Evaluating the diagnostic accuracy of magnetic resonance angiography versus digital subtraction angiography for spinal dural arteriovenous fistulas: a systematic review and meta-analysis.

Spinal dural arteriovenous fistulas (SDAVFs) are underrecognized, and delayed diagnosis can result in irreversible neurological deficits. Magnetic resonance angiography (MRA) provides noninvasive, high-resolution vascular imaging of SDAVFs. Digital subtraction angiography (DSA) is the diagnostic gold standard; however, it is invasive and may yield nondiagnostic results. This is the first study to systematically compare the diagnostic accuracy of DSA and MRA using surgically confirmed SDAVFs as the reference standard. Following PROSPERO registration (CRD42023474375), PubMed, Embase, and Web of Science were searched from inception to March 2024 for adult studies reporting MRA and/or DSA alongside surgically confirmed vertebral-level localization, using PRISMA guidelines. Primary outcomes were pooled sensitivity, positive predictive value (PPV), and accuracy for lesion detection and localization. Thirteen studies comprising 133 patients met inclusion criteria. Across these studies, 102 patients underwent MRA and 83 underwent DSA. Pooled PPV was 0.806 for MRA (95% CI 0.734-0.879) and 0.873 for DSA (95% CI 0.765-0.982; p = 0.277). Pooled sensitivity was higher for MRA (0.984; 95% CI 0.951-1.00) than DSA (0.860; 95% CI 0.793-0.928; p = 0.008), whereas pooled accuracy favored DSA (0.769; 95% CI 0.670-0.868) over MRA (0.644; 95% CI 0.503-0.784; p = 0.041). Using surgical confirmation as the reference standard, MRA demonstrates superior sensitivity to DSA for SDAVF detection, whereas DSA provides higher accuracy, and both have comparable PPV for precise vertebral‑level localization. Standardized, prospective multicenter studies are warranted to refine integrated imaging strategies.