Neurotransmitter-Informed Connectome Approach to Language Impairment After Stroke.

Hornberger, Tom; Schulz, Robert; Koch, Philipp J; Feldheim, Jan F; Wróbel, Paweł P; Thomalla, Götz; Magnus, Tim; Saur, Dorothee et al. · Stroke · 2026

cross_sectional · Level IV

Where this comes from

Abstract

Variability in poststroke language outcomes remains insufficiently explained by established clinical system neuroscience concepts. This study examined whether damage to neurotransmitter-informed structural networks is associated with poststroke language impairment. Two openly available cohorts of patients with left-hemispheric stroke were analyzed: the Washington Stroke Cohort (St. Louis), including patients after a first symptomatic stroke (acute phase), and the Aphasia Recovery Cohort (South Carolina), focusing on chronic recovery. Language performance was assessed cross-sectionally using either a comprehensive language battery (Washington Stroke Cohort; 1-2 weeks poststroke) or the Western Aphasia Battery-Revised (Aphasia Recovery Cohort; chronic stage). Individual stroke lesion masks were embedded into normative connectomes weighted by positron-emission tomography-derived density maps of 16 neurotransmitter receptors/transporters. Partial least squares regression and adjusted linear regressions (age, sex, lesion volume, and time poststroke) identified predictors of language functioning. Two hundred seventy patients were included. Washington Stroke Cohort (n=44): mean age, 54.2±12.3 years; 45.5% female; median, 12 days poststroke (interquartile range, 10-14). Aphasia Recovery Cohort (n=226): mean age, 57.8±11.2 years; 38.4% female; and median, 721 days poststroke (interquartile range, 403-1765). Across both cohorts, partial least squares analyses converged on a neurochemical profile in which damage to networks related to serotonergic (5-HT<sub>1a</sub> and 5-HT<sub>2a</sub>) and dopaminergic (D1) receptor distributions showed the strongest associations with poorer language performance. Damage to 5-HT<sub>1a</sub> and D1 networks remained significant in fully adjusted models, improving fit over covariate-only models (all <i>P</i><sub>FDR</sub><0.001; Washington Stroke Cohort: ∆AIC<sub>5-HT1a</sub>=<sub>1.77</sub> and ∆AIC<sub>D1</sub>=0.96; Aphasia Recovery Cohort: ∆AIC<sub>5-HT1a</sub>=25.29 and ∆AIC<sub>D1</sub>=19.96). The disruption of large-scale serotonergic (5-HT<sub>1a</sub>) and dopaminergic (D1) networks is associated with language impairment in acute to subacute and chronic stroke. Neurotransmitter-related network damage, based on normative positron-emission tomography-derived maps serving as a structural proxy of neurotransmitter systems, explained additional variability beyond clinical variables and lesion burden, providing a neurochemically informed network framework for understanding variability in poststroke aphasia. However, given the indirect nature of the measures, implications for clinical translation and targeted rehabilitation strategies remain preliminary.