Many restorative therapies that promote brain repair are under development. Stroke is very heterogeneous, highlighting the need to identify target populations and to understand inter-subject differences in treatment response. Several neuroimaging measures have shown promise as biomarkers and predictors, including measures of structure and function, in gray matter and white matter. Choice of biomarker and predictor can vary with the content of therapy and with the population under study, for example, contralesional hemisphere measures may be of particular importance in patients with more severe injury. Studies of training effects in healthy subjects provide insights useful to brain repair. Limitations of published studies include a focus on chronic stroke, however the brain is most galvanized to respond to restorative therapies in the early days post-stroke. Multimodal approaches might be the most robust approach for stratifying patients and so for optimizing prescription of restorative therapies after stroke.

Many different measures have been found to be related to behavioral outcome after stroke. Preclinical studies emphasize the importance of brain injury and neural function. However, the measures most important to human outcomes remain uncertain, in part because studies often examine one measure at a time or enroll only mildly impaired patients. The current study addressed this by performing multimodal evaluation in a heterogeneous population. Patients (n = 36) with stable arm paresis 3-6 months post-stroke were assessed across 6 categories of measures related to stroke outcome: demographics/medical history, cognitive/mood status, genetics, neurophysiology, brain injury, and cortical function. Multivariate modeling identified measures independently related to an impairment-based outcome (arm Fugl-Meyer motor score). Analyses were repeated (1) identifying measures related to disability (modified Rankin Scale score), describing independence in daily functions and (2) using only patients with mild deficits. Across patients, greater impairment was related to measures of injury (reduced corticospinal tract integrity) and neurophysiology (absence of motor evoked potential). In contrast, (1) greater disability was related to greater injury and poorer cognitive status (MMSE score) and (2) among patients with mild deficits, greater impairment was related to cortical function (greater contralesional motor/premotor cortex activation). Impairment after stroke is most related to injury and neurophysiology, consistent with preclinical studies. These relationships vary according to the patient subgroup or the behavioral endpoint studied. One potential implication of these results is that choice of biomarker or stratifying variable in a clinical stroke study might vary according to patient characteristics.

No abstract

Background and Purpose Behavioral measures are often used to distinguish subgroups of stroke patients, e.g., to predict treatment gains, stratify clinical trial enrollees, or select rehabilitation therapy. In studies of the upper extremity, measures of brain function using fMRI have also been found useful, but this approach has not been examined for the lower extremity. The current study hypothesized that an fMRI-based measure of cortical function would significantly improve prediction of treatment-induced lower extremity behavioral gains. Biomarkers of treatment gains were also explored. Methods Patients with hemiparesis 1-12 months post-stroke were enrolled in a double-blind, placebo-controlled, randomized clinical trial of ropinirole+physical therapy vs. placebo+physical therapy, results of which have previously been reported (NCT00221390). Primary endpoint was change in gait velocity. Enrollees underwent baseline multimodal assessment that included 19 measures spanning five assessment categories (medical history, impairment, disability, brain injury, and brain function), and also underwent reassessment three weeks after end of therapy. Results In bivariate analysis, eight baseline measures belonging to four categories (medical history, impairment, disability, and brain function) significantly predicted change in gait velocity. Prediction was strongest, however, using a multivariate model containing two measures (leg Fugl-Meyer score and fMRI activation volume within ipsilesional foot sensorimotor cortex). Increased activation volume within bilateral foot primary sensorimotor cortex correlated positively with treatment-induced leg motor gains. Conclusions A multimodal model incorporating behavioral and fMRI measures best predicted treatment-induced changes in gait velocity in a clinical trial setting. Results also suggest potential utility of fMRI measures as biomarkers of treatment gains.

Vector auto-regressive (VAR) models typically form the basis for constructing directed graphical models for investigating connectivity in a brain network with brain regions of interest (ROIs) as nodes. There are limitations in the standard VAR models. The number of parameters in the VAR model increases quadratically with the number of ROIs and linearly with the order of the model and thus due to the large number of parameters, the model could pose serious estimation problems. Moreover, when applied to imaging data, the standard VAR model does not account for variability in the connectivity structure across all subjects. In this paper, we develop a novel generalization of the VAR model that overcomes these limitations. To deal with the high dimensionality of the parameter space, we propose a Bayesian hierarchical framework for the VAR model that will account for both temporal correlation within a subject and between subject variation. Our approach uses prior distributions that give rise to estimates that correspond to penalized least squares criterion with the elastic net penalty. We apply the proposed model to investigate differences in effective connectivity during a hand grasp experiment between healthy controls and patients with residual motor deficit following a stroke.