Slide Slam A8
Functional Cortical Reorganization Supporting Therapy-Induced Language Recovery in Chronic Aphasia: A Large-Scale Clinical Trial
Sigfus Kristinsson1, Anja Zgodic2, Dirk B. den Ouden1,3, Chris Rorden1,4, Argye Hillis1,5, Leonardo Bonilha1,6, Greg Hickok1,7, Alex McLain2, Julius Fridriksson1,3; 1Center for the Study of Aphasia Recovery, University of South Carolina, Columbia, SC, USA, 2Department of Epidemiology and Biostatistics, University of South Carolina, Columbia, SC, USA, 3Department of Communication Sciences and Disorders, Columbia, SC, USA, 4Department of Psychology, University of South Carolina, Columbia, SC, USA, 5Department of Neurology and Physical Medicine & Rehabilitation, Johns Hopkins School of Medicine; Department of Cognitive Science, Johns Hopkins University, Baltimore, MD, USA, 6Department of Neurology, Medical University of South Carolina, Charleston, SC, USA, 7Department of Cognitive Sciences and Language Science, University of California, Irvine, CA, USA
Introduction The neural mechanisms supporting language reorganization following aphasia therapy remain elusive (e.g., Schevenels, Price, Zink, De Smedt, & Vandermosten, 2020; Wilson & Schneck, 2020). The lack of concrete evidence may, in part, stem from the fact that most prior studies have been grossly underpowered due to small sample sizes (all 32 studies in Schevenels et al.’s review included N < 30). Here, we compared competing hypotheses regarding the optimal functional reorganization supporting therapy-induced language recovery in aphasia by modeling therapy response based on baseline activity and activity changes in (a) left hemisphere language regions, (b) right hemisphere homologue regions, (c) bilateral domain-general regions, and (d) perilesional regions. Method A total of 93 patients with chronic aphasia (>12-months post-stroke) underwent baseline language assessments followed by 30-hours of anomia therapy (15 hours of semantic and phonological therapy in a cross-over design). Task-based functional magnetic resonance imaging (fMRI) was carried out at baseline and after language therapy. Participants were asked to name common noun pictures and to stay silent when presented with abstract images; functional activity specific to naming was obtained by contrasting naming-related activity and activity elicited by abstract image viewing. Therapy response was predicted from baseline functional activity and pre- to post-therapy activity changes in preselected regions-of-interest (Blank et al., 2014; Fridriksson et al., 2012; Fridriksson et al., 2016). The following factors were identified as independent predictors of therapy response and were therefore included as covariates in all prediction models: initial aphasia severity (Western Aphasia Battery-Revised Aphasia Quotient; Kertesz, 2007), participants’ age at testing, presence/absence of diabetes, and lesion location. Results Overall therapy gain was most accurately predicted by baseline activity in right hemisphere language region homologs (R2 = 0.474, 95% CI: 0.231 – 0.697) and residual left hemisphere language regions (R2 = 0.476, 95% CI: 0.235 – 0.703). Activity changes in residual left hemisphere language regions (R2 = 0.427, 95% CI: 0.191 – 0.671) and right hemisphere homolog regions (R2 = 0.435, 95% CI: 0.194 – 0.683) similarly yielded accurate predictions of therapy response. Baseline activity in the right hemisphere angular gyrus (ß = 0.284) and left hemisphere posterior middle temporal gyrus (ß = 0.380), and activity changes in the left hemisphere insula (ß = -0.078) and right hemisphere posterior insula (ß = 0.511) emerged as the strongest predictors of therapy response in these models, respectively. Functional activity in domain-general and perilesional regions was not found to reliably predict therapy response in this cohort. Conclusion The current study represents the largest fMRI study of therapy-induced language recovery in aphasia to date. Our findings revealed that baseline activity in both left hemisphere language regions and contralateral regions in the right hemisphere predicts therapy response to a similar degree when variability explained by common behavioral variables and lesion damage is accounted for. Our findings emphasize the need to consider individual factors such as lesion size and location in relation to functional integrity to inform the neurobiological mechanisms of treated aphasia recovery. Future research should examine neural reorganization in subsamples guided by lesion characteristics.