|UFDC Home||myUFDC Home | Help|
This item has the following downloads:
1 MEASUREMENT OF ACTIVITY SPECI FIC BEHAVIORAL RECOVERY IN CHRONIC STROKE By MARK GOODMAN BOWDEN A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2009
2 2009 Mark Goodman Bowden
3 To the women in my family who have unconditionally supported me throughout this endeavor. To my wife, Fletcher Timmerman Bowden, who has endured countless hours away from home, family, and responsibilities but has never wavered in her optimism and encouragement. To my daughters, Margaret Goodman Bo wden and Anna Delaney Bowden for constantly grounding me, teaching me daily what is important in my life, and for forcing happiness and love to overwhelm any other emotion that I may be feeling. Las tly, to my mother, Nancy Goodman Bowden, who started me in my life as a reha bilitationist by daily instructing me through her actions and deeds how to care for others with l ove, compassion and selflessness.
4 ACKNOWLEDGMENTS I would first like to acknowledge the suppor t of my committee members for shepherding me through this process and for making it such a meaningful and productive journey. My chair, Steven Kautz, accepted me as a student after I sp ent eight years as a clinical therapist and had barely heard of a ground reaction force; he ha s molded me into someone who can converse fluently with therapists and e ngineers alike regarding the mechan isms of hemiparetic gait. Dr. Kautz has instilled in me a completely nove l level of understanding and inquiry into the determinants of human movement. My co-chair Andrea Behrman, has consistently modeled in her own life and work the passion re quired to conduct clinic al research that can directly impact the lives of those with disabilities. Dr. Be hrmans unique contributions have shaped my transition from clinician to scientist and have in spired a lifetime of work dedicated to affecting the lives of those we serve. Jay Rosenbek has been an instructor, co llaborator, and mentor throughout my tenure in Florida as a student and employee. He has instilled in me a deep appreciation of the understanding of different elements of th e human condition and how these different elements interact in i ndividuals to determine their over all health status. Lastly, Jeff Kleim has consistently challenged my concepts of recovery versus compensation at both body structure and behavioral levels, and these conve rsations have led directly to the overall framework of this dissertation a nd future research directions. Although not a formal member of my committ ee, I would also like to acknowledge the contributions of Carolynn Patten. Dr. Patten was in strumental in setting up the instrumentation for the experiment in Chapter 4 and provided some of the equipment to make that experiment possible. Perhaps more importantly, I appreciate the frequent spontaneous discussions regarding spinal level reflexes and contro l of movement, and I am grateful to her contributions to my understanding of the science.
5 My academic career and professional growth would not have been possible without the consistent and ongoing support of th ose at the Brain Rehabilitation Research Center (BRRC). I can think of no better place to have grown as a re searcher, but more importa nt has been the care and support fostered by the family atmosphere w ithin. While there have been too many people involved in my development to name individually, I would be remiss if I did not name several influential people. First, Leslie Gonzalez Rothi ha s been superlative as a mentor of the research process and as a center director, bu t it is her care for me as an individual that allowed me to juggle work and school responsibil ities while offering me the wealth of resources within the BRRC. My family and I will forever be grateful for the opportunity I had to pursue an advanced degree in the manner in which I was allowed. Sandy Davis, an extraordinary woman and therapist, was a source of constant support a nd friendship, and I count her among the truest friends that I have ever had the privilege of knowing. Ryan Knight and Cameron Nott have been instrumental in facil itating my understanding of the engineering underpinnings of our work, and without their assistance, many of our best ideas would have died a quiet death prior to implementation. Lastly, Chris Gregory has been both a friend and a mentor, spirited in encouragement and grounded in reality, without wh om it is doubtful that I might have emerged from this process with any degree of sanity. Many people have contributed to the data collect ion of the material presented here: Lise Worthen and Maria Kim from the Palo Alto, CA Veterans Affairs Medical Center; Chitra K. Balasubramanian, Erin Carr, and Helen Emery fr om the Gainesville, FL Veterans Affairs Medical Center; and numerous stude nts and volunteers who have assist ed us with our research at the BRRC.
6 I would also like to acknowledge Paul Zehr a nd his staff at the Neur oRehab Laboratory at the University of Victoria, British Columbia fo r their mentoring of studies associated with cutaneous relfexes. Particularly helpful was Ma rc Klimstra, a brilliant doctoral candidate who fielded many, many questions without ever once making me feel like the neuroscience novice that I was. Through collaboration with Dr. Zehr and Marc, I have been exposed to windows into the nervous system of which I had previously only dreamed, and I look forward to future collaborations as we continue to tackle comple x questions regarding ne urological function that drive us all. I would be remiss if I did not al so acknowledge Nicole Tester, my partner in the study of cutaneous reflexes, and it is through my many hours in development and discussion with Nicole that Chapter 4 of th is dissertation was possible. I have been most fortunate to work with mentors who are very well funded, and I have been supported in part by the following grants : VA Rehabilitation R&D Center of Excellence grant F2182C (Brain Rehabil itation Research Center); VA Re habilitation R&D Grants B3983R and B2748R; NIH R01 HD46820 and R01 HD37996; NIH K01 HD01348; and the Craig H. Neilsen Foundation.
7 TABLE OF CONTENTS page ACKNOWLEDGMENTS...............................................................................................................4 LIST OF TABLES................................................................................................................. ..........9 LIST OF FIGURES................................................................................................................ .......10 ABSTRACT....................................................................................................................... ............11 CHAPTER 1 LITERATURE REVIEW.......................................................................................................13 Introduction................................................................................................................... ..........13 Significance of Problem........................................................................................................ .15 Need for Quantification of Re habilitation Research Findings...............................................16 International Classification of Func tioning, Disability, and Health (ICF).............................18 Biomechanical Examination of Hemiparetic Walking...........................................................22 Electromyography...........................................................................................................22 Moments........................................................................................................................ ..24 Power.......................................................................................................................... .....25 Work........................................................................................................................... .....25 Examination of Post-stroke Motor Control............................................................................27 Examination of Spinal Cutaneous Reflexes...........................................................................31 Task Dependency of Cutaneous Reflexes.......................................................................32 Phase Dependency of Cutaneous Reflexes......................................................................32 Functional Role of Cutaneous Reflexes during Walking................................................33 Role of Reflexes in Study of Activity Dependent Neuroplasticity.................................34 Conclusion..................................................................................................................... .........35 2 ANTERIOR-POSTERIOR GROUND REACTION FORCES AS A MEASURE OF PARETIC LEG CONTRIBUTION IN HEMIPARETIC WALKING...................................39 Introduction................................................................................................................... ..........39 Methods........................................................................................................................ ..........41 Participants................................................................................................................... ...41 Variables...................................................................................................................... ....42 Statistical Analysis..........................................................................................................4 4 Results........................................................................................................................ .............44 Gait Characteristics.........................................................................................................44 Pedaling Characteristics..................................................................................................45 Discussion..................................................................................................................... ..........46
8 3 EVALUATION OF ABNORMAL SYNERGY PATTERNS POST-STROKE: RELATIONSHIP OF CLINICAL E XAMINATION TO HEMIPARETIC LOCOMOTION..................................................................................................................... 52 Introduction................................................................................................................... ..........52 Methods........................................................................................................................ ..........55 FM-LE Stratification.......................................................................................................56 Non-negative Matrix Factorization.................................................................................56 Clinical and Biomechanical Assessment Tools...............................................................57 Statistical Analysis..........................................................................................................5 8 Results........................................................................................................................ .............58 Fugl-Meyer Assessment and EMG Activation Patterns..................................................59 Assessing Walking EMG Patterns with FMS Severity...................................................59 Fugl-Meyer and Walking Performance...........................................................................60 Non-Negative Matrix Factoriza tion and Walking Performance.....................................60 Discussion..................................................................................................................... ..........60 4 MODULATION OF CUTANEOUS REFLEXES POST-STROKE: RELATIONSHIP TO WALKING PERFORMANCE AND INTERLIMB COORDINATION........................70 Introduction................................................................................................................... ..........70 Methods........................................................................................................................ ..........74 Participants................................................................................................................... ...74 Walking Assessment Measures.......................................................................................74 Kinematics and Kinetics..................................................................................................75 Electromyography...........................................................................................................76 Nerve Stimulation............................................................................................................76 Data Processing...............................................................................................................7 7 Statistical Analysis..........................................................................................................7 8 Results........................................................................................................................ .............78 Reflex Responses and Walking Performance Measures.................................................78 Phase Dependent Reflex Modulation..............................................................................79 Differences between Paretic and Non-Paretic Stimulation.............................................80 Discussion..................................................................................................................... ..........80 5 DISCUSSION................................................................................................................... ......92 Background..................................................................................................................... ........92 Currently Defined Patterns of Recovery.................................................................................93 Neurobiological Control of Walking......................................................................................97 Relationship of Neural Control of Walking to Experiments..................................................99 Conclusion..................................................................................................................... .......102 LIST OF REFERENCES............................................................................................................. 106 BIOGRAPHICAL SKETCH.......................................................................................................116
9 LIST OF TABLES Table page 2-1 Correlations of gait char acteristics with walking sp eed and stroke severity.....................49 2-2 Walking and pedaling correlations....................................................................................49 3-1 Fugl-Meyer and walking performance measures...............................................................66 3-2 NNMF correlations with walking assessment measures...................................................66 4-1 Subject demographics....................................................................................................... .85 4-2 Correlations between reflex modulation and walking parameters.....................................85 4-3 Paretic reflex activity wi th paretic leg stimulation............................................................89 4-4 Paretic reflex activity with non-paretic leg stimulation.....................................................90 4-5 Comparisons between paretic and non-pa retic stimulation pare tic reflex activity............91
10 LIST OF FIGURES Figure page 1-1 ICF model of rehabilitation................................................................................................ 35 1-2 Central control centers for control of walking...................................................................36 1-3 Reflex reversal............................................................................................................ .......37 1-4 Afferent signaling to reflex interneurons...........................................................................37 1-5 Post training modulation of cuta neous reflexes in spinalized cats....................................38 2-1 Comparison of the anterior-posterior ground reaction forces for the paretic and nonparetic legs of subjects of di ffering hemiparetic severity..................................................50 2-2 Comparison of the average propulsion (expressed as a percent of the total propulsion) generated by the paretic and non-paretic legs of subjects of differing hemiparetic severity........................................................................................................... 51 2-3 Individual walking speed data for subjects in each of the hemiparetic severity groups....51 3-1 Phase descriptions for the gait cycl e for someone with right hemiparesis........................67 3-2 Tibialis anterior (TA) bursting patterns during right isolated DF (A) and during walking (B).................................................................................................................... ....67 3-3 FM-LE and EMG activation..............................................................................................68 3-4 Walking EMG patterns with FMS severity.......................................................................69 4-1 Relationship between soleus modula tion and paretic propulsion deviation (stimulating the non-paretic leg)........................................................................................86 4-2 Relationship between tibialis anterior modulation and walking speed (stimulating the non-paretic leg)............................................................................................................... ...87 4-3 Normalized middle latency re flex and background EMG and ACRE150 for targeted muscles........................................................................................................................ .......88 5-1 Power curves for the hip (A) and ankle (B).....................................................................104 5-2 Motor homunculus........................................................................................................... 105
11 Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy MEASUREMENT OF ACTIVITY SPECI FIC BEHAVIORAL RECOVERY IN CHRONIC STROKE By Mark Goodman Bowden May 2009 Chair: Steven A. Kautz Cochair: Andrea L. Behrman Major: Rehabilitation Science Current examinations of post-stroke motor control rely on interp retation of isolated movements performed during standardardized clini cal examinations. Due to the complexity of human walking, however, evaluation of walking specific motor control likely needs to be conducted as the patient is ambulating. Task spec ific evaluations may enha nce the ability of the clinician to distinguish recove ry of functional behavior by means of compensation for motor control deficits from true rest itution of walking specific motor control. The purpose of this dissertation is to examine assessments by which individuals post stroke may be examined during walking in order to distinguish restitution of physical function from compensatory responses. Current clinical examinations do not differentiate motor activity during the task or in walking and correlate poorly with functional and biomechan ical walking performance measures. Factor analysis of electromyographic moto r patterns, however, yields speci fic modules of activity that correlate significantly with each of the examin ed walking performance measures. Voluntary, discrete activities may be inadequate to captu re the complex motor behavior in walking, and walking specific measures are required to descri be the efficacy of reha bilitation on behavioral recovery. One such measure is derived from the anterior-posterior ground reaction forces
12 generated during walking. These forces are resp onsible for the propulsion of the center of mass anteriorly and we propose a measure (paretic pr opulsion) that allows for examination of the paretic leg contribution to overa ll propulsion. This measure is positively correlated both with speed and with severity of hemiparesis. Perhap s most importantly, paretic propulsion allows the investigator to distinguish functional compen sation from physiological restitution by providing a measure of coordinated output of the paretic le g. Additionally, we assesse d measures of spinal level reflex activity to examine the degree to wh ich those with post stroke hemiparesis modulate sensory input. While healthy controls modulate in a systematic and reproducible fashion, those with stroke demonstrate substantia lly more variable responses. We determined that paretic leg responses differ depending on side stimulated, indicat ing that stroke leads to altered function at the level of the spinal cord during gait. Furt her exploration is require d to fully understand the motor control and clinical implications.
13 CHAPTER 1 LITERATURE REVIEW Introduction The present moment may be the most optimal time in history to be involved in rehabilitation research. Technology is progre ssing rapidly, improving our ability to provide interventions and capture necessary, quantifiable data to measure capacity and performance. Translational research is bridgi ng the gap between basic and clinic al sciences, and partnerships spanning research domains are a llowing scientists to grow knowle dge at an unprecedented rate. Neuroscientists and clinicians have not always used the same language to communicate findings and suggestions, however, and it is imperative that the rehabilitation community institute a common framework to guide communication for all individuals participating in the broad spectrum of activities from scie ntific inquiry to patient care. A potential mechanism by which such a framework may be achieved is by providing definitions within the structure and understanding of the rehabilitation models that are currently accepted as crit ical in international rehabilitation research. As Alan Jette stated in a recent review of rehabilitation models, a framework demonstrates great potential to provide the rehabilitation disciplines with a universal language with which to discuss disability and re lated phenomena (Jette, 2006). Application of current growth is dependent upon a structural framework that allows researchers and clinicians to understand and communicate common interests and findings in a way such that science may be moved forward optimally. One area in which the nomenclature creates conf usion is the discussion of recovery versus compensation as it relates to neuroanatomy, phys ical impairments after neurologic injury, and functional performance. For example, the term recovery has been used somewhat interchangeably to refer to the amelioration of st ructural deficits within the nervous system and
14 the improvement seen at a functional level as th e result of a neuroreh abilitation intervention (Levin et al., 2008). This functional recovery, however, does not often distinguish between a return to a pre-pathological pattern of functioning and an adaptive compensatory response relying on altered performance of varying effectors within the system. Examining functional improvements to distinguish restitution of physi cal performance from compensatory adaptations requires a level of measurement that is currently sparse and largely unapplied to the field of rehabilitation in general, and neurorehabilitation specifically. Current tools for assessing the functionality of walking are based on physical performance measures such as walking speed, distance walked, physical independence, use of ad aptive equipment and/or orthotic devices, and observational methods of balance control. Howe ver, improvements in all of the above may be attained via compensation of other limbs and body segments, and in fact tell the observer very little of the degree of functional restitution. The purpose of this dissertation is to provide assessments by which individuals post stroke may be examined during the functional activity of walking in order to distinguish restitution of physical function from compensatory responses. This literature review aims to introduce: 1) the significance of the problem of im paired walking post stroke; 2) the need for quantification of physical performance measures; 3) the International Classifica tion of Functioning, Disability, and Health (ICF) and how this model can help define recovery versus compensation; 4) the literature relative to examination of biomechanical factors of hemiparetic walking that assist in explaining walking specific motor control poststroke; 5) a standardized measure of motor control post-stroke, the Fugl-Meyer Assessment (FMA) and how quantifi cation of RMA results may elucidate walking specific motor control pos t-stroke; and 6) a proposal for a method of
15 exploring if impairments in modulation of spinal level cutaneous reflexes relate to impaired biomechanical and functional walking performance in a sample of individuals post-stroke. Significance of Problem Stroke is one of the most debilitating medical conditions in America, affecting approximately 750,000 people each year with a su rviving cohort of n early 8 million people (Gresham et al., 1995). Hemiparesis, strictly defi ned as a muscular weakness or partial paralysis of half of the body, is seen in three-quarters of persons post-st roke. Seventy-three percent of those surviving stroke will have some degree of long term disability (Gresham et al., 1995), and less than 50% of survivors progr ess to independent community am bulation (Perry et al., 1995a). Even among those who achieve i ndependent ambulation, significant residual deficits persist in balance and gait speed, with 60% of persons poststroke reporting limitations in mobility related to walking (Gresham et al., 1995) Given that in persons post-stroke, improving walking speed is 1) independently related to overa ll health status (Studenski et al., 2002); 2) a strong predictor of functional recovery (Richards et al., 1995c); 3) reflective of both physiological and functional changes (Bowden et al., In Press); and 4) th e most often stated goal during rehabilitation following a stroke (Bohannon et al ., 1988), interventions aimed at improving walking speed, and by implication functional walking status, are an im portant goal. This goa l for effective therapy creates a critical need to measure the effect of the intervention as well as predict appropriate candidates for therapy. As many contemporary th erapeutic approaches are based on activity based therapies targeting available plasticity of the central nervous system, it becomes even more paramount to develop and utilize measures that accurately capture the results of task specific interventions.
16 Need for Quantification of Rehabilitation Research Findings Accurate and reliable measures are vital in or der to interpret findings accurately, regardless of the domain of rehabilitation models in whic h one focuses his or her research. Without appropriate measures, all research findings are subject to threats of internal validity (the degree to which changes in the dependent variable are caused by careful and controlled manipulation of the independent variable) and external validity (the degree to which you can make inferences from a study to the general population). Measurement lies in the ability to represent the targeted construct accurately, and too often existing outcomes represent different or multiple constructs. True objective measurement may be defined as r epetition of a unit amount which maintains its size with an allowable range of error no matter which instrument is used to measure the variable of interest (test free) and no matter who or what relevant person or thing is being measured (sample free) (personal communication, Craig Velozo). In the above definition, all measurement is an approximation of a perfect concept, which itself is an abstraction. An example would be the concept of length, which is a perfect abstraction as it has neither a maximum nor a minimum value. The ruler, a co mmon tool to measure le ngth, is not perfect as there will always be a degree of error associated with its use. However, the ruler is an appropriate representation of the c onstruct of length. The ruler is standardized, sample free, test free, efficient, and precise, thus meeting the criteria defined above. Much of the historical measurement in reha bilitation resides at th e level of rankings and scores based on observation. Research focuse d on body structures and functioning, however, requires precise instrumentation and quantification techniques. Biomechanical measurements of joint angles, ground reaction forces, velocity, and acceleration are appropriat e quantifications of the proposed constructs associated with human movement and force production. Similarly, EMG signal amplitude, frequency, and timing are appropriate for measur ing the construct of
17 neural activation to a muscle. Ho wever, unclear definitions of c onstructs and the use of ordinal level measurement continue to exist, even within the level of body function research. In the past decade, researchers have begun us ing quantifiable biomechanical measures as outcome measures for interventions based on am elioration of impairments such as decreased strength (Gregory et al., 2007; Pa rvataneni et al., 2007; Teixeira -Salmela et al., 1999) and for task-specific interventions such as locomoto r training (McCain et al., 2008; Sullivan et al., 2006). However, these assessments are technological ly and monetarily cos tly and require a great deal of time to complete, making their application in the clinic almost im possible. In fact, a 2001 survey of European rehabilitation centers, only six out of 68 centers (~9 %) were using any type of technological assessment tool routinely, and these were primarily limited to video analysis and goniometry (van Wijck et al., 2001). Interestingly, five of these six centers were associated with educational institutions, and only 2/3 of the responding centers ev en had access to any type of technological assessment tool. Barriers to im plementation included a lack of money, lack of training, insufficient technical suppo rt, and a lack of information. Perhaps even more troubling is the fact that only approximately half of the centers used at least one non-quantifiable assessmen t scale in at least 75% of the cases (van Wijck et al., 2001). The most common pattern seen in neurorehabilitation clinics was a basket of assessment tools including the Modified Ashworth Scale (for spasticity assessme nt), the Functional Independence Measure (to assess independence in activities), an d the Fugl-Meyer Assessment (to assess strokespecific motor control) (van Wijck et al., 2001) Barriers to implementation of measurement included a shortage of time, a lack of informati on, a lack of training, or the fact that the tests were too cumbersome. This particular survey did not address how tools were used, nor did it address the issue of standard ization, but previous reports have documented a lack of
18 standardization between rehabilita tion facilities (Wade, 1992). Of great importance is how these instruments may provide valid information regard ing severity of the construct or how such information may guide interventions. Clearly, more work needs to be completed wi thin the rehabilitation research community not only to develop quantifiable tools to captu re meaningful information about individuals impairments and performance, but also to deve lop clinical analogues th at delineate similar information. Furthermore, necessary education, training, and support need to be provided to clinicians in conjunction with the academic community to assist th e use of quantifiable information in the treatment planning and progr am implementation for those with neurological injury. International Classification of Func tioning, Disability, and Health (ICF) In 2001, the World Health Organization (WHO) attempted to clarify the original rehabilitation model put forth by Saad Nagi (Nag i, 1965) in order to achieve the following: 1) redefine the components of the model; 2) make th e model multidirectional; 3) clearly define the environmental and personal factors that contri bute to the model; and 4) combine medical and social models of disability. In addition, the IC F model was designed as a classification system to be used in concert with the WHOs Internati onal Classification of Di seases, Tenth Revision (ICD-10). Combined, these system s systematically group health a nd health related domains in an attempt to develop a meaningful picture of the health of people or populations in order to be used for decision-making purposes (WHO, 2001). The new definitions of the ICF are as follows (Figure 1-1) (WHO, 2001): BODY FUNCTIONS. physiological or psychological functions of body systems (WHO, 2001). BODY STRUCTURES: anatomic parts of the body such as organs, limbs, and their components (WHO, 2001).
