Can Predict Late Encephalopathy by Diffusion Tensor Imaging in Acute Phase of Carbon Monoxide Poisoning?

Bang, Seung Jae; Kim, Gap Teog

Abstract

PURPOSE
Approximately 30% of patients with acute carbon monoxide (CO) poisoning display chronic neurological symptoms due to late encephalopathy, 1/3 in delayed neuropsychiatric sequelae (DNS) and 2/3 in persistent neurological symptoms (PNS). This study was conducted in order to determine whether diffusion tensor imaging (DTI) can sensitively represent damage to the brain in the acute phase after CO poisoning.
METHODS
This prospective study included 132 patients with CO poisoning from February 2013 to February 2014. The patients were classified according to three groups: patients with asymptomatics; patients with PNS; and patients with DNS. Mean fractional anisotrophy (FA) and apparent diffusion coefficient (ADC) of regions of interest (ROI) in the acute phase were compared between the DNS or PNS groups and the asymptomatic group. DNS was considered present when the patient had clinical symptoms and signs of DNS within six weeks after exposure to CO poisoning.
RESULTS
Late encephalopathy occurred at a rate of 18.9% (12.1% in the DNS, 6.7% in the PNS). DTI cannot sensitively represent brain damage of the DNS groups in the acute phase after CO poisoning. But low FA values of high frontal subcortex and low ADC values of globus pallidus in the DNS group clearly differed from those in the asymptomatic group. The PNS group showed significantly low FA and ADC values in centrum semiovale, several subcortics, globus pallidus, corpus callosum, and internal capsule. No significant difference in ADC was observed between patient groups.
CONCLUSION
In the acute phase of CO poisoning, FA or ADC values sensitively represented damage to white matter in the some subcortex. However it had limited value in predicting the DNS because of no significant reduction of FA value in the centrum semiovale in the acute phase. However low values of FA and ADC in diffuse lesions in deep white matter and subcortex is a powerful predictor of the PNS.