Proposal of Venous Drainage-Based Classification System for Carotid Cavernous Fistulae With Validity Assessment in a Multicenter Cohort.

Thomas, Ajith J; Chua, Michelle; Fusco, Matthew; Ogilvy, Christopher S; Tubbs, R Shane; Harrigan, Mark R; Griessenauer, Christoph J

BACKGROUND Carotid cavernous fistulae (CCFs) are most commonly classified based on arterial supply. Symptomatology and treatment approach, however, are largely influenced by venous drainage. OBJECTIVE To propose an updated classification system using venous drainage. METHODS CCFs with posterior/inferior drainage only, posterior/inferior and anterior drainage, anterior drainage only, and retrograde drainage into cortical veins with/without other drainage channels were designated as types 1, 2, 3, and 4, respectively. CCFs involving a direct connection between the internal carotid artery and cavernous sinus were designated as type 5. This system was retrospectively applied to 29 CCF patients. RESULTS Our proposed classification was significantly associated with symptomatology (P < .001). Type 2 was significantly associated with coexisting ocular/orbital and cavernous symptoms only (P < .001), type 3 with ocular/orbital symptoms only (P < .01), and type 4 demonstrated cortical symptoms with/without ocular/orbital and cavernous symptoms (P < .01), respectively. There was a significant association of our classification system with the endovascular treatment approach (P < .001). Types 1 and 2 were significantly associated with endovascular treatment through the inferior petrosal sinus (P < .01). Type 3 was significantly associated with endovascular treatment through the ophthalmic vein (P < .01) and type 5 with transarterial approach (P < .01), respectively. Types 2 (27.6%) and 3 (34.5%) were most prevalent in this series, whereas type 1 was rare (6.9%), suggesting that some degree of thrombosis is present, with implications for spontaneous resolution. Type 2 CCFs demonstrated a trend toward partial resolution after endovascular treatment (P = .07). CONCLUSION Our proposed classification system is easily applicable in clinical practice and demonstrates correlation with symptomatology, treatment approach, and outcome.