Alterations to dual stream connectivity predicts response to aphasia therapy following stroke

Background: Predicting aphasia recovery is difficult due to a high variability in treatment response. Detailed measures of treatment response are compounded by a dearth of information that examine brain connections that contribute to clinical improvement. In this study we measure alterations to cortical connectivity pathways during a therapy paradigm to detect whether key brain connections that contribute to language recovery can be detected prior to therapy.

Methods: We conducted a case-control trial with twenty-three adults including eight adults with chronic, post-stroke aphasia. Aphasia patients underwent 12 naming therapy sessions over 4 weeks, consisting of semantic and phonological treatment approaches. High-density electroencephalography (128 channel EEG) was measured prior to therapy and immediately following treatment in patients with aphasia. Analysis via a dynamic causal modelling (DCM) was used to assess which cortical connections significantly correlated with therapy response.

Results: Altered cortical responses in aphasia patients measured bilaterally in a dual stream DCM connectivity model were predictive of treatment-induced improvement in naming. Pre-treatment DCM coupling (i.e., strength of cortical connections) significant correlated with naming improvement for items treated with semantic therapy, as indicated by increased connection strengths between left inferior parietal lobule (LIPL) and inferior frontal gyrus (LIFG, r = .63, p_FDR = .016). In particular, the mediating role of contralateral regions significantly influences overall treatment improvement in the latter stages of stroke recovery.

Conclusions: Our findings identify a potential means to stratify larger cohorts of patients in neurorehabilitation settings into distinct treatments that are tailored to their individual language deficit.