19 ACTIVITY: execution of a task or involvement in a life situation in a uniform environment (capacity) (WHO, 2001). PARTICIPATION: execution of a task or involve ment in a life situation in an individuals current environm ent (performance) (WHO, 2001). ENVIRONMENTAL FACTORS: the physical, social and at titudinal environment in which people live and conduct their lives (WHO, 2001 ). Environmental factors may either be individual (including home, workplace, and school as well as the direct contact one may have with others in this environment) or societal (other social structures, services, and systems that may have an impact on an individual). PERSONAL FACTORS: the background of an indi viduals life and living, and are defined by the features of the individual that are not part of a hea lth condition or health state(WHO, 2001). Examples of personal fa ctors include gender, race age, lifestyle habits, upbringing, coping styles, social ba ckground, education, expe riences, character style, and psychological assets. One of the clear changes in th e ICF model compared to previ ous models of disablement is the inclusion of environmental and personal factors as contributors to the overall health condition of an individual. While other models imply an environmental impact and contribution, the ICF defines and clearly delineates the contributions of the environmental and personal factors. In contrast to the Nagi model and the first genera tion WHO model, where th ere is a progression as the environment becomes disabling, the ICF rec ognizes that environmenta l and personal factors are pervasive throughout the model and can aff ect body structures and functioning as well as activity and participation. Becau se of the interconnectedness of the model, intervention at a particular area has the potential of impacting any other area of the model. In further contrast to the Nagi model, the ICF is not limited to discussion of the pathological condition, but rather is designed to describe the health status of any individual. In a healthy population, the model describes norma l body structure and function, activities, and participation whereas they describe impairm ents, activity limitations, and participation restrictions in those in an abnormal health condition (WHO, 2001). Furthermore, one does not have to demonstrate a progression linearly throu gh the model. As examples: 1) someone with a
20 physical disfigurement (impairment) may not ha ve activity or partic ipation limitations; 2) societal stigma associated with a diagnosis may limit participati on in the absence of impairments; and 3) the use of adaptive equipm ent may reduce or decrease activity limitations and participation restrictions even in the presence of profound impairment. Lastly, in contrast to earlier models, the ICF allows for the presence of indirect impairments, which are not caused by a disease process directly, but inst ead are sequelae to disease and processes. As an example, an acute spinal cord injury does not lead directly to decubitus ul cer formation, but instead ulcers form secondary to factors such as prolonged immobility, reduced nourishment, and maceration. The ICF makes unique contributions in its abi lity to 1) define health status for both impaired and non-impaired individuals; 2) allow for entrance into the model at any point and not rely on progression from one extreme to the ot her; and 3) account for indirect effects of pathological states. For these r easons, the ICF has been adopted as the model of choice locally by the College of Public Health and Health Professions and was recently adopted nationally by the American Physical Therapy Association as its model of choice to guide rehabilitation efforts. For my particular area of research interest, th e above advantages of the ICF are particularly salient in defining and quantifying the subtasks of human locomotion. Within the context of the above definitions of body structures and activity, the concepts of recovery and compensation may take on different meanings. At a neuronal level, most researchers would agree that the term recovery in stroke connot ates reactivation in brain areas previously nonactivated by the circulatory event (Levin et al., 2008). When discussing functional recovery, however, researchers and clin icians alike fail to distinguish if recovery occurs at a body structure or activity leve l. Current assessment tools focus on task accomplishment as opposed to discriminating the m echanistic underpinnings of how the task is
21 performed. As Levin recently st ated, we have to demonstrate that functional motor outcomes are superior when therapeutic intervention is aimed at the reacquisition of motor elements underlying functional task accomplishment (Levin et al., 2008). If patients are identified who fail to recover at a functional level, we ma y determine either that the intervention was insufficient to promote recovery or that some patients may lack the capacity for task-specific motor learning. This failure to distinguish adequa tely the effect of the interventions not only limits determination of therapeu tic efficacy, but it also fails to distinguish those who may maximally benefit from recovery-based interv entions from those who would perhaps optimize functional performance with a compensatory appro ach. The ability to customize interventions to maximize the motor learning potential of the individual is dependent upon accurate understanding and measurement of structural im pairments as well as f unctional performance. Building on the ICFs definition of body struct ure/functioning, Levin posits that recovery at this level is characterized by restoring the ab ility to perform a movement is the same manner as it was performed before injury relying on pre-morbid levels of strength, range of motion and movement patterns (Levin et al., 2008). Conve rsely, compensation at th e body structure/function level requires performing a movement in a new manner and is characterized by altering degrees of freedom through co-contracti on, altered timing, and altere d combination of movements (abnormal synergies). Similarly, recovery of activity level performance revolves around successful task accomplishment using pre-morbid effectors, while compensation allows for functional success with altered effector usage when compared to pre-morbid or healthy control patterns. Clinical examinations such as manua l muscle testing and spasticity assessment may give accurate information as to the recovery of body structure impairments while giving very little information regarding how this structural recovery rela tes to functional recovery.
22 Conversely, functional assessments such as walki ng speed provide a great deal of information about expected functionality (P erry et al., 1995b) but does not tell the examiner how that function was attained. A full battery of examin ation is required at both the body structure and activity levels, measuring both r ecovery and functional capacity in order to move therapeutic interventions forward for true cust omization and optimization of care. The experiments described within this proposal focus primarily on measures of recovery at the activity level of the rehabilitation mode l and the relationship between recovery and performance based measures of walking performa nce. This framework offers the potential to further understand walking performance and thus improve upon the next generation of interventions designed to assist those with stroke relation hemi paresis improve walking function. Biomechanical Examination of Hemiparetic Walking As deficits in walking perfor mance post-stroke are related to a reduced ability to produce forces necessary to advance the center of ma ss and limbs forward during walking, quantifiable techniques are required to assess this ability during the task. One mechanism by which this may be accomplished is through examination of the surface electromyographic (EMG). Electromyography EMG illustrates the electrical signal that oc curs when the motoneuron communicates with the muscle to activate muscle firing. EMG analys is has been used for many years as a way to evaluate neuronal activation of the muscles. This activation can be described temporally, defining the firing of different pa tterns of activity in locomotor-specific patterns during the gait cycle, and amplitude, which attempts to quantif y neuronal input by the am ount of voltage in the electrical signal. EMG signals are alternating curr ent, so initially signals are rectified to arrange all of the signals in the positive direction and th en filtered to remove unrepresentative spikes in amplitude. EMG signals also contain a great deal of noise and variability making interpretations
23 on a single or few gait cycles problematic at best (Knutson and Soderberg, 1995). For this reason, the preferable analysis i nvolves averaging over several gait cycles, an analysis that is made easier by the multiple gait cycles that are collected in walking trials on an instrumented treadmill. To complete this analysis, however, the data must be normalized temporally, meaning that each cycle must be identified from 0% to 100% by some external indicator. In the experiments in the Human Motor Performance Labor atory, this is achieved by identifying foot contact via GRFs on the treadmill. Once temporal ly, normalized, separate gait cycles may be averaged together, creating an ensemble average of linear envelops, cont rolling for variability and noise through the averaging of multiple trials. Additionally, background EMG must be taken in to account to glean out resting electronic signals from those associated with walking beha vior. Simply taking the lowest activity in the cycle as rest is insufficient as postural stabiliz ation muscles must be activated, and this activity is not normal particularly in the neurologically impaired population (Knutson and Soderberg, 1995). Resting EMG in our protocols is quantifie d during a quiet sitting activity. EMG by itself only relates one mechanism associated with walki ng behavior. For example, electrical activity does not distinguish between concentric and ecc entric contractions, no r does it explain the kinesiological correlates that are happening concurrently.(Knutson and Soderberg, 1995) Additional factors such as energy absorption and generation, limb positioning, and joint moments must be c onsidered as well. A second mechanism of invest igating motor control biomech anically incorporates an analysis of ground reaction forces (GRF). These investigations lead to the ability to examine movement of the center of mass as well as to calculate joint moments, powers, and work relative
24 to specific phase of the walking cycle. Defini tions and importance of moments, work, and power will be described later in this section. GRFs are derived from Newtons third law, st ating that for every action there is an equal and opposite reaction. In the study of human motion, this law indicates that all surfaces provide a reaction force and that the indivi dual is acted upon by that force when s/he is in contact with that surface (Hamill and Knutzen, 1995). In human motion analysis, the study of this reaction is achieved via the use of a force plate which meas ures the vector describing the force acting upon the individual. This vector is comprised of th ree orthogonal components: the vertical force; the anterior/posterior force; and the medial and lateral force. Analysis of these individual components within the context of a free body diagra m allows investigators to summarize all of the forces acting on a system at any particular time and describe the jo int reaction force at a particular joint. Extrapolation of these forces in a distal to proximal direction allows for the calculation of subsequent joint re action forces and is the basis for the inverse dynamics approach to human motion analysis (Hamill and Knut zen, 1995). These calculations allow for determination of the previously me ntioned moments, power, and work. Moments During walking, moments describe the summary torque that results around any joint during the gait cycle and is required to maintain dynami c stability of the system by preventing collapse when gravity leads to instability. These moment s are calculated as the product between the GRF vector and the limb segment vector when analyzed together in a kinetic chain. For example, a plantarflexor moment occurs duri ng the gait cycle as the tibia pa sses over the foot stationary on the ground. The muscle activity at this time pr oduces plantarflexion for stability, even though the movement at the time is in a dorsiflexion direction.
25 Power Power is computed by calculating the product between moments and angular velocity. When moments and angular velocity are both pos itive, then the power is positive indicating a net production of energy. When the moments and ve locity have a different sign, as when the moment is positive and the angular velocity is negative, the power is negative indicating a net absorption of energy, although this absorption may not be at the same rate as the previously described energy expenditure (Beltman et al., 2004). Considering the example of the tibia moving over the stationary foot, the moment is plantarflexion, but the velocity is slowing, stabilizing the movement and preparing for li mb advancement, indicative of an a power absorptive phase. Work Mechanical work describes the amount of fo rce used to move an object a particular distance as defined by the formula W=F x s (distance). For example, in an isometric contraction in which no movement occurs, no work is done even though a considerable amount of force may be generated. In the field of motion analysis it is often a simpler calculation, however to describe work as a product of power and tim e which may be calculated by determining the integral of the power curve between two define d time points (eg. the beginning and the end of a stance phase). Over the past 20 years, investigators ha ve begun using biomechanical analyses to investigate the contributing mechanisms to impair ed walking post-stroke. In particular, power and work are often used to describe the capac ity to produce the required mechanics for steady state walking. In 1991, Olney published a semina l description of work and power in the hemiparetic population (Olney et al., 1991). In this study, Olney describes both positive work (integral of positive power curves) and negative (integral of negative work curves), as both are
26 indicative of energy expenditure. In particular the following phases of the gait cycle contribute to mechanical work being performed by each leg: positive work by the ankle plantarflexors during terminal stance/push-off (A2); negative work by the knee extensors during weight acceptance (K1) and at push-off (K3) and positiv e work during midstance (K2); negative work by the knee flexors during terminal swing (K4) ; positive work by the hip extensors in early stance (H1) and early swing (H3, al so termed pull-off), and negative work through mid to late stance (H2). These values we re calculated for both the paretic and non-paretic legs. As a result of this work, Olney determined th at regardless of walking speed, the paretic leg contributes approximately 40% of the positive wo rk being performed (Olney et al., 1991). The profiles between the two legs visu ally appear very similar, with the primary difference being one of amplitude, and Olney concludes that the compensation assumed be tween the legs is minimal. These mechanical work estimates, however, re ly on assumptions as to the recovery of mechanical work from individual sources and intercompensation between sources (Aleshinsky, 1986) and do not describe what functional task s are being accomplished by the work. Forward dynamic simulation models have demonstrated th at more of the mechanical work performed during a gait cycle is done in ea rly single leg stance in order to raise the bodys center of mass (COM) than occurs in double limb support (DLS ) prior to swing, which primarily provides forward propulsion and swing initiation (Neptune et al., 2003). The power produced during DLS likely decreases even more in those with slower walking speeds, implying that an even higher proportion of work is done to raise the COM. The mechanical work estimates provided by Ol ney thus may not describe the task of propelling the body forward during walking, which is an essential requ irement of locomotion (Shumway-Cook and Woollacott, 2001), and instead may more accurately describe the role of
27 the hemiparetic leg to support body weight during stance. Subseque ntly, additional measures are required to describe adequately the critical role of moving the body forward during locomotion and to assess if these measures are sensitive to levels of hemiparetic severity. Therefore, the purpose of the first study in this dissertation is to examine an terior/posterior GRFs as an appropriate method of measuring th e contribution of the paretic le g to the coordinated task of forward propulsion during walking. Examination of Post-stroke Motor Control Motor recovery post-stroke is difficult to measure, and theories surrounding motor function post-stroke have been dominated by the concept of progres sing through predictable stages of recovery (Brunnstrom, 1966; Twitc hell, 1951). This progres sion is based on the organization of reflex behavior, th eorizing that severe impairments reflect a return to previously assimilated primitive reflexes. According to this theory, primitive reflexes provide the necessary background for more complicated voluntary mo vements (Fugl-Meyer et al., 1975). Someone with non-flaccid paralysis (preservation of reflex es) presents with a reco very of motor function in a regular sequence in which initial moveme nts are dependent on reflex-based synergistic movements. Patients recovering from stroke gradually develop fully integrated voluntary movement patterns, relying less on reflexive beha vior (Brunnstrom, 1966). Based on this theory, Fugl-Meyer in 1975 developed a measurement inst rument reflecting this reflex hierarchy to quantify recovery post-stroke (Fugl-Meyer et al., 1975). This instrument is divided into upper extremity and lower extremity components focusi ng on distinct constructs such as reflexes, movement control, coordination and speed, with an additional section specific to balance recovery (Fugl-Meyer et al., 1975). Specifically the lower extremity motor evaluation (FM-LE) consists of a total score of 34 poi nts with 17 items scored on a 0-2 scale. In addition, an 11-item
28 (22 points) sub-section of the FM-LE is dedicat ed to analysis of abnormal movement synergy patterns (FM-S) and excludes the reflex and coordination/speed parameters. The FM-LE, however, is based on voluntary, discrete tasks based on the dominant influence of cortical input on motor control. In addition, the FM-LE exam ines motor control in three theoretically progressive positions: supine, sitting, and standing. However, the motor control deficits that the FM meas ures may differ from deficits s een during task specific activities such as walking. A recent study investigated the abnormal movement patterns seen post-stroke by analyzing strength deficits and movement patterns from a functionally relevant standing position (Neckel et al., 2006). The authors found although those with stroke were significantly weaker than neurologically heal thy control subjects. Those with hemiplegia and controls used similar strategies to achieve movements. In fact, only during maxi mal hip abduction did a significant secondary movement of hip flexion emerge in those with hemiplegia, mimicking the abnormal synergy patterns described by Brunnstr om and Fugl-Meyer (Brunnstrom, 1966; FuglMeyer et al., 1975). These resu lts suggest that the primary impairment in post-stroke motor control is weakness, and that correct interpreta tion of post-stroke motor control can only be gleaned from positioning that is re levant to the targeted behavior These findings are consistent with task specific approaches to studying behavior such as walking. Based on the original work by Graham Brown, (Brown, 1911) scientists have more recently investigated the role of the spinal co rd in the control of walking. After long-term training involving manual assistance from trainers cats with severed thoracic spinal cords were able to step on a treadmill with full weight bearing at vary ing speeds (Barbeau and Rossignol, 1987; Lovely et al., 1986). Spatiotemporal char acteristics, kinematics, and EMG responses all approximated normal cats at comparable speeds (Barbeau and Rossignol, 1987). These studies
29 also indicate the importance of sp ecificity of the intervention as t hose cats that were trained to walk could do so but demonstrated limited capac ity for static standing, while those cats standtrained were successful in standing but not wa lking (Hodgson et al., 1994) Furthermore, the training effects were reversible as stand-trained cats could successfully complete walking training, and those cats trained to walk were cap able of stand-training Both groups that were retrained demonstrated reduction of the originally trained skil l (Hodgson et al., 1994). These discoveries were in sharp contrast with the pr evious assumption of the immutability of spinal cord function and laid the founda tion for interventions aimed at the possibility of the spinal cords ability to modulate peripheral input and to learn motor tasks. The presence of walking capacity in the absen ce of any supraspinal i nput gave rise to the description of central pa ttern generators (CPGs). CPGs are t hought to be located at the level of the spinal cord and can be the controller of rhythmic patterned behavior such as walking (MacKay-Lyons, 2002) and breathing (Barlow and Estep, 2006). CPGs can coordinate cyclic activity in the lower extremities and may be driven supraspinally or peripherally, and peripheral sensory signaling provide cues that enable the human lumbos acral spinal cord to modulate efferent output in a manner that may facilitate th e generation of stepping (Harkema et al., 1997). This peripheral afferent input is sufficient in the absence of s upraspinal control to drive the rhythmicity of the pattern (Zehr, 2005). Afferent input comes in the form of cutaneous feedback, vibratory sense, proprioception, lo ad, and muscle stretch and provi des information to assist in modulating the behavior (Nielsen, 2003). Within the CPG, there likely exists a complex pool of interneurons that assist in shaping the beha vior, and it has been hypothesized that perhaps a single pattern generator may exist for a variety of rhythmic activities such as walking, biking, or swimming that is modulated by the interneuron pool.(Zehr, 2005) In corporating knowledge
30 from animal models and the theory of CPGs, Ba rbeau first described training on a treadmill with body-weight support in humans in 1987 (Barbeau et al., 1987). Reviews of the existing studies are available in the literature a nd describe the theoretic framework associated with LT (Barbeau et al., 2006; Dietz and Harkem a, 2004), describing it as one of the evidence-based clinical approaches that will be used in the 21st century to enhance recovery of posture and locomotion in stroke, SCI subjects and in many other neurological conditions (Barbeau, 2003b). These studies illustrate the increasi ng understanding that training of th e spinal cord is dependent on optimal stimulation of the necessary afferent inpu ts needed to train spinal circuits responsible for producing desired rhythmic motor patters su ch as walking (Dromerick et al., 2006). Human walking is an incredibly complex task involving a multitude of degrees of freedom and an immense number of combinations of musc le activity. While it is perhaps staggering to consider that the brain can volunt arily control all of these variab les during a cyclic, rhythmical task, definitive studies of the sp inal control of walking in human s are difficult to conduct due to the inability to directly assay the human central nervous syst em. In addition, modifications that humans must make for the demands of upright bipe dalism make direct tran slation from studies of quadruped CPGs exceedingly problematic (Nielsen, 2003). As Nielsen stated in a recent review of the central control of muscle activity during walki ng, it is the task of the whole central nervous system to generate this muscle activity, to ensure that it is optimally coordinated, to ensure that it is adjusted to the immediate environment, and to modify it when required (Figure 1-2) (Nielsen, 2003). Nielsen co ncludes by saying that there is no reason to suggest that human walking is controlled exclus ively by the spinal cor d, nor is there a reas on to imply that the motor cortex alone is responsible for activation of muscles during walking. Instead, this activity
31 related to walking must rely on an integration of spinal neuronal circuitry, afferent signals and descending motor commands (Nielsen, 2003). This complex integration of motor control requires increased complexity of assessment tools to distinguish true recove ry and response to neuroplasticbased interventions. Clinical examinations, however, continue to rely on non-ta sk specific, voluntary ac tivation of movement patterns to describe motor control post stroke, and it is suggested that these current clinical measures may be insufficient. The purpose of the second study of th is dissertation is to quantitatively analyze if the FM-LE examination adequately assesses poststroke motor control relative to walking, or if additional assessment t ools are required to ca pture this integrated capacity of the human nervous system to produce motor control required for walking. As a posthoc analysis, we will examine additional methods of assessing spinally modulated control of walking and relate both assessments to performan ce in clinical and biomechanical measures of walking performance. Examination of Spinal Cutaneous Reflexes The term reflex has often been used to desc ribe distinct, stereot yped motion in response to fixed peripheral nerve stimulations. Howeve r, recent research has illustrated that the connotation that reflexes are stereotyped and immutable is pa tently false (Zehr and Stein, 1999). Reflexes, including those elicited from peripheral cutaneous afferents from skin mechanoreceptors, demonstrate task, phase, and in tensity dependence for modulation of reflex amplitude. Perhaps the best working definition of a reflex is a response evoked with great probability by particular stimuli (Brooks, 1986). The purpose of th is section is to introduce the spinal cutaneous reflex by desc ribing its modulations, effect on walking performance, and usefulness in assessing activity -dependent neuroplasticity.
