Stroke Recovery—Moving in an EXCITE-ing Direction

Stroke produces well-known motor and cognitive impairment and high levels of disability. Currently used stroke rehabilitation treatments generally are based on empirical approaches rather than scientifically validated therapies.^{1 - 3} A wide variety of poorly standardized and unevaluated traditional treatments aim to recondition spinal or muscular targets to prevent⁴ or accommodate⁵ impaired motor control. The results of the EXCITE (Extremity Constraint Induced Therapy Evaluation) trial reported by Wolf and colleagues⁶ in this issue of JAMA contribute to a new understanding of how to approach disability in stroke survivors. A careful look at this trial and its background opens new directions for clinical care, stroke recovery, rehabilitation care delivery, and research. Such new directions are of utmost importance for stroke survivors who hope to reacquire abilities lost after ischemic brain injury.

The EXCITE trial is the first multisite randomized study to demonstrate the efficacy of a rehabilitative intervention. It therefore moves neurorehabilitative care into the area of evidence-based medicine. The intervention, constraint-induced therapy for 2 weeks, including up to 6 hours of physical therapy assistance per workday and wearing a restraining mitt on the less affected arm for 90% of waking hours for 14 days, improved functional use of the affected arm and reduced disability. These benefits were maintained for at least 12 months. The control group followed customary care provided by their treating physician and also improved arm function but significantly less so than the constraint-induced therapy group.

Constraint-induced therapy was developed based on the theory of learned nonuse of the paretic arm in the deafferented monkey model.⁷ Deafferentiation leads to adaptations in cortical and subcortical motor circuits, a form of neuroplasticity that has negative consequences for the deafferented arm. By restraining the intact arm (forced use), the therapy aims at reverting learned nonuse. Constraint-induced therapy combines forced use with repetitive training and shaping, techniques that are known to induce “positive” neuroplastic adaptations that lead to improved performance in the trained tasks. While a number of interventions following neuroplasticity approaches have shown some efficacy, constraint-induced therapy applied according to the study protocol is the first neuroplasticity therapy for which controlled multicenter evidence is now available. However, the single vs combined effects of these components, forced use or repetitive training and shaping, remain unresolved.

As the first large controlled trial in neurorehabilitation, the study by Wolf et al leaves important questions for future trials. According to the EXCITE protocol, constraint-induced therapy is only applicable to patients with limited disability (preserved ≥10° active wrist extension, ≥10° thumb abduction/extension, and ≥10° of extension in ≥2 additional digits). Different interventions may be necessary for more severely disabled patients. In low-functioning individuals, the safety of constraint-induced therapy needs to be considered; eg, the restraining mitt promoting falls in patients with gait impairment. Given the sample size of the trial, low-frequency events such as these may not be well-estimated.

Although adherence to the protocol was rigidly controlled in this trial, compliance may pose a problem in routine use because the therapy will increase patients' burdens in activities of daily living. High dropout rates (approximately 24% in the report by Wolf et al) are a common problem in rehabilitation trials because of population age and comorbidities. It remains to be investigated whether in clinical practice a similar rate of patients are lost to follow-up and whether this attrition has an unfavorable effect on outcome or secondary cardiovascular prevention. Other rehabilitative interventions, performed before or during the study period, may have played a facilitative role that is undefined by the current study design. Furthermore, imaging data was not gathered; therefore, stroke location as a potential predictor for efficacy could not be addressed.^{8 - 9}

The trial by Wolf et al confirmed the persistence of the therapeutic effect over 12 months, but longer follow-up data are needed to truly estimate patient benefits and to assess the durability of the response. The time point of administration of constraint-induced therapy was 3 to 9 months after the index stroke. Given that constraint-induced therapy shows some efficacy in patients with chronic disability,⁷ the optimal time point for the intervention needs to be defined. Similarly, the minimum intensity for effective constraint-induced therapy is unknown and will be of interest when treatment cost becomes an issue in clinical routine.

