Objective: To demonstrate the clinical utility, sensitivity, and specificity of standard magnetic resonance imaging (MRI) sequences in differentiating temporal bone cerebrospinal fluid leaks from all other middle ear effusions.
Study design: Retrospective imaging review.
Setting: Academic medical center.
Subjects: Patients with cerebrospinal fluid leaks or other middle ear effusions who also underwent MRI.
Methods: Patients were assigned to cerebrospinal fluid leak and other effusion cohorts based on clinical course, findings at surgery/myringotomy, and beta-2 transferrin fluid analysis. Reviewers blinded to the clinical outcome examined T1-weighted, T2-weighted, diffusion-weighted, fluid-attenuated inversion recovery (FLAIR), and 3-dimensional (3D) acquired T2-weighted MRI sequences. For each sequence, fluid imaged in the temporal bone was graded as either similar or dissimilar in signal intensity to cerebrospinal fluid in the adjacent subarachnoid space. Signal similarity was interpreted as being diagnostic of a leak. Test characteristics in predicting the presence of a leak were calculated for each series.
Results: Eighty patients met criteria (41 leaks, 39 other effusions). The 3D T2 series was 76% sensitive and 100% specific in diagnosing a leak, and FLAIR was 44% sensitive and 100% specific. The T1-weighted (73% sensitive, 69% specific), T2-weighted (98% sensitive, 5.1% specific), and diffusion-weighted (63% sensitive, 66% specific) series were less useful.
Conclusions: MRI, with attention to 3D T2 and FLAIR series, is a noninvasive and highly specific test for diagnosing cerebrospinal fluid leak in the setting of an indeterminate middle ear effusion.
Keywords: cerebrospinal fluid leak; imaging; magnetic resonance imaging; skull base; temporal bone.