Abstract 2879: Acute Injury to Specific Neural Systems Predicts Outcome of Individual Behaviors

Abstract

INTRODUCTION: Improved methods are needed for early prediction of modality-specific outcomes. While global outcome measures are important after stroke, modality-specific endpoints such as arm motor function or language are also important. Different modalities recover to different extents, post-acute treatment often revolves around specific modalities (e.g., physiotherapy for weakness), and patients want to know what they will be able to do once recovery plateaus. Modality-specific endpoints are the primary outcome measure in many restorative stroke trials and indeed have been the basis for regulatory approval. The current study hypothesized that measuring injury to a specific neural system will improve prediction of the modality-specific behavior arising from that system, as compared to a global injury measure such as infarct volume. METHODS: For 74 patients in the GAIN studies who received an MRI <6 hr after stroke onset and survived to 3 months, predictors of global (NIHSS total) and modality-specific (arm motor function, NIHSS Q5; leg motor, Q6) outcome measures were examined using stepwise logistic regression modeling. Predictors in each model included respective baseline scores, infarct volume, % corticospinal tract injury, age, hypertension, diabetes, hypercholesterolemia, and AFib. RESULTS: Patients were age 65 +− 13 yrs, and had median NIHSS score =15 at baseline and =05 at 3mo. For the model predicting whether total NIHSS score at 3mo was 0, R ∧ 2 =0.27 (p=0.0006), with baseline total NIHSS score and infarct volume remaining as significant predictors. For the model predicting whether arm motor function was normal at 3mo (NIHSS Q5=0), R ∧ 2 was 0.38 (p<0.0001), with baseline NIHSS-Q5 subscore and % corticospinal tract injury being significant predictors. For the model predicting whether leg motor function was normal at 3 mo (NIHSS Q6=0), R ∧ 2 was 0.16 (p<0.002), with baseline NIHSS-Q6 subscore and % corticospinal tract injury being significant predictors. Voxel lesion symptom mapping in these 74 subjects supported above results and also suggested other brain areas whose injury might influence motor outcome. CONCLUSIONS: An acute measure of injury to a specific neural system is useful for predicting 3mo outcome of the behavior arising from that system. On the other hand, global outcomes are best predicted by global measures. These findings are of value because early prediction of the 3mo values for modality-specific measures can potentially aid prognosis, guide treatment decisions, provide mechanistic insights, and inform entry criteria in restorative stroke trials.