Utility of MRI in spinal arteriovenous fistula

Abstract

Spinal arteriovenous fistula (SAVF) is a rare but treatable cause of myelopathy. The diagnostic accuracy of MRI for detecting SAVF is unknown. Our objective was to determine the sensitivity and specificity of MRI in the diagnosis SAVF and characterize its radiographic features. We conducted a retrospective case-control study of all SAVF treated at our institution from 1995 to 2010, including patients who presented with myelopathy, had MRIs available for review, and underwent either spinal angiogram or had another diagnosis confirming test. Two blinded board-certified radiologists reviewed a series of MRIs and listed the most likely diagnoses, radiologic findings, and recommended follow-up. Sensitivities and specificities of MRI compared to spinal angiogram were calculated. We additionally conducted a literature review of cases describing MRI findings in spinal dural and perimedullary arteriovenous fistula. We identified 36 cases of SAVF (median age 56, 67% male) and 32 controls (median age 54, 44% male). MRI was sensitive in identifying SAVF as the primary diagnosis in 94% (radiologist A, 95% confidence interval [CI] 0.87-1.02) and 89% (radiologist B, 95% CI 0.79-0.99) of cases. The sensitivity of spinal cord T2 hyperintensity or flow voids was 100% and the specificity of T2 hyperintensity and flow voids was 97%. Among patients with myelopathy, spinal angiography is mandatory in the presence of both T2 hyperintensity and flow voids but may be unnecessary if both of these findings are absent.