Citation
Reliability and utility of transcranial magnetic stimulation to assess activity-dependent plasticity in human stroke

Material Information

Title:
Reliability and utility of transcranial magnetic stimulation to assess activity-dependent plasticity in human stroke
Creator:
Malcolm, Matthew Paul
Publication Date:
Language:
English
Physical Description:
viii,180 leaves : ill. ; 29 cm.

Subjects

Subjects / Keywords:
Cogs ( jstor )
Electrodes ( jstor )
Genetic mapping ( jstor )
Hands ( jstor )
Maps ( jstor )
Motor ability ( jstor )
Motor cortex ( jstor )
Neurons ( jstor )
Strokes ( jstor )
Upper extremity ( jstor )
Cerebrovascular Accident ( mesh )
Department of Rehabilitation Science thesis Ph.D ( mesh )
Dissertations, Academic -- College of Health Professions -- Department of Rehabilitation Science -- UF ( mesh )
Evoked Potentials, Motor ( mesh )
Motion Therapy, Continuous Passive -- methods ( mesh )
Motor Cortex ( mesh )
Neuronal Plasticity ( mesh )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 2003.
Bibliography:
Includes bibliographical references (leaves 170-179).
Additional Physical Form:
Also available online.
General Note:
Typescript.
General Note:
Vita.
Statement of Responsibility:
by Matthew Paul Malcolm.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
The University of Florida George A. Smathers Libraries respect the intellectual property rights of others and do not claim any copyright interest in this item. This item may be protected by copyright but is made available here under a claim of fair use (17 U.S.C. §107) for non-profit research and educational purposes. Users of this work have responsibility for determining copyright status prior to reusing, publishing or reproducing this item for purposes other than what is allowed by fair use or other copyright exemptions. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder. The Smathers Libraries would like to learn more about this item and invite individuals or organizations to contact the RDS coordinator (ufdissertations@uflib.ufl.edu) with any additional information they can provide.
Resource Identifier:
030445411 ( ALEPH )
54726876 ( OCLC )

Downloads

This item has the following downloads:


Full Text










RELIABILITY AND UTILITY OF TRANSCRANIAL MAGNETIC STIMULATION TO
ASSESS ACTIVITY-DEPENDENT PLASTICITY IN HUMAN STROKE













By

MATTHEW PAUL MALCOLM












A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2003






















To April and Grandma













ACKNOWLEDGMENTS

A doctoral dissertation cannot occur at the hands of a single person. The success of all aspects of the project is intimately tied to the support and guidance of many individuals. In light of this, I am extremely grateful to all of those involved with my education, research and life.

I must first thank my wonderful wife, April, for her unending patience,

support, rationality, and love. She moved away from her family to follow me to UF, endured the ups and downs of my academic career, and helped me to enjoy life as much as a graduate student can. April has been my angel during this whole process, and I could not have done it without her.

I owe a debt of gratitude to the members of my committee: Dr. Leslie

Gonzalez-Rothi, Dr. William Triggs, Dr. Orit Shechtman, and Dr. Kathye E. Light. I thank Leslie for being supportive during my pre-doctoral fellowship at the Brain Rehabilitation Research Center, and for her words of advice throughout my job search. I am fortunate to have known Bill Triggs on both a personal and professional level, as he is a true gentleman and scholar. He was both a TMS and musical 'sensei' to me and I look forward to future collaborations with him. Orit was the first faculty member to take me under her wing, and has taught me so much about educating students. She also stuck with me through many dissertation topics and always lent a supportive ear to my occasional frustrations.


iii








Last but not least, I must thank one of the most influential people of my life, Kathye Light. Kathye is a wonderful mentor who fills her graduate students with confidence and strength. I am blessed to have learned so much about science, the politics of academia, and life from Kathye. I owe much of my academic and research success to her, and will never forget her unselfish contributions to my life. I also thank Dr. Carl Kukulka, Dr. Mark Bishop, Dr. William Mann, Dr. Pam Duncan, Dr. Andrea Behrman, Dr. Sam Wu, Kimberly Reid, and Dick Moss for their support, assistance and guidance during my doctoral studies.

One key to surviving a doctoral program is to share the process with

wonderful fellow students. For example, I am fortunate to have been one half of the "brother and sister" research duo known as "Matt and Stacy: the CIMT kids". Stacy was a great person to bounce questions, ideas and problems off of, and she really challenged me to be a better student. I thank Gauri Khandekar, as her assistance during data collection and analysis proved to be instrumental in the success of this dissertation. I also owe a debt of gratitude to all of the people working in the stroke rehabilitation lab: Tara, Shalaka, Vicky, Sharon, and Cristina. I thank them for doing such a great job training the study participants. Finally, I thank Sheryl, Po, Amy, Arlene, Emily, Tiffany and Michelle for being the best fellow students a guy could ask for.

From the time I was able to read and tell time, my family and friends have encouraged my pursuit for knowledge. My parents taught me the value of education and helped me to see my potential, even when I doubted my



iv








intelligence or will. I thank my sister, Lucy, for saying, "you are so smart" with complete sincerity and pride. The loudest cheers during my academic pursuits have come from my grandparents, who have bragged to anyone and everyone who will listen. Some of you are in heaven now, but I have felt you watching over me all of the while (sorry about all of the swearing). Cheryl and Chris were awesome to believe in me, even though my time in Florida was difficult on their family. I thank Sue and Drew for being the first people to encourage my decision to return to college, and Lisa and Ryan for offering support when the going got rough. Finally, I appreciate Pete, Oscar, Joe, Jill, Jimmy, Jeff, Mary, and the entire Buffalo Gang for making life enjoyable during my 9 years of college.

I also wish to acknowledge those financial contributions that helped me pursue my research and academic goals: the VA Division of Rehabilitation Research and Development, the VA Brain Rehabilitation Research Center, the Shands Hospital Board of Directors, and the Departments of Occupational and Physical Therapy at the University of Florida.

Last and most importantly, I thank God for this. I am eternally grateful

for the 30 years of guidance that He has provided in my life. In addition, I thank Him for creating the beauty that is the human mind.









v













PREFACE

The reader should note that those persons involved in the experiment on stroke are not referred to as patients, but rather as participants or individuals. This is an important distinction that is often not made in rehabilitation research. The word patient implies that the person is ill or in need of medical care. Labeling survivors of stroke as patients only furthers the stereotype that these people are disadvantaged and dependent on others. Although the participants in the second experiment of this study had sustained a stroke, they were otherwise healthy, competent and motivated individuals.























vi














TABLE OF CONTENTS
Page

ACKNOWLEDGMENTS................................................................................... III

PREFACE ..................................................................vi

ABSTRACT ................................................................. .................. x

CHAPTER

1 INTRODUCTION AND BACKGROUND ..................................................... 1

Specific Aim s and Hypotheses ............................................................... 7
Background and Literature Review ................................................... 8
Neural Control of Hand Movement ......................................... 8
Activity-Dependent Neuroplasticity .................................................13
Transcranial Magnetic Stimulation .................................................. 21
Post-Stroke Activity-Dependent Plasticity and Neurorehabilitation ......29
Sum m ary...................................................... ............. .............35

2 EXPERIMENT I: RELIABILITY OF TRANSCRANIAL MAGNETIC
STIMULATION ........................................................................ 36

M ethods........................................................ ........................................41
Results................................................................. ......... ..........50
Discussion........................... ................ ................. .................65

3 EXPERIMENT II: ACTIVITY-DEPENDENT PLASTICITY IN STROKE .....79

Methods....................................................... ............. ..............85
Results ......................................................................................... 97
D iscussion......................................................... ........................... .. 118

4 GENERAL SUMMARY AND CONCLUSIONS .............................................38

Experiment I Summary ..................................... 138
Experim ent II Sum m ary ................................................................ 141
General Conclusions ...................................... 146



vii









APPENDIX

A PILOT DATA .............................................................................. ... 148

B BEHAVIORAL EVALUATION FORMS ............................................... 157

C INFORMED CONSENT TO PARTICIPATE IN RESEARCH ...................... 161

REFERENCE LIST ................................................................................. ...170

BIOGRAPHICAL SKETCH ....................................................................... ...... 180








































viii














Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy

RELIABILITY AND UTILITY OF TRANSCRANIAL MAGNETIC STIMULATION TO
ASSESS ACTIVITY-DEPENDENT PLASTICITY IN HUMAN STROKE By

Matthew Paul Malcolm

December 2003

Chair: Kathye E. Light, Ph.D.
Cochair: Orit Shechtman, Ph.D.
Major Department: Rehabilitation Science

Motor cortex (Ml) neuroplasticity is of primary interest to rehabilitation scientists, as this process may underlie post-stroke recovery of movement. Of the several neurophysiologic techniques, transcranial magnetic stimulation (TMS) is the most appropriate for studying M1 plasticity. Numerous studies have used TMS, however, limited evidence exists for the reliability of this technique. In Experiment 1, we sought to establish test-retest reliability of several TMS measures of M1 organization and excitability. In Experiment 2, we used TMS to assess plasticity related to recovery of upper extremity function in stroke survivors engaged in constraint-induced (CI) movement therapy. We hypothesized that (1) TMS measures would demonstrate good test-retest reliability, and (2) that stroke survivors would demonstrate M1 plasticity following a course of CI therapy. Participants in Experiment 1 were 20 healthy volunteers.

ix








Participants in Experiment 2 were 23 individuals who were 10 months to 10.75 years post-stroke. In both experiments, the following TMS variables were investigated in two hand and two forearm muscle representations: motor map size, motor map volume, map center of gravity, recruitment curve slope, and motor threshold. Participants in both experiments were tested on two sessions separated by 2 weeks. In Experiment 2, subjects also underwent an evaluation of upper extremity function. Participants in the second experiment received CI therapy during the 2-week testing interval. The intraclass correlation coefficient was used to assess test-retest reliability. Separate t-tests were used to assess pre- to post-CI therapy differences. Association between TMS and functional outcomes was determined using the Pearson r. Group differences between high and low functioning subjects were assessed using separate ANOVAs. Noteworthy findings from Experiment 1 include generally moderate to high reliability for the TMS measures. In Experiment 2, we found significant changes in some M1 muscle representations, which were paralleled by, but not directly correlated with, functional improvements. We found generally small differences between high and low functioning groups. TMS is a reliable measure of M1 organization and excitability, and may be used to investigate activity-dependent plasticity associated with intensive upper limb training in individuals post-stroke.









x












CHAPTER 1
INTRODUCTION AND BACKGROUND

One of the most important properties of the human brain is its capacity to adapt to a variety of demands. Pioneering studies in both nonhuman primate and human models demonstrate that learning, practice, and recovery from neurological injuries are associated with reorganization of primary motor cortex

(Ml) representations. Alterations of M1 representations of upper limb movement are of particular interest, as these changes subserve the acquisition or recovery of complex, skilled movements. For example, the learning of a new fine-motor skill is associated with an enlargement of the motor maps of the involved digits after several days of intensive practice (Karni et al. 1995, Pascual-Leone et al. 1995a). Similarly, recovery of hand function in individuals post-stroke has been associated with an enlargement or shifting of the hand representation in the damaged hemisphere (Cicinelli et al. 1997; Liepert et al. 1998b, 2000; Rossini et al. 1998; Traversa et al. 1997). Such activity-dependent alterations in M1 representations probably involve changes in synaptic efficacy, unmasking of silent synapses, and/or shifts in the excitatory-inhibitory balance (Hallett 2001). These changes are also paralleled by improved movement skill, implicating activity-dependent neuroplasticity as a substrate to skill mastery during novel learning or during the recovery of movement after stroke.



1





2

Three primary techniques exist to investigate the organization and

plasticity of M1. Functional magnetic resonance imaging (fMRI) and positron emission tomography (PET) are neuroimaging techniques that measure blood flow or metabolic changes that are linked to function-related activity of neurons. When mapping a motor representation, fMRI and PET may be used to visualize the entire neural network subserving a particular movement. There are, however, a number of limitations in both the fMRI and PET techniques. These include limited temporal resolution, the inability to differentiate between activity related to excitation or inhibition, and the requirement of the subject to perform a discrete movement (which may be difficult in the case of stroke). Given these limitations, fMRI and PET may not be the most sensitive and appropriate techniques to track discrete plastic changes associated with the acquisition or recovery of specific motor skills.

A third neurophysiological technique, transcranial magnetic stimulation

(TMS), is perhaps better suited to investigate motor nervous system changes. In mapping studies, TMS is used to indirectly excite corticospinal neurons within M1. In this manner, TMS can be used to outline the location, size, and excitability of the motor representation subserving a particular muscle. The advantage of using TMS over fMRI and PET is that this technique maintains high temporal resolution, activates both excitatory and inhibitory inputs to the corticospinal tract, and does not require the subject to produce any movement. Since the introduction of TMS, numerous studies have been performed to





3


investigate the somatotopic organization of M1 (Aimonetti et al. 2002, BrasilNeto et al. 1992b, Roricht et al. 1999, Wassermann et al. 1992); stimulusresponse characteristics of the corticospinal system (Boroojerdi et al. 2001a, Devanne et al. 1997, Ray et al. 2002, Ridding & Rothwell 1997, Thickbroom et al. 1998); and alterations in the motor system following skill practice (Liepert et al. 1998a, Pascualleone et al. 1995, Pascual-Leone et al. 1995b) or after stroke (Cicinelli et al. 1997; Liepert et al. 1998b, 2000, 2001; Traversa et al. 1997; Trompetto et al. 2000).

The extensive number of TMS studies supports the utility of this

technology to investigate the motor nervous system. Additionally, several of these investigations report the occurrence of activity-dependent plastic changes related to skill practice and recovery from stroke. Results of these studies, however, should be considered with caution, as the reliability of TMS assessments of M1 representations has received limited attention. Only five published studies have investigated the reproducibility of TMS measures (Carroll et al. 2001, McMillan et al. 1998, Miranda et al. 1997, Mortifee et al. 1994, Uy et al. 2002). Although these are important early studies, most of them have methodological and statistical issues that cast doubt on their validity. Moreover, the range of normal variation in TMS measures of cortical organization and excitability has not been adequately determined. Establishing the reliability of TMS and limits of normality within the nervous system are essential factors that need to be determined before further claims of activity-dependent





4


neuroplasticity. Given such consideration, TMS could be a useful tool in aiding researchers to uncover the impact of intensive task practice or therapy on the normal and stroke-damaged brain.

The ability of TMS to assess activity-dependent plasticity is important to

research on the recovery of skilled upper extremity function after stroke. Clearly, there is a critical need to identify the neural correlates to long-term recovery from stroke, as this disease is the leading cause of disability in the United States (Dobkin 1995). Stroke affects approximately 730,000 Americans annually (Dobkin 1995), and leaves more than half of these individuals with mild to severe movement deficits in the arm and hand (Taub et al. 1999). Until recently, recovery of upper extremity function was considered unrecoverable after 6 months had passed since stroke onset (Hallett, 2001). As a result, current models of clinical practice are often based upon theories that hold the nervous system to be rigid, hierarchical and compartmentalized (Held 2000). Additionally, the window of effective treatment is often considered to be within the first 6 months post-stroke. This line of thinking influenced traditional therapies such that the focus of stroke rehabilitation has often been to use compensatory rather than restorative techniques to access function (Held 2000). As result, the long-term potential for recovery is often not optimized or explored.

The recent explosion of research on the topic of plasticity, however, has

begun to change traditional views of the brain's capabilities following injury. This work has demonstrated that even the damaged brain is capable of functional





5

change and that the neural circuitry may be refined through intensive therapy. Basic science research in a variety of animal models of stroke has identified the mechanisms of these neural changes down to the cellular level (Jagodzinski & Hess, 2001; Rioult-Pedotti et al., 1993; Greenough, Larson & Withers, 1985). Researchers are now poised to translate basic science evidence of neuroplasticity to clinical investigations of the neural correlates of therapy in the human stroke model. Within such a framework, brain damage is now viewed as less catastrophic in that the individual has a greater potential for recovery than was once believed (Held 2000). Recovery of upper extremity function and the underlying activity-dependent plasticity appears to occur by engaging the involved limb in massed practice of meaningful tasks.

Constraint induced (CI) movement therapy is a recently developed stroke therapy that utilizes an intensive program of practice to treat upper extremity hemiparesis after stroke (Taub et al. 1993, Taub et al. 1999). The key therapeutic factor in CI therapy is massed practice of functional motor tasks. Several reports demonstrate the effectiveness of CI therapy to improve coordination, movement speed, and amount and quality of use of the hemiparetic arm and hand in chronic stroke survivors (Kunkel et al. 1999, Miltner et al. 1999, Taub et al. 1993). CI therapy is also believed to drive activitydependent neuroplasticity in the primary motor cortex. Two published TMS studies report alterations in the neural representations of affected hand muscles following a course of CI therapy (Liepert et al. 1998b, 2000). Liepert's





6

preliminary work provides initial evidence for the potential of therapy-induced reorganization of the damaged hemisphere after stroke. Other neuroimaging studies provide additional support to the theory of post-stroke neuroplasticity in the primary motor cortex (Cramer & Bastings 2000, Kopp et al. 1999, Levy et al. 2001, Traversa et al. 1997).

Although TMS and other neuroimaging studies report a link between recovery and plasticity, there are a number of theoretical and methodological limitations in the current published research. One major limitation is that the reliability of techniques, such as TMS, has not been adequately established. Because of the variability inherent in complex neurophysiological tools and the human nervous system, researchers must first separate extraneous variability from true changes in the brain. Claims of neuroplasticity are unfounded until this preliminary step is addressed and established. The role of therapy is seldom considered in brain plasticity studies, despite the fact that animal research describes functional neuroplasticity as 'activity dependent' (Johansson 2000; Liepert et al. 1998b, 2000; Nudo et al. 2001). The existing neuroimaging work also fails to identify a direct association between neural and behavioral changes. Functional measures directly related to changes in the primary motor cortex, which supports skilled movement, should be used to provide a more precise relationship between plasticity and a specific function.

The primary purposes of this study were two-fold. First, we sought to assess the test-retest reliability of multiple and parallel TMS measures of the





7


location, organization and excitability of upper extremity representations in M1.

The second purpose was to use TMS to investigate the impact of CI therapy on

motor representations in chronic stroke survivors (> 9 months post-stroke).

These experiments evolved from the following specific aims and hypotheses.

Specific Aims and Hypotheses

Aim #1: To demonstrate the reproducibility of the transcranial magnetic

stimulation (TMS) measurements across two testing sessions in neurologically

intact individuals.

Hypotheses: The following physiological characteristics of the primary
motor cortex (Ml), as measured by TMS, will demonstrate good test-retest
reliability when assessed over two testing sessions in neurologically intact
individuals:
(a) size (or area) of motor maps of muscle representations
(b) volume of motor maps
(c) map location (center of gravity)
(d) threshold for excitation
(e) slope of the stimulus-response (or recruitment) curve.

Aim #2: To determine the neural correlates to therapy-induced recovery of

upper extremity function in the affected M1 of individuals 9 months or greater

post-stroke.

Hypotheses: An intensive therapy program directed at the hemiparetic upper limb will produce the following neurophysiological changes in the
contralateral M1, as measured by TMS:
(a) enlargement of motor maps of muscle representations
(b) increase in the volume of motor maps
(c) lateral or medial shift of the maps, as evidenced by a change in the
location of the map center of gravity
(d) decrease in the threshold for excitation
(e) increase in recruitment curve slope.





8

Aim 3: To determine the association between neurophysiological changes in the affected M1 and improvements in hemiparetic arm and hand function following constraint-induced movement therapy.

Hypotheses: Neurophysiological changes will be highly correlated with
functional improvements in the following areas:
(a) dexterity
(b) movement speed
(c) amount of use
(d) strength.

Background and Literature Review

The background and literature review are divided into five primary sections. First, the neural control of upper extremity motor function is considered, emphasizing the relationship between movement control and the role of the primary motor cortex and corticospinal system. Second, animal and human evidence for activity-dependent plasticity were reviewed, as well as some of the neural mechanisms of neuroplasticity in the motor cortex. Third, the utility and reliability of transcranial magnetic stimulation as an effective technique for studying activity-dependent plasticity is addressed. The final section considers the role of therapy in driving functional plasticity of the strokedamaged motor cortex, as measured by transcranial magnetic stimulation. Neural Control of Hand Movement

Highly coordinated movement of the upper extremity is dependent upon a diffuse network of nervous system components. As such, the neural codes that initiate and control skilled movement cannot be isolated to one exclusive region. Although the primary motor cortex is but one player in this network, its strong





9


and relatively direct connection with the peripheral motor system suggests that this region plays a pivotal role during skill learning and practice. As will be discussed, the primary motor cortex contains representations with multiple cell assemblies that code for a variety of features inherent in movement. These representations and their specific assemblies are modified through repetition and practice.

Motor control

The representations and plans for movement are thought to be stored in generalized motor programs (Schmidt & Lee 1999). During voluntary movement, these motor programs specify invariant features such as speed, acceleration, and relative force (Schmidt & Lee 1999). Motor programs also tell the nervous system how to respond to sensory input related to the task (Kandel et al. 2000). Essentially, these programs are plans that specify the kinematic and dynamic features of a movement, and specify how the system should adjust based upon sensory feedback. Complex motor skills like prehension, writing, typing, and drawing are governed by motor programs that have a significant representation in the primary motor cortex.

Practice and learning can affect the distribution of a motor program's neural representation. For example, the performance of novel motor tasks tends to be dependent upon neural control from the supplementary motor area (Hikosaka et al. 1996). As the task is practiced and learned it becomes more automatic (that is, it requires limited cognitive control) (Schmidt & Lee 1999). With this





10

automaticity in motor skill, the neural control of performance of the task shifts from the supplementary motor area to the primary motor cortex (Hikosaka et al. 1996, Kandel et al. 2000).

Although the primary motor cortex is not the only player in the generation and control of voluntary movement, the learning and mastery of skilled tasks are highly dependent upon it. Damage to areas outside of the motor cortex will affect but not abolish movement, whereas damage involving the primary motor cortex typically results in paralysis and moderate to severe deficits in upperextremity motor function.

The primary motor cortex and corticospinal system

Fine-motor control of the hand is largely under control of the corticospinal system. Corticospinal fibers arise from a number of different brain regions that are involved with the planning, initiation and modulation of movement. The majority of these fibers (some 40%) arise from cortical Layer V of the primary motor cortex (Kandel et al. 2000). From this layer, the corticospinal fibers pass through the corona radiata and converge to enter the posterior limb of the internal capsule. This tract then descends through the brain stem where 85 to 90% of the fibers cross in the pyramidal decussation at the medullospinal junction (Haines 2002). Corticospinal fibers that originate from M1 travel in the lateral funiculus of the spinal cord where they primarily terminate in the intermediate zone and anterior horn (laminae VII-IX). The larger corticospinal fibers may have only a monosynaptic connection with the alpha motor neurons








that innervate intrinsic hand muscles, while some synapse with interneurons within the spinal cord. The latter, indirect connections are important for coordination of larger groups of muscles, as required during reaching (Kandel et al. 2000). Highly coordinated movements, involving intrinsic and extrinsic hand muscles, are dependent upon these corticospinal projections.

The seminal work of Penfield and Rasmussen (1950) described a

somatotopic organization for M1. During neurosurgery with the brain exposed, these scientists found that muscular responses could be elicited by electrically stimulating Broadmann's area #4 in the precentral gyrus. This area corresponds to the primary motor cortex. These responses occurred in an organized fashion, such that the neural representations of adjacent body parts were typically located next to each other. These orderly representations may be joined together to form the motor homunculus. The disproportion of body size, as represented by the homunculus, reflects the density and distribution of corticospinal fibers devoted to muscle control. The largest representations (i.e., hand and face) subserve movements requiring the greatest precision (Rossini & Pauri 2000).

Highly skilled movement involves a wide distribution of brain regions. Early concepts of motor control viewed M1 simply as the switchboard in the motor nervous system (Haines 2002). These theories were based upon electrical stimulation studies, which found that individual muscle responses could be generated during stimulation (Kandel et al. 2000, Penfield & Rasmussen 1950).





12

Further animal and human investigations, however, identify divergence and convergence in the system, suggesting that the M1 is capable of coordinating multiple activations of different cells to create multiple movements (Hallett 2001, Rossini & Pauri 2000). Additionally, muscle representations are diffuse and distributed among several highly specialized cells.

Evarts (1968) provided the initial evidence that M1 plays a complex and

directed role in controlling voluntary movement. In this now classic study, Evarts showed that, during a wrist flexion task, the firing of M1 neurons varied with the amount of force required to move the hand against resistance, but not with degree of movement. The activity of these neurons, therefore, is to specifically signal the amount of force required to produce a movement rather than actual displacement of the limb. Similarly, Maier et al. (1993) discovered that neurons active during precision grip are silent during power grip. M1 neurons also work together to produce coordinated movements. For example, Georgopoulos (1982) demonstrated that specific classes of M1 neurons work together to coordinate the spatial demands of a task, as required during activities like reaching. M1 neurons appear to be involved with planning the movement, as they become active prior to muscle initiation (Evarts 1968, Leonard 1998).

Leonard (1998) describes M1 as a dynamic structure that is involved in the spatial transformation (i.e., sensory perceptions to motor acts), trajectory control and the modulation of other sub-cortical structures involved in movement. The conclusion of Leonard's research and other similar work is that





13

the primary motor cortex is highly specialized and imminently involved in the programming, execution and modulation of skilled movement. Furthermore, M1 appears to have two levels of functional organization: a lower level that directly controls groups of muscles that can be brought together in task specific combinations; and a higher level control system that encodes the more global features of movement, such as force and spatiotemporal characteristics (Kandel et al. 2000).

Activity-Dependent Neuroplasticity

The motor cortex, like much of the central nervous system, is a competitive system. The 'competition' for resources within M1 neural representations is a fundamental principle of cortical plasticity (Rossini & Pauri 2000). Munk (1881) labeled this phenomenon as "vicariation of function". In the absence of input or the demand for output, a neural representation may be taken over by the adjacent representation. For example, cutting of the facial nerve results in a rapid decrease in the facial neural representation, with a rapid increase in the size of the forearm and hand motor representations (Sanes et al. 1988). On the other hand, intensive and purposeful activity extends the representation for the muscle and/or limb in use. Scientists have demonstrated that regular performance of a skilled motor task results in an enlargement of the cortical representation for the muscles involved, as seen for the fingers in string players (Elbert et al. 1995). Similarly, the M1 representation of the reading finger is





14

expanded in blind Braille readers (Pascual-Leone et al. 1993) and, furthermore, fluctuates with changes in reading activity patterns (Pascual-Leone et al. 1995a).

Lesion studies in sub-human primates also provide evidence for the relationship between activity-dependent plasticity and skill training. These studies demonstrate that disuse after injury results in alterations of the neural system that once supported function. For example, small surgically induced lesions in the M1 hand area of squirrel monkeys created immediate deficits in prehensile abilities and a resultant decreased use of the impaired hand (Nudo & Milliken 1996). Non-use of the affected limb resulted in an extension of the lesion, such that there was further territorial loss in the hand representation. Similar changes related to disuse have been demonstrated in human studies involving amputees (Roricht et al. 1999) and during limb immobilization (Liepert et al. 1995). In the case of the stroke-affected monkeys, retraining of prehension in the affected hand reduced extension of the lesion, and induced further reorganization in the surrounding undamaged M1. These plastic changes were associated with near-complete recovery of skilled hand use.

Activity-dependent plasticity is a descriptor that broadly encompasses many intricate neuronal mechanisms. Changes in the size, location and excitability of M1 representations likely reflect alterations in synaptic communication, neurotransmitter function, and neuronal morphology. Neuroplasticity is typically characterized in terms of short and long-term changes. Short-term or more rapid changes primarily involve alterations in the behavior of synapses, while





15

long-term changes include morphological changes in the neural circuitry (Johansson 2000). Both short and long-term alterations are evident in the practice and development of motor skill, and are not mutually exclusive (Hallett 2001).

Plasticity may take the form of a change in synaptic efficacy, that is, an

improvement in synaptic communication. Hebb (1949) postulated that increases in synaptic efficacy occur when the firing of one neuron repeatedly produces firing in another neuron to which it is connected. In other words, an association of pre- and post-synaptic activity in two neurons results in some change in these neurons such that the synaptic connection between them is strengthened (Hebb 1949). The Hebbian concept of activity-dependent modification of synaptic strength is the primary example of plasticity of synaptic communication. This mechanism allows for the refinement of neural circuitry through activitydependent means at the synaptic level. Much of the understanding of learning and memory processes are based on Hebb's concept (Turrigiano 1999). Repetitive activity or practice drives synaptic efficacy and the threshold for activation by refining the temporal structure and synchronization of impulse arrival and neuronal firing (Rossini 2000). For example, long-term potentiation and long-term depression, two phenomena involved in memory formation and learning, follow Hebbian rules (Maren & Baudry 1995).

Long-term potentiation (LTP) represents a relatively fast strengthening of existing synapses (Hallett, 2000). LTP is different than short-term potentiation





16

(STP) and post-tetanic potentiation (PTP) in that it persists for hours or days, while the later represent more short-term changes in synaptic strength (i.e., seconds or minutes) (Maren & Baudry 1995). Additionally, the long-term characteristics of LTP are dependent upon a unique cellular mechanism. Although hippocampal LTP has been a primary focus of research on memory and learning (Kandel et al. 2000, Maren & Baudry 1995), LTP also occurs in the diffuse network that subserves movement (Hess 1996). Rioult-Pedotti et al. (1998) demonstrated that intensive practice of a reaching task in rats induced LTP in intracortical connections of M1. LTP may also be induced directly in Layer V of the motor cortex, which indicates that direct connections to corticospinal nerve cell bodies may be influenced by LTP (Jagodzinski & Hess 2001). Furthermore, motor cortex plasticity is substantially reduced when LTP is blocked by NDMA receptor blockers or by GABAergic disinhibitors (Boroojerdi et al. 2001a). These findings suggest a common mechanistic link between LTP and activity-dependent plasticity in the motor system.

Long-term depression (LTD) is another communicative mechanism that underlies memory and motor learning. LTD involves a relatively fast weakening of existing synapses (Hallett 2000). LTD has been primarily studied in the cerebellum, but like LTP, it occurs elsewhere in the brain (Maren & Baudry 1995). In the cerebellum, LTD involves the activity of parallel and climbing fibers. Following several pairings of parallel and climbing fiber stimulation, synaptic responses in the parallel fiber neurons exhibit an enduring depression





17

(Maren & Baudry 1995). Hess (1996) demonstrated that LTD may be induced in the horizontal connections in Layers II and III of the rat primary motor cortex, and further suggests that the regulation of this neuronal mechanism is activity dependent.

The occurrence of both potentiation and depression in the nervous system illustrates the delicate balance between change and stability. Together, these processes alter and stabilize the properties of neural circuits (Turrigiano 1999). Plasticity, through Hebbian mechanisms such as LTP and LTD, serves to refine the neural circuitry involved in skill learning and performance. This has been illustrated by neuroimaging studies that reveal large and diffuse brain activation during initial motor skill learning (Leonard 1998). With repeated practice and a degree of automaticity, a reduction in the number of regions involved in performing the task is seen. Additionally, the primary motor cortex appears to 'take over' the neural representation for motor skill performance when the task becomes more automatic (Kandel et al. 2000). The change in 'skill representation' may occur because of a change in the brain's strategy or because of a change in synaptic communication. In either case, plasticity is a process dependentupon activity. Thus, neuronal associativity and cooperativity involved in the learning of skill (or perhaps the refinement of motor programs) are developed through repetitive action and practice.

Rapid alterations in cortical representations are the result of short-term changes in the connectivity of neuronal networks (Johansson 2000). Such





18

alterations are not, per se, the result of anatomical changes, but rather shortterm adaptations in the existing circuitry. These represent the brain's attempt to make rapid adaptations in the face of a behavioral demand. The unmasking of 'silent' connections, and alterations in overall excitation and inhibition represent such short-term alterations.

The unmasking of silent synapses may be a passive or an active process. Passive unmasking has been demonstrated to occur in persons following amputation (Dobkin 1993, Ramachandran 1993). Within four weeks of amputation, representational maps in the primary sensory cortex (Sl) are noted to be reorganized (Ramachandran 1993). When the patient's face or residual limb was touched, he reported sensation in the missing hand, suggesting that sensory input from the face had now invaded the adjacent hand area in the sensory homunculus. Similar findings have been noted in studies using shortterm ischemic nerve blocks and result in nearly instantaneous alterations in M1 and S1 representations (Boroojerdi et al. 2001b). Such changes in cortical representations probably arise by the passive unmasking of previously silent thalamocortical and intracortical synapses (Dobkin 1993). Although this form of unmasking does not rely on practice or action, it could still be considered activity-dependent, in the sense that the brain is likely to change with a discrete loss of afferent activity.

Active unmasking of silent connections more clearly represents activitydependent plasticity. Unmasking through processes involving concentrated





19

activity has been demonstrated in subjects learning a skilled movement task over a period of 5 days (Pascual-Leone et al. 1995a). As the subjects became more skilled in a five-finger piano exercise, the size of the M1 hand representation increased. Such rapid changes in cortical maps likely represent the unmasking of weak or secondary synaptic connections, and are driven by concentrated practice.

Short-term alterations in the functioning of neural circuitry also involve changes in the inhibitory and excitatory characteristics of the brain. For example, the excitability of the neuronal membrane may be altered as a function of the behavior of sodium ion channels (Halter et al. 1995). Such a change accounts for a lower excitation threshold for a given neuron (Chen 2000). Unmasking or changes in membrane excitability are caused partly by the removal of tonic inhibition (Jacobs & Donoghue 1991). In the case of brain damage or stroke, increased inhibition is noted (Liepert et al. 2000). Plasticity is enhanced with the release of tonic inhibition, which may be evidence for the cortex's attempt to repair itself (Hallett 2001).

