Abstract TP125: Retina and Optic Nerve Diffusion Restriction in Acute Central Retinal Artery Occlusion: A Case-Control Study

Abstract

Objective: To assess whether diffusion restriction (DR) of the retina and optic nerve (ON) can be accurately and reliably identified on standard stroke protocol brain magnetic resonance diffusion-weighted imaging (DWI-MRI) in patients presenting with acute non-arteritic central retinal artery occlusion (CRAO). Background: Retinal DR (RDR) and ONDR in patients with acute non-arteritic CRAO have been described with various diagnostic accuracies and often identified by experienced neuroradiologists with fair inter-rater agreement. Methods: This retrospective case-control study included patients that presented to our academic center from 2013-2021 with acute non-arteritic CRAO or acute cerebral ischemia (controls, age and gender-matched) and had brain MRI performed within 14 days from symptom onset. Two neuroradiologists, blinded to the site of CRAO and diagnosis, independently reviewed MRI of CRAO cases and controls to assess for the presence of RDR and ONDR. Their inter-rater agreement was calculated (prevalence and bias- adjusted kappa coefficient) and their final consensus was used to perform sensitivity and specificity analyses. Results: We included 128 patients with CRAO (mean age 69 years; 50% males; median time from CRAO to MRI 2 days (IQR 1-5)) and 128 matched controls with acute cerebral ischemia (mean age 68 years; 51% males). After neuroradiologists’ consensus, DR was correctly identified in the retina or ON in 51/128 (39.8%), retina alone 35/128 (27.3%), ON alone 31/128 (24.2%), both retina and ON 15/128 (11.7%), with excellent inter-rater agreement for retina (K=0.91) and ON (K=0.83). There were no cases of RDR or ONDR in the control group. After neuroradiologists’ consensus, sensitivity, specificity, PPV and NPV were 28.1%, 100%, 100%, and 58.2% for RDR and 27.3%, 100%, 100%, and 57.9% for ONDR. Conclusion: Though RDR and ONDR can be identified with excellent interrater reliability, they are infrequent findings and their role in diagnosing acute CRAO is unclear. Observational prospective studies with large cohort of patients, optimization of dedicated and standardized orbit DWI-MRI protocols, and education of neurology and neuroradiology providers are needed to facilitate more accurate identification of RDR and ONDR in hyperacute CRAO.