Critical periods after stroke study: translating animal stroke recovery experiments into a clinical trial

And if we had ANY leaders in stroke they would be tackling these difficult translational problems. But no, we have jackshit leaders, WAITING FOR SOMEONE ELSE TO SOLVE THE PROBLEM
http://journal.frontiersin.org/article/10.3389/fnhum.2015.00231/full?

Alexander W. Dromerick^1,2*, Matthew A. Edwardson^1,2, Dorothy F. Edwards³, Margot L. Giannetti¹, Jessica Barth¹, Kathaleen P. Brady¹, Evan Chan¹, Ming T. Tan⁴, Irfan Tamboli⁵, Ruth Chia⁵, Michael Orquiza⁵, Robert M. Padilla⁵, Amrita K. Cheema⁶, Mark E. Mapstone⁷, Massimo S. Fiandaca^2,5, Howard J. Federoff^2,5 and Elissa L. Newport^1,2

• ¹Department of Rehabilitation Medicine, Center for Brain Plasticity and Recovery, Georgetown University and MedStar National Rehabilitation Hospital, Washington, DC, USA
• ²Department of Neurology, Georgetown University, Washington, DC, USA
• ³Department of Kinesiology and Occupational Therapy, University of Wisconsin, Madison, WI, USA
• ⁴Department of Biostatistics, Georgetown University, Washington, DC, USA
• ⁵Department of Neuroscience, Georgetown University, Washington, DC, USA
• ⁶Departments of Oncology and Biochemistry, Georgetown University, Washington, DC, USA
• ⁷Department of Neurology, University of Rochester, Rochester, NY, USA

Introduction: Seven hundred ninety-five thousand Americans will have a stroke this year, and half will have a chronic hemiparesis. Substantial animal literature suggests that the mammalian brain has much potential to recover from acute injury using mechanisms of neuroplasticity, and that these mechanisms can be accessed using training paradigms and neurotransmitter manipulation. However, most of these findings have not been tested or confirmed in the rehabilitation setting, in large part because of the challenges in translating a conceptually straightforward laboratory experiment into a meaningful and rigorous clinical trial in humans. Through presentation of methods for a Phase II trial, we discuss these issues and describe our approach.

Methods: In rodents there is compelling evidence for timing effects in rehabilitation; motor training delivered at certain times after stroke may be more effective than the same training delivered earlier or later, suggesting that there is a critical or sensitive period for strongest rehabilitation training effects. If analogous critical/sensitive periods can be identified after human stroke, then existing clinical resources can be better utilized to promote recovery. The Critical Periods after Stroke Study (CPASS) is a phase II randomized, controlled trial designed to explore whether such a sensitive period exists. We will randomize 64 persons to receive an additional 20 h of upper extremity therapy either immediately upon rehab admission, 2–3 months after stroke onset, 6 months after onset, or to an observation-only control group. The primary outcome measure will be the Action Research Arm Test (ARAT) at 1 year. Blood will be drawn at up to 3 time points for later biomarker studies.

Conclusion: CPASS is an example of the translation of rodent motor recovery experiments into the clinical setting; data obtained from this single site randomized controlled trial will be used to finalize the design of a Phase III trial.

Background

Using animal models of stroke, substantial scientific progress has been made in the understanding of the neural substrates of recovery after brain injury. Experimental studies of motor training after injury show that motor function can be improved significantly when a number of recovery and training variables are controlled. The experiment of Biernaskie et al. (2004) has been particularly intriguing given the finding of a sensitive period after experimental stroke in which rodents are most responsive to motor training in a specific time window soon after stroke. This finding has provoked much discussion in the stoke rehabilitation research community, since of course one wants to rehabilitate stroke patients at the time after stroke when therapies can be most effective. In this paper, we discuss the challenges faced by clinical trialists in translating a conceptually straightforward rodent experiment into a stroke rehabilitation clinical trial. We present our methods for the Critical Periods after Stroke Study (CPASS) as one example of the choices that can be made in testing whether promising findings in rodents have relevance in rehabilitation of patients with stroke.

The CPASS trial is designed to translate important findings from the rodent motor recovery literature into the human clinical trial setting. Adapting the critical elements of the rodent studies to the stroke rehabilitation setting requires a series of decisions and accommodations. In this paper, we review and discuss these considerations and how we have addressed them. Where possible we have retained essential elements of the rodent studies, including manipulation of intervention timing, randomization, standardized motor training paradigm based on a highly salient reward, and the use of motor performance measures. Data obtained from this randomized controlled trial will be used to formulate more effective treatments to better focus on the needs of individuals with stroke.

