximately
threefold higher (3.14; 95% CI, 1.22–10.65) for women compared with men, while the
trend of higher risk in subjects aged <60 years did not reach statistical significance
(2.14; 95% CI, 0.83–6.78).[14 ] It is therefore important to mention that cases of secondary immune thrombocytopenia
(ITP) occurring in recipients of mRNA-based COVID-19 vaccines have been widely reported
in the current scientific literature,[15 ] though it remains to be determined whether this association is causal or merely
coincidental.
Both currently approved adenovirus-based vaccines seem to be associated with this
pathology, even if CVT episodes appear to be more frequently reported in recipients
of AstraZeneca (3.6 per million doses; 99% CI, 2.7–4.8 per million doses) than in
those of Johnson & Johnson vaccines (0.9 per million doses; 99% CI, 0.2–2.3 per million
doses).[16 ] An additional analysis of cases so far described has evidenced that there may be
some important differences between the vaccine-induced thrombosis caused by either
adenovirus-based vaccine. Compared with Johnson & Johnson, the thrombotic syndrome
observed in AstraZeneca recipients seems to develop earlier, with higher D-dimer values
and higher likelihood of being positive for functional platelet activation testing,
but with a lower risk of developing intracerebral bleeding. No other major clinical
and laboratory differences could be seen, beside a higher propensity to develop internal
jugular vein thrombosis in Johnson & Johnson vaccine recipients.[17 ] More recently, Krzywicka et al conducted a systematic analysis of all cases of post-COVID-19
vaccination CVT reported to the EudraVigilance database of the European Medicines
Agency (EMA).[18 ] Overall, 213 reports could be found, 187 (87.8%) in recipients of adenovirus-based
AstraZeneca vaccine, 25 (11.7%) in mRNA-based Pfizer vaccine recipients, and the remaining
one (0.5%) in a recipient of mRNA-based Moderna vaccine. Importantly, of all patients
in whom the outcome had been reported, the death rate was 37.6% (44/177) in recipients
of adenovirus-based AstraZeneca vaccine compared with 20% (2/10) in recipients of
either mRNA-based vaccine, thus considerably higher than that reported for cases of
pre-COVID-19 CVT (3%; 3/100).
Therefore, despite needing more time to garner a final epidemiological picture of
CVT-associated VITT, especially because the metrics remain widely heterogeneous across
different reports (epidemiologic data are referred to vaccine doses administered,
recipients, person-months, etc.), it seems reasonable to conclude that this syndrome
mostly involves younger individuals (i.e., aged 60 years or younger), potentially
with a female predominance, develops within 1 to 3 weeks after receiving the first
vaccine dose, and is indeed more prevalent after receiving an adenovirus-based COVID-19
vaccine. Nonetheless, as recently highlighted by an expert consensus on vaccine-induced
immune thrombotic thrombocytopenia,[19 ] the risk of VITT remains substantially lower than that of dying from COVID-19, so
that the favorable effects of vaccination largely offset the risk of developing severe
forms and dying from COVID-19, even at a younger age (i.e., the risk of VITT is 18–21
per million vaccine recipients compared with a risk of dying for COVID-19 up to 100
per million people aged 20–49 years in areas with high prevalence of SARS-CoV-2 infections).
Pathogenesis of VITT
The pathogenesis of CVT has been recently reviewed by Ulivi et al.[20 ] Briefly, this pathology, which is typically caused by occlusion of cerebral venous
sinuses (i.e., cerebral venous sinus thrombosis) or smaller cortical veins (i.e.,
cortical vein thrombosis), mostly develops in young females (in the third decade of
age, with 2:1 female:male ratio). The use of estrogen-containing oral contraceptives
is the most prevalent risk factor for CVT, associated with a prothrombotic condition
(either inherited or acquired) in the vast majority of patients. The most frequent
thrombophilic conditions are factor V Leiden, prothrombin gene polymorphism G20210A,
along with deficiencies of physiological inhibitors of blood coagulation (i.e., protein
C or S, antithrombin). With regard to the acquired conditions other than oral contraceptives,
an increased burden of antiphospholipid antibody syndrome, obesity, cancer, local
infections, and head trauma has been reported, while higher frequency is also seen
during pregnancy and puerperium. Although the prognosis is mostly favorable in the
vast majority of patients with CVT, around 15% of them may have long-term consequences
or even die of its consequences.