32 Task Dependency of Cutaneous Reflexes Early human studies of cutaneous reflexes demonstrated that subnoxious stimulation of cutaneous nerves resulted in alterations in musc le spindle firing rates without changes in motor neuron firing rates, a result that was only seen when in a stan ding position (Aniss et al., 1990). Burke et al later discovered inhibitions of the tibia lis anterior, soleus, biceps femoris, and vastus lateralis within 100 ms of a stimulus, but only when the muscles were activated. Additionally, the early latency reflexes (60-80 ms) were modulat ed differently depending on the stability of the posture (sitting, standing, and perturbed stan ding), finding increased amplitudes with more unstable postures (Burke et al., 1991). Furthermore, positional changes may not only alter the magnitude of the response, but also may revers e the role. For example, stimulation of the posterior tibial nerve in standing inhibits the sole us reflex, while the same reflex is stimulated while prone. Interestingly, the pr one response reverses to suppressi on if a pressure is applied to the sole of the foot, indicating th e importance of load receptors and cutaneous input to the sole of the foot in reflex modulation (Abbruzzese et al., 1996). Phase Dependency of Cutaneous Reflexes As in the prone versus standing example a bove, cutaneous reflex responses may change role from suppression to excitati on or vice versa depending on the phase of the gait cycle during which the stimulation was applied (De Serres et al., 1995; Duysens et al., 1992; Yang and Stein, 1990). The phenomenon is known as a reflex reversal and can be seen in Figure 3. For example, in the tibialis anterior, stimulation during swing produces an excitatory response while a similar stimulation produces an inhibitory response in the swing to stance transition (Figure 1-3). DeSerres et al. utilized single motor unit analyses to determine that reflex reversals were likely due to competing parallel interneuronal pathways to the alpha motor neuron and that the reflex pathway depended on the phase dependent route th rough available interneurons (Figure 1-4) (De
33 Serres et al., 1995). Typically, th ese reflex reversals are only witn essed in muscles such as the tibialis anterior and biceps femoris that have two patterns of EMG bursti ng during the course of the gait cycle when stimulating nerves that are predominantly cutaneous (Stein, 1991). Functional Role of Cutane ous Reflexes during Walking Cutaneous reflexes appear to have a func tional significance of a stumbling corrective response, which has been documented in cats (Drew and Rossignol, 1987; Forssberg, 1979) as well as humans (Zehr et al., 1997). Specifically, Duysens first obser ved kinematic alterations in ankle dorsiflexion in humans as a result of the swing phase excitatory modulation, but instead of attributing the response to a stumbling correctiv e response, concluded that the responses are instead related to the opening and closing of refl ex pathways to a central pattern generator used in human locomotion (Duysens et al., 1992). Zehr later performed a more extensive analysis of the kinematics of human walking and demonstrat ed correlations to knee flexion during early swing, ankle dorsiflexion in late stance, and ankl e plantarflexion during late swing (Zehr et al., 1997). These movement patterns are all consistent with prevention of tripping, promotion of smooth transition of the swing limb, and preparat ion for early weight acceptance (Zehr et al., 1997). It has been hypothesized that the cutaneous reflex is most predominant in both stance to swing and swing to stance transitions and acts with other reflexes to maintain stability during active movement (Zehr and Stein, 1999). In a subsequent study, those with post-stroke hemiparesis demonstrated reflex modulation, although the predominant pattern was inhibition, and these reflexes failed to correlate with ki nematic responses in the same manner as healthy controls (Zehr et al., 1998). It should be noted, however, that mechanical alterations such as increased joint stiffness may have contributed to the lack of kinema tic correlations in those with stroke. In spinal cord injur y, reflex modulation was also main tained, and differed from healthy controls in that the predomin ant response was excitatory (Jones and Yang, 1994). Kinematic
34 correlations were not performed in this study, bu t excessive excitation may explain many of the gait abnormalities seen in those with spastic gait patterns after sp inal cord injury. At this point, it is unclear if cutaneous reflexes maintain thei r functional significance in those with central nervous system injury. Role of Reflexes in Study of Ac tivity Dependent Neuroplasticity Presently, a great opportunity exis ts to develop studies inves tigating the potential of the human nervous system to modify reflexes as a re sult of activity-dependent rehabilitation in those with central nervous system injuri es. While H-reflex modulation ha s been studied as a result of locomotor rehabilitation post spinal cord inju ry (Trimble et al., 2001; Trimble et al., 1998), investigators have not yet utilized the potenti al of examining multisynaptic cutaneous reflex modulation in this population. Similarly, cutaneous reflexes have been minimally studied in the stroke population, and current activ ity based therapies create a n eed to understand the alteration of cutaneous reflexes as the result of an in tervention. However, a single study was recently completed in the spinalized cat model undergoing locomotor training (Cote and Gossard, 2004). In this study, 10 of the 71 pathways studied demonstrated some degree of post-training modulation. Six of these 10 path ways involved the medial planta r nerve, which innervates the plantar surface of the foot, i ndicating that ground contact may pl ay an important role in the modulation of the reflexes (Cote and Gossard, 20 04). Figure 1-5 illustrates an example of a medial gastroc motoneuron and the reduction of inhibition that is seen as a response post locomotor training (Figure 1-5). This study demonstrates promise for cutaneous reflex modulation post nervous system lesion and provides a rational for study in the human population.
35 Conclusion In summary, the following experiments allo w the possibility of examining individuals post-stroke during the func tional activity of walking in an effort to measure elements of recovery at the activity level. Biomechanical patterni ng in ground reaction force generation, analysis of EMG patterning, and spinal level reflex modulation may all reflect ways in which individuals post-stroke achieve the behavior of walking. These assessment tools, while computationally intensive, may provide preliminary evidence of the ability to disti nguish recovery from compensatory adaptation at an activity level. These measurement tools could therefore serve as a framework by which clinical analogs could be developed to improve clinicians ability to optimize clinical decision making and interpret intervention efficacy. Figure 1-1. ICF mode l of rehabilitation.
36 Figure 1-2. Central control centers for contro l of walking. (from Niel sen JB. How we walk: central control of muscle activity during human walking. Neuroscientist. Jun 2003;9(3):195-204.) Muscle activ ation comes directly from the cortex as well as spinal CPGs and is modulated at each location by afferent input.
37 Figure 1-3. Reflex reversal. (from Zehr EP, St ein RB. What functions do reflexes serve during human locomotion? Prog Neurobiol. Jun 1999;58(2):185-205.) Notice the lack of response during stance but a switch from exc itation to inhibition in the middle latency going from swing to stance initiation. Figure 1-4. Afferent sign aling to reflex interneu rons. (From Zehr EP. Training-induced adaptive plasticity in human somato sensory reflex pathways. J Appl Physiol 2006; 101: 178394.) Reflex reversals may be due to compe ting interneuronal pathways to the alpha motor neuron.
38 Figure 1-5. Post training modula tion of cutaneous reflexes in sp inalized cats. (From Cote MP, Gossard JP. Step training-dependent plas ticity in spinal cutaneous pathways. J Neurosci 2004; 24: 11317-27.) Note the ea rly latency excitation and later latency decrease of inhibition.
39 CHAPTER 2 ANTERIOR-POSTERIOR GROUND REACTION FORCES AS A MEASURE OF PARETIC LEG CONTRIBUTION IN HEMIPARETIC WALKING Introduction Post-stroke hemiparesis results in a unilateral primary impairment of the paretic leg that results in a disrupted walking patt ern. Therapeutic strategies fo r walking recovery have largely focused on the paretic leg, as it exhibits a reduced muscular output, which is evidenced by decreased maximal voluntary contraction (Mulro y et al., 2003), reduced EMG amplitudes during walking (Knutsson, 1981), and a decline in mech anical work performed (Olney et al., 1991). However, there are limitations in available qua ntitative measures of strength (they are not specific to the walking task), EMG (it is difficult to synthesize information from multiple muscles) and mechanical work (e.g., requires mu ltiple assumptions (Aleshinsky, 1986)) such that currently available measures give an incomple te assessment of the coordinated output of the paretic leg during walking. Walking speed is the most widely used measure of performance; however, compensatory action by th e non-paretic leg can result in a relatively functional walking speed despite poor coordination of the paretic le g. While a training prog ram may increase ones walking speed by making a compensatory strategy more effective, current neurorehabilitation philosophies based on the princi ples of neuroplasticity are di rected at the restitution of neurological deficits. Thus, a qua ntitative measure of the coordina ted output of the paretic leg might predict the outcome of therapy; assist in evaluating the specific muscle coordination changes associated with various therapeutic interventions; and correlate with structural and functional studies of the nervous system such that the underlying mechanisms can be better understood. Olney et al (1991) suggested that the paretic leg, regardle ss of hemiparetic severity, performs approximately 40% of the mechanical work of walking, as calculated from kinetic
40 analyses based on intersegmental joint powers (O lney et al., 1991). These mechanical work estimates, however, rely on assumptions as to th e recovery of mechanical work from individual sources and intercompensation between sources (Aleshinsky, 1986), and do not describe what functional tasks are being accomp lished by the work. Forward dynamic simulation models have demonstrated that more of the mechanical work performed during a gait cycle is done in early single leg stance in order to raise the bodys COM than occu rs in double limb support (DLS) prior to swing, which primarily provides forwar d propulsion and swing ini tiation.(Neptune et al., 2003) The power produced during DL S likely decreases even more in those with slower walking speeds, implying that an even higher proportion of work is done to raise the COM. Currently, no adequate measure exists to examine the paretic leg contributions to the task of propelling the body forward during walking, which is an essen tial requirement of locomotion (Shumway-Cook and Woollacott, 2001). Paretic limb work production has previously been studied utiliz ing a well-controlled pedaling paradigm (Kautz and Brown, 1998; Kaut z and Patten, 2005). This model has shown that the paretic leg produces significantly less mechanical work output than do healthy, agematched controls (Brown and Kautz, 1999). This net decrease in mechanical work output is a product of less positive work and more negative work being done by the paretic leg (Brown and Kautz, 1999). Of particular inte rest is that the pedaling-derive d measures of mechanical work seem to assess coordinated output independent of the need to support the body by providing postural stability as a result of the seated posture As a result, the contribution of the paretic leg to the pedaling task in the most impaired subj ects was found to be near zero, or even negative (i.e., hindering task accomplishment that require s additional work done by the non-paretic leg).
41 Thus in contrast to the findings of Olney et al. (1991) in walking, we were able to link hemiparetic severity to motor perfor mance (Brown and Kautz, 1999). Similarly, the anterior-posteri or ground reaction forces (A -P GRFs) may represent an appropriate method of measuring th e contribution of the paretic le g to the coordinated task of forward propulsion during walking. Previous studies have implem ented A-P GRF as a measure of the forward propulsion and braking in people with hemiparesis utiliz ing a cane for ambulatory assistance (Chen et al., 2001), but the A-P GRFs have not been ut ilized as a measure of the mechanical contribution of the paretic and non-pa retic legs. In additi on, we propose comparing the paretic leg coordinated output in walking to our previously derived measures for pedaling and propose that pedaling measures will provid e confirming evidence of the importance and robustness of these walking measures. Specifica lly, we hypothesize that m easures derived from the A-P GRF impulse (paretic and non-paretic propulsive impulse, pa retic and non-paretic braking impulse, and paretic net propulsion) will correlate with gait speed, hemiparetic severity, and the positive and negative work measurements in pedaling. Methods The A-P GRF data presented in this study were collected (but not reported) as part of a larger study that investigated th e links between gait characteristic s and bone density in chronic stroke survivors (Worthen et al., 2005). Participants Individuals presenting with chro nic stroke were recruited for this study at the Palo Alto Department of Veterans Affa irs Medical Center. Forty-seve n individuals with chronic hemiparesis (41 male, 6 female; ages = 62.4 10.2 (SD) years; time since stroke (yrs) = 4.3 3.8; affected side: left = 25, right = 22) participated in the stu dy. A subset of this population also
42 participated in a pedaling study in our lab (Kautz et al., 2003). This sample of 16 included 14 male, 2 female; 9 with left hemiparesis, 7 with right hemiparesis; average age of 63.5 + 6.6 years; average chronicity was 3.0 + 1.4 years. Written informed consent was obtained from all participants for each study and the Stanford Un iversity Administrative Panel on Human Subjects in Medical Research ap proved both protocols. Inclusion criteria included the following: unila teral weakness; less th an 85 years of age; time since stroke greater than 12 months; if female at least 5 years past the onset of menopause; and ability to walk 10 meters in 50 seconds or le ss without contact assistan ce. Exclusion criteria included the following: more than 1 previous cerebral vascular incident; inability to provide informed consent; use of osteoporosis drug or hormone replacement therapy within the past 5 years; history of leg fracture or pain; and the ex istence of any other medical condition that could affect bone mass. Participants were characterized according to th eir level of hemiparetic severity, identified on the basis of the Brunnstrom motor recovery stages (Brunnstrom, 1966). These participants demonstrated a range of abilities to perform movements within and outside of extensor and flexor synergy patterns (Brunnstrom, 1966). Severe hemiparesis was defined as subjects rated as Brunnstrom stage 3 (n=19), moderate hemiparesis was defined as subjects rated as either a Brunnstrom stage 4 or 5 (n=18), and mild hemipa resis was defined as subjects rated Brunnstrom stage 6 (n=10). Variables Walking speeds were measured while each participant walked on a 4.3 meter long GAITRite portable walkway system (CIR System s, Inc). Additionally, GRF were measured throughout the stance phase for both the paretic a nd non-paretic legs as each participant walked at their self-selected speed along a 10-meter walkway equipped with embedded force platforms
43 (Advanced Medical Technology, Inc and Bertec). GRF data were acquired at 200 Hz and were filtered with a lowpass fourth order Butterworth filter at 20 Hz forward and backward in time. The A-P GRF component (normalized by each in dividuals body weight) was used in the subsequent analysis. Four to fi fteen trials were collected in or der to assure adequate contact on the force platforms to determine the GRF and walking speed for each participant. When possible, multiple foot contacts were averaged to generate GRF values, but in one participant, only one trial could be analyzed due to inc onsistent foot striking on the force plate. A subset of the sample also participated in a separate experiment in which they were assessed on a cycle-ergometer to evaluate work production generated by each leg. The positive, negative, and total work (sum of positive and nega tive) were calculated for each lower extremity. The pedaling evaluations were completed with a custom two-servomotor ergometer, which has been described previously in the literature (Kautz and Patten, 2005). In the present pedaling trials, the servomotors were programmed to em ulate conventional two-legged pedaling, and toe clips were utilized to allow the hip flexor s to generate power during the cycle. The stance phase was separated into four bins in order to analyze impulse generation at various time points in the gait cycl e: 1) double limb support after pare tic foot strike, 2) the first 50% of paretic single limb stance, 3) the second 50% of paretic single l imb stance, and 4) double limb support prior to paretic swing. Customized Matlab programs were written to process the data, and when possible raw data from two cons ecutive heel strikes were analyzed to examine the temporal relationship with the contralateral leg. When consecutive heel strikes were not collected, average temporal relationships were assu med to be representative of a subject's gait. Variables derived from the A-P GRF were defined as follows: 1) propulsive impulse is the time integral of the positive A-P GRF, 2) braking impulse is the time integral of the negative A-P
44 GRF, and 3) net impulse is the sum of the propulsive impulse plus the braking impulse for each leg. Propulsive and braking impulses we re also calculated within ea ch bin. The percentage of total propulsion generated by the paretic leg, referred to as paretic propulsion (Pp), was calculated by dividing the propulsi ve impulse of the paretic leg by the sum of the paretic and non-paretic propulsive impulses. Statistical Analysis Correlations between parametric variables were analyzed using Pearsons correlation coefficient, while correlations with hemiparetic severity levels were performed using the nonparametric Spearmans correlation. All statistic s were run using SPSS version 11.0 (SPSS, Inc.). Results Gait Characteristics Figure 2-1 illustrates the A-P GRF tracing for th ree representative participants. In steady state (constant speed) walking, the area under the curve in the positive direction (propulsion) should roughly equal the area under the curve in the negative direction (braking) and the two legs tracings should be similar in shape and magn itude. Note that is the top participant, the paretic and non-paretic legs were fairly symmetrical. However, the more impaired subjects were asymmetrical. In order to maintain stea dy state walking speeds, reduced net propulsion by the paretic leg has to be offset by increased pr opulsion in the non-paretic leg. As the graphs progress from mild to severe hemiparesis, a sm aller percentage of total propulsion was generated by the paretic leg (PP), corresponding with a slower walking speed. Walking speed Walking speed was positively correla ted with paretic propulsive impulse, non-paretic braking impulse, paretic net impulse, non-paretic net impulse, a nd the net impulse of Bin4 on the paretic leg (Table 2-1).
45 Hemiparetic severity. Severity was significantly corr elated with propulsive impulse, net impulse, and bin4 net impulse in the paretic le g, and propulsive impulse, braking impulse, and net impulse in the non-paretic leg (Table 2-1). Analyses then examined the eff ect of hemiparetic severity on PP. Figure 2-2 illustrates that for those categorized as mild severity, the mean PP is approximately 49%. Those participants who demonstrated moderate severity had a mean PP of 36%, whereas thos e participants with severe hemiparesis had a mean PP of only 16%. All of the participants in this study were allowed to walk with the assistive and/or orthotic device that they normally use in everyday walk ing. Eight of 19 participants with severe hemiparesis used an ankle-foot orthosis (AFO), and 11 of the 19 used some form of unilateral assistive device. Analyses were done to see if or thotic and/or assistive device usage affected the PP. There was not a significant difference (p =0.176) between those that used an AFO (PP=12.07%) and those who did not (PP =19.32%). There was a non-significant difference (p=0.74) between those us ing assistive devices (PP =12.45%) and those who did not (PP=21.52%). Those participants who used neit her an AFO nor an assistive device had a PP of 21.14%. PP was significantly correlated with both speed (r=0.551, p=0.000) and with hemiparetic severity (r=0.737, p=0.000). Note, however, that five individuals with severe hemiparesis walked faster than the functionally signifi cant 0.8 m/s(Perry et al., 1995b) and all had PP < 25% (Figure 2-3). Additionally, three individuals with mild severity walked more slowly than 0.8 m/s and all had PP > 49%. These eight participants are indi cated in Figure 2-3 with filled markers. Pedaling Characteristics Pedaling subset. Sixteen individuals completed the full pedaling and gait evaluations, including work and force production results.
46 Correlations of pedaling and walking. Measurements of work production in pedaling (total work, positive work, and negative work) we re correlated with the impulse generated during walking (paretic propulsive impulse, paretic brak ing impulse, paretic net impulse, and paretic net Bin4 impulse) (Table 2-2). Correlations of pedaling and walk ing with hemiparetic severity. Hemiparetic severity was positively correlated with total work done in pedaling (r=0.798, p=0.000), with positive work done in pedaling (r=0.588, p=0.017), and negative work done in pedaling (r=0.791, p=0.000). Discussion PP was found to provide a quantit ative measure of the coordina ted output of the paretic leg that is sensitive to hemiparetic severity. The correlation between PP and hemiparetic severity was strong and there was a drama tic difference between those with severe hemiparesis (16% PP) and a mild hemiparesis (49% PP). These findings demonstrate th at Pp levels are sensitive to hemiparetic severity, differing from mechanical work estimates that calculate work at a constant 40% regardless of severity (O lney et al., 1991). Reduced PP by those with severe hemiparesis does not seem to be attributable solely to the use of an assistive and/or orthotic device, as even those with severe hemiparesis not using assistive and/or orthotic devices demonstrated markedly reduced Pp (21.14%). In addition to PP, the raw measures of walki ng presented, i.e. paretic propulsive impulse, net paretic impulse, and net im pulse in Bin4 are all se nsitive to the severity of the hemiparesis, implying that the A-P GRF qu antities are appropriate measures for assessing the paretic leg contribution to walking that are sensitive to severity. Additionally, these results also corroborate previous findings correlating hemi paretic severity with the pedaling measures. The overall aim of this paper was to further our understanding of th e contribution of the paretic leg to forward propulsion in hemipare tic walking. Previous work on hemiparetic
47 interlimb coordination and contri butions to total work using a pedaling paradigm has indicated that work measurements are sensitive to hemiparetic severity (Kautz and Brown, 1998). Specifically, those with more se vere hemiparesis produce less to tal work and encounter more resistance from the paretic limb than do those with less severe hemipare sis. Comparing A-P GRF to pedaling work, we see some similarities in the task of walking as both total and positive work strongly correlate with propulsive impulse, net paretic impulse and net impulse generated during Bin4, which is the most propulsive phase of the gait cycle (Neptune et al., 2001). It is important to realize that, in addition to the active generation of propulsive forces by muscles, there were also direct mechanical influe nces associated with the value of Pp achieved because of the expected strong relationship be tween foot placement and amount of propulsive impulse. If the leg were to act purely as a rigi d strut (i.e., GRF vector parallel to long axis of leg), as in an idealized inverted pendulum, the A-P GRF would be directly related to the position of the foot relative to the bodys COM (i.e., an terior foot position produc es posterior GRF, the posterior foot position produces an anterior force, and asymme try between the percent of the stance phase with the foot anterior versus pos terior would introduce a similar asymmetry in propulsive impulse). Thus, reduced propulsion in Bin4 is likely rela ted to an inability to achieve adequate hip extension (e.g., sufficiently posterior foot position) in add ition to reduced active generation of propulsive GRF by the muscle forces. Note that the resistive impulse late in Bin4 that follows the generation of so me propulsive impulse (Figure 21, bottom tracing) is unlikely to be related to the direct mechanics as the f oot is likely behind the COM in this phase. Consequently, direct mechanical effects are not sufficient to explain Pp. In addition, the biomechanics underlie one of the differences between walking and the pedaling paradigm.
