Abstract T MP44: Different Predictors of Treatment Gains in Lacunar and Non-lacunar Stroke
Introduction: Numerous predictors of treatment-induced behavioral gains after stroke have been identified. However, only a few types of predictors have generally been examined at a time. The current study examined multiple categories of established predictors in parallel among patients enrolled in a motor therapy study, hypothesizing that brain injury and function are of greatest predictive value. Furthermore, given that clinical relationships often differ according to stroke pathophysiology, the hypothesis was also tested that predictors of treatment gains vary according to stroke subtype.
METHODS: Patients with age>18 yr, stable arm paresis, and stroke 11-26 wks prior were enrolled in a study (NCT01244243) that provided 3 weeks of standardized robotic arm therapy. The primary outcome measure was change in Fugl-Meyer (FM) and Action Research Arm Test (ARAT) scores. A total of 6 categories of baseline assessments were evaluated as potential predictors of treatment gains:  demographics/history,  behavior,  genetics,  neurophysiology (TMS),  brain injury (3T MRI), and  brain function (3T fMRI). For each category, the strongest bivariate predictor of gains (when significant) was advanced into a multivariate model.
RESULTS: At baseline, the 29 patients had substantial motor deficits (FM=36±15), medium-sized infarcts (33±48cc; 8 lacunar, 21 non-lacunar), and were 4.3 months post-stroke. Patients improved significantly (p<0.0005) on the FM (3.7±3.5) and ARAT (4.1±6.3) scores. In non-lacunar stroke, extent of corticospinal tract injury and baseline behavior survived as predictors of treatment gains in the multivariate model (r=0.60, p=0.02). In lacunar stroke, the model found that extent of activation within ipsilesional primary motor cortex was the sole predictor of gains (r=0.79, p=0.02).
CONCLUSIONS: Measures of brain injury, brain function, and behavior are the strongest predictors of arm motor gains among patients in the early phase of chronic stroke. Different factors predict gains across stroke subtypes. Current results may be useful for patient selection in a clinical trial setting, where effective stratification can reduce variance and increase study power.
Author Disclosures: E. Burke: None. L. Dodakian: None. J. See: None. J.D. Riley: None. A. McKenzie: None. V. Le: None. S.C. Cramer: Consultant/Advisory Board; Modest; GlaxoSmithKline, MicroTransponder.
- © 2014 by American Heart Association, Inc.