What has clearly emerged since the description of the first cases, is that the CVT
developing in patients with VITT seems to be different in terms of pathogenesis and
evolution from its more traditional form. The clearest evidence, as originally underpinned
by Greinacher et al[4 ] and Schultz et al,[5 ] is that no clear prothrombotic risk factors were present. Of the five VITT patients
described by Schultz et al, four (80%) were female, one was using oral contraceptives,
one contraceptive vaginal ring and one hormone-replacement therapy, though all were
found to have thrombocytopenia (between 10–70 × 109 /L), elevated D-dimer values (>13 mg/L), along with high values of IgG antibodies
against PF4-polyanion complexes. None had known pre-existing prothrombotic conditions
and three (60%) of these patients had fatal outcome. Similarly, in only one of the
11 VITT cases described by Greinacher et al was a pre-existing pro-thrombotic condition
able to be identified (FVL), while in all in whom a PF4-antibody ELISA test was performed,
the result was positive. All patients had thrombocytopenia (between 8 and 107 × 109 /L) and elevated D-dimer values (between 1.8 and 142 mg/L). Importantly, 6/11 (55%)
of these patients had fatal outcome, and almost all samples tested from a further
series of patients with suspected VITT (22/24; 92%) displayed platelet activation
after challenge with PF4. Similar evidence was provided in a 12, U.S.-based, case
series of adenovirus-based (Johnson & Johnson) COVID-19 vaccine recipients, as reported
by See and co-authors.[21 ] Overall, use of oral contraceptives was reported in only one of the patients (six
were obese), and only one could be safely discharged at home. As now could be predicted,
all patients tested had thrombocytopenia (between 9 and 127 × 109 /L), elevated D-dimer values (between 1.1 and 112 mg/L) and were positive for PF4-antibody
ELISA. In the analysis of the EudraVigilance database conducted by Krzywicka et al,[18 ] the prevalence of prothrombotic risk factors in VITT cases was also considerably
low (11% in adenovirus-based COVID-19 vaccines recipients and 15% in those who received
mRNA-based COVID-19 vaccines, compared with 64% in CVT cases reported pre-COVID-19).
Similarly, the use of oral contraceptives was nearly threefold lower (20 and 14% in
recipients of adenovirus-based and mRNA-based COVID-19 vaccines, respectively) compared
with the CVT cases reported pre-COVID-19 (54%).
Several other cases have since been described, as recently reviewed elsewhere.[6 ] All of these shared similar aspects of being positive for anti-PF4 antibodies as
measured by ELISA assays, but not with other immunological tests that are usually
positive in patients with heparin-induced thrombocytopenia with thrombosis (HITT).
Therefore, anti-PF4 antibodies are probably the main drivers of this life-threatening
pathology, so that the presence of other (either mild or severe) pro-thrombotic factors
does not additionally contribute to an otherwise already devastating thrombotic mechanism.
Beside the well-established role of anti-PF4 antibodies, the potential mechanisms
underlying their generation after COVID-19, which may also encompass neutrophil activation
and consequent release of the pro-thrombotic neutrophil extracellular traps (NETs),
is still a matter of speculation ([Fig. 1 ]).
Fig. 1 Pathogenetic mechanisms potentially involved in vaccine-induced thrombocytopenia
and thrombosis in recipients of adenovirus-based COVID-19 (coronavirus disease 2019)
vaccines. PF4, platelet factor 4.
Goldman and Hermans[22 ] have speculated that, after intramuscular injection of adenoviruses-based COVID-19
vaccine, endothelial cells may be infected and induced to synthesize the spike protein
of SARS-CoV-2, which could then bind with heparan sulfate proteoglycans at the luminal
side or released by injured cells. To this end, the spike proteins could bind to angiotensin
converting enzyme 2 (ACE2) at platelet surface, thus leading to their activation and
PF4 release, which may then assume immunogenic properties after linkage with heparan
sulfate proteoglycans shed from the same endothelial cells. Possible support to this
theory was provided by studies showing that administration of adenovirus-based vaccines
was effective to induce native-like post-translational processing and assembly of
SARS-CoV-2 spike protein contained with the adenovirus vector. The spike protein is
hence finally expressed at cell surface with the trimeric prefusion conformation capable
to bind to its natural host receptors.[23 ] The fact that COVID-19 vaccination may induce measurable amount of circulating spike
protein, thus playing a possible role in the pathogenesis of VITT, is supported by
data published by Ogata and colleagues.[24 ]
Another possible interpretation has been provided by Huynh and colleagues.[25 ] Briefly, it was first found that the binding of anti-PF4 antibodies in the sera
of VITT patients involved eight surface amino acids of PF4, all located within the
heparin binding site of the protein, but was different from the antigenic moiety recognized
by the anti-PF4 antibodies developing in patients with HITT. These VITT anti-PF4 antibodies
could instead effectively bind PF4, thus generating platelet-activating immune complexes,
in the absence of heparin; this in turn would favor a cross-linking of FcγRIIa receptors
on platelet surface, a phenomenon associated with activation of intracellular signaling
events ultimately triggering platelet activation and aggregation. Another potential
mechanism has been hypothesized by Kowarz and colleagues.[26 ] Specifically, the authors postulated that adenovirus-based COVID-19 vaccines may
trigger the transcription of wild-type and codon-optimized SARS-CoV-2 spike protein
open reading frames. This would enable alternative splice events, generating C-terminal
truncated and soluble SARS-CoV-2 Spike protein variants, which may bind to ACE2-expressing
cells in the circulation, possibly platelets and endothelial cells, thus leading to
their activation and thus finally favoring the generation of a pro-thrombotic milieu.