48 Negative work in pedaling is likely due to resi stance from the paretic leg, while braking in walking may be a mechanical response to havi ng taken a longer stride with the paretic leg. Strong correlations were seen with net impul se (e.g., propulsion resistance) in Bin4, the double limb support prior to the swing of the paretic limb. Bin4 is important in the attainment of speed during walking, and DeQuervain illustrated that those with hemipa resis with the slowest gait velocity spend the most tim e in Bin4 (De Quervain et al., 1996). This phase may be of particular importance in the act of progressing the body forwards as it coincides with the burst of ankle power (Olney et al., 1991), much of which is stored mechanical energy from gastrocnemius and soleus activity and acts to propel the body fo rwards (Neptune et al., 2001). Both total work and positive work performed during pedaling showed th e strongest correlation with GRF impulses in Bin4 and this phase also had the strongest correlation with hemiparetic severity. Lastly, it may be possible to utilize the PP to document compensatory gait patterns. For example, 5 of those with a severe hemiparesis had a walking speed greater than 0.8 m/s, but had a PP of < 25%, implying that they were achieving their velocity through some mechanism other than paretic leg contribution. These compensa tions to achieve functi onal velocities are not appreciated when examining walking speed alone, but may be important in assessing outcomes for a therapeutic intervention. A training program may increas e ones walking speed by only making a compensatory strategy more effective, but current neurorehab ilitation philosophies based on the principles of neuroplasticity are directed at the restitution of neurological deficits. Thus, PP may be an effective tool in distinguishing functional co mpensation from physiological restitution.
49 Table 2-1. Correlations of ga it characteristics with walking speed and stroke severity Outcome Measure Walking Speed Hemiparetic Severity # Paretic Propulsive Impulse r= 0.641 p=0.000 r=-0.650 p=0.000 Paretic Braking Impulse r=-0.235 p=0.112 r=-0.162 p=0.276 Non-Paretic Propulsive Impulse r= 0.140 p=0.350 r= 0.467 p=0.001 Non-Paretic Braking Impulse r= 0.696 p=0.000 r= 0.507 p=0.000 Paretic Net Impulse r= 0.363 p=0.012 r=-0.659 p=0.000 Non-Paretic Net Impulse r=-0.431 p=0.002 r= 0.753 p=0.000 Paretic Bin4 Net Impulse r= 0.748 p=0.000 p=-0.681 p=0.000 *Pearson Correlation Coefficient, #Spearman Correlation Coefficient Table 2-2. Walking and pedaling correlations Paretic LE: Pedaling Paretic LE: Walking r Sig. Propulsive Impulse 0.588 0.017 Net Paretic Impulse 0.550 0.027 Total Work Net Bin4 Impulse 0.748 0.001 Propulsive Impulse 0.613 0.012 Net Paretic Impulse 0.631 0.009 Positive Work Net Bin4 Impulse 0.661 0.005 Negative Work Braking Impulse 0.064 0.813
50 Figure 2-1. Comparison of the anterior-posteri or ground reaction forces for the paretic (red lines) and non-paretic legs (blue lines) of subjects of differing hemiparetic severity. Positive values represent propulsion, and the positive area under the curve is the propulsive impulse. PHS = paretic heel strike, NTO = non-paretic toe off, NHS = non-paretic heel strike, and PTO = paretic to e off. Increased hemiparetic severity was associated with decreased PP and decr eases in self-selected walking speed.
51 Figure 2-2. Comparison of the average propulsion (expressed as a percent of the total propulsion) generated by the paretic (blue ba rs, note that this is defined as Pp) and non-paretic legs (red bars) of subjects of differing hemiparetic severity. There are substantial differences in th e percent of total propulsion generated by the paretic leg (PP) in those with severe and moderately severe hemiparesis when compared with those with mild severity. Compensation by the non-paretic leg is noticeable in the asymmetry shown by the moderate and severe groups. Error bars indicate standard deviation for each variable. Figure 2-3. Individual walking speed data for subjects in each of the hemiparetic severity groups. While there is a weak overall tr end for walking speed to decrease with severity, the variability is substantial and the speed ranges in each of the groups overlap substantially. Five participants w ith severe hemiparesis are still able to achieve normal walking speeds (>0.8 m/s), but all demonstrate substantial decreases in PP. Conversely, 3 participants with m ild hemiparesis walked <0.8 but have normal PP (> 49%). These eight participants are indicated with filled markers.
52 CHAPTER 3 EVALUATION OF ABNORMAL SYNERGY PATTERNS POST-STROKE: RELATIONSHIP OF CLINICAL EXAMINATION TO HEMIPARETIC LOCOMOTION Introduction Motor recovery post-stroke is difficult to measure, and theories surrounding motor function post-stroke have been dominated by the concept of progres sing through predictable stages of recovery (Brunnstrom, 1966; Twitc hell, 1951). This progres sion is based on the organization of reflex behavior, th eorizing that severe impairments reflect a return to previously assimilated primitive reflexes. According to th is theory, primitive motor pathways accessible by reflex activation provide a f oundation for more complicated vo luntary movements (Fugl-Meyer et al., 1975) and someone with non-flaccid hemi paresis (preservation of reflexes with no voluntary movement) would present with a recovery of motor function in a regular sequence in which initial voluntary movements are dependent on primitive motor pathways and synergistic movements. Patients fully recovering from stroke are thought to gradually develop more complex motor behaviors, fully integrating volunta ry movement patterns out side of stereotypical abnormal synergy patterns (Brunnstrom, 1966). Based on this theory, Fugl-Meyer in 1975 developed a measurement instrument reflecting this hierarchy of the emergence of complex motor control behaviors in order to quantify recovery of motor function post-stroke (Fugl-Meyer et al., 1975). This instrument is divided into upper extremity and lower extremity components focusing on distinct constructs such as reflex es, voluntary control of isolated movement, coordination and speed, with an additional section speci fic to balance recove ry (Fugl-Meyer et al., 1975). Specifically, the lower extremity motor evaluation (FM-LE) consis ts of a total score of 34 points with 17 items sc ored on a 0-2 scale. Since its inception, the FM-LE has been stud ied extensively to doc ument stroke-related motor impairment and recovery. The FM-LE ex am has undergone scrutiny via reliability and
53 validity studies (Duncan et al., 1983; Fugl-Meyer et al ., 1975), and has been used to validate other instruments (Lamontagne et al., 2001; Wa ng et al., 2002). Furthermore, the FM-LE has been utilized to measure the efficacy of novel therapeutic approaches (Dobkin et al., 2004), and has been included in models attempting to pr edict functional recovery (Kollen et al., 2005; Nadeau et al., 1999). However, the motor control de ficits that the FM measures may differ from deficits seen during activiti es such as walking. Walking is a ta sk specific activity that is at least partially controlled by spinal cord level automa ticity via the presence of a complex system of spinal interneurons, or centra l pattern generator (CPG) (Bar beau, 2003a; Dietz and Harkema, 2004; Edgerton et al., 2004). The CPG has been pr oposed to act in concert with both peripheral afferent input and supraspinal control to produ ce functional and coordi nated walking behavior (Nielsen, 2003; Yang and Gorassini, 2006; Zehr, 2005). As such, evaluation of voluntary motor impairment using FM-LE may have limited abilit y to measure the neural determinants of walking dysfunction. In particul ar, control of isolated volunt ary force and movement differs markedly from the cyclical pa tterns of walking, which rely h eavily on spinal circuits and sensorimotor integration. Recent advancements have been made in th e use of EMG factoriz ation procedures to identify shared patterns of activation among groups of muscles. Such approaches are consistent with the historical clinical pe rspective of functiona l muscle groupings, but allow more objective determination of muscle groupings under any movement condition including complex functional activities. Ivanenko et al. have used a princi pal component analysis to identify five basic underlying factors that explain 90 % of the variance of muscle EM G during gait activities that are modulated by both descending and proprioceptiv e signals (Ivanenko et al., 2004). Using jumping, swimming, and walking patterns of frogs, dAvella and Bizzi utilized EMG recordings
54 to identify a mixture of synergie s that may be isolated or sh ared to account for all of the movement variability (d'Avella et al., 2003). Th is control is speculated as being downstream of the processes that generate motor activation (i.e. cortical inputs) (Ivanenko et al., 2006). Using microstimulation, iontophoresis, a nd behavioral analysis, Bizzi et al. localized this modular organization at the level of the sp inal cord, and defined the ability to generate a specific pattern of motor output with a specific pa ttern of input into these spinal modules (Bizzi et al., 2008). Additionally, non-negative matrix f actorization (NNMF) analysis has demonstrated that these synergistic activation patterns produce previously reported stereo typical responses to postural perturbations to promote balance equilibrium (Ting and Macpherson, 2005; Torres-Oviedo and Ting, 2007). Recent work in our laboratory has utilized the NNMF to identify the number and composition of EMG modules ( NNMF factors) accounting for lower extremity EMG during walking in healthy adults and adul ts post-stroke (Clark et al., 2008) Emergence of these patterns post-stroke may reflect an emergence of comple x behavior and neuromot or organization during the task of walking that cannot be identified during isolated single and multi-joint movements. The purpose of this study is to test whethe r the motor impairment measured by the FM-LE is indicative of motor dysfunction during walking in adults with post-stroke hemiparesis. Specifically, we hypothesize that the FM-LE and FM-S will demons trate weak correlations with biomechanical and clinical measurements of walking performance. As both the FM-LE and NNMF have been used to probe the underlying neural determinants of motor function, a secondary analysis will be to quantitatively an alyze whether these two methods are measuring the same construct. Further, we test whether cl assification of stroke patients based on each of these approaches yields associa tions between the level of classi fication and deficits in walking performance, hypothesizing that NNMF classificati on will better stratify walking performance.
55 Methods Individuals with chronic (great er than six-months post-stroke ) hemiparesis participated in a study at the Department of Veterans Affairs ( VA) Medical Center in Ga inesville, FL. Thirtyfour individuals (15 female and 19 male), age 60 + 12.2 years (standard deviation), 13 with right and 21 with left hemiparesis part icipated in the study. Particip ants had a history of a single unilateral stroke, were ambulatory without contact assistance, were able to follow a multiple step command, and did not have other medical issues interfering with their ability to walk. In addition, 17 healthy controls (3 male, 14 female, mean age 65.1 + 10.7 years) participated in the study and walked at 0.6 m/s to match the average wa lking speed of the hemiparetic participants. All participants signed written informed consen t approved by University of Florida Institutional Review Board/Gainesville VA Subcomm ittee for Clinical Investigation. Each participant walked for two 30 sec ond trials on an instrumented treadmill (Techmachine, Andrezieux Boutheon, France), to collect ground reaction forces (GRF) and kinematic data. GRF data were acquired at 200 Hz and were filtered with a low pass fourth order Butterworth filter at 20 Hz forward and backward in time. The A-P GRF component (normalized by each individuals body weight) was us ed in the subsequent analysis. Surface EMG (Konigsberg Instruments, Pasadena, CA) was acquired using bipolar Ag-AgCl surface electrodes (Vermed, Inc., Bellows Falls, VT) during treadmill walking and the FM-LE from eight different muscles: tibiali s anterior (TA); soleus (SOL); gastrocnemius (GAS); vastus medialis (VM); rectus femoris (RF); biceps fe moris (BF); semimembranosus (SM), and gluteus medius (GM). Reference electr odes were place over the electr ically neutral patella. EMG signals were filtered with a 40Hz high pass filter and then a 20Hz low pass filter for averaging multiple steps of walking data or a 4Hz low pass filter to smooth one-trial data for the subsequent FM-LE analysis.
56 FM-LE Stratification Each participant was stratified by severity acco rding to 22-point sub-section of the FM-LE that examines the ability to perform voluntary isolated m ovement independent from mass patterns of whole-limb co-activ ation (FM-S) and excludes the reflex and coordination/speed parameters (severe, n=11; moderate, n=14; mild, n=9) and examined with a battery of walkingspecific clinical and biomechanical assessment tool s. In this stratifica tion, a FM-S score of < 15 characterized severe hemiparesis; 15-19 ch aracterized moderate hemiparesis; and > 20 determined mild hemiparesis (Kautz and Patte n, 2005). These cutoffs are based on theoretical limitations of moving within abnormal synergy pa tterns, combining syner gy patterns, or moving at least partially outside of the patterns. Non-negative Matrix Factorization For each subject, EMG were combined into an m x t matrix (EMGo), where m indicates the number of muscles and t is the time base (t = # of strides x 101). A non-negative matrix factorization algorithm was then applied to this matrix for a set of consecutive gait cycles because inherent stride-to-stride variability contai ns structured information that is critical to differentiating between independe nt factors and establishing r obust factor definitions. NNMF defined the factors by populating two matrices: 1) an m x n matrix indicating the relative weighting of each muscle within each factor; an d 2) an n x t matrix reflecting the activation timing profile of the factor across the gait cycle. NNMF assumes that the weightings remain fi xed over the entire gait cycle, and that muscles may belong to more than one factor. The two matrices were multiplied to produce an m x t matrix of reconstructed EMG (EMGr), which was then compared to the original EMGo and the agreement quantified by calculating the sum of the squared errors: (EMGoEMGr) 2. Within
57 this framework, NNMF performed an iterative optimization procedure by adjusting the weightings and timings until they converged on factor definitions that minimized the error. Separate NNMF analyses were performed with the output constrained to 1, 2, 3, 4 and 5 factors. To determine how many factors were actually needed for each leg of each subject, we calculated the vari ability accounted for (VAF = 1 (EMGoEMGr)2/EMGo 2). VAF was calculated for each muscle across the entire gait cycle and for al l muscles within each of six phases of the gait cycle (calculate d as the cumulative VAF for all muscles within each phase). Beginning with a single factor, if VAF was greater than or equal to 90% for all 14 conditions (8 muscles, 6 phases), then it was concluded that ad ditional factors were not needed. Otherwise, the number of factors was increas ed until all conditions achieve d 90% VAF or until adding an additional factor did not substant ially increase VAF for the muscle(s) and/or phase(s) with the lowest VAF. Clinical and Biomechanical Assessment Tools Walking speed. Self-selected walking speed overground was measured on an instrumented walkway (GAITRite, CIR Systems, Inc, Havertown, PA). Berg Balance Test (BBT). The BBT is a 14-item test th at requires an individual to perform everyday tasks of increasing difficulty such as sitting, moving from one chair to another, standing up, turning around, picking up an item fro m the floor, as well as the performance of more challenging tasks such as standing on one foot (Berg et al., 1992a; Berg et al., 1992b). Dynamic Gait Index (DGI). The DGI rates performance from 0 (poor) to 3 (excellent) on eight different gait tasks, including gait on ev en surfaces, gait when changing speeds, gait and head turns in a vertical or hor izontal direction, stepping over or around obstacles, and gait with pivot turns, and steps (Shumw ay-Cook and Woollacott, 2001).
58 Paretic propulsion (Pp). Pp is a quantitative measure of the coordinated output of the paretic leg, which describes the contribution of the paretic leg in propelling the center of mass forward during walking and is defined as the pe rcentage of propulsion performed by the paretic leg.(Bowden et al., 2006) Statistics were r un on the absolute devi ation from normal (0.5). Paretic step ratio (PSR). PSR is defined as the percenta ge of stride length performed by the paretic step (Balasubramanian et al., 2007). Statistics were run on the absolute deviation from normal (0.5). Paretic pre swing (PPS). The percentage of the gait cycle spent in the double-limb support phase prior the paretic preswing (De Quervain et al., 1996). The gait cycle was divided into six phases for data analysis (Figure 3-1). The percentage of EMG activity in each phase was compared between each FMS severity level and for controls. Statistical Analysis Group analyses were completed using a non-pa rametric Kruskal-Wallis H Test with Rank Sums Tests post-hoc analyses. FM-LE and FMS valu es as well as the number of NNMF factors explaining EMG variability were correlated to the walking assessment battery using the nonparametric Spearmans correlation coefficient. Significance for all tests was set at alpha < 0.05. All statistics were run usi ng SPSS version 15.0 (SPSS, Inc.). Results Of the 34 participants, several demonstrated higher activation of the TA during walking than during isolated dorsiflexion tasks in supine (n=12, 35%), dur ing sitting voluntary dorsiflexion (n=17, 50%), and dur ing standing dorsiflexion (n=18, 53%). An example of this decreased TA activation during isolated tasks is seen in Figure 3-2, in which an individual exhibited no active dorsiflexion during the isol ated FM task but demonstrated active DF movement during walking accompanied by discrete bursting of the TA EMG (Figure 3-2). The x
59 and y scaling for the EMG tracing was kept consis tent for the purposes of visual comparison. The accompanying images illustrate the inabilit y to dorsiflex the ankle in item 10.a. and functional dorsiflexion during the walking cycle in item 10.b. Fugl-Meyer Assessment and EMG Activation Patterns When stratifying according the FMS severity, five of the six FM-LE tasks show varying degrees of differences between severity levels (F igure 3-3). In the supi ne extension task, TA (p=0.046), MG (p=0.048), BF (p=0.005), SM (p= 0.001), and GM (p=0.008) demonstrate main effects, and of these BF (p=0.004), SM (p =0.002), and GM (p=0.0008) show significant increases in EMG activity in the moderate group co mpared to the mild group. In the sitting knee flexion task, only BF demonstrated a main effect (p=0.05) with si gnificantly greater activation in the mild group compared to the moderate (p=0.037) TA demonstrated a main effect in both the sitting dorsiflexion (DF) (p=0.017) and standi ng DF (p=0.013) tasks, although only the sitting DF demonstrated a significant in crease in activation for the m oderate group, compared to the severe (p=0.018). Standing knee flexion demons trated a significant main effect among a majority of muscles: SOL (p=0.32), MG (p =0.19), VM (p=0.013), BF (p=0.0.1), SM (p=0.011) and GM (p=0.002). Of these, the BF (p=0.008), SM (p=0.006), and GM (p=0.001) demonstrated significantly higher activation in the mild hemiparesis group. Ev en though there are significant differences among muscles, there are no mass extension or mass flexion patterns noticed in the more complex tasks for the moderate and seve re groups, and generally mild hemiparesis is represented by increased activa tion of the primary movers. Assessing Walking EMG Patterns with FMS Severity Among those with hemiparesis, only Phase 2 (0.026) and Phase 4 (0.011) of the RF demonstrate a main effect among the severity groups (Figure 3-4). In Phase 2, the moderate and severe groups demonstrate a decrease in activ ation, while these same groups demonstrate an
60 increase in activation in Phase 4 compared to those with mild hemiparesis (RF Phase 2 p=0.027 and RF Phase 4 p=0.044). Observationally, large bursts are seen consis tently in Phases 1 and 4, representative of paretic limb loading and swing initiation, respec tively. The control curve, however, is much more differentiated and shows add itional bursting in the MG and GAS consistent with late stance plantarflexor activity. Additionally, the control cu rves demonstrate increased late swing activity (Phase 6) in the BF and SM consistent with li mb deceleration, preparatio n for initial contact. Fugl-Meyer and Walking Performance Of the six clinical and biomech anical measures of walking performance examined, walking speed, BBT, and PSR were signif icantly correlated with the FM-LE, while only the walking speed and PSR correlated significan tly with the FMS (Taable 3-1). Non-Negative Matrix Factorization and Walking Performance When participants were characterized by the number of NNMF factor s required to explain their EMG variability, there wa s strong evidence for correlation with all variables (p<.025) (Table 3-2). Discussion When examined against a battery of clinical and biomechanical walking measures, the FMLE and FMS both correlated significantly with se lf-selected walking speed and PSR, and the FM-LE was additionally correlated with the BBT The correlation with walking speed is consistent with previous reports in the litera ture(Nadeau et al., 1999) and likely reflects the general motor impairment that is present in this clinical population. Simi larly, while establishing correlations between the FMA and both self-selec ted and fastest comfortable walking speeds, Nadeau et al. also performed a regression an alysis using the FMA (balance, FM-LE, and sensation portions entered sepa rately), a spasticit y index, isometric dynamometry scores, and
61 spatiotemporal analysis values into a multiple regression model to examine specific contributions to the walking speed. Hip flexor strength, balance, and FMA were all significantly correlated to both self-selected and maximal walking speeds, but the multiple regression analysis indicated that only the hip flexor strength was predictiv e of self-selected walk ing speed. Hip flexor strength, sensation of the lower extremities, and plantarflexor strength were all predictive of maximal walking speed, but again, the FMA was not part of the predictive model.(Nadeau et al., 1999) This failure of the FM-LE to contribute significantly to a predictor model is true for walking speed as well as functional walking profiles such as the Functional Ambulation Categories (FAC) (Kollen et al., 2005). In this study, assessments were taken longitudinally 18 times during the first year post-stroke and incl uded the following measurements: FAC, FM-LE, Motricity index leg score, letter cancellation task (LCT), FM-balance, and the timed balance test (TBT). The primary outcome measure of the stu dy was the change over time of the FAC and the contribution of other outcomes to the regression m odel. All of the covari ates listed above were significantly correlated with the FAC change score when analyzed with a bivariate regression model, with TBT having the st rongest relationship, followed by FM-balance, FMA, LCT, and Motricity index. Multivariate modeling indicated that when a ll of the above factors were combined into a single regression model, the model only predicted 18% of the change in the FAC (Kollen et al., 2005). It is inconsistent with previous literature that Pp did not significantly correlate with hemiparetic severity, as previ ous reports from our laborator y found significant correlations between Pp and Brunnstrom levels (Bowden et al ., 2006). However, this earlier sample had a higher level of ambulatory function (0.77 + 0.34 m/s in the earlier sample versus 0.57 + 0.24 m/s
62 in the current sample). In addition, Pp was calculated in the current manuscript using the absolute deviation from the normal value of 0.5, wh ereas in the previous sample, Pp was treated as a continuous variable and raw values were correlated with Brunnstrom levels and not raw Fugl-Meyer scores. Another pos sible cause of this difference is the calculation of paretic propulsion from treadmill walking, as some inves tigators have argued that TM walking differs from overground walking (Harris-Love et al., 2004). However, comparisons between treadmill walking and overground walking in healthy cont rols yield no significan t differences in the anterior propulsive forces (Goldberg et al., 2008). Additionally, the instrumented treadmill used in this experiment was recently demonstrated to be valid for laboratory gait analysis in that ground reactions, hip, knee, and ankle sagittal ro tations, torques, power, and surface EMG from four thigh and leg muscles were all not signifi cantly different than ove rground walking with the exception of an 8% decrease in st ride length (Tesio and Rota, 2008). When examining muscle activation patterns w ithin the FM-LE, the clinical examination failed to distinguish hemiparetic se verity consistently based on mu scle activation patterns. Only supine extension and standing knee flexion demonstr ated significant main effects in more than one muscle, and both showed group differences only in the BF, SM, and GM. Other than in these two tasks, EMG activity within the FM-LE is fairly consistent, regardless of severity group. Therefore, it may be inferred that four of the six tasks within the FM-LE offer practically no information regarding hemiparetic severity, at leas t as it relates to muscle activation patterns. The supine extension task illustrates the interest ing finding of increased BF, SM and GM activity in the most mildly hemiparetic group, indicating that these muscle groups may be active as hip extensors during the task. A limita tion of the FM-LE is that there are no other extension tasks to which the supine extension results may be co mpared. Conversely, the BF, SM and GM are
63 significantly more activ e in the standing knee flexion task in the mild hemiparesis group, indicating increased muscle activity during a flexor phase. This flexibility of response seen in the mild group may reflect adaptability within the nervous system that is not seen in those with moderate and severe hemiparesis. Alternatively, those with more severe hemiparesis may not be able to activate the BF, SM and GM during the tasks, but there is no evidence of mass extension and flexor patterns as the theory behind the FM -LE would assert. In analyses of voluntary single-plane motions while in a functionally significant standing position, Neckel et al. demonstrated that while indivi duals post stroke produce reduced torque in six of the eight motions in the paretic leg, they used similar st rategies to controls in seven of the motions (Neckel et al., 2006). The only evidence of an abnormal synergy patter n producing the desired movement emerged with maximal hip abduction when hip flexor torque was also recorded in the stereotyped flexor synergy activity (Neckel et al., 2006). EMG analysis of walking further illustrates the inability of the FM-LE to differentiate between hemiparetic severities as only the RF demonstrates any differences among the severity groups, as the mild group differs from the moderate in phases 2 and 4. This analysis also fails to illustrate any type of mass extension or flexi on strategy in the severe and moderate groups. What is seen during the walking trials is a cons istent burst of activity in paretic loading and paretic pre-swing. This reflects a non-differentiated burst of activ ity across all measured muscles when activation is required to stabilize the body or prepare for swing initiation. Although only significant in the RF, several other muscles (T A and GM) demonstrate an increased peak in Phase 4 for the severe hemiparetic group, even wh en activity in those muscles is not generally associated with pre-swing activity. The cont rols, on the other hand demonstrate additional peaks, namely during Phase 3 for the SOL and GAS and Phase 6 for the BF and SM.
