Brain Diffusion Tensor Imaging: Useful in Predicting Neurologic Outcomes After Cardiac Arrest?

Determining prognosis of comatose patients following a cardiac arrest is difficult. A detailed neurologic exam to predict neurologic outcome is challenging because standard early predictors of poor outcome after cardiac arrest (e.g., absence of brain stem reflexes and motor response) are unreliable due to the effects of sedatives and neuromuscular blocking agents used during the first week. Among patients with a good clinical outcome, it is impossible to separate those who will recover completely from those with significant neurologic sequelae. Other predictors, such as serum concentration of neuron-specific enolase, tau protein, and somatosensory evoked potentials, have substantial limitations and are not completely reliable.

For this prospective, observational, cohort study done in 14 centers in France, Italy, and Belgium, 185 patients aged 18 years or older at time of cardiac arrest who remained unconscious at 7 days were enrolled in a derivation cohort, of whom 150 patients had an interpretable multimodal MRI and were included in the analysis. Favorable neurologic outcome (Glasgow-Pittsburgh Cerebral Performance Category 1 or 2) at 6 months occurred in 33 patients (22%). Prognostic accuracy was significantly higher with the normalized whole brain white matter fractional anisotropy (WWM-FA) value than with standard criteria for unfavorable outcome and other MRI sequences. In a validation cohort of 50 patients, a normalized WWM-FA value lower than 0.91, set from the derivation cohort, had a negative predictive value of 71.4% (95% confidence interval, 41.9–91.6) and a positive predictive value of 100% (90.0–100) with 89.7% sensitivity (75.8–97.1) and 100% specificity (69.1–100) for prediction of unfavorable outcome.