Early Acquisition Timing and CSF Venous Fistula Conspicuity on Lateral Decubitus CT Myelography

BACKGROUND AND PURPOSE: The optimal acquisition timing of lateral decubitus CT myelography (LD-CTM) for visualizing CSF venous fistulas (CVFs) remains uncertain. We compared known-site CVF conspicuity on acquisitions initiated 1 and 2 minutes after the start of intrathecal contrast injection to determine whether timing affected fistula visibility. MATERIALS AND METHODS: This retrospective paired-image study included 35 patients with LD-CTM findings classified as high-probability or definite CVF according to the Duke CSF-Venous Fistula Confidence Score. Paired axial image sets spanning the depicted fistula were selected from the two acquisitions at matched anatomic levels. Three neurointerventional radiologists, blinded to acquisition timing and clinical information, independently assigned a visibility score to each acquisition-specific image set, ranging from 0 (not visible) to 3 (clearly visible). Scores of 2 or 3 met the prespecified threshold for visibility. Reviewers also selected which image set better depicted the fistula. The primary outcome was the difference in patient-level median visibility scores between acquisitions. Rating-level threshold visibility was analyzed using mixed-effects logistic regression, with an exploratory model additionally including measured intrathecal attenuation. RESULTS: The patient-level median visibility score was higher on the 1-minute than on the 2-minute phases (3 [IQR, 2-3] versus 1 [IQR, 0-2]; P < .001). At the reviewer level, threshold visibility was achieved in 92/105 ratings (87.6%) at 1 minute compared with 48/105 ratings (45.7%) at 2 minutes. The odds of threshold visibility were higher at 1 minute (OR, 9.73; 95% CI, 4.70 to 20.15; P < .001). After inclusion of measured fistula-level intrathecal attenuation, a residual phase association remained (attenuation-adjusted OR, 5.02; 95% CI, 2.25 to 11.19; P < .001). The 1-minute acquisition was preferred in 26/35 patients (74.3%), the 2-minute acquisition in 5/35 (14.3%), and the phases were rated equal in 4/35 (11.4%). CONCLUSIONS: Under this standardized fixed-delay protocol, the 1-minute acquisition provided greater known-site CVF conspicuity overall, while the 2-minute acquisition provided complementary information in a minority of patients. These findings support multiphasic imaging but do not establish a universally optimal acquisition time.