64 These four peaks seen in the control group are consistent with NNMF analysis which indicates four factors explaining normal locomotion (Clark et al., 2008). These four factors are associated with weight acceptance (at approx imately 10% of the gait cycle), propulsion (at approximately 45% of the gait cycle), ground cleara nce (at approximately 70% of the gait cycle), and leg deceleration during the end of swing (at approximately 95% of the gait cycle). These factors and their timing correspond very well to th e peaks seen in phases 1, 3, 4, and 6. As those with hemiparesis have fewer peaks of activity, this too may reflect a decrease in the number of factors required to explain the EMG activity in walking. Seventeen indi viduals with stroke required only two factors, 15 requi red three, and only two required four factors to explain the variance in their EMG. Table 32 illustrates the degree to whic h classification by NNMF factors differentiates walking performan ce measures, which is more highly effective than by FM-LE or FMS (Table 3-1). These data st rongly imply that an increase in NNMF factors is associated with increased complexity of the motor pattern and differentiation of muscle activation. Perhaps because NNMF is based on data collected while a participant is walking, its construct may more closely reflect walking performance than an a ssessment whose construc t is based on voluntary, isolated movements as in the FM-LE. Inter-subject differences in the complexity of the walking pattern revealed by NNMF may reflect the differences in the in teraction of supraspinal, spinal and peripheral input following stroke. Current evidence does not suggest either that human walking is controlled exclusively by the spinal cord or that the mo tor cortex alone is responsible for activation of muscles during walking (Nielsen, 2003). Emergence of an increased number of factors explaining EMG variability and the relationship of those factors with clinical and biomechanical measures of performance may reflect the integr ity of descending motor pathways. However, this emergence
65 of complexity of behavior may also relate to activity in the periphery such as integration of spinal neuronal circuitry and processing of affere nt signals. While additional work is necessary to delineate the role and neural mechanisms of spinal modules of motor control, it may be that these modules are integral to the coordination of multiple inputs for the control of human locomotion. At this time, the number of modules as determ ined by NNMF is not appropriate to serve as a clinical measure because of the need for EM G data and detailed mathematical analyses. However, the FM-LE appears to be insufficient to capture necessary information about walking performance, and its use as an outcome measure for pos t-stroke motor control should be limited to non-walking related ac tivities. The FMAs effectiveness as a measure of upper extremity motor control may be related to the more dire ct corticospinal connections to the arms and decreased reliance on patterned, spinally-modulated movement (N akayama et al., 1994; Sanford et al., 1993). Future work developing clinical analogs to assess and m onitor presence and/or emergence of NNMF factors may greatly assist clinicians in accurately describing walkingspecific motor control post-stroke.
66 Table 3-1. Fugl-Meyer and wa lking performance measures SPEED BBT DGI Pp Deviation PSR Deviation PPS r=0.588 r=0.369 r=0.116 r=-0.075 r=-0.357 r=-0.291 FM-total LE p<0.001 p=0.032 p=0.534 p= 0.674 p= 0.038 p= 0.126 r= 0.456 r=0.325 r=0.058 r=-0.149 r=-0.365 r=-0.258 FM-Synergy p=0.007 p=0.061 p=0.758 p= 0.400 p= 0.034 p= 0.177 Table 3-2. NNMF correlations with walking assessment measures SPEED BBT DGI Pp Deviation PSR Deviation PPS r= 0.451 r= 0.504 r= 0.545 r=-0.389 r=-0.558 r=-0.398 NNMF Factors p=0.008 p=0.003 p=0.002 p= 0.023 p= 0.001 p= 0.020
67 Figure 3-1. Phase descriptions for the gait cycle for someone with right hemiparesis. The first double support phase defines Phase 1. Phases 2 and 3 are the firs t and second 50% of the single limb support. The second double limb support (paretic pre-swing) is defined as Phase 4. Phases 5 and 6 are the first and second 50% of the swing phase. Figure 3-2. Tibialis anterior (TA) bursting patterns during ri ght isolated DF (A) and during walking (B). Axes are identical for the tw o tracings. Notice th e higher amplitude and clear bursting pattern in walking (B) compared to the fairly tonic activity in isolated movements (a). In addition, notice the lack of DF movement in (a) compared to functional right DF during th e walking cycle in (b). A. TA activity during isolated DF in standing B. TA activity during walking at selfselected speed
68 Figure 3-3. FM-LE and EMG activation. Bars reflect standard de viations. Significant differences between FM-S groups are only not ed for TA in the sitting dorsiflexion task and for BF, SM, and GM in supine extension and standing knee flexion. Additionally, BF is significantly different in sitting knee fl exion. However, there are no mass extension or mass flexion patterns not iced in the more complex tasks for the moderate and severe groups and genera lly mild hemiparesis is represented by increased activation of the primary m overs. (*denotes significant differences)
69 Figure 3-4. Walking EMG patterns with FMS severity. Significant differences between FM-S groups are only noted for RF for Phases 2 and 4, demonstrating th at those in differing FM-S generally activate similarly during walk ing. Differences from control subjects (black line) can be clearly noticed in TA SOL, GAS, and GM. (*denotes significant differences)
70 CHAPTER 4 MODULATION OF CUTANEOUS REFLEXES POST-STROKE: RELATIONSHIP TO WALKING PERFORMANCE AND INTERLIMB COORDINATION Introduction Human walking is thought to be a combinati on of supraspinal input, peripheral sensory signals, and activity from a comple x system of spinal interneurons often referred to as a central pattern generator (CPG) (Nie lsen, 2003; Zehr, 2005). Da ting back to Sherringtons groundbreaking work in the early part of the 20th Century (Sherrington, 1906), the role of sensory inputs in elic iting reflex responses has provided a mechanism by which researchers can assay the central nervous system, and consider able recent work has focused on the Hoffmann reflex (H reflex, for review see Zehr 2002 a nd Wolpaw 2007) (Wolpaw, 2007; Zehr, 2002) and the cutaneous afferent reflex pathways (Zehr, 200 6; Zehr and Stein, 1999). The H-reflex is a very effective tool in exploring spinal cord excitabi lity due to its mostly monosynaptic nature and ability to describe the 1a pre-synaptic refl ex regulation. H-reflex amplitude, however, is modulated by both afferent feedb ack and central motor output. In contrast, the cutaneous reflex is not modulated by 1a afferent feedback and occu rs only as a result of central motor output and specific locomotor relate d activity (Duysens et al., 1992; Ro ssi et al., 1996). Background EMG can be subtracted to yield a predominant re sidual reflex response (Z ehr and Stein, 1999). Additionally, examination of cutaneous reflexes allows the potential to analyze different latencies of responses, reflec ting polysynaptic communications among differing levels of the spinal cord and potential supraspi nal circuitry (Nie lsen et al., 1997). The amplitude regulation of cutaneous reflexes is very responsive to the present functional and behavioral state of the body. These reflex es are task dependent (e.g. standing versus walking tasks) (Duysens et al., 1993 ), have contralatera l as well as ipsilateral effects (Haridas and Zehr, 2003; Tax et al., 1995), and are phase de pendent (eg. swing versus stance) (Duysens et
71 al., 1992; Duysens et al., 1990; Yang and Stein, 1990). The phase dependency of the gait cycle also is accompanied by reflex reversals in which the same cutaneous input evokes both facilitatory and inhibitory res ponses in the same muscle at differing phases of the gait cycle (Duysens et al., 1992; Yang and Stein, 1990). Fo r example, the tibialis anterior (TA) demonstrates a middle latency (80-120 ms) excita tory response when stim ulated in early swing and an inhibitory response in the swing to stance tr ansition (Duysens et al., 1992) (see Duysens and Tax 1995 for review) (Duysens et al., 1995). Based on post stimulus time histograms of individual motor units, these reflex reversals are likely due to co mpeting parallel facilitative and inhibitory pathways to the motor neuron (De Serres et al., 1995). This reflex reversal is most often seen in those muscles which have two pe riods of bursting during the course of the gait cycle (Stein, 1991). Furthermore, cutaneous reflex es appear to have a functional significance of a stumbling corrective response, which has b een documented in cats (Drew and Rossignol, 1987; Forssberg, 1979) as well as humans (Zehr et al., 1997). Lastly, the neuronal mechanisms controlling cutaneous reflex respon ses is maintained during other locomotor tasks such as incline walking and stair climbing (Lamont and Zehr, 2006) and contribute to the maintenance of stability during walking (Haridas et al., 2005). The activity of cutaneous refl exes during walking in populatio ns with impaired nervous systems has been less thoroughly studied, although the reflexes are shown to be affected by central nervous system lesion. The first such study examined the cutaneous reflex behavior during walking with spinal co rd injury, and found that refl exes were modulated, although exaggerated excitation predominated (Jones and Yang, 1994). Specifically, the TA demonstrated excitation throughout the gait cycle, while soleus demonstrated inhibition in stance and abnormal excitation responses during swing (Jones and Ya ng, 1994). Zehr demonstrated phase dependent
72 reflex regulation in those post st roke, indicating at least some ma intenance of spinal cord level processing similar to what was seen in the contro l participants (Zehr et al., 1998). However, the specific pattern of regulation differed in those post stroke in that TA was strongly suppressed through early swing, soleus and vastus lateralis were suppressed through out stance, and biceps femoris was excited only during early parts of the swing phase (Zehr et al., 1998). Interestingly, the reflex responses in stroke failed to achieve the same kinematic significance as the reflex responses in healthy controls (Z ehr et al., 1998). More recentl y, Duysens et al. demonstrated decreased reflex activity in thos e with hereditary spastic parapa resis, suggesting cortico-spinal involvement in the regulation of middle latenc y reflexes (Duysens et al., 2004). In upper extremity assessments of individu als with acute stroke, middle a nd late latency reflexes were suppressed in all of the patients studied and la tencies were exaggerate d in the majority of individuals with sensory impairments (Chen et al., 1998). With follow-up testing over a two year period, all of the patients increased the peak to peak am plitude between excitatory and inhibitory responses (reflex modulation), alt hough abnormalities were st ill apparent in two patients with motorically complete recovery. Na dler et al. found opposite results in a two year longitudinal study, demonstrating that reflex amplitude did not change over time, and found exaggerated reflexes in those th at demonstrated the least motori c recovery (Nadler et al., 2004). To date, the relationship between lower extrem ity cutaneous reflexes and functional and/or motoric recovery has not been investigated. Furthermore, examination of interlimb respons es in those with post-stroke hemiparesis may provide insight into the functionality of in terneuronal circuitry. In studying the interlimb cutaneous reflexes of the contra lateral limb during walking in hea lthy controls, Haridas and Zehr demonstrated phase dependent re gulation and reflex reversals th at were significantly correlated
73 with kinematic changes when stimulating the su perficial peroneal nerv e (Haridas and Zehr, 2003). Interlimb coordination as measured with ot her means, however, is impaired in those with stroke. Movement amplitudes and cycle speed ar e decreased in tasks i nvolving all four limbs compared to unilateral tasks in individuals post-st roke, and no facilitation of the impaired limb was found in bilateral tasks (Garry et al., 2005). In addition, Ka utz and Patten illustrated that individuals post-stroke performi ng a leg pedaling task increased paretic leg EMG deficits when the non-paretic leg was activated (Kautz a nd Patten, 2005). The suppression of interlimb influences seen in controls and those post-st roke during static and discrete tasks was not demonstrated in the rhythmical leg cycling ta sks post-stroke. The use of cutaneous nerve stimulation may further elucidate the impact of stroke on interlimb coordination and provide information regarding central reorganization and plasticity associated with activity-dependent locomotor retraining. The purposes of this study are to: 1) examin e the relationship between cutaneous reflex modulation and measures of behavioral recovery and functional performance in individuals poststroke; and 2) investigate post-st roke interlimb reflex response s by examining cutaneous reflex modulation in both legs when stimuli are applie d separately to the nonparetic leg and paretic leg. We hypothesized that individuals post-st roke will demonstrate decreased modulation of cutaneous reflexes and that the degree of modul ation will positively correla te with measures of locomotor function. In addition, we hypothesized th at individuals post-str oke will demonstrate decreased modulation of cutaneous reflexes and decreased magnitude of responses in the paretic leg during stimulation of the non-paretic lower ex tremity than during stimulation of the paretic lower extremity.
74 Methods Participants Fourteen individuals with ch ronic stroke (greater than six months post-stroke) were recruited for this study at the Gainesville, FL Department of Veterans Affairs (VA) Medical Center. Participants have a history of a singl e unilateral stroke, are am bulatory without contact assistance, are able to follow a multiple st ep command, and have no other medical issues interfering with their ability to walk. All par ticipants signed written informed consent approved by the University of Florida Institutional Revi ew Board and Gainesville VA Subcommittee for Clinical Investigation. Each participant underwent a walking asse ssment of the following measures illustrating a measure of restitution of the walking pattern (paretic propulsion), compensation of the walking pattern (walking speed), and the clinic al analog of paretic propulsion (paretic step ratio). Participants included nine with left hemiparesi s and ten males with a mean chronicity of 51.5 + 46.2 months (range 13 to 163 months), a m ean walking speed of 0.72m + 0.30 m/s (range 0.31 to 1.26 m/s), and a mean stance phase of 68.51 + 4.74% of the gait cycle (range 60.3 to 76.5) (Table 4-1). Walking Assessment Measures Paretic propulsion (Pp). Pp is a quantitative measure of the coordinated output of the paretic leg, describing the contribu tion of the paretic leg in prope lling the center of mass forward during walking; it is defined as the percentage of propulsion performed by the paretic leg (Bowden et al., 2006). As Pp illustrates that many of those with normal or near normal walking speeds continue to exhibit substa ntial motor control deficits, Pp may be an effective tool in distinguishing functional compensation from physi ological restitution. St atistics will be run on the absolute deviation from normal (0.5).
75 Walking speed. Self-selected walking speed overground will be measured on an instrumented walkway (GAITRite, CIR Systems, Inc, Havertown, PA). Self-selected walking speed has proven to be an important measure of str oke recovery because it is simple to measure, reflects both functional and physiological change s (Perry et al., 1995; Richards et al., 1995), remains reliable and sensitive to change even as recovery advances (Richards et al., 1995), and is a predictor of health status (Studenski et al., 2003). Paretic step ratio (PSR). PSR is defined as the percenta ge of stride length performed by the paretic leg and has been demonstrated to differentiate types of walking impairment poststroke (Balasubramanian et al., 2007). PSR demo nstrates a strong negative correlation (p=-0.78) with Pp, indicating that those with high PSR gene rate very little propulsi on (Balasubramanian et al., 2007). As such, we consider PSR to be a pot ential clinical analog of Pp for those without access to a biomechanics laboratory. Statistics will be run on the absolute deviation from normal, symmetrical value (0.5). Kinematics and Kinetics Each participant completed a 30-second tria l of walking on an instrumented treadmill (Medical Development, Tecmachine Hef, Andrezie ux Boutheon, France) at a self selected speed to collect ground reaction forces (GRF) and ki nematic data using a modified Helen Hayes marker set. A full biomechanical analysis was generated for each partic ipant. GRF data was acquired at 200 Hz and was filtered with a low pa ss fourth order Butterworth filter at 20 Hz forward and backward in time. The A-P GRF co mponent was used for an alysis of the paretic propulsion. Step cycles will be identified by initial ground contact on the force plate on the paretic leg.
