Frontiers in Psychology (Apr 2014)
White Matter Correlates of Lexical Access in Aphasia
Abstract
INTRODUCTION Neurolinguistic models have coalesced around the view that two distinct pathways support different kinds of processing (Hickok & Poeppel ,2004; Saur et al., 2008): A ventral stream (VS) maps sound to meaning, while a dorsal stream (DS) maps sound to articulation. Dell, Schwartz, Nozari, Faseyitan, & Coslett (2013) found correspondences between this framework and the interactive two-step model of lexical access (Foygel & Dell, 2000). However, there is evidence that the arcuate fascicle (AF), a major component of the DS, also contributes to semantic processing (Glasser & Rilling, 2008; Fernandez-Miranda et al., 2014). We present preliminary results from a study applying advanced fiber mapping techniques to investigate white matter correlates of word production in aphasia. METHOD Fifteen participants with aphasia due to left hemisphere stroke were given the Philadelphia Naming Test (mean %correct = 70%, sd = 23%) and response-type counts were entered into WebFit (http://langprod.cogsci.illinois.edu/cgi-bin/webfit.cgi) to estimate the semantic (s) and phonological (p) connection weight parameters of the two-step model. Diffusion spectrum imaging data (acquired via Siemens 3T Tim Trio Scanner, 32-channel coil, 257 directions with twice-refocused spin-echo EPI sequence, TR = 9616ms, TE = 152ms, voxel size = 2.4mm3, bmax = 7000s/mm2, FOV = 231x231mm) were reconstructed by generalized q-sampling, a high angular resolution-based approach. Orientation distribution functions (directional probability of diffusion) were used to calculate quantitative anisotropy (QA) values for each tract using whole-brain seeding and defined ROIs. ROIs for the AF were the superior, middle, and inferior temporal gyri. We estimated two regression models with s and p parameters as dependent variables and QA values for the left VS (VSQA; inferior fronto-occipital fascicle, extreme capsule, uncinate fascicle) and DS (DSQA; AF, superior longitudinal fascicle) as predictors. RESULTS VSQA and DSQA were both significant positive predictors of s-weight (p < 0.05, multiple r-squared = 0.60, 95%CI: 0.07, 0.88; VSQA semi-partial r-square = 0.31, 95%CI: 0.1, 0.6; DSQA semi-partial r-squared = 0.19, 95%CI: 0.001, 0.55) DSQA was a significant negative predictor of p-weight (p = 0.009, r-squared = 0.41, 95%CI = 0.08, 0.7), with higher values predicting lower p-weights. DISCUSSION We found significant relationships between VS and DS white matter lesions and semantic word production impairments in aphasia. This is consistent with neurolinguistic models that propose a role for the AF in semantic processing (Glasser & Rilling, 2008; Fernandez-Miranda et al., 2014). They are also consistent with Dell and colleagues (2013), who found that s-weight was associated with lesions in both VS and DS cortical areas. However, when they controlled for lesion volume, relationships with DS areas were absent. Our results must be confirmed in a larger sample with control of lesion volume. We also found a counterintuitive negative relationship between phonological processing and left DS integrity. This suggests that reliance on this pathway may be maladaptive for phonological processing in aphasia. Perhaps patients with more intact dorsal pathways continue to rely on this (still damaged) network, while patients with greater damage are more successful at reorganizing function (cf. Parkinson, Raymer, Chang, FitzGerald, & Crosson, 2009).
Keywords
