Ilana Spokoyny, MD
The ability to predict post-stroke functional and cognitive outcomes would be very useful in guiding patients and families as to the course of their post-stroke recovery. Specific anatomic locations, predominantly in the left hemisphere, affect language and cognitive processing and it is plausible that injury to those specific areas would disproportionately affect cognitive outcomes. However, damage to other areas may disrupt network functioning and cause global cognitive impairment.
This study sought out to identify and characterize the contribution of functionally important brain areas on cognitive and functional outcomes at 3 months post-stroke. Patients were included who had with no previous disability (modified Rankin Scale/mRS=0) with supratentorial ischemic strokes. Patients with severe cognitive impairment at baseline were excluded. Each patient had an MRI done between 24-72 hours post stroke. The authors determined functionally eloquent areas of brain by calculating the probability that functional scores (in this case mRS or MoCA-Montreal Cognitive Assessment) would worsen if that voxel of brain tissue was affected by an ischemic lesion. Then, maps were created based on this information, a technique known as voxel-based lesion symptom mapping. The information from 215 of the 289 patients was used to create the maps, and the other patients were used as an independent population for validation of the method. The maps for both mRS and MoCA showed a predominance of eloquent brain tissue in the left hemisphere. This lateralization was stronger in the MoCA maps, likely due to language localization.
Multivariable analyses were performed to determine whether stroke location (using number of eloquent voxels as per the VLSM maps), age, NIHSS, and stroke volume predicted good 3 month functional (mRS 0-1) and cognitive (MoCA>25) outcomes. Another cognitive score was calculated, sMoCA, which removed naming and language from the MoCA, to isolate the non-aphasia components.
Stroke location did not hold up to multivariable analysis as an independent predictor of mRS at 3 months; functional outcome was mostly predicted by NIHSS. Stroke location was, however, the strongest independent predictor of cognitive outcome (MoCA score) at 3 months. In patients with MoCA >25 (good outcome), a median of only 2 eloquent voxels were affected, compared with 224 eloquent voxels for MoCA ≤ 25. Including stroke location in the model also improved the predictive ability of non-language cognitive performance (represented by sMoCA), demonstrating that overall cognitive function, not just aphasia, was influenced by stroke location.
Limitations of this study include the use of a somewhat crude functional outcome score (mRS). It is possible that using a more detailed motor or ADL score would have been better predicted by stroke location. In the future, it would be interesting to create a map based on a larger population, and determine the predictive ability of stroke location on cognitive and functional outcomes.