Approaching the Translation of Animal Experiments into Clinical Trials

Table 1 displays the many advantages of rodent experiments; these advantages allow exacting study of the biology of mammalian brain recovery, and the most unequivocal demonstration of the impact of putative motor training interventions. The rodents can be healthy young animals predictably available through breeding or purchase, eliminating confounds of differences in rearing, medical conditions and post-injury mortality. The ability to test a group of subjects simultaneously eliminates any drift in study or training procedures. Heterogeneity across animal subjects can be limited by the use of a single gender and a genetically homogenous strain. Brain lesions can be standardized and made in a brain that is otherwise pristine. Motor training protocols can be uniform and timed exactly. Food can be used as a highly motivating reward, and subjects are not lost to follow-up. Biological mechanisms can be studied using tissue and molecular techniques requiring sacrifice of the animals.

TABLE 1

Table 1. Issues in translation from rodent experiments to human clinical trials.

Designing a human stroke motor recovery trial tightly linked to the methods used in rodent motor recovery experiments involves a series of adaptations. These adaptations attempt to minimize the real-world limitations of clinical research and to maximize the clinical and scientific utility. The middle column of Table 1 displays some of the challenges faced by clinical trialists as they adapt these experiments to the clinical setting. A simple direct translation of rodent methods into humans can result in a trial that would be straightforward to design, but impractical to execute. For example, an investigator may want to insist that a single, specific lesion type be present for an individual to enroll in a clinical trial. This insistence might be scientifically justifiable, but impossible to execute because of the difficulty of finding sufficient numbers of individuals who suffered the needed infarct, meet other inclusion criteria, and are willing and able to participate in a trial. Similarly, challenges exist in enforcing exact timing of treatments, the content of treatments, obtaining motivated participation in training, and simply locating the individual to collect outcome measures. Approaching the biology of brain recovery in humans is also more challenging because of the infeasibility of recovering brain tissue; even lumbar punctures limit large scale participation in trials.

Clinical trial methods can mitigate the limitations of the stroke rehabilitation clinical setting; many examples are listed in the third column of Table 1. For example, the problem of identifying large numbers of participants can be limited through the use of adaptive trial designs, ensuring that participants will be randomized only to study arms that are promising. Less stringent inclusion/exclusion criteria can increase participant accrual, and the accompanying increase in heterogeneity across subjects can be managed using adaptive randomization strategies to minimize differences between study groups. Treatments that begin on a single preselected day in animals are not realistic in the fluid and unpredictable clinical setting, but can be replaced by treatment initiation intervals, allowing flexibility to the participant and research team. In other cases, investigators must simply make choices based on knowledge of the population, clinical setting, or treatment techniques. The number of choices can be quite large, and often the importance of individual choices is visible only in retrospect at the end of an expensive multiyear effort to answer what to all initial appearances is a straightforward question.

Laboratory-based Work in Critical Periods after Stroke

In current practice, as it becomes possible for the patient to participate after stroke, rehabilitation begins. This rehabilitation is initially superimposed on a background of resolving brain edema, inflammation and apoptosis, which are not thought to be materially influenced by experiences such as motor training (Carmichael, 2006; Cramer, 2008).

In contrast, rehabilitation itself is a mixture of compensation and learning. New learning, particularly that obtained via activity-based therapies (ABT's) (Dromerick et al., 2006), is thought to be accomplished by experience driven neuroplasticity (Kleim and Jones, 2008; Carter et al., 2010). The patient relearns prior methods of accomplishing everyday tasks and when necessary, learns new ways to accomplish goals through a combination of newly acquired compensatory strategies (Nakayama et al., 1994) and restoration of motor, sensory, and cognitive function in uninjured tissues (Lum et al., 2004, 2009; Levin et al., 2009). Since these processes often do not return the patient back to pre-stroke levels of function, understanding and exploiting animal findings of critical or sensitive periods in rehabilitation is an important approach to improving treatment.