Irrespective of the precise mechanism, an important aspect in the pathogenesis of
VITT is that virtually all patients with this syndrome have shown the presence of
anti-PF4 antibodies. However, as analogous to heparin-induced thrombocytopenia (HIT),
not all patients who develop anti-PF4 antibodies after COVID-19 vaccination have thrombosis,
as clearly shown in different published works.[14 ]
[27 ] Interestingly, although the SARS-CoV-2 spike protein and PF4 share at least one
similar epitope, the anti-PF4 antibodies found in VITT patients do not seemingly cross-reacted
with SARS-CoV-2 spike protein.[27 ] It is hence conceivable that the association between anti-PF4 antibodies, thrombocytopenia
and/or thrombosis may occur in a minority of predisposed patients, who may have some
still unknown biological predisposition, or in those who develop a class of anti-PF4
antibodies that has larger platelet-activating potency. Last but not least, a direct
interplay between the adenoviral vector and the platelets of the COVID-19 vaccine
recipient cannot be ruled out, since it has been reported that some of these viruses
can bind to platelet receptors (e.g., coxsackie and adenovirus receptor [CAR], CD46),
thus triggering platelet activation, aggregation, and PF4 secretion.[28 ] Importantly, it was also recently shown that although the development of anti-PF4
antibodies could be observed in most adenovirus-based AstraZeneca COPVID-19 vaccine
recipients, these are non-platelet activating, are present in low titers and, last
but not least, their serum level seems to correlate significantly with those of anti-SARS-CoV-2
neutralizing antibodies (r = 0.50; p < 0.017).[29 ]
It is also important to mention here that although all patients with CVT who tested
positive for anti-PF4 antibodies in the study of Schulz et al had received the adenovirus-based
AstraZeneca vaccine,[14 ] none of those who received the Pfizer mRNA vaccine had a VITT risk score >2, meaning
that the thrombotic events recorded in those who received Pfizer vaccine may represent
another pathogenetic mechanism or may only be casually associated with vaccine administration,
thus being perhaps more appropriately categorized as having vaccine-associated thrombosis
(VAT) rather than VITT. To this end, it is noteworthy that although VITT seems to
occur more commonly after receiving adenovirus-based COVID-19 vaccines, a limited
number of individuals may also develop a form of acquired thrombotic thrombocytopenic
purpura (TTP) after receiving mRNA vaccines, as also highlighted by Maayan et al.[30 ] These cases are typically defined as being relatively young, and having considerably
decreased platelet count, low ADAMTS-13 (a disintegrin and metalloprotease with a
thrombospondin type 1 motif 13) activity and high anti-ADAMTS-13 antibody levels,
thus confirming that different pathogenetic mechanisms may be involved in the so-called
mRNA COVID-19 vaccines-associated thrombosis. The fact that the pathogenetic mechanisms
involved in the cases of thrombocytopenia developing after adenovirus-based COVID-19
vaccination and those seen in COVID-19 mRNA vaccine recipients may be different has
also been addressed by Lee et al,[15 ] who described 20 patients (median age 41 years; 55% women) who developed secondary
ITP after receiving either mRNA COVID-19 vaccine (i.e., 9 Pfizer and 11 Moderna).
Fourteen of such patients (70%) experienced significant hemorrhagic symptoms (bruising,
petechiae, or mucosal bleeding) before hospitalization, while none developed clinically
significant thrombosis.