76 Electromyography To collect electromyographic (EMG) data, the skin was shaved, abraded and cleaned with alcohol before attaching bipolar Ag-AgCl surface electrodes (Neu rolog, Digitimer Ltd., Hertfordshire, England). EMG recordings were co llected from four different muscles bilaterally: tibialis anterior (TA); soleus (SOL); rectus fe moris (RF); and biceps femoris (BF). Data were collected at 1000 Hz and filtered with a 10 Hz hi gh pass filter and 500 Hz low pass filter. Gains were variable from 1000 to 20,000 Hz depending on individual muscle signals. Reference electrodes will be placed over the electrically neutral patella of each leg. Nerve Stimulation Bilateral superficial peroneal (SP) nerves we re stimulated independently on the paretic and non-paretic anterior foot/ankle usi ng a bipolar configuration of fl exible disposable 1cm Ag/AgCl electrodes (Vermed, Inc., Bellows Falls, VT). Stimul ations were provided in trains of five x 1.0 ms pulses at 300 Hz using an isolated constant current stimulation (Grass S88 stimulator with Grass SIU5 and CCU1 isolation and constant current units). The intensity of the stimulation was 2.0-2.5 times the radiating threshold of each foot, not to exceed the level of noxious stimulation (Zehr et al., 1997). Stimulation should create a non-painful stimulus radi ating on the dorsum of the foot into the distal second and third phalanx to assure cutaneous reflex responses and not pure flexor withdrawal responses that would be seen with a painful response. On-line assessments will be conducted to assure achieveme nt of a reflex response during sitting with a sub-maximal isometric contraction of the TA on the stimulated leg using customized Matlab programming (the Mathworks, Natick, MA). Each participant walked at their sel f-selected speed on a treadmill while applying stimulations to the SP nerve at a randomized frequency ranging fr om three to five seconds to assure that stimulations will not occur during co nsecutive steps. Each individual walked until
77 100 stimulations were applied, allowing for rest breaks as necessary if individuals were not capable of completing 100 stimulations consecutivel y. Data collection resu lted in approximately 10-15 minutes of total walking. Data Processing EMG data was collected at 1000 Hz with a 12bit A/D converter (National Instruments, BNC 2090) connected to a PC running customi zed LabVIEW virtual instruments (National Instruments, Austin, TX). Offline analysis using customized Matlab programming (the Mathworks, Natick, MA) averaged all of the st ep trials, dividing them into four equal proportions of the step cycle begi nning with foot strike of the pa retic leg. Each phase of the cycle included approximately 20 -25 sweeps during which the stimul us was applied. Data were collected for each stimulation from 100 ms prestimulus to 200 ms post stimulus. The raw EMG signals will be rectified and filtered with a low-pass 3rd order Butterworth filter. For each phase of the gait cycle, EMG signals from the non-stimulated sweeps will be subtracted from the average of the sweeps containi ng stimulations to yield the subtracted reflex trace, from which net reflex response were calcul ated. A significant refl ex response was defined as exceeding two standard deviations above or below the non-stimulated EMG. Analyses were conducted on the middle (~75-125 ms) latency respons es for the following variables: 1) peak amplitude normalized to the peak subtracted EM G for the middle latency period will be used for subsequent analyses; and 2) reflex modulation for each muscle was defined as the difference between a significant positive and negative reflex. Baseline EMG signals were compared to EM G obtained post-stimulation and the Average Cumulative Reflex EMG after 150 ms (ACRE150) was calculated to determine the net EMG reflex effects of the stimulation. For this meas ure, the EMG value at 150 ms post stimulation is divided by the time interval of integration in or der to quantify a summary reflex response (Zehr
78 et al., 1998). The ACRE150 was normalized to the peak undi sturbed (non-stimulated) EMG signal at 150 ms post stimulation for each muscle. Detailed descriptions of phase averaging and subtracting baseline EMG signals have been presented elsewhere in the litera ture (Haridas and Zehr, 2003). Statistical Analysis Correlations between walking pe rformance measures and 1) reflex modulation; and 2) ACRE150 were completed using Pearsons Correl ation Coefficient. Comparison of reflex responses between stimulation of the paretic and non-paretic legs were completed using a pairedsample t-test. Results Reflex Responses and Walking Performance Measures When comparing the magnitude of reflex modul ation in the paretic le g with measures of walking performance, Pp deviation only signifi cantly correlated with SOL modulation when the non-paretic leg was stimulated (r=-0.581, r=0.037, Table 4-2). Se lf-selected walking speed was significantly correlated with TA modulation when the non-pa retic leg was stimulated (r=-0.709, p=0.007, Table 4-2). PSR was not significantly co rrelated with the reflex modulation of any muscle, and none of the reflex modulations re sulting from stimulation of the paretic leg significantly correlated with any of the walking measures. Th ere were no significant correlations between the ACRE150 and any of the walking performance measures. The association between Pp deviation and th e SOL modulation is c onsistent with the hypothesis that high degrees of walking impairment (large Pp deviation values) would be associated with smaller (shallow) modulation (F igure 4-1). The relationship between walking speed and TA modulation, however, represents a strong correlation in the direction opposite of
79 that hypothesized, with those wa lking most quickly demonstrat ing the shallowest modulation (Figure 4-2). Phase Dependent Reflex Modulation When responses from all subjects were aver aged together, the pare tic TA showed strong inhibitory responses, regardless of which leg was stimulated (Figure 4-3). In contrast, the paretic SOL demonstrated inhibition with paretic leg stimulation and excitation with non-paretic leg stimulation, and both of these responses were stro nger in bins 1 and 2 (stance phase of the gait cycle) than in bins 3 and 4. RF and BF demonstr ated more variable responses, with peak reflex activity occurring in bins 1 and 4 for the RF. In the TA, the ACRE150 was positive for all four bins when stimulating the non-paretic leg and ne gative for paretic leg stimulating (Figure 4-3). Similar responses were seen in the SOL, with peaks occurring during stance (bins 1 and 2), while peaks for the TA occurred during the stance to swing transition (bin 3). Differences between stimulating sides were less ev ident for both the RF and BF. In order to further examine the contributors to the reflex modulation discussed above, each of the four bins for each muscle and each participant was examined for stimulation of the paretic leg (Table 4-3) and the non-pare tic leg (Table 4-4). There was a high degree of variability with the stimulation of each leg, with very few cons istent patterns emerging. When stimulating the paretic leg, TA bins 3 and 4 as well as SOL bins 1 and 2 were predominantly inhibitory (highlighted red). Conversel y, when stimulating the non-pare tic leg, SOL bins 1 and 2 are predominately excitatory (highlighted red). Thes e patterns were defined as having at least half the participants demonstrating significant reflexes in these bins with one or zero occurring in the opposite direction from the predominate pattern In summary, the non-paretic leg stimulation produced reflex responses that were 70.59% excita tory, while the paretic leg stimulation elicited reflexes that were overall 68.07% inhibitory. Furthermore, stimul ation of the paretic leg yielded
80 23 reflex reversals from a posit ive or negative significant refl ex to one of the opposite sign within the same muscle, including five participants who demonstrated more than one reversal per cycle. In contrast, stimulation of the non-paretic leg elicited only 11 significant reflex reversals, with only one participant demonstr ating more than one reversal in a muscle within a gait cycle. Differences between Paretic and Non-Paretic Stimulation When reflex responses were analyzed to compare stimulation of the paretic leg to stimulation of the non-paretic leg, all of the sign ificant differences occurred in the TA and SOL (Table 4-5). Significant diffe rences were noted for the TA bin 3 (P=0.025), TA modulation (0.013), SOL bin 1 (p=0.001), SOL bin 2 (p=0.001) and SOL bin 3 (p=0.009). This may be limited, however, to the variability of the reflex response as evidenced by the excessively large standard deviations relative to the mean in Table 4-5. Discussion The primary hypothesis of this experiment, th at reflex modulation would correlate with measures of walking performance, is only ma rginally supported by the results. The SOL modulation resulting from non-pare tic stimulation demonstrated a significant correlation with Pp deviation (r=-0.581, p=0.037) and th e TA modulation resulting fr om non-paretic stimulation yielded a significant co rrelation with speed (r =-0.709, p=0.007) (Table 4-2). These relationships, however, occur in the opposite di rection of each other, with SOL modulation increasing with decreased Pp deviation (closer to symmetrical or normal propulsion) and speed increasing as TA modulation decreases. While no other correlations with reflex modulation were significant (including none with paretic le g stimulation), these results should not be overstated, but the completely opposite relationship between speed a nd Pp deviation is intriguing. As Pp is purported to be a measure of motor coordination impairment and speed is at least partially a compensatory behavior (Bowden et al., 2006), it is notable that our physiological measure of
81 spinal level modulation demonstrates the more pred ictive pattern with a recovery based measure. As Pp and speed have been shown to correlate significantly in the past (Bowden et al., 2006) and approach significance in this sample (r=-0.518, p= 0.058), these correlations may have very little to do with the mechanisms of how each meas ure is produced. The results of the current experiment are far from definitive and more research would need to be performed to address specifically the contribution of spinal reflex modulation to coordinative and compensatory measures of walking performance. The minimal number of significant correlations between reflex modulation and walking performance measures was surprising and was pe rhaps due to the variability of the reflex responses in this sample. As evidenced in Tables 4-3 and 4-4, there is an absence of consistent significant reflex responses across patients and across bins. This randomness of response is particularly prevalent in the proximal musc ulature (RF and BF) with no specific pattern emerging from stimulation either to the pare tic or non-paretic leg. The only patterns that emerged were from the distal musculature (TA and SOL), particularly TA3 and 4 when stimulating the paretic leg and SOL1 and 2 when s timulating either leg. In previous studies of individuals post stroke, co-activa tion patterns between the BF and RF are shown to be relatively invariant throughout recovery and ar e not responsive to gait interv entions as applied in the study (Den Otter et al., 2006). As suc h, it may be that post stroke these muscles are less susceptible to reflex modulation due to the compensatory de mands of hemiparetic walking. Conversely, the distal musculature may be more amenable to stimulation influences, although the functional relevance is not clear in the current investig ation. Specifically, pare tic stimulation yields inhibitory responses in the SOL during stan ce when the task requirements would indicate excitation would be required for improved stability of the stance phase. In addition, paretic leg
82 stimulation produces predominately inhibitory re sponses in the TA during terminal stance and swing when responses are expected to be excitatory to facilitate to e clearance. These deviations imply a strong irregularity in th e activation of appropriate spin al interneuronal pools to produce the functionally suitable responses. The difference between paretic and non-paretic leg stimulation was statistically significant for TA3, TA modulation, and SOL1-3 (Table 45). These differences may be due to the variation in activation of spinal level excitatory and inhibitory interneurons. Previous work examining interlimb coordination has speculated that this influence may be partially due rhythmic activity at the level of the spinal co rd (Ferris et al., 2004; Kawashima et al., 2005). Additionally, Kautz and Patten sp eculated that the corticoreticular-reticulospinal-spinal interneuronal system may play a major role in bilateral coordination a nd specifically that the non-paretic leg can contribute greatly to the cont rol of the paretic locomotor pattern via these interneuronal mechanisms (Kautz and Patten, 2005) Alternatively, mechanical positioning of the stimulated limb may also prov ide a substantial sensory input into the activation of interlimb coordination effects. In the present study, stim ulation of the paretic and non-paretic legs occurred anti-phase to each other as the reco rded output was always dependent on the position of the paretic leg. This mechanical influence ca nnot account for the lack of consistent reflex responses observed in the proximal musculature but may account for some of the phase specific differences seen in TA and SOL. Variability within ambulation patterns poststroke likely contribu te greatly to the variability in response observed in this study. Walking post-stroke is mo re variable than in healthy controls for step length, swing, pre-swing and stride times; paretic swing time variability is greater than that in the non-pa retic leg; and additional interlimb differences were observed for
83 those with more impaired gait patterns (Bal asubramanian et al., 2009). In addition, EMG patterns in healthy controls demons trate a considerable amount of variability, particularly in the proximal musculature (Winter and Yack, 1987) and it is likely that EMG patterns post-stroke present even greater variance in responses. Furthermore, animal models have demonstrated that intersubject variability is inconsistent, dem onstrating varying degrees of locomotor and cutaneous reflex variance in cat hindlimbs, representing a func tional heterogeneity in normal walking (Loeb, 1993). The above aspects of variab ility likely contribute in varying degrees to the lack of consistent responses to cutaneous s timulation observed in the current study. As the sample included many with a high degree of wa lking impairment (as determined by speed and Pp profiles), these results reflect a significant cha llenge in applying cutaneous reflexes studies to a broad spectrum of the hemiparetic population. The current study is limited by the absence of exact biomechanical positions of the nonparetic limb as the gait observations were tri ggered by landing of the foot on the forceplate identified with the paretic side. Additional tri ggers for toe off on the paretic side as well as triggers for non-paretic initial co ntact and toe off would be require d to correctly describe all of the bilateral gait ev ents. This methodology, however, would re quire twice as many gait events to get an adequate number of stimulations in each bin. As such, the respondent burden may have been excessive for many of the pa rticipants in the present study as the gait performance abilities ranged from fully independent to househol d ambulators as defined by Perrys speed classification standards (Perry et al., 1995b). A dditionally, variability of the reflex responses required collapse into four bins containing 20-2 5 stimulations per bin in order to observe adequate reflex activity for data analysis. This collapse further impairs the ability to describe specific gait events, particularly in bin3 which contained a mixture of stance, pre-swing and
84 swing events leading to a likely averaging of exc itatory and inhibitory re flex responses within a bin. This is particularly true of the present sample in which the percentage of the gait cycle encompassing stance phase ranged from 60.3% to 76.5% (subject 10, for whom toe off actually occurred in bin4). Expansion to a larger number of gait bins again would have necessitated proportionately more step cycles to assure an ad equate number of stimulated steps in each bin. In addition to increasing the number of gait ev ent markers, understa nding the relationship between cutaneous reflex responses and kinetic events may add clar ity the present analysis. Full kinetic evaluations to determine ankle and hip po wers throughout the gait cycle were collected but analysis was not included in the present stud y. Additionally, it may be helpful to relate the ability to differentiate comple x motor control to the ability to modulate cutaneous reflex responses. Non-negative matrix factorization of EMG walking patterns ma y assist in defining the differentiation of motor activ ity and assist in delineating t hose with normalized responses from those with a greater degree of variance. Ag ain, data have been collected for this analysis, but are not included in the present study. Lastly, therapeutic programs that target the available neuroplasticity within the central nervous system may incr ease the amplitude of the reflex response, increase the modulati on between positive and negative significant reflexes, and increase the sensitivity of the measure by reducing the amount of variability in the responses. Current evaluations are ongoing to assess the spinal cutaneous re flexes pre and post a locomotor training intervention in order to examine the pote ntial for using cutaneous reflex studies as a marker for neuroplastic cha nges in those post stroke.
85 Table 4-1. Subject demographics SUBJECT PARETIC SIDE GENDER CHRONICITY (months) SPEED (m/s) % STANCE 1 Left Male 14 0.31 65.2 2 Right Female 60 0.63 73.5 3 Left Male 14 0.75 67.7 4 Left Male 43 0.97 72.6 5 Left Male 94 1 70.8 6 Right Male 163 0.49 72.9 7 Right Female 13 0.33 65.4 8 Left Male 28 0.44 67.3 9 Right Male 41 1.11 64.8 10 Left Female 30 0.61 76.5 11 Left Male 18 0.63 65.7 12 Left Male 130 1.26 73.2 13 Left Male 26 0.49 60.3 14 Right Female 47 1 63.2 9L, 5R 10M, 4F 51.5 0.72 68.51 Table 4-2. Correlations between reflex modulation and walking parameters Pp deviation SPEED PSR NP stim TA_mod r=-0.004 p=0.991 r= -0.709 p= 0.007 r=0.420 p=0.153 NP stim SOL_mod r= -0.581 p= 0.037 r=-0.083 p=0.788 r=-0.205 p=0.501 NP stim RF_mod r=-0.325 p=0.279 r=0.353 p=0.236 r=-0.423 p=0.149 NP stim BF_mod r=-0.314 p=0.297 r=0.349 p=0.242 r=-0.093 p=0.763 P stim TA_mod r=0.340 p=0.234 r=-0.387 p=0.172 r=-0.038 p=0.897 P stim SOL_mod r=-0.078 p=0.792 r=-0.215 p=0.460 r=-0.021 p=0.944 P stim RF_mod r=0.291 p=0.313 r=-0.393 p=0.165 r=0.077 p=0.793 P stim BF_mod r=0.282 p=0.328 r=-0.175 p=0.549 r=-0.166 p=0.572
86 Figure 4-1. Relationship between soleus modulation and paretic propulsion deviation (stimulating the non-paretic leg). Increased modulation is significantly associated with smaller Pp deviations from symmetry. r=-0.581 p=0.037
87 Figure 4-2. Relationship between tibialis ante rior modulation and walking speed (stimulating the non-paretic leg). Increased modulation is significantly associated with faster walking speeds. r=-0.709 p =0.007
88 Figure 4-3. Normalized middl e latency reflex and background EMG (Left Column), and ACRE150 (Right Column) for targeted muscle s. Blue bars and lines indicate stimulation of the paretic leg, while orange bars and lines indicate stimulation of the non paretic leg. Note the discrepancy in re sponses between stimulation sides for both the individual reflexes amplitudes a nd the cumulative reflex response (ACRE150). TA: ACRE150-25.00 -20.00 -15.00 -10.00 -5.00 0.00 5.00 10.00 Paretic Stimulation Non Paretic Stimulation SOL: ACRE150-25.00 -20.00 -15.00 -10.00 -5.00 0.00 5.00 10.00 Paretic Stimulation Non Paretic Stimulation RF: ACRE150-25.00 -20.00 -15.00 -10.00 -5.00 0.00 5.00 10.00 Paretic Stimulation Non Paretic Stimulation BF: ACRE150-25.00 -20.00 -15.00 -10.00 -5.00 0.00 5.00 10.00 Paretic Stimulation Non Paretic Stimulation TA-60.00 -40.00 -20.00 0.00 20.00 40.00 60.00 80.00 100.00 Paretic Stimulation Non Paretic Stimulation Background EMG SOL-60.00 -40.00 -20.00 0.00 20.00 40.00 60.00 80.00 100.00 Paretic Stimulation Non Paretic Stimulation Background EMG RF-60.00 -40.00 -20.00 0.00 20.00 40.00 60.00 80.00 100.00 Paretic Stimulation Non Paretic Stimulation Background EMG BF-60.00 -40.00 -20.00 0.00 20.00 40.00 60.00 80.00 100.00 Paretic Stimulation Non Paretic Stimulation Background EMG
89 Table 4-3. Paretic reflex activity with paretic leg stimulation Tibialis Anterior Soleus Rectus Femoris Biceps Femoris 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 TOTAL + TOTAL Reversals Subj1 + + + + 4 9 1 Subj2 + + + 3 7 4 Subj3 + + + + 4 1 1 Subj4 + + + + 4 4 4 Subj5 + + 2 7 0 Subj6 + + + 3 6 1 Subj7 + + + + + + 6 7 5 Subj8 + + + + 4 10 3 Subj9 0 5 0 Subj10 + + + 3 5 2 Subj11 + + 2 5 1 Subj12 + 1 5 0 Subj13 + 1 4 1 Subj14 + 1 6 0 Total 38 81 23 68.07% inhibitory
90 Table 4-4. Paretic reflex activity with non-paretic leg stimulation Tibialis Anterior Soleus Rectus Femoris Biceps Femoris 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 TOTAL + TOTAL Reversals Subj1 + + + + + + 6 1 1 Subj2 + + + + + 5 1 1 Subj3 + + + + 4 1 1 Subj4 + + + + + + + + + 9 0 0 Subj5 + + + 2 2 1 Subj6 + + + + 4 4 1 Subj7 + + + + + + + 7 1 0 Subj8 + + + + 4 4 3 Subj9 + + + 3 2 0 Subj10 + + + + 4 1 0 Subj11 + + + 3 5 2 Subj12 + + + + 4 3 1 Subj13 Subj14 + + + + + 5 0 0 Total 60 25 11 70.59% excitatory
91 Table 4-5. Comparisons between pareti c and non-paretic st imulation paretic reflex activity Mean Std. Deviation t Sig. Tibialis Anterior bin 1 2.50132.4350.2780.786 Tibialis Anterior bin 2 -11.41323.521-1.7490.106 Tibialis Anterior bin 3 -35.49149.843-2.5670.025 Tibialis Anterior bin 4 -15.91630.253-1.8970.082 Tibialis Anterior Modulation 30.22637.5492.9020.013 Soleus bin 1 -58.41545.066-4.6740.001 Soleus bin 2 -51.22840.804-4.5270.001 Soleus bin 3 -12.56014.482-3.1270.009 Soleus bin 4 -14.85031.367-1.7070.114 Soleus Modulation -5.62624.173-0.8390.418 Rectus Femoris bin 1 -15.61545.011-1.2510.235 Rectus Femoris bin 2 8.50532.7850.9350.368 Rectus Femoris bin 3 1.43922.3140.2320.820 Rectus Femoris bin 4 7.88325.8241.1010.293 Rectus Femoris Modulation 19.80035.1102.0330.065 Biceps Femoris bin 1 -8.75945.003-0.7020.496 Biceps Femoris bin 2 -8.34324.006-1.2530.234 Biceps Femoris bin 3 0.53520.9970.0920.928 Biceps Femoris bin 4 -7.92636.281-0.7880.446 Biceps Femoris Modulation 5.89231.3730.6770.511
92 CHAPTER 5 DISCUSSION Background Recovery post stroke has traditionally been de fined by impairment level clinical scales or observations of functional behavior and has been limited by expectations associated with preconceived notions of the time-dependent patterns of recovery. It is unclear, however, how current clinical scales delineate restitution of pre-morbid mo tor behavior from compensated functional pattern of movements. Walking speed, for example, is an often used clinical measure of stroke recovery because it is simple to measure, reflects both functional and physiological changes (Perry et al., 1995b; Richards et al., 199 5a), remains reliable and sensitive to change even as recovery advances (Richa rds et al., 1995a), and is a predicto r of health status (Studenski et al., 2003). Self-selected walki ng speed has been associated with discrimination of potential for rehabilitation (Richards et al., 1995b), prediction of falls and fear of falling (Maki, 1997), and as functional health in the aging popul ation (Studenski et al., 2003). However, gait speed may be increased by a variety of mechanisms, includi ng but not limited to improved motor activation, increased peripheral strength, improved cardiov ascular capacity, and improved confidence of performance. While examination of walking spee d gives an indication of the functionality and independence of the task, and examination of the task underpinnings gives evidence regarding the composite mechanisms, very few measures exis t to evaluate within ta sk restitution of the behavior. In recent years therapy has shifted from being based predominately on treatment of impairments to being based on task specific activ ities encouraging patients to activate their paretic limbs and not rely on compensatory techni ques. This task specific, or activity based, training approach is largely driven by our current knowledge of neur oplasticity. Previous
93 theoretical frameworks have been dominated by the notion that the central nervous system is hard-wired and incapable of self-repair (Ramon y Cajal, 1928) However, recent technological advances in the domain of measurement have allo wed scientists to disc over that the nervous system has a tremendous ability to adapt both to damage and to behavioral interventions. Examples of these discoveries include the following: 1) different patterns of brain activation via functional imaging studies; 2) mo rphological changes in nerve syna pses; 3) dendritic and axonal growth, contraction, and sprouting; 4) modulati on of neurotransmitters; and 5) modulation of spinal level reflexes (Nudo, 2006). This knowledge of the nervous systems ability to adapt has led to a paradigm shift in the field of neuror ehabilitation which is curr ently under extensive study in an attempt to take advantage of available ad aptability and modulation potential (Behrman et al., 2006; Nadeau, 2002). Task specific interv entions also imply a need for task specific evaluative tools, as the interven tions may not be targeting impair ments per se, but instead target the nervous systems potential ability to restore functional behavior to its pre-morbid state. Currently Defined Patterns of Recovery Currently, descriptions of recovery post str oke are limited to measurement of impairments at the body structure/function leve l and measurements of functiona l behavior at the activity and participation level of the Intern ational Classification of Functi oning, Disability, and Health (ICF Model) (Figure 1-1). Recovery as currently measured increases sharply in the first six weeks after stroke and generally levels off from thr ee to six months when measured by impairment level scales such as the Fugl-Meyer (Duncan et al., 1994; Kollen et al., 20 05). In the large scale Copenhagen Stroke Study, recovery was defined as the ability to discharge home, and 64% of patients were able to discharge to the home envi ronment, with 11% having severe deficits, 11% with moderate deficits, and 78% with mild deficits as determin ed by the locomotor portion of the Barthel Index (Jorgensen et al., 1995a; Jorgensen et al., 1995b; Jorg ensen et al., 1995c). In this
94 study, functional recovery was comp leted within 12.5 weeks in 95% of the patients (Jorgensen et al., 1995a), which may speak stro ngly either to recovery or to the ceiling effect present within the Barthel Index. The best recovery was noted in 8.5 weeks in those w ith the mildest strokes and 20 weeks in those with the most severe. It has been hypothesized that this early functional return is strongly linked to natu ral neurologic restitution of peri -infarct tissue, recovery of diaschisis, and improved neurotra nsmission near and distal to th e infarct site (Kwakkel et al., 2004). This spontaneous neurolog ic recovery has been demonstr ated via regression models to account for 16-42% of the improvements in body stru cture function and activit ies in the first two to three months post stroke (Kwakkel et al., 2006). In contrast to the ordinal clin ical scales and subscales, walk ing speed is shown to reflect changes in functional recovery for up to 12-18 m onths post stroke, reflec ting both increases in cadence and step length (Richards et al., 1992). Kollen later conf irmed these results by showing longitudinal increases in walking speed over the course of a year, with self selected speed increasing from 0.037 m/s to 0.64 m/s (Kollen et al., 2006). Additionally, Kollen demonstrated that the Functional Ambulation Category (FAC) in creased longitudinally over the course of a year, reflecting additional functiona lity associated with the increas e in walking speed (Kollen et al., 2005). However, concurrently collected clin ical scores such as the Fugl-Meyer lower extremity subscore, the Fugl-Meyer balance subsco re, the Letter Cancella tion Task, the Motricity Index, and the Timed Balance Test combined to account for only 18% of the variance in the FAC change, even though each demonstrated significan t correlations with th e FAC (Kollen et al., 2005). These results underscore tw o points: 1) that functional im provement may continue even as impairment level measurement plateaus; and 2) that correlation of clinical findings to functional performance cannot be inte rpreted as a cause for the change.