These putative periods of greatest responsiveness after stroke have been hypothesized to be analogous to the “critical periods” in normal development (Murphy and Corbett, 2009). In the developing brain, critical periods are defined as times of greatest sensitivity to exogenous influences or experiences. Critical periods for the effect of experience on the formation of neural circuits and on the behaviors they control have been demonstrated, for example, in the establishment of ocular dominance columns and stereopsis in the visual system (Hubel and Wiesel, 1970), in the formation of attachment and species identification in a variety of avian species (Hess, 1973), and in vocal learning in songbirds (Marler, 1970) and in humans (Johnson and Newport, 1989; Newport, 1990). The molecular mechanisms underlying the opening and closing of developmental critical periods are beginning to be well understood (Hensch, 2005), and there are now even examples of “reopening” early critical periods during adulthood (Bavelier et al., 2010; Zhou et al., 2011).

The work of Biernaskie et al. (2004) suggests that certain periods after stroke may constitute a period of enhanced plasticity, analogous to a critical period during which the recovering brain is most sensitive to exogenous stimuli and experience. Thus, there may be an optimal time when stroke patients might show the largest improvement from therapy; and, should stroke patients not receive this optimally timed therapy, it is possible that the opportunity for optimal recovery could be irrevocably lost. Given the discontinuities in US health care, it is common for patients' therapy to be delayed for personal, medical or insurance reasons (Ostwald et al., 2009); even inpatient rehabilitation admission does not guarantee substantial amounts of motor training (Lang et al., 2009). Carefully executed studies demonstrating the optimal timing of therapies will help clinicians and policymakers ensure delivery of effective rehabilitation.

Most of the evidence regarding the timing effects of post-stroke motor training focuses on the behavioral, cellular, and molecular mechanisms of neuroplasticity. More recently, animal models demonstrate that genes involved in normal development (and that are quiescent in adulthood) are expressed at high levels in the first weeks after stroke and then decline, with distinct temporal patterns of gene expression after injury (Carmichael, 2003, 2006). This pattern of gene expression is consistent with the notion of an injury-induced recapitulation of development-like processes which occur during a period of enhanced plasticity. Most of these findings focus on the first weeks after stroke; our study design has three relevant time points (early/acute, subacute, and chronic), in order to best assess and locate such an effect, if indeed it occurs in human patients.

There are two major findings regarding treatment timing in animal models of stroke. First is the work of Schallert (Kozlowski et al., 1996; Humm et al., 1998) and others (Bland et al., 2000) showing that very early and intensive training can reduce recovery after experimental stroke and enlarge lesions. This may have been confirmed in humans in our own work (Dromerick et al., 2009), when we found that very intense motor training early after stroke led to worse outcomes. Second, and more optimistic is the work of Biernaskie et al. (2004) where the question of timing effects was directly addressed. They randomized lesioned animals to receive focused motor training at 5, 14, or 30 days after lesioning. They found that the best response to training started at 5 days after lesioning; an intermediate response was present when training was initiated at 14 days; and therapy beginning at 30 days resulted in the same motor outcome as controls who were not trained at all. This powerful pattern of results suggests that critical periods in stroke recovery do exist in adult mammals (Murphy and Corbett, 2009).

Human and Clinical Data regarding Timing Effects in Rehabilitation Treatment

Whether and how the results of Biernaskie et al. translate to human stroke patients is unknown. Few prospective human studies directly address optimal timing of rehabilitation. Natural history studies show that recovery after stroke in humans is fastest in the first weeks (Wade and Hewer, 1987; Jorgensen et al., 1995a,b); this period coincides with both the onset of rehabilitation treatment and the time that homeostasis is re-established, as described above. Clinicians have written for decades regarding the features of motor recovery that seem to resemble patterns of normal motor development (Cramer and Chopp, 2000; Pollock et al., 2007; Kollen et al., 2009). Retrospective data from clinical populations suggest that early initiation of rehabilitation is associated with better outcome (Wylie, 1970; Feigenson et al., 1977; Kotila et al., 1984; Rossi et al., 1997). However, these studies are confounded because patients who present late to rehabilitation are generally sicker and more severely affected, and thus less likely to improve regardless of timing of care (Ween et al., 1996). Some, but not all (Gagnon et al., 2006) newer studies using case control methods (Paolucci et al., 2000) or large multicenter cohorts (Maulden et al., 2005) have also found better responses early.

