Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Reading efficiency after stroke relates to lesions, structural disconnections, and activation in spared language and cognitive nodes
Poster Session A, Friday, September 12, 11:00 am - 12:30 pm, Field House
Ryan Staples1, Elizabeth J. Anderson1, Andrew T. DeMarco1, Peter E. Turkeltaub1,2; 1Center of Brain Plasticity and Recovery, Georgetown University Medical Center, 2Research Division, National Rehabilitation Hospital
The neural systems that underlie reading are frequently damaged in left hemisphere stroke, leading to alexia. Alexia is thought to result not only from direct anatomical damage to the language/reading network, but also dysfunction of unlesioned network nodes. However, few studies of alexia combine both measures of residual function and anatomical lesion damage. Here, we leverage a large sample of left-hemisphere stroke survivors and controls to identify 1) the necessary neural structures for oral reading via lesion-symptom mapping and 2) where residual fMRI reading/language activation supports oral reading. Controls (N=52) and stroke survivors (N=87) read aloud 200 single-syllable words, crossed on regularity, imageability, and frequency. Reading efficiency (mean accuracy/median RT) was calculated for each participant. Stroke survivors also performed behavioral tasks assessing semantic and phonological processing. fMRI activation was measured during an adaptive written semantic relatedness task in all participants. A reading/language network, consisting of eight left-hemisphere frontal and temporal ROIs, and a right cerebellar ROI, were identified from a sample of 15 held-out controls. We used support-vector regression voxelwise lesion-symptom mapping (SVR-VLSM) and connectome lesion-symptom mapping (SVR-CLSM) to identify lesions and structural disconnections associated with reduced reading efficiency, covarying for demographics and lesion size. Separately in controls and stroke survivors, used partial Spearman correlations between ROI-wise fMRI activation and reading efficiency to identify regions where activation supports reading ability, controlling for demographics. For stroke survivors, lesion volume, chronicity, motor speech impairment, and ROI-wise lesion load were also covaried. Exploratory correlations examined the relationship between activation and efficiency on eight categories of words (high/low frequency, imageability, regularity; 100 words/cell). Finally, within the stroke survivors, we correlated semantic and phonological processing measures with nodewise fMRI activation. SVR-VLSM identified a left posterior superior temporal gyrus lesion location, extending into the angular and middle and inferior temporal gyri, that reduced reading efficiency (p=0.03). SVR-CLSM identified disconnections within a network spanning left anterior and middle temporal gyrus, fusiform gyrus, angular gyrus, and supramarginal gyrus (edgewise p’s<0.001). Controls outperformed stroke survivors in semantic decision (t(126.5))=7.7, p<0.0001) and reading (t(119.2)=10.5, p<0.001). Anterior inferior temporal gyrus (aITG) activity correlated with reading efficiency in both groups (controls: r=0.45, p=0.013, stroke: r=0.385, p=0.019), and with semantic processing in stroke survivors (r=0.32, p=0.017). Exploratory Spearman correlations found that dorsomedial prefrontal (dmPFC) activation was related to low-frequency word efficiency, regardless of regularity or imageability (all r’s>0.29, p’s<0.05) in stroke survivors only. The left posterior temporal cortex is a critical lexical convergence zone for reading, where damage or disconnections impair reading efficiency. Separately, activation of the aITG is associated with better reading in typical readers and stroke survivors and better semantic performance after stroke, supporting previous findings linking this region to semantics. In stroke survivors only, dmPFC activation was specifically related to low-frequency word efficiency, suggesting additional cognitive resources may be needed to retrieve infrequent lexical representations after stroke. Together, these results suggest that reading is served by a critical network of temporoparietal regions, supported by task-sensitive activation in semantic and, for stroke survivors, prefrontal cognitive control regions.
Topic Areas: Reading, Disorders: Acquired