Baseline Predictors of Response to Repetitive Task Practice in Chronic Stroke

Abstract

Background: Repetitive task practice reduces mean upper extremity motor impairment in populations of patients with chronic stroke, but individual response is highly variable. A method to predict meaningful reduction in impairment in response to training based on biomarkers and other data collected prior to an intervention is needed to establish realistic rehabilitation goals and to effectively allocate resources.

Objectives: To identify prognostic factors and better understand the biological substrate for reductions in arm impairment in response to repetitive task practice among patients with chronic (≥6 months) post-stroke hemiparesis.

Methods: The intervention is a form of repetitive task practice using a combination of robot-assisted therapy and functional arm use in real-world tasks. Baseline measures include the Fugl-Meyer Assessment, Wolf Motor Function Test, Action Research Arm Test, Stroke Impact Scale, questionnaires on pain and expectancy, MRI, transcranial magnetic stimulation, kinematics, accelerometry, and genomic testing.

Results: Mean increase in FM-UE was 4.6 ± 1.0 SE, median 2.5. Approximately one-third of participants had a clinically meaningful response to the intervention, defined as an increase in FM ≥ 5. The selected logistic regression model had a receiver operating curve with AUC = .988 (Std Error = .011, 95% Wald confidence limits: .967-1) showed little evidence of overfitting. Six variables that predicted response represented impairment, functional, and genomic measures.

Conclusion: A simple weighted sum of 6 baseline factors can accurately predict clinically meaningful impairment reduction after outpatient intensive practice intervention in chronic stroke. Reduction of impairment may be a critical first step to functional improvement. Further validation and generalization of this model will increase its utility in clinical decision-making.

Trial registration: ClinicalTrials.gov NCT02658630.

Keywords: MRI; TMS; prognosis; rehabilitation; robot; stroke.

Conflict of interest statement

Conflict of Interest Statement

The authors declare that there is no conflict of interest. Dr. Wittenberg was on a Clinical Advisory Board for the Battelle Memorial Foundation, and currently for Myomo, Inc. Both companies are involved in rehabilitation technology with general relevance to this study but none of that technology was used.

Figures

Figure 1.

Study flow (consolidated standards of reporting trials diagram).

Figure 2.

Fugl-Meyer (FM) change scores as a function of total score at Baseline. Points above the horizontal line indicate Responders (n=16) to therapy (change in FM score >= 5 points); points below the line are Non-Responders (n=26).

Figure 3.

ROC curves for the final predictive model (Model) and cross-validation (ROC1).

Figure 4.

Extent of lesions across participants. Responders (n=14) are displayed on the top, Non-Responders (n=23) on the bottom; all images were flipped to show lesions in the Left hemisphere. The heat map demonstrates lesion-affected-voxels shared in the range from 1 participant to the maximum number of participants that shared that voxel with the range being 1–7 in Responders and 1–16 in Non-Responders. Notably, both groups had significant overlap of lesions within the deep motor white matter of the affected hemisphere.