Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Short timescale flexibility of functional networks facilitates long-term connectivity changes underlying treatment-induced post-stroke aphasia recovery
Poster Session B, Friday, October 25, 10:00 - 11:30 am, Great Hall 3 and 4
Isaac Falconer1, Maria Varkanitsa2, Anne Billot3,4, Swathi Kiran2,5; 1Chobanian & Avedisian School of Medicine, Boston University, Boston, MA, USA, 2Center for Brain Recovery, Boston University, Boston, MA, USA, 3Department of Neurology, Massachusetts General Hospital & Harvard Medical School, Charlestown, MA, USA, 4Department of Psychology, Center for Brain Science, Harvard University, Cambridge, MA, USA, 5Department of Speech, Language, and Hearing Sciences, Sargent College of Health and Rehabilitation Sciences, Boston University, Boston, MA, USA
Post-stroke aphasia (PSA) is a very frequent sequela of stroke with major implications for quality of life. Recovery trajectories of language ability after stroke varies greatly, with factors like initial aphasia severity and lesion characteristics being among the most reliable predictors. However, a significant portion of this variation remains unexplained. With the goal of elucidating factors and mechanisms underlying interindividual differences in recovery trajectories, our previous work used resting-state functional MRI to investigate dynamic functional connectivity (dFC), which refers to time-varying interregional correlations in the blood-oxygen-level dependent (BOLD) signal, in PSA. Greater temporal variability (TV) of dFC at baseline predicted greater treatment-induced improvement in picture naming. We proposed the following mechanism for this finding: (1) Transient inter-regional synchronization facilitates synaptic plasticity between regions and (2) greater TV represents a greater diversity of connectivity configurations sampled over time, producing more opportunities for plasticity. This study sought to test this mechanism by investigating the relationships between TV, treatment-induced network changes, and behavioral treatment response. This study consisted of retrospective analyses of behavioral and functional imaging data from a cohort of 30 individuals with chronic PSA due to left hemisphere strokes. Each participant received up to 24 2-hr sessions of semantic aphasia therapy. Picture naming accuracy was assessed at baseline, at each session, and post-treatment. Resting-state functional MRI scans were obtained at baseline and post-treatment and were used both for estimation of dFC using a sliding window analysis and for calculation of region-of-interest (ROI or “node”) based static functional connectivity (sFC) matrices. Treatment-induced network connectivity changes were quantified using graph metrics of static functional connectivity (sFC, i.e., time-invariant interregional BOLD signal correlations). Relationships between (1) baseline TV of dFC, (2) treatment-induced network connectivity changes (i.e., sFC graph measure changes), and (3) behavioral treatment response were evaluated using multiple linear regression models accounting for potential confounders such as lesion size and initial severity. Additionally, simulations of healthy (i.e., non-lesioned) neural dynamics were performed using a parametric mean field model to further investigate the influence of short-term dynamics on long-term connectivity changes. Each simulation included one of several alternative Hebbian plasticity rules whereby transient coactivation leads to changes in connection strength. Higher baseline TV was found to be predictive of treatment-induced decreases in node-level strength (i.e., sum of sFC of a node’s connections) (β=-0.282, p=3.04e-05), which were in turn associated with greater behavioral treatment gains (β=-3.38, p=0.00318). These decreases in node-level strength were also significantly associated with global increases in small-worldness (β=-0.0529, p=9e-07), a measure of balance between local clustering of nodes and global network efficiency, which was in turn significantly associated with better behavioral treatment response (β=1.15, p=0.0034). Simulation results were consistent with these findings showing that only plasticity rules that drove down connection strengths between non-hub nodes produced increases in global graph measures similar to those seen in patients with greater treatment gains. Overall, both the experimental findings and simulations provided support for higher TV facilitating node-level changes that result in global network changes that support recovery.
Topic Areas: Disorders: Acquired, Speech-Language Treatment