Poster C65, Friday, August 17, 10:30 am – 12:15 pm, Room 2000AB
Post-stroke impairment of auditory comprehension is associated with changes in cerebral blood flow
Grigori Yourganov1, Jaleel Jefferson1, Julius Fridriksson1, Brielle Stark1, Christopher Rorden1;1University of South Carolina
INTRODUCTION Lesion-symptom mapping is the most common analysis of the relationship between post-stroke brain damage and behavioral impairments. However, this analysis is strongly dependent on the underlying vasculature: it can only detect the regions that are lesioned in a sufficient number of participants. Theoretically, a brain region which might be behaviorally important but spared by the stroke is not likely to be identified by lesion-symptom mapping. We tested this hypothesis by analyzing regional cerebral blood flow (CBF) data, performed on a set of participants with post-stroke aphasia. METHODS 141 people with middle cerebral artery (MCA) strokes and various degrees of post-stroke speech impairment were evaluated using Western Aphasia Battery. Their structural brain damage was assessed by T1- and T2-weighted imaging, and CBF was assessed with Arterial Spin Labeling technique. Binary lesion maps were manually demarcated on T2-weighted images. We used the NiiStat MATLAB toolbox to see how the participants’ auditory comprehension scores were related to CBF maps, and, separately, to lesion maps. For this purpose, we used univariate analysis (VLSM for lesion maps, GLM for CBF maps) as well as multivariate analysis (Support Vector Regression for both CBF and lesion maps, with a leave-one-out framework for out-of-sample prediction). For lesion analysis, only the voxels damaged in at least 10 participants were considered; CBF analysis was performed on the whole brain. RESULTS Using multivariate analysis, auditory comprehension scores could be predicted from lesion and CBF maps with approximately the same accuracy (Pearson’s correlation coefficient between actual and predicted scores: r = 0.6534 for CBF, and r = 0.6541 for lesion maps; p < 10^-18). For lesion analysis, this prediction was largely driven by damage to superior and middle temporal regions (stretching from temporo-parietal junction to temporal pole) and supramarginal gyrus. In CBF analysis, in addition to these regions, we found that impairment of auditory comprehension was associated with reduced CBF to cingulate gyrus, cuneus, and basal ganglia, as well as with increased CBF to several right-hemisphere regions (among others, ventromedial prefrontal cortex and cingulate and inferior frontal gyri). These findings were also supported by univariate analysis. DISCUSSION Compared with lesion maps (which are usually drawn on high-resolution T1- or T2-weighted images), the ASL-derived CBF maps have poor spatial resolution and signal-to-noise ratio. However, they are less dependent on the underlying vasculature, and can help us to identify the brain regions that might be involved in speech but are not commonly lesioned. We showed this by identifying a set of regions where reduced CBF is associated with impairment of auditory comprehension; these regions are perfused by anterior and posterior cerebral arteries and therefore cannot be identified by lesion analysis of our sample of MCA stroke survivors. In addition, CBF analysis revealed potential “luxury perfusion” in the right hemisphere, i.e. increased blood flow associated with behavioral impairment. We suggest that CBF maps can provide the information about the association between brain damage and behavioral impairment which is not available in lesion maps.
Topic Area: Methods