Abstract 3752: Performance Of Color ADC Maps As A Prognostic Tool In Comatose Post-cardiac Arrest Patients
Background: Predicting outcome for comatose post-cardiac arrest survivors is challenging and compounded by the use of therapeutic hypothermia and sedative agents in recent years. Previous studies suggest that brain abnormalities on MRI are predictive for poor outcome. MRI based predictive factors are attractive because they are not affected by drugs or metabolic derangements; however, most of the methods proposed require image post-processing with specialized software. We assessed the prognostic value of color apparent diffusion coefficient (ADC) maps in a prospective study.
Methods: Consecutive patients who remained comatose after cardiac arrest were prospectively enrolled. Color ADC maps were created by assigning computed ADC values to 8 colors of spectrum ranging from red to blue (Figure). The color ADC maps were not available to the clinical teams caring for the patient. Two raters (a neurocritical care/stroke neurologist and a medical student) independently and blinded reviewed the color ADC maps and predicted 3 month outcome as poor (Glasgow Outcome Scale (GOS) 1 or 2), impaired (GOS=3) or good (GOS of 4 or 5). Both raters were “trained” by viewing 4 examples of patients with good, impaired and poor outcomes. A 3 month GOS of 3-5 was considered a favorable outcome. The agreement between raters and the predictive performance of the color ADC maps were assessed.
Results: 112 color ADC maps of 94 patients (56% with poor, 12% with impaired, and 32% with good outcome) were reviewed: age 59±15 years, 36% females, 69% underwent therapeutic hypothermia, median (IQR) arrest duration 20min (14-30), and time between the arrest and MRI 82hours (60-141). Kappa with quadratic weighting for agreement on predicting all 3 levels of outcomes was 0.74, while kappa for favorable vs. unfavorable outcome was 0.76. For the two reviewers, the sensitivity for predicting poor outcome was 0.85 (95%CI 0.73-0.92) and 0.78 (0.66-0.87), the specificity 0.81(0.66-0.90) and 0.74(0.59-0.86), and the true positive predictive rate 86% (74-93%) and 81% (69-89%), respectively. After excluding early (≤24 hours) and late (>120 hours) scans (ADC changes are time dependent and most apparent after day 1 and before day 6), the specificity improved to 0.87 (0.68-0.96) and 0.77 (0.57-0.89), respectively.
Conclusion: MRI color ADC maps hold promise as a useful and easy to interpret adjunct for predicting outcome of comatose post-cardiac arrest patients in the first few days after the arrest. Since these maps do not require post-processing and can be created in real-time, they can easily be implemented in the clinical setting.
- © 2012 by American Heart Association, Inc.