We read with interest the article by Wijdicks et al (1) on the role of MR imaging as a confirmatory study in survivors of cardiac arrest. The authors, using fluid-attenuated inversion recovery (FLAIR) and diffusion-weighted (DW) imaging, demonstrated diffuse signal intensity abnormalities in the cerebellum, thalamus, frontoparietal cortices, and hypothalamus that were not initially visualized on CT scans. This study and several others stress the important new role of neuroimaging in predicting the neurologic outcome after global ischemia (2–4).
We would like to draw the attention of the authors to a previously published study (3) in which CT findings were used as predictors of the outcome after cardiac arrest. Wijdicks et al report an interesting discrepancy between MR imaging and CT results. This discrepancy supports our previous findings (5) that a loss of gray matter (GM)–white matter (WM) differentiation (GWMD) on CT scans, when visually assessed, is not a reliable predictor of a poor outcome or death after cardiac arrest. We found a poor interrater reliability (κ = 0.3). Interrater agreement was only 58% with an actual agreement beyond chance of 20%. This finding prompted us to use a more quantitative assessment of the loss of GWMD. Hence, we measured the Hounsfield unit density of GM and WM in several regions of interest on nonenhanced CT scans of the brain. The most important area was at the basal ganglial level (BGL). Using an analysis of the receiver operating characteristic curve, we determined that a Hounsfield unit ratio of GM (caudate) to WM (posterior limb of the internal capsule) of less than 1.18 at the BGL was 100% predictive of death with an odds ratio of death of 21.67 (3).
Although there is no doubt that MR and DW imaging are far more accurate than CT scanning in identifying early ischemic changes, the use of this technique is still limited by several factors, such as the lack of availability and expertise needed to interpret DW images in several centers and the long duration of the test, which may expose critically ill patients to undue risks. Wijdicks et al were able to obtain MR images in only 37% of their patients, with additional advantage of having the anesthesia and MR imaging departments located on the same level as the intensive care unit. This level of support is not readily available in most hospitals.
We believe that, until MR imaging technique overcomes some of these limitations, the quantitative measurement of the loss of GWMD with CT may be a good substitute in several medical centers.
- Copyright © American Society of Neuroradiology