95 Using more detailed instrumentation, rese archers demonstrated that electromyographic (EMG) measurements may be used to evaluate defi cits in a patients motor control. Using EMG output patterns, Knutsson and Richards identified three types of hemiparetic walking based on dependence on spasticity, amount of voluntary activation, and the amount of co-contraction (Knutsson and Richards, 1979). EMG patterns have b een utilized in longit udinal assessments of patients and demonstrate improvement over a two y ear period that are related to changes in walking speed (Richards et al., 1995a). While bur sting of the triceps su rae (gastrocnemius and soleus) does not appear to change over time, the amplitude of tibialis ante rior (TA) bursting does show longitudinal improvement. Those with fa ster walking speeds demonstrated improved movement profiles associated with increased EMG activity, while those with slower walking speeds maintain more constant profiles (Richard s et al., 1995a). Proximal musculature such as the biceps femoris remains more invariant, although the quadriceps al so improve their EMG activation patterns beginning at approximately one year post stroke (Richa rds et al., 1995a). More recent explorations of longitudinal EMG changes we re conducted in two studies investigating activation patterns resulting from therapeutic interventions in sub-acute populations. Ambulatory independence, walking speed, and mobility incr eased, but patterns of co-activation in the biceps femoris and rectus femoris remained relatively constant up to 10 weeks post stroke, creating consistent asymmetry in the swing phase of the gait pattern (Den Otter et al., 2006). As function improved in the absence of alte red activation profiles, it was interpreted that EMG patterning changes are not necessary for functional gain. A more recent analysis demonstrated similar findings up to 24 weeks post-stroke, showing that surface EMG patterns did not change in spite of significant improvements in walking function (Buurke et al.,
96 2008). Both of these studies in terpret these findings as a sour ce of advocacy for compensatory based approaches to walking rehabilitation as improved muscle activity coordination is not associated with functional recovery. However, both studies utilized a traditional approach to walking rehabilitation, and it remains to be seen if contemporary task specific approaches targeting available central nervous system plasticity will yield similar responses. As the end goal of muscle activation is to produce appropriately sized and timed rotational forces around the hips, knees, and ankles, ki netic analyses of hemiparetic walking are additionally important in understa nding the achievement of impr oved function. Power profiles (Figure 5-1) were collected befo re and two months afte r a task oriented in tervention and the H1 (hip extension during initial st ance), H3 (hip flexion duri ng early swing), and A2 (ankle plantarflexion during late stan ce and pre-swing) (Winter, 1990) all increased after therapy and the A2 and H3 bursts were significantly correlate d with walking speed (Richards et al., 2004). Furthermore, the A2 peak burst improvement was responsible for 25% of the gain in walking. Examining the interlimb coordination effects dem onstrated that pre-trai ning, paretic A2 and H3 accounted for 84% of the variance in walking speed, while post-training the non-paretic H3 replaces the paretic H3 to account for 82% of the post-training va riance in patients that improved from 0.40 m/s to 0.58 m/s as a result of the inte rvention (Richards et al ., 2004). Parvataneni demonstrated a similar interlimb coordination e ffect, illustrating that the non-paretic H1 was equally important as the paretic H1 in account ing for gait speed changes from 0.69 m/s to 0.83 m/s as the result of an intervention (Parvataneni et al., 2007). These results demonstrate two primary findings: 1) that pareti c H1, H3, and A2 and non-paretic H1 and H3 collectively account for changes in walking speed (Parvataneni et al., 2007); and 2) that conventional training programs are effective in training the non-paretic leg as well as the paretic. Activity based
97 therapies focusing on encouraging use of the pare tic leg may demonstrate a different balance of kinetic profiles, and measures need to have the capability of assimilati ng the coordinative effects of power generation. Neurobiological Control of Walking The historical viewpoint of the bodys potentia l to restore walking function emerged from the belief that all movements were originated and controlled within the cerebral cortex. This top-down approach is rooted in the functional neuroanatomy of the brain and its descending tracts. As can be seen in the model of the human motor homunculus (Figure 5-2), the portion of the motor cortex that is dedicated to the lower extremities largely lies deep in the interhemispheric fissure between th e two cerebral hemispheres. The primary source of input into the corticospinal (CS) tract comes from the prim ary motor cortex, or Brodmans area 4 in the pre-central gyrus (Nadeau et al., 2004). Other pr ojections to the CS trac t emerge from the premotor cortex (Brodmans area 6) as well as th e post-central gyrus, or somatosensory cortex (Brodmans areas 3, 2, and 1). In addition to the pre-motor a nd somatosensory cortexes, the primary motor cortex receives input from the pr imary sensory cortex, thalamus, putamen (basal ganglia), dorsal column nuclei, and the cerebellum (Nadeau et al ., 2004). These fibers emerge from the primary motor cortex and travel through the corona radiat a in the periventricualar white matter, through the internal capsu le, cerebral peduncle s, pons, medulla (pyramids) and to the contralateral spinal cord (Nadeau et al., 2004). From there, th e CS synapses with peripheral nerves and produce the observed motor evoked potentials. In contrast, recent advances in the field of locomotor recovery afte r spinal cord injury argue compellingly for a bottom up approach in which much of automatic walking and rhythm generation is controlled at the leve l of the spinal cord. Animal work over the past 30 years has proven that alternating activity o ccurs in antagonistic muscles even in the presence of complete
98 spinalization (Grillner, 1985; Ro ssignol, 2000). The presence of s ub-cortical rhythmical control of walking gave rise to the description of central pattern generators (CPG s). CPGs are thought to be located at the level of the spinal cord and can be the controller of rhythmic patterned behavior such as walking (MacKay-Lyons, 2002) and breat hing (Barlow and Estep, 2006). This spinal network can be activated by des cending mesencephalic and subthalamic locomotor regions in a non-specific way and is capable of recruiting locomotor muscles in a coordinated fashion (Nielsen, 2003). In humans, it was demonstrated as early as 1997 that the lumbosacral spinal cord can coordinate cyclic activity in the lowe r extremities and may be driven supraspinally or peripherally (Harkema et al., 1997). This peripheral sensory signali ng provide cues that enable the human lumbosacral spinal cord to modulate efferent output in a manner that may facilitate the generation of stepping (Harkema et al., 1997). Sensory feedback is thought to be critical in the effective functioning of the CPG (Nielsen, 2003). While purely cortically driven activation may be true of non-patterned movements of the upper extremity and gait events that require adapta tion to steady state patterns, the presence of sub-cortical rhythmical cont rol may necessitate an understa nding of the spinal cords contribution to walking. Huma n locomotion, however, is much more functionally impaired by lesions to the spinal cord than is quadripedal locomo tion, and lesions of the pyramidal tract may have much more devastating effect on humans than other animals (Porter and Lemon, 1993). Based on clinical trials of indi viduals with varying degrees of supraspinal control, those with clinically complete injuries de monstrate severely limited functiona l recovery (Dietz et al., 1995; Wirz et al., 2001), and it may be deduced that cont rol of walking exists w ithin some merger of the top down and bottom up approaches. As Niel sen stated in a recent review of the central control of muscle activit y during walking, it is the task of the whole central nervous system to
99 generate this muscle activity, to ensure that it is optimally coordinated, to ensure that it is adjusted to the immediate environment, and to modify it when required (Nielsen, 2003). Nielsen concludes by saying that there is no reason to suggest that human walking is controlled exclusively by the spin al cord, nor is there a reason to imply that the motor cortex alone is responsible for activation of muscles dur ing walking. Instead, th is activity related to walking must rely on an integrat ion of spinal neuronal circuitry, afferent signals and descending motor commands (Nielsen, 2003). Relationship of Neural Contro l of Walking to Experiments The purpose of the experiment described in Chapter 3 was to test whether the motor impairment measured by the Fugl Meyer is i ndicative of motor dysfunction during walking in adults with post-stroke hemipare sis. Observations in our la boratory have indicated that individuals may not be able to achieve active dorsiflexion du ring a supine, sitting, and/or standing position even though they have detectable bursting of EMG activity during the task. Approximately half of these individuals, however are able not only to in crease the magnitude of tibialis anterior (TA) bursting dur ing the walking cycle, but also are able to achieve near normal dorsiflexion while ambulating. Th ese findings reflect the comple x control of human locomotion described above and may provide some observati onal evidence of the cont rol of walking beyond a pure cortically driven phenomenon. As a re sult, we would infer th at voluntary, discrete activities as performed in the Fugl Meyer may be inadequate to capture the complex motor behavior in walking. In contrast, walking specific measures are required to describe the effect of walking rehabilitation interventi ons on behavioral recovery. One such task specific measure is paretic propul sion (Pp), described in detail in Chapter 2. As described, Pp represents a coordinated outpu t of the paretic leg by representing the paretic legs contribution to the overall anterior gr ound reaction force (propulsion). Subsequent
100 analyses have demonstrated that this mechanism occurs partly due to aberrant EMG activity of the paretic leg as it is associated strongly a nd positively with the plantarflexor activity of the triceps surae and is associated negatively with ac tivity of the rectus femo ris (Turns et al., 2007). This EMG activity demonstrates that those with less severe hemiparesis ac tivate plantarflexors in the late stance phase and paretic pre-swing (whe n propulsion would ideally occur) while those with more severe hemiparesis demonstrate ear lier flexor activity during terminal stance (as opposed to paretic pre-swing and swing when flexor activity shoul d take place). Pp may also be associated with the combina tions of various power bursts associated with the walking cycle. Propulsion of the center of mass (COM) occurs through some combination of the three primary positive power bursts H1, H3, and A2 By themselves, these power bursts may also reflect intralimb and interlimb compen sations. For example, a weak A2 during double limb support may be compensated for by a stronger H1 in the contralateral lead leg, effectively pulling the COM anteriorly. Similarly, in the tr ail leg, a weak A2 burst may be compensated by a stronger H3, effectively to create a pull off in stead of a push off in the trail leg. A pull off of the paretic leg would lead to a stronger ground reaction for ce during contrala teral single leg stance as that would be the only force of contact at that moment. These patterns of compensation are clearly seen in kinetic analyses of walki ng post stroke (Olney et al., 1991). However, examination of these power bursts alone would not demonstrate the compensatory behavior and the overall contribution of the paretic leg. In fact, integration of the power curves yielding work profiles demonstrate that the paretic leg performs 40% of the work during walking regardless of hemiparetic severity (Olney et al., 1991). This is in direct contrast to previous reports of mechanical work in pedaling post stroke, where those with severe hemiparesis produce markedly
101 less mechanical work than healthy controls, and often achieve no positive work from the paretic leg (Kautz and Brown, 1998). Perhaps the most intriguing concept regard ing Pp and compensation emerges from the relationship with walking speed. While the correla tion is significant, an alyses of Pp clearly illustrate that those with normal walking speed s of greater than 0.8 m/s have a deficit of coordinated motor control of less than 25% of total propulsion. Unpublished work from our laboratory has also demonstrated that Pp is sp eed insensitive, and the percentage of propulsion from the paretic leg is independent of the speed at which one walks. This is in contrast to power, which may be specific to the speed at which one walks. As hip fle xor moments and angular velocities both increase with wa lking speed, it should be assume d that powers would increase concordantly. Thus, Pp not only accounts for th e intralimb and interlimb compensations present in a power analysis, it also controls for speed related increases in power. Results from Chapter 4 illustrate that Pp, spec ifically the deviation from the symmetrical value of 50%, is negatively correlated with reflex modulation of the soleus muscle. The presented analysis does not allow for a direct de terminant of the functiona l significance of this correlation, but as the soleus is on e of the main contributors to the A2 power burst, it is of note that the impaired nervous system is able to pr ovide heightened modulati on in those with more coordinated motor output. Conversely, the co mpensatory based measure of walking speed demonstrates a negative correla tion with tibialis anterior modul ation, indicating that those who walk most quickly demonstrate the poorest modulat ion of the primary dorsiflexor of the ankle. Further exploration is required to glean the exact functional significance of these modulation effects.
102 The cutaneous reflex experiment does not prov ide any large degree of understanding as to the role of spinal level mechanisms in the cont rol of hemiparetic walking nor does it illustrate a purely supraspinal mechanism. The results primar ily highlight the degree of variability within the impaired central nervous system and the inability to consistently activate either excitatory or inhibitory spinal interneurons as required. The technique is still promising, however, for the illumination of a less random system as a result of interventions targeting available neuroplasticity of the nervous sy stem. Such activity based therapies may increase the amplitude of the reflex response, incr ease the modulation between posit ive and negative significant reflexes, and increase the sensitivity of the m easure by reducing the amount of variability in the responses. Conclusion Control of walking ability post stroke is a coordinated function relying on multiple elements of the central nervous system, periphera l motor and sensory systems, skeletal muscle function, cardiovascular potential, as well as personal and environm ental factors contributing to self-efficacy. Previous reports of walking recove ry post stroke have examined impairment level contributions to stroke recovery, with recent adva nces in biomechanical contributions in terms of EMG, kinematic, and kinetic profiles demonstr ating mechanistic characteristics of improved walking. Even these mechanistic approaches, ho wever, illustrate substantial intralimb and interlimb coordination effects and fail to adequately assess restitution of walking performance as opposed to functional compensation. These experi ments highlighted one of the historical goldstandards of motor control in detail as it re lates to walking perform ance and found it to be inadequate to describe walking specific restitu tion. We proposed a coordinative measure of walking performance that takes into account th e compensatory limitations of looking at power burst separately, and that arguably provides a single measure to desc ribe restitution of
103 performance specific to gait activity. The interpla y between multiple neurobiological controls of walking is still poorly delineated, and further work is ongoing in the BRRC/University of Florida Human Motor Performance Laboratory and colla borating laboratories to examine how these influences might be addressed with future neuror ehabilitation techniques. The recent onset of novel therapies, including those based on activ ity based neuroplasticity, necessitates that researchers and clinicians not rely on historic al constructs to examine restitution of walking performance so that clinicians may more e ffectively understand intervention results and maximize the capacity of those we treat.
104 Figure 5-1. Power curves for the hip (A) and ankle (B) (From Winter DA. Biomechanics and motor control of human movement. New Yo rk, NT: John Wiley & Sons, Inc., 1990). The positive power bursts H1, H3, and A2 c ontribute to propulsion of the center of mass. A. HIP POWER B. ANKLE POWER
105 Figure 5-2. Motor homunculus (From Nadeau S, Fe rguson T, Valenstein E, Vierck C, Petruska J, Streit W, et al. Medical Neuroscience. Philadelphia: Saunders, 2004). Note the minor size of the lower extremity representation and its location in the interhemispheric fissure.
106 LIST OF REFERENCES Abbruzzese M, Rubino V, Schieppati M. Task-d ependent effects evoked by foot muscle afferents on leg muscle activity in humans. El ectroencephalogr Clin Neurophysiol 1996; 101: 339-48. Aleshinsky SY. An energy 'sources' and 'fracti ons' approach to the mechanical energy expenditure problem--V. The mechanical energy expenditure reduction during motion of the multi-link system. J Biomech 1986; 19.: 311-5. Aniss AM, Diener HC, Hore J, Burke D, Gandevia SC. Reflex activation of muscle spindles in human pretibial muscles during st anding. J Neurophysiol 1990; 64: 671-9. Balasubramanian CK, Bowden MG, Neptune RR, Kautz SA. Relationship between step length asymmetry and walking performance in subjects with chronic hemiparesis. Arch Phys Med Rehabil 2007; 88: 43-9. Balasubramanian CK, Neptune RR, Kautz SA. Variab ility in spatiotemporal step characteristics and its relationship to walking performance post-stroke. Gait Post ure 2009; 29: 408-14. Barbeau H. Locomotor training in neurorehab ilitation: emerging reha bilitation concepts. Neurorehabil Neural Repair 2003a; 17: 3-11. Barbeau H. Locomotor training in neurorehab ilitation: emerging reha bilitation concepts. Neurorehabil Neural Repair 2003b; 17: 3-11. Barbeau H, Nadeau S, Garneau C. Physical determinants, emerging concepts, and training approaches in gait of individuals with spin al cord injury. J Neurotrauma 2006; 23: 571-85. Barbeau H, Rossignol S. Recovery of locomotion after chronic spinaliza tion in the adult cat. Brain Res 1987; 412: 84-95. Barbeau H, Wainberg M, Finch L. Descripti on and application of a system for locomotor rehabilitation. Med Biol Eng Comput 1987; 25: 341-4. Barlow SM, Estep M. Central pattern genera tion and the motor infrastructure for suck, respiration, and speech. J Co mmun Disord 2006; 39: 366-80. Behrman AL, Bowden MG, Nair PM. Neuroplasticity after spinal cord injury and training: an emerging paradigm shift in rehabilitation and walking recovery. Phys Ther 2006; 86: 1406-25. Beltman JG, van der Vliet MR, Sargeant AJ, de Haan A. Metabolic cost of lengthening, isometric and shortening contractions in maxi mally stimulated rat skeletal muscle. Acta Physiol Scand 2004; 182: 179-87.
107 Berg KO, Maki BE, Williams JI, Holliday PJ Wood-Dauphinee SL. Clinical and laboratory measures of postural balance in an elderl y population. Arch Phys Med Rehabil 1992a; 73: 1073-80. Berg KO, Wood-Dauphinee SL, Williams JI, Maki B. Measuring balance in the elderly: validation of an instrument. Can J P ublic Health 1992b; 83 Suppl 2: S7-11. Bizzi E, Cheung VC, d'Avella A, Saltiel P, Tr esch M. Combining modules for movement. Brain Res Rev 2008; 57: 125-33. Bohannon RW, Andrews AW, Smith MB Rehabilitation goals of patie nts with hemiplegia. Int J Rehabil Res 1988; 11: 181-183. Bowden MG, Balasubramanian CK, Behrman AL Kautz SA. Validation of a speed-based classification system using quantitative meas ures of walking performance post-stroke. Neurorehabil Neural Repair In Press. Bowden MG, Balasubramanian CK, Neptune RR, Kautz SA. Anterior-posterior ground reaction forces as a measure of paretic leg contributi on in hemiparetic walking. Stroke 2006; 37: 872-6. Brooks VB. The Neural Basis of Motor Contro l. Oxford: Oxford University Press, 1986. Brown DA, Kautz SA. Speed-dependent reductions of force output in people with poststroke hemiparesis. Physical Therapy 1999; 79: 919-930. Brown GT. The intrinsic factors in the act of progre ssion in the mammals. Proc. Roy. Soc. 1911; 84: 308. Brunnstrom S. Motor testing procedures in he miplegia: based on sequential recovery stages. Phys Ther 1966; 46: 357-75. Burke D, Dickson HG, Skuse NF. Task-dependent changes in the respon ses to low-threshold cutaneous afferent volleys in the human lower limb. J Physiol 1991; 432: 445-58. Buurke JH, Nene AV, Kwakkel G, Erren-Wolters V, Ijzerman MJ, Hermens HJ. Recovery of gait after stroke: what changes? Neuror ehabil Neural Repa ir 2008; 22: 676-83. Chen CC, Chen JT, Wu ZA, Kao KP, Liao KK. Cuta neous reflexes in patients with acute lacunar infarctions. J Neurol Sci 1998; 159: 28-37. Chen CL, Chen HC, Wong MK, Tang FT, Chen RS. Te mporal stride and force analysis of caneassisted gait in people with hemiplegic str oke. Arch Phys Med Rehabil 2001; 82: 43-8. Clark DJ, Subramanian S, Neptune RR, SA K. Fe wer basic activation patterns account for lower extremity EMG during walking in adults post-st roke compared to hea lthy controls. Neural Control of Movement Annual Meeting. Naples, FL, 2008.
