Dural Arteriovenous Fistula Formation Secondary to Cerebral Venous Thrombosis: Longitudinal Magnetic Resonance Imaging Assessment Using 4D-Combo-MR-Venography

Abstract

Abstract Background and Purpose Dural arteriovenous fistulae (DAVFs) can develop secondary to cerebral venous thrombosis (CVT). The incidence of DAVF has not yet been investigated prospectively. Methods Between July 2012 and January 2018, combined static and dynamic 4D MR venography (4D-combo-MRV) was performed in 24 consecutive patients at diagnosis of CVT and after 6 months. 3 Tesla magnetic resonance imaging with time of flight and contrast-enhanced magnetization-prepared rapid acquisition with gradient echo were performed at baseline to evaluate the extent of thrombosis and affected vessel segments. Baseline and follow-up 4D-combo-MRV were assessed for signs of DAVF. Interrater reliability of DAVF detection and the extent of recanalization were analyzed with kappa statistics. Results DAVFs were detected in 4/30 CVT patients (13.3%, 95% confidence interval [CI] 3.3–26.7). Two of 24 patients (8.3%, 95% CI: 0–20.8) had coincidental DAVF with CVT on admission. At follow-up, de novo formation of DAVF following CVT was seen in 2/24 patients (8.3%, 95% CI: 0–20.8). Both de novo DAVFs were low grade and benign fistulae (Cognard type 1, 2a), which had developed at previously thrombosed segments. Endovascular treatment was required in two high degree lesions (Cognard 2a + b) detected at baseline and in one de novo DAVF (Cognard 1) because of debilitating headache and tinnitus. Thrombus load, vessel recanalization, and frequency of cerebral lesions (hemorrhage, ischemia) were not associated with DAVF occurrence. Conclusion This exploratory study showed that de novo DAVF formation occurs more frequently than previously described. Although de novo DAVFs were benign, 75% of all detected DAVFs required endovascular treatment. Therefore, screening for DAVF by dynamic MRV, such as 4D-combo-MRV, seems worthwhile in CVT patients.