The present article seeks to raise awareness of a potential correlation between the occurrence of DAVF and the current global epidemiological context, capitalized by COVID-19. We found a significant increase in the incidence of bDAVF during what we could define as the COVID-19 era (2020-2021) compared with a historic cohort collected from the previous 9 years. The features of these recently diagnosed bDAVFs were not significantly different from those of the precedent years. Neither the size and characteristics of the population nor the potential risk factors, which could eventually drive the development of bDAVFs, were uneven between the two considered periods. Establishing a relation of causality between COVID-19 or the new vaccines designed to prevent it and these vascular malformations would require further studies, longer periods of observation and the pooling of data from larger populations. However, we consider that our findings deserve to be reported and interpreted with caution. The following discussion will focus on the potential involvement of vaccines in the development of bDAVF since the rise in their incidence was remarkably superior after the kick off of massive vaccination, which in our region took place on January 4th, 2021.
According to the classic paper “The environment and disease: association or causation?” by A. Bradford Hill, there are nine main principles that must be considered to suggest causality between two facts: Strength, Temporality, Specificity, Analogy, Plausibility, Coherence, Biological Gradient, Experiment and Consistency 14. Some of these criteria might not be possible to elucidate due to the nature of the involved elements or the lack of tools or experiments to build the required evidence. However, the presence of a majority of them would allow us to elaborate a coherent argumentation to suggest a possible or probable correlation between two facts.
The strength of the association was considered by A.B. Hill as the most important factor to demonstrate the association between a disease and an epidemiological fact 14. In our case, this is given by the increase in the incidence, which was five times higher during 2021 than the average incidence of the previous nine years. This rise is even higher when compared to previous reports in which the incidence of bDAVF is estimated to be between 0.15-0.29 per 100,000 per year1,4,15,16.
Temporality is demonstrated by the fact that the suspected cause precedes the consequence in a reasonable timeline. Except for two cases that suffered COVID-19 infection prior to the vaccine and to the diagnosis of bDAVF, the remaining cases received their first dose 90 to 200 days before the diagnosis. If a prothrombotic phenomenon is suspected as a potential mechanism for vaccines or COVID to lead to the occurrence of bDAVFs, it is important to note that it has been reported that the incidence of bDAVF significantly increases during the first 6 months following CVT4,17. For COVID-19 infections, the time frame in our cohort was 60-80 days. Indeed, the time from COVID symptoms to CVT reported by other authors is estimated to be between 1-2 weeks18,19. The occurrence of CVTs would precede the diagnosis of bDAVFs in up to 6 months, which agrees with our findings.
Plausibility, coherence and analogy are discussed together in an attempt to provide a credible explanation supported by, or at least, without conflicting the available evidence. Accordingly, our rationale is built on the evidence of vaccines causing immune-mediated VITT, which elevates the risk of CVT, and the fact that CVTs often precede or coexist with bDAVFs. VITT is a rare syndrome that belongs to a wider spectrum of disorders characterized by the activation of platelets via anti-PF4 antibodies induced by an immunizing stimulus, often heparin or heparin-like molecules20. VITT was first related to adenoviral vector-based vaccines against COVID (ChAdOx1 CoV-19 vaccine, AstraZeneca and Ad26.COV2. vaccine, Janssen)8,20,21. Later, other reports also described this syndrome in patients receiving mRNA-based vaccines (BNT162b2, Pfizer-BioNTech)10. VITT is characterized by mild to severe thrombocytopenia, a low to normal range of fibrinogen, elevated D-dimer and normal or mildly increased coagulation times22. Some authors prefer the term thrombosis with thrombocytopenia syndrome in the absence of laboratory-proven immunological mechanisms. Similarly, pre-VITT syndrome has been reserved for patients who present with clinical and laboratory findings of VITT without imaging of thrombosis following vaccination23,24. A remarkable feature of thrombosis in VITT is that, in addition to occurring in typical sites of venous thrombosis (lungs and lower limbs), it shows a predilection for unusual locations, such as cerebral and ophthalmic veins22,25. The reaction induced by anti-PF4 antibodies in VITT is not limited to platelets; instead, it spreads, provoking a pancellular activation that involves monocytes, neutrophils and endothelial cells, which express their respective cytokines8,20,25. Overall, this global activation magnifies the thrombosis risk and might act as a trigger of neoangiogenic mechanisms leading to DAVFs, as suggested by the evidence derived from animal models26–28. The actual pathogenesis of bDAVF is unclear, but growing evidence suggests that neoangiogenesis induced by vascular endothelial growth factor (VEGF) release following venous hypertension might be a plausible mechanism26–28. The retrospective nature of this report and the wide range of time that may asymptomatically elapse from VITT to the diagnosis of bDAVF hinder the active investigation of coagulation or hematologic abnormalities. In those cases, in which blood and coagulation tests had been performed somewhen between the vaccine or COVID diagnosis and bDAVF diagnosis, the data were compatible with VITT 9(Table 5). However, compatibility is a vague concept, and confirmation of VITT requires further evidence, such as PF-4 antibody testing20.
The specificity of the association might be the most controversial item of the current argumentation. The specificity of the suggested association lies in the evidence of a proven ability of vaccines, and ultimately SARS-CoV-2, to provoke thrombosis through an immune-mediated pathophysiology. However, this fact alone might not be enough to state that vaccines or COVID elevate the risk of developing bDAVFs. Many other factors that coexist with COVID or vaccines could surreptitiously cause CVTs or bDAVFs by other unknown mechanisms in this specific period of time. For instance, a more sedentary lifestyle due to lockdown regimes or the increment of teleworking could lead to a higher risk of thrombotic phenomena.
Biological Gradient and Experiment do not apply for the present case since there is no evidence of a dose-dependent effect and the phases of experimentation for commercialized vaccines are over.
Consistency would require the repeated observation of this association by various researchers in other centers. Indeed, the purpose of this warning is precisely to raise awareness in this field to prompt other physicians to report their findings to contest or validate our observations. Undoubtedly, vaccines are the best available therapy to prevent the disease and tackle the pandemic. The incidence of bDAVFs in the vaccinated population is still very low. If confirmed, our results would only result in the communication of a very rare complication of vaccination that would not probably change at all current policies. In this sense, we have tried to temper our assertions and insist on the need for further studies.
The present report harbors several limitations. First, it is a retrospective series of patients from a single center in which the risk of selection bias is not negligible. Indeed, the epidemiological context, where a majority of the population has been vaccinated or exposed to the virus, precludes comparing the cohort of 2021 with a control group of patients who have not been exposed to any of the suspected risk factors. The retrospective design of the current report prevented us from actively looking for inflammatory biomarkers, immune-mediated prothrombotic factors or coagulation disorders. Second, it remains unclear whether CVT precedes bDAVF or is a consequence of venous hypertension caused by fistulous flow. Evidence suggests that CVT and cerebral venous sinus abnormalities are associated with dural and cerebral arteriovenous malformations29. Experiments in animal models have demonstrated that the occlusion of a cerebral venous sinus significantly increases the blood levels of VEGF and that VEGF is involved in the subsequent angiogenic response that might eventually develop into pathological arteriovenous communication27,28. However, the rate of CVTs was not higher in the vaccinated cohort than in the historical cohort. Nonetheless, as previously said, CVT could act as a trigger for neoangiogenic cascades leading to bDAVF and would not necessarily require coexisting with them. Indeed, it has been reported that the rate of diagnosed venous thrombosis in VITT is highly exceeded by the number of undiagnosed ones. Finally, other confounding or coexisting factors that might contribute to explaining the observed rise in the incidence of bDAVF could not be ruled out.