108 Cote MP, Gossard JP. Step training-dependent plasticity in spinal cutaneous pathways. J Neurosci 2004; 24: 11317-27. d'Avella A, Saltiel P, Bizzi E. Combinations of muscle synergies in the construction of a natural motor behavior. Nat Neurosci 2003; 6: 300-8. De Quervain IA, Simon SR, Leurgans S, Peas e WS, McAllister D. Gait pattern in the early recovery period after stroke. J Bone Joint Surg Am 1996; 78: 1506-14. De Serres SJ, Yang JF, Patrick SK. Mechanism fo r reflex reversal during walking in human tibialis anterior muscle revealed by single mo tor unit recording. J P hysiol 1995; 488 ( Pt 1): 249-58. Den Otter AR, Geurts AC, Mulder T, Duysens J. Ga it recovery is not associated with changes in the temporal patterning of muscle activity during treadmill walking in patients with poststroke hemiparesis. Clin Neurophysiol 2006; 117: 4-15. Dietz V, Colombo G, Jensen L, Baumgartner L. Lo comotor capacity of spinal cord in paraplegic patients. Ann Neurol 1995; 37: 574-82. Dietz V, Harkema SJ. Locomotor activity in spin al cord-injured persons. J Appl Physiol 2004; 96: 1954-60. Dobkin BH, Firestine A, West M, Saremi K, Woods R. Ankle dorsi flexion as an fMRI paradigm to assay motor control for walking during rehabilitation. Neuroima ge 2004; 23: 370-81. Drew T, Rossignol S. A kinematic and electr omyographic study of cutaneous reflexes evoked from the forelimb of unrestrained wa lking cats. J Neurophysiol 1987; 57: 1160-84. Dromerick AW, Lum PS, Hidler J. Activit y-based therapies. NeuroRx 2006; 3: 428-38. Duncan PW, Goldstein LB, Horner RD, Landsman PB, Samsa GP, Matchar DB. Similar motor recovery of upper and lower extremities after stroke. Stroke 1994; 25: 1181-8. Duncan PW, Propst M, Nelson SG. Reliability of the Fugl-Meyer assessment of sensorimotor recovery following cerebrovascular accident. Phys Ther 1983; 63: 1606-10. Duysens J, Baken BC, Burgers L, Plat FM, den Otter AR, Kremer HP. Cutaneous reflexes from the foot during gait in heredi tary spastic paraparesis. C lin Neurophysiol 2004; 115: 1057-62. Duysens J, Tax AA, Nawijn S, Berger W, Proko p T, Altenmuller E. Gating of sensation and evoked potentials following foot stimulation during human gait. Exp Brain Res 1995; 105: 423-31. Duysens J, Tax AA, Trippel M, Dietz V. Phas e-dependent reversal of reflexly induced movements during human gait. E xp Brain Res 1992; 90: 404-14.
109 Duysens J, Tax AA, Trippel M, Dietz V. Incr eased amplitude of cutaneous reflexes during human running as compared to standing. Brain Res 1993; 613: 230-8. Duysens J, Trippel M, Horstmann GA, Dietz V. Ga ting and reversal of reflexes in ankle muscles during human walking. Exp Brain Res 1990; 82: 351-8. Edgerton VR, Tillakaratne NJ, Bigbee AJ, de Leon RD, Roy RR. Plasticity of the spinal neural circuitry after injury. Annu Rev Neurosci 2004; 27: 145-67. Ferris DP, Gordon KE, Beres-Jones JA, Harkem a SJ. Muscle activation during unilateral stepping occurs in the nonsteppi ng limb of humans with clini cally complete spinal cord injury. Spinal Cord 2004; 42: 14-23. Forssberg H. Stumbling corrective reaction: a phase-dependent compensatory reaction during locomotion. J Neurophysiol 1979; 42: 936-53. Fugl-Meyer AR, Jaasko L, Leyman I, Olsson S, St eglind S. The post-stroke hemiplegic patient. 1. a method for evaluation of physical perfor mance. Scand J Rehabil Med 1975; 7: 13-31. Garry MI, van Steenis RE, Summers JJ. Interl imb coordination following stroke. Hum Mov Sci 2005; 24: 849-64. Goldberg EJ, Kautz SA, Neptune RR. Can treadmill walking be used to assess propulsion generation? J Biomech 2008; 41: 1805-8. Gregory CM, Bowden MG, Jayaraman A, Shah P, Behrman A, Kautz SA, et al. Resistance training and locomotor recovery after incomplete spinal cord injury: a case series. Spinal Cord 2007. Gresham G, Duncan PW, Stason WB. Post Stroke rehabilitation guideli nes technical report. Agency for health care policy a nd research. Rockville, MD, 1995. Grillner S. Neurobiological bases of rhythmic mo tor acts in vertebrates. Science 1985; 228: 143-9. Hamill J, Knutzen KM. Biomechanical Basis of Human Movement. Media, PA: Limppincott Wiliams & Wilkins, 1995. Haridas C, Zehr EP. Coordinated interlimb compensatory responses to electrical stimulation of cutaneous nerves in the hand and foot dur ing walking. J Neurophysiol 2003; 90: 2850-61. Haridas C, Zehr EP, Misiaszek JE. Postural uncertainty leads to dynamic control of cutaneous reflexes from the foot during human walking. Brain Res 2005; 1062: 48-62. Harkema SJ, Hurley SL, Patel UK, Requejo PS, Dobkin BH, Edgerton VR. Human lumbosacral spinal cord interprets loading dur ing stepping. J Neur ophysiol 1997; 77: 797-811.
110 Harris-Love ML, Macko RF, Whitall J, Forrester LW. Improved hemiparetic muscle activation in treadmill versus overground walking. Neurorehabil Neural Repair 2004; 18: 154-60. Hodgson JA, Roy RR, de Leon R, Dobkin B, Edgerton VR. Can the mammalian lumbar spinal cord learn a motor task? Med Sc i Sports Exerc 1994; 26: 1491-7. Ivanenko YP, Poppele RE, Lacquaniti F. Five basic muscle activation patterns account for muscle activity during human locomotion. J Physiol 2004; 556: 267-82. Ivanenko YP, Poppele RE, Lacquaniti F. Motor control programs and walking. Neuroscientist 2006; 12: 339-48. Jette AM. Toward a common language for function, disability, and health. Phys Ther 2006; 86: 726-34. Jones CA, Yang JF. Reflex behavi or during walking in incomplete spinal-cord-injured subjects. Exp Neurol 1994; 128: 239-48. Jorgensen HS, Nakayama H, Raaschou HO, Olsen TS. Recovery of walking function in stroke patients: the Copenhagen St roke Study. Arch Phys Med Rehabil 1995a; 76: 27-32. Jorgensen HS, Nakayama H, Raaschou HO, Vive-L arsen J, Stoier M, Olsen TS. Outcome and time course of recovery in stroke. Part I: Outcome. The Copenhagen Stroke Study. Arch Phys Med Rehabil 1995b; 76: 399-405. Jorgensen HS, Nakayama H, Raaschou HO, Vive-L arsen J, Stoier M, Olsen TS. Outcome and time course of recovery in stroke. Part II: Ti me course of recovery. The Copenhagen Stroke Study. Arch Phys Med Rehabil 1995c; 76: 406-12. Kautz SA, Brown DA. Relationshi ps between timing of muscle excitation and impaired motor performance during cyclical lower extremity movement in post-stroke hemiplegia. Brain 1998; 121: 515-26. Kautz SA, Patten C. Interlimb influences on pare tic leg function in poststroke hemiparesis. J Neurophysiol 2005; 93: 2460-73. Kautz SA, Patten C, Neptune RR, Worthen LC, Ki m CM. Bilateral coordi nation deficits differ with post-stroke motor recovery status (A bstract). Annual Meeting of the Society for Neuroscience. New Orleans, LA, 2003. Kawashima N, Nozaki D, Abe MO, Akai M, Na kazawa K. Alternate leg movement amplifies locomotor-like muscle activity in spinal cord injured persons. J Neurophysiol 2005; 93: 777-85. Knutson LM, Soderberg GL. EMG methodology. In : Craik RL and Oatis CA, editors. Gait analysis: theory and application. St. Loui s, MO: Mosby-Year Book, Inc., 1995: 293-306. Knutsson E. Gait control in hemiparesi s. Scand J Rehabil Med 1981; 13: 101-108.
111 Knutsson E, Richards C. Different types of di sturbed motor control in gait of hemiparetic patients. Brain 1979; 102: 405-30. Kollen B, Kwakkel G, Lindeman E. Hemiplegic ga it after stroke: is measurement of maximum speed required? Arch Phys Med Rehabil 2006; 87: 358-63. Kollen B, van de Port I, Lindeman E, Twisk J, Kwakkel G. Predicting im provement in gait after stroke: a longitudinal prospect ive study. Stroke 2005; 36: 2676-80. Kwakkel G, Kollen B, Lindeman E. Understanding the pattern of functi onal recovery after stroke: facts and theories. Restor Neurol Neurosci 2004; 22: 281-99. Kwakkel G, Kollen B, Twisk J. Impact of time on improvement of outcome after stroke. Stroke 2006; 37: 2348-53. Lamont EV, Zehr EP. Task-speci fic modulation of cutaneous refl exes expressed at functionally relevant gait cycle phases during level and in cline walking and stair climbing. Exp Brain Res 2006; 173: 185-92. Lamontagne A, Malouin F, Ri chards CL. Locomotor-specific measure of spasticity of plantarflexor muscles after stroke. Ar ch Phys Med Rehabil 2001; 82: 1696-704. Levin MF, Kleim JA, Wolf SL. What Do Moto r "Recovery" and "Compe nsation" Mean in Patients Following Stroke? Neur orehabil Neural Repair 2008. Loeb GE. The distal hindlimb musculature of the cat: interanimal variability of locomotor activity and cutaneous reflexes. Exp Brain Res 1993; 96: 125-40. Lovely RG, Gregor RJ, Roy RR, Edgerton VR. Effects of training on the recovery of fullweight-bearing stepping in the adult sp inal cat. Exp Neurol 1986; 92: 421-35. MacKay-Lyons M. Central pattern generation of locomotion: a review of the evidence. Phys Ther 2002; 82: 69-83. Maki BE. Gait changes in older adul ts: predictors of falls or indi cators of fear. J Am Geriatr Soc 1997; 45: 313-20. McCain KJ, Pollo FE, Baum BS, Coleman SC, Ba ker S, Smith PS. Locomotor treadmill training with partial body-weight support before overground gait in adults with acute stroke: a pilot study. Arch Phys Med Rehabil 2008; 89: 684-91. Mulroy S, Gronley J, Weiss W, Newsam C, Perry J. Use of cluster an alysis for gait pattern classification of patients in th e early and late recovery phase s following stroke. Gait Posture 2003; 18: 114-25. Nadeau S, Arsenault AB, Gravel D, Bourbonnais D. Analysis of the clinic al factors determining natural and maximal gait speeds in adults with a stroke. Am J Phys Med Rehabil 1999; 78: 123-30.
112 Nadeau S, Ferguson T, Valenstein E, Vierck C, Petruska J, Streit W, et al. Medical Neuroscience. Philadelphia: Saunders, 2004. Nadeau SE. A paradigm shift in neuroreh abilitation. Lancet Neurol 2002; 1: 126-30. Nadler MA, Harrison LM, Stephens JA. Cutaneom uscular reflexes following stroke: a 2-year longitudinal study. J Neurol Sci 2004; 217: 195-203. Nagi S. Some conceptual issues in disability and rehabilitation. In: Sussman M, editor. Sociotogy and Rehabititation. Washington, DC: American Sociological Association, 1965: 100-113. Nakayama H, Jorgensen HS, Raaschou HO, Olsen TS. Recovery of upper extremity function in stroke patients: the Copenhagen Stroke St udy. Arch Phys Med Rehabil 1994; 75: 394-8. Neckel N, Pelliccio M, Nichols D, Hidler J. Quantification of functional weakness and abnormal synergy patterns in the lower limb of individua ls with chronic stroke. J Neuroeng Rehabil 2006; 3: 17. Neptune RR, Kautz SA, Zajac FE. Contributions of the individual ankle plantar flexors to support, forward progression and swing ini tiation during walking. J Biomech 2001; 34: 1387-98. Neptune RR, Zajac FE, Kautz SA. Muscle fo rce redistributes segmental power for body progression during walking. Gait Posture 2003. Nielsen J, Petersen N, Fedirchuk B. Evidence su ggesting a transcortical pathway from cutaneous foot afferents to tibialis anterior motoneur ones in man. J Physiol 1997; 501 (Pt 2): 473-84. Nielsen JB. How we walk: central control of muscle activity during human walking. Neuroscientist 2003; 9: 195-204. Nudo RJ. Plasticity. NeuroRx 2006; 3: 420-7. Olney SJ, Griffin MP, Monga TN, Mc Bride ID. Work and power in ga it of stroke patients. Arch Phys Med Rehabil 1991; 72: 309-14. Parvataneni K, Olney SJ, Brouwer B. Changes in muscle group work associated with changes in gait speed of persons with stroke. Clin Biomech (Bristol, Avon) 2007; 22: 813-20. Perry J, Garrett M, Gronley JK, Mulroy SJ. Clas sification of walking ha ndicap in the stroke population. Stroke 1995a; 26: 982-989. Perry J, Garrett M, Gronley JK, Mulroy SJ. Clas sification of walking ha ndicap in the stroke population. Stroke 1995b; 26: 982-9. Porter R, Lemon R. Corticospinal function and voluntary movement. New York: Oxford University Press, 1993.
113 Ramon y Cajal S. Degeneration and Regenerati on of the Nervous System. London: Oxford University Press, 1928. Richards C, Malouin F, Dumas F, Tardiff D. Gait velocity as an outcome measure of locomotor recovery after stroke. In: Craik R and Oatis C, ed itors. Gait analysis: theory and applications. St. Louis: Mosby, 1995a: 355-364. Richards C, Malouin F, Wood-Daup hinee S. Gait velocity as an outcome measure of locomotor recovery after stroke. In: Craik R and Oatis C, editors. Gait Analysis: Theory and Applications. St Louis: Mosby, 1995b: 355-64. Richards CL, Malouin F, Bravo G, Dumas F, Wood-Dauphinee S. The role of technology in task-oriented training in persons with subacute stroke: a ra ndomized controlled trial. Neurorehabil Neural Repair 2004; 18: 199-211. Richards CL, Malouin F, Dumas F, Tardiff D. Ga it velocity as an outcome measure of locomotor recovery after stroke. In: Oatis C, editor. Gait analysis: theory and applications. St. Louis: Mosby, 1995c: 355-364. Richards CL, Malouin F, Dumas F, Wood-Dauphinee S. The relatinsh ip of gait speed to clinical measures of function and muscle activations during recover post-stroke. Proceeding of the North AmericanCongress of Biomechanics, 1992: 299-302. Rossi A, Zalaffi A, Decchi B. Interaction of nociceptive and non-nocicepti ve cutaneous afferents from foot sole in common reflex pathways to tibialis anterior motoneur ones in humans. Brain Res 1996; 714: 76-86. Rossignol S. Locomotion and its recovery after spinal injury. Curr Opin Neurobiol 2000; 10: 708-16. Sanford J, Moreland J, Swanson LR, Stratford PW, Gowland C. Re liability of the Fugl-Meyer assessment for testing motor performance in pa tients following stroke. Phys Ther 1993; 73: 447-54. Sherrington CS. The Integrativ e Action of the Nervous System. Oxford: Oxford University Press, 1906. Shumway-Cook A, Woollacott M. Motor Contro l: Theory and Practical Applications. Philadelphia: Lippincott Williams and Wilkins, 2001. Stein RB. Reflex modulation during locomotion: functioinal significance. In: Patla AE, editor. Adaptability of Human Gait. Am sterdam: Elsevier Science, 1991. Studenski S, Perera S, Wallace D, Chandler JM, Duncan PW, Rooney E, et al. Physical performance measures in the clinical setting. J Am Geriatr Soc 2003; 51: 314-22. Studenski S, Wallace D, Chandler J, Duncan PW, E R, M F, et al. Gait speed as a clinical vital sign in the care of older adults. J.Am.Geriatr.Soc. 2002.
114 Sullivan K, Klassen T, Mulroy S. Combined ta sk-specific training and strengthening effects on locomotor recovery post-stroke: a case study. J Neurol Phys Ther 2006; 30: 130-41. Tax AA, Van Wezel BM, Dietz V. Bipedal reflex coordination to tactile stimulation of the sural nerve during human running. J Neurophysiol 1995; 73: 1947-64. Teixeira-Salmela LF, Olney SJ, Nadeau S, Br ouwer B. Muscle strengthening and physical conditioning to reduce impairment and disability in chronic str oke survivors. Arch Phys Med Rehabil 1999; 80: 1211-8. Tesio L, Rota V. Gait analysis on split-belt forc e treadmills: validation of an instrument. Am J Phys Med Rehabil 2008; 87: 515-26. Ting LH, Macpherson JM. A limited set of muscle synergies for force control during a postural task. J Neurophysiol 2005; 93: 609-13. Torres-Oviedo G, Ting LH. Muscle synergies characterizing human postural responses. J Neurophysiol 2007; 98: 2144-56. Trimble MH, Behrman AL, Flynn SM, Thigpen MT Thompson FJ. Acute effects of locomotor training on overground walking speed and Hreflex modulation in individuals with incomplete spinal cord injury. J Spinal Cord Med 2001; 24: 74-80. Trimble MH, Kukulka CG, Behrman AL. The eff ect of treadmill gait training on low-frequency depression of the soleus H-reflex: comparis on of a spinal cord injured man to normal subjects. Neurosci Lett 1998; 246: 186-8. Turns LJ, Neptune RR, Kautz SA. Relationships between muscle activity and anteroposterior ground reaction forces in hemiparetic walki ng. Arch Phys Med Rehabil 2007; 88: 1127-35. Twitchell TE. The restoration of motor functi on following hemiplegia in man. Brain 1951; 74: 443-80. van Wijck FM, Pandyan AD, Johnson GR, Barnes MP Assessing motor deficits in neurological rehabilitation: patterns of instrument usage. Neurorehabil Neural Repair 2001; 15: 23-30. Wade DT. Measurement in neurol ogical rehabilitation. Oxford: Oxford University Press, 1992. Wang CH, Hsieh CL, Dai MH, Chen CH, Lai YF. Inte r-rater reliability and validity of the stroke rehabilitation assessment of movement (str eam) instrument. J Rehabil Med 2002; 34: 20-4. WHO. International Classification of Functi oning, Disability and Health: ICF. Geneva, Switzerland: World Health Organization, 2001. Winter DA. Biomechanics and motor control of human movement. New York, NT: John Wiley & Sons, Inc., 1990.
115 Winter DA, Yack HJ. EMG profiles during normal human walking: stride -to-stride and intersubject variability. Electroencephalogr Clin Neurophysiol 1987; 67: 402-11. Wirz M, Colombo G, Dietz V. Long term effect s of locomotor training in spinal humans. J Neurol Neurosurg Psychiatry 2001; 71: 93-6. Wolpaw JR. Spinal cord plastici ty in acquisition and maintenance of motor skills. Acta Physiol (Oxf) 2007; 189: 155-69. Worthen LC, Kim CM, Kautz SA Lew HL, Kiratli BJ, Beaupre GS. Key characteristics of walking correlate with bone density in indivi duals with chronic str oke. J Rehabil Res Dev 2005; 42: 761-8. Yang JF, Gorassini M. Spinal and brain control of human walking: implications for retraining of walking. Neuroscienti st 2006; 12: 379-89. Yang JF, Stein RB. Phase-dependent reflex reve rsal in human leg muscles during walking. J Neurophysiol 1990; 63: 1109-17. Zehr EP. Neural control of rhythmic human movement: the common core hypothesis. Exerc Sport Sci Rev 2005; 33: 54-60. Zehr EP. Training-induced adaptive plasticity in human somatose nsory reflex pathways. J Appl Physiol 2006; 101: 1783-94. Zehr EP, Fujita K, Stein RB. Reflexes from th e superficial peroneal nerve during walking in stroke subjects. J Neurophysiol 1998; 79: 848-58. Zehr EP, Komiyama T, Stein RB. Cutaneous refl exes during human gait: electromyographic and kinematic responses to electrical stim ulation. J Neurophysiol 1997; 77: 3311-25. Zehr EP, Stein RB. What functions do reflexes serve during human loco motion? Prog Neurobiol 1999; 58: 185-205. Zehr PE. Considerations for use of the Hoffmann re flex in exercise studies. Eur J Appl Physiol 2002; 86: 455-68.
116 BIOGRAPHICAL SKETCH Mark Bowden graduated from Salisbury High School in Salisbury, NC in 1987. He attended Duke University in Durham, NC from 1987 to 1991, graduating with a B.S. in psychology. After graduation, Mark worked as a research assistant in the Veterans Affairs Medical Center in Durham, NC. In 1993, Mark returned to Duke University in the masters program in physical therapy, gradua ting with an M.S. in 1995. Mark worked for three years as a staff physical therapist at the Ch arlotte Institute of Rehabilitation in Charlotte, NC, leaving in April, 1998 to become Therapy Manager of the Spinal Cord Injury Program at Methodist Rehabilitation Center in Jackson, MS. He left clinical practice in September 2002 to take the role of Research Physical Therapist at the Brai n Rehabilitation Research Center in Gainesville, FL, and enrolled in the rehabilita tion science doctoral program at the University of Florida in January 2004, where he graduated in May 2009 w ith a Ph.D. in rehabi litation science and a concentration in movement dysfunction.