Short-term physiological changes such as unmasking, LTP or LTD, and

shifts in excitability may eventually give rise to structural changes (Kandel et al. 2000). Although physiological and morphological plasticity operate in different time periods, they are not mutually exclusive (Hallett 2001). Activity drives short-term physiological changes, which then may drive certain alterations in the morphology of the nervous system. Examples of these anatomical changes are





20

the pruning back of existing synaptic connections, growth of new synaptic connections, dendritic arborization, and the proliferation of non-neural cells (Johansson 2000). As with short-term changes, the critical element for morphological plasticity is activity.

Long-term anatomical changes have been identified in studies on environmental enrichment in animals. Rats housed in complex, enriched environments with access to toys, activities, and socialization show more morphological brain changes than animals housed individually or in cages without varied physical activity (Johansson 2000). These animals develop more dendritic branching and more synapses per neuron within the sensorimotor cortex (Johansson 2000). Dendritic spines, sites that receive most of the excitatory synaptic inputs, are formed in greater numbers and are continuously modified in animals reared in enriched environments (Fischer et al. 1998).

Alterations in the size, location and excitability of M1 representations are likely the result of any or all of the above-mentioned mechanisms of plasticity. Changes in the gross characteristics of representations may be successfully tracked with neurophysiologic techniques such as TMS. Importantly, this technology does not possess the ability to differentiate between the various types of mechanisms of neuroplasticity. More detailed investigations of the cellular events related to plasticity are not yet possible or appropriate in human studies. By studying more general alterations in motor representations,





21

however, we assume that these specific changes are occurring at the neuronal level.

Transcranial Magnetic Stimulation

Barker et al. (1985) first described TMS as an alternative to transcranial electric stimulation (TES). Unlike TES, TMS provides a non-painful means to stimulate the brain. TMS works by producing a large and brief electric current, which is passed through a heavily insulated wire coil that is placed on the skull over the area that corresponds to the location of M1 (Bastings et al. 1998). This transient current produces a large, time-varying magnetic field. The magnetic field passes through the skull relatively unimpeded and creates a perpendicular electric field in the underlying neural tissue, which activates neurons in M1 (Weber & Eisen 2002). Motor physiology studies indicate that TMS indirectly activates corticospinal neurons by directly activating horizontally oriented interneurons in the motor cortex (Di Lazzaro et al. 1998). When applied over a muscle representation in the motor cortex, TMS generates a motor evoked potential (MEP), which may be quantified by means of electromyography.

TMS has several advantages as a neurophysiological tool. The technique: is non-invasive; does not require movement from the subject; has high temporal and spatial resolution; and is relatively inexpensive to administer. While PET and fMRI take up to two or three hours to administer, a complete map of the hand and forearm area of M1 may be performed in one to one and a half hours using TMS. Additionally, TMS may be performed while the subject is comfortably





22

seated in a reclined position (where as fMRI requires the subject to be in a loud and potentially claustrophobic space). A major benefit of TMS is that it requires no movement from the subject. This benefit is especially valuable when studying stroke survivors, as the ability to move may be limited and highly variable from person to person. TMS also boasts high-resolution abilities for mapping. TMS has a spatial resolution of 5 mm, and a temporal resolution on the order of a few milliseconds (Weiller 1998).

Motor mapping

TMS may be used to map muscle representations by stimulating over a broad region of the motor cortex. For example, the representation for an intrinsic thumb flexor muscle may be mapped by stimulating over the 'hand area' of the motor homunculus. As in the case of Penfield and Rasmussen's electrical stimulation studies (Penfield & Rasmussen 1950), TMS is used to outline the extent of a map by assessing the number of stimulating positions that elicit an MEP in the muscle of interest. Essentially, the number of excitable positions can be determined in order to create a representational map of a particular muscle. The spatial resolution of TMS for motor mapping is improved by using a figureof-eight shaped magnetic coil (Wassermann et al. 1992), providing relatively focal stimulation at the intersection of the loops (Jalinous 1991, Triggs et al. 1999). When using a figure-of-eight coil, the spatial resolution of TMS is approximately 5 mm (Brasil-Neto et al. 1992c). In addition to providing information about the spatial area of a representation, motor maps may be





23


translated into three-dimensional volumetric representations. Map volume is represented as the sum total of MEP peak-to-peak amplitudes or areas of all active positions within the representation (Mortifee et al. 1994, Wassermann et al. 1992). This measurement is one assessment of the overall excitability of a representation.

Map center of gravity

The center of gravity (CoG) is the position of a motor map that yields the highest amplitude weighted response to stimulation (Wassermann et al. 1992). Assessment of CoG provides information about the somatotopic orientation of one muscle representation to another. Somatotopic differences between the CoGs of proximal and distal upper extremity muscles (Wassermann et al. 1992), and even between different intrinsic hand muscles (Wilson et al. 1993) have been found. Animal studies suggest that one correlate to the recovery of motor function following stroke is the recruitment of motor areas adjacent to the original representation (Nudo 1996, Nudo & Milliken 1996). Changes in the CoG of a representation may reflect such recruitment (Liepert et al. 1998b, 2000), although the existing literature is equivocal on the direction and relevancy of a shift in representation location.

Recruitment curve

TMS may also be used to investigate the input-output properties of the corticospinal system. Also known as stimulus-response curves, the recruitment curve depicts the change in MEP size as a function of stimulus intensity. This





24

measurement is thought to assess neurons other than those in the core neuronal population that are activated at motor threshold and during motor mapping (Chen 2000, Hallett 2000). The threshold for these neurons is higher either because they are less excitable or because they are further from the site of stimulation (Hallett 2000). Recruitment curves are steeper in muscles with a low threshold to stimulation (i.e., intrinsic hand muscles), and are affected by different pharmacological interventions (Boroojerdi et al. 2001a, Chen 2000). The slope and shape of these curves provide information about the neurophysiological strength of intracortical and corticospinal connections (Devanne et al. 1997, Ridding & Rothwell 1997). The slope of the recruitment curve may also be modulated by ischemic limb deafferentation (Brasil-Neto et al. 1992a), amputations (Ridding & Rothwell 1997) and by motor learning (PascualLeone et al. 1999).

Motor threshold

Motor threshold represents the lowest TMS intensity that elicits a small MEP (usually 50 pV). Motor threshold is lower for intrinsic hand muscles compared to proximal arm, trunk, and leg muscles (Chen 2000). This is probably due to differences in the strength of corticospinal projections (Chen 2000) and the larger absolute number of corticospinal neurons devoted to intrinsic hand muscles (Wassermann et al. 1992). Motor thresholed is raised by drugs that block sodium channels, but is not affected by drugs that alter GABAergic or glutaminergic systems (Ziemann et al. 1996). Given these pharmacological





25

effects, motor threshold likely reflects neuronal membrane excitability (Chen 2000). Several TMS studies in stroke survivors have shown that motor threshold is often higher in the damaged hemisphere relative to the undamaged hemisphere (Trompetto et al. 2000). Stroke survivors with a high motor threshold and/or absence of MEPs from the affected hemisphere typically have a poor functional recovery, while those with a lower motor threshold in the damaged hemisphere (relative to undamaged) have a "good" functional outcome (Trompetto et al. 2000). Based on these findings, we predicted that intensive practice of motor skills would lower motor threshold in chronic stroke survivors. Such a change would provide evidence for increased excitability in the stroke affected M1.

Reliability of TMS measurements

Several published studies have utilized TMS to investigate the organization and excitability of the motor cortex. In addition, a number of these studies report neuroplastic changes related to the practicing of skills (Pascual-Leone et al. 1995a), intensive hand use (Pascual-Leone et al. 1995b), amputation (Roricht et al. 1999), and recovery from stroke (Cicinelli et al. 1997; Liepert et al. 1998b, 2000; Pennisi et al. 2002; Rossini et al. 1998; Traversa et al. 1997; Traversa et al. 1998; Trompetto et al. 2000). The results of these studies, however, should be considered with caution, as the reliability of TMS mapping and recruitmentcurve generation is not well established. Only five published studies have investigated the reproducibility of these TMS measures (Carroll et al. 2001,





26

McMillan et al. 1998, Miranda et al. 1997, Mortifee et al. 1994, Uy et al. 2002). Although these are important early studies, most of them have methodological and statistical issues that cast doubt on their validity. Half of these existing studies do not utilize sufficient statistical analyses to demonstrate reliability. Previous reports (McMillan et al. 1998, Miranda et al. 1997, Uy et al. 2002) have utilized a comparison of means statistic (i.e., ANOVA) or coefficient of variation to assess reliability. These analyses do not consider the degree of association and agreement between subject scores. A more rigorous and preferred assessment of test-retest reliability is the intraclass correlation coefficient (ICC) because it reflects both the degree of correspondence and agreement among individual scores obtained over multiple testing sessions (Portney & Watkins 2000). A reliability study that examines multiple TMS assessments of multiple muscle representations using a rigorous statistical and methodological design is lacking.

Mortifee and colleagues (1994) published the first study on the

reproducibility of TMS mapping. These authors reported that the area and volume (amplitude weighted sum of MEPs) of motor maps for two intrinsic hand muscles were relatively stable across two testing sessions. Although this is an important and seminal work with a fairly robust method for statistical analysis, there are a number of limitations with the study. The results are based upon an investigation of only 6 subjects across two testing sessions. Only two intrinsic hand muscles were studied, which does not address if proximal (i.e., forearm)





27

muscle representations remain stable. Furthermore, only one of the two muscles studied was reliable above a threshold significance level (i.e., 0.75). These authors used a circular coil rather than the more focal figure-of-eight coil, which is the standard of practice for motor mapping. To this end, the results of Mortifee and colleagues' work may not be applicable to most mapping studies. This group also did not average responses over a set number of stimuli per site, which is again, another standard of practice in mapping research. Finally, although Mortifee et al. appear to have devised an interesting experiment, there are methodological issues, which are unclear: how was the stimulating grid referenced from session to session? Did motor threshold change from session to session? Was stimulus intensity kept constant from session to session? Each of these issues would likely affect the motor map. In the case of the latter two, transient changes in motor threshold could affect the size and shape of the motor map. With an n=6, the effect of such transient changes could not be adequately accounted for.

More recently, Uy et al. (2002) published a TMS reliability study. This is an important investigation, but one that also has similar methodological and statistical limitations. The sample size was small (n=8), although subjects were tested over four sessions, which increases the statistical power of the investigation. The authors, however, did not use an adequate statistical analysis to determine reliability. Test-retest reliability is best assessed using the intraclass correlation coefficient (ICC). The ICC has an advantage over other





28

parametric analyses in that it looks at association and agreement of outcomes from two separate testing sessions (Portney and Watkins, 2000). This limitation in Uy and colleagues' work is quite apparent upon examination of the study results. The authors used a repeated measures ANOVA to determine significant differences between any of the four testing sessions. They report an F-score equal to 2.4, which has a probability value equal to 0.094. These results indicated that there was no statistically significant difference between the testing sessions (although the p-value was still relatively small, suggesting that there might have been a trend for a difference). The repeated measures ANOVA, however, lacks three important features that are important in determining testretest reliability: (1) what is the magnitude of the similarity of the results between testing sessions, (2) what is the degree of association between these data, and (3) what is the degree of agreement between these data? The intraclass correlation coefficient accounts for all three of these variables, and thus provides the most robust method for assessing test-retest reliability. Future investigations of the reproducibility of TMS measures should use the ICC in assessing test-retest reliability.

Motor mapping alone only provides one component of neuroplasticity related to practice and recovery. As previously described, shifts in excitability also tells an important story about change in the nervous system. Recruitment curve and motor threshold analyses provide this sort of information. In fact, shifts in excitability alone may affect the area and volume of a motor map





29


(Thickbroom et al. 1998). Only a single investigation on the reliability of the recruitment curve measurement exists. To our knowledge, no study has explicitly examined the reproducibility of motor threshold across multiple test sessions.

TMS is a powerful tool that may provide insight into the workings of the nervous system. In a search for the neural correlates of recovery from stroke, rehabilitation scientists and neuroscientists are employing techniques like TMS to investigate the neural correlates to recovery from stroke. Clearly, there are a number of methodological and statistical issues that must be first investigated to determine the reliability of TMS. This is an important first step to any investigation of activity-dependent neuroplasticity. Establishing the reproducibility of TMS measures will also aide researchers in discovering the true impact of neurorehabilitation on the damaged brain. Post-Stroke Activity-Dependent Plasticity and Neurorehabilitation

Until recently, scientists and clinicians assumed that the adult brain was incapable of long-term recovery from neurological damage caused by a stroke. This line of thinking has influenced traditional therapies such that the focus of stroke rehabilitation has often been to use compensatory rather than restorative techniques to access function. In addition, clinicians often operate under the assumption that the potential recovery ends at 6 months post-stroke onset. Under these assumptions, the capability of the nervous system to continue to recover and change in the chronic phase has not been optimized. For example,





30

therapists often teach stroke patients to utilize one-handed techniques to perform activities of daily living (ADL) and other functional tasks using the unaffected upper limb. Although the individual may become proficient in managing basic self-care, his or her repertoire of skills is reduced. As a result, many stroke survivors remain dependent on others for many areas of daily living, and some 84% are unable to resume their previous personal, familial and worker roles (Dobkin 1995).

The challenging goal of rehabilitation after stroke is to improve skills like reaching, grasping and manipulating once the acute period of spontaneous recovery has passed. Because of financial and time limitations, therapists often rely upon compensatory training as a means to increase independence in their clients. The down side of such an intervention is that it encourages disuse. As the previously mentioned neuroplasticity studies have shown, there are neurological consequences for this disuse.

The primary element for the induction of plasticity is intensive practice.

Studies in animal and human models indicate that plastic changes and functional improvements are driven by sustained and repeated practice of a skill or set of skills. This relationship between practice, plasticity and improvements in skill provides rehabilitation scientists with the opportunity to re-evaluate current treatments and to create more effective ones.

Constraint-induced (CI) movement therapy is an example of a treatment that employs intensive practice to remediate upper extremity function following





31

stroke (Taub et al. 1993). CI therapy is believed to be successful for two primary reasons: (1) it reverses the psychological process of learned non-use of the affected limb and (2) it induces enduring plastic changes that provide the neural substrates of recovery (Taub et al. 1999). Similar treatment paradigms are being applied to gait training (Dobkin 1993, Taub et al. 1999), and focal hand dystonia (Taub et al. 1999).

CI therapy evolved from basic research with somatosensory

deafferentation in monkeys and is based on a behavioral theory of learned nonuse (Taub et al. 1993). Learned non-use is a psychological process that involves conditioned suppression of movement (Morris et al. 2001). In the case of human stroke, individuals learn to avoid using their affected upper extremity because such attempts result in no movement at all or clumsy, inefficient movement. At the same time, these individuals learn to compensate with the unaffected upper extremity. CI therapy seeks to reverse the process of learned non-use by increasing an individual's motivation to use the stroke affected arm and hand. The intervention involves two critical elements: intensive practice and constraint of the unaffected limb. The therapy program entails 6 hours per day of treatment for 2 weeks, during which participants are engaged in massed practice of functional motor skills. Positive knowledge of performance and results feedback is provided, which encourages appropriate motor learning. To further encourage use, the participant is also required to wear a padded restraining mitt on the unaffected limb during 90% of waking hours.





32

Taub and others believe that this experimental paradigm serves to reverse the process of learned nonuse (Taub et al. 1993). The intense 6-hour training sessions are designed to have two-fold results. First, the repetitive and concentrated treatment truly presents the subject with massed practice in learning to use their hemiparetic arm. In retraining novel motor tasks, massed practice facilitates motor learning in the early stages of skill acquisition (Schmidt & Lee 1999). Second, the therapeutic sessions are designed to make the subjects struggle, but also to achieve some level of success. Therefore, the person with a stroke begins to associate movement with some degree of success, rather than failure. In so doing, the subject is more encouraged to use the paretic limb in his daily routine at home.

The end results of CI therapy have been remarkable. Subjects have consistently demonstrated large increases in the amount and quality of use, coordination, speed of movement, and spontaneous use of the affected upper limb (Kunkel et al. 1999, Liepert et al. 1998b, Miltner et al. 1999, Taub et al. 1998). Taub and colleagues (1993) also demonstrated that these functional improvements were maintained up to 2 years post-CI therapy. In addition, Miltner et al. (1999) found that even very chronic stroke survivors (post-stroke times of 9-17 years) were amenable to CI therapy, and did as well as those individuals who were much closer in time to their stroke. Clearly, CI therapy has a dramatic effect on recovery of function in stroke, the results of which hold a relative permanency and are apparently not influenced by the time since stroke.





33

Researchers are now beginning to explore the link between intensive

neurorehabilitation, like CI therapy, and activity-dependent plasticity. This work has begun to extend the animal research on training induced neural changes. Neurophysiological techniques such as TMS allow researchers to study the organization, excitability and patterns of activation of the brain. Furthermore, these techniques provide an opportunity for neuroscientists and rehabilitation scientists to examine the neural correlates to recovery, and how therapy influences this process.

Using TMS, Liepert and colleagues (1998, 2000) demonstrated that the motor representation of the hemiparetic hand in chronic stroke survivors was reduced in size as compared to the intact hemisphere. Following a course of intensive movement therapy, the cortical map of the hemiparetic hand expanded into adjacent areas, increasing significantly from the pre-treatment testing. Using electroencephalography (EEG), Kopp et al. (1999) reported a shift of activation from the primary motor cortex towards the supplementary motor area from pre- to post-therapy. Nelles et al. (2001) used positron emission tomography (PET) to investigate the effect of therapy on behavioral and cortical changes. This group found increased activation in premotor and primary sensory areas subsequent to a standard therapy program. Levy and colleagues (2001) have provided the only fMRI evidence to support the neurological effects of intensive therapy aimed at the hemiparetic limb. In this pilot study, activity related to sensory processing and simple motor task performance increased





34

around the rim of the infarct following treatment. These subjects also demonstrated increased activation in the supplementary motor areas after training.

Although these studies and techniques have provided preliminary

evidence for neuroplasticity following stroke, the results are not clearly related to improvements in function. In other words, there may be motor cortex changes that occur as a person is progressing towards recovery, but what is the relationship between this neuroplasticity and improvements in the person's repertoire of skills? A major limitation of the existing neuroimaging work is that limited or no behavioral measures were used to support the notion that brain reorganization underlies functional recovery. Despite identifying cortical changes, Nelles and colleagues (2001) did not find any significant improvement in use of the hemiparetic limb. Liepert et al. (1998) and Liepert et al. (2000) demonstrated a marked increase in the cortical representation of the affected hand subsequent to therapy. The only behavioral measure used, however, was a subject questionnaire. No objective measures of functional recovery were performed. Levy and colleagues' (2001) pilot study provides support for the theory of activity-induced cortical reorganization as a mechanism for improved use of the hemiparetic limb. Subjects in this experiment demonstrated modest improvements in speed and quality of movement, strength, and real world use of the affected arm (Levy et al. 2001). These results were obtained in only two subjects, which limits the ability to generalize from this study.





35

Summary

Animal and human neurophysiological studies demonstrate that motor cortex representations for movement are altered with intensive practice. The acquisition of skills is clearly related and probably dependent on this plasticity. Several methods exist to investigate brain plasticity, but TMS is perhaps the bestsuited technique to examine changes in the motor nervous system. Although numerous TMS studies exist, the reliability of the technique has not received adequate attention. Establishing the reproducibility of TMS measures is a critical first step in developing an understanding of the true impact of therapy on the damaged nervous system.

The experiments performed in this study address many of the theoretical and technical issues that are evident in the existing literature on neuroplasticity. The first purpose of these experiments was to establish the reliability of TMS as a tool for studying the motor nervous system. The second purpose was to provide mechanistic evidence for the direct influence of CI therapy on both neurological and behavioral changes. The final purpose of this study was to identify the relationship between neuroplasticity and recovery of specific motor functions.













CHAPTER 2
EXPERIMENT I: RELIABILITY OF TRANSCRANIAL MAGNETIC STIMULATION

Transcranial magnetic stimulation (TMS) is a neurophysiologic technique that may be used to study the human motor nervous system. Utilizing TMS, researchers have investigated the organization and excitability of the corticospinal system that subserves voluntary movement. Until the advent of this technology, non-invasive methods for testing excitatory thresholds and for mapping the primary motor cortex (Ml) were not suitable for human studies. Since the introduction of TMS, however, numerous researchers have investigated the somatotopic organization of M1 (Aimonetti et al. 2002, Brasil-Neto et al. 1992a, Roricht et al. 1999, Wassermann et al. 1992), stimulus-response characteristics of the corticospinal system (Boroojerdi et al. 2001, Devanne et al. 1997, Ray et al. 2002, Ridding & Rothwell 1997, Thickbroom et al. 1998), and alterations in the motor system following stroke (Cicinelli et al. 1997; Liepert et al. 1998b, 2000, 2001; Traversa et al. 1997; Trompetto et al. 2000) or following skill practice (Liepert et al. 1998a; Pascual-Leone et al. 1995a, 1995b). Despite this extensive TMS research, limited attention has been directed toward identifying the reliability of TMS measurement techniques for mapping neural representations or generating stimulus-response curves. The purpose of this study was to determine the test-retest reliability of various TMS measures to



36





37

examine M1 neural representations and corticospinal excitability in normal subjects.

Barker et al. (1985) first described TMS as an alternative to transcranial electric stimulation (TES). Unlike TES, TMS provides a non-painful means to stimulate the brain. TMS works by producing a large and brief electric current, which is passed through a heavily insulated wire coil that is placed on the skull over the area that corresponds to the location of M1 (Bastings et al. 1998). This transient current produces a large, time-varying magnetic field. The magnetic field passes through the skull relatively unimpeded and creates a perpendicular electric field in the underlying neural tissue, which activates neurons in M1 (Weber & Eisen 2002). Motor physiology studies indicate that TMS indirectly activates corticospinal neurons by directly activating horizontally oriented interneurons in the motor cortex (Di Lazzaro et al. 1998). When applied over a muscle representation in the motor cortex, TMS generates a motor evoked potential (MEP), which may be quantified and qualified by means of electromyography.

TMS may be used to map muscle representations by stimulating over a broad region of the motor cortex. For example, the representation for an intrinsic thumb flexor muscle may be mapped by stimulating over the "hand area" of the motor homunculus. As in the case of Penfield and Rasmussen's electrical stimulation studies (Penfield & Rasmussen 1950), TMS is used to outline the extent of a map by assessing the number of stimulating positions that





38

elicit an MEP in the muscle of interest. Essentially, the number of excitable positions can be determined in order to create a representational map of a particular muscle. The spatial resolution of TMS for motor mapping is improved by using a figure-of-eight shaped magnetic coil (Wassermann et al. 1992), providing relatively focal stimulation at the intersection of the loops (Jalinous 1991, Triggs et al. 1999). When using a figure-of-eight coil, the spatial resolution of TMS is approximately 5 mm (Brasil-Neto et al. 1992b). In addition to providing information about the spatial area of a representation, motor maps may be translated into three-dimensional volumetric representations. Map volume is represented as the sum total of MEP peak-to-peak areas of all active positions within the representation (Mortifee et al. 1994, Wassermann et al. 1992). This measurement is one assessment of the overall excitability of a representation.

The TMS measurement, center of gravity (CoG), is the position of a motor map that yields the highest amplitude weighted response to stimulation (Wassermann et al. 1992). Assessment of CoG provides information about the somatotopic orientation of one muscle representation to another. Somatotopic differences exist between the CoGs of proximal and distal upper extremity muscles (Wassermann et al. 1992), and even between different intrinsic hand muscles (Wilson et al. 1993). Animal studies suggest that one correlate to recovery of motor function following stroke is the recruitment of motor areas adjacent to the original representation (Nudo 1996, Nudo & Milliken 1996).





39


Changes in the CoG of a representation may reflect such recruitment (Liepert et al. 1998b, 2000), however, the range of normal variation in CoG location within the healthy brain has not been established.

TMS may also be used to investigate the input-output properties of the corticospinal system. Also known as stimulus-response curves, the recruitment curve depicts the change in MEP size as a function of stimulus intensity. This measurement is thought to assess neurons other than those in the core neuronal population that is activated at motor threshold and during motor mapping (Chen 2000, Hallett 2000). The threshold for these neurons is higher either because they are less excitable or because they are further from the site of stimulation (Hallett 2000). Recruitment curves are steeper in muscles with a low threshold to stimulation (i.e., intrinsic hand muscles), and are affected by different pharmacological interventions (Boroojerdi et al. 2001, Chen 2000). The slope of these curves provides information about the neurophysiological strength of intracortical and corticospinal connections (Devanne et al. 1997, Ridding & Rothwell 1997).

The multiple capabilities of TMS make it a useful technique for providing a comprehensive assessment of motor nervous system organization and excitability. Indeed, numerous published studies have utilized TMS to investigate motor cortex changes related to the practicing of skills (Pascual-Leone et al. 1995a), intensive hand use (Pascual-Leone et al. 1995b), amputation (Roricht et al. 1999), and recovery from stroke (Cicinelli et al. 1997; Liepert et al. 1998b,





40

2000; Pennisi et al. 2002; Rossini et al. 1998; Traversa et al. 1997; Traversa et al. 1998; Trompetto et al. 2000). The results of these studies, however, should be considered with caution, as limited evidence exists for the reliability of the TMS measures employed. Only five published studies have investigated testretest reliability of TMS assessments of M1 organization or excitability (Carroll et al. 2001, McMillan et al. 1998, Miranda et al. 1997, Mortifee et al. 1994, Uy et al. 2002). Although these are important early studies, most of them have methodological and statistical issues that cast doubt on their validity. Half of these existing studies do not utilize sufficient statistical analyses to demonstrate reliability. Previous reports (McMillan et al. 1998, Miranda et al. 1997, Uy et al. 2002) have utilized a comparison of means statistic (i.e., ANOVA) or coefficient of variation to assess reliability. These analyses do not consider the degree of association and agreement between subject scores. A more rigorous and preferred assessment of test-retest reliability is the intraclass correlation coefficient (ICC) because it reflects both the degree of correspondence and agreement among individual scores obtained over multiple testing sessions (Portney & Watkins 2000). A reliability study that examines multiple TMS assessments of multiple muscle representations, while employing a rigorous statistical and methodological design, is lacking.

TMS is a powerful tool that provides insight into the workings of the

nervous system. There are, however, numerous extraneous factors that could potentially confound the results. Example confounds include: spontaneous rapid





41

alterations in cortical outflow (Ridding & Rothwell 1997) coil position (PascualLeone et al. 1994), EMG data collection and processing (Hermens et al. 2000, McMillan et al. 1998), and transient changes in motor threshold (Thickbroom et al. 1998). These phenomena may create noise in the technique that could be mistaken for change within the nervous system. Before activity-related changes in the motor nervous system may be studied, the reliability and general variability of TMS must be established.

The purpose of this study was to assess the test-retest reliability of

multiple and parallel TMS parameters. Specifically, we hypothesized that motor map area, map volume, CoG, recruitment curve slope, and the threshold for excitation would demonstrate high test-retest reliability across two testing sessions separated by 2 weeks. By studying multiple parameters, this study provides broad but detailed information about the variability in the organization and excitability of the corticospinal system, as determined by TMS.

Methods

Subjects and Preparation

The participants were 20 healthy volunteers (10 females, 10 males)

between the ages of 21 and 36 years of age (mean 26.9 4.5 years). These individuals met the following inclusion and exclusion criteria: (1) age 21 years or greater, (2) normal motor function, (3) normal cognition, (4) right hand dominant, (5) no history of neurological damage or disease, (6) no recent history of injury or disease involving the dominant upper extremity, (7) no history of





42

psychiatric disease, (8) no history of seizures or epilepsy, (9) no pacemaker or other metal implants in the upper body, and (10) negative pregnancy test (for women of childbearing potential). All aspects of participant recruitment and testing met the approval of the local Institutional Review Board prior to initiation of this project. Each participant signed a written informed consent prior to his or her involvement with any of the study procedures.

The participant was comfortably seated in a semi-reclined modified dental chair with a pillow support placed beneath the dominant forearm and hand. Passive bipolar surface electrodes were applied over the first dorsal interosseous (FDI), abductor pollicis brevis (APB), extensor digitorum communis (EDC), and flexor carpi radialis (FCR) muscles in the right upper limb in a belly-tendon arrangement. Correct placement of the electrodes was verified by asking the subject to maximally contract the muscle while the investigator monitored for an EMG output amplitude of approximately 0.8 mV. The inter-electrode distance was fixed at 20 mm for all muscles. EMG signals were filtered with a bandpass set at 2-10 kHz, rectified, and amplified with a Viking II Electromyograph (Nicolet Biomedical, Madison, WI). Audio feedback from the electromyograph was routinely monitored to ensure muscle relaxation during the testing session. A latex swim cap was placed on the participant's head so that a coordinate system could be clearly marked. The vertex (Cz) was marked as the intersection of the nasion-inion and interaural lines. Measurement of these lines was recorded to





43


ensure consistent location of the Cz across testing sessions. All TMS stimulation points were recorded in reference to the Cz (see Figure 2-1). General TMS Testing Procedure

Five primary TMS variables were investigated: (1) motor map area, (2) FDI motor map volume, (3) location of the center of gravity (CoG) of the motor map, (4) slope of the recruitment curves, and (5) motor threshold. Additionally, the compound motor action potential (CMAP) for the right FDI, using supramaximal electrical stimulation of the ulnar nerve at the wrist, was determined for data normalization purposes in the map volume analysis for this muscle. Two different stimulators were used during the TMS procedures. Stimulation during optimal position locating, motor threshold testing at the optimal position and motor mapping was delivered using a Magstim Rapid (Magstim Company Limited, UK) magnetic stimulator through a 5 cm mean loop diameter figure-of-eight shaped magnetic coil. Stimulation during the recruitment curve procedure and in determining the motor threshold at the Cz was performed using a Magstim 200 (Magstim Company Limited, UK) magnetic stimulator through a 9 cm mean loop diameter circular shaped coil. All assessments were performed during two separate testing sessions, separated by

2 weeks.

Determination of Optimal Position and Motor Threshold

The technique for stimulation was performed as described by

Wassermann et al. (1992). The coil handle was oriented sagittaly, with the





44


handle pointing posteriorly and the figure-of-eight coil situated tangential to the skull (Figure 2-1). Stimulation was delivered over the left hemisphere, contralateral to the muscles of interest (i.e., in the dominant, right arm). With the stimulator set at its maximum output and with the subject relaxed, the 'optimal position' for stimulation was identified and its location recorded in relation to the Cz. The optimal position is defined as the stimulation point that elicits the largest amplitude MEPs. Once the optimal position was determined, motor threshold was assessed in a step-wise fashion. Motor threshold is defined as the lowest stimulation intensity that elicits discernable MEPs in at least 5 of 10 consecutive stimulations using an oscilloscope gain of 200 IpV per cm (Wassermann et al. 1992). To compensate for possible initial heightened arousal levels and/or startle responses, which might affect MEP threshold; several trial stimulating runs were performed prior to the final assessment of motor threshold at the optimal position.

TMS Mapping

A 5 X 5 cm grid was marked on the swim cap and centered around the optimal position (25 points, separated by 1 cm; see Figure 2-1). The stimulator was set at 115% of the motor threshold and five stimuli were delivered to each grid point at a frequency of 1Hz. The EMG responses from these five stimuli/grid point were rectified and averaged online using Viking II nerve conduction software (Nicolet Biomedical, Madison, WI). After all grid positions were stimulated, the grid was extended, as necessary, until the area from which MEPs





45

were elicited was surrounded by stimulated sites that did not elicit MEPs discernible at an oscilloscope display gain of 200 pjV per cm in any muscle. This method ensures that the full extent of the motor map is captured (Triggs et al. 1999). In considering the test-retest reliability of mapping motor representations with TMS, we recognized that changes in motor threshold between testing sessions could occur and might influence the mapping results. Since changes in motor threshold between testing sessions may reflect differences in coil position or corticospinal excitability, we elected to use the same intensity of stimulation (115% of motor threshold determined in testing session 1) for both testing sessions, and to determine the test-retest reliability of both motor maps and motor threshold as individual variables. Recruitment Curve Procedure

The recruitment curve procedure was performed after the mapping procedure, and in a manner previously described by Ray et al. (2002) and Boroojerdi et al. (2001). The subjects were prepped in the same manner as in TMS mapping. Magnetic stimulation was delivered over the Cz using the circular coil. The coil was placed tangential to the skull, with the handle oriented sagittaly and side 'A' facing up to direct the magnetic outflow towards the left hemisphere. Magnetic stimuli were applied at 5% steps between 30% and 100% of the maximum stimulator output. Five stimuli were delivered at each intensity, at a rate of 0.2 to 0.3 Hz. The MEPs produced at each stimulation intensity were rectified and averaged online over the five stimulation trials, as





46

during the mapping procedure. These data were used to construct a stimulusresponse curve, with MEP peak-to-peak amplitude plotted as a function of TMS intensity. The slope of this curve was analyzed for test-retest reliability.











5 cm
Cz
...... 5 CM ........ .......... ...



