Language and Attention Networks Have Distinct Roles in Language Improvement Following an Intensive Comprehensive Aphasia Program

Attention is known to play an important role in language, and attentional deficits have been associated with language impairments in people with aphasia (PWA). A prior study by our laboratory indicated that behavioral measures for PWA participating in an intensive comprehensive aphasia program (ICAP) clustered into 1 language and 1 attention-related factor, with each factor correlated with independent resting state functional connectivity (rsFC) networks. The present study includes additional attention measures and participants to better assess the relationship between attention, language, and rsFC. PWA participated in 120 hours of ICAP therapy over 4 weeks, between April 2018 and August 2022. Participants were evaluated with the Western Aphasia Battery and Conners' Continuous Performance Test, pre- and post-ICAP. rsFC data were collected during functional magnetic resonance imaging scans pre- and post-ICAP. Principal component analysis identified behavioral score associations pre- and post-ICAP. FC matrices and graph measures were evaluated for both PWA and healthy controls. Twenty-three PWA participated (19 included in the final analysis). Data for 179 age and sex-matched healthy controls were taken from a public data set. The principal component analysis indicated 1 language and 2 attention-related principal components (PCs). The attention PCs were labeled Attention Accuracy and Attention Rate, in accordance with their best-associated behavioral measures. Importantly, each PC was associated with distinct networks, including higher-order networks, so-called because of their involvement in higher-order cognition. Notably, the language PC was significantly correlated with the ventral attention-memory-retrieval network rsFC (r=0.67, P=0.003) and the default mode-frontoparietal network rsFC (r=-0.56, P=0.019). Attention Accuracy was significantly correlated with cingulo-opercular-subcortical network rsFC (r=0.60, P=0.011). Attention Rate was significantly negatively correlated with visual-somatomotor network rsFC (r=-0.74, P=0.0007). This study supports the notion that language improvements for PWAs participating in an ICAP are associated with distinct network changes. Importantly, these networks are not restricted to language networks but also include attention, and task-control networks.