Slide Slam S3 Sandbox Series
Longitudinal lesion-symptom mapping of recovery of expressive grammar in aphasia
Veronika Vadinova1, Aleksi Sihvonen1, Kim Garden1, Kate O'Brien1, Tracy Roxbury1, Greig de Zubicaray2, Marcus Meinzer3, Katie McMahon2, David Copland1, Sonia Brownsett1; 1University of Queensland, 2Queensland University of Technology, 3University Medicine Greifswald
Introduction: Previous lesion-symptom mapping (LSM) studies have suggested that a large network of regions, incorporating frontal and temporal, and cortical and subcortical regions, are associated with grammatical impairments when using tools designed to specifically assess discrete componets of grammar (e.g., sentence production task) (Ouden et al. 2019, Lukic et al. 2021). However, such discrete measures do not assess a person’s ability to generate and produce grammatically correct speech during realistic communicative scenarios. For the purpose of this study, we define the ability to produce grammatically correct speech during such a scenario as 'expressive grammar'. One approach to evaluating expressive grammar in aphasia is through the assessment of its production during Connected Speech (CS) samples. LSM studies that have used CS to focus on grammatical impairments, have typically invetigated surrogate measures, such as mean length of utterance (MLU) or proportion of closed-class words (Borovsky et al. 2007, Mirman et al. 2019). Importantly, all LSM studies of CS to date have been conducted in chronic post-stroke aphasia, beyond the period of spontaneous recovery. The neural correlates of expressive grammar deficits in early post stroke aphasia remain unknown. Longitudinal LSM analyses, that illuminate the neural correlates of favourable versus unfavourable chronic recovery of a given function, are also missing. In contrast to single timepoint LSM analyses, longitudinal LSM studies investigate the relationship between anatomical damage and the change in impairment between the early and chronic time-points (i.e., recovery index score) (Karnath al. 2011). Research question & Hypotheses: We aim to assess grammatical properties of CS in a group of Individuals With Aphasia (IWA) that have been followed up from early subacute to chronic stage of recovery. We hypothesize, based on evidence yielded in previous studies, that damage to left-hemisphere frontal white matter (particularly the arcuate fasciculus) and several temporal cortical regions (particularly the superior temporal gyrus) will play a pivotal role in unfavourable recovery of expressive grammar in chronic aphasia. In addition, we hypothesize that the neural correlates of early and chronic expressive grammar deficits may differ as a reflection of successful behavioural recovery in some IWA. Methods: 38 IWA a) experienced a single left hemisphere stroke and were diagnosed with aphasia, b) English as first language, and c) had a 3D T1- weighted image acquired at baseline. Lesion masks were manually delineated on T1 sequences. IWA completed the Western Aphasia Battery picture description task at two timepoints: early-subacute stage (2-6 weeks poststroke) and chronic stage (6 months poststroke). CS samples were transcribed, and the Computerized Language Analysis software (CLAN) was used to derive the grammaticality measure (MLU x proportion of grammatical errors). To examine lesion-symptom associations, we will apply a multivariate method, support vector regression LSM (SVR-LSM) (Zhang et al. 2014). We aim to control for the influence of lesion volume, education, naming and white matter hyperintensities. Several SVR-LSM models will be built to explore a) the anatomy of early expressive grammar deficits; b) the anatomy of chronic expressive grammar deficits; c) the anatomy of recovery of expressive grammar deficits.