Stimulating coil







Figure 2-1. Location of the vertex (Cz), interaural and nasion-inion lines,
stimulating grid, and figure-of-eight coil orientation on the head. Motor Threshold Procedure

Motor threshold was studied for two different purposes, and thus was assessed by two different means. As described above, the first purpose for assessing motor threshold was to determine the stimulation intensity to be used during the TMS mapping. For this purpose, motor threshold was determined at





47


the optimal position. The second purpose for assessing motor threshold was to determine the reliability of this measure over multiple testing sessions. We considered that a more precise means of identifying motor threshold for this purpose might be achieved by stimulating with a large circular coil placed over the Cz. As at the optimal position, motor threshold at the Cz was assessed in a step-wise fashion as the lowest TMS intensity to elicit an MEP in at least 5 of 10 consecutive stimulations using an oscilloscope gain of 200 pV per cm. To account for subject adaptation to TMS, motor threshold at the Cz was re-checked at the end of the testing session.

Compound Motor Action Potential

The compound motor action potential (CMAP) was determined for each subject for the purposes of data normalization during the assessment of FDI motor map volume. The CMAP is generated by electrically stimulating a peripheral nerve to produce an MEP in a distal muscle. A MEP produced in this manner represents stimulation of 100% of the motor neurons supplying the muscle of interest. The CMAP procedure was studied in all subjects, and was performed in the following manner. Passive, bipolar surface EMG electrodes were applied over the FDI with a reference electrode located on the dorsal aspect of the hand. A bipolar stimulating electrode was prepared with conducting gel and was placed over the ulnar nerve at the distal-medial aspect of the forearm. Stimulation was delivered at a rate of 0.7 Hz while intensity was





48

gradually increased until the CMAP waveform ceased to increase in amplitude, producing a maximum M-wave.

Data Analysis

Viking II nerve conduction software was used to determine the mean peak-to-peak area of rectified MEPs elicited during the motor mapping and recruitment curve procedures. For motor mapping, the mean area of MEPs elicited at each stimulation site were normalized to that of the stimulus site which produced the largest MEPs. In this manner, the muscle representation area was quantified as the number of stimulation sites that elicited MEPs of area >10% of the MEP area for the stimulation site that produced the largest MEPs. This method of normalization eliminates EMG output due to spontaneous muscle activity unrelated to the stimulation. Map volume was expressed as the sum of the MEP areas, normalized to the CMAP peak-to-peak areas, for the FDI maps. Center of gravity was represented as the maximum amplitude-weighted position, and was calculated as follows: for each stimulating position on the map, the amplitude-weight was computed as the amplitude at that position divided by the sum of peak-to-peak MEP areas recorded for the map. The weight at any stimulating position was interpreted as the proportion of the total map area contributed by that location. For the recruitment curve data, MEP area was plotted as a function of stimulus intensity (i.e., percent of stimulator output). These data were fitted to a linear model. Although the recruitment curve is typically sigmoid in shape, using the linear model does not require the higher





49

plateau values in the curve. This consideration avoids the use of high stimulus intensities to produce plateau values, which result in subject discomfort.

We assessed the test-retest reliability of map area, map volume, optimal position location, recruitment curve slope, and motor threshold across two testing sessions separated by 2 weeks. The intraclass correlation coefficient (ICC) was used to assess reliability. This statistic is the preferred index of reliability, as it reflects both the degree of association and agreement between pre- and posttest session findings (Portney and Watkins 2000). We report the ICC model (C,2), which indicates the reliability of each of the variables when averaged over two sessions. An ICC > 0.75 is generally considered high, while those below 0.75 are indicative of moderate to poor reliability (Portney & Watkins 2000).

Differences between the locations of all four muscle representation CoGs were assessed for each testing session using separate one-way ANOVAs [one factor (muscle) with four levels (APB, FDI, EDC and FCR)]. The Tukey's Honestly Significant Difference test was performed post-hoc to look for differences in the location of each individual CoG from the others. Finally, the ability of motor threshold at the Cz to predict motor threshold at the optimal position was investigated using linear regression.





50

Results

Motor Map Area

The mean number of active positions and intraclass correlation coefficient results for the motor map area of each muscle are shown in Table 2-1. As is indicated by the reliability analysis and Figures 2-2 and 2-3, the muscle representation area remained relatively stable across the two testing sessions. For motor map area, the ICCs were: APB=0.68, FDI=0.60, EDC=0.86, and FCR=0.85. Analysis of the pooled data for all subjects revealed that the EDC had the largest mean area (15.65 active positions), while the representations for the remaining three muscles were similar (APB=9.75, FDI=11.45, FCR=8.5 active positions).

Table 2-1. Comparison of pre- and post-session mean motor map area and ICC
results.
mean area (S.D.)
Muscle pre post ICC
APB 9.0 (4.8) 10.5 (6.7) 0.676 FDI 12.9 (4.5) 10.0 (5.5) 0.599 EDC 15.6 (5.5) 15.7 (6.4) 0.858* FCR 9.0 (6.1) 8.0 (6.8) 0.848*
*Met or exceeded the threshold value for a significant replication of results of ICC > 0.75.

Motor Map Volume: FDI

Motor map volume for the FDI was assessed as an indicator of the

stability of relative size of the MEPs from session to session. Each MEP peak-topeak area was expressed as a percent of that of the CMAP. The average FDI map volume for all subjects remained very stable across the two testing sessions, only changing by 0.9% (Figure 2-4). Map volume differences between






51


subjects were highly variable with large pre- (60.0) and post-session (47.8) standard deviations. These data, however, demonstrated good test-retest reliability within subjects, as demonstrated by a high ICC of 0.85. Sample pre and post FDI map volume outputs from a representative subject are depicted in Figure 2-5.


Mean Motor Map Area








10 Ipre

m8
1U
862
E
4

2

0
APB FDI EDC FCR muscle

Figure 2-2. Comparison of mean motor map area between testing sessions and
by each muscle. Area is expressed as the total number of active
stimulating positions, which elicited an MEP with a peak-to-peak area
>10% of that of the maximum MEP.







52











I"



-I


APB Pre Map-Subject 19 1 APB Post Map-Subject 10



4 3.5







7.5










3 2 1 0 -1 -2 2 1 0 -1 -2






4.5 3.5
c--ant. (cml Dost.-.> --ant. (cm) post.-->

Fg Pr asap-subJctf FDI PotMap-Tubject y






5.5 U 4.5 E E o 0o 4


6.5 ; 5.5 V




7.5 &,5 3,5 2.5 1.5 0.5 -0.5 2 1 0 -1 -2
<--ant. (cm) post.--> <--ant. (cm) post.-->

Figure 2-3a. Pre- and post-session motor maps of the APB and FDI. The y-axis

represents lateral distance from the vertex (Cz). The x-axis represents anterior (+x) and posterior (-x) distance from the

interaural line. Each grid square represents a stimulation position

oriented over the pre-central gyrus. Marked squares indicate

positions that produced MEPs. Although the shape of pre- and post
maps differ, the absolute area is similar.








53













EDC Pre M ap-Subje ct 1 EDC Post M ap-Subject 1










U U o 0




44




-3


45 35 25 15 05 3 2 0
<.-ant. (cm) post.--> <--ant. (cm) post.-->



FCR P. Mp.Subje ct 12 FCRPost M ap.Subject 12





S25

















4 3 2 1 0 -1 -2 4 3 2 1 0
<--ant. (m) past.-> <--ant. (cm) post.,-Figure 2-3b. Pre- and post-session motor maps of the EDC and FCR for

representative subjects. The y-axis represents lateral distance (in cm) from the vertex (Cz). The x-axis represents anterior (+x) and

posterior (-x) distance from the interaural line. Each position on the

grid represents a stimulation position, roughly oriented over the lateral aspect of the pre-central gyrus. Marked squares indicate

positions that produced an MEP with significant peak-to-peak area (i.e., >10% of the maximum MEP). Although the shape of pre- vs.

post maps differ, the absolute area is similar, if not equal.







54










Mean FDI Motor Map Volume 60





50





40
40.4
39.5




30-30





20





10





0
pre post ICC=0.85*




Figure 2-4. Mean pre- and post-session motor map volume for the FDI. Map
volume is expressed as the sum total of MEP peak-to peak areas,
which were normalized as a percent of the CMAP. *Met or exceeded
the threshold value for a significant replication of results of ICC 2
0.75.








55






FDI Pre Map Volume-Subject 6












6-7


6 4-5
4 %CMAP
3 02-3
01-2
2 00-1


65
45 0 25 0
0.5 3 2
4
ant.lpost. distance (cm) 61 5
-35 8 7 lat. distance from Cz (cm) 10

FDI Post Map Volume-Subject 6












78
6-7
5R5

4 % CMAP
03-4
3 02-3
2 01-2 6 00-1

4 -0








10

Figure 2-5. Pre and Post FDI motor map volume in a single representative

subject. The mean MEP peak-to-peak area for each active position is

expressed as a percent of the CMAP peak-to-peak volume. The yaxis represents lateral distance from the vertex (Cz), measured in

cm. The x-axis represents anterior (+x) and posterior (-x) distance

in cm from the interaural line.





56

Center of Gravity

The lateral (y, distance from Cz) and anterior-posterior (x, distance from interaural line) coordinates of the CoG of each motor map were analyzed separately for test-retest reliability of position. Mean location of the CoG and results of the reliability analysis are presented in Table 2-2. The lateral distance

(y) of the CoG remained stable for the APB, FDI and EDC, as demonstrated by high ICCs that ranged from 0.82 (APB) to 0.86 (EDC). The FCR was moderately reliable (ICC=0.69). The anterior-posterior CoG coordinate did not meet the threshold for significant ICC, as these values ranged from 0.37 (APB) to 0.70 (EDC).

We sought to determine if there were CoG location differences between each of the four muscle representations. As demonstrated in Table 2-2 and Figure 2-6, the mean CoG position changed little from pre- and post-session locations, with the exception of the FDI. These positions also followed a predicted somatotopic organization, with the APB and FDI CoGs located lateral to those of the EDC and FCR. There was a trend for differences in the lateral location between each muscle's CoG on the pretest (F-2.3, p=0.084, d=0.084, power=0.60), and significant differences in location of these on the posttest (F-5.2, p=0.003, d=0.182, power=0.91). There were no significant differences in the anterior-posterior location of the CoGs. A post-hoc analysis (Tukey's Honestly Significant Difference) was performed to determine which of the muscle's CoGs differed on their lateral location for the posttest data. This





57


analysis revealed that the APB CoG significantly differed from the location of the EDC and FCR. No other significant locational differences of CoG between muscles were found. See Table 2-3 for post-hoc analysis results on the post-session CoG locations.

Table 2-2. Mean CoG coordinates for all subjects combined and respective ICC
values.
mean lateral M-L Coord. mean ant./post. A-P Coord.
muscle coordinate (s.d.) ICC coordinate (s.d.) ICC
APB 5.41 (0.74) 0.824* 1.17 (0.77) 0.374 FDI 5.11 (0.70) 0.852* 1.25 (0.79) 0.380 EDC 4.88 (0.70) 0.860* 1.05 (0.67) 0.702 FCR 4.73 (0.63) 0.685 1.14 (0.76) 0.530
The lateral coordinate is the distance (in cm) from the Cz, while the anteriorposterior coordinate is the anterior (+x) or posterior (-x) distance from the interaural line. *Met or exceeded the threshold value for a significant replication of results of ICC > 0.75.

Motor Threshold

Motor threshold was assessed while stimulating over the optimal

stimulating position and at the Cz for 19 of the 20 subjects. The data on one subject was removed due to a missing motor threshold value for the post Cz assessment. There was little change in motor threshold at either of these stimulating positions across testing sessions (Table 2-4). The ICCs were high for the optimal and Cz motor threshold at 0.97 and 0.90, respectively. Figure 2-7 presents the average motor threshold at both locations for all subjects combined. Linear regression revealed a relatively low ability for motor threshold values at the Cz to predict motor threshold at the optimal position on pre- (r2=0.19) and post-session (r2=0.31) assessments (Figure 2-8).






58




Mean CoG Position: Pre vs. Post
4.60
PRE POS1
7FCf 4.70
0 EDCI
0 FDI | 4.80
/ APBA
O 4.90

5.00

5.10

5.20

5.30

A 5.40

S5.50
2 1.8 1.6 1.4 1.2 1 0.8 0.6 0.4 0.2 0 <-ant. (cm) post.-->



Figure 2-6. Comparison of mean CoG position for each muscle between pre- and
post-sessions. These positions were relatively similar between preand post-sessions, with the exception of the FDI, which
demonstrated a dramatic anterior-posterior shift. These positions
also followed a relative somatotopic organization, which was
maintained across the two sessions.

Table 2-3. Analysis of differences between individual muscle CoG locations.

Post-hoc Analysis
Post-session locations

mean
difference standard
(cm) error sig.
APB:EDC 0.59* 0.19 0.017 APB:FCR 0.74* 0.20 0.002 APB:FDI 0.38 0.20 0.216 FDI:EDC 0.21 0.19 0.696 FDI:FCR 0.36 0.20 0.279 EDC:FCR 0.15 0.20 0.869
*Significant difference between muscle locations, Tukey's HSD post-hoc analysis.





59




Motor Threshold

65
60 --.
55:::: 58 58
50 45
S 40 42::::5:::::: 3
o 35 o pre
!!5i Ig post
........ .. ..... .............
25
20 15 .
10


0
MT at optimal MT at vertex location of MT assessment

Figure 2-7. Comparison of mean pre- and post-session motor threshold values
obtained during stimulation at the optimal point and Cz. Motor threshold values are described as the lowest TMS intensity that
produced a discernable MEP at a display gain of 200 qV per cm on 5
out of 10 trials. These data represent the mean motor threshold obtained at both locations for all subjects combined. There was
very little mean change in motor threshold between pre- and postsession testing at either location. Motor threshold at the optimal
point was assessed using a Magstim Rapid magnetic stimulator with a figure-of-eight coil, while motor threshold at the Cz was assessed
using a Magstim 200 stimulator with a circular coil. Absolute
differences in motor threshold at these two locations, therefore, is
at least partly related to different output intensities of the two
different stimulators, with the Magstim 200 having a higher output
intensity.







60




Pre-MT
90

80 A R2 0.1886

70

S60 U50 E A O30 20 10 0
0 10 20 30 40 50 60 70 80 MT at vertex (% stim. output) Post-MT
70 60
R2 = 0.3108 50


40 30 S20 10



0 10 20 30 40 50 60 70 80 90 MT at vertex (%stim. output)



Figure 2-8. Relationship between motor thresholds (MT) obtained at the Cz and
optimal point for pre- and post-sessions. Motor threshold values are
described as the lowest stimulus intensity that produced a
discernable MEP. Each data point represents the intersection of
motor threshold at the Cz with motor threshold at the optimal point for each individual subject. Linear regression revealed a low ability
of motor threshold at the Cz to predict motor threshold at the
optimal position, as noted by the low r2 values.





61

Recruitment Curves

When analyzing the recruitment curve results, the mean peak-to-peak MEP area was plotted against TMS intensity and fitted with a line of best fit. Statistical analyses were, therefore, based upon the slope of this line for each muscle and in each subject. The mean recruitment curve slopes and ICC results are presented in Table 2-4. The EDC, FCR and FDI demonstrated good testretest reliability as the ICCs of the recruitment curve slopes for these muscles ranged between 0.85 (FCR) and 0.91 (EDC). The APB slope data was highly variable, however, resulting in poor test-retest reliability. This high variability was the result of large differences in the raw MEP peak-to-peak area values within and between subjects. Representative pre and post-session recruitment curves from individual subjects are presented in Figure 2-9a and 2-9b. Table 2-4. Test-retest reliability analysis of recruitment curve slope by muscle.

Mean Slope (S.D.)
Musde pre post ICC
APB 0.16 (0.23) 0.21 (0.23) 0.255 EDC 0.11 (0.09) 0.12 (0.12) 0.910* FCR 0.06 (0.06) 0.06 (0.06) 0.853* FDI 0.25 (0.27) 0.22 (0.19) 0.892*

*Met or exceeded the threshold value for a significant replication of results of ICC .75.








62





APB Recruitment Curves-Subject 14


7 6 5




A --- pre
a- -- post
W3



2



1 A




30 35 40 45 50 55 60 65 70 75 80 85
TMS Intensity (%)




FDI Recruitment Curves-Subject 20 45 40 35 30


25

------post
S20


15 10






30 35 40 45 50 55 60 65 70 75 TM S intensity (%)



Figure 2-9a. Representative pre- and post-session recruitment curves for the

APB and FDI in Subjects 14 and 20, respectively. MEP peak-topeak area was plotted as a function of TMS intensity (% of

stimulator output).







63




EDC Recruitment Curves-Subject 13 120 100



80


-B-E- EDCpre
60 -l__ --EDCpost
I 60


40 20 0
30 35 40 45 50 55 60 65 70 75 80 85 90 95 100 TMS intensity (%) FCR Recruitment Curves-Subject 20 12 10



8
E
a0 -pre
6
--+--post


4


2


0
30 35 40 45 50 55 60 65 70 75 TMS intensity (%) Figure 2-9b. Representative pre- and post-session recruitment curves for the
EDC and FCR in Subjects 17 and 20, respectively. MEP peak-topeak area was plotted as a function of TMS intensity (% of
stimulator output).





64

Table 2-5. Summary of ICCs for all TMS assessments. Assessment APB FDI EDC FCR Map Area: moderate* moderate good good (.676) (.599) (.858) (.848) Recruitment Curve slope: poor good good good (.255) (.892) (.910) (.853) CoG, y-coord.: good good good moderate (.824) (.852) (.860) (.685) CoG, x-coord.: poor poor moderate moderate (.374) (.380) (.702) (.530) Map Volume: good (.846)

MT-optimal position: good MT-vertex: good (.971) (.899)
*Qualitative scores are based on the following scale of ICC values: 20.75 = good, 0.50-0.74 = moderate, <0.50 = poor reliability (Portney and Watkins 2000). Actual ICC values are included parenthetically. Summary of Results

In general, the results of this experiment support the hypotheses.

Measurement of motor map area demonstrated moderate to good test-retest reliability, with better reproducibility in the forearm muscle representations as

compared to the intrinsic hand muscles. The FDI motor map volume assessment was highly reliable. We found good test-retest reliability for locating the CoG, but only in regard to its lateral distance (y-coordinate) from the Cz. The anterior-posterior (x-coordinate) location of the CoG demonstrated poor to moderate test-retest reliability. When assessed at the optimal position or over the Cz, motor threshold was highly reliable across test sessions. Finally, with the exception of the APB, recruitment curve slope demonstrated high test-retest reliability. The poor reliability for the APB recruitment curves was likely due to





65

high within-subject differences in the peak-to-peak MEP area during this procedure.

Discussion

The results of this study demonstrate that TMS measures of motor

representation size, organization and excitability are reasonably reliable across multiple testing sessions. Although we did not find large changes in motor map area, test-retest reliability of this assessment may be dependent upon the muscle being studied, as map area was more reproducible in forearm muscles compared to intrinsic hand muscles. The results also indicate that motor map volume is a reliable measure of muscle representation excitability, especially when the evoked muscle responses are normalized to a CMAP. This process of normalization compensates for variations in the size of MEPs across testing sessions, which may occur due to differences in electrode placement or subject arousal. The locations of the motor maps, as measured by map CoG, followed a predicted somatotopic orientation that remained relatively stable in the mediallateral direction across testing sessions. Motor threshold was reliable when assessed at the optimal position and Cz, indicating that corticospinal excitability remains stable across testing sessions in normal subjects. Finally, the results demonstrate that recruitment curves provide a reliable measure of the inputoutput properties of muscle representations. Recruitment curves may lack sufficient reliability, however, if large differences in MEP size occur across testing sessions, as was observed for the APB curves. The following sections provide an





66

in-depth discussion on the study findings for each TMS measure employed in this investigation.

Motor Mapping

We found that the test-retest reliability of motor map area was better for forearm muscle representations than for intrinsic hand muscle representations. Other reports have also found map areas of intrinsic hand muscles to be only moderately reliable (Mortifee et al. 1994), with nearly significant differences (F-2.4, p-0.094, n=8) in map area across multiple testing sessions (Uy et al. 2002). In the following discussion on motor mapping, we first deal with issues related to the lower reliability of intrinsic hand muscle representations, and then discuss why motor maps of forearm muscle representations are more reproducible across testing sessions.

There are several potential reasons, both neurophysiological and

methodological, that map areas for the APB and FDI were only moderately reliable. These muscles may be more affected by these factors than the EDC and FCR, given the greater complexity of intrinsic hand muscle representations. First, M1 representations for these muscles might change due to spontaneous variations in the M1 cortical outflow. Studies using reversible deafferentation with a blood pressure cuff confirm that rapid alterations in MEP amplitude (Brasil-Neto et al. 1992a) and motor map area (Ridding & Rothwell 1997) occur within minutes of the onset of deafferentation. Although the mechanisms for motor map changes in ischemic deafferentation is likely different than in the





67

present study, these findings indicate that the modulation of motor outputs from M1 occur quite rapidly. The cortical maps of intrinsic hand muscles are presumably more subject to these rapid changes than forearm muscles due to the higher number of corticospinal neurons devoted to hand muscle representations (Phillips & Porter 1977, Wassermann et al. 1992).

Daily activity patterns have also been shown to affect the size of motor maps of intrinsic hand muscles. In a study on Braille proofreaders, PascualLeone et al. (1995) found that the size of FDI motor maps varied according to activity level. The maps were larger on workdays (i.e., concentrate hand use) than on weekends off of work. In the present study, we found that nearly half of the subjects typed on a computer for a significant portion of their workday. As with the Braille proofreaders, the intrinsic muscle maps in our subjects may have been influenced by variations in the time spent typing during the 2 weeks between testing sessions. The wrist flexors and extensors (EDC and FCR) maps would presumably be less subject to the effects of concentrated typing, as these muscles are less involved in the performance of the varied and repetitive movements required during this activity. Our finding of reduced reliability of intrinsic muscle map area demonstrates the limits for obtaining reproducible motor maps across multiple testing sessions. Future investigations, which attempt to use changes in map area as a marker for plasticity within the motor system, should consider that alterations in these maps may occur secondary to spontaneous fluctuations in the cortico-motoneuronal output or to variations in





68

general activity patterns over time. Multiple baseline assessments would account for such variability prior to the introduction of a controlled behavioral intervention. Despite the apparent variability in intrinsic muscle maps, the investigation of these representations should continue, as they are elemental in the performance of highly skilled fine-motor tasks.

EMG-related issues also may affect motor map area. In a study on the reliability of TMS to map the human masseter muscle, McMillan et al. (1998) found that maps were reproducible, but only if the electrodes were left in place between testing sessions. When the electrodes were removed, map area was significantly different between testing runs. The results of the present study may be influenced by electrode placement as in the McMillan et al. (1998) study. We placed electrodes over the muscle bellies in parallel with the muscle fibers and with an inter-electrode distance of the recommended 20 mm (Hermens et al. 2000). Despite our attention to inter-electrode distance, we found that electrode placement was difficult to replicate across testing sessions separated by such a long time period. Again, the greater complexity of hand muscle representations (Phillips & Porter 1977) combined with between-sessions electrode placement differences, may have resulted in higher variability in motor maps across testing sessions for these muscles. Intrinsic hand muscles are smaller and have a higher density of motor units than the forearm muscles. These muscles also have a comparatively more complex neural representation than the forearm muscles; with a greater number of cortico-motoneuronal connections (Brasil-Neto et al.





69

1992a, Phillips & Porter 1977). If the electrodes were over different populations of motor units, then the corresponding representations would likely vary more in the APB and FDI than in the EDC and FCR. To account for variability secondary to electrode placement in the small hand muscles, precise marking and recording of electrode placement is necessary. Needle or fine wire electrodes could also be used to investigate single motor units, which are not contaminated by the responses of other motor units (Turker 1993). Although these electrodes offer greater precision, their invasiveness and potential to cause subject discomfort should also be considered.

Map Volume

Map volume was assessed in the FDI as the sum of all MEP peak-to-peak areas, each of which was expressed as a percent of the CMAP. The reproducibility of FDI map volume in this investigation was similar (Mortifee et al. 1994) or better (Miranda et al. 1997, Uy et al. 2002) than in previous studies. We found map volume to be highly reliable between two testing sessions; which was an interesting result given that FDI map area was only moderately reliable. The higher reliability for FDI map volume was likely related to the fact that the MEPs were expressed as a percent of the CMAP for this measurement. Normalizing the raw waveforms in this manner corrects for variations in the size of the MEPs, which may occur due to differences in electrode placement across testing sessions. Because map volume is normalized to a known muscle response, this measure provides a good assessment of corticospinal excitability,





70


which is less subject to errors in electrode placement than the assessment of map area.

Center of Gravity

Separate reliability analyses were performed on the lateral (y) and anterior-posterior (x) coordinates of the CoG, each of which represents the amplitude-weighted center of the motor map. The relationship of the different CoG locations, especially in the y-direction, may be used to determine the somatotopic organization of the motor cortex. In the present study we found that the CoGs did fall in a predicted somatotopic orientation (Penfield & Rasmussen 1950) that was maintained across the two testing sessions. The lateral distance from the Cz (y) was highly reliable for all muscles, with the exception of the FCR, which demonstrated a moderate ICC. Several potential factors may have resulted in the lowered test-retest reliability of the FCR lateral CoG coordinate. Coil position during TMS mapping over the motor cortex can influence the size of MEPs and concurrently, the size, shape and location of the map (Mills et al. 1992, Miranda et al. 1997, Pascual-Leone et al. 1994). We mapped all 4 muscles simultaneously; with the figure-of-eight coil handle positioned parallel to the midsagittal with a backward flowing inducing current. This position has been shown to produce optimal responses in the APB and FDI, while allowing for easier replication of coil position between stimulus sites (Pascual-Leone et al. 1994). The optimal coil position for the FCR, however, was previously shown to be at a 45-degree angle to the midsagittal (Pascual-Leone et





71

al. 1994). This orientation produces a maximal induced current that flows at right angles to the central sulcus (Mills et al. 1992). Horizontal interneurons are oriented perpendicular to the central sulcus (Pascual-Leone et al. 1994) and are preferentially excited by TMS (Hallett 2000, Mills et al. 1992). A less than optimal coil orientation for stimulating the FCR motor representation could, therefore, affect the location and size of the map, as well as the size of the MEPs elicited. Indeed, we found the FCR to have an average maximum MEP area that was 3 times less than the other three muscles, a smaller average map size, and the highest variability in CoG location between sessions. The smaller MEP size is likely to be the most influential factor in the decreased reliability of the ycoordinate in the FCR. At a less than optimal coil position, the stimulation that reaches the representation for this muscle could be at or just slightly above the threshold for excitation. Since variability of the MEP response is inversely related to stimulus intensity, the variability in the size of FCR MEPs is more likely to be affected by rapid and spontaneous fluctuations in corticospinal and segmental motor neuron excitability levels (Kiers et al. 1993). The effect of these fluctuations is that stimulation sites may produce a positive response on one occasion and a null response on another (Weber & Eisen 2002). This variability would affect the distribution of the map, and therefore, the location of the CoG. Mapping the FCR at higher intensity and/or with the coil oriented at 45 to 50 degrees from the midsagittal should alleviate the MEP variability and increase the reliability of CoG location for this muscle. Based upon our findings and those of





72

Pascual-Leone et al. (1994), future investigations that assess CoG of the FCR should employ separate coil positioning and testing in order to obtain optimal stimulation of this muscle's cortical representation.

Interestingly, reproduction of the CoG in the anterior-posterior direction was fairly low for all four muscles. This was an unexpected finding, as the primary motor cortex is distributed in a medial to lateral rather than an anterior to posterior direction. Coil orientation may again play a role in the inter-session variability of the CoG x-coordinate. In normal subjects, map shape often mirrors the pattern of induced current flow, being elongated along the axis of the coil (Wilson et al. 1993). As a result, there is a larger area over which the CoG may fall in the anterior-posterior direction when the coil handle is oriented parallel to the midsagittal, increasing the room for variation of this measure across sessions. Our finding poor reproducibility of CoG in the anterior-posterior direction contradicts a previous report that reported no significant change, in 3 subjects across three testing sessions (Miranda et al. 1997). Nonetheless, we found the mean change in CoG location for each muscle, both in the x and y directions, to still be less than what has been reported in the only other studies on CoG replication (Miranda et al. 1997, Uy et al. 2002). Further investigations should consider the effect of coil orientation on the distribution of map CoGs. Motor Threshold

Motor threshold, or the smallest intensity level to produce a discernable MEP, demonstrated high test-retest reliability when assessed at the optimal





73


position and over the Cz. These findings indicate that the threshold for excitation of the corticospinal system remains relatively constant across multiple testing sessions in normal subjects. The fact that motor threshold was stable is relevant to the motor map area findings, as changes in motor threshold could affect the apparent size of these maps (Thickbroom et al. 1998). In this regard, we suggest that small changes in especially the APB and FDI map areas across testing sessions were not the result of fluctuations in motor threshold. Recruitment Curves

The slope of the recruitment curves was found to have high test-retest reliability, with the exception of the APB, which displayed very low reliability. The APB is especially susceptible to replication of electrode placement because of its small size and proximity to other thenar muscles. As with the motor maps, this may produce high variability in the size of MEP responses between sessions. When comparing the mean slope of the recruitment curves of all subjects, the APB had the largest mean difference in slope and the largest standard error of the mean. To adjust for the variability in MEP between sessions, these data should be normalized as a percent of the CMAP. This procedure would correct for differences in the raw MEP sizes from session to session, as was demonstrated in the FDI map volume data.

The generation of recruitment curves involves measuring the excitability of neurons other than those in the core region of excitability that is activated at threshold. These neurons have a higher motor threshold, either because they





74


are intrinsically less excitable by TMS or because they are distant from the point of stimulation (Hallett 2001). Recruitment curves have been found to be steeper in intrinsic hand muscles (Chen 2000) when compared to leg muscles (Devanne et al. 1997). Indeed, we found the mean slope of the APB and FDI to be slightly more elevated than in the forearm muscles. The different input-output characteristics of these muscles suggest that the slope of a recruitment curve is related to the amount of corticospinal input to a particular muscle, with the intrinsic hand muscles being more easily excited by TMS.

The test-retest analysis of recruitment curve slope indicates that this is a reliable measure across test sessions, especially when the data is normalized to the maximum MEP area. We chose to use a linear model to represent these data even though the responses produced a sigmoid-shaped curve. The linear model might be the most practical way to describe the data. Ray et al. (2002) found that the s-model produces impractical results, returning inflated curve values at higher stimulus intensities. Similarly, Carroll et al. (2001) noted that linear slope was more reliable than the peak slope obtained using a sigmoid function. Finally, the linear model does not require a plateau value in the curve. This avoids using high stimulus intensities that result in subject discomfort. Strengths and Limitations

Upon review of the literature, only five studies have explicitly examined the reliability of TMS (Carroll et al. 2001, McMillan et al. 1998, Miranda et al. 1997, Mortifee et al. 1994, Uy et al. 2002) only one of which addressed the





75


reproducibility of recruitment curves (Carroll et al. 2001). Although one might argue that this is a sufficient body of work to establish test-retest reliability, there are methodological and statistical issues in these studies that need to be strengthened. The present study attempted to address some of these issues by using a more rigorous statistical analysis, coupled with multiple different assessments to provide parallel information on studying the nervous system with TMS.

We assessed the reproducibility of numerous TMS measures with the ICC, as it is the most robust and preferred method to evaluate test-retest reliability (Portney and Watkins 2000). The historical approach to testing reliability involves the use of the Pearson correlation (r), which provides information on the degree of association between two covariates. The Pearson r does not, however, provide a measure of agreement between the variables. The ICC is a more powerful assessment of reliability as it addresses association and agreement among multiple scores. The ICC essentially describes the proportion of the variance within a data set that is attributable to each of the independent factors in the data set. Mortifee et al. (1994) was the only group to use the ICC to describe test-retest reliability of motor maps. Using an ANOVA, Uy et al. (2002) found no significant (F=2.4, p=0.094) change in map area across four testing sessions of varying inter-session testing intervals. McMillan et al. (1998) also employed an ANOVA, while Miranda et al. (1997) used descriptive statistics to characterize reliability during mapping. The use of an ANOVA or descriptive





76

statistics alone does not address the degree of association and agreement between each factor separately and for each individual subject. The results of these studies do not clearly indicate if there were differences between individual sessions or individual muscle maps across sessions. The magnitude of reliability of these individual variables was also not demonstrated. In the present investigation, we found generally high ICCs for the map area, map volume and recruitment curve linear slope, which were similar or higher than ICCs reported by Mortifee et al. (1994) and Carroll et al. (2001). A large sample size was utilized in the present study, which strengthens the statistical power of the reliability analysis.

The use of several assessments in four muscles helped to provide parallel information about the reliability of TMS procedures. Based on our findings, recruitment curve slope and motor threshold over the Cz might be the most reliable measures to assess changes in cortical excitability. Map area was most reliable in forearm muscles and is dependent upon replication of electrode placement across testing sessions. Thickbroom et al. (1998) suggests that changes in motor representations are probably a function of increases or decreases in motor cortex excitability, and should therefore be characterized by recruitment curves or motor threshold assessment. Our finding that FDI map volume was highly reproducible, despite only moderate map area reliability supports the notion that changes in excitability may influence map area, as map volume is dependent upon the sum total of MEP areas regardless of the number





77

of active positions. Although the map area for the APB and FDI did not meet threshold for reliability, the lateral orientation of CoG for these muscle representations did. In addition, all muscle CoGs were organized in a predicted somatotopic orientation that was maintained across testing sessions.

A limitation of the present study may be that the same stimulus intensity was used during the mapping procedure on the pre- and posttest sessions. Transient differences in motor threshold (i.e., the level of cortical excitability) could have occurred between the two sessions. Such a change would influence map area, volume, and possibly the location of CoGs, which could account for some of the occasions of reduced reliability. We did not, however, see large changes in motor threshold at the optimal position, so this may have had only a small effect on the findings.

