Heavily T2-weighted MR myelography vs CT myelography in spontaneous intracranial hypotension

To assess the diagnostic accuracy of heavily T2-weighted magnetic resonance myelography (MRM) in patients with spontaneous intracranial hypotension (SIH). Patients with SIH were recruited prospectively, and first underwent MRM and then computed tomographic myelography (CTM). The results of MRM were validated with the gold standard, CTM, focusing on 1) CSF leaks along the nerve roots, 2) epidural CSF collections, and 3) high-cervical (C1-3) retrospinal CSF collections. Comparisons of these 3 findings between the 2 studies were made by kappa statistics and agreement rates. Targeted epidural blood patches (EBPs) were placed at the levels of CSF leaks if supportive treatment failed. Nineteen patients (6 men and 13 women, mean age 37.9 +/- 8.6 years) with SIH completed the study. MRM did not differ from CTM in the detection rates of CSF leaks along the nerve roots (84% vs 74%, p = 0.23), high-cervical retrospinal CSF collections (32% vs 16%, p = 0.13), and epidural CSF collections (89% vs 79%, p = 0.20). MRM demonstrated more spinal levels of CSF leaks (2.2 +/- 1.7 vs 1.5 +/- 1.5, p = 0.011) and epidural collections (12.2 +/- 5.9 vs 7.1 +/- 5.8, p < 0.001) than CTM. The overall level-by-level concordance was substantial for CSF leaks along the nerve roots (C1-L3) (kappa = 0.71, p < 0.001, agreement = 95%) and high-cervical retrospinal CSF collections (C1-3) (kappa = 0.73, p < 0.001, agreement = 92%), and moderate for epidural CSF collections (C1-L3) (kappa = 0.47, p < 0.001, agreement = 72%). Ten of the 14 patients (71%) receiving targeted EBPs experienced sustained symptomatic relief after a single attempt. Heavily T2-weighted magnetic resonance myelography was accurate in localizing CSF leaks for patients with spontaneous intracranial hypotension. This noninvasive technique may be an alternative to computed tomographic myelography before targeted epidural blood patches.