Relationship between cortical connectivity and aphasia therapy response

Abstract

Introduction Individual response to aphasia therapy is highly variable and difficult to predict. While different therapies for word retrieval are often given based on assumptions regarding the underlying breakdown, behavioural symptoms do not consistently predict which type of naming therapy will be most effective. Measures of cortical connectivity derived from EEG before and after therapy may provide additional important information for predicting and understanding response to therapy. Methods Twenty-three adults were recruited including eight adults with chronic, post-stroke aphasia who successfully completed 12 therapy sessions over 4 weeks, alternating between semantic and phonological approaches to word retrieval based on semantic feature analysis and phonological components analysis. Fifteen age-matched healthy controls were included for comparison. High-density electroencephalography (128 channel EEG) was measured during a picture-word judgement task performed before and after treatment for the aphasia group. Analysis of EEG via a dynamic causal modelling (DCM) approach was used to assess semantic and phonological processing pre and post therapy. DCM was applied using five fronto-temporo-parietal regions of interest based on source imaging analysis in the control and aphasia groups. Results Post-treatment naming improvement was associated with cortical responses measured bilaterally in a DCM connectivity model. Specifically, naming improvement for items treated with semantic feature analysis was correlated with (a) increased pre-treatment coupling between the left inferior parietal lobule and left inferior frontal gyrus (r = 0.63, pFDR=0.016) (b) increased coupling from left to right inferior parietal lobule (r = .77, FDRp = .0005) and (c) decreased pre-treatment coupling between the right inferior parietal lobule and right anterior middle temporal gyrus (r = -0.76, pFDR = 0.03). Post-treatment, reduced coupling between right inferior frontal gyrus and right posterior superior temporal gyrus was also significantly correlated with naming improvement for semantic treatment items (r= −0.53, pFDR=0.010). Conclusions This preliminary study on a small cohort of individuals with chronic aphasia has demonstrated the potential of DCM connectivity models derived from EEG data to predict and understand aphasia therapy response. In addition to highlighting the importance of connectivity within left hemisphere networks and inter-hemispheric coupling for treatment, the observation of reduced right intrahemispheric coupling in those individuals with improved outcomes supports the view that right hemisphere mechanisms may not invariably support treatment-induced recovery. Elucidating effective ipsilateral and contralateral connectivity before and after aphasia treatment provides new insights into therapy-induced reorganisation of cortical networks associated with successful therapy outcomes.