The results of this study suggest that motor map area and volume, center of gravity, motor threshold, and recruitment curve generation a reasonably reliable measures. Despite finding APB and FDI map area to be only moderately reliable, the inter-session reproducibility was similar to other reports (Miranda et al. 1997, Mortifee et al. 1994, Uy et al. 2002). Future studies should employ a rigidly fixed inter-electrode distance to control influences of cross-talk, as well as skin marking that would last over the course of the study. Adequate data normalization, appropriate positioning of the coil angle relative to the midsagittal, and multiple assessments of motor threshold throughout the testing session will also increase the reliability of TMS investigations that occur over multiple testing





78

sessions. In addition, the results demonstrate that there is at least a minimal degree of unexplained variability in especially intrinsic hand muscle representations. Future investigations should, therefore, establish the test-retest reliability and limits of normality of the TMS measures prior to study initiation. Given these considerations and based upon these results, TMS could be very useful in studying stable characteristics and/or activity-related changes of the human motor nervous system.













CHAPTER 3
EXPERIMENT II: ACTIVITY-DEPENDENT PLASTICITY IN STROKE

More than half of the 730,000 Americans who sustain a stroke each year are left with chronic deficits in upper extremity function (Taub et al. 1999). Although physical rehabilitation may assist the stroke survivor in learning to compensate for these functional deficits, traditional therapeutic approaches have demonstrated limited efficacy to remediate motor skill in the affected arm and hand (Dobkin 1995). Research on a newly developed treatment, however, is providing initial evidence that recovery of skilled upper-extremity function is possible for individuals in the chronic post-stroke phase. This intervention, known as constraint-induced (CI) movement therapy, employs intensive practice of motor skills for several hours per day over a 2-week period. To encourage further use of the hemiparetic limb, CI therapy participants also wear a constraining mitt on the unaffected hand. The combination of the intensive practice and constraint is believed to reverse the process of learned disuse of the hemiparetic limb (Taub et al. 1993, Taub et al. 1999). Several recent reports indicate that CI therapy results in significant improvements in movement speed, quality and coordination, as well as increases in the amount of use of the affected limb (Kunkel et al. 1999, Miltner et al. 1999, Taub et al. 1993, Taub et al. 1999). In addition, a two-year follow-up study indicated that these



79





80

improvements were maintained well beyond the therapeutic intervention (Taub et al. 1993).

Animal models of stroke indicate that intensive treatment programs similar to CI therapy alter the organization and excitability of neural representations for movement (Friel et al. 2000, Friel & Nudo 1998, Nudo 1996, Nudo et al. 1997). For example, in squirrel monkeys with surgically induced cortical lesions, forced affected upper limb use caused an expansion of the cortical representations for muscles involved in massed task practice (Nudo & Milliken 1996). Accordingly, one tenet of CI therapy is that the intensive treatment regimen induces activitydependent plasticity in the affected primary motor cortex (Ml), and that the changes in the organization and excitability of these muscle representations subserve recovery of skilled movement (Taub et al. 1999). Evidence from human neurophysiology studies indicates that massive M1 reorganization does occur following stroke (Cicinelli et al. 1997; Cramer & Bastings 2000, Cramer et al. 1997, 2000, 2001a, 2001b; Levy et al. 2001; Liepert et al. 1998b, 2000, Nelles et al. 1997; Rossini et al. 1998; Traversa et al. 1998; Weiller C 1992; Zemke & Cramer 2002), however, the relationship between these changes and recovery of specific motor skills requires further investigation.

Using transcranial magnetic stimulation (TMS), Liepert and colleagues (1998, 2000) demonstrated that the motor representation of the hemiparetic hand in chronic stroke survivors was reduced in size as compared to the intact hemisphere. Following a course of intensive movement therapy, the cortical map





81

of the hemiparetic hand expanded into adjacent areas, increasing significantly from the pre-treatment testing. Similar findings have been demonstrated in studies employing electroencephalography (EEG), positron emission tomography (PET), and functional magnetic resonance imaging (fMRI) (Kopp et al. 1999, Levy et al. 2001, Nelles et al. 2001). Although these studies and techniques have provided preliminary evidence for M1 neuroplasticity following stroke, the results are not clearly related to improvements in function. In other words, there may be M1 changes that occur as a person is progressing towards recovery, but what is the relationship between this neuroplasticity and improvements in the person's repertoire of skills? One major limitation of the existing neuroimaging work is that limited or no behavioral measures were used to support the notion that brain reorganization underlies functional recovery. Despite identifying cortical changes, Nelles and colleagues (2001) did not find any significant improvement in use of the hemiparetic limb. Liepert et al. (1998) and Liepert et al. (2000) demonstrated a marked increase in the cortical representation of the affected hand subsequent to therapy. The only behavioral measure used, however, was a subject questionnaire. No objective measures of functional recovery were performed. Levy and colleagues' (2001) pilot study provides support for the theory of activity-induced cortical reorganization as a mechanism for improved use of the hemiparetic limb. Subjects in this experiment demonstrated modest improvements in speed and quality of movement, strength, and real world use of the affected arm (Levy et al., 2001). These





82


results were obtained in only two subjects, which limits the ability to generalize from this study.

Of the several neurophysiological techniques available to investigate the brain, TMS is perhaps the best suited to track plastic changes in the motor cortex. In mapping studies, TMS is used to indirectly excite corticospinal neurons within Ml. In this manner, TMS can be used to outline the location, size and excitability of the motor representation subserving a particular muscle. The advantage of using TMS over fMRI and PET is that this technique maintains high temporal resolution, activates both excitatory and inhibitory inputs to the corticospinal tract, and does not require the subject to produce any movement. Since the introduction of TMS, numerous studies have been performed to investigate the somatotopic organization of M1 (Aimonetti et al. 2002, BrasilNeto et al. 1992, Roricht et al. 1999, Wassermann et al. 1992), stimulusresponse characteristics of the corticospinal system (Boroojerdi et al. 2001, Devanne et al. 1997, Ray et al. 2002, Ridding & Rothwell 1997, Thickbroom et al. 1998), and alterations in the motor system following skill practice (Liepert et al. 1998a; Pascual-Leone et al. 1995a, 1995b) or following stroke (Cicinelli et al. 1997; Liepert et al. 1998b, 2000, 2001; Traversa et al. 1997; Trompetto et al. 2000).

In addition to providing information about the size and extent of M1 representations, TMS may also be used to track the expansion of a representation into adjacent cortical territories. Expansion of motor





83


representations may be assessed by tracking movement of the center of gravity (CoG). The CoG represents the amplitude-weighted center of a motor representation (Wassermann et al. 1992), and is thought to be the low threshold area where corticospinal neurons projecting to the muscles of interest are most concentrated (Escudero et al. 1998). Evidence from the animal literature demonstrates that representations "invade" adjacent and perhaps less active cortical areas under the conditions of massed practice (Nudo & Milliken 1996), resulting in a change of the CoG location. A similar process may occur in human stroke following CI therapy, although the existing literature is equivocal on the direction and relevancy of a shift in representation location.

One additional benefit of TMS is that it may be used to study precise changes in corticospinal excitability. Several TMS studies in stroke survivors have shown that the level of excitability is often depressed in the damaged hemisphere relative to the undamaged hemisphere (Trompetto et al. 2000). Stroke survivors with a high threshold to stimulation and/or absence of motor evoked potentials (MEPs) from the affected hemisphere; typically have a poor functional recovery. Individuals with a lower threshold to stimulation in the damaged hemisphere, however, demonstrate a better functional outcome (Trompetto et al. 2000). To investigate changes in corticospinal excitability, TMS may be used to characterize the stimulus-response properties of the damaged motor cortex. Also known as stimulus-response curves, the recruitment curve represents how MEP amplitude increases as a function of stimulus intensity. This





84

measurement is thought to assess the excitability of neurons other than those in the core neuronal population that is activated at MT (Chen 2000, Hallett 2000). The threshold for these neurons is higher either because they are less excitable or because they are further from the site of stimulation (Hallett 2000). Assessment of recruitment curves following skill practice/learning offers another means to study plasticity in the nervous system.

The purposes of this study were three-fold. First, we sought to

investigate the impact of CI therapy on motor cortex representations for two intrinsic hand muscles and two forearm muscles in the affected upper extremity. By examining four muscles with different functions, this investigation demonstrates which of the muscle representations are most altered by CI therapy. In addition, multiple neurophysiological measures were employed to provide a comprehensive description of how the organization, location and excitability of M1 representations are altered by intensive practice. The second purpose of this study was to determine the relationship between any neurophysiological changes and specific improvements in motor skill as well as real-world use of the affected arm and hand. Finally, because our subjects naturally fell into either a high functioning or low functioning category, we sought to explore differences in neuroplasticity and recovery of function in these sub-groups. Based upon the animal literature and Liepert et al.'s findings, we predicted that motor maps would increase in area and volume, the slope of the recruitment curve would increase, and that the threshold for excitation of the





85

motor cortex would decrease following CI therapy. In all cases of plasticity, we predicted that such neurophysiological changes would be correlated with improvements in motor skill.

Methods

Subjects

The participants were 23 individuals (10 female, 13 male) between the

ages of 49 and 83 years (mean 65.8 10.3 years), who were 10 months to 10.75 years post-stroke (mean 4.0 3.3 years) upon entering the study. Demographic details of these subjects are presented in Table 3-1. These individuals met the following inclusion criteria: history of a clinically diagnosed stroke, met minimal active movement criteria in the affected fingers and wrist (see below); ability to follow simple instructions, a score of 24 or higher on the Mini Mental State Exam; the ability to sit independently without back or arm support for 5 minutes; the ability to actively participate for 6 hours of therapy without long rest periods; and passive range of motion of all upper extremity motions of at least half the normal range. These criteria have been demonstrated as effective parameters to select appropriate candidates for an intensive therapy program (Taub et al. 1993). Exclusion criteria were: history of seizures and/or epilepsy; presence of pacemaker or other metal devices in or around the head and upper thorax; tactile/proprioceptive sensory deficit; presence of major depression or other psychiatric disorder; and history of other neurological disease (i.e., multiple sclerosis). These subjects were classified as either high or low functioning based





86

upon specific motor criteria. High functioning subjects demonstrated the ability to actively move the wrist through at least 200 of flexion-extension range and the thumb and fingers for 100 of active flexion-extension range at the MP and PIP joints. Low functioning subjects were only able and required to demonstrate at least a trace amount of active wrist extension and at least trace extension in any two fingers at the MCP and IP joints. These motor criteria were previously established as an effective means to categorize hand function in individuals with stroke (Taub et al. 1999). Based upon these criteria, 12 participants were classified as "high functioning" and 11 participants as "low functioning". All aspects of participant recruitment and testing met the approval of the local Institutional Review Board prior to initiation of this project. Each participant signed a written informed consent prior to his or her involvement with any of the study procedures.

General Study Procedure

All of the participants underwent both TMS and behavioral assessments

prior to the study intervention. These subjects then participated with 2 weeks of CI therapy. TMS and behavioral assessments were again administered immediately following completion of the 2-week therapy program. Further details of the study intervention and testing procedures are provided below.





87


Table 3-1. Study participant demographics including age, gender, involved hemisphere, affected limb (dominant or non-dominant), and time since stroke.
Age Gerder Side of S9oke Donirnce Tire since sroke (yrs)
1 68 M Rbrain Nocdom 4.83 2 52 M L brain Dorninat 1.75 3 72 M L brain Dorrinart 10.33 4 83 F R brain Non-dom 2.5 5 65 M L brain Dorrinart Z83 6 76 F R brain Noxdom 0.83 7 75 F Lbrain NorKbm 5.5 8 67 M Lbrain NlnxIom 2
9 76 M L brain Doninant 1.83
10 80 F L brain Donirnat 1
11 49 F L brain Donirat 2.83 12 54 M L brain Doinant 4.5
13 78 F Rbrain Non-om 10.75 14 71 F L brain Nondob 2.16 15 60 M Rbrain Nor~xom 8.5 16 59 M Rbrain Nondom 5.67 17 74 M L brain Dnrrinant 1.16 18 59 M Rbrain Noniom 2.08 19 70 M Rbrain Noicxn 0.75 20 54 F Rbrain Dorrinart 1.42 21 49 M L brain Dorrinart 7.25 22 65 F R brain Dornirant 1.83 23 58 F R brain Non-dom 10.16
Mean age: 10 Ferrales 11 R Brain 12 Doninart Mean tire since C'A
65.8 (10.3) yeas 4.0 (3.3) years





88

TMS Procedures

Subject preparation for TMS testing

The participant was comfortably seated in a semi-reclined modified dental chair with a pillow support placed beneath the affected limb. Passive bipolar surface electrodes were applied over the first dorsal interosseous (FDI), abductor pollicis brevis (APB), extensor digitorum communis (EDC), and flexor carpi radialis (FCR) muscles in the affected limb in a belly-tendon arrangement. Correct placement of the electrodes was verified by asking the subject to maximally contract the muscle while the investigator monitored for an EMG output amplitude of approximately 0.4 mV, or by palpation when the subject was unable to voluntarily activate a muscle. The inter-electrode distance was fixed at 20 mm for all muscles. EMG signals were filtered with a bandpass set at 2-10 kHz, rectified, and amplified with a Viking II Electromyograph (Nicolet Biomedical, Madisson, WI). Audio feedback from the electromyograph was monitored to ensure muscle relaxation during the testing session. A latex swim cap was placed on the participant's head so that a coordinate system could be clearly created. The vertex (Cz) was marked as the intersection of the nasioninion and interaural lines. Measurement of these lines (in cm) was recorded to ensure consistent location of the Cz across testing sessions. All TMS stimulation points were recorded in reference to the Cz (see Figure 3-1).





89

General TMS testing procedure

Five primary TMS variables were investigated: (1) size (area) of motor maps, (2) volume of motor maps, (3) location of the center of gravity (CoG) of the motor map, (4) recruitment curves, and (5) motor threshold. Two different stimulators were used during the TMS procedures. Stimulation during motor mapping was delivered using a Magstim Rapid (Magstim Company Limited, UK) magnetic stimulator through a 5 cm mean loop diameter figure-of-eight shaped magnetic coil. Stimulation during the recruitment curve procedure and in determining the motor threshold at the vertex was performed using a Magstim 200 (Magstim Company Limited, UK) magnetic stimulator through a 9 cm mean loop diameter circular shaped coil. All assessments were performed during two separate testing sessions, separated by 2 weeks. Determination of optimal position and motor threshold

The technique for stimulation was performed as described by

Wassermann et al. (1992). The coil handle was oriented sagitally, with the handle pointing posteriorly and the figure-of-eight coil situated tangential to the skull. Stimulation was delivered over the affected hemisphere, contralateral to the muscles of interest (i.e., in the affected upper extremity). With the stimulator set at its maximum output and with the subject relaxed, the "optimal position" for stimulation was identified and its location recorded in relation to the vertex. The optimal position is defined as the optimal stimulating point for eliciting the largest amplitude MEPs. Once the optimal position was





90

determined, motor threshold (MT) was assessed in a step-wise fashion. MT is defined as the lowest stimulation intensity that elicits discernable MEPs in at least

5 of 10 consecutive stimulations using an oscilloscope gain of 200 pV per cm (Wassermann et al. 1992). (Wassermann et al. 1992). To compensate for possible initial heightened arousal levels and/or startle responses, which might affect MEP threshold, several trial stimulating runs were performed prior to the final assessment of MT at the optimal position. TMS mapping

A 5 X 5 cm grid was marked on the swim cap and centered at the optimal position (25 spots, separated by 1 cm; see Figure 3-1). The stimulator was set at 115% of the MT and five stimuli were delivered to each spot at a frequency of 1Hz. The EMG responses from these five stimuli/spot were averaged online using Viking II nerve conduction software (Nicolet Biomedical, Madison, WI). After all grid positions were stimulated, the grid was extended, as necessary, until the area from which MEPs were elicited was surrounded by stimulated sites that did not elicit MEPs discernible at an oscilloscope display gain of 200 IpV per cm in any muscle. This method ensures that the full extent of the motor map is captured (Triggs et al. 1999).

Recruitment curve procedure

The recruitment curve procedure was performed following the mapping procedure, and in a manner previously described by Ray et al. (2002) and Boroojerdi et al. (2001). Due to technical issues, the recruitment curve




Full Text
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID ERWXVJ3PN_2YBQA6 INGEST_TIME 2014-12-15T21:32:02Z PACKAGE AA00026654_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES



PAGE 1

RELIABILITY AND UTILITY OF TRANSCRANIAL MAGNEHC STIMULATION TO ASSESS ACTIVITY-DEPENDENT PLASTICITY IN HUMAN STROKE By MATTHEW PAUL MALCOLM A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2003

PAGE 2

To April and Grandma

PAGE 3

ACKNOWLEDGMENTS A doctoral dissertation cannot occur at tine hands of a single person. The success of all aspects of the project is intimately tied to the support and 1 guidance of many individuals. In light of this, I am extremely grateful to all of those involved with my education, research and life. I must first thank my wonderful wife, April, for her unending patience, support, rationality, and love. She moved away from her family to follow me to UP, endured the ups and downs of my academic career, and helped me to enjoy life as much as a graduate student can. April has been my angel during this whole process, and I could not have done it without her. j I owe a debt of gratitude to the members of my committee: Dr. Leslie I Gonzalez-Rothi, Dr. William Triggs, Dr. Orit Shechtman, and Dr. Kathye E. Light. I thank Leslie for being supportive during my pre-doctoral fellowship at the Brain Rehabilitation Research Center, and for her words of advice throughout my job I search. I am fortunate to have known Bill Triggs on both a personal and professional level, as he is a true gentleman and scholar. He was both a TMS and musical 'sensei' to me and I look forward to future collaborations with him. Orit was the first faculty member to take me under her wing, and has taught me so much about educating students. She also stuck with me through many dissertation topics and always lent a supportive ear to my occasional frustrations.

PAGE 4

Last but not least, I must thank one of the most influential people of my life, Kathye Light. Kathye is a wonderful mentor who fills her graduate students with confidence and strength. I am blessed to have learned so much about science, the politics of academia, and life from Kathye. I owe much of my academic and research success to her, and will never forget her unselfish contributions to my life. I also thank Dr. Carl Kukulka, Dr. Mark Bishop, Dr. William Mann, Dr. Pam Duncan, Dr. Andrea Behrman, Dr. Sam Wu, Kimberly Reid, and Dick Moss for their support, assistance and guidance during my doctoral studies. One key to surviving a doctoral program is to share the process with wonderful fellow students. For example, I am fortunate to have been one half of the "brother and sister" research duo known as "Matt and Stacy: the CIMT kids". Stacy was a great person to bounce questions, ideas and problems off of, and she really challenged me to be a better student. I thank Gauri Khandekar, as her assistance during data collection and analysis proved to be instrumental in the success of this dissertation. I also owe a debt of gratitude to all of the people working in the stroke rehabilitation lab: Tara, Shalaka, Vicky, Sharon, and Cristina. I thank them for doing such a great job training the study participants. Finally, I thank Sheryl, Po, Amy, Arlene, Emily, Tiffany and Michelle for being the best fellow students a guy could ask for. From the time I was able to read and tell time, my family and friends have encouraged my pursuit for knowledge. My parents taught me the value of education and helped me to see my potential, even when I doubted my iv

PAGE 5

intelligence or will. I thank my sister, Lucy, for saying, "you are so smart" with complete sincerity and pride. The loudest cheers during my academic pursuits have come from my grandparents, who have bragged to anyone and ever/one who will listen. Some of you are in heaven now, but I have felt you watching over me all of the while (sorry about all of the swearing). Cheryl and Chris were awesome to believe in me, even though my time in Florida was difficult on their family. I thank Sue and Drew for being the first people to encourage my decision to return to college, and Lisa and Ryan for offering support when the going got rough. Finally, I appreciate Pete, Oscar, Joe, Jill, Jimmy, Jeff, Mary, and the entire Buffalo Gang for making life enjoyable during my 9 years of college. I also wish to acknowledge those financial contributions that helped me pursue my research and academic goals: the VA Division of Rehabilitation Research and Development, the VA Brain Rehabilitation Research Center, the Shands Hospital Board of Directors, and the Departments of Occupational and Physical Therapy at the University of Florida. Last and most importantly, I thank God for this. I am eternally grateful for the 30 years of guidance that He has provided in my life. In addition, I thank Him for creating the beauty that is the human mind. V

PAGE 6

PREFACE The reader should note that those persons involved In the experiment on stroke are not referred to as patients, but rather as participants or Individuals. This is an important distinction that is often not made in rehabilitation research. The word patient Implies that the person Is ill or in need of medical care. Labeling survivors of stroke as patients only furthers the stereotype that these people are disadvantaged and dependent on others. Although the participants in the second experiment of this study had sustained a stroke, they were otherwise healthy, competent and motivated Individuals. vi

PAGE 7

TABLE OF CONTENTS Page ACKNOWLEDGMENTS iii PREFACE vi ABSTRACT ix CHAPTER 1 INTRODUCTION AND BACKGROUND 1 Specific Aims and Hypotheses 7 Background and Literature Review 8 Neural Control of Hand Movement 8 Activity-Dependent Neuroplasticity 13 Transcranial Magnetic Stimulation 21 Post-Stroke Activity-Dependent Plasticity and Neurorehabilitation 29 Summary 35 2 EXPERIMENT I: RELIABILITY OF TRANSCRANIAL MAGNETIC STIMULATION 36 Methods 41 Results 50 Discussion 65 3 EXPERIMENT II: ACTIVFTY-DEPENDENT PLASTICTIY IN STROKE 79 Methods 85 Results 97 Discussion 118 4 GENERAL SUMMARY AND CONCLUSIONS 138 Experiment I Summary 138 Experiment II Summary 141 General Conclusions 146 vii

PAGE 8

APPENDIX A PILOT DATA 148 B BEHAVIORAL EVALUATION FORMS 157 C INFORMED CONSENT TO PARTICIPATE IN RESEARCH 161 REFERENCE LIST 170 BIOGRAPHICAL SKETCH 180 viii

PAGE 9

Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy RELIABIinY AND UTIinY OF TRANSCRANIAL MAGNETIC STIMULATION TO ASSESS ACTIVITY-DEPENDENT PLASTICITY IN HUMAN STROKE By Matthew Paul Malcolm December 2003 Chair: Kathye E. Light, Ph.D. Cochair: Orit Shechtman, Ph.D. Major Department: Rehabilitation Science Motor cortex (Ml) neuroplasticlty is of primary interest to rehabilitation scientists, as this process may underlie post-stroke recovery of movement. Of the several neurophysiologic techniques, transcranial magnetic stimulation (TMS) is the most appropriate for studying Ml plasticity. Numerous studies have used TMS, however, limited evidence exists for the reliability of this technique. In Experiment 1, we sought to establish test-retest reliability of several TMS measures of Ml organization and excitability. In Experiment 2, we used TMS to assess plasticity related to recovery of upper extremity function in stroke survivors engaged in constraint-induced (CI) movement therapy. We hypothesized that (1) TMS measures would demonstrate good test-retest reliability, and (2) that stroke survivors would demonstrate Ml plasticity following a course of CI therapy. Participants in Experiment 1 were 20 healthy volunteers. ix

PAGE 10

Participants in Experiment 2 were 23 individuals wfio were 10 months to 10.75 years post-stroke. In both experiments, the following TMS variables were investigated in two hand and two forearm muscle representations: motor map size, motor map volume, map center of gravity, recruitment curve slope, and motor threshold. Participants in both experiments were tested on two sessions separated by 2 weeks. In Experiment 2, subjects also underwent an evaluation of upper extremity function. Participants in the second experiment received CI therapy during the 2-week testing interval. The intraclass correlation coefficient was used to assess test-retest reliability. Separate t-tests were used to assess preto post-CI therapy differences. Association between TMS and functional outcomes was determined using the Pearson r. Group differences between high and low functioning subjects were assessed using separate ANOVAs. Noteworthy findings from Experiment 1 include generally moderate to high reliability for the TMS measures. In Experiment 2, we found significant changes in some Ml muscle representations, which were paralleled by, but not directly correlated with, functional improvements. We found generally small differences between high and low functioning groups. TMS is a reliable measure of Ml organization and excitability, and may be used to investigate activity-dependent plasticity associated with intensive upper limb training in individuals post-stroke. X

PAGE 11

CHAPTER 1 INTRODUCTION AND BACKGROUND One of the most important properties of the human brain is its capacity to adapt to a variety of demands. Pioneering studies in both nonhuman primate and human models demonstrate that learning, practice, and recovery from neurological Injuries are associated with reorganization of primary motor cortex (Ml) representations. Alterations of Ml representations of upper limb movement are of particular interest, as these changes subserve the acquisition or recovery of complex, skilled movements. For example, the learning of a new fine-motor skill is associated with an enlargement of the motor maps of the involved digits after several days of intensive practice (Kami et al. 1995, Pascuai-Leone et al. 1995a). Similarly, recovery of hand function in individuals post-stroke has been associated with an enlargement or shifting of the hand representation in the damaged hemisphere (Cicinelli et al. 1997; Liepert et al. 1998b, 2000; Rossini et al. 1998; Traversa et al. 1997). Such activity-dependent alterations in Ml representations probably involve changes in synaptic efficacy, unmasking of silent synapses, and/or shifts in the excitatory-inhibitory balance (Hallett 2001). These changes are also paralleled by improved movement skill, implicating activity-dependent neuroplasticity as a substrate to skill mastery during novel learning or during the recovery of movement after stroke. 1

PAGE 12

2 Three primary techniques exist to investigate the organization and plasticity of Ml. Functional magnetic resonance imaging (fMRI) and positron emission tomography (PET) are neuroimaging techniques that measure blood flow or metabolic changes that are linked to function-related activity of neurons. When mapping a motor representation, fMRI and PET may be used to visualize the entire neural network subserving a particular movement. There are, however, a number of limitations in both the fMRI and PET techniques. These include limited temporal resolution, the inability to differentiate between activity related to excitation or inhibition, and the requirement of the subject to perform a discrete movement (which may be difficult in the case of stroke). Given these limitations, fMRI and PET may not be the most sensitive and appropriate techniques to track discrete plastic changes associated with the acquisition or recovery of specific motor skills. A third neurophysiological technique, transcranial magnetic stimulation (TMS), is perhaps better suited to investigate motor nervous system changes. In mapping studies, TMS is used to indirectly excite corticospinal neurons within Ml. In this manner, TMS can be used to outline the location, size, and excitability of the motor representation subserving a particular muscle. The advantage of using TMS over fMRI and PET is that this technique maintains high temporal resolution, activates both excitatory and inhibitory inputs to the corticospinal tract, and does not require the subject to produce any movement. Since the introduction of TMS, numerous studies have been performed to

PAGE 13

3 investigate the somatotopic organization of Ml (Aimonetti et al. 2002, BrasllNeto et al. 1992b, Roricht et al. 1999, Wassermann et al. 1992); stimulusresponse characteristics of the corticospinal system (Boroojerdi et al. 2001a, Devanne et al. 1997, Ray et al. 2002, Ridding & Rothwell 1997, Thickbroom et al. 1998); and alterations in the motor system following skill practice (Liepert et al. 1998a, Pascualleone et al. 1995, Pascual-Leone et al. 1995b) or after stroke (Cicinelli et al. 1997; Liepert et al. 1998b, 2000, 2001; Traversa et al. 1997; Trompetto et al. 2000). The extensive number of TMS studies supports the utility of this technology to investigate the motor nervous system. Additionally, several of these investigations report the occurrence of activity-dependent plastic changes related to skill practice and recovery from stroke. Results of these studies, however, should be considered with caution, as the reliability of TMS assessments of Ml representations has received limited attention. Only five published studies have investigated the reproducibility of TMS measures (Carroll et al. 2001, McMillan et al. 1998, Miranda et al. 1997, Mortifee et al. 1994, Uy et al. 2002). Although these are important early studies, most of them have methodological and statistical issues that cast doubt on their validity. Moreover, the range of normal variation in TMS measures of cortical organization and excitability has not been adequately determined. Establishing the reliability of TMS and limits of normality within the nervous system are essential factors that need to be determined before further claims of activity-dependent

PAGE 14

4 neuroplasticity. Given such consideration, Ti^S could be a useful tool in aiding researchers to uncover the impact of intensive task practice or therapy on the normal and stroke-damaged brain. The ability of TMS to assess activity-dependent plasticity is important to research on the recovery of skilled upper extremity function after stroke. Clearly, there is a critical need to identify the neural correlates to long-term recovery from stroke, as this disease is the leading cause of disability in the United States (Dobkin 1995). Stroke affects approximately 730,000 Americans annually (Dobkin 1995), and leaves more than half of these individuals with mild to severe movement deficits in the arm and hand (Taub et al. 1999). Until recently, recovery of upper extremity function was considered unrecoverable after 6 months had passed since stroke onset (Hallett, 2001). As a result, current models of clinical practice are often based upon theories that hold the nervous system to be rigid, hierarchical and compartmentalized (Held 2000). Additionally, the window of effective treatment is often considered to be within the first 6 months post-stroke. This line of thinking influenced traditional therapies such that the focus of stroke rehabilitation has often been to use compensatory rather than restorative techniques to access function (Held 2000). As result, the long-term potential for recovery is often not optimized or explored. The recent explosion of research on the topic of plasticity, however, has begun to change traditional views of the brain's capabilities following injury. This work has demonstrated that even the damaged brain is capable of functional

PAGE 15

5 change and that the neural circuitry may be refined through intensive therapy. Basic science research in a variety of animal models of stroke has identified the mechanisms of these neural changes down to the cellular level (Jagodzinski & Hess, 2001; Rioult-Pedotti et al., 1993; Greenough, Larson & Withers, 1985). Researchers are now poised to translate basic science evidence of neuroplasticity to clinical investigations of the neural correlates of therapy in the human stroke model. Within such a framework, brain damage is now viewed as less catastrophic in that the individual has a greater potential for recovery than was once believed (Held 2000). Recovery of upper extremity function and the underlying activity-dependent plasticity appears to occur by engaging the involved limb in massed practice of meaningful tasks. Constraint induced (CI) movement therapy is a recently developed stroke therapy that utilizes an intensive program of practice to treat upper extremity hemiparesis after stroke (Taub et al. 1993, Taub et al. 1999). The key therapeutic factor in CI therapy is massed practice of functional motor tasks. Several reports demonstrate the effectiveness of CI therapy to improve coordination, movement speed, and amount and quality of use of the hemiparetic arm and hand in chronic stroke survivors (Kunkel et al. 1999, Miltner et al. 1999, Taub et al. 1993). CI therapy is also believed to drive activitydependent neuroplasticity in the primary motor cortex. Two published TMS studies report alterations in the neural representations of affected hand muscles following a course of CI therapy (Liepert et al. 1998b, 2000). Liepert's

PAGE 16

6 preliminary work provides initial evidence for the potential of therapy-induced reorganization of the dannaged hemisphere after stroke. Other neuroimaging studies provide additional support to the theory of post-stroke neuroplasticity in the primary motor cortex (Cramer & Bastings 2000, Kopp et al. 1999, Levy et al. 2001, Traversa et al. 1997). Although TMS and other neuroimaging studies report a link between recovery and plasticity, there are a number of theoretical and methodological limitations in the current published research. One major limitation is that the reliability of techniques, such as TMS, has not been adequately established. Because of the variability inherent in complex neurophysiological tools and the human nervous system, researchers must first separate extraneous variability from true changes in the brain. Claims of neuroplasticity are unfounded until this preliminary step is addressed and established. The role of therapy is seldom considered in brain plasticity studies, despite the fact that animal research describes functional neuroplasticity as 'activity dependent' (Johansson 2000; Liepert et al. 1998b, 2000; Nudo et al. 2001). The existing neuroimaging work also fails to identify a direct association between neural and behavioral changes. Functional measures directly related to changes in the primary motor cortex, which supports skilled movement, should be used to provide a more precise relationship between plasticity and a specific function. The primary purposes of this study were two-fold. First, we sought to assess the test-retest reliability of multiple and parallel TMS measures of the

PAGE 17

7 location, organization and excitability of upper extremity representations in Ml. The second purpose was to use TMS to investigate the impact of CI therapy on motor representations in chronic strol 9 months post-stroke). These experiments evolved from the following specific aims and hypotheses. Specific Aims and Hypotheses Aim #1: To demonstrate the reproducibility of the transcranial magnetic stimulation (TMS) measurements across two testing sessions in neurologically intact individuals. Hypotheses: The following physiological characteristics of the primary motor cortex (Ml), as measured by TMS, will demonstrate good test-retest reliability when assessed over two testing sessions in neurologically intact individuals: (a) size (or area) of motor maps of muscle representations (b) volume of motor maps (c) map location (center of gravity) (d) threshold for excitation (e) slope of the stimulus-response (or recruitment) curve. Aim #2: To determine the neural correlates to therapy-induced recovery of upper extremity function In the affected Ml of individuals 9 months or greater post-stroke. Hypotheses: An Intensive therapy program directed at the hemiparetic upper limb will produce the following neurophyslologlcal changes In the contralateral Ml, as measured by TMS: (a) enlargement of motor maps of muscle representations (b) increase in the volume of motor maps (c) lateral or medial shift of the maps, as evidenced by a change in the location of the map center of gravity (d) decrease In the threshold for excitation (e) increase in recruitment curve slope.