Despite the number of unanswered questions, constraint-induced therapy, if it were a drug, would probably receive approval based on the EXCITE results. Integration of constraint-induced therapy into routine patient care has begun to occur outside of the United States.^{10 - 12} Even though it remains unknown whether constraint-induced therapy fully exploits the potential for functional recovery, the improvements patients experience are substantial and merit strong consideration of constraint-induced therapy by those responsible for practice guidelines and insurance benefits.

The findings from this trial also raise questions that merit further research. For instance, it is becoming certain that frequent repetition of simple movements (representing part or all of a specific task) has been underperformed in traditional clinical interventions, resulting in diminished physical function. Although the EXCITE trial suggests that patients with stroke have been undertreated, the amount of disability that can be reversed is yet unknown. Future neurorehabilitative trials must define a recovery target and explore whether disability can be reduced further. Modifications to constraint-induced therapy or supportive interventions such as drug therapy¹³ may increase its efficacy. Therefore, a critical task for investigation and clinical development is to maximize the benefit of these interventions. In this context, the minimum effective intensity must be established (ie, duration of training, frequency of repetition, and the number of distinct movements that must be trained). This seems particularly important after the EXCITE trial, as current clinical evaluations suggest that routine therapy does not approach the intensity of treatment applied in this study.¹⁴

Another major research direction involves understanding the biological mechanism of the observed motor improvements. Does constraint-induced therapy alter limb mechanical function, or does its repetitive nature of training access neural circuit plasticity by exceeding a repetitive threshold? Early mechanistic work suggests that constraint-induced therapy modifies brain circuitry, particularly motor and premotor cortices in the affected hemisphere, to produce its benefit.^{15 - 16} Repetitive interventions were shown to recruit sensory-motor cortex in the undamaged hemisphere and cerebellum as interconnected circuits.¹⁷ If these circuits are the biological substrate for the recovery effects of rehabilitation, clinical and mechanistic tests must be designed to understand how to maximally stimulate these areas and to determine if lesions involving these areas limit the effectiveness of therapies such as constraint-induced therapy. The EXCITE trial has moved the clinical and basic science world of recovery investigation past the age in which untested neurodevelopmental principles¹⁸ drove intervention design into an era in which improvements can be built on an efficacious treatment.

Evidence-based clinical guidelines evaluating recovery therapies will need to consider the EXCITE trial results. Given that constraint-induced therapy apparently produces functionally meaningful gains in arm function, guidelines reflecting the position that constraint-induced treatment is appropriate for stroke patients with moderate impairment but preserved motor control in the hand are needed. Clinicians should recognize that a biological rationale for a much more vigorous therapeutic approach is rapidly gaining momentum. Specifically, task-directed treatments that use repetition are likely to produce modest but functionally significant gains after the 6-month poststroke “plateau” or “spontaneous recovery” periods. For patients with stroke who also have concurrent medical illness for which basic medical stability requires longer than 3 months, specific rehabilitation treatments may have value. Moreover, in contrast with acute stroke, where “time is brain,” physicians and other health care professionals must accept that time after stroke does not appear to be a limiting factor for recovery. In addition, for motivated patients, task repetition should be encouraged in the home to enhance recovery and for long-term maintenance of function. In stroke recovery, “training is new brain.”

The EXCITE trial clearly suggests that more recovery after stroke is possible than neuroscientists currently teach in professional schools or training programs and than clinicians have been telling patients to expect. Recovery of lost or impaired motor functions has been demonstrated in animal models and is biologically feasible for humans, and now has been confirmed by data from a well-designed randomized clinical trial. Knowledgeable physicians and scientists must reject the nihilism that bases the projection of best recoveries on observations of therapies without defined benefit. If training is new brain, many more therapeutic and mechanistic clinical trials of stroke recovery must be conducted. Despite decades of research and discovery, there is still no clear idea as to the maximal amount of benefit that can be achieved with interventions that harness the learning powers of the human brain. A little more excitement in the lives of stroke survivors can only be good.