Accuracy of routine fat-suppressed FLAIR and diffusion-weighted images in detecting clinically evident acute optic neuritis

Background: Contrast-enhanced, fat-suppressed T1-weighted imaging (CET1WI) magnetic resonance imaging (MRI) is quite sensitive in detecting acute optic neuritis (ON), but ON remains a clinical diagnosis. MRI is indicated to evaluate demyelinating brain lesions rather than the optic nerves, while "routine" brain protocols typically include axial FLAIR and DWI.

Purpose: To evaluate the accuracy of axial, fat-suppressed FLAIR and DWI sequences used for our routine brain imaging in detecting acute ON, as compared to CET1WI and the clinical diagnosis.

Material and methods: The clinical data and MRI examinations were retrospectively reviewed of 60 patients presenting to a neuro-ophthalmologist for various visual symptoms. Each patient underwent dedicated neuro-ophthalmologic examination, with axial 5 mm fat-suppressed FLAIR and DWI (part of "routine" brain MRI protocol), as well as 3 mm axial and coronal fat-suppressed CET1WI (part of dedicated orbit MRI protocol). Two neuroradiologists independently evaluated FLAIR and DWI, while CET1WI was reviewed by consensus.

Results: Thirty-one patients were clinically positive, 29 negative for ON (total = 34 positive and 86 negative nerves). The sensitivities of FLAIR, DWI, and CET1WI for ON were 75.7-77.3%, 77.3%, and 89.5%, respectively; the specificities were 90.5-93.5%, 80.4-82.7%, and 86.0%, respectively; the accuracies were 85.7-88.2%, 79.5-81.1%, and 87.0%, respectively. Inter-observer kappa was 0.783 for FLAIR, and 0.605 for DWI; intra-observer kappa was 0.746-0.816 for FLAIR, and 0.674-0.699 for DWI (each P < 0.0001).

Conclusion: Being more specific, but not as sensitive, as dedicated CET1WI in acute ON, axial fat-suppressed FLAIR likely has additional value in evaluating for acute ON in "routine" brain MR protocols evaluating for demyelinating disease, while DWI may be hampered by artifacts.