Association between the CT myelography-based Duke cerebrospinal fluid venous fistula confidence score and outcomes after fistula embolization.

Abstract

The CT myelography (CTM)-based Duke CSF-Venous Fistula Confidence Score (DCCS) standardizes diagnostic assessment of suspected CSF-venous fistulas (CVFs) in spontaneous intracranial hypotension (SIH). We evaluated whether confidence categories are associated with response after transvenous embolization (TVE), and whether renal contrast accumulation adds prognostic information in equivocal cases. This retrospective, single-center cohort included patients with SIH attributed to suspected or confirmed CVFs who underwent TVE between May 2023 and January 2026. Based on the DCCS, pretreatment CTM studies were categorized as definite, high-probability, or low-probability groups. The primary outcome was overall clinical improvement, defined as complete or partial symptom improvement without additional intervention at 3 months. Renal contrast accumulation was evaluated as an adjunctive marker. Overall clinical improvement occurred in 44/63 patients (69.8%), including complete resolution in 32 (50.8%) and partial improvement in 12 (19.0%). Improvement rates differed by Duke category: 24/30 (80.0%) in the definite, 10/12 (83.3%) in the high-probability, and 10/21 (47.6%) in the low-probability group (p=0.02). High-confidence CTM findings (comprising definite and high-probability) remained independently associated with overall improvement after adjustment for preoperative Bern score (adjusted OR 4.47; 95% CI 1.15 to 20.19; p=0.03). A renal accumulation modified model showed similar discrimination to conventional CTM grading (area under the curve (AUC) 0.691 vs 0.683; p=0.87), but lower corrected Akaike Information Criterion (AICc)/Bayesian Information Criterion (BIC) and higher sensitivity (90.9% vs 77.2%). The CTM-based DCCS categories were associated with response after TVE. Low-probability cases had less predictable outcomes but improved in nearly half of patients. Renal contrast accumulation may provide adjunctive information in equivocal cases but requires prospective validation.