PAGE 18

8 Aim 3: To determine the association between neurophysiological changes in the affected Ml and improvements in hemiparetic arm and hand function following constraint-induced movement therapy. Hypotheses: Neurophysiological changes will be highly correlated with functional improvements in the following areas: (a) dexterity (b) movement speed (c) amount of use (d) strength. Background and Literature Review The background and literature review are divided into five primary sections. First, the neural control of upper extremity motor function is considered, emphasizing the relationship between movement control and the role of the primary motor cortex and corticospinal system. Second, animal and human evidence for activity-dependent plasticity were reviewed, as well as some of the neural mechanisms of neuroplasticity in the motor cortex. Third, the utility and reliability of transcranial magnetic stimulation as an effective technique for studying activity-dependent plasticity is addressed. The final section considers the role of therapy in driving functional plasticity of the strokedamaged motor cortex, as measured by transcranial magnetic stimulation. Neural Control of Hand Movement Highly coordinated movement of the upper extremity is dependent upon a diffuse network of nervous system components. As such, the neural codes that initiate and control skilled movement cannot be isolated to one exclusive region. Although the primary motor cortex is but one player in this network, its strong

PAGE 19

9 and relatively direct connection with the peripheral motor system suggests that this region plays a pivotal role during skill learning and practice. As will be discussed, the primary motor cortex contains representations with multiple cell assemblies that code for a variety of features inherent in movement. These representations and their specific assemblies are modified through repetition and practice. Motor control The representations and plans for movement are thought to be stored in generalized motor programs (Schmidt & Lee 1999). During voluntary movement, these motor programs specify invariant features such as speed, acceleration, and relative force (Schmidt & Lee 1999). Motor programs also tell the nervous system how to respond to sensory input related to the task (Kandel et al. 2000). Essentially, these programs are plans that specify the kinematic and dynamic features of a movement, and specify how the system should adjust based upon sensory feedback. Complex motor skills like prehension, writing, typing, and drawing are governed by motor programs that have a significant representation in the primary motor cortex. Practice and learning can affect the distribution of a motor program's neural representation. For example, the performance of novel motor tasks tends to be dependent upon neural control from the supplementary motor area (Hikosaka et al. 1996). As the task is practiced and learned it becomes more automatic (that is, it requires limited cognitive control) (Schmidt & Lee 1999). With this

PAGE 20

10 automaticity in motor skill, the neural control of performance of the tasl< shifts from the supplementary motor area to the primary motor cortex (Hil
PAGE 21

11 that innervate intrinsic hand muscles, while some synapse with interneurons within the spinal cord. The latter, indirect connections are important for coordination of larger groups of muscles, as required during reaching (Kandel et al. 2000). Highly coordinated movements, involving intrinsic and extrinsic hand muscles, are dependent upon these corticospinal projections. The seminal worl< of Penfield and Rasmussen (1950) described a somatotopic organization for Ml. During neurosurgery with the brain exposed, these scientists found that muscular responses could be elicited by electrically stimulating Broadmann's area #4 in the precentral gyrus. This area corresponds to the primary motor cortex. These responses occurred in an organized fashion, such that the neural representations of adjacent body parts were typically located next to each other. These orderly representations may be joined together to form the motor homunculus. The disproportion of body size, as represented by the homunculus, reflects the density and distribution of corticospinal fibers devoted to muscle control. The largest representations (i.e., hand and face) subserve movements requiring the greatest precision (Rossini & Pauri 2000). Highly skilled movement involves a wide distribution of brain regions. Early concepts of motor control viewed Ml simply as the switchboard in the motor nervous system (Haines 2002). These theories were based upon electrical stimulation studies, which found that individual muscle responses could be generated during stimulation (Kandel et al. 2000, Penfield & Rasmussen 1950).

PAGE 22

12 Further animal and human investigations, however, identify divergence and convergence in the system, suggesting that the Ml is capable of coordinating multiple activations of different cells to create multiple movements (Hallett 2001, Rossini & Pauri 2000). Additionally, muscle representations are diffuse and distributed among several highly specialized cells. Evarts (1968) provided the initial evidence that Ml plays a complex and directed role in controlling voluntary movement. In this now classic study, Evarts showed that, during a wrist flexion task, the firing of Ml neurons varied with the amount of force required to move the hand against resistance, but not with degree of movement. The activity of these neurons, therefore, is to specifically signal the amount of force required to produce a movement rather than actual displacement of the limb. Similarly, Maier et al. (1993) discovered that neurons active during precision grip are silent during power grip. Ml neurons also work together to produce coordinated movements. For example, Georgopoulos (1982) demonstrated that specific classes of Ml neurons work together to coordinate the spatial demands of a task, as required during activities like reaching. Ml neurons appear to be involved with planning the movement, as they become active prior to muscle initiation (Evarts 1968, Leonard 1998). Leonard (1998) describes Ml as a dynamic structure that is involved in the spatial transformation (i.e., sensory perceptions to motor acts), trajectory control and the modulation of other sub-cortical structures involved in movement. The conclusion of Leonard's research and other similar work is that

PAGE 23

the primary motor cortex is liighly specialized and imminently involved in the programming, execution and modulation of skilled movement. Furthermore, Ml appears to have two levels of functional organization: a lower level that directly controls groups of muscles that can be brought together in task specific combinations; and a higher level control system that encodes the more global features of movement, such as force and spatiotemporal characteristics (Kandel et al. 2000). ActivityDependent Neuroplasticity The motor cortex, like much of the central nervous system, is a competitive system. The 'competition' for resources within Ml neural representations is a fundamental principle of cortical plasticity (Rossini & Pauri 2000). Munk (1881) labeled this phenomenon as "vicariation of function". In the absence of input or the demand for output, a neural representation may be taken over by the adjacent representation. For example, cutting of the facial nerve results in a rapid decrease in the facial neural representation, with a rapid increase in the size of the forearm and hand motor representations (Sanes et al. 1988). On the other hand, intensive and purposeful activity extends the representation for the muscle and/or limb in use. Scientists have demonstrated that regular performance of a skilled motor task results in an enlargement of the cortical representation for the muscles involved, as seen for the fingers in string players (Elbert et al. 1995). Similarly, the Ml representation of the reading finger is

PAGE 24

14 expanded in blind Braille readers (Pascual-Leone et al. 1993) and, furthermore, fluctuates with changes in reading activity patterns (Pascual-Leone et al. 1995a). Lesion studies in sub-human primates also provide evidence for the relationship between activity-dependent plasticity and skill training. These studies demonstrate that disuse after injury results in alterations of the neural system that once supported function. For example, small surgically induced lesions in the Ml hand area of squirrel monkeys created immediate deficits in prehensile abilities and a resultant decreased use of the impaired hand (Nudo & Milliken 1996). Non-use of the affected limb resulted in an extension of the lesion, such that there was further territorial loss in the hand representation. Similar changes related to disuse have been demonstrated in human studies involving amputees (Roricht et al. 1999) and during limb immobilization (Liepert et al. 1995). In the case of the stroke-affected monkeys, retraining of prehension in the affected hand reduced extension of the lesion, and induced further reorganization in the surrounding undamaged Ml. These plastic changes were associated with near-complete recovery of skilled hand use. Activity-dependent plasticity is a descriptor that broadly encompasses many intricate neuronal mechanisms. Changes in the size, location and excitability of Ml representations likely reflect alterations in synaptic communication, neurotransmitter function, and neuronal morphology. Neuroplasticity is typically characterized in terms of short and long-term changes. Short-term or more rapid changes primarily involve alterations in the behavior of synapses, while

PAGE 25

15 long-term changes include morphological changes in the neural circuitry (Johansson 2000). Both short and long-term alterations are evident in the practice and development of motor skill, and are not mutually exclusive (Hallett 2001). Plasticity may take the form of a change in synaptic efficacy, that is, an improvement in synaptic communication. Hebb (1949) postulated that increases in synaptic efficacy occur when the firing of one neuron repeatedly produces firing in another neuron to which it is connected. In other words, an association of preand post-synaptic activity in two neurons results in some change in these neurons such that the synaptic connection between them is strengthened (Hebb 1949). The Hebbian concept of activity-dependent modification of synaptic strength is the primary example of plasticity of synaptic communication. This mechanism allows for the refinement of neural circuitry through activitydependent means at the synaptic level. Much of the understanding of learning and memory processes are based on Hebb's concept (Turrigiano 1999). Repetitive activity or practice drives synaptic efficacy and the threshold for activation by refining the temporal structure and synchronization of impulse arrival and neuronal firing (Rossini 2000). For example, long-term potentiation and long-term depression, two phenomena involved in memory formation and learning, follow Hebbian rules (Maren & Baudry 1995). Long-term potentiation (LTP) represents a relatively fast strengthening of existing synapses (Hallett, 2000). LTP is different than short-term potentiation

PAGE 26

16 (STP) and post-tetanic potentiation (PTP) in that it persists for hours or days, while the later represent more short-term changes in synaptic strength (i.e., seconds or minutes) (Maren & Baudry 1995). Additionally, the long-term characteristics of LTP are dependent upon a unique cellular mechanism. Although hippocampal LTP has been a primary focus of research on memory and learning (Kandel et al. 2000, Maren & Baudry 1995), LTP also occurs in the diffuse network that subserves movement (Hess 1996). Rioult-Pedotti et al. (1998) demonstrated that intensive practice of a reaching task in rats induced LTP in intracortical connections of Ml. LTP may also be induced directly in Layer V of the motor cortex, which indicates that direct connections to corticospinal nerve cell bodies may be influenced by LTP (Jagodzinski & Hess 2001). Furthermore, motor cortex plasticity is substantially reduced when LTP is blocked by NDMA receptor blockers or by GABAergic disinhibitors (Boroojerdi et al. 2001a). These findings suggest a common mechanistic link between LTP and activity-dependent plasticity in the motor system. Long-term depression (LTD) is another communicative mechanism that underlies memory and motor learning. LTD involves a relatively fast weakening of existing synapses (Hallett 2000). LTD has been primarily studied in the cerebellum, but like LTP, it occurs elsewhere in the brain (Maren & Baudry 1995). In the cerebellum, LTD involves the activity of parallel and climbing fibers. Following several pairings of parallel and climbing fiber stimulation, synaptic responses in the parallel fiber neurons exhibit an enduring depression

PAGE 27

17 (Maren & Baudry 1995). Hess (1996) demonstrated that LTD may be induced in the horizontal connections in Layers II and III of the rat primary motor cortex, and further suggests that the regulation of this neuronal mechanism is activity dependent. The occurrence of both potentiation and depression in the nen/ous system illustrates the delicate balance between change and stability. Together, these processes alter and stabilize the properties of neural circuits (Turrigiano 1999). Plasticity, through Hebbian mechanisms such as LTP and LTD, serves to refine the neural circuitry involved in skill learning and performance. This has been illustrated by neuroimaging studies that reveal large and diffuse brain activation during initial motor skill learning (Leonard 1998). With repeated practice and a degree of automaticity, a reduction in the number of regions involved in performing the task is seen. Additionally, the primary motor cortex appears to 'take over' the neural representation for motor skill performance when the task becomes more automatic (Kandel et al. 2000). The change in 'skill representation' may occur because of a change in the brain's strategy or because of a change in synaptic communication. In either case, plasticity is a process dependent upon activity. Thus, neuronal associativity and cooperativity involved in the learning of skill (or perhaps the refinement of motor programs) are developed through repetitive action and practice. Rapid alterations in cortical representations are the result of short-term changes in the connectivity of neuronal networks (Johansson 2000). Such

PAGE 28

18 alterations are not, per se, the result of anatomical changes, but rather shortterm adaptations in the existing circuitr/. These represent the brain's attempt to make rapid adaptations in the face of a behavioral demand. The unmasking of 'silent' connections, and alterations in overall excitation and inhibition represent such short-term alterations. The unmasking of silent synapses may be a passive or an active process. Passive unmasking has been demonstrated to occur in persons following amputation (Dobkin 1993, Ramachandran 1993). Within four weeks of amputation, representational maps in the primary sensory cortex (SI) are noted to be reorganized (Ramachandran 1993). When the patient's face or residual limb was touched, he reported sensation in the missing hand, suggesting that sensory input from the face had now invaded the adjacent hand area in the sensory homunculus. Similar findings have been noted in studies using shortterm ischemic nerve blocks and result in nearly instantaneous alterations in Ml and SI representations (Boroojerdi et al. 2001b). Such changes in cortical representations probably arise by the passive unmasking of previously silent thalamocortical and intracortical synapses (Dobkin 1993). Although this form of unmasking does not rely on practice or action, it could still be considered activity-dependent, in the sense that the brain is likely to change with a discrete loss of afferent activity. Active unmasking of silent connections more clearly represents activitydependent plasticity. Unmasking through processes involving concentrated

PAGE 29

19 activity has been demonstrated in subjects learning a skilled movement task over a period of 5 days (Pascual-Leone et al. 1995a). As the subjects became more skilled in a five-finger piano exercise, the size of the Ml hand representation increased. Such rapid changes in cortical maps likely represent the unmasking of weak or secondary synaptic connections, and are driven by concentrated practice. Short-term alterations in the functioning of neural circuitry also involve changes in the Inhibitory and excitatory characteristics of the brain. For example, the excitability of the neuronal membrane may be altered as a function of the behavior of sodium ion channels (Halter et al. 1995). Such a change accounts for a lower excitation threshold for a given neuron (Chen 2000). Unmasking or changes in membrane excitability are caused partly by the removal of tonic inhibition (Jacobs & Donoghue 1991). In the case of brain damage or stroke, increased inhibition is noted (Liepert et al. 2000). Plasticity is enhanced with the release of tonic inhibition, which may be evidence for the cortex's attempt to repair itself (Hallett 2001). Short-term physiological changes such as unmasking, LTP or LTD, and shifts in excitability may eventually give rise to structural changes (Kandel et al. 2000). Although physiological and morphological plasticity operate in different time periods, they are not mutually exclusive (Hallett 2001). Activity drives short-term physiological changes, which then may drive certain alterations in the morphology of the nervous system. Examples of these anatomical changes are

PAGE 30

the pruning back of existing synaptic connections, growth of new synaptic connections, dendritic arborization, and the proliferation of non-neural cells (Johansson 2000). As with short-term changes, the critical element for morphological plasticity is activity. Long-term anatomical changes have been identified in studies on environmental enrichment in animals. Rats housed in complex, enriched environments with access to toys, activities, and socialization show more morphological brain changes than animals housed individually or in cages without varied physical activity (Johansson 2000). These animals develop more dendritic branching and more synapses per neuron within the sensorimotor cortex (Johansson 2000). Dendritic spines, sites that receive most of the excitatory synaptic inputs, are formed in greater numbers and are continuously modified in animals reared in enriched environments (Fischer et al. 1998). Alterations in the size, location and excitability of Ml representations are likely the result of any or all of the above-mentioned mechanisms of plasticity. Changes in the gross characteristics of representations may be successfully tracked with neurophysiologic techniques such as TMS. Importantly, this technology does not possess the ability to differentiate between the various types of mechanisms of neuroplasticity. More detailed investigations of the cellular events related to plasticity are not yet possible or appropriate in human studies. By studying more general alterations in motor representations.

PAGE 31

21 however, we assume that these specific changes are occurring at the neuronal level. Transcranial Magnetic Stimulation Barker et al. (1985) first described TMS as an alternative to transcranial electric stimulation (TES). Unlike TES, TMS provides a non-painful means to stimulate the brain. TMS works by producing a large and brief electric current, which is passed through a heavily insulated wire coil that is placed on the skull over the area that corresponds to the location of Ml (Bastings et al. 1998). This transient current produces a large, time-varying magnetic field. The magnetic field passes through the skull relatively unimpeded and creates a perpendicular electric field in the underlying neural tissue, which activates neurons in Ml (Weber & Eisen 2002). Motor physiology studies indicate that TMS indirectly activates corticospinal neurons by directly activating horizontally oriented interneurons in the motor cortex (Di Lazzaro et al. 1998). When applied over a muscle representation in the motor cortex, TMS generates a motor evoked potential (MEP), which may be quantified by means of electromyography. TMS has several advantages as a neurophysiological tool. The technique: is non-invasive; does not require movement from the subject; has high temporal and spatial resolution; and is relatively inexpensive to administer. While PET and fMRI take up to two or three hours to administer, a complete map of the hand and forearm area of Ml may be performed in one to one and a half hours using TMS. Additionally, TMS may be performed while the subject is comfortably

PAGE 32

22 seated in a reclined position (winere as fMRI requires tlie subject to be in a loud and potentially claustrophobic space). A major benefit of TMS is that it requires no movement from the subject. This benefit is especially valuable when studying stroke survivors, as the ability to move may be limited and highly variable from person to person. TMS also boasts high-resolution abilities for mapping. TMS has a spatial resolution of 5 mm, and a temporal resolution on the order of a few milliseconds (Weiller 1998). Motor mapping TMS may be used to map muscle representations by stimulating over a broad region of the motor cortex. For example, the representation for an intrinsic thumb flexor muscle may be mapped by stimulating over the 'hand area' of the motor homunculus. As in the case of Penfield and Rasmussen's electrical stimulation studies (Penfield & Rasmussen 1950), TMS is used to outline the extent of a map by assessing the number of stimulating positions that elicit an MEP in the muscle of interest. Essentially, the number of excitable positions can be determined in order to create a representational map of a particular muscle. The spatial resolution of TMS for motor mapping is improved by using a figureof-eight shaped magnetic coil (Wassermann et al. 1992), providing relatively focal stimulation at the intersection of the loops (Jalinous 1991, Triggs et al. 1999). When using a figure-of-eight coil, the spatial resolution of TMS is approximately 5 mm (Brasil-Neto et al. 1992c). In addition to providing information about the spatial area of a representation, motor maps may be

PAGE 33

23 translated into three-dimensional volumetric representations. Map volume is represented as the sum total of MEP peak-to-peak amplitudes or areas of all active positions within the representation (Mortifee et al. 1994, Wassermann et al. 1992). This measurement is one assessment of the overall excitability of a representation. Map center of gravity The center of gravity (CoG) is the position of a motor map that yields the highest amplitude weighted response to stimulation (Wassermann et al. 1992). Assessment of CoG provides information about the somatotopic orientation of one muscle representation to another. Somatotopic differences between the CoGs of proximal and distal upper extremity muscles (Wassermann et al. 1992), and even between different intrinsic hand muscles (Wilson et al. 1993) have been found. Animal studies suggest that one correlate to the recovery of motor function following stroke is the recruitment of motor areas adjacent to the original representation (Nudo 1996, Nudo & Milliken 1996). Changes in the CoG of a representation may reflect such recruitment (Liepert et al. 1998b, 2000), although the existing literature is equivocal on the direction and relevancy of a shift in representation location. Recruitment curve TMS may also be used to investigate the input-output properties of the corticospinal system. Also known as stimulus-response curves, the recruitment curve depicts the change in MEP size as a function of stimulus intensity. This

PAGE 34

24 measurement is thought to assess neurons other than those in the core neuronal population that are activated at motor threshold and during motor mapping (Chen 2000, Hallett 2000). The threshold for these neurons is higher either because they are less excitable or because they are further from the site of stimulation (Hallett 2000). Recruitment curves are steeper in muscles with a low threshold to stimulation (i.e., intrinsic hand muscles), and are affected by different pharmacological interventions (Boroojerdi et al. 2001a, Chen 2000). The slope and shape of these curves provide information about the neurophysiological strength of intracortical and corticospinal connections (Devanne et al. 1997, Ridding & Rothwell 1997). The slope of the recruitment curve may also be modulated by ischemic limb deafferentation (Brasil-Neto et al. 1992a), amputations (Ridding & Rothwell 1997) and by motor learning (PascualLeone et al. 1999). Motor threshold Motor threshold represents the lowest TMS intensity that elicits a small MEP (usually 50 pV). Motor threshold is lower for intrinsic hand muscles compared to proximal arm, trunk, and leg muscles (Chen 2000). This is probably due to differences in the strength of corticospinal projections (Chen 2000) and the larger absolute number of corticospinal neurons devoted to intrinsic hand muscles (Wassermann et al. 1992). Motor thresholed is raised by drugs that block sodium channels, but is not affected by drugs that alter GABAergic or glutaminergic systems (Ziemann et al. 1996). Given these pharmacological

PAGE 35

25 effects, motor threshold likely reflects neuronal membrane excitability (Chen 2000). Several TMS studies in stroke survivors have shown that motor threshold is often higher in the damaged hemisphere relative to the undamaged hemisphere (Trompetto et al. 2000). Stroke survivors with a high motor threshold and/or absence of MEPs from the affected hemisphere typically have a poor functional recovery, while those with a lower motor threshold in the damaged hemisphere (relative to undamaged) have a "good" functional outcome (Trompetto et al. 2000). Based on these findings, we predicted that intensive practice of motor skills would lower motor threshold in chronic stroke survivors. Such a change would provide evidence for increased excitability in the stroke affected Ml. Reliability of TMS measurements Several published studies have utilized TMS to investigate the organization and excitability of the motor cortex. In addition, a number of these studies report neuroplastic changes related to the practicing of skills (Pascual-Leone et al. 1995a), intensive hand use (Pascual-Leone et al. 1995b), amputation (Roricht et al. 1999), and recovery from stroke (Cicinelli et al. 1997; Liepert et al. 1998b, 2000; Pennisi et al. 2002; Rossini et al. 1998; Traversa et al. 1997; Traversa et al. 1998; Trompetto et al. 2000). The results of these studies, however, should be considered with caution, as the reliability of TMS mapping and recruitmentcurve generation is not well established. Only five published studies have investigated the reproducibility of these TMS measures (Carroll et al. 2001,

PAGE 36

McMillan et al. 1998, Miranda et al. 1997, Mortifee et al. 1994, Uy et al. 2002). Although these are important early studies, most of them have methodological and statistical issues that cast doubt on their validity. Half of these existing studies do not utilize sufficient statistical analyses to demonstrate reliability. Previous reports (McMillan et al. 1998, Miranda et al. 1997, Uy et al. 2002) have utilized a comparison of means statistic (i.e., ANOVA) or coefficient of variation to assess reliability. These analyses do not consider the degree of association and agreement between subject scores. A more rigorous and preferred assessment of test-retest reliability is the intraclass correlation coefficient (ICC) because it reflects both the degree of correspondence and agreement among individual scores obtained over multiple testing sessions (Portney & Watkins 2000). A reliability study that examines multiple TMS assessments of multiple muscle representations using a rigorous statistical and methodological design is lacking. Mortifee and colleagues (1994) published the first study on the reproducibility of TMS mapping. These authors reported that the area and volume (amplitude weighted sum of MEPs) of motor maps for two intrinsic hand muscles were relatively stable across two testing sessions. Although this is an important and seminal work with a fairly robust method for statistical analysis, there are a number of limitations with the study. The results are based upon an investigation of only 6 subjects across two testing sessions. Only two intrinsic hand muscles were studied, which does not address if proximal (i.e., forearm)

PAGE 37

27 muscle representations remain stable. Furthermore, only one of the two muscles studied was reliable above a threshold significance level (i.e., 0.75). These authors used a circular coil rather than the more focal figure-of-eight coil, which is the standard of practice for motor mapping. To this end, the results of Mortifee and colleagues' work may not be applicable to most mapping studies. This group also did not average responses over a set number of stimuli per site, which is again, another standard of practice in mapping research. Finally, although Mortifee et al. appear to have devised an interesting experiment, there are methodological issues, which are unclear: how was the stimulating grid referenced from session to session? Did motor threshold change from session to session? Was stimulus intensity kept constant from session to session? Each of these issues would likely affect the motor map. In the case of the latter two, transient changes in motor threshold could affect the size and shape of the motor map. With an n=6, the effect of such transient changes could not be adequately accounted for. More recently, Uy et al. (2002) published a TMS reliability study. This is an important investigation, but one that also has similar methodological and statistical limitations. The sample size was small (n=8), although subjects were tested over four sessions, which increases the statistical power of the investigation. The authors, however, did not use an adequate statistical analysis to determine reliability. Test-retest reliability is best assessed using the intraclass correlation coefficient (ICC). The ICC has an advantage over other

PAGE 38

parametric analyses in that it looks at association and agreement of outcomes from two separate testing sessions (Portney and Watlcins, 2000). This limitation in Uy and colleagues' work is quite apparent upon examination of the study results. The authors used a repeated measures ANOVA to determine significant differences between any of the four testing sessions. They report an F-score equal to 2.4, which has a probability value equal to 0.094. These results indicated that there was no statistically significant difference between the testing sessions (although the p-value was still relatively small, suggesting that there might have been a trend for a difference). The repeated measures ANOVA, however, lacks three important features that are important in determining testretest reliability: (1) what is the magnitude of the similarity of the results between testing sessions, (2) what is the degree of association between these data, and (3) what is the degree of agreement between these data? The intraclass correlation coefficient accounts for all three of these variables, and thus provides the most robust method for assessing test-retest reliability. Future investigations of the reproducibility of TMS measures should use the ICC in assessing test-retest reliability. Motor mapping alone only provides one component of neuroplasticity related to practice and recovery. As previously described, shifts in excitability also tells an important story about change in the nervous system. Recruitment curve and motor threshold analyses provide this sort of information. In fact, shifts in excitability alone may affect the area and volume of a motor map

PAGE 39

29 (Thickbroom et al. 1998). Only a single investigation on the reliability of the recruitnnent curve measurement exists. To our knowledge, no study has explicitly examined the reproducibility of motor threshold across multiple test sessions. TMS is a powerful tool that may provide insight into the workings of the nervous system. In a search for the neural correlates of recovery from stroke, rehabilitation scientists and neuroscientists are employing techniques like TMS to investigate the neural correlates to recovery from stroke. Clearly, there are a number of methodological and statistical issues that must be first investigated to determine the reliability of TMS. This is an important first step to any investigation of activity-dependent neuroplasticity. Establishing the reproducibility of TMS measures will also aide researchers in discovering the true impact of neurorehabilitation on the damaged brain. Post-Stroke ActivityDependent Plasticity and Neurorehabilitation Until recently, scientists and clinicians assumed that the adult brain was incapable of long-term recovery from neurological damage caused by a stroke. This line of thinking has influenced traditional therapies such that the focus of stroke rehabilitation has often been to use compensatory rather than restorative techniques to access function. In addition, clinicians often operate under the assumption that the potential recovery ends at 6 months post-stroke onset. Under these assumptions, the capability of the nervous system to continue to recover and change in the chronic phase has not been optimized. For example.

PAGE 40

30 therapists often teach stroke patients to utilize one-handed techniques to perform activities of daily living (ADL) and other functional tasks using the unaffected upper limb. Although the individual may become proficient in managing basic self-care, his or her repertoire of skills is reduced. As a result, many stroke survivors remain dependent on others for many areas of daily living, and some 84% are unable to resume their previous personal, familial and worker roles (Dobkin 1995). The challenging goal of rehabilitation after stroke is to improve skills like reaching, grasping and manipulating once the acute period of spontaneous recovery has passed. Because of financial and time limitations, therapists often rely upon compensatory training as a means to increase independence in their clients. The down side of such an intervention is that it encourages disuse. As the previously mentioned neuroplasticity studies have shown, there are neurological consequences for this disuse. The primary element for the induction of plasticity is intensive practice. Studies in animal and human models indicate that plastic changes and functional improvements are driven by sustained and repeated practice of a skill or set of skills. This relationship between practice, plasticity and improvements in skill provides rehabilitation scientists with the opportunity to re-evaluate current treatments and to create more effective ones. Constraint-induced (CI) movement therapy is an example of a treatment that employs intensive practice to remediate upper extremity function following

PAGE 41

31 stroke (Taub et al. 1993). CI therapy is believed to be successful for two primary reasons: (1) it reverses the psychological process of learned non-use of the affected linnb and (2) it induces enduring plastic changes that provide the neural substrates of recovery (Taub et al. 1999). Similar treatment paradigms are being applied to gait training (Dobkin 1993, Taub et al. 1999), and focal hand dystonia (Taub et al. 1999). CI therapy evolved from basic research with somatosensory deafferentation in monkeys and is based on a behavioral theory of learned nonuse (Taub et al. 1993). Learned non-use is a psychological process that involves conditioned suppression of movement (Morris et al. 2001). In the case of human stroke, individuals learn to avoid using their affected upper extremity because such attempts result in no movement at all or clumsy, inefficient movement. At the same time, these individuals learn to compensate with the unaffected upper extremity. CI therapy seeks to reverse the process of learned non-use by increasing an individual's motivation to use the stroke affected arm and hand. The intervention involves two critical elements: intensive practice and constraint of the unaffected limb. The therapy program entails 6 hours per day of treatment for 2 weeks, during which participants are engaged in massed practice of functional motor skills. Positive knowledge of performance and results feedback is provided, which encourages appropriate motor learning. To further encourage use, the participant is also required to wear a padded restraining mitt on the unaffected limb during 90% of waking hours.

PAGE 42

32 Taub and others believe that this experimental paradigm serves to reverse the process of learned nonuse (Taub et al. 1993). The intense 6-hour training sessions are designed to have two-fold results. First, the repetitive and concentrated treatment truly presents the subject with massed practice in learning to use their hemiparetic arm. In retraining novel motor tasks, massed practice facilitates motor learning in the early stages of skill acquisition (Schmidt & Lee 1999). Second, the therapeutic sessions are designed to make the subjects struggle, but also to achieve some level of success. Therefore, the person with a stroke begins to associate movement with some degree of success, rather than failure. In so doing, the subject is more encouraged to use the paretic limb in his daily routine at home. The end results of CI therapy have been remarkable. Subjects have consistently demonstrated large increases in the amount and quality of use, coordination, speed of movement, and spontaneous use of the affected upper limb (Kunkel et al. 1999, Liepert et al. 1998b, Miltner et al. 1999, Taub et al. 1998). Taub and colleagues (1993) also demonstrated that these functional improvements were maintained up to 2 years post-CI therapy. In addition, Miltner et al. (1999) found that even very chronic stroke survivors (post-stroke times of 9-17 years) were amenable to CI therapy, and did as well as those individuals who were much closer in time to their stroke. Clearly, CI therapy has a dramatic effect on recovery of function in stroke, the results of which hold a relative permanency and are apparently not influenced by the time since stroke.

PAGE 43

33 Researchers are now beginning to explore the linl< between intensive neurorehabilitation, like CI therapy, and activity-dependent plasticity. This work has begun to extend the animal research on training induced neural changes. Neurophysiological techniques such as TMS allow researchers to study the organization, excitability and patterns of activation of the brain. Furthermore, these techniques provide an opportunity for neuroscientists and rehabilitation scientists to examine the neural correlates to recovery, and how therapy influences this process. Using TMS, Liepert and colleagues (1998, 2000) demonstrated that the motor representation of the hemiparetic hand in chronic stroke survivors was reduced in size as compared to the intact hemisphere. Following a course of intensive movement therapy, the cortical map of the hemiparetic hand expanded into adjacent areas, increasing significantly from the pre-treatment testing. Using electroencephalography (EEG), Kopp et al. (1999) reported a shift of activation from the primary motor cortex towards the supplementary motor area from preto post-therapy. Nelles et al. (2001) used positron emission tomography (PET) to investigate the effect of therapy on behavioral and cortical changes. This group found increased activation in premotor and primary sensory areas subsequent to a standard therapy program. Levy and colleagues (2001) have provided the only fMRI evidence to support the neurological effects of intensive therapy aimed at the hemiparetic limb. In this pilot study, activity related to sensory processing and simple motor task performance increased

PAGE 44

34 around the rim of the infarct following treatment. These subjects also demonstrated increased activation in the supplementary motor areas after training. Although these studies and techniques have provided preliminary evidence for neuroplasticity following stroke, the results are not clearly related to improvements in function. In other words, there may be motor cortex changes that occur as a person is progressing towards recovery, but what is the relationship between this neuroplasticity and improvements in the person's repertoire of skills? A major limitation of the existing neuroimaging work is that limited or no behavioral measures were used to support the notion that brain reorganization underlies functional recovery. Despite identifying cortical changes, Nelles and colleagues (2001) did not find any significant improvement in use of the hemiparetic limb. Liepert et al. (1998) and Liepert et al. (2000) demonstrated a marked increase in the cortical representation of the affected hand subsequent to therapy. The only behavioral measure used, however, was a subject questionnaire. No objective measures of functional recovery were performed. Levy and colleagues' (2001) pilot study provides support for the theory of activity-induced cortical reorganization as a mechanism for improved use of the hemiparetic limb. Subjects in this experiment demonstrated modest improvements in speed and quality of movement, strength, and real world use of the affected arm (Levy et al. 2001). These results were obtained in only two subjects, which limits the ability to generalize from this study.

PAGE 45

35 Summary Animal and human neurophysiological studies demonstrate that motor cortex representations for movement are altered with intensive practice. The acquisition of skills is clearly related and probably dependent on this plasticity. Several methods exist to investigate brain plasticity, but TMS is perhaps the bestsuited technique to examine changes in the motor nervous system. Although numerous TMS studies exist, the reliability of the technique has not received adequate attention. Establishing the reproducibility of TMS measures is a critical first step in developing an understanding of the true impact of therapy on the damaged nervous system. The experiments performed in this study address many of the theoretical and technical issues that are evident in the existing literature on neuroplasticity. The first purpose of these experiments was to establish the reliability of TMS as a tool for studying the motor nervous system. The second purpose was to provide mechanistic evidence for the direct influence of CI therapy on both neurological and behavioral changes. The final purpose of this study was to identify the relationship between neuroplasticity and recovery of specific motor functions.

