Clinical Value of Dorsal Medulla Oblongata Involvement Detected With Conventional Magnetic Resonance Imaging for Prediction of Outcome in Children With Enterovirus 71-Related Brainstem Encephalitis

BACKGROUND: Brainstem encephalitis is the most common neurologic complication after enterovirus 71 infection. The involvement of brainstem, especially the dorsal medulla oblongata, can cause severe sequelae or death in children with enterovirus 71 infection. We aimed to determine the prevalence of dorsal medulla oblongata involvement in children with enterovirus 71-related brainstem encephalitis (EBE) by using conventional magnetic resonance imaging (MRI) and to evaluate the value of dorsal medulla oblongata involvement in outcome prediction. METHODS: Forty-six children with EBE were enrolled in the study. All subjects underwent a 1.5 Tesla MRI examination of the brain. The disease distribution and clinical data were collected. Dichotomized outcomes (good vs. poor) at longer than 6 months were available for 28 patients. Logistic regression was used to determine whether the MRI-confirmed dorsal medulla oblongata involvement resulted in improved clinical outcome prediction when compared with other location involvement. RESULTS: Of the 46 patients, 35 had MRI evidence of dorsal medulla oblongata involvement, 32 had pons involvement, 10 had midbrain involvement and 7 had dentate nuclei involvement. Patients with dorsal medulla oblongata involvement or multiple area involvement were significantly more often in the poor outcome group than in the good outcome group. Logistic regression analysis showed that dorsal medulla oblongata involvement was the most significant single variable in outcome prediction (predictive accuracy, 90.5%), followed by multiple area involvement, age and initial Glasgow Coma Scale score. CONCLUSIONS: Dorsal medulla oblongata involvement on conventional MRI correlated significantly with poor outcomes in EBE children, improved outcome prediction when compared with other clinical and disease location variables, and was most predictive when combined with multiple area involvement, Glasgow Coma Scale score and age.