Sensitivity of orbital magnetic resonance imaging in acute demyelinating optic neuritis

ABSTRACT Objective To determine the sensitivity of orbital magnetic resonance imaging (MRI) in acute demyelinating optic neuritis (ON) in routine clinical practice, and the added value of a dedicated neuroradiology interpretation. Design Retrospective chart review. Participants Patients with clinically proven ON evaluated between 2004 and 2014 in the University of Michigan neuro-ophthalmology clinics. Inclusion criteria involved visual recovery and orbital MRI completed within 30 days of symptom onset and before corticosteroid treatment. Methods Demographics, clinical examination, and MRI report data (high T2 signal, gadolinium contrast enhancement) were abstracted for each eligible eye. Every MRI was reinterpreted by a neuroradiologist masked to the affected side. Descriptive statistics summarized patient and eye characteristics. Interrater agreement between the neuroradiologist and the radiology report for the radiographic diagnosis of ON was assessed with Cohen's kappa statistic. Results Of 92 patients who met all inclusion criteria, 70 (76.1%) were reported to have at least 1 MRI feature consistent with ON. After dedicated review by a neuroradiologist, 77 (83.7%) were determined to have a positive MRI for ON. Agreement between the neuroradiologist and MRI report was moderate (κ = 0.63). Gadolinium enhancement was the most common feature in MRI positive ON (72 [78.3%] of neuroradiology reviewed MRIs; 66 [71.7%] of clinical MRI reports). Conclusions The sensitivity of MRI in ON was lower than previously reported and confirms the importance of making a clinical diagnosis of ON without relying on neuroimaging for confirmation. MRI interpretation by a skilled neuroradiologist increased sensitivity, underscoring the complexity of orbital MRI interpretation.