PAGE 46

CHAPTER 2 EXPERIMENT I: RELIABILITY OF TRANSCRANIAL MAGNEHC STIMULATION Transcranial magnetic stimulation (TMS) is a neurophysiologic technique that may be used to study the human motor nervous system. Utilizing TMS, researchers have investigated the organization and excitability of the corticospinal system that subserves voluntary movement. Until the advent of this technology, non-invasive methods for testing excitatory thresholds and for mapping the primary motor cortex (Ml) were not suitable for human studies. Since the introduction of TMS, however, numerous researchers have investigated the somatotopic organization of Ml (Aimonetti et al. 2002, Brasil-Neto et al. 1992a, Roricht et al. 1999, Wassermann et al. 1992), stimulus-response characteristics of the corticospinal system (Boroojerdi et al. 2001, Devanne et al. 1997, Ray et al. 2002, Ridding & Rothwell 1997, Thickbroom et al. 1998), and alterations in the motor system following stroke (Cicinelli et al. 1997; Liepert et al. 1998b, 2000, 2001; Traversa et al. 1997; Trompetto et al. 2000) or following skill practice (Liepert et al. 1998a; Pascual-Leone et al. 1995a, 1995b). Despite this extensive TMS research, limited attention has been directed toward identifying the reliability of TMS measurement techniques for mapping neural representations or generating stimulus-response curves. The purpose of this study was to determine the test-retest reliability of various TMS measures to 36

PAGE 47

37 examine Ml neural representations and corticospinal excitability in normal subjects. Barker et al. (1985) first described TMS as an alternative to transcranial electric stimulation (TES). Unlike TES, TMS provides a non-painful means to stimulate the brain. TMS works by producing a large and brief electric current, which is passed through a heavily insulated wire coil that is placed on the skull over the area that corresponds to the location of Ml (Bastings et al. 1998). This transient current produces a large, time-varying magnetic field. The magnetic field passes through the skull relatively unimpeded and creates a perpendicular electric field in the underlying neural tissue, which activates neurons in Ml (Weber & Eisen 2002). Motor physiology studies indicate that TMS indirectly activates corticospinal neurons by directly activating horizontally oriented interneurons in the motor cortex (Di Lazzaro et al. 1998). When applied over a muscle representation in the motor cortex, TMS generates a motor evoked potential (MEP), which may be quantified and qualified by means of electromyography. TMS may be used to map muscle representations by stimulating over a broad region of the motor cortex. For example, the representation for an intrinsic thumb flexor muscle may be mapped by stimulating over the "hand area" of the motor homunculus. As in the case of Penfield and Rasmussen's electrical stimulation studies (Penfield & Rasmussen 1950), TMS is used to outline the extent of a map by assessing the number of stimulating positions that

PAGE 48

elicit an MEP in the muscle of interest. Essentially, the number of excitable positions can be determined in order to create a representational map of a particular muscle. The spatial resolution of TMS for motor mapping is improved by using a figure-of-eight shaped magnetic coil (Wassermann et al. 1992), providing relatively focal stimulation at the intersection of the loops (Jalinous 1991, Triggs et al. 1999). When using a figure-of-eight coil, the spatial resolution of TMS is approximately 5 mm (Brasil-Neto et al. 1992b). In addition to providing information about the spatial area of a representation, motor maps may be translated into three-dimensional volumetric representations. Map volume is represented as the sum total of MEP peak-to-peak areas of all active positions within the representation (Mortifee et al. 1994, Wassermann et al. 1992). This measurement is one assessment of the overall excitability of a representation. The TMS measurement, center of gravity (CoG), is the position of a motor map that yields the highest amplitude weighted response to stimulation (Wassermann et al. 1992). Assessment of CoG provides information about the somatotopic orientation of one muscle representation to another. Somatotopic differences exist between the CoGs of proximal and distal upper extremity muscles (Wassermann et al. 1992), and even between different intrinsic hand muscles (Wilson et al. 1993). Animal studies suggest that one correlate to recovery of motor function following stroke is the recruitment of motor areas adjacent to the original representation (Nudo 1996, Nudo & Milliken 1996).

PAGE 49

39 Changes in the CoG of a representation may reflect such recruitment (Liepert et al. 1998b, 2000), however, the range of normal variation in CoG location within the healthy brain has not been established. TMS may also be used to investigate the input-output properties of the corticospinal system. Also known as stimulus-response curves, the recruitment curve depicts the change in MEP size as a function of stimulus intensity. This measurement is thought to assess neurons other than those in the core neuronal population that is activated at motor threshold and during motor mapping (Chen 2000, Hallett 2000). The threshold for these neurons is higher either because they are less excitable or because they are further from the site of stimulation (Hallett 2000). Recruitment curves are steeper in muscles with a low threshold to stimulation (i.e., intrinsic hand muscles), and are affected by different pharmacological interventions (Boroojerdi et al. 2001, Chen 2000). The slope of these curves provides information about the neurophysiological strength of intracortical and corticospinal connections (Devanne et al. 1997, Ridding & Rothwell 1997). The multiple capabilities of TMS make it a useful technique for providing a comprehensive assessment of motor nervous system organization and excitability. Indeed, numerous published studies have utilized TMS to investigate motor cortex changes related to the practicing of skills (Pascual-Leone et al. 1995a), intensive hand use (Pascual-Leone et al. 1995b), amputation (Roricht et al. 1999), and recovery from stroke (Cicinelli et al. 1997; Liepert et al. 1998b,

PAGE 50

2000; Pennisi et al. 2002; Rossini et al. 1998; Traversa et al. 1997; Traversa et al. 1998; Trompetto et al. 2000). The results of these studies, however, should be considered with caution, as limited evidence exists for the reliability of the TMS measures employed. Only five published studies have investigated testretest reliability of TMS assessments of Ml organization or excitability (Carroll et al. 2001, McMillan et al. 1998, Miranda et al. 1997, Mortifee et al. 1994, Uy et al. 2002). Although these are important early studies, most of them have methodological and statistical issues that cast doubt on their validity. Half of these existing studies do not utilize sufficient statistical analyses to demonstrate reliability. Previous reports (McMillan et al. 1998, Miranda et al. 1997, Uy et al. 2002) have utilized a comparison of means statistic (i.e., ANOVA) or coefficient of variation to assess reliability. These analyses do not consider the degree of association and agreement between subject scores. A more rigorous and preferred assessment of test-retest reliability is the intraclass correlation coefficient (ICC) because it reflects both the degree of correspondence and agreement among individual scores obtained over multiple testing sessions (Portney & Watkins 2000). A reliability study that examines multiple TMS assessments of multiple muscle representations, while employing a rigorous statistical and methodological design, is lacking. TMS is a powerful tool that provides insight into the workings of the nervous system. There are, however, numerous extraneous factors that could potentially confound the results. Example confounds include: spontaneous rapid

PAGE 51

41 alterations in cortical outflow (Ridding & Rothwell 1997) coil position (PascualLeone et al. 1994), EMG data collection and processing (Hermens et al. 2000, McMillan et al. 1998), and transient changes in motor threshold (Thicl
PAGE 52

42 psychiatric disease, (8) no history of seizures or epilepsy, (9) no pacemaker or other metal implants in the upper body, and (10) negative pregnancy test (for women of childbearing potential). All aspects of participant recruitment and testing met the approval of the local Institutional Review Board prior to initiation of this project. Each participant signed a written informed consent prior to his or her involvement with any of the study procedures. The participant was comfortably seated in a semi-reclined modified dental chair with a pillow support placed beneath the dominant forearm and hand. Passive bipolar surface electrodes were applied over the first dorsal interosseous (FDI), abductor pollicis brevis (APB), extensor digitorum communis (EDC), and flexor carpi radialis (FCR) muscles in the right upper limb in a belly-tendon arrangement. Correct placement of the electrodes was verified by asking the subject to maximally contract the muscle while the investigator monitored for an EMG output amplitude of approximately 0.8 mV. The inter-electrode distance was fixed at 20 mm for all muscles. EMG signals were filtered with a bandpass set at 2-10 kHz, rectified, and amplified with a Viking II Electromyograph (Nicolet Biomedical, Madison, WI). Audio feedback from the electromyograph was routinely monitored to ensure muscle relaxation during the testing session. A latex swim cap was placed on the participant's head so that a coordinate system could be clearly marked. The vertex (Cz) was marked as the intersection of the nasion-inion and interaural lines. Measurement of these lines was recorded to

PAGE 53

ensure consistent location of tine Cz across testing sessions. All TMS stimulation points were recorded in reference to the Cz (see Figure 2-1). General TMS Testing Procedure Five primary TMS variables were investigated: (1) motor map area, (2) FDI motor map volume, (3) location of the center of gravity (CoG) of the motor map, (4) slope of the recruitment curves, and (5) motor threshold. Additionally, the compound motor action potential (CMAP) for the right FDI, using supramaximal electrical stimulation of the ulnar nerve at the wrist, was determined for data normalization purposes in the map volume analysis for this muscle. Two different stimulators were used during the TMS procedures. Stimulation during optimal position locating, motor threshold testing at the optimal position and motor mapping was delivered using a Magstim Rapid (Magstim Company Limited, UK) magnetic stimulator through a 5 cm mean loop diameter figure-of-eight shaped magnetic coil. Stimulation during the recruitment curve procedure and in determining the motor threshold at the Cz was performed using a Magstim 200 (Magstim Company Limited, UK) magnetic stimulator through a 9 cm mean loop diameter circular shaped coil. All assessments were performed during two separate testing sessions, separated by 2 weel
PAGE 54

44 handle pointing posteriorly and the figure-of-eight coil situated tangential to the skull (Figure 2-1). Stimulation was delivered over the left hemisphere, contralateral to the muscles of interest (i.e., in the dominant, right arm). With the stimulator set at its maximum output and with the subject relaxed, the 'optimal position' for stimulation was identified and its location recorded in relation to the Cz. The optimal position is defined as the stimulation point that elicits the largest amplitude MEPs. Once the optimal position was determined, motor threshold was assessed in a step-wise fashion. Motor threshold is defined as the lowest stimulation intensity that elicits discernable MEPs in at least 5 of 10 consecutive stimulations using an oscilloscope gain of 200 pV per cm (Wassermann et al. 1992). To compensate for possible initial heightened arousal levels and/or startle responses, which might affect MEP threshold; several trial stimulating runs were performed prior to the final assessment of motor threshold at the optimal position. TMS Mapping A 5 X 5 cm grid was marked on the swim cap and centered around the optimal position (25 points, separated by 1 cm; see Figure 2-1). The stimulator was set at 115% of the motor threshold and five stimuli were delivered to each grid point at a frequency of IHz. The EMG responses from these five stimuli/grid point were rectified and averaged online using Viking II nerve conduction software (Nicolet Biomedical, Madison, WI). After all grid positions were stimulated, the grid was extended, as necessary, until the area from which MEPs

PAGE 55

were elicited was surrounded by stimulated sites that did not elicit MEPs discernible at an oscilloscope display gain of 200 |jV per cm in any muscle. This method ensures that the full extent of the motor map is captured (Triggs et al. 1999). In considering the test-retest reliability of mapping motor representations with TMS, we recognized that changes in motor threshold between testing sessions could occur and might influence the mapping results. Since changes in motor threshold between testing sessions may reflect differences in coil position or corticospinal excitability, we elected to use the same intensity of stimulation (115% of motor threshold determined in testing session 1) for both testing sessions, and to determine the test-retest reliability of both motor maps and motor threshold as individual variables. Recruitment Curve Procedure The recruitment curve procedure was performed after the mapping procedure, and in a manner previously described by Ray et al. (2002) and Boroojerdi et al. (2001). The subjects were prepped in the same manner as in TMS mapping. Magnetic stimulation was delivered over the Cz using the circular coil. The coil was placed tangential to the skull, with the handle oriented sagittaly and side W facing up to direct the magnetic outflow towards the left hemisphere. Magnetic stimuli were applied at 5% steps between 30% and 100% of the maximum stimulator output. Five stimuli were delivered at each intensity, at a rate of 0.2 to 0.3 Hz. The MEPs produced at each stimulation intensity were rectified and averaged online over the five stimulation trials, as

PAGE 56

46 during the mapping procedure. These data were used to construct a stimulusresponse curve, with MEP peak-to-peak amplitude plotted as a function of TMS intensity. The slope of this curve was analyzed for test-retest reliability. Figure 2-1. Location of the vertex (Cz), interaural and nasion-inion lines, stimulating grid, and figure-of-eight coil orientation on the head. Motor Threshold Procedure Motor threshold was studied for two different purposes, and thus was assessed by two different means. As described above, the first purpose for assessing motor threshold was to determine the stimulation intensity to be used during the TMS mapping. For this purpose, motor threshold was determined at

PAGE 57

47 the optimal position. The second purpose for assessing nnotor threshold was to determine the reliability of this measure over multiple testing sessions. We considered that a more precise means of identifying motor threshold for this purpose might be achieved by stimulating with a large circular coil placed over the Cz. As at the optimal position, motor threshold at the Cz was assessed in a step-wise fashion as the lowest TMS intensity to elicit an MEP in at least 5 of 10 consecutive stimulations using an oscilloscope gain of 200 pV per cm. To account for subject adaptation to TMS, motor threshold at the Cz was re-checked at the end of the testing session. Compound Motor Action Potential The compound motor action potential (CMAP) was determined for each subject for the purposes of data normalization during the assessment of FDI motor map volume. The CMAP is generated by electrically stimulating a peripheral nerve to produce an MEP in a distal muscle. A MEP produced in this manner represents stimulation of 100% of the motor neurons supplying the muscle of interest. The CMAP procedure was studied in all subjects, and was performed in the following manner. Passive, bipolar surface EMG electrodes were applied over the FDI with a reference electrode located on the dorsal aspect of the hand. A bipolar stimulating electrode was prepared with conducting gel and was placed over the ulnar nerve at the distal-medial aspect of the forearm. Stimulation was delivered at a rate of 0.7 Hz while intensity was

PAGE 58

48 gradually increased until the CMAP waveform ceased to increase in amplitude, producing a maximum M-wave. Data Analysis Viking II nerve conduction software was used to determine the mean peak-to-peak area of rectified MEPs elicited during the motor mapping and recruitment curve procedures. For motor mapping, the mean area of MEPs elicited at each stimulation site were normalized to that of the stimulus site which produced the largest MEPs. In this manner, the muscle representation area was quantified as the number of stimulation sites that elicited MEPs of area >10% of the MEP area for the stimulation site that produced the largest MEPs. This method of normalization eliminates EMG output due to spontaneous muscle activity unrelated to the stimulation. Map volume was expressed as the sum of the MEP areas, normalized to the CMAP peak-to-peak areas, for the FDI maps. Center of gravity was represented as the maximum amplitude-weighted position, and was calculated as follows: for each stimulating position on the map, the amplitude-weight was computed as the amplitude at that position divided by the sum of peak-to-peak MEP areas recorded for the map. The weight at any stimulating position was interpreted as the proportion of the total map area contributed by that location. For the recruitment curve data, MEP area was plotted as a function of stimulus intensity (i.e., percent of stimulator output). These data were fitted to a linear model. Although the recruitment curve is typically sigmoid in shape, using the linear model does not require the higher

PAGE 59

plateau values in the curve. This consideration avoids the use of high stimulus intensities to produce plateau values, which result in subject discomfort. We assessed the test-retest reliability of map area, map volume, optimal position location, recruitment curve slope, and motor threshold across two testing sessions separated by 2 weeks. The intraclass correlation coefficient (ICC) was used to assess reliability. This statistic is the preferred index of reliability, as it reflects both the degree of association and agreement between preand posttest session findings (Portney and Watkins 2000). We report the ICC model (C,2), which indicates the reliability of each of the variables when averaged over two sessions. An ICC > 0.75 is generally considered high, while those below 0.75 are indicative of moderate to poor reliability (Portney & Watkins 2000). Differences between the locations of all four muscle representation CoGs were assessed for each testing session using separate one-way ANOVAs [one factor (muscle) with four levels (APB, FDI, EDC and FCR)]. The Tukey's Honestly Significant Difference test was performed post-hoc to look for differences in the location of each individual CoG from the others. Finally, the ability of motor threshold at the Cz to predict motor threshold at the optimal position was investigated using linear regression.

PAGE 60

50 Results Motor Map Area The mean number of active positions and intraciass correlation coefficient results for the motor map area of each muscle are shown in Table 2-1. As is indicated by the reliability analysis and Figures 2-2 and 2-3, the muscle representation area remained relatively stable across the two testing sessions. For motor map area, the ICCs were: APB=0.68, FDI=0.60, EDC=0.86, and FCR=0.85. Analysis of the pooled data for all subjects revealed that the EDC had the largest mean area (15.65 active positions), while the representations for the remaining three muscles were similar (APB=9.75, FDI=11.45, FCR=8.5 active positions). Table 2-1. Comparison of preand post-session mean motor map area and ICC results. mean area (S.D.) M uscle pre post ICC APB 9 .0 (4.8) 10.5 (6.7) 0 .676 FDI 12 .9 (4.5) 10.0 (5.5) 0 .599 EDC 15 .6 (5.5) 15.7 (6.4) 0 .858* FCR 9 .0 (6.1) 8.0 (6.8) 0 .848* *Met or exceeded the threshold value for a significant replication of results of ICC > 0.75. Motor Map Volume: FDI Motor map volume for the FDI was assessed as an indicator of the stability of relative size of the MEPs from session to session. Each MEP peak-topeak area was expressed as a percent of that of the CMAP. The average FDI map volume for all subjects remained very stable across the two testing sessions, only changing by 0.9% (Figure 2-4). Map volume differences between

PAGE 61

51 subjects were highly variable with large pre(60.0) and post-session (47.8) standard deviations. These data, however, demonstrated good test-retest reliability within subjects, as demonstrated by a high ICC of 0.85. Sample pre and post FDI map volume outputs from a representative subject are depicted in Figure 2-5. Mean Motor Map Area 18 16 14 a 12 (A O | 10 ra S. 8 £ n ra 6 E APB i EDC FCR muscle Figure 2-2. Comparison of mean motor map area between testing sessions and by each muscle. Area is expressed as the total number of active stimulating positions, which elicited an MEP with a peak-to-peak area >10% of that of the maximum MEP.

PAGE 62

52 APB Pre Map-Subject 19 1 r ^ 1 r ^ k. Jk. J r -V ^ k. Jk^'j 2 10-1 e-ant. (cml DOSt."> E 5 E o 6 i c 7 5 APB Post Map-Subject 19 r k. J 2 10-1 <"ant. (cm) post."> 3.5 4.5 S. N u E o 5 5 6 .5 "5 7.5 FDI Pre Map-Subject 6 FOI Post Map-Subject 6 r ir ir i k. JL J r 1 r ^ 3.5 2.5 1.5 0.5 <~ant. (cm) posl."> 4.5 5.5 O E 6 5 7.5 -0.5 1 0 -1 <"ant. (cm) post.-> • r ^ i • 3.5 4.5 O E o 5.5 6.5 Figure 2-3a. Preand post-session motor maps of the APB and FDI. The y-axis represents lateral distance from the vertex (Cz). The x-axis represents anterior (-hx) and posterior (-x) distance from the interaural line. Each grid square represents a stimulation position oriented over the pre-central gyrus. Marked squares indicate positions that produced MEPs. Although the shape of preand post maps differ, the absolute area is similar.

PAGE 63

53 EDC Pre M ap-Subje ct 1 B3C Post M ap-Subje ct 1 4 i ^ > 4 ^ r>4 3 5 2 5 15 <--tnt. (cm ) poit.--> 2 1 0 <--nt. {cm ) pott."> FCRPre Map-Subject 12 fCRPost M ap-SubjccI 12 r ^ ^ JL.' Jk. JL. JL. Jl. j r ^ -^r yuK 4 4 ^ ^I^^L JL JL J 1 r ^ L. JL. ^ ^ JL. J m -anl. (cm ) pott."> ant. (cm } poat."> Figure 2-3b. Preand post-session motor maps of the EDC and FCR for representative subjects. The y-axis represents lateral distance (in cm) from the vertex (Cz). The x-axis represents anterior (+x) and posterior (-x) distance from the interaural line. Each position on the grid represents a stimulation position, roughly oriented over the lateral aspect of the pre-central gyrus. Marked squares indicate positions that produced an MEP with significant peak-to-peak area (i.e., >10% of the maximum MEP). Although the shape of prevs. post maps differ, the absolute area is similar, if not equal.

PAGE 64

54 Mean FDI Motor Map Volume 60 T — 50 pre pott ICC=0.85* Figure 2-4. Mean preand post-session motor map volume for the FDI. Map volume is expressed as the sum total of MEP peal<-to peak areas, which were normalized as a percent of the CMAP. *Met or exceeded the threshold value for a significant replication of results of ICC > 0.75.

PAGE 65

55 FDl Pre Map Volume-Subject 6 %CMAP ant./post. distance (cm) g lat. distance from Cz (cm) FDl Post Map Volume-Subject 6 ant./post. distance (em) lat. distance from Cz (cm) 6-7 5-6 4-S 3-4 2-3 1-2 0-1 7-8 6-7 5-6 4-5 3-4 2-3 1-2 0-1 Figure 2-5. Pre and Post FDl motor map volume in a single representative subject. The mean MEP peak-to-peak area for each active position is expressed as a percent of the CIMAP peak-to-peak volume. The yaxis represents lateral distance from the vertex (Cz), measured in cm. The x-axis represents anterior (+x) and posterior (-x) distance in cm from the interaural line.

PAGE 66

56 Center of Gravity The lateral (y, distance from Cz) and anterior-posterior (x, distance from interaural line) coordinates of the CoG of each motor map were analyzed separately for test-retest reliability of position. Mean location of the CoG and results of the reliability analysis are presented in Table 2-2. The lateral distance (y) of the CoG remained stable for the APB, FDI and EDC, as demonstrated by high ICCs that ranged from 0.82 (APB) to 0.86 (EDC). The FCR was moderately reliable (ICC=0.69). The anterior-posterior CoG coordinate did not meet the threshold for significant ICC, as these values ranged from 0.37 (APB) to 0.70 (EDC). We sought to determine if there were CoG location differences between each of the four muscle representations. As demonstrated in Table 2-2 and Figure 2-6, the mean CoG position changed little from preand post-session locations, with the exception of the FDI. These positions also followed a predicted somatotopic organization, with the APB and FDI CoGs located lateral to those of the EDC and FCR. There was a trend for differences in the lateral location between each muscle's CoG on the pretest (^2.3, yD=0.084, cfcO.084, power=0.60), and significant differences in location of these on the posttest (/^5.2, yC7=0.003, £7^0.182, power=0.91). There were no significant differences in the anterior-posterior location of the CoGs. A post-hoc analysis (Tukey's Honestly Significant Difference) was performed to determine which of the muscle's CoGs differed on their lateral location for the posttest data. This

PAGE 67

57 analysis revealed that the APB CoG significantly differed from the location of the EDC and FCR. No other significant locational differences of CoG between muscles were found. See Table 2-3 for post-hoc analysis results on the post-session CoG locations. Table 2-2. Mean CoG coordinates for all subjects combined and respective ICC values. muscle mean lateral coordinate (s.d.) M-L Coord. ICC mean ant./post. coordinate (s.d.) A-P Coord. ICC APB 5.41 (0.74) 0.824* 1.17 (0.77) 0.374 FDI 5.11 (0.70) 0.852* 1.25 (0.79) 0.380 EDC 4.88 (0.70) 0.860* 1.05 (0.67) 0.702 FCR 4.73 (0.63) 0.685 1.14 (0.76) 0.530 The lateral coordinate is the distance (in cm) from the Cz, while the anteriorposterior coordinate is the anterior (+x) or posterior (-x) distance from the interaural line. *Met or exceeded the threshold value for a significant replication of results of ICC > 0.75. Motor Threshold Motor threshold was assessed while stimulating over the optimal stimulating position and at the Cz for 19 of the 20 subjects. The data on one subject was removed due to a missing motor threshold value for the post Cz assessment. There was little change in motor threshold at either of these stimulating positions across testing sessions (Table 2-4). The ICCs were high for the optimal and Cz motor threshold at 0.97 and 0.90, respectively. Figure 2-7 presents the average motor threshold at both locations for all subjects combined. Linear regression revealed a relatively low ability for motor threshold values at the Cz to predict motor threshold at the optimal position on pre(r^=0.19) and post-session (r^=0.31) assessments (Figure 2-8).

PAGE 68

58 RRE POST] EDd 0 FDit A APBA Mean CoG Position: Pre vs. Post o o A 1.8 1.6 1.4 1.2 1 0.8 <-ant. (cm) post.-> 0.6 0.4 0.2 4.60 4.70 4.80 4.90 1 o 5.00 I o 5.10 i s l 5.20 ^ 5.30 5.40 5.50 Figure 2-6. Comparison of mean CoG position for each muscle between preand post-sessions. These positions were relatively similar between preand post-sessions, with the exception of the FDI, which demonstrated a dramatic anterior-posterior shift. These positions also followed a relative somatotopic organization, which was maintained across the two sessions. Table 2-3. Analysis of differences between individual muscle CoG locations. Post-hoc Analysis Post-session locations mean difference (cm) standard error sig. APB EDC 0.59* 0.19 0.017 APB FCR 0.74* 0.20 0.002 APB FDI 0.38 0.20 0.216 FDI: EDC 0.21 0.19 0.696 FDI: FCR 0.36 0.20 0.279 EDC :FCR 0.15 0.20 0.869 Significant difference between muscle locations, Tukey's HSD post-hoc analysis. 4

PAGE 69

59 Motor Threshold 65 m at optimal MT at vertex location of MT assessment Figure 2-7. Comparison of mean preand post-session motor threshold values obtained during stimulation at the optimal point and Cz. Motor threshold values are described as the lowest TMS intensity that produced a discernable MEP at a display gain of 200 liV per cm on 5 out of 10 trials. These data represent the mean motor threshold obtained at both locations for all subjects combined. There was very little mean change in motor threshold between preand postsession testing at either location. Motor threshold at the optimal point was assessed using a Magstim Rapid magnetic stimulator with a figure-of-eight coil, while motor threshold at the Cz was assessed using a Magstim 200 stimulator with a circular coil. Absolute differences in motor threshold at these two locations, therefore, is at least partly related to different output intensities of the two different stimulators, with the Magstim 200 having a higher output intensity.

PAGE 70

60 Pre-MT 10 20 30 40 50 MT at vertex (% stim. output) 60 o i 40 .§ 30 S 20 Post-MT R^ = 0.3108 30 40 50 MT at vertex (Vo stim. output) Figure 2-8. Relationship between motor thresholds (MT) obtained at the Cz and optimal point for preand post-sessions. Motor threshold values are described as the lowest stimulus intensity that produced a discernable MEP. Each data point represents the intersection of motor threshold at the Cz with motor threshold at the optimal point for each individual subject. Linear regression revealed a low ability of motor threshold at the Cz to predict motor threshold at the optimal position, as noted by the low r^ values.

PAGE 71

Recruitment Curves When analyzing the recruitment curve results, the mean peak-to-peak MEP area was plotted against TMS intensity and fitted with a line of best fit. Statistical analyses were, therefore, based upon the slope of this line for each muscle and in each subject. The mean recruitment curve slopes and ICC results are presented in Table 2-4. The EDC, FCR and FDI demonstrated good testretest reliability as the ICCs of the recruitment curve slopes for these muscles ranged between 0.85 (FCR) and 0.91 (EDC). The ARB slope data was highly variable, however, resulting in poor test-retest reliability. This high variability was the result of large differences in the raw MEP peak-to-peak area values within and between subjects. Representative pre and post-session recruitment curves from individual subjects are presented in Figure 2-9a and 2-9b. Table 2-4. Test-retest reliability analysis of recruitment curve slope by muscle. Mean Slope (S.D.) Muscle pre post ICC APB 0.16 (0.23) 0.21 (0.23) 0.255 EDC 0.11 (0.09) 0.12 (0.12) 0.910* FCR 0.06 (0.06) 0.06 (0.06) 0.853* FDI 0.25 (0.27) 0.22 (0.19) 0.892* *Met or exceeded the threshold value for a significant replication of results of ICC > .75.

PAGE 72

62 APB Recruitment Curves-Subject 14 -A pre post FDI Recruitment Curves-Subject 20 -pre -post 35 40 45 50 55 TM S Intensity (%) 65 70 Figure 2-9a. Representative preand post-session recruitment curves for the APB and FDI in Subjects 14 and 20, respectively. MEP peal<-topeal< area was plotted as a function of TI^S intensity (% of stimulator output).

PAGE 73

63 EDC Recruitment Curves-Subject 13 120 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100 TMS intensity (%) FCR Recruitment Curves-Subject 20 12 10 x 8 (0 E ra LU ^ 1 1 1 1 1 1 -pre post 30 35 40 45 50 55 60 TMS intensity (%) 65 70 75 Figure 2-9b. Representative preand post-session recruitment curves for the EDC and FCR in Subjects 17 and 20, respectively. MEP peal<-topeak area was plotted as a function of TMS intensity (% of stimulator output).

PAGE 74

64 Table 2-5. Summary of ICCs for all TMS assessments. Assessment ARB FDI EDC FCR Map Area: moderate* moderate good f~\ r\ r\ r\ gooa ( 6761 i 599) (.858) (.848) Recruitment Curve slope: poor good gooo y UUU ( 255) (.892) (.910) (.853) CoG, y-coord.: good good good moderate (.824) (.852) (.860) (.685) CoG x-CQord poor poor moderate moderate (.374) (.380) (.702) (.530) Map Volume: good (.846) MT-optimal position: good MT-vertex: good (.971) (.899) *Qualitative scores are based on the following scale of ICC values: >0.75 = good, 0.50-0.74 = moderate, <0.50 = poor reliability (Portney and Watkins 2000). Actual ICC values are included parenthetically. Summary of Results In general, the results of this experiment support the hypotheses. Measurement of motor map area demonstrated moderate to good test-retest reliability, with better reproducibility in the forearm muscle representations as compared to the intrinsic hand muscles. The FDI motor map volume assessment was highly reliable. We found good test-retest reliability for locating the CoG, but only in regard to its lateral distance (y-coordinate) from the Cz. The anterior-posterior (x-coordinate) location of the CoG demonstrated poor to moderate test-retest reliability. When assessed at the optimal position or over the Cz, motor threshold was highly reliable across test sessions. Finally, with the exception of the APB, recruitment curve slope demonstrated high test-retest reliability. The poor reliability for the APB recruitment curves was likely due to

PAGE 75

65 high within-subject differences in the peal<-to-peal< MEP area during this procedure. Discussion The results of this study demonstrate that TI^S measures of motor representation size, organization and excitability are reasonably reliable across multiple testing sessions. Although we did not find large changes in motor map area, test-retest reliability of this assessment may be dependent upon the muscle being studied, as map area was more reproducible in forearm muscles compared to intrinsic hand muscles. The results also indicate that motor map volume is a reliable measure of muscle representation excitability, especially when the evoked muscle responses are normalized to a CMAP. This process of normalization compensates for variations in the size of MEPs across testing sessions, which may occur due to differences in electrode placement or subject arousal. The locations of the motor maps, as measured by map CoG, followed a predicted somatotopic orientation that remained relatively stable in the mediallateral direction across testing sessions. Motor threshold was reliable when assessed at the optimal position and Cz, indicating that corticospinal excitability remains stable across testing sessions in normal subjects. Finally, the results demonstrate that recruitment curves provide a reliable measure of the inputoutput properties of muscle representations. Recruitment curves may lack sufficient reliability, however, if large differences in I^EP size occur across testing sessions, as was observed for the APB cun/es. The following sections provide an

PAGE 76

in-depth discussion on the study findings for each TMS measure employed in this investigation. Motor Mapping We found that the test-retest reliability of motor map area was better for forearm muscle representations than for intrinsic hand muscle representations. Other reports have also found map areas of intrinsic hand muscles to be only moderately reliable (Mortifee et al. 1994), with nearly significant differences {F=1A, /7=0.094, n=8) in map area across multiple testing sessions (Uy et al. 2002). In the following discussion on motor mapping, we first deal with issues related to the lower reliability of intrinsic hand muscle representations, and then discuss why motor maps of forearm muscle representations are more reproducible across testing sessions. There are several potential reasons, both neurophysiological and methodological, that map areas for the APB and FDI were only moderately reliable. These muscles may be more affected by these factors than the EDC and FCR, given the greater complexity of intrinsic hand muscle representations. First, Ml representations for these muscles might change due to spontaneous variations in the Ml cortical outflow. Studies using reversible deafferentation with a blood pressure cuff confirm that rapid alterations in MEP amplitude (Brasil-Neto et al. 1992a) and motor map area (Ridding & Rothwell 1997) occur within minutes of the onset of deafferentation. Although the mechanisms for motor map changes in ischemic deafferentation is likely different than in the

PAGE 77

67 present study, these findings indicate that the modulation of motor outputs from 1^1 occur quite rapidly. The cortical maps of intrinsic hand muscles are presumably more subject to these rapid changes than forearm muscles due to the higher number of corticospinal neurons devoted to hand muscle representations (Phillips & Porter 1977, Wassermann et al. 1992). Daily activity patterns have also been shown to affect the size of motor maps of intrinsic hand muscles. In a study on Braille proofreaders, PascualLeone et al. (1995) found that the size of FDI motor maps varied according to activity level. The maps were larger on workdays (i.e., concentrate hand use) than on weekends off of work. In the present study, we found that nearly half of the subjects typed on a computer for a significant portion of their workday. As with the Braille proofreaders, the intrinsic muscle maps in our subjects may have been influenced by variations in the time spent typing during the 2 weeks between testing sessions. The wrist flexors and extensors (EDC and FCR) maps would presumably be less subject to the effects of concentrated typing, as these muscles are less involved in the performance of the varied and repetitive movements required during this activity. Our finding of reduced reliability of intrinsic muscle map area demonstrates the limits for obtaining reproducible motor maps across multiple testing sessions. Future investigations, which attempt to use changes in map area as a marker for plasticity within the motor system, should consider that alterations in these maps may occur secondary to spontaneous fluctuations in the cortico-motoneuronal output or to variations in

PAGE 78

68 general activity patterns over time. Multiple baseline assessments would account for such variability prior to the introduction of a controlled behavioral intervention. Despite the apparent variability in intrinsic muscle maps, the investigation of these representations should continue, as they are elemental in the performance of highly skilled fine-motor tasks. EMG-related issues also may affect motor map area. In a study on the reliability of TMS to map the human masseter muscle, McMillan et al. (1998) found that maps were reproducible, but only if the electrodes were left in place between testing sessions. When the electrodes were removed, map area was significantly different between testing runs. The results of the present study may be influenced by electrode placement as in the McMillan et al. (1998) study. We placed electrodes over the muscle bellies in parallel with the muscle fibers and with an inter-electrode distance of the recommended 20 mm (Hermens et al. 2000). Despite our attention to inter-electrode distance, we found that electrode placement was difficult to replicate across testing sessions separated by such a long time period. Again, the greater complexity of hand muscle representations (Phillips & Porter 1977) combined with between-sessions electrode placement differences, may have resulted in higher variability in motor maps across testing sessions for these muscles. Intrinsic hand muscles are smaller and have a higher density of motor units than the forearm muscles. These muscles also have a comparatively more complex neural representation than the forearm muscles; with a greater number of cortico-motoneuronal connections (Brasil-Neto et al.

