Prevalence and Associations of Dural Arteriovenous Fistulae in Cerebral Venous Thrombosis: Analysis of ACTION-CVT

Article information

J Stroke. 2024;26(2):325-329

Publication date (electronic) : 2024 May 30

doi : https://doi.org/10.5853/jos.2023.02264

Aaron Shoskes^,1, Liqi Shu², Thanh N. Nguyen³, Mohamad Abdalkader⁴, James Giles⁵, Jordan Amar⁵, James E. Siegler⁶, Nils Henninger^7,8, Marwa ElNazeir⁷, Sami Al Kasab^9,10, Piers Klein³, Mirjam R. Heldner¹¹, Kateryna Antonenko¹¹, Marios Psychogios¹², David S. Liebeskind¹³, Thalia Field¹⁴, Ava Liberman¹⁵, Charles Esenwa¹⁶, Alexis Simpkins¹⁷, Grace Li¹⁷, Jennifer Frontera¹⁸, Lindsey Kuohn¹⁸, Aaron Rothstein¹⁹, Ossama Khazaal¹⁹, Yasmin Aziz²⁰, Eva Mistry²⁰, Pooja Khatri²⁰, Setareh Salehi Omran²¹, Adeel S. Zubair²², Richa Sharma²², Robert M. Starke²³, Jacques J. Morcos²³, Jose G. Romano¹, Shadi Yaghi², Negar Asdaghi¹

¹Department of Neurology, University of Miami, Miami, FL, USA

²Department of Neurology, Brown University, Providence, RI, USA

³Department of Neurology, Boston Medical Center, Boston University School of Medicine, Boston, MA, USA

⁴Department of Radiology, Boston Medical Center, Boston University School of Medicine, Boston, MA, USA

⁵Department of Neurology, Washington University, Saint Louis, MO, USA

⁶Cooper Neurologic Institute, Cooper University, Camden, NJ, USA

⁷Department of Neurology, University of Massachusetts, Worcester, MA, USA

⁸Department of Psychiatry, University of Massachusetts, Worcester, MA, USA

⁹Department of Neurology, Medical University of South Carolina, Charleston, SC, USA

¹⁰Department of Neurosurgery, Medical University of South Carolina, Charleston, SC, USA

¹¹Department of Neurology, Inselspital, University Hospital Bern, University of Bern, Bern, Switzerland

¹²Department of Interventional and Diagnostic Neuroradiology, Clinic of Radiology and Nuclear Medicine, University Hospital Basel and University of Basel, Basel, Switzerland

¹³Department of Neurology, University of California at Los Angeles, Los Angeles, CA, USA

¹⁴Division of Neurology, University of British Columbia, Vancouver, BC, Canada

¹⁵Department of Neurology, Weill Cornell Medical Center, New York, NY, USA

¹⁶Department of Neurology, Montefiore Medical Center, New York, NY, USA

¹⁷Department of Neurology, University of Florida, Gainesville, FL, USA

¹⁸Department of Neurology, New York University, New York, NY, USA

¹⁹Department of Neurology, University of Pennsylvania, Philadelphia, PA, USA

²⁰Department of Neurology and Rehabilitation Medicine, University of Cincinnati, OH, USA

²¹Department of Neurology, University of Colorado School of Medicine, Aurora, CO, USA

²²Department of Neurology, Yale School of Medicine, New Haven, CT, USA

²³Department of Neurological Surgery, University of Miami, FL, USA

Correspondence: Aaron Shoskes Department of Neurology, University of Miami Miller School of Medicine, 1120 NW 14th Street, Floor 13, Miami, FL 33136, USA Tel: +1-801-585-7575 E-mail: aaron.shoskes@hsc.utah.edu

Received 2023 July 13; Revised 2024 January 1; Accepted 2024 February 15.

Dear Sir:

Dural arteriovenous fistulae (DAVFs) represent approximately 10% of all intracranial vascular malformations [1] with an incidence of 0.16–1.04 per 100,000 person-years [2,3]. Prior studies identified an association between cerebral venous thrombosis (CVT) and DAVF although the direction of causality remains uncertain. In one series, 39% of DAVF patients had a CVT identified within an adjacent or downstream sinus [4], and patients with DAVF have a higher prevalence of thrombophilia, implicating thrombosis in its pathogenesis [5]. We aimed to determine the prevalence and associated factors of DAVF in a large multicenter study of treated CVT.

The Anticoagulation in the Treatment of Cerebral Venous Thrombosis (ACTION-CVT) study was a multicenter retrospective study comparing outcomes among CVT patients treated with a vitamin K antagonist versus direct oral anticoagulants [6]. The study included consecutive patients with acute CVT from January 1, 2015 through December 31, 2020. Patients were identified using the International Classification of Diseases (ICD)-9/10 codes, and diagnoses were confirmed by retrospective review of records and imaging studies. Institutional review board (IRB) approval was obtained at each center and informed consent was waived by IRB.

Abstracted patient variables are included in Table 1. Radiographic variables at time of presentation included venous infarction, cerebral edema, intracranial hemorrhage, and CVT location divided into cortical veins, superficial sinuses (superior sagittal, transverse, or sigmoid sinus), and deep sinuses (internal cerebral veins, straight sinus, or vein of Galen). Information about recanalization was abstracted from radiology reports of imaging obtained subsequent to the initial hospitalization for CVT and was subdivided into complete (full recanalization without residual thrombus), partial (improved opacification/flow but residual thrombus), and no recanalization (no change or worsening in opacification/flow). Characteristics of DAVF (laterality, location, timing, Borden classification, and treatment of DAVF) were retrospectively reviewed by individual sites.

Table 1.

