P4‐147: Deep subcortical white matter disease is a major risk factor for late post stroke cognitive impairment

the fluent type of PPA (a.k.a. semantic dementia). More recently, we have identified errorless learning ((i.e., preventing patients from making errors during learning process)as the most fruitful rehabilitation paradigm in semantic dementia (Jokel & Anderson, submitted). The literature on recovery of function in post-stroke aphasia shows that intervention not only improves patients’ communication abilities but it also induces lasting and measurable effects in the brain (Abramsky et al., 2001; Peck et al., 2004; Sidaros et al., 2008). There is also emerging evidence that behavioural interventions for various neurologically-based conditions can ameliorate pre-existing neural deficits (De Nil et al. 2003, Leger et al. 2002, Temple et al. 2003, Han et al. 2007). However, neuroimaging evidence on the effects of langauge interventions for PPA is lacking. Our study has been carefully designed to determine whether a successful language intervention for people with PPA can induce measurable effects in brain activity. Methods: Three patients with fluent PPAwere recruited from Sunnybrook Health Sciences Centre in Toronto. Their demographic and language data are presented in Table 1. Intensive treatment for naming impairments was conducted via an errorless learning paradigm, whereby participants were presented with information about each item and were asked to repeat its name. The stimuli set consisted of 80 pictures of common nouns that participants were unable to name prior to the study. Twenty nouns were re-trained with a phonological approach (i.e., initial sound/letter, rhyming word, number of syllables), and 20 with a semantic approach (i.e., description of item’s appearance, location, and function). The remaining 40 words comprised the control set. Each patient received 10 hourly treatment sessions for each condition. Treatment effects were measured upon completion of each treatment condition on a confrontation naming test with no cueing of all treated and untreated items. Each participant was scanned before and after treatment while performing two linguistic tasks (1) Phonological task: matching of letter strings based on their sound, (e.g., beaw-beew1⁄4match, stom-spom1⁄4 no match), (2) Semantic task: wordmatching (swim-sail1⁄4match, throw-sit1⁄4 nomatch). Imaging utilized a 3T MRI system (TIM Trio, Siemens Medical Solutions Inc., vb15 software). Participants viewed visual stimuli displayed on the screen through an angled mirror in the Matrix coil. Responses were recorded using a Fiber-Optic Response Pad system (Current Designs Inc.). E-prime (Psychological Software Tools, Inc.) was used to deliver stimuli and record responses. Results: Imaging: Increased activation in the left anteriortemporal region was observed in all three participants post-treatment. Statistical analyses are under way to determine the magnitude of that change. In addition, some reorganization of activation in other brain regions was observed in all three participants. Figure 1 shows pre-and post-treatment images of one of the participants, CW, who showed moderate gains in treatment. Naming: All participants improved their naming accuracy from the baseline of 0 to 50-100% accuracy in both phonological and semantic conditions (p<.001, McNemar Change Test). Participants benefitted from the two approaches to different degrees. While MK benefitted mostly from semantic cues (p<.01, t-Test), CWand HD did not show the same preference and performed equally well with either cue type. Accuracy of the inscanner tasks: Varied accuracy on matching in-scanner tasks was observed. There was significant improvement on post-tx semantically matched pairs (p<.05, t-Test), marginal improvement on phonologically matched pairs (p1⁄4.06, t-Test), and no difference between post-tx semantic and post-tx phonological condition. Feedback from participants indicated that, despite equal length, the pairs of letter strings were more difficult to process than real words on the semantic task. The accuracy of the in-scanner task did not correlate with the behavioural outcomes of language therapy. Conclusions: This is the first study addressing post-treatment gains in PPA, a progressive language disorder, reflected in a detectible change in brain activity. Our study provides important novel information regarding neural underpinnings of the learning process in PPA. Inspection of participants’ fMRI images showed increased post-treatment activation in the left anterior temporal brain regions that are crucial for processing semantic information. This correlated with improved naming in each participant. As our therapy was impairment-based (as opposed to a compensatory approach), we can conclude that our language therapy was largely responsible for the increase in activation on post-treatment scans. Our findings have both clinical and theoretical implications. Firstly, we demonstrated that patients with mildto-moderate language difficulties due to PPA can benefit from naming therapy. This is important, given that currently patients do not receive impairment-based treatment for language disorders in neurodegenerative conditions. Second, the increased activation noted on post-treatment scans shows initial promise that a non-pharmaceutical intervention is capable of inducing positive changes in neural activation in a progressive disorder.