The first few months of post-stroke recovery is a critical time period where spontaneous changes in language functions and underlying neural processes are observed. Imaging brain changes during this time may help clinicians identify the recovery of neural processes, but must be done in a way that is not confounded by stroke- induced cerebrovascular changes. This is important, as evidence suggests that cerebrovascular insult will elicit a cascade of changes that leads to vascular remodeling in the first few months of post-stroke recovery. This CDA2 proposal addresses this gap through development of a set of integrated multi-modal neuroimaging methodologies to dissociate neural and vascular changes during recovery of language functions from early sub- acute (2-6 weeks post-stroke) to late sub-acute phase (12-16 weeks post-stroke) in patients with aphasia. In the first aim, we will determine if regressing out the vascular signals (CVR and CBF) from task Blood Oxygen Level Dependent (BOLD) functional MRI (fMRI) activity at each time point (early and late phase) will improve the relationship between change in task-BOLD activity and change in lexical decision behavior. Our approach will be to track task-BOLD fMRI activity from early to late sub-acute phase while the patients participate in an auditory word recognition task (lexical decision) in both phases. We will apply our sensitization scheme for regressing out vascular signals. The change in BOLD amplitude (from sensitized and unsensitized/standard task BOLD fMRI activity) will then be related to changes in the lexical decision behavior. We expect to see that the neuro-sensitized task BOLD fMRI activity will have a stronger correlation with lexical decision measures than the unsensitized/standard BOLD fMRI signal. In the second aim, we will determine if removing the vascular signals from resting state BOLD (rs-BOLD) acquired from residual language network at each time point (early and late phase) will improve the relationship between change in rs-BOLD network measures and change in language (domain specific and domain general) measures. Our approach will be to acquire rs-BOLD fMRI scans during the early and late sub-acute phase in patients with aphasia. We will then carry out a whole brain voxel-wise network analysis (i.e. modularity) choosing apriori seed ROI from the residual language- related brain areas. To sensitize the rs-BOLD fMRI signals to neural connections, we will carry out the same sensitization scheme (as described in Aim 1) at a voxel level. We will then identify changes in language network measures (from sensitized and unsensitized/standard rs-BOLD fMRI) and correlate them with language behavior (domain specific and domain general). After completing Aim 2, it is our expectation that the proposed sensitization scheme enhances the sensitivity of rs-BOLD network measures to brain network reorganization in language specific and language-nonspecific cognitive domains. In the third aim, we will determine if vascular measures (CBF and CVR) from early phase can predict change in language behavior (Western Aphasia Battery measure of comprehension and Philadelphia Naming Test. Our approach will be to recruit patients with aphasia in the early sub-acute phase and measure their vascular and behavioral recovery at both early and late time points. Specifically, we will acquire voxel-wise CBF and CVR measures at the early time point, and using multiple regression approach, we will model the vascular measures to predict change in clinically-relevant language behavior. After completing Aim 3, it is our expectation that we will have a clinically translatable data driven (vascular physiology-based) prediction model that can be used to identify brain regions that must change to support improvements in language behavior. The expected outcome of this CDA2 is an integration of multimodal neuroimaging tools to more accurately predict longitudinal recovery of language functions in sub- acute patients with aphasia. The long-term goal is to determine how neuroimaging tools can best be used to provide accurate individualized language recovery trajectories and predict treatment outcome.
Every year about 15,000 Veterans are hospitalized for stroke (Cowper 2004), and one-third of those patients develop stroke-related language disorders (aphasia) (Brady 2012). The first few months of post-stroke recovery is a critical time period where spontaneous changes in language functions and underlying neural processes are observed. Imaging brain changes during this time may help clinicians identify the recovery of neural processes, but must be done in a way that is not confounded by stroke-induced cerebrovascular changes. This CDA2 proposal addresses this gap through development of a set of integrated multi-modal neuroimaging methodologies to accurately identify neural systems associated with language recovery from early sub-acute (2- 6 weeks post-stroke) to late sub-acute phase (12-16 weeks post-stroke) of aphasia recovery.
