Corpus callosum involvement in language ability after left-hemispheric stroke


The left hemisphere (LH) is dominant for language in the majority of the healthy population. Patients with LH-damage may show global right-hemisphere (RH) activity for language. This makes interhemispheric transfer a good candidate for a brain plasticity mechanism through which speaking abilities may recover. However, the brain structures enabling this functional reorganization remain largely unknown. In this study, we investigate the involvement of the major white-matter commissural pathway, corpus callosum (CC), in language performance after stroke. The study sample consisted of 19 left-hemispheric stroke patients (mean age = 56, SD = 11) and 22 age, sex and education-matched healthy controls (mean age = 59, SD = 12). All participants were right-handed and underwent high resolution diffusion weighted imaging (DWI). Participants carried out a picture-word interference task (PWI) and the Amsterdam-Nijmegen Everyday Language Test (ANELT), which investigates functional communication. We focused on three portions of the CC: genu, anterior midbody and splenium. Anatomical definition of the three corresponding regions of interest (ROIs) was specified on the MNI brain template following the parcellation proposed by Hofer and Frahm (2006). Subsequently, these ROIs were transferred to each participant’s native space DWI and their corresponding white-matter circuitry was obtained using probabilistic tractography implemented in FSL Probtrackx. The outputs were thresholded at 50% of the robust range and binarized, then used to extract microstructural (Fractional Anisotropy, FA) and macrostructural (Volume) properties of the three CC portions. Picture naming times were shorter for controls than for patients (p = .021). When naming pictures with a distractor word relative to a congruent distractor, there was a significant difference in the congruency effect between patients and controls (p = .030). Finally, patients scored lower on functional communication (p = .013). Significant differences were found between patients and controls’ FA for genu (p = .025) and for anterior midbody (p = .008), but not for splenium. To summarise the location of the lesions across the entire sample, hierarchical clustering was used on the proportion of damage to major gray matter gyri. Preliminary results suggested two separate clusters: one with smaller lesions located mainly in the temporal lobe, the other one with bigger lesions encompassing both frontal and temporal lobes. For the genu, the mean FA value in controls was 0.461 (SD = 0.022), for patients in the small-lesions cluster 0.449 (SD = 0.022) and for patients in the big-lesions cluster 0.430 (SD = 0.030). For the anterior midbody, the mean FA in controls was 0.466 (SD = 0.026), for the small-lesions cluster 0.450 (SD = 0.031) and for the big-lesions cluster 0.407 (SD = 0.065). Finally, the mean FA of splenium for controls was 0.532 (SD = 0.032), for the small-lesions cluster 0.527 (SD = 0.024) and for the big-lesions cluster 0.520 (SD = 0.043). To check for possible relationships between the structural integrity of the CC portions and performance in the behavioral tests, we will adopt an exploratory approach, using a series of Spearman correlations.

Poster presentation at SNL 2020, Sandbox Series
Joanna Sierpowska
Former post-doc; current collaborator