2026 年 13 巻 p. 471-475
Cervical spinal epidural arteriovenous fistulas with intradural venous reflux are exceptionally rare, and endovascular treatment is particularly challenging when the arterial feeder arises directly from the vertebral artery. We report a 69-year-old man with progressive myelopathy caused by a cervical type A spinal epidural arteriovenous fistula with a feeder arising directly from the left vertebral artery at C6, treated by dual balloon-assisted transarterial n-butyl cyanoacrylate embolization. A balloon guiding catheter was used for proximal vertebral artery flow arrest, whereas an occlusion balloon catheter was positioned across the feeder origin for distal protection and flow-directed microcatheter stabilization. After confirming complete segmental flow arrest, 0.15 mL of 25% n-butyl cyanoacrylate mixed with Lipiodol was injected. Complete angiographic obliteration was achieved without complications, and neurological function gradually improved with marked resolution of spinal cord edema on follow-up magnetic resonance imaging at 3 months. Reported series of transarterial embolization for these fistulas show high obliteration rates but carry a risk of reflux into the parent artery when the feeder arises from a major vessel. In this anatomically rare lesion, the dual balloon configuration provided simultaneous proximal flow arrest, distal protection across the feeder origin, and stabilization of the feeder microcatheter, allowing controlled n-butyl cyanoacrylate injection without proximal reflux. This technique may be a useful endovascular option for selected cases of cervical spinal epidural arteriovenous fistula with direct vertebral artery feeders, although its broader efficacy requires confirmation in larger series.
Spinal epidural arteriovenous fistulas (SEAVFs) are uncommon vascular shunt lesions characterized by abnormal arteriovenous connections within the epidural venous plexus. On the basis of venous drainage pattern, SEAVFs are classified into type A lesions with intradural venous reflux, causing venous congestive myelopathy, and type B lesions with predominantly extradural drainage producing compressive symptoms.1-3) Cervical SEAVFs with intradural venous reflux (type A) are exceptionally uncommon in an already rare entity.4) When the arterial feeder arises directly from the vertebral artery (VA) and shares a short common trunk, transarterial liquid embolization carries a substantial risk of embolic reflux into the parent artery, which may cause posterior circulation ischemia. This anatomical configuration therefore demands precise procedural control.
Balloon-assisted embolization techniques have been used to improve flow control and procedural safety in complex spinal arteriovenous lesions.5,6) A dual balloon protection strategy combining a balloon guiding catheter with a distal balloon microcatheter has also been reported for n-butyl cyanoacrylate (NBCA) embolization through the meningohypophyseal trunk, indicating the value of dual flow arrest with maintained proximal protection during liquid embolic injection.7) However, to our knowledge, a systematic description of a dual balloon strategy specifically addressing the angioarchitectural challenges of direct VA feeders in cervical SEAVFs has not been previously reported. We report a rare case of cervical SEAVF with a feeder arising directly from the dominant VA, successfully treated by dual balloon-assisted transarterial NBCA embolization, and discuss the technical considerations relevant to this anatomically challenging setting.
A 69-year-old man presented with a 4-month history of progressive gait disturbance, bilateral lower extremity weakness, and urinary dysfunction. He was able to ambulate only with a cane. Neurological examination revealed symmetric moderate weakness in the lower extremities (Medical Research Council [MRC] grade 3/5), normal upper extremity strength (MRC grade 5/5), decreased pinprick and light touch sensation below the T10 level with preserved vibration and proprioception, bilateral brisk deep tendon reflexes, positive Babinski signs, and increased post-void residual volume. There was no history of trauma, infection, or prior spinal procedures.
Magnetic resonance imaging (MRI) revealed longitudinally extensive T2-weighted hyperintensity from C7 to T6 with serpiginous perimedullary flow voids (Figure 1), consistent with venous congestive myelopathy.1-3) Spinal angiography confirmed a type A SEAVF supplied by a segmental branch arising directly from the left VA at C6. Multiple feeders converged into an epidural venous pouch at C6-C7, which drained retrogradely through a radiculomedullary vein into the perimedullary venous system (Figure 2). Epidural venous outflow was limited. The left VA was dominant, without significant collateral contribution from the contralateral VA or the posterior inferior cerebellar artery, underscoring the importance of avoiding prolonged proximal VA occlusion.
Pre- and post-treatment spinal magnetic resonance imaging (MRI) findings. (A) Preoperative sagittal T2-weighted MRI indicating longitudinally extensive spinal cord hyperintensity (asterisk) from C7 to T6, disproportionate to degenerative changes, consistent with venous congestive myelopathy. (B) Preoperative axial T2-weighted MRI showing a dilated perimedullary vein (arrow) and an epidural venous pouch (arrowhead). (C) Follow-up sagittal T2-weighted MRI at 3 months showing marked resolution of spinal cord edema. (D) Follow-up axial T2-weighted MRI at 3 months showing disappearance of abnormal vessels.