Secondary analyses of existing clinical trials are mixed. The EXCITE trial (Wolf et al., 2006) evaluated whether constraint therapy was superior to an uncharacterized “usual and customary care” (UCC) control in improving UE motor impairment; secondary analyses suggested that the participants treated earlier had a better motor outcome than those treated later (Wolf et al., 2010). The LEAPS trial (Duncan et al., 2011) of body-weight supported treadmill training for gait did not confirm a timing effect. LEAPS found that there were persistent treatment responses at both time points tested (2 and 6 months), but there were no significant outcome differences between the earlier and later groups. VECTORS, a single center Phase II trial (Dromerick et al., 2009) of constraint therapy early after stroke addressed dosing and therapy content rather than timing, but the results at this earlier time period suggested an inverse dose response relationship (at high doses, more therapy led to less motor recovery). A more recent study testing additional rehabilitation therapy early after stroke did not confirm this inverse dose phenomenon and suggested greater ipsilesional cortical activation on functional MRI in those randomized to extra therapy (Hubbard et al., 2014). A recent trial in ICH patients suggested a possible mortality benefit with early therapy (Liu et al., 2014). Preliminary data from AVERT (Bernhardt et al., 2008), an early mobilization RCT, are promising but enrollment is still ongoing.

Study Rationales and Hypotheses

The overall goal for the CPASS trial is to identify a critical period after stroke in which patients are particularly responsive to motor training interventions. We hope simply to elicit a signal that a critical period exists; optimization of dosing or treatment strategy would come in subsequent studies.

Our approach is to use a bolus of standardized motor therapy to elicit a motor improvement during a specific time period indicative of a critical period. Our hypothesis for the CPASS Phase II trial is that, compared to individuals randomized to the control condition or to the subacute (2–3 months after onset) or chronic (6–9 months after onset) time points, persons randomized to early intensive motor training will show greater UE motor improvement measured at 1 year. In addition, we will use the opportunity presented to collect peripheral blood to perform a proof of principle study exploring molecular signals associated with response to treatment and overall motor recovery. See Figure 1 for a diagram of study design.

In order to adapt the rodent experimental design to the clinical delivery patterns in the United States, we made two major decisions. First was the choice of time periods in which study-related treatment would be delivered. Precisely how post-stroke days compare between humans and rodents is unknown, and we attempted to balance fidelity to the Biernaskie et al. design with the pragmatics of accommodating existing treatment venues. These venues are not under the control of the research team. Choosing the exact time points and a single day window to initiate therapies such as was used in the rodent study meant that participants would need to be consented within 72 h of stroke onset so that baseline measures could be collected and the participant randomized with the possibility of treatment beginning exactly on post-stroke Day 5. Though conceptually not impossible, this choice would lead to several complexities including unavailability of patients to undergo study related therapy during a time when diagnostic testing must take first priority, medical complications and fatigue preventing therapy participation, uncertainty about the trajectory of motor recovery, and uncertainty as to whether and where the patient might be referred for inpatient rehabilitation. We chose instead three more flexible windows of time for study related treatments: early (<30 days, corresponding to the inpatient rehabilitation period), subacute (60–90 days, corresponding to typical outpatient therapy delivery), and chronic (6 months, by which time most US stroke patients will have been discharged from therapy). These times were chosen as analogous to the 5, 14, and 30 day times used in the Biernaskie et al. study. By using those existing clinical treatment venues, any improvements in efficacy that results from this line of work can improve the effectiveness of those venues without requiring a major change in how care is delivered. Thus, the translation to actual treatment would not be hindered by the need for policy and reimbursement changes.

The second major decision was that of how much study related treatment was necessary to observe a detectable effect of a critical period. Dose-response data for motor training are particularly lacking in the first few weeks after stroke onset. We chose 20 h of additional motor therapy because our previous work has shown that a difference of 10 h of treatment is sufficient to alter motor outcomes; this amount of additional therapy should thus provide an adequate signal indicating a critical period, if there is one (Dromerick et al., 2009). Moreover, should we find a large difference in outcomes in one group, it seems feasible to deliver 20 h more training to stroke patients in the current healthcare environment. Several studies document persistent motor improvements post-stroke with treatments of similar or even less intensity (Sivenius et al., 1985; Sunderland et al., 1992; Whitall et al., 2000; Page et al., 2001; Michaelsen and Levin, 2004; Michaelsen et al., 2006; Woldag et al., 2010; Han et al., 2013).

More at link.