Characterization of Anti-PF4 Antibodies in VITT
Characterization of Anti-PF4 Antibodies in VITT
Since the pathogenesis of VITT is essentially based on development of anti-PF4 antibodies,
their identification and quantification in plasma are essential for achieving a confirmation
of the diagnosis. In this respect, is has been now clearly demonstrated that only
ELISA-based anti-PF4 assays are able to identify the presence of such antibodies,
since all the other automated commercial (“rapid”) immunoassay display rather poor
sensitivity.[6 ] It is also important to highlight that not all ELISAs may reach 100% sensitivity
in detecting these antibodies, as demonstrated by Platton et al.[31 ] Therefore, when a single ELISA is negative in the presence of strong clinical suspicion,
another test should be used to complement the results of the former. Clear evidence
that ELISA methods are to be preferred for detecting anti-PF4 antibodies in patients
with VITT has also been provided by a comprehensive report of the UK National External
Quality Assessment Scheme, highlighting that all centers performing ELISAs reported
positive result on both lyophilized and liquid VITT samples, while all centers using
chemiluminescence, latex immunoassay, and lateral flow assays reported negative result
on VITT lyophilized samples.[32 ] Furthermore, an additional study published by Vayne and colleagues also found that
the sensitivity of the methods for detecting anti-PF4 IgG antibodies depends on the
antigen target, since significant levels of these antibodies could only be detected
in the assays using PF4–poly(vinyl sulfonate) complex as antigenic target.[33 ] As an additional supplement to ELISA assays for anti-PF4 antibodies, consideration
should also be given to performance of functional platelet activation assays. There
exists a plethora of potential assays; however, some variable positivity may arise
under the influence of technical test aspects. For example, the serotonin release
assay can be modified by adding PF4 to increase diagnostic sensitivity.[33 ] Furthermore, assay steps used in the assay to increase sensitivity to HITT, namely
use of a therapeutic dose of heparin, seems to attenuate and sometimes even abolish
the response in VITT.[21 ] Many of these important technical aspects have been recently reviewed,[6 ] where it was further stressed that the sensitivity of rapid chemiluminescent automated
anti-PF4 immunoassays is indeed too low compared with that of more conventional ELISA
techniques, so that only the latter should be used for this purpose. Yet, no ELISA
method is like another and so, even among ELISA techniques, the diagnostic sensitivity
may vary widely.[6 ]
[31 ] Another area of uncertainty is how long these antibodies will persist and whether
or not they may be associated with an increased risk of thrombosis after receiving
adenovirus-based vaccines for other infectious diseases.[34 ]
Conclusion
Unquestionable evidence has been provided about the binding of SARS-CoV-2 to platelets
with a pathway that also involves the spike protein-ACE2 interaction, and that this
binding is the followed by platelet activation, followed by necroptosis and apoptosis,[35 ] while no similar evidence has been provided for the adenovirus-based AstraZeneca
COVID-19 vaccine, in that challenge of citrate anticoagulated blood was not associated
with platelet activation or aggregation, nor with binding of autologous antibodies
to platelets.[36 ]
[37 ] It is hence not really surprising that some forms of SARS-CoV-2 recombinant spike
protein which were produced after human cells have been transfected by COVID-19 vaccination
may enter the circulation and then trigger a similar cascade of events (i.e., platelet
hyper-aggregation, which may then be followed by thrombosis, along with severe platelet
injury, which may then cause thrombocytopenia) in a small proportion of vaccine recipients.
Overall, VITT is a rare but potentially serious consequence of vaccination against
COVID-19 with certain adenovirus vaccines, namely AstraZeneca and Johnson and Johnson
(Janssen). The most serious consequential thromboses are represented by CVT and splanchnic
vein thrombosis. VITT represents “an immune response, leading to a condition similar
to one seen sometimes in patients treated with heparin,” that is HIT.[38 ] However, there are some clinical and laboratory differences between VITT and HITT.
Key to differentiating VITT and HITT is identification of prior heparin exposure,
and laboratory testing using modifications of existing assays. For example, use of
therapeutic heparin levels may augment both ELISA and functional testing (e.g., by
serotonin release assay) in HITT, but would more typically lower ELISA and functional
platelet responses in VITT. Also, rapid automated (e.g., AcuStar) plus ELISA anti-PF4
antibody tests would be positive in HITT, whereas in VITT only ELISA anti-PF4 antibody
tests are typically positive.
