Abstract 193: Imaging Biomarkers to Predict Fluency and Naming in Chronic Nonfluent Aphasic Patients

Abstract

We previously showed that a combined variable of lesion site and size, the Arcuate Fasciculus lesion load (AF-LL), predicted speech fluency in chronic aphasic patients. In the current study, we compare the correlations between speech fluency and naming measures with two lesion markers in a large group of patients: the previously established structural lesion marker (i.e., AF-LL) and a new functional lesion marker (i.e., the functional grey matter lesion load (fGM-LL)). fGM-LL was constructed from functional brain imaging of overt speaking tasks in a group of twelve healthy elderly control participants. High-resolution structural brain images were acquired to reconstruct the arcuate fasciculus (AF) as a structural map. In addition to these two canonical maps, a combined AF-fGM was derived from summing fGM and AF maps. Each canonical map was overlaid with lesion masks of 50 chronic aphasic patients with varying degrees of impairment in speech production to calculate a functional and structural lesion load value for each patient. Lesion loads were then regressed with measures of speech fluency and naming. We found that both AF-LL and fGM-LL independently predicted speech fluency and naming ability (p<0.01); however, AF lesion load explained most of the variance for both measures. The combined AF-fGM lesion load did not have a higher predictability than either AF-LL or fGM-LL alone. Clustering and classification methods confirmed AF lesion load was best at stratifying patients into severe and non-severe outcome groups of speech fluency with 96% accuracy and naming with 86% accuracy. Our classification model showed that an AF-LL of greater than 4cc was the critical threshold that determined poor fluency and naming outcomes. Thus, surrogate biomarkers of language impairments may help healthcare providers make more accurate predictions of outcome and help to stratify patients for experimental studies.