Microsurgical versus endovascular treatment of spinal epidural arteriovenous fistulas with intradural venous drainage: a multicenter study of 81 patients.

[OBJECTIVE] Spinal arteriovenous shunts are rare vascular lesions and are classified into 4 types (types I-IV). Due to rapid advances in neuroimaging, spinal epidural AVFs (edAVFs), which are similar to type I spinal dural AVFs (dAVFs), have recently been increasingly reported. These 2 entities have several important differences that influence the treatment strategy selected. The purposes of the present study were to compare angiographic and clinical differences between edAVFs and dAVFs and to provide treatment strategies for edAVFs based on a multicenter cohort.

[METHODS] A total of 280 consecutive patients with thoracic and lumbosacral spinal dural arteriovenous fistulas (dAVFs) and edAVFs with intradural venous drainage were collected from 19 centers. After angiographic and clinical comparisons, the treatment failure rate by procedure, risk factors for treatment failure, and neurological outcomes were statistically analyzed in edAVF cases.

[RESULTS] Final diagnoses after an angiographic review included 199 dAVFs and 81 edAVFs. At individual centers, 29 patients (36%) with edAVFs were misdiagnosed with dAVFs. Spinal edAVFs were commonly fed by multiple feeding arteries (54%) shunted into a single or multiple intradural vein(s) (91% and 9%) through a dilated epidural venous plexus. Preoperative modified Rankin Scale (mRS) and Aminoff-Logue gait and micturition grades were worse in patients with edAVFs than in those with dAVFs. Among the microsurgical (n = 42), endovascular (n = 36), and combined (n = 3) treatment groups of edAVFs, the treatment failure rate was significantly higher in the index endovascular treatment group (7.5%, 31%, and 0%, respectively). Endovascular treatment was found to be associated with significantly higher odds of initial treatment failure (OR 5.72, 95% CI 1.45-22.6). In edAVFs, the independent risk factor for treatment failure after microsurgery was the number of intradural draining veins (OR 17.9, 95% CI 1.56-207), while that for treatment failure after the endovascular treatment was the number of feeders (OR 4.11, 95% CI 1.23-13.8). Postoperatively, mRS score and Aminoff-Logue gait and micturition grades significantly improved in edAVFs with a median follow-up of 31 months.

[CONCLUSIONS] Spinal epidural AVFs with intradural venous drainage are a distinct entity and may be classified as type V spinal vascular malformations. Based on the largest multicenter cohort, this study showed that primary microsurgery was superior to endovascular treatment for initial treatment success in patients with spinal edAVFs.