Slide Slam P10
Functional network changes in post-stroke and primary progressive aphasia: Shared and distinctive features
Yuan Tao1, Kyrana Tsapkini2, Brenda Rapp1; 1Department of Cognitive Science, Johns Hopkins University, 2Department of Neurology, Johns Hopkins School of Medicine
Very similar aphasia profiles can result from stroke (PSA) and neurodegenerative diseases (e.g., primary progressive aphasia -- PPA). However, it is not known whether the two aphasia etiologies produce similar functional network disruption. Moreover, it is not understood how the different temporal characteristics of the etiologies (static, focal vs. progressive) affect functional network (re)organization. To address this issue, we compared the functional network properties of PSA and PPA using graph-theoretic methods. Importantly, both groups had overlapping damage epicenters (left inferior frontal lobe) and language characteristics, allowing us to attribute functional network differences to the different pathological processes. Methods We examined resting-state fMRI in 18 individuals with PSA, 18 with non-fluent PPA, and 14 age-matched healthy controls (HC). PSA and PPA groups were matched in age, education, gender and damage epicenter. Functional connectomes were constructed for each participant and for each hemisphere (LH connectome: 111 nodes, RH connectome: 122). Nodes with substantial damage in each participant were excluded. We examined the graph-theoretic metrics: normalized global efficiency (GE) and clustering coefficient (CC), calculated from binary graphs with fixed edge density (10%-40). GE quantifies the extent of global integration of nodes, with higher values indicating that, on average, fewer steps (connections) separate each node from every other node. CC measures local functional integration, with higher average values indicating greater pervasiveness of small cliques (triangles). First for both measures we carried out a mixed-effects regression analysis to compared the hemisphere averages of each aphasic group to the HC’s and evaluated if effects were consistent across hemispheres. Second, in regional analyses, we investigated whether hemispheric effects were driven by the GE and CC characteristics of specific brain regions. Results In terms of average hemispheric effects we found: (1) for GE, both aphasic groups had similarly lower values than HCs in both hemispheres (PSA: p=0.08; PPA: p=0.07); (2) For CC, there was a hemisphere-by-group interaction with normal CC levels in the LH (HC-PSA: p=0.68, PPA p=0.43) but significantly higher levels in the RH (HC-PSA: p=0.05, PPA: p=0.0008). The results revealed that both aphasic groups exhibited more segregated, less integrated functional network organization, not only in the lesioned left hemisphere but also in the right. In terms of regional effects, we found: (1) In terms of GE, for PSA lower values were concentrated in LH peri-lesional and in RH contra-lesional areas. In contrast, for PPA, decreased GE was observed in bilateral dorsal parietal areas; (2) With regard to CC, for PSA, increases were concentrated in contralesional RH areas. Summary Despite very similar lesion foci, both PSA and PPA groups exhibited very distinctive regional effects. This suggests that functional network changes in PSA may be driven by the focal and static nature of the damage to neural tissue whereas in PPA the temporally extended time-course of the disease may play a critical role. Nonetheless, changes in both aphasia types resulted in overall more segregated, less integrated networks, which may reflect a common re-organization mechanism in response to neural injury, regardless of etiology.