PAGE 79

69 1992a, Phillips & Porter 1977). If the electrodes were over different populations of motor units, then the corresponding representations would likely vary more in the ARB and FDI than in the EDC and FCR. To account for variability secondar/ to electrode placement in the small hand muscles, precise marking and recording of electrode placement is necessary. Needle or fine wire electrodes could also be used to investigate single motor units, which are not contaminated by the responses of other motor units (Turker 1993). Although these electrodes offer greater precision, their invasiveness and potential to cause subject discomfort should also be considered. Map Volume Map volume was assessed in the FDI as the sum of all MEP peak-to-peak areas, each of which was expressed as a percent of the CMAP. The reproducibility of FDI map volume in this investigation was similar (Mortifee et al. 1994) or better (Miranda et al. 1997, Uy et al. 2002) than in previous studies. We found map volume to be highly reliable between two testing sessions; which was an interesting result given that FDI map area was only moderately reliable. The higher reliability for FDI map volume was likely related to the fact that the MEPs were expressed as a percent of the CMAP for this measurement. Normalizing the raw waveforms in this manner corrects for variations in the size of the MEPs, which may occur due to differences in electrode placement across testing sessions. Because map volume is normalized to a known muscle response, this measure provides a good assessment of corticospinal excitability.

PAGE 80

70 which is less subject to errors in electrode placement than the assessment of map area. Center of Gravity Separate reliability analyses were performed on the lateral (y) and anterior-posterior (x) coordinates of the CoG, each of which represents the amplitude-weighted center of the motor map. The relationship of the different CoG locations, especially in the y-direction, may be used to determine the somatotopic organization of the motor cortex. In the present study we found that the CoGs did fall in a predicted somatotopic orientation (Penfield & Rasmussen 1950) that was maintained across the two testing sessions. The lateral distance from the Cz (y) was highly reliable for all muscles, with the exception of the FCR, which demonstrated a moderate ICC. Several potential factors may have resulted in the lowered test-retest reliability of the FCR lateral CoG coordinate. Coil position during TMS mapping over the motor cortex can influence the size of MEPs and concurrently, the size, shape and location of the map (Mills et al. 1992, Miranda et al. 1997, Pascual-Leone et al. 1994). We mapped all 4 muscles simultaneously; with the figure-of-eight coil handle positioned parallel to the midsagittal with a backward flowing inducing current. This position has been shown to produce optimal responses in the APB and FDI, while allowing for easier replication of coil position between stimulus sites (Pascual-Leone et al. 1994). The optimal coil position for the FCR, however, was previously shown to be at a 45-degree angle to the midsagittal (Pascual-Leone et

PAGE 81

71 al. 1994). This orientation produces a maximal induced current that flows at right angles to the central sulcus (Mills et al. 1992). Horizontal interneurons are oriented perpendicular to the central sulcus (Pascual-Leone et al. 1994) and are preferentially excited by TIMS (Hallett 2000, Mills et al. 1992). A less than optimal coil orientation for stimulating the FCR motor representation could, therefore, affect the location and size of the map, as well as the size of the MEPs elicited. Indeed, we found the FCR to have an average maximum MEP area that was 3 times less than the other three muscles, a smaller average map size, and the highest variability in CoG location between sessions. The smaller MEP size is likely to be the most influential factor in the decreased reliability of the ycoordinate in the FCR. At a less than optimal coil position, the stimulation that reaches the representation for this muscle could be at or just slightly above the threshold for excitation. Since variability of the MEP response is inversely related to stimulus intensity, the variability in the size of FCR MEPs is more likely to be affected by rapid and spontaneous fluctuations in corticospinal and segmental motor neuron excitability levels (Kiers et al. 1993). The effect of these fluctuations is that stimulation sites may produce a positive response on one occasion and a null response on another (Weber & Eisen 2002). This variability would affect the distribution of the map, and therefore, the location of the CoG. Mapping the FCR at higher intensity and/or with the coil oriented at 45 to 50 degrees from the midsagittal should alleviate the MEP variability and increase the reliability of CoG location for this muscle. Based upon our findings and those of

PAGE 82

72 Pascual-Leone et al. (1994), future investigations that assess CoG of the FCR should employ separate coil positioning and testing in order to obtain optimal stimulation of this muscle's cortical representation. Interestingly, reproduction of the CoG in the anterior-posterior direction was fairly low for all four muscles. This was an unexpected finding, as the primary motor cortex is distributed in a medial to lateral rather than an anterior to posterior direction. Coil orientation may again play a role in the inter-session variability of the CoG x-coordinate. In normal subjects, map shape often mirrors the pattern of induced current flow, being elongated along the axis of the coil (Wilson et al. 1993). As a result, there is a larger area over which the CoG may fall in the anterior-posterior direction when the coil handle is oriented parallel to the midsagittal, increasing the room for variation of this measure across sessions. Our finding poor reproducibility of CoG in the anterior-posterior direction contradicts a previous report that reported no significant change, in 3 subjects across three testing sessions (Miranda et al. 1997). Nonetheless, we found the mean change in CoG location for each muscle, both in the x and y directions, to still be less than what has been reported in the only other studies on CoG replication (Miranda et al. 1997, Uy et al. 2002). Further investigations should consider the effect of coil orientation on the distribution of map CoGs. Motor Threshold Motor threshold, or the smallest intensity level to produce a discernable MEP, demonstrated high test-retest reliability when assessed at the optimal

PAGE 83

position and over the Cz. Tliese findings indicate that the threshold for excitation of the corticospinal system remains relatively constant across multiple testing sessions in normal subjects. The fact that motor threshold was stable is relevant to the motor map area findings, as changes in motor threshold could affect the apparent size of these maps (Thickbroom et al. 1998). In this regard, we suggest that small changes in especially the APB and FDI map areas across testing sessions were not the result of fluctuations in motor threshold. Recruitment Curves The slope of the recruitment curves was found to have high test-retest reliability, with the exception of the APB, which displayed very low reliability. The APB is especially susceptible to replication of electrode placement because of its small size and proximity to other thenar muscles. As with the motor maps, this may produce high variability in the size of MEP responses between sessions. When comparing the mean slope of the recruitment curves of all subjects, the APB had the largest mean difference in slope and the largest standard error of the mean. To adjust for the variability in MEP between sessions, these data should be normalized as a percent of the CMAP. This procedure would correct for differences in the raw MEP sizes from session to session, as was demonstrated in the FDI map volume data. Ttie generation of recruitment curves involves measuring the excitability of neurons other than those in the core region of excitability that is activated at threshold. These neurons have a higher motor threshold, either because they

PAGE 84

74 are intrinsically less excitable by TMS or because they are distant fronn the point of stimulation (Hallett 2001). Recruitment curves have been found to be steeper in intrinsic hand muscles (Chen 2000) when compared to leg muscles (Devanne et al. 1997). Indeed, we found the mean slope of the APB and FDI to be slightly more elevated than in the forearm muscles. The different input-output characteristics of these muscles suggest that the slope of a recruitment curve is related to the amount of corticospinal input to a particular muscle, with the intrinsic hand muscles being more easily excited by TMS. The test-retest analysis of recruitment curve slope indicates that this is a reliable measure across test sessions, especially when the data is normalized to the maximum MEP area. We chose to use a linear model to represent these data even though the responses produced a sigmoid-shaped curve. The linear model might be the most practical way to describe the data. Ray et al. (2002) found that the s-model produces impractical results, returning inflated curve values at higher stimulus intensities. Similarly, Carroll et al. (2001) noted that linear slope was more reliable than the peak slope obtained using a sigmoid function. Finally, the linear model does not require a plateau value in the curve. This avoids using high stimulus intensities that result in subject discomfort. Strengths and Limitations Upon review of the literature, only five studies have explicitly examined the reliability of TMS (Carroll et al. 2001, McMillan et al. 1998, Miranda et al. 1997, Mortifee et al. 1994, Uy et al. 2002) only one of which addressed the

PAGE 85

75 reproducibility of recruitment curves (Carroll et al. 2001). Although one might argue that this is a sufficient body of work to establish test-retest reliability, there are methodological and statistical issues in these studies that need to be strengthened. The present study attempted to address some of these issues by using a more rigorous statistical analysis, coupled with multiple different assessments to provide parallel information on studying the nervous system with TMS. We assessed the reproducibility of numerous TMS measures with the ICC, as it is the most robust and preferred method to evaluate test-retest reliability (Portney and Watkins 2000). The historical approach to testing reliability involves the use of the Pearson correlation (r), which provides information on the degree of association between two covariates. The Pearson r does not, however, provide a measure of agreement between the variables. The ICC is a more powerful assessment of reliability as it addresses association and agreement among multiple scores. The ICC essentially describes the proportion of the variance within a data set that is attributable to each of the independent factors In the data set. Mortifee et al. (1994) was the only group to use the ICC to describe test-retest reliability of motor maps. Using an ANOVA, Uy et al. (2002) found no significant (F=2.4, p=0.094) change in map area across four testing sessions of varying inter-session testing intervals. McMillan et al. (1998) also employed an ANOVA, while Miranda et al. (1997) used descriptive statistics to characterize reliability during mapping. The use of an ANOVA or descriptive

PAGE 86

76 statistics alone does not address tlie degree of association and agreement between each factor separately and for each individual subject. The results of these studies do not clearly indicate if there were differences between individual sessions or individual muscle maps across sessions. The magnitude of reliability of these individual variables was also not demonstrated. In the present investigation, we found generally high ICCs for the map area, map volume and recruitment curve linear slope, which were similar or higher than ICCs reported by Mortifee et al. (1994) and Carroll et al. (2001). A large sample size was utilized in the present study, which strengthens the statistical power of the reliability analysis. The use of several assessments in four muscles helped to provide parallel information about the reliability of TMS procedures. Based on our findings, recruitment curve slope and motor threshold over the Cz might be the most reliable measures to assess changes in cortical excitability. Map area was most reliable in forearm muscles and is dependent upon replication of electrode placement across testing sessions. Thickbroom et al. (1998) suggests that changes in motor representations are probably a function of increases or decreases in motor cortex excitability, and should therefore be characterized by recruitment curves or motor threshold assessment. Our finding that FDI map volume was highly reproducible, despite only moderate map area reliability supports the notion that changes in excitability may influence map area, as map volume is dependent upon the sum total of MEP areas regardless of the number

PAGE 87

77 of active positions. Although the map area for the APB and FDI did not meet threshold for reliability, the lateral orientation of CoG for these muscle representations did. In addition, all muscle CoGs were organized in a predicted somatotopic orientation that was maintained across testing sessions. A limitation of the present study may be that the same stimulus intensity was used during the mapping procedure on the preand posttest sessions. Transient differences in motor threshold (i.e., the level of cortical excitability) could have occurred between the two sessions. Such a change would influence map area, volume, and possibly the location of CoGs, which could account for some of the occasions of reduced reliability. We did not, however, see large changes in motor threshold at the optimal position, so this may have had only a small effect on the findings. The results of this study suggest that motor map area and volume, center of gravity, motor threshold, and recruitment curve generation a reasonably reliable measures. Despite finding APB and FDI map area to be only moderately reliable, the inter-session reproducibility was similar to other reports (Miranda et al. 1997, Mortifee et al. 1994, Uy et al. 2002). Future studies should employ a rigidly fixed inter-electrode distance to control influences of cross-talk, as well as skin marking that would last over the course of the study. Adequate data normalization, appropriate positioning of the coil angle relative to the midsagittal, and multiple assessments of motor threshold throughout the testing session will also increase the reliability of TMS investigations that occur over multiple testing

PAGE 88

sessions. In addition, the results demonstrate that there is at least a minimal degree of unexplained variability in especially intrinsic hand muscle representations. Future investigations should, therefore, establish the test-retest reliability and limits of normality of the TMS measures prior to study initiation. Given these considerations and based upon these results, TIMS could be very useful in studying stable characteristics and/or activity-related changes of the human motor nervous system.

PAGE 89

CHAPTER 3 EXPERIMENT II: ACTIVHY-DEPENDENT PLASTICITY IN STROKE More than half of the 730,000 Americans who sustain a stroke each year are left with chronic deficits in upper extremity function (Taub et ai. 1999). Although physical rehabilitation may assist the stroke survivor in learning to compensate for these functional deficits, traditional therapeutic approaches have demonstrated limited efficacy to remediate motor skill in the affected arm and hand (Dobkin 1995). Research on a newly developed treatment, however, is providing initial evidence that recovery of skilled upper-extremity function is possible for individuals in the chronic post-stroke phase. This intervention, known as constraint-induced (CI) movement therapy, employs intensive practice of motor skills for several hours per day over a 2-week period. To encourage further use of the hemiparetic limb, CI therapy participants also wear a constraining mitt on the unaffected hand. The combination of the intensive practice and constraint is believed to reverse the process of learned disuse of the hemiparetic limb (Taub et al. 1993, Taub et al. 1999). Several recent reports indicate that CI therapy results in significant improvements in movement speed, quality and coordination, as well as increases in the amount of use of the affected limb (Kunkel et al. 1999, Miltner et al. 1999, Taub et al. 1993, Taub et al. 1999). In addition, a two-year follow-up study indicated that these 79

PAGE 90

80 improvements were maintained well beyond the therapeutic intervention (Taub et al. 1993). Animal models of stroke indicate that intensive treatment programs similar to CI therapy alter the organization and excitability of neural representations for movement (Friel et al. 2000, Friel & Nudo 1998, Nudo 1996, Nudo et al. 1997). For example, in squirrel monkeys with surgically induced cortical lesions, forced affected upper limb use caused an expansion of the cortical representations for muscles involved in massed task practice (Nudo & Milliken 1996). Accordingly, one tenet of CI therapy is that the intensive treatment regimen induces activitydependent plasticity in the affected primary motor cortex (Ml), and that the changes in the organization and excitability of these muscle representations subsen/e recovery of skilled movement (Taub et al. 1999). Evidence from human neurophysiology studies indicates that massive Ml reorganization does occur following stroke (Cicinelli et al. 1997; Cramer & Bastings 2000, Cramer et al. 1997, 2000, 2001a, 2001b; Levy et al. 2001; Liepert et al. 1998b, 2000, Nelles et al. 1997; Rossini et al. 1998; Traversa et al. 1998; Weiller C 1992; Zemke & Cramer 2002), however, the relationship between these changes and recovery of specific motor skills requires further investigation. Using transcranial magnetic stimulation (TMS), Liepert and colleagues (1998, 2000) demonstrated that the motor representation of the hemiparetic hand in chronic stroke survivors was reduced in size as compared to the intact hemisphere. Following a course of intensive movement therapy, the cortical map

PAGE 91

81 of the hemiparetic hand expanded into adjacent areas, increasing significantly from the pre-treatment testing. Similar findings have been demonstrated in studies employing electroencephalography (EEG), positron emission tomography (PET), and functional magnetic resonance imaging (fMRI) (Kopp et al. 1999, Levy et al. 2001, Nelles et al. 2001). Although these studies and techniques have provided preliminary evidence for Ml neuroplasticity following stroke, the results are not clearly related to improvements in function. In other words, there may be Ml changes that occur as a person is progressing towards recovery, but what is the relationship between this neuroplasticity and improvements in the person's repertoire of skills? One major limitation of the existing neuroimaging work is that limited or no behavioral measures were used to support the notion that brain reorganization underlies functional recovery. Despite identifying cortical changes, Nelles and colleagues (2001) did not find any significant improvement in use of the hemiparetic limb. Liepert et al. (1998) and Liepert et al. (2000) demonstrated a marked increase in the cortical representation of the affected hand subsequent to therapy. The only behavioral measure used, however, was a subject questionnaire. No objective measures of functional recovery were performed. Levy and colleagues' (2001) pilot study provides support for the theory of activity-induced cortical reorganization as a mechanism for improved use of the hemiparetic limb. Subjects in this experiment demonstrated modest improvements in speed and quality of movement, strength, and real world use of the affected arm (Levy et al., 2001). These

PAGE 92

results were obtained in only two subjects, which limits the ability to generalize from this study. Of the several neurophysiological techniques available to investigate the brain, TMS is perhaps the best suited to track plastic changes in the motor cortex. In mapping studies, TMS is used to indirectly excite corticospinal neurons within Ml. In this manner, TMS can be used to outline the location, size and excitability of the motor representation subsen/ing a particular muscle. The advantage of using TMS over fMRI and PET is that this technique maintains high temporal resolution, activates both excitatory and inhibitory inputs to the corticospinal tract, and does not require the subject to produce any movement. Since the introduction of TMS, numerous studies have been performed to investigate the somatotopic organization of Ml (Aimonetti et al. 2002, BrasilNeto et al. 1992, Roricht et al. 1999, Wassermann et al. 1992), stimulusresponse characteristics of the corticospinal system (Boroojerdi et al. 2001, Devanne et al. 1997, Ray et al. 2002, Ridding & Rothwell 1997, Thickbroom et al. 1998), and alterations in the motor system following skill practice (Liepert et al. 1998a; Pascual-Leone et al. 1995a, 1995b) or following stroke (Cicinelli et al. 1997; Liepert et al. 1998b, 2000, 2001; Traversa et al. 1997; Trompetto et al. 2000). In addition to providing information about the size and extent of Ml representations, TMS may also be used to track the expansion of a representation into adjacent cortical territories. Expansion of motor

PAGE 93

83 representations may be assessed by tracking movement of tfie center of gravity (CoG). The CoG represents the amplitude-weighted center of a motor representation (Wassermann et al. 1992), and is thought to be the low threshold area where corticospinal neurons projecting to the muscles of interest are most concentrated (Escudero et al. 1998). Evidence from the animal literature demonstrates that representations "invade" adjacent and perhaps less active cortical areas under the conditions of massed practice (Nudo & Milliken 1996), resulting in a change of the CoG location. A similar process may occur in human stroke following CI therapy, although the existing literature is equivocal on the direction and relevancy of a shift in representation location. One additional benefit of TMS is that it may be used to study precise changes in corticospinal excitability. Several TMS studies in stroke survivors have shown that the level of excitability is often depressed in the damaged hemisphere relative to the undamaged hemisphere (Trompetto et al. 2000). Stroke survivors with a high threshold to stimulation and/or absence of motor evoked potentials (MEPs) from the affected hemisphere; typically have a poor functional recovery. Individuals with a lower threshold to stimulation in the damaged hemisphere, however, demonstrate a better functional outcome (Trompetto et al. 2000). To investigate changes in corticospinal excitability, TMS may be used to characterize the stimulus-response properties of the damaged motor cortex. Also known as stimulus-response cun/es, the recruitment curve represents how MEP amplitude increases as a function of stimulus intensity. This

PAGE 94

measurement is thought to assess the excitability of neurons other than those in the core neuronal population that is activated at MT (Chen 2000, Hallett 2000). The threshold for these neurons is higher either because they are less excitable or because they are further from the site of stimulation (Hallett 2000). Assessment of recruitment curves following skill practice/learning offers another means to study plasticity in the nervous system. The purposes of this study were three-fold. First, we sought to investigate the impact of CI therapy on motor cortex representations for two intrinsic hand muscles and two forearm muscles in the affected upper extremity. By examining four muscles with different functions, this investigation demonstrates which of the muscle representations are most altered by CI therapy. In addition, multiple neurophysiological measures were employed to provide a comprehensive description of how the organization, location and excitability of Ml representations are altered by intensive practice. The second purpose of this study was to determine the relationship between any neurophysiological changes and specific improvements in motor skill as well as real-world use of the affected arm and hand. Finally, because our subjects naturally fell into either a high functioning or low functioning category, we sought to explore differences in neuroplasticity and recovery of function in these sub-groups. Based upon the animal literature and Liepert et al.'s findings, we predicted that motor maps would increase in area and volume, the slope of the recruitment curve would increase, and that the threshold for excitation of the

PAGE 95

85 motor cortex would decrease following CI therapy. In all cases of plasticity, we predicted that such neurophysiological changes would be correlated with improvements in motor skill. Methods Subjects The participants were 23 individuals (10 female, 13 male) between the ages of 49 and 83 years (mean 65.8 10.3 years), who were 10 months to 10.75 years post-stroke (mean 4.0 3.3 years) upon entering the study. Demographic details of these subjects are presented in Table 3-1. These individuals met the following inclusion criteria: history of a clinically diagnosed stroke, met minimal active movement criteria in the affected fingers and wrist (see below); ability to follow simple instructions, a score of 24 or higher on the Mini Mental State Exam; the ability to sit independently without back or arm support for 5 minutes; the ability to actively participate for 6 hours of therapy without long rest periods; and passive range of motion of all upper extremity motions of at least half the normal range. These criteria have been demonstrated as effective parameters to select appropriate candidates for an intensive therapy program (Taub et al. 1993). Exclusion criteria were: history of seizures and/or epilepsy; presence of pacemaker or other metal devices in or around the head and upper thorax; tactile/proprioceptive sensory deficit; presence of major depression or other psychiatric disorder; and history of other neurological disease (i.e., multiple sclerosis). These subjects were classified as either high or low functioning based

PAGE 96

86 upon specific motor criteria. High functioning subjects demonstrated the ability to actively move the wrist through at least 20 of flexion-extension range and the thumb and fingers for 10 of active flexion-extension range at the MP and PIP joints. Low functioning subjects were only able and required to demonstrate at least a trace amount of active wrist extension and at least trace extension in any two fingers at the MCP and IP joints. These motor criteria were previously established as an effective means to categorize hand function in individuals with stroke (Taub et al. 1999). Based upon these criteria, 12 participants were classified as "high functioning" and 11 participants as "low functioning". All aspects of participant recruitment and testing met the approval of the local Institutional Review Board prior to initiation of this project. Each participant signed a written informed consent prior to his or her involvement with any of the study procedures. General Study Procedure All of the participants underwent both TMS and behavioral assessments prior to the study intervention. These subjects then participated with 2 weeks of CI therapy. TMS and behavioral assessments were again administered immediately following completion of the 2-week therapy program. Further details of the study intervention and testing procedures are provided below.

PAGE 97

87 Table 3-1. Study participant demograpliics including age, gender, involved hemisphere, affected limb (dominant or non-dominant), and time since stroke. Gender blue or btroKe Lxrnnanoe 1 llTE Sinuc SUUKc ^^yrbj 1 68 M R brain iNon-oom 2 52 M M L brain LJornnanc 1 7"^ i./j 3 72 M L brain uomnanc in 4 83 r R brain INOn-QCXn Z. 0 5 65 M M L brsin UDrnnanL 6 76 R brain NOTHjOm n fn 7 75 L brain iNon-aom J.J 8 67 Lorain Non-dom z 9 75 L brain uomnanc i.tJJ 10 80 L brain uomnarx 1 1 11 49 L brain uomnanc Z.OJ 12 54 L brain uomnanc t. J 13 "TO 78 R brain iNCXrClOm iU. /j 14 71 L brain iNorHX)m Z.iD 15 60 M R brain iNOTmorn O.J 16 59 M M K Drain INOTmOrn J.O/ A —J 17 74 M L brain uomnarx 1 1fi X.ID 18 59 M R brain iNOTHJOm Z.Ud 19 70 M R brain I'^JorKlom 0.75 20 54 F R brain Dominant 1.42 21 49 M L brain Dorrinant 7.25 22 65 F R brain Dorrinart 1.83 23 58 F R brain |Nion-dom 10.16 Maanage: 65.8 (10.3) years 10 Females 11 R Brain 12Doninart Mean time since CVA: 4.0 (3.3) years

PAGE 98

88 TMS Procedures Subject preparation for TMS testing The participant was comfortably seated in a semi-reclined modified dental chair with a pillow support placed beneath the affected limb. Passive bipolar surface electrodes were applied over the first dorsal interosseous (FDI), abductor pollicis brevis (APB), extensor digitorum communis (EDC), and flexor carpi radialis (FCR) muscles in the affected limb in a belly-tendon arrangement. Correct placement of the electrodes was verified by asking the subject to maximally contract the muscle while the investigator monitored for an EMG output amplitude of approximately 0.4 mV, or by palpation when the subject was unable to voluntarily activate a muscle. The inter-electrode distance was fixed at 20 mm for all muscles. EMG signals were filtered with a bandpass set at 2-10 kHz, rectified, and amplified with a Viking II Electromyograph (Nicolet Biomedical, Madisson, WI). Audio feedback from the electromyograph was monitored to ensure muscle relaxation during the testing session. A latex swim cap was placed on the participant's head so that a coordinate system could be clearly created. The vertex (Cz) was marked as the intersection of the nasioninion and interaural lines. Measurement of these lines (in cm) was recorded to ensure consistent location of the Cz across testing sessions. All TMS stimulation points were recorded in reference to the Cz (see Figure 3-1).

PAGE 99

General TMS testing procedure Five primary TMS variables were investigated: (1) size (area) of motor maps, (2) volume of motor maps, (3) location of the center of gravity (CoG) of tine motor map, (4) recruitment curves, and (5) motor threshold. Two different stimulators were used during the TMS procedures. Stimulation during motor mapping was delivered using a Magstim Rapid (Magstim Company Limited, UK) magnetic stimulator through a 5 cm mean loop diameter figure-of-eight shaped magnetic coil. Stimulation during the recruitment curve procedure and in determining the motor threshold at the vertex was performed using a Magstim 200 (Magstim Company Limited, UK) magnetic stimulator through a 9 cm mean loop diameter circular shaped coil. All assessments were performed during two separate testing sessions, separated by 2 weeks. Determination of optimal position and motor threshold The technique for stimulation was performed as described by Wassermann et al. (1992). The coil handle was oriented sagitally, with the handle pointing posteriorly and the figure-of-eight coil situated tangential to the skull. Stimulation was delivered over the affected hemisphere, contralateral to the muscles of interest (i.e., in the affected upper extremity). With the stimulator set at its maximum output and with the subject relaxed, the "optimal position" for stimulation was identified and its location recorded in relation to the vertex. The optimal position is defined as the optimal stimulating point for eliciting the largest amplitude MEPs. Once the optimal position was

PAGE 100

90 determined, motor threshold (MT) was assessed in a step-wise fashion. MT is defined as the lowest stimulation intensity that elicits discernable MEPs in at least 5 of 10 consecutive stimulations using an oscilloscope gain of 200 |jV per cm (Wassermann et al. 1992). (Wassermann et al. 1992). To compensate for possible initial heightened arousal levels and/or startle responses, which might affect MEP threshold, several trial stimulating runs were performed prior to the final assessment of MT at the optimal position. TMS mapping A 5 X 5 cm grid was marked on the swim cap and centered at the optimal position (25 spots, separated by 1 cm; see Figure 3-1). The stimulator was set at 115% of the MT and five stimuli were delivered to each spot at a frequency of IHz. The EMG responses from these five stimuli/spot were averaged online using Viking II nerve conduction software (Nicolet Biomedical, Madison, WI). After all grid positions were stimulated, the grid was extended, as necessary, until the area from which MEPs were elicited was surrounded by stimulated sites that did not elicit MEPs discernible at an oscilloscope display gain of 200 pV per cm in any muscle. This method ensures that the full extent of the motor map is captured (Triggs et al. 1999). Recruitment curve procedure The recruitment curve procedure was performed following the mapping procedure, and in a manner previously described by Ray et al. (2002) and Boroojerdi et al. (2001). Due to technical issues, the recruitment curve

PAGE 101

procedure was only performed in the last 11 subjects to participate in the study. The subjects were prepped in the same manner as in TMS mapping. Magnetic stimulation was delivered over the vertex using the circular coil. The coil was Figure 3-1. Location of the vertex (Cz), interaural and nasion-inion lines, stimulating grid, and figure-of-eight coil orientation on the head. placed tangential to the skull, with the handle oriented sagittaly. Magnetic stimuli were applied at 5% steps between 30 and 100% of the maximum stimulator output. Five stimuli were delivered at each intensity, at a rate of 0.2 to 0.3 Hz. The MEPs produced at each stimulation intensity were rectified and averaged online. These data were used to construct a stimulus-response curve.

PAGE 102

92 with MEP peak-to-peak amplitude plotted as a function of TMS intensity. The slope of this curve was analyzed for preto post-CI therapy changes. TMS data analysis Viking II nerve conduction software was used to determine the mean area of rectified MEPs elicited during the motor mapping and recruitment curve procedures. For motor mapping, the mean area of MEPs elicited at each stimulation site were normalized to that of the stimulus site which produced the largest MEPs. In this manner, the muscle representation area was quantified as the number of stimulation sites that elicited MEPs of area >10% of the MEP area for the stimulation site that produced the largest MEPs. For map volume data, the waveform peak-to-peak area of each MEP was expressed as a percent of the maximum MEP peak-to-peak area. These normalized MEP waveform areas were then summed to produce a volumetric measurement of each muscle's motor map. CoG was represented as the maximum amplitude-weighted position, and was calculated as follows: for each stimulating position on the map, the amplitude-weight was computed as the amplitude at that position divided by the sum of peak-to-peak MEP areas recorded for the map. The weight at any stimulating position was interpreted as the proportion of the total map area contributed by that location. For the recruitment curve data, MEP area was plotted as a function of stimulus intensity (i.e., percent of stimulator output). These data were then fit to a line of best fit for a linear model. Descriptions of each TMS outcome measure are summarized in Table 3-2.

PAGE 103

93 Table 3-2. A brief description of each TMS outcome measure. TMS Measure Data Outcome Measure Description Quantitation Motor Map Area Size of motor map Number of active stimulating points Motor Map Volume Excitability of motor map Sum of MEP peak-to-peak areas* Center of Gravity Location Tracks shifts in map Distance (mm) from vertex (y) locations and intra-aural line (x) Recruitment Curve Slope Input-output properties of MEP peak-to-peak area as a a motor representation function of stimulus intensity Motor Threshold Threshold of corticospinal Lowest stimulus intensity to excitation produce a discernable MEP *Peak-to-peak areas were normalized as a percent of tlie peak-to-peak area of the largest MEP. CI Therapy Procedures and Behavioral Tests All subjects were enrolled in a program of intensive upper extremity rehabilitation known as constraint-induced (CI) movement therapy. The CI therapy protocol entailed a 2-week (10 consecutive weekdays) period of intensive treatment. For this period, the unaffected hand and arm were immobilized with a padded mitt for 90% of the subject's waking hours. The mitt was used at all times except when certain agreed upon activities were being performed (e.g., excretory functions, naps, when the unaffected limb was used for an assistive device in walking, other circumstances when safety might be compromised). This immobilization of the unaffected hand and arm induced greatly increased use of the affected extremity. During weekdays of this 2-week period, subjects received supervised task practice using their affected hand and arm on a variety of activities for 6 hours. These activities included food preparation, eating, grooming, home maintenance, games and hobbies, etc. The focus of the treatment was for the subjects to perform frequent repetitions of

PAGE 104

movements in the context of functional activities. A particular emphasis was placed on performing activities that require finger and hand manipulations while maintaining active wrist stabilization. To this end, the hand and forearm muscles studied during TMS sessions were predicted to be involved in repetitive practice during the CI therapy intervention. Behavioral measures related to upper extremity function were performed immediately prior to and following the CI therapy intervention. These functional assessments measured performance in fine-motor control, rapid movement, strength, and real-world use of the affected upper extremity. These assessments are detailed below and summarized in Table 3-3. Box and Block Test The Box and Block Test (BBT) is a fast repetitive task that measures grasp, transport, and release of small objects. Test-retest reliability was tested at 6-month intervals and is .94 and .98 (Mathiowetz et al. 1985). Performance on the BBT was measured as the number of blocks moved in one minute. Fitt's Tapping Task The Fitt's Tapping Task is an open-loop continuous tapping task that generally measures information processing. For the purposes of this investigation, the Fitt's task was used to evaluate rapid flexion and extension movement at the wrist in the affected upper limb. Subjects were required to rapidly tap a stylus on a sensor, producing the greatest number of taps possible in three 10-second trials.

