Poster D51, Friday, August 17, 4:45 – 6:30 pm, Room 2000AB
Rethinking Effects of White Matter Tract Disconnection on Post-Stroke Language Impairment
Jason Geller1, Melissa Thye1, Daniel Mirman1;1University of Alabama-Birmingham
Background : Lesion-symptom mapping studies have generally focused on associations between behavioral deficits and grey matter damage, though there is also recognition that white matter damage contributes to language impairment following left hemisphere stoke. White matter damage can be measured with a continuous variable (i.e., percentage of the white matter tract that has been destroyed by the lesion, known as “lesion load”) or a binary variable – whether the lesion has severed the tract producing a structural disconnection. Recently, Hope et al (2016) demonstrated that high lesion load does not always lead to white matter tact disconnection, and vice versa. Further, Hope et al. found that a simple binary estimate of structural disconnection explained more of the variance in fluency and naming scores than did lesion load. In the current study, we attempt to replicate and extend this finding by examining lesion load and tract disconnection in three white matter tracts that have been implicated in language processing: arcuate fasciculus, uncinate fasciculus, and inferior fronto-occipital fasciculus (IFOF) using a different sample of stroke patients and language measures. Method: Behavioral and structural neuroimaging data for 128 participants with aphasia following left hemisphere stroke were drawn from a large-scale study. We examined the effect of lesion load and tract disconnection on overall aphasia severity (Western Aphasia Battery – Aphasia Quotient), picture naming (Philadelphia Naming Test), and composite measures of speech production ability (word and nonword repetition, phonological errors in picture naming, etc.) and semantic cognition (Camel and Cactus Test, synonym judgments, semantic category discrimination, Peabody Picture Vocabulary Test, etc.). The composite measures were based on a factor analysis closely related to prior work that captured the independent sub-systems of language. The white matter tracts were taken from the tractography atlas published by Thiebaut de Schotten et al. (2011). Lesions and tracts were spatially normalized to the same stereotaxic space (MNI). Registration and calculation of lesion load and tract disconnection were conducted using the ANTsR package in R. Results: Regression analyses were conducted for lesion load and tract disconnection in each white matter tract. For each language score, we performed forward and backward selection to determine the best fitting regression model. Lesion load damage to the arcuate fasciculus significantly predicted speech production deficit severity and overall aphasia severity, and damage to the IFOF predicted semantic deficit. However, lesion load in the uncincate did not predict any of the language outcomes. Tract disconnection did not predict any of the language scores. Conclusion: In contrast to Hope et al., lesion load in the arcuate fasciculus and IFOF were predictive of language impairment, but structural white matter disconnection was not. Our results are consistent with a graded disconnection model in which white matter tract effectiveness is related to the overall integrity of the tract. It is also possible that aligning lesion images with a tractography atlas provides a viable estimate of graded tract damage but is not precise enough to reliably estimate tract disconnection, which may require direct individual measurements of white matter integrity.
Topic Area: Language Disorders