Characteristics of patients with and without dural arteriovenous fistulae

Clinical, imaging, and anticoagulation characteristics of patients with versus without DAVF were compared using chi-square, Fisher’s exact, and rank sum tests as appropriate. Stepwise multivariable logistic regression analysis including clinical variables with P<0.2 in univariate analysis were used to determine factors associated with DAVF, and variables with P<0.05 from the stepwise regression were included in the final logistic regression to identify factors associated with DAVF. Missing data were not imputed for the initial multivariable analysis, and patients with missing data were excluded. In additional analyses, we conducted mean imputation for missing data and built multivariable binary logistic regression models to determine predictors of DAVF using the imputed values.

A total of 1,024 patients met inclusion criteria of whom 62.8% were female with median age of 44 years, and 751 (73.3%) had follow-up imaging available after hospitalization. Over a median follow-up of 284 days (interquartile range [IQR] 111–685 days), 19 patients (1.9%, 95% confidence interval [CI] 1.0%–2.7%) had DAVF. Patients with DAVF had a median age of 55 years, 36.8% were female, and 72.2% were White (Table 1). Patients with DAVF were older (median age 55 [IQR 45–64] vs. 44 [IQR 32–58], P=0.03), more likely to be male (63.2% vs. 36.7%, P=0.03), and less likely to have headache as a presenting CVT symptom (47.4% vs. 75.1%, P=0.01). Other clinical variables did not vary between the two groups (Table 1). Absent venous recanalization on follow-up imaging was more common among patients with DAVF compared to those without (50.0% vs. 14.9%, P<0.01) (Table 2), while there was no difference in CVT location or the presence of venous infarction, hemorrhage, or cerebral edema. In the stepwise binary logistic regression analysis including clinical variables with P<0.2 in univariate analysis (age, sex, headache, seizure, cortical vein involvement, and lack of recanalization), only cortical vein involvement (P=0.01) and absent venous recanalization (P=0.01) met the threshold for significance and in the final regression analysis both cortical vein involvement (OR 8.00, 95% CI 1.63–39.27, P=0.01) and absent venous recanalization on follow-up imaging (OR 6.28, 95% CI 1.60–24.68, P=0.01) were independently associated with detection of DAVF. In additional analyses imputing missing data, headache inversely correlated with DAVF (OR 0.31, 95% CI 0.11–0.85, P=0.020) and no recanalization on follow-up imaging remained a significant predictor of DAVF (OR=5.37, 95% CI 1.56–18.3, P=0.008).

Table 2.

Imaging characteristics of patients with and without dural arteriovenous fistulae

DAVF was detected at a median of 44 days from diagnosis of CVT (IQR 0–242) (Table 3). Overall, 8 (42.1%) patients were diagnosed with DAVF within 7 days of CVT diagnosis and 11 (57.9%) were noted to have the DAVF on follow-up imaging (median 194 days). Eight fistulae were Borden type I (42.1%), five were Borden type II (26.3%), and six were Borden type III (31.5%). The majority of patients (n=15, 78.9%) underwent surgical or endovascular treatment of the DAVF.

Table 3.

Characteristics of dural arteriovenous fistulae identified in cerebral venous thrombosis patients

In this international multicenter cohort of 1,024 patients diagnosed with CVT, 19 (1.9%) DAVF were identified including 11 that were not recognized during initial hospitalization for CVT. Cortical vein thrombosis and absent recanalization on follow-up imaging were independently associated with DAVF.

Our findings are in keeping with prior studies on this subject demonstrating a higher prevalence of DAVF among patients with CVT as compared to the reported incidence rate of DAVF in population-based studies [2,3]. In an International Cerebral Venous Thrombosis Consortium study, DAVF was detected in 2.4% of CVT patients and was associated with male gender, chronic CVT, and older age [7]. Several pathophysiological events have been proposed to underly the de novo formation of DAVF after CVT including neovascularization secondary to venous hypertension caused by CVT [8] or enlargement of pre-existing shunts between meningeal arteries and dural venous sinuses. The observation that venous hypertension appears to play a role in DAVF formation offers an explanation as to why absent CVT recanalization is a risk factor of DAVT. Cortical vein thrombosis has not been associated with DAVF in prior studies, but has previously been described in 17%–61% of patients with CVT [9,10], and involvement of cortical veins is associated with higher prevalence of venous infarction/hemorrhage and worse outcomes [10].

Despite the strengths of a large international multicenter cohort, our study has several limitations. Limited information regarding characteristics of DAVFs was available, restricting our ability to comment on clinical symptoms suggestive of DAVF or the decision to treat or observe the DAVF. Furthermore, ACTIONCVT did not capture all lab values of systemic hypercoagulability (such as d-dimer) and given the known association between hypercoagulability and DAVF, this reflects an area of potential future study. Follow-up imaging was not systematically obtained for all patients (performed in 751/1,024 [73.9%] within the cohort), different sites had variable imaging practices, and the indication for choosing the imaging modality was not reported. Only 4% of patients underwent digital subtraction angiography (DSA) in our study (which may result in under-recognition as it is the gold standard for detection of DAVF) and it is likely that those that underwent DSA did so due to concern for DAVF on initial imaging. Finally, the direction of causality between DAVF and CVT remains uncertain given the possibility of a DAVF being missed on initial imaging, and it is possible that in some cases that the presence of a DAVF was the reason for lack of venous recanalization and not the result of lack of recanalization of a CVT.

In this large multicenter study of patients with CVT, cortical vein thrombosis and lack of venous recanalization at follow-up were independently associated with DAVF. The present study reinforces the association between CVT and DAVF as both concomitant diagnoses as well as the potential for DAVF formation after CVT identification.