You are viewing the SNL 2018 Archive Website. For the latest information, see the Current Website.

Poster A55, Thursday, August 16, 10:15 am – 12:00 pm, Room 2000AB

Beyond left hemisphere versus right hemisphere in aphasic language recovery: Evidence from effective connectivity during lexical-semantic processing

Erin Meier1, Jeffrey Johnson1, Yue Pan1, Swathi Kiran1;1Boston University

Traditional models of neural reorganization in persons with chronic aphasia (PWA) propose that optimal, satisfactory, and poor language recovery are characterized, respectively, by reinstatement of premorbid left hemisphere (LH) language activation patterns; recruitment of perilesional LH issue outside canonical language cortex; and reliance on right hemisphere (RH) homologues of damaged LH regions (Heiss & Thiel, 2006; Anglade, Theil, & Ansaldo, 2014). While some support for this hierarchy exists, a great deal of debate revolves around the compensatory capacity of the RH and perilesional LH regions for language recovery. Activation studies that pit LH activation again RH recruitment have not resolved this debate. Therefore, we used fMRI and dynamic causal modeling (DCM; Friston, Harrison, & Penny, 2003) to interrogate bilateral effective connectivity in 30 PWA and 18 age-matched controls during a semantic feature decision task. During the task, participants decided via button press whether written semantic features applied to pictured items. The DCM model space included seven ROIs (i.e., bilateral inferior frontal gyrus, pars triangularis [IFGtri], bilateral posterior temporoparietal cortex [TPC], posterior inferior temporal gyrus [pITG] and left middle frontal gyrus [LMFG]) selected a priori from the activation literature that were activated at a 2nd-level in PWA and/or controls. DCM models were constructed to test whether task-based connectivity patterns in chronic patients aligned with Heiss and Theil’s (2006) hierarchy. Various combinations of task-modulated intra-LH, intra-RH and interhemispheric connections were modeled, resulting in 14 models split into four families (i.e., Family #1: LH-lateralized connectivity [“normal” processing]; Family #2: bilateral anterior connectivity [posterior damage]; Family #3: bilateral posterior connectivity [anterior damage]; and Family #4: RH-lateralized connectivity [extensive LH damage]). Family-wise Bayesian model selection (Penny et al., 2010) revealed “normal” LH-lateralized models (i.e., family #1) best fit control data (exceedance probability [xp]=0.949). In PWA, model fit was split between family #1 (xp=0.568) and family #3: bilateral posterior connectivity (xp= 0.424). One-sample t-tests on parameters weighted by model evidence across all families revealed significant task-modulated connections in controls included LITG-LTPC, LITG-LIFGtri, LITG-LMFG and LITG-RITG. For PWA, the aforementioned connections were also significant (although weaker than controls) as were LIFGtri-LMFG, LIFGtri-RIFGtri, RIFGtri-RITG and RITG-RTPC. Critically, connection strength was not related to the amount of spared tissue in LH ROIs (p>0.05 across connections). Connections that significantly predicted patients’ fMRI task accuracy were identified via backward stepwise regression and re-entered into a multivariate regression model. This model was significant F(10,18)=2.575, p=0.039, adjusted R-squared=0.360) and included positive predictors, where greater strength of certain connections (i.e., LIFGtri-LMTG, LITG-LMFG, LMFG-LIFGtri, RIFGtri-LIFGtri, RTPC-RIFGtri) was indicative of better task accuracy, as well as negative predictors, where greater strength of other connections (i.e., LMFG-LMTG, LMTG-LMFG, RITG-RIFGtri) was related to poorer task accuracy. In all, task-modulated connections that related to better task accuracy were not confined to interactions between canonical language regions (LIFGtri, LITG, LTPC) nor were all intra-RH and interhemispheric connections maladaptive for task performance. As such, traditional structure-function-behavior recovery models of chronic aphasia must be re-evaluated in light of recent connectivity studies that highlight more nuanced recovery patterns in PWA.

Topic Area: Language Disorders