Dual balloon-assisted transarterial embolization. The balloon guiding catheter provided proximal vertebral artery (VA) flow arrest; the occlusion balloon catheter, positioned across the feeder origin, provided distal protection and flow-directed microcatheter stabilization. (A) Early arterial phase angiogram showing a segmental feeder arising directly from the VA (arrow). (B) Late arterial phase angiogram showing an epidural venous pouch (asterisk) with retrograde intradural venous drainage. (C) Intraoperative fluoroscopic image showing n-butyl cyanoacrylate cast (arrow) and the inflated occlusion balloon catheter (arrowhead) positioned across the feeder origin. (D) Post-embolization angiogram showing complete obliteration of the fistula and absence of early venous drainage.
Endovascular treatment was performed under general anesthesia. A balloon guiding catheter (OPTIMO, Tokai Medical Products, Inc., Kasugai, Aichi, Japan) was positioned in the left VA for proximal flow control. An occlusion balloon catheter (Scepter C, 4 mm× 10 mm, MicroVention, Inc. [Terumo], Tustin, CA, USA) was advanced through the balloon guiding catheter and positioned in the VA across the feeder origin, with the balloon segment spanning it. Correct positioning was confirmed by roadmap angiography. A flow-directed microcatheter (DeFrictor Nano, Medico's Hirata Inc., Osaka, Japan) was then navigated coaxially through the balloon guiding catheter and advanced selectively into the arterial feeder under roadmap guidance.
With all catheters in position, the occlusion balloon catheter was inflated first to provide distal protection and stabilize the flow-directed microcatheter in the feeder, followed by inflation of the balloon guiding catheter to achieve proximal VA flow arrest. Complete segmental flow arrest between the 2 balloons was confirmed by test injection through the balloon guiding catheter. No angiographic evidence of collateral feeder recruitment was observed under dual balloon occlusion. Under these conditions, 0.15 mL of 25% NBCA mixed with Lipiodol was injected through the flow-directed microcatheter under continuous fluoroscopic roadmap guidance. The embolic material penetrated the epidural venous pouch and proximal draining vein without reflux into the parent VA. After injection, the flow-directed microcatheter was withdrawn slowly under maintained dual balloon inflation to prevent distal embolism from catheter tip adhesion. After the flow-directed microcatheter had been safely retracted into the guiding catheter, the occlusion balloon catheter was deflated first; approximately 10 mL of blood was then aspirated through the balloon guiding catheter twice (total 20 mL), a volume substantially exceeding the balloon guiding catheter dead space of approximately 2.0-2.5 mL and therefore considered sufficient to clear any potentially mobilized NBCA debris, and the balloon guiding catheter was deflated last to restore antegrade VA perfusion. Final angiography showed complete obliteration of the fistula (Figure 2).
Neurological function gradually improved postoperatively. Follow-up MRI at 3 months revealed marked resolution of spinal cord edema and disappearance of abnormal perimedullary vessels (Figure 1). No recurrence has been observed on follow-up MRI. Written informed consent was obtained from the patient for publication of this article and accompanying images.
Cervical SEAVFs with intradural venous reflux and a feeder arising directly from the VA are exceptionally uncommon, and reports describing successful endovascular treatment for this specific anatomy remain limited. The clinical presentation may also be initially misattributed to degenerative spine disease because progressive myelopathy and gait disturbance overlap with cervical spondylotic myelopathy.8) In the present case, longitudinally extensive T2 hyperintensity from C7 to T6 with perimedullary flow voids was the key MRI pattern that prompted spinal angiography, on which a type A SEAVF with a feeder arising directly from the dominant V2 segment of the left VA at C6 was identified.9)
For SEAVFs with intradural venous reflux, the treatment goal is complete disconnection of the arteriovenous shunt and the refluxing intradural venous pathway, achievable by both microsurgical and endovascular approaches.1-3,10) A multicenter study in 81 patients indicated that primary microsurgery achieved higher initial obliteration rates than endovascular treatment for type A SEAVFs, and the authors emphasized the importance of careful case selection and angioarchitectural evaluation when an endovascular-first strategy is considered.11) However, this multicenter analysis included almost no cases of cervical SEAVFs with feeders arising directly from the V2 segment of the VA. To our knowledge, microsurgical disconnection of such a fistula is rarely reported because direct surgical manipulation of feeders originating from the dominant VA carries a substantial risk of massive bleeding, VA injury, and concomitant spinal cord ischemia.1) In the present case, the angioarchitecture was also unfavorable for safe surgical exposure: the feeder arose directly from the dominant V2 segment of the left VA at C6, where the artery runs within the transverse foramen and is surrounded by the venous plexus. We therefore judged that endovascular treatment offered the only realistic curative option, and dual balloon-assisted transarterial NBCA embolization was selected accordingly.
