Localization of spinal dural arteriovenous fistulas from the spatial relationships of perimedullary vessels on standard MRI

Abstract

OBJECTIVE The goal in this study was to explore the spatial relationship of perimedullary vessels visualized on MRI to localize the side and the site of spinal dural arteriovenous fistula (SDAVF). METHODS A retrospective analysis of 30 consecutive patients diagnosed with SDAVF on MRI was undertaken. Two experienced reviewers blinded to all reports and angiographic images analyzed T2-weighted as well as postcontrast T1-weighted sequences. A focal prominent zone of perimedullary vessels with lateralization to one side in the thecal space was evaluated to locate the side and the site of the fistula. Spinal digital subtraction angiography served as the gold standard technique. RESULTS Good interrater agreement (κ = 0.77) was shown for the diagnosis of SDAVF with perimedullary vessels on T2-weighted MRI. Flow voids on T2-weighted MRI demonstrated a sensitivity of 1.0 (95% CI 1.0–1.0) and an accuracy of 0.87 (95% CI 0.79–0.95) to identify the presence of fistula. The flow voids on T2-weighted MRI also demonstrated 0.88 (95% CI 0.71–1.03) sensitivity and 0.81 (95% CI 0.70–0.92) accuracy to identify the side of SDAVF. Furthermore, flow voids on T2-weighted MRI showed 0.87 (95% CI 0.71–1.03) sensitivity and 0.87 (95% CI 0.79–0.95) accuracy to identify the site of SDAVF within 3 vertebral levels above or below the actual site. Area under the receiver operating characteristic curve demonstrated significant results (0.87 [95% CI 0.73–1.0]; p < 0.001) for flow voids on T2-weighted MRI to identify the site of shunts within 3 vertebral levels in the cranial or caudal direction. CONCLUSIONS Spatial distribution of perimedullary vessels observed on standard MRI show promise to locate the side and the site of fistula in patients with SDAVF.