Long-term Outcome After Endovascular Treatment of Cavernous Sinus Dural Arteriovenous Fistula and a Literature Review

Background: The long-term efficacy of endovascular treatment (EVT) for cavernous sinus dural arteriovenous fistulae (CS-dAVF) was assessed with a special focus on residual shunts after initial EVT.

Patients And Methods: This retrospective survey included 50 patients who had undergone EVT and were followed for 1 month or longer (median follow-up 56 months).

Results: Common preoperative symptoms were chemosis (78%), extra-ocular motor palsy (72%), exophthalmos (66%), and tinnitus (26%). CS-dAVF were addressed by transvenous embolization (tVE, n = 48), tVE only was used in 43 instances and tVE plus transarterial embolization (tAE) in five. Two patients underwent tAE only. Procedure-related morbidity (brainstem infarction) was recorded in one patient (2%) and transient symptom exacerbation (paradoxical worsening) in 12 patients (24%). Postoperative digital subtraction angiography showed no major retrograde shunt or cortical venous reflux in any of the 50 patients. Anterograde or minor retrograde residual shunt was observed in 17 patients (34%); three of these underwent additional tVE and four had Gamma Knife surgery. The shunt flow disappeared in all 17 patients 12.6 ± 13.4 (mean ± SD) months after initial EVT. At the latest follow-up, 65.7 ± 52.6 months after the initial operation, no shunt flow was observed in any of the 50 patients. None had remaining or newly developed chemosis or tinnitus on follow-up. The rate of persistent cavernous sinus symptoms at the latest follow-up was higher in patients with than without post-procedural paradoxical worsening (5/12, 41.7% vs. 2/38, 5.3%, p = 0.0059 by Fisher's exact test).

Conclusions: Long-term follow-up showed that EVT, especially tVE, is an efficient and safe treatment for CS-dAVF. It resulted in the eventual disappearance of shunt flow. Residual shunt without major retrograde flow or cortical venous reflux can be monitored without additional treatment.