Presentation

Search Abstracts | Symposia | Slide Sessions | Poster Sessions | Poster Slams

Accounting for interindividual variability in normative language network organization to explain post-stroke aphasia deficits

Poster A33 in Poster Session A, Thursday, October 6, 10:15 am - 12:00 pm EDT, Millennium Hall

W. Tyler Ketchabaw¹, Candace M. van der Stelt¹, Alycia B. Laks¹, Sachi Paul¹, Elizabeth L. Dvorak¹, Sarah F. Snider¹, Andrew T. DeMarco¹, John D. Medaglia², Peter E. Turkeltaub^1,3; ¹Georgetown University School of Medicine, Washington, D.C., ²Drexel University, Philadelphia, PA, ³MedStar National Rehabilitation Hospital, Washington, D.C.

Introduction: Language function relies on a network of connected brain regions, and there is a high degree of interindividual variability in language network organization in the typical population. Given this variability, the same lesion might cause different outcomes depending on idiosyncrasies of the brain in which it occurs. Yet, the potential impact of this natural variability has not been considered in prior studies of post-stroke aphasia. To account for this variability, we employ a cohort of typically-aging adults to derive a set of normative language networks (NLNs), which catalog the combinations of nodes that activate during a language task in each typically-aging participant. We then quantify lesion-induced disruption by calculating graph theory measures of each NLN in lesioned brains, using connectivity data from a cohort of chronic left-hemisphere stroke survivors. Finally, to investigate if NLN disruption underpins aphasia deficits, we correlate graph theory measures to language scores. Methods: NLNs were derived from 51 typically-aging adults (27 female, mean age 60.0 ± 11.8 years). Mean activation on a semantic decision fMRI task was calculated in all 234 nodes from the Lausanne 2007 parcellation. Each control participant’s NLN was defined as the top 10% highest activating nodes. To investigate lesion-induced disruption, functional connectivity (FC) data was collected from 52 chronic left-hemisphere stroke survivors (20 female, mean age 60.7 ± 11.8 years). We performed a graph theory analysis on each NLN’s combination of nodes, calculating clustering, global efficiency, and modular FC to quantify segregation and integration in each stroke survivor and the control participant from whom the NLN was derived. This procedure was repeated for all NLNs, such that all control-stroke survivor pairs were examined. Disruption of the language network, factoring in normative variability, was then calculated as the mean difference in graph measures between each stroke survivor and all controls. We then performed a principal component analysis (PCA) on 16 tests of language function in the stroke survivors. Finally, we correlated (Spearman’s rho) each principal component score with measures of network disruption in stroke survivors. Results: Our PCA revealed three principal components explaining 83% of the variance in language test performance. These components are summarized as Naming/Word-finding (NWF); Repetition (REP); and Comprehension (COMP). Correlations showed significant relationships (all P < .005, uncorrected) between NWF and clustering (⍴ = .401); COMP and FC within left temporoparietal cortex [LTPC] (⍴ = .454); and COMP and FC between default mode network and LTPC (⍴ =.404). FC within LTPC also trended towards a significant relationship with REP (⍴ = .358, P = .009). After controlling for lesion size, the relationships with COMP survive but relationships with NWF and REP do not. Conclusions: The effect of a stroke on brain connectivity depends on properties of both the lesion and premorbid network organization. Accounting for variability in the latter may be critical for understanding observed deficits. Deficits may arise from effects on the language network as a whole, as we observed with naming and fluency, or from disruption of specific subnetworks, as we observed with comprehension and repetition.

Topic Areas: Disorders: Acquired, Methods