Longitudinal prediction of naming and connected speech in post-stroke aphasia: Contribution of acute white matter characteristics

People with aphasia (PWA) display considerable variability in naming and connected speech during language recovery. While white matter lesion-related, micro- and macro-structural characteristics in the subacute and chronic phases have been associated with these language functions, the predictive value of white matter predictors acquired in the acute phase remains largely unknown. We aimed to assess how acute features of the arcuate fasciculus (AF), inferior longitudinal fasciculus (ILF), inferior fronto-occipital fasciculus (IFOF), uncinate fasciculus (UF), and frontal aslant tract (FAT) contribute to naming and connected speech fluency and efficiency over time. Twenty-eight PWA underwent anatomical and diffusion-weighted magnetic resonance imaging within the first week post-stroke. Naming and connected speech (quantified using correct information units per minute, CIUs/min) were assessed across the acute (0-5 days), subacute (7-15 days), and chronic (>6 months) phases. We conducted correlation and regression analyses using demographics, acute clinical scores, and characteristics of the language-related bundles of interest. Better subacute naming was correlated with higher acute naming, smaller lesions, and lower IFOF-lesion load. Subacute connected speech was correlated positively to acute CIUs/min and negatively to IFOF-lesion load. Regression models revealed that acute naming predicted subacute naming, and a combination of acute CIUs/min and IFOF-lesion load predicted subacute CIUs/min. Acute aphasia severity, as reflected in acute naming or connected speech performances, is a key predictor of subacute outcomes. Furthermore, the contribution of acute white matter tracts appears to vary across symptoms and recovery stages.