Magnetic resonance imaging (MRI) data of the brain may improve the prediction of neurological recovery from cardiac arrest (CA), according to study results published in Neurology.
Study researchers in Belgium analyzed data from the Neuroprotect Post-CA Study (Clinicaltrials.gov Identifier: NCT02541591). Patients (N = 102) who had out-of-hospital CA and who remained unconscious on hospital admission were given an MRI 4 to 6 days after their CA and a neurological exam on day 5. On day 180, patient recovery was assessed using the Cerebral Performance Category scale.
An MRI was performed in a total of 79 patients and clinical data were available in 58. A neurologist blind to MRI data predicted a poor prognosis in 23 patients (39.7%). A total of 33 patients (57%) had a poor prognosis and 25 (43%) a good prognosis.
Continue reading
Patients with a good and bad prognosis had a median age of 61 years (interquartile range [IQR], 52-71) and 66 (IQR, 58-74) years and 64% and 76%, respectively, were men. Patients with a good prognosis were more likely to have ventricular fibrillation (P <.01), a higher Glasgow Coma Scale (P <.01), a higher mean apparent diffusion coefficient (ADC) (P <.01), and fewer brain voxels with ADC <450 × 10-6 mm2 / s (P <.01).
The prediction of poor neurological outcome based on the percentage of brain voxels <650 × 10 -6 mm2 / s had an area under the receiving operator characteristic curve (AUC) of 0.59 (95% CI, 0.45-0, 72). A cutoff of 23.4% corresponded to a specificity of 100% and a sensitivity of 38.5%.
To predict good neurological outcomes, the percentage of brain voxels must be <450×10-6 mm2/s had an AUC of 0.67 (95% CI, 0.55-0.79). The optimum cutoff was ADC >931 x 10-6 mm2 / s with 100% sensitivity and 38% specificity.
The best predictors of good neurological recovery were the mean ADC of the postcentral cortex (AUC 0.78; 95% CI 0.68-0.88) and the percent voxels in the temporal cortex with an ADC <450 × 10-6 mm2 / s (AUC, 0.73; 95% CI, 0.61-0.84).
The final predictive model included 3 clinical features and 4 MRI features and had an AUC of 0.96 (95% CI, 0.91-1.00; P = 0.03) and a false positive of 27%. The model without MRI data had an AUC of 0.89 (95% CI, 0.81-0.97; false positive, 39%). The addition of brain phenotypes improved the prediction of neurological recovery in 7% of patients.
This study could contain some selection bias due to the large amount of missing data.
Based on their results, the study researchers concluded that “the brain MRI is predictive of good neurological recovery …
reference
Wouters A, Scheldeman L, Plessers S, et al. Added value of the quantitative apparent diffusion coefficient values for neuroprognostics after cardiac arrest. Neurology. 2021; 96 (21): e2611-e2618. doi: 10.1212 / WNL.0000000000011991
