Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Factor-based lesion-symptom mapping of pyramids and palm trees in chronic post-stroke aphasia reveals multiple semantic subsystems of the brain
Poster Session D, Thursday, October 1, 4:30 - 6:30 pm, Wangari Maathai
William Matchin1, Scott Decker1, Rutvik Desai1, Ryan Dixon2, Leonard Bonilha1, Julius Fridriksson1; 1University of South Carolina, 2Medical College of Georgia
Lesion-symptom mapping in stroke-based aphasia on the lesion correlates of semantic deficits has primarily focused on linguistic processing, such as comprehension deficits and semantic paraphasias (Kalenine et al., 2010; Schwartz et al., 2009). However, lesion-symptom mapping of total scores on non-linguistic semantic matching tests such as the Pyramids and Palm Trees Test (PPT; Howard & Patterson, 1992) in stroke-based aphasia have shown variable and inconclusive results, not typically identifying regions classically associated with conceptual-semantic representation and processing (Han et al., 2013; Lacey et al., 2017; Dixon et al., in prep) and show questionable validity (Klein & Buchanan, 2009). We suggest that this inconclusiveness derives in part from the diversity of items in these measures, some which may be better aligned with underlying semantic constructs. Therefore, we performed item response theory (IRT) and factor analysis (FA) in order to identify distinct latent task components and used lesion-symptom mapping to determine their lesion correlates. 110 participants with chronic post-stroke aphasia completed the Pyramids and Palm Trees test and variety of other behavioral measures; lesion data were available for a subset of 100 participants. We performed FA with oblique rotation for polychoric data and IRT of binary responses for all 52 individual items of the PPT. Participant-wise factor loadings were entered into correlational analyses with behavioral measures of language comprehension (WAB-R Auditory comprehension, WAB-R comp; Kertesz, 2007 and NAVS sentence comprehension, NAVS; Cho-Reyes & Thompson, 2012), general intelligence (WAIS-IV Matrix), and total lesion volume (TLV). We then performed univariate region- and voxel-based lesion-symptom mapping within seven semantics-related regions derived from the meta-analysis reported by Binder et al. (2009) (Bonferroni correction): dorsomedial prefrontal cortex (DMPFC), ventromedial prefrontal cortex (VMPFC), inferior frontal gyrus pars orbitalis, posterior cingulate cortex (postCing), angular gyrus, lateral temporal lobe, and medial temporal lobe (medTemp). Fourteen factors with eigenvalues larger relative to parallel analysis were retained. Amount of variance explained by each factor did not strongly predict its lesion and behavioral correlations. Factor 1 was associated with language comprehension deficits and damage to VMPFC. Factor 2 was not robustly associated with behavioral deficits, but was associated with TLV and had robust lesion effects in VMPFC, DMPFC, and postCing. Both of those factors may reflect aspects of semantic control and/or executive function. Factor 8 was associated with language comprehension deficits and damage to medTemp, perhaps reflecting impacted perceptual-conceptual representations. Factor 13 was not significantly associated with damage to any ROI, but revealed substantial parietal lobe damage in voxel-wise analysis and correlated with language comprehension measures as well as TLV, perhaps reflecting semantic knowledge or other processing ability. IRT revealed wide variability of discriminability (range: 0.02-20.08, mean: 2.43, SD: 3.49) and difficulty (range: -10.04-11.17, mean: -1.96, SD: 2.32). Our results suggest that at least some items of the PPT are aligned with dissociable underlying mechanisms and potentially enable identification of distinct semantic subsystems of the brain. Deeper analysis of individual items loading on different factors may enable the development of alternative and potentially more valid measures of non-linguistic semantic processing in stroke-based aphasia.
Topic Areas: Meaning: Lexical Semantics, Disorders: Acquired