White matter microstructure and verbal fluency

Poor performance on verbal fluency tasks is associated with an increased risk of post-stroke cognitive impairment. Grey matter regions supporting verbal fluency have been identified via lesion–symptom mapping, but the links between verbal fluency and white matter structure remain less well described. We examined white matter correlates of semantic (Category Fluency Animals) and phonemic or lexical fluency (COWAT FAS) after stroke, accounting for stroke severity measured with the National Institutes of health Stroke Scale (NIHSS), age, sex, and level of education. White matter fibre density and cross-section measures were automatically extracted from 72 tracts, using MRtrix and TractSeg software in 72 ischaemic stroke survivors assessed 3 months after their event. We conducted regression analyses separately for phonemic and semantic fluency for each tract. Worse semantic fluency was associated with lower fibre density in several tracts, including the arcuate fasciculus, superior longitudinal fasciculus, inferior occipito-frontal fasciculus, inferior longitudinal fasciculus, optic radiation, striato-occipital, thalamo-occipital tracts, and inferior cerebellar peduncle. Our stroke sample was heterogenous with largely non-overlapping and predominantly right-lateralised lesions (lesion distribution: left N = 27, right N = 43, bilateral N = 2), dissimilar to previous studies of verbal fluency. Yet, the tracts we identified as correlates of semantic fluency were all left-lateralised. No associations between phonemic fluency performance and fibre density metrics in any of the white matter tracts we extracted survived correction for multiple comparisons, possibly due to the limitations in the selection of tracts and sample characteristics. We conclude that when accounting for the effects of stroke severity, sex, age, and education, semantic fluency is associated with white matter microstructure in the left arcuate fasciculus, superior longitudinal fasciculus, and several occipital tracts, possibly reflecting the disconnection in the sagittal stratum. Our results obtained with fixel-based analysis, complement previous findings obtained with lesions–symptom mapping and neurodegenerative approaches.