Slide Slam D10
White matter tracts underlying verbal fluency after stroke
Natalia Egorova-Brumley1,2, Chen Liang1, Amy Brodtmann1,2; 1The University of Melbourne, 2The Florey Institute of Neuroscience and Mental Health
Verbal fluency is known to be associated with stroke risk (Brady et al., 2001) and declines after stroke (M. Babulal, 2017). Previous studies using lesion-symptom mapping have identified a number of gray matter regions associated with verbal fluency, highlighting the importance of frontal and temporal areas (Baldo et al., 2006) and dissociating semantic and phonemic fluency (Biesbroek et al., 2016). Yet a very recent study (Thye et al., 2021) has implicated damage to white matter tracts as the primary correlate for both deficits (particularly semantic fluency), rather than frontal or temporal cortical regions. It further underscored a substantial overlap between both types of fluency, suggesting that semantic fluency and letter fluency largely rely on the same neural system. This study, however, did not control for the general stroke severity, although previous evidence suggests that phonemic fluency relies more on executive processes (Faroqi-Shah and Milman, 2018). Furthermore, while lesion-symptom mapping studies in stroke offer valuable insights, their resolution - mostly dissociating frontal and temporal lesions and the use of cohorts with homogeneous lesions make them potentially less sensitive. We here focused on white matter correlates of semantic (COWAT Animals) and phonemic fluency (COWAT FAS) after stoke, with the goal of identifying white matter tracts associated with semantic and phonemic fluency, taking into account stroke severity, age, sex and level of education, as the latter 3 are known to influence performance (Loonstra et al., 2001). White matter fibre density measure was extracted from 72 tracts, using MRTRIX and TRACTSEG software in 72 participants assessed 3 months post-stroke. The results of regressions conducted separately for COWAT FAS and COWAT Animals for each tract showed that phonemic fluency was associated with the left superior longitudinal fascicle III fiber density (p=0.018), with the left arcuate fascicle and superior longitudinal fascicle II trending (p=0.059 and p=0.054). Importantly, 59 of the 72 tracts showed an association with stroke severity. No results of the COWAT FAS task survived FDR correction for multiple comparisons. For the COWAT Animals, fiber density in 17 tracts was associated with semantic fluency, of which 7 survived the FDR correction, including the arcuate fascicle, inferior cerebellar peduncle, inferior occipito-frontal fascicle, inferior longitudinal fascicle, optic radiation, superior longitudinal fascicle III, striato-occipital and thalamo-occipital tracts. All of the identified tracts were in the left hemisphere. There was an FDR significant effect of education on all tracts, and several tracts showing an association with NIHSS and sex, that did not survive correction. We conclude that verbal fluency after stroke is associated with a set of left-lateralised white matter tracts that show some overlap between semantic and phonemic fluency, namely in the arcuate fascicle and the superior longitudinal fascicle III. However, our findings emphasise that COWAT FAS is primarily associated with stroke severity rather than specific white matter tract microstructure. In contrast, while COWAT Animals task is significantly dependent on the level of education, a number of fronto-temporal tracts are specifically related to performance on the task, highlighting the important role of white matter connections for semantic fluency.