Within endovascular treatment, the choice of balloon configuration also required careful consideration. Prior balloon-assisted reports for SEAVFs used a single balloon for proximal flow arrest (Uyama et al.5)) or for mechanical stabilization during coil embolization (Park et al.6)); however, in the present case, the feeder arose directly from the dominant VA, requiring simultaneous proximal and distal balloon occlusion. As Tega et al.7) described, the dual balloon protection technique is particularly valuable when the feeder microcatheter is unstable or when NBCA must be injected close to the parent artery, regardless of the specific lesion treated. The dual balloon configuration adopted here created a temporary, reversible zone of segmental flow arrest, which simultaneously prevented proximal NBCA reflux into the parent VA, stabilized the feeder microcatheter, and suppressed competitive antegrade flow. This configuration enabled controlled NBCA injection, leading to complete angiographic obliteration of the fistula without complications and substantial neurological recovery with marked resolution of spinal cord edema on follow-up MRI.
The reported outcomes of endovascular treatment help to clarify what dual balloon protection adds. In contemporary series of SEAVF with venous congestive myelopathy, transarterial embolization has achieved complete obliteration in approximately 73-94% of cases, with neurological improvement in up to 91% and a generally low periprocedural complication rate.12) Two factors, however, recurrently limit the efficacy and safety of transarterial embolization. First, durable cure requires the liquid embolic to penetrate beyond the fistulous point into the proximal draining vein; embolization limited to the arterial feeder tends to recur through recruitment of new feeders.12) Second, when the feeder arises directly from a major parent artery, antegrade flow and the short interval to the parent vessel predispose to reflux of liquid embolic into that artery, which in the present anatomy could have caused posterior circulation infarction.
A dual balloon configuration addresses these 2 factors mechanistically. Proximal flow arrest with the balloon guiding catheter, combined with a distal occlusion balloon catheter spanning the feeder origin, eliminates competitive antegrade flow across the shunt and thereby facilitates penetration of NBCA into the venous side. When the microcatheter cannot be advanced distally to a position close to the shunt, the proximal balloon limits reflux of NBCA into the parent artery, and the segmental flow arrest established between the 2 balloons helps the microcatheter achieve a wedged position that allows more controlled injection. Maintaining the proximal balloon until microcatheter retrieval and aspiration are complete further protects the parent artery from cast migration during withdrawal. In the present case, this configuration achieved complete obliteration with venous penetration and without any embolic complication.
These advantages are mechanistic rationales rather than proved benefits. As a single-case experience, our report cannot establish the general efficacy or safety of the dual balloon strategy, and its reproducibility warrants evaluation in larger series. Nonetheless, in selected high-risk anatomies in which the feeder arises directly from a major parent artery and a single-balloon technique provides incomplete protection, dual balloon protection may be a reasonable option.
The sequential order of balloon inflation and deflation merits careful consideration. The occlusion balloon catheter was inflated first to anchor the feeder microcatheter before hemodynamic changes occurred, followed by inflation of the balloon guiding catheter; we acknowledge that inflating the balloon guiding catheter first, as reported by Tega et al.,7) may also be acceptable, given the order of inflation is unlikely to differ substantially in safety provided that complete segmental flow arrest is achieved before NBCA injection. The deflation sequence, in contrast, has direct safety implications. After NBCA injection and slow withdrawal of the flow-directed microcatheter under maintained dual balloon protection, the occlusion balloon catheter was deflated first; the balloon guiding catheter was then aspirated twice with approximately 10 mL each (total 20 mL), a volume substantially exceeding the balloon guiding catheter dead space of approximately 2.0-2.5 mL and therefore considered sufficient to clear any potentially mobilized NBCA debris; the balloon guiding catheter was deflated last to restore antegrade VA perfusion. This deflation sequence aligns with that described by Tega et al.,7) in which the proximal balloon guiding catheter functions as an "airbag" maintained until microcatheter retrieval and proximal aspiration are complete. Maintaining proximal protection until aspiration was completed is particularly important because Lipiodol-mixed NBCA may damage the distal microballoon, and any cast fragment dislodged during microcatheter withdrawal could migrate into the parent VA if proximal protection were released prematurely. The selection of NBCA concentration and injection volume also warrants attention. A relatively high concentration (25%) was chosen to achieve rapid polymerization and limit distal migration under flow-arrested conditions, whereas a small injection volume (0.15 mL) targeted the epidural venous pouch and proximal draining vein without propagation into the perimedullary venous system. Continuous roadmap monitoring and immediate cessation on any sign of unwanted propagation are mandatory, and pre-injection roadmap angiography under dual balloon occlusion is recommended to detect any collateral feeder recruitment before committing to injection, particularly in cases with multiple potential radicular feeders.
Several limitations of this report warrant acknowledgment. Because this is a single-case experience, the dual balloon strategy may possibly be useful for selected cervical SEAVFs with direct VA feeders, but its general efficacy cannot be established from this report; our intent is to share technical considerations that may be helpful in similar high-risk anatomies. The reproducibility and broader applicability of the dual balloon strategy require validation in additional cases. The technique is inherently demanding, requiring precise catheter positioning across the feeder origin and careful coordination of balloon inflation timing; it should be performed only at centers with substantial experience in spinal vascular interventions. Furthermore, long-term follow-up beyond 3 months has not yet been obtained, and continued surveillance is necessary to confirm durable obliteration.
The authors thank the medical staff involved in the care of the patient.
All authors have no conflict of interest.