PAGE 105

95 Grip and pinch strength dynamometry Grip and pinch strengtli of the affected hand was assessed using a standard Jamar dynamometer and a standard pinch meter, respectively. Subjects were instructed to produce a maximum contraction against the dynamometer on three consecutive trials, and the average of these trials was used in the data analysis. Performance was expressed in kilograms. Wolf Motor Function Test In general, the Wolf Motor Function Test (WMFT) is comprised of a series of 15 timed tasks and two strength tasks that require movement at all joints of the affected arm. Three specific tasks that involve fine-motor control were selected as measures of dexterity in the affected hand. These included: the soda can, pencil and paperclip pick-up tasks. For the can pick-up task, the subjects were required to take a 12 oz. soda can from a standard table position and bring it to within 1 inch of the mouth. This task requires the subject to first extend the fingers and abduct the thumb, and then to close them around the can in a cylindrical grasp pattern. For the pencil and paperclip tasks, the subjects were required to move the hand from the lap and pick-up these items from a standardized position on a table in front of them. The pencil and paperclip tasks require finger extension and either 3-jaw chuck or precision pinch prehensile patterns. Subjects were asked to complete these three tasks as quickly as possible. Performance was scored as the difference (in seconds) between completing the task with the unaffected and affected hands. We also assessed

PAGE 106

96 the mean difference between affected and unaffected hand trials on all 15 timed tasks on the WMFT. Motor Activity Log The Motor Activity Log (MAL) is a structured interview that assesses the subject's perception about "how much" he or she performed 30 functional activities at home using the hemiparetic limb. The MAL is a typical assessment of Yeal-world use' of the affected (hemiparetic) upper extremity that is used in CIMT studies (Taub et al. 1999). Using a 6-point scale, subjects were asked to rate the amount of use of the affected upper extremity for each of the 30 activities. The average rating on the amount scale was used in the data analysis. Statistical Analysis The paired t-test was used to analyze mean differences on preand postCI therapy assessments for TMS and behavioral measures. We predicted that mean map area and volume, recruitment curve slope, BBT and Fitt's performance, strength, amount of limb use would be significantly increased following CI therapy. Motor threshold and task completion time (on the WMFT) were predicted to decrease following CI therapy. CoG location was predicted to shift, but the direction of the shift was not specified. Differences between the location of all muscle's CoG were assessed for each testing session using separate one-way ANOVAs. Separate two-way ANOVA's (group x testing session) were conducted for each outcome measure to identify significant differences between the high and low functioning groups. To determine the

PAGE 107

97 relationship between behavioral and neurophysiological changes, functional outcome measures were cross-correlated with TMS outcome measures using a Pearson's correlation matrix. Table 3-3. A brief description of each behavioral outcome measure. Behavioral Measure Data Outcome Measure Description Quantitation Box and Block Test Grasp, transport & release Number of blocks moved in 1 min. Fitt's Tapping Task Rapid wrist flexion and Avg. number of taps in 10 sec. Trial extension to tap stylus Strength dynamometry Grip and lateral pinch Kilograms strength Can pick-up task (WMFT) Gross (cylindrical) grasp Time to pick up soda can and bring ability to mouth Pencil pick-up (WMFT) Three-jaw pinch ability Time to pick up pencil from table Paperclip pick-up (WMFT) Precision pinch ability Time to pick up paperclip from table Wolf Motor Function Test Upper extremity capacity Mean time difference (affected minus unaffected) to complete tasks Results We were unable to elicit any MEPs in the affected limb of two subjects in the high functioning group. Data from these individuals were not included In the analysis of the TMS and behavioral results. We also found in a number of subjects, that MEPs could only be elicited in some but not all of the four muscles studied. For this reason, the analyses of TMS measures by muscle were only based upon those instances in which at least one significant MEP could be elicited on the pre and/or post testing session. In the following sections, we first present the TMS and behavioral results of all 21 subjects, and then detail the findings obtained when the subjects were grouped by functional level (i.e., high versus low motor function).

PAGE 108

98 TMS Results A summary of TMS results by muscle Is presented in Table 3-4. Half of these neurophyslological assessments revealed statistically significant differences between preand post-CI therapy testing sessions. Significant findings had moderate to high effect sizes. The power analysis for non-significant findings revealed that large sample sizes would have been required to achieve 70% power at an alpha level of 0.05. Specific findings are detailed in the following sections. Table 3-4. Summary of TMS assessments results and power analysis. Number of participants needed for TMS 70% Power Outcome Mean Power and IVIeasure Muscle n Change f P d (%) alpha = .05 Map area APB 20 2.4 1.740 0.049* 0.55 53 32 (# active FDI 20 0.4 0.279 0.392 0.09 9 1200 stimulating EDC 20 0.7 0.700 0.247 0.22 17 194 positions) FCR 20 1.5 1.038 0.156 0.33 27 87 Map volume APB 20 125.0 3.702 0.001* 1.18 98 8 (sum of waveform FDI 20 47.8 0.660 0.259 0.21 16 215 peak-to-peak EDC 20 33.9 0.639 0.265 0.20 15 235 areas) FCR 20 163.7 2.040 0.028* 0.65 65 23 CoG (mm) APB 15 7.1 5.152 <0.001* 0.97 93 11 (Tracks shifts in FDI 13 4.8 4.304 0.001* 0.96 77 11 map location) EDC 14 5.3 5.770 <0.001* 1.32 96 7 FCR 7 9.2 2.886 0.028* 0.66 32 22 Recruitment APB 10 0.16 0.733 0.241 0.06 6 2645 Curve FDI 10 0.03 1.253 0.129 0.06 6 2645 (slope of curve) EDC 10 0.08 0.128 0.451 0.52 30 36 FCR 10 0.31 2.171 0.033* 0.56 33 31 Motor Threshold 17 -4.7 -2.644 0.009* 0.91 83 12 (lowest stimulation intensity to produce a significant MEP) *Significant at a=0.05.

PAGE 109

Motor map area and volume The mean pre and post map area results are presented in Figure 3-2. Following CI therapy, the number of active stimulating positions increased by 21% for the APB, 3% for the FDI, 7% for the EDC, and 27% for the FCR motor maps. Only the APB map area, however, demonstrated a significant increase in motor map size (^=1.74, yC7=0.049, £^0.55). Motor map volume increased in all muscle representations, with significant changes occurring in the APB (^=3.702, /7=0.001, £7^=1.17) and FCR (^=2.04, /7=0.028, £^0.65) representations. Map volume data is also displayed in Figure 3-3. Exemplary motor maps and map volumes from representative subjects are presented in Figures 3-4 and 3-5, respectively. Map Area 18.0 T — — 1 APB FDI EDC FCR Muscle Figure 3.2. Comparison of mean motor map area between pre-and post-CI therapy assessments. Area is expressed as the total number of active stimulating positions, which elicited an MEP with a peak-topeak area >10% of that of the maximum MEP. *Map area significantly increased in the APB (/C=0.049) following CI therapy.

PAGE 110

100 Map Volume z w 400.0 E 200.0 1 APB FDI EDC FCR Figure 3-3. Comparison of mean motor map volume between preand post-CI therapy assessments. Volume is expressed as the sum of the normalized (% of max response) peak-to-peak MEP areas within a given motor map. ^Significant increases in the APB (/7=0.001) and FCR (/7=0.028) map volumes were noted. Center of gravity The amplitude-weighted center of activation sites or CoG of the APB, FDI, EDC, and FCR representations showed small but significant shifts in location from pre to post assessments. The absolute movement of CoG ranged from 5.3 mm {t=SJ7, yCK.OOl) in the EDC representation to 9.2 mm (^^2.89, p=0.028) in the FCR representation. Results of the absolute CoG shift for each muscle representation are presented in Table 3-5. When considering the direction of shift (i.e., +y=lateral, -y=medial), the mean shifts for the FDI, EDC, and FCR CoGs occurred in the lateral direction, while the APB CoG shifted slightly medial. The lateral shift of the FCR CoG approached significance (^^ 1.955, /7=0.098,

PAGE 111

101 0^=0.51, power=0.13), while changes in the location of the CoG for the APB {t=0.134, /7=0.895), FDI (^=0.322, yC7=0.753) and EDC (^^0.542, ^7=0.597) were clearly not significant. Table 3-6 demonstrates the direction of CoG shifts for each muscle representation in each subject. As Table 3-6 shows, the CoGs tended to shift laterally in approximately one half of the subjects and medially in the other half. APB Motor Map Representations— Subject 5. Pre-CI therapy Post-CI therapy Figure 3-4. Preand post-CI therapy motor maps of the APB in a representative subject. Marked squares indicate positions that produced an MEP with significant peak-to-peak area (i.e., >10% of the maximum MEP). The shading of each square represents mean MEP area elicited at each stimulation site as a percentage of the largest MEPs elicited in the APB. Note the increase in APB area for this subject following CI therapy.

PAGE 112

102 FCR Pre-CI Therapy Map Volume-Subject 5 80-100 60-80 40-60 20-40 0-20 FCR Post-CI Therapy Map Volume-Subject 5 80-100 60-80 40-60 20-40 0-20 Figure 3-5. Preand post-CI therapy motor map volume of the FCR in a representative subject. The mean MEP peak-to-peak area for each active position is expressed as a percent of the max MEP peak-topeak area. The y-axis represents lateral distance from the vertex (Cz). The x-axis represents anterior (-i-x) and posterior (-x) distance in cm from the intra-auralline. The total volume of the FCR representation increased from 437.5% at pre-CI therapy to 1153.8% at post-CI therapy for this subject.

PAGE 113

103 Within subjects, a lateral shift in one muscle representation's CoG was typically accompanied by lateral shifts in the other CoG's, although this was not a strict rule in all subjects. No significant shifts were noted in the anterior-posterior direction. As is shown in Figure 3-6, the locations of each CoG followed a predicted somatotopic organization, with the APB and FDI CoG's located lateral to those of the EDC and FCR. This ordering was preserved from pre to post assessments, with the exception of the FCR CoG, which shifted lateral to all other CoGs on the post assessment. Separate one-way ANOVAs revealed that there were no significant differences in the lateral location of each muscle's CoG on the pre (/^0.305, /7=0.82) or post {F=0306, p=0.82) assessment. This finding suggests that although there is a somatotopic order of the CoGs, there is also significant overlap of the neural representations for these four muscles. Figure 3-7 depicts the degree of overlap of neural representations of the four muscles studied. Table 3-5. Absolute movement of the center of gravity for each muscle representation. Muscle Center of Gravity Movement (s.d.) t sig. Effect Size Power APB 7.1 mm (5.3) 5.152 p<.001* 1.34 0.99 FDI 6.1 mm (5.1) 4.304 p^.oor 1.19 0.95 EDC 5.3 mm (3.5) 5.770 p<.001* 1.54 0.998 FCR 9.2 mm (8.4) 2.886 p=.028* 1.09 0.53 ^Represents a significant difference at a=0.05.

PAGE 114

104 Shift direction APB FDI EDC FCR lateral 7 4 8 5 medial 7 9 6 2 no change 1 0 0 0 no response 5 7 6 13 Counts under each muscle represent the number of subjects who experienced a lateral shift, medial shift, or no shift in CoG following CI therapy. The "no response" row represents individuals who did not exhibit significant MEPs in the muscle of interest on either the preor posttesting sessions. 4 8 4.9 5,1 ^ E 5 2 N E o L_ <^ o u c S in 5 *i PRE POST 5 3 5 4 5.5 5.6 5.7 ^ 5.8 0 fcr4 EDCl 0 FDii 0.5 0 -0.5 <--ant. (cm ) post.--> Figure 3-6. Comparison of preand post-CI therapy mean CoG locations for each muscle. Scale is in centimeters lateral to the vertex (y-axis) and anterior (-i-x) or posterior (-x) to the intra-aural line. These positions followed a relative somatotopic organization, which was generally maintained across testing sessions (except for FCR).

PAGE 115

105 Relative Motor Map Sizes and Overlap <-ant. (cm) post.-> 0.2 0 -0.2 0.4 -0.6 Figure 3-7. Comparison of relative motor map sizes, locations and overlap for motor cortex representations of each muscle. Scale as in Figure 3-6. Due to the degree of overlap in these four muscle representations, significant differences in the location of CoGs were not found. Motor threshold Motor threshold, or the lowest stimulation intensity level to produce a discernable MEP, was assessed in order to determine stimulation intensity during the mapping procedure. We also used motor threshold as an indicator of corticospinal excitability. Following CI therapy, motor threshold was significantly lower /7=0.009, cf=0.91, power=0.97) than on the pre-CI therapy assessment. This finding indicates that lower stimulation intensities were required to produce significant MEPs following the treatment intervention.

PAGE 116

106 Recruitment curves For the recruitment curve results, the mean peak-to-peak MEP area was plotted against TMS intensity and fitted with a line of best fit. Statistical analyses were, therefore, based upon the slope of this line for each muscle. As predicted, the slope of the recruitment curves increased following CI therapy for all muscles. Only the FCR recruitment curve slopes demonstrated a significant increase on the post assessment (^^2.171, p=0.033, d=0.S6, power=0.18). Mean slopes and t-test results are presented in Table 3-7, and mean pre and post recruitment curves for each muscle are presented in Figures 3-8a and 3-8b. Table 3-7. Mean recruitment curve slope results for preand post-CI therapy testing sessions. MEAN Recruitment Curve Slope MEAN MUSCLE Pre (S.E.) Post (S.E.) CHANGE t sig. ARB 1.48 (0.24) 1.64 (0.31) 0.16 0.733 0.241 FDI 1.8 (0.28) 1.83 (0.19) 0.03 1.253 0.129 EDC 1.72 (0.23) 1.8 (0.14) 0.08 0.128 0.451 FCR 1.06 (0.15) 1.37 (0.13) 0.31 2.171 0.033* *Significant at a=0.05.

PAGE 117

107 APB Recruitment Curves TMSinten>lty(%) 'APBpre ~ APBposI Linear (APBpre) ----' Linear (APBpost) EDC Recruitment Curves 120 I — — — — Figure 3-8a. Preand post-CI therapy mean recruitment curves for the APB and EDC (n=ll). MEP peak-to-peak area was normalized as a percentage of that of the maximum MEP, and then plotted as a function of TMS intensity (% of stimulator output). A line of best fit reveals the relative slope of these curves. Mean slope did increase slightly and non-significantly for these recruitment curves.

PAGE 118

108 FCR Recruitment Curves 120 T — -40 TMS Intensity ('/tj >FCRpre"" FCRposl Linear (FCRpfe) Unear (FCRposI) FDI Recruitment Curves -20 J TMS intensity (%) 'FDIpre FDipost Linear (FDIpre) -----Linear (FDiposI) Figure 3-8b. Preand post-CI therapy mean recruitment curves for the FCR and FDI (n=ll). MEP peak-to-peak area was normalized as a percentage of that of the maximum MEP, and then plotted as a function of TMS intensity (% of stimulator output). A line of best fit reveals the relative slope of these curves. Mean slope of the FCR demonstrated a significant increase (p=0.033) following CI therapy, and increased slightly but non-significantly on the post-CI therapy FDI recruitment curve.

PAGE 119

109 Behavioral Test Results Mean performance on most of the behavioral measures were significantly better on the post-CI therapy evaluation as compared to pre-therapy performance. Grasp and transport ability on the BBT improved by an average of 3 blocks (f=4.379, yC7<0.0001) (Figure 3-9a). Rapid repetitive tapping during the Fitt's tapping task increased on the post-CI therapy evaluation by an average of 3.8 taps per 10-second trial (f^2.921, /7=0.008) (Figure 3-9b). There was a trend for faster performance on the pencil (^^-1.731, ^0.099) and paperclip (f=-1.821, p=0.084) pick-up tasks in the WMFT, however, soda can pick-up time did not significantly change between preand post-CI therapy evaluations {t^0.599, yC7=0.582) (Figure 3-lOa). Despite not finding statistically significant changes in these components of the WMFT, the overall mean time to complete all tasks on the WMFT significantly decreased following CI therapy (f=-2.932, p=0.027). Pinch strength in the affected hand significantly increased (f=2.52, /7=0.021) following CI therapy, along with a nearly significant increase in grip strength (^=1.952, /7=0.066) (Figure 3-lOb). The amount of real-world use of the affected upper extremity significantly increased from a pre-treatment score of 0.7 to a post-treatment score of 2.0 on the MAL amount scale (f=5.884, /kO.0001) (Figure 3-11). Although this is a significant increase on the MAL, the clinical significance of this improvement may be questionable. A score of 2.0 on

PAGE 120

110 the MAL amount scale indicates that, on average, these individuals rarely used their affected arm and hand to perform common daily activities. (A) Box and Block Test 16 14 •a 12 o 10 a V) c tra 8 u 6 o 4 pre-CI Therapy post-CI Therapy (B) Fitt's Tapping Task 30 25 I pre-CI Therapy post-CI Therapy Figure 3-9. Mean preand post-CI therapy performance on the Box and Block Test (A) and Fitt's Tapping Task (B). *Significant increases in performance were found in both evaluations following CI therapy. Performance Time on 3 WIVIFT Tasks 120.0 100.0 0) 0) a E o •S 40.0 0) S 80.0 60.0 20.0 pre-CI Therapy post-CI Therapy pencil paperclip Grip and Pinch Strength 12.0 10.0 8.0
PAGE 121

Ill Motor Activity Log pre-CI Therapy post-CI Therapy Figure 3-11. Mean preand post-CI therapy scores on the Motor Activity Log amount of affected arm use scale. The mean subject rating of affected arm use prior to CI therapy was 0.7, which represents a score of "very rare" use of the affected arm. Following CI therapy, the mean rating significantly increased (p<0.0001) to 2.0, which indicates "rare" use of the affected arm in common daily activities. Group Results: High vs. Low Functioning In general, motor maps for all muscles were larger in the high functioning subjects (Figure 3-12a). APB map area significantly increased when controlled by level of function and the interaction between functional level and test session (Fi8=5.66, /C7=0.029, d=0.2A, power=61%). There was also a significant interaction effect between group and test session factors (Fi8=5.66, /7=0.028, 0^=0.24, power=61%), with APB map area increasing from preto posttest sessions in the high functioning, and no change in the low functioning group. Although no other significant changes in map area were found, we observed that

PAGE 122

112 map area did increase for all four muscle representations in the high functioning group. Both the APB and EDC map areas did not change in the lower functioning subjects across test sessions. APB map volume also increased when controlled by level of function and the interaction between functional level and test session (Fi8= 19.09, /?<0.001, d=0.52, power=0.96). There was a significant interaction effect (/i8=4.48, /7=0.049, d=0.20, power=0.52), with APB map volume increasing to a greater degree in the high functioning subjects (Figure 3-12b). The absolute movement of each CoG was larger for the low function group, although no significant group or interaction effects were found. These data were influenced by high between subjects variability, which may mask actual group differences. The lower functioning group demonstrated the largest preto post-CI therapy FCR CoG change, but these data were highly variable on the posttest evaluation. Figure 3-13a displays the mean absolute CoG movement of each muscle representation for high and low functioning groups. FDI recruitment curves significantly differed between the two groups (/^=5.02, p<0.055, £3^=0.39, power=0.50), with the low functioning group demonstrating larger (steeper) curve slopes. There was also a trend for a preto post-therapy increase in FCR recruitment curve (^6=4.48, yC7=0.079, cf=0A7, power=0.50). In general, we observed that recruitment curve slope was higher in the low functioning group (Figure 3-13b). Also, the high function group's recruitment curve slopes actually decreased in three of the four muscles studied

PAGE 123

113 (APB, FDI and EDC) following CI therapy, which is opposite to what was predicted. Motor threshold at the optimal position significantly decreased when controlled by level of function and the interaction between functional level and test session (/i5=7.07, /C7=0.018, cf=0.32, power=0.70). Motor threshold at the optimal position tended to be slightly lower for individuals in the higher function group, although no significant interaction effect was found. Scores on the BBT, Fitt's task, MAL amount, WMFT and pinch strength assessments all significantly increased when controlling for level of function and interaction effects. Under these same conditions, there was a trend for paperclip pick-up time to decrease and for grip strength to increase following CI therapy. We also found significant interaction effects between group and test session factors for the BBT and Fitt's assessments, with the high function group demonstrating a larger increase from preto posttest sessions (Figure 3-14a, 314b). Descriptive and ANOVA results of the behavioral assessments are presented in Table 3-8. TMS and Behavioral Results Cross-Correlation Data from the nine behavioral measures were cross-correlated with the data of 18 TMS measures (4 assessments of all 4 muscles and 2 assessments of motor threshold). Significant correlations are presented in Table 3-9 and are graphically displayed in Figure 3-15. Improvements on the Fitt's tapping task were moderately correlated with FDI map area (/^0.53, p=0.016) and volume

PAGE 124

114 {r^OAV, yC7=0.035). There was a fair correlation between MAL amount score and EDC map area (r=0.42, p=0.067). EDC map area increases were also inversely related to mean completion time of all tasks on the WMFT (r=-0.42, yC>=0.067). Improvement in grip strength was moderately correlated with EDC map area (r=0.57, ^0.012) and fairly correlated with EDC map volume (r=0.42, yC7=0.073). Grip strength was also negatively correlated with FCR recruitment slope (r=-0.79, yC7=0.035), however, these data were strongly influenced by one subject who demonstrated a very large increase in strength and decrease in recruitment curve slope following CI therapy. Additional fair to good correlations were found between some measures, however, these were highly influenced by a single outlier and should not be considered as significant. No other significant correlations were found. (A) Mean Map Area-HF vs. LF Groups LOW PRE LOW POST HIGH PRE HIGH POST APB FDI EDC FCR area area area area (B) Mean Map Volume-HF vs. LF Groups LOW PRE LOW POST HIGH PRE HIGH POST APB FDI EDC FCR vol. vol. vol. vol. Figure 3-12. Mean map area (A) and map volume (B) in high (HF) and low functioning (LF) groups. Both the size and volume of these representations were larger in the HF individuals. *Significant test session and interaction effects were found for the APB map area and volume.

PAGE 125

115 (A) CoG Movement: High vs. Low Functioning 2.5 E E ? 1.5 1 0.5 ill |OLOW iBHIGH APB FDI EDC FCR CoG CoG CoG CoG (B) Mean Recruitment Curve Siope-HF vs. LF Groups APB FDI EDC FCR slope slope slope slope LOW PRE LOW POST HIGH PRE HIGH POST Figure 3-13. Mean CoG (A) and recruitment curve slope (B) differences between high (HF) and low functioning (LF) groups. *There was a significant group effect for FDI recruitment curve slope, with LF subjects demonstrating steeper slopes. (A) 35 30 25 I 20 XI S 15 s 10 5 0 BBT-HF vs. LF t Pre-CI therapy .1 Post-CI therapy LF HF (B) Fitt's Tapping Task-HF vs. LF 40 35 30 25 20 15 10 5 -I 0 Pre-CI therapy Post-CI therapy LF HF Figure 3-14. Comparison of performance on the Box and Block Test (A) and Fitt's Tapping Task (B) in HF and LF groups. Significant interaction effects were found on both tests, with HF individuals showing a larger performance improvement from preto post-CI therapy.

PAGE 126

116 Table 3-8. Comparison of high (HF) and low functioning (LF) groups on behavioral assessments: descriptive statistics and ANOVA model effects. LFQap(n=125 l-FQap(n=G!) If QcLp(rp]2) l-FQap(n=g) TestSEsariBfel QapBfet IrtBaaicnBfedt l^(SE) h*m(SE) htei(SE) htei(SE) F 99 F 99 F 99 BT 17.57 (SS0) 30D(16L) 2241(691) 265
PAGE 127

117 Pitt's Performance and FDI Map Area in a. s o in 'H -30= 0 3232 -46FDI area (# active positions) IVIAL Amount and EDC IVlap Volume 8 (A O E _I < 2 -500 0 500 EDC volume (sum of %max) Grip Strength and EDC Map Area 33 1000 c EDC map area (# of active positions) Fitt's Tapping and FDI Map Volume -40= 0.2247 500 1000 15D0 FDI Map Volume (sum %max) WMFT and EDC Map Area EDC map area (# of active positions) Grip Strength and EDC Map Volume O) c
PAGE 128

118 Discussion The findings of this study support our hypotheses. First, in general, alterations in motor representation size, volume, location and excitability occurred following CI therapy, although not always to a statistically significant degree. The largest changes in motor map organization and excitability were primarily seen in the APB and FCR, indicating that these muscles may be more active during CI therapy than the EDC and FDI. In regards to the second aim of the study, neurophysiological changes were paralleled by improvements in coordination, movement speed, strength, and amount of use of the hemiparetic upper extremity. Although the EDC neurophysiological results did not reach significance, the motor map changes for this muscle were more strongly correlated with improvements in upper extremity function than the other muscles. Finally, clear differences in both neurophysiological changes and motor skill performance emerged when comparing high and low functioning individuals. Participants with a less severe motor deficit tended to demonstrate more robust alterations in the motor representation of the hand, and higher levels of function as determined by the behavioral assessments. Further interpretation of the study findings is detailed in the following discussion. Resolution of the First Hypothesis: Neurophysiological Alterations Increase in the size of motor representations The results of this study indicate that the size of certain upper extremity muscle representations in the motor cortex increase following CI therapy.

PAGE 129

119 Increases in the size of motor maps could be a result of increased use of the hemlparetic hand during the treatment intervention. An analogous study in adult squirrel monkeys with surgically induced Ml stroke demonstrated that upper extremity restraint training (similar to CI therapy) resulted in an expansion in both size and location of the motor representations of both intrinsic hand and forearm muscles (Nudo & Milliken 1996). These neurophysiological changes were paralleled by near complete recovery of hand function. Similarly, in humans, Liepert et al. (1998, 2000) found APB representations to increase following CI therapy, which were also accompanied by an increase in the amount of self-perceived real-world use of the hemlparetic limb. Our findings are in agreement with the results reported by Liepert and colleagues, in that we found an increase in map size to be paralleled by improvements in affected upper extremity dexterity, movement speed, strength, and amount of real-world use in a larger sample of subjects. Technical limitations in the TMS procedure should be addressed when considering activity-related changes in motor map area. First, TMS mapping of intrinsic hand muscle representations demonstrates only moderate test-retest reliability when assessed over two sessions. Based upon our findings in normal subjects (Experiment 1), map area of hand muscle representations changes across testing sessions, even in the absence of any behavioral intervention. These small changes in motor maps could be the result of rapid alterations in cortico-motoneuronal outputs (Kiers et al. 1993), or due to differences in EMG

PAGE 130

120 electrode placement between testing sessions (Hermens et al. 2000, Miranda et al. 1997). Such technical limitations of TMS mapping indicate that some of the preto post-CI therapy changes could be due to factors other than activitydependent plasticity of the nervous system. We presume, however, that variability due to these spurious factors was equivalent on both testing sessions, and that the expansion of motor maps were related to activity-dependent changes in the cortico-motoneuronal system. Furthermore, our results do not suggest that differences in map area occurred by chance, as we found post-CI therapy map area increases in the majority of the participants. To our knowledge, the impact of intensive therapy on FDI, EDC or FCR muscle representations has not been previously studied. In regard to the FDI, alterations in its motor map occur following intensive hand use in Braille proofreaders (Pascual-Leone et al. 1995) and during motor skill learning in healthy volunteers (Pascualleone et al. 1995). We did not, however, find any change in FDI map area in subjects undergoing CI therapy. As previously mentioned, the mapping of intrinsic hand muscle representations is subject to higher variability, and is accordingly somewhat limited by its moderate test-retest reliability (Experiment 1). High variability and lower reliability reduce the sensitivity of the measure and, therefore, may mask true differences that occur between testing sessions for the FDI representation. In regard to EDC and FCR motor maps, animal and human cortical electrostimulation studies demonstrate that the representations for the forearm

PAGE 131

121 muscles are smaller than those of intrinsic hand muscles (Nudo & Milliken 1996). Indeed, we also found that the EDC and FCR maps were 23% to 54% smaller than the hand muscle maps in our study participants, respectively. Perhaps due to the smaller map size and higher motor threshold of the EDC and FCR representations, we were unable to elicit MEPs in the EDC in 4 subjects nor in the FCR in 6 subjects. Because of the neurophysiological characteristics (i.e., relative size and excitability) of the EDC and FCR representations, as well as the lower number of subjects who demonstrated MEPs in these muscles, the detection of subtle yet meaningful changes in their motor maps may be difficult. In addition, there was large between-subject variability in EDC and FCR map size changes. Changes in the size of motor representations are likely dependent upon how active the muscles of interest are during CI therapy. This intervention is very task-oriented, with greater emphasis placed upon increasing the amount of use of the affected arm and hand, rather than the normalization of movement. To this end, subjects may have used gross shoulder or elbow movements to compensate for impaired wrist flexion, extension and stabilization. This is a commonly seen strategy in individuals post-stroke (Held 2000), especially in lower functioning individuals. With a less intensive dose of practice, the representations of the wrist stabilizers (i.e., EDC and FCR) are less likely to show activity-dependent changes. The FDI, on the other hand, should have been highly involved in the therapeutic sessions. This muscle is not only responsible

PAGE 132

122 for abducting the first finger; it is also active during precision grip (|viaier et al. 1993) and during in-hand manipulation of objects. I^any typical activities in CI therapy involve precise grasping and manipulation of objects, so the activity level of the FDI should have been relatively high. We are therefore surprised that there was no change in the FDI map following CI therapy. We noted, however, that more than half of the subjects demonstrated very minimal isolated finger movement, and thus relied upon a very gross and non-precise grasp to manipulate objects. In this way, the FDI's role during the intensive task practice was likely reduced, as compared to the APB, which is more involved in gross manipulation. Increase in the volume of motor representations The volume of motor maps increased following CI therapy. The largest and significant increases in map volume occurred in the APB and FCR representations. These results indicate that the overall excitability of the neural network supporting these representations increased following therapy. Stroke survivors typically demonstrate a higher threshold to stimulation and relatively smaller amplitude MEPs as compared to healthy subjects (Byrnes et al. 1999). This higher threshold and reduced MEP size could be due to a reduction in the neural network supporting muscle activation, but could also result from a reduced excitability of the motor cortex secondary to an imbalance between inhibitory and excitatory inputs to the corticospinal tract. The increase in APB and FCR map volume, therefore, indicates that the intensive training employed in

PAGE 133

the CI therapy influenced the excitability of the interneurons that synapse on pyramidal cells, or the corticospinal tract proper. Interestingly, the largest changes in motor map area and volume occurred in the APB and FCR. The explanation for this finding is not clear, although we suspect that these muscles were more active during the CI therapy intervention than the FDI and EDC. Individuals with hemiparesis often rely upon gross hand grasp with wrist flexion to manipulate objects. The APB and FCR are primarily involved in this type of movement. Stroke survivors typically have difficulty performing fine-motor manipulations, a function in which the FDI is involved. Likewise, wrist and finger extension (functions of the EDC) is commonly difficult to perform for this population, and is often substituted by compensatory shoulder and elbow movements. To this end, our findings suggest that CI therapy preferentially trained more gross patterns of manipulation, and that activity level was thus greater for the APB and FCR than the FDI and EDC. The behavioral results also support this notion, as the participants did not demonstrate significant improvements in the more demanding fine-motor tasks (i.e., the pencil and paperclip tasks). Decrease in motor threshold at optimal position As predicted, motor threshold at the optimal stimulating position was significantly lower following CI therapy. Like map volume, motor threshold is intimately related to the level of excitability within the area of motor cortex being stimulated during TMS. In a relaxed muscle, the motor threshold reflects the

PAGE 134

124 global excitability of the corticospinal pathway, including the large pyramidal cells, excitatory and inhibitory interneurons, and spinal motoneurons (Weber & Eisen 2002). Shifts in the level of excitation are, therefore, evident by changes in motor threshold. One might argue that motor threshold changes were the result of transient shifts in subject arousal within a session. To account for such a possibility, we assessed motor threshold after location of the optimal position, which allowed the subject adequate time to accommodate to the TMS. Based upon the results of Experiment 1, motor threshold should remain relatively stable across testing sessions when no intervention is applied. Our finding of a lower motor threshold, therefore, suggests that an activity-induced increase in corticospinal excitability occurred following CI therapy. Shift in center of gravity location We found highly significant differences in the medial-lateral location of the CoGs for each of the four muscle representations following CI therapy. This finding implies that the observed changes in motor map area are not only due to the increase in corticomotor excitability, as exhibited by the increases in volume of the maps and decrease in motor threshold. The shift in CoG suggests that areas adjacent to the original map location were recruited, as was found in an analogous study in squirrel monkeys (Nudo & Milliken 1996). Liepert and colleagues' (1998, 2000) also described differences in CoG location for the APB following CI therapy, which are consistent with our observations in a larger number of subjects. Although we found absolute differences in CoG location

PAGE 135

125 post-therapy, the shift did not preferentially occur in either the medial or lateral direction. In addition, a lateral shift in the EDC CoG, for exanaple, was sometimes accompanied by medial shifts in the remaining muscle CoGs within subjects. To this end, the present study does not indicate whether a lateral shift In a motor cortex representation is more or less advantageous than a medial shift. Small differences in CoG location could be due to technical limitations of TMS related to the spatial resolution of the technique (Brasil-Neto et al. 1992). The amount of inter-session movement of CoG in normal subjects (Experiment 1) ranges from 4.0 to 4.6 mm. These CoG shifts are smaller than the limitations of the spatial resolution of TMS, which has been reported at 5 mm (Brasil-Neto et al. 1992). In the present experiment in stroke survivors, we found larger shifts in CoG position, ranging from 5.3 to 9.2 mm. These changes are greater than the limitations of spatial resolution, and therefore suggest significant shifts in the location of muscle representations following CI therapy. Resolution of Second Hypothesis: Relationship Between Behavioral and TMS Results Behavioral outcomes We predicted that the observed neurophysiological changes would be correlated with improvements in hand dexterity, strength and in the amount of use. The behavioral measures revealed significant or nearly significant changes following CI therapy, indicating that this intervention resulted in some recovery of motor function in the hemiparetic limbs of these subjects. These findings

PAGE 136

126 parallel those of other reports, which have demonstrated similar improvements in chronic stroke survivors following CI therapy (Kunl
PAGE 137

127 Correlation between behavioral and TMS outcomes We found changes in the FDI representation to be moderately associated with improvements on the Fitt's tapping tasl<. The Pitt's tapping task assesses the rapidity of fractionated movement in the hand, a function that is highly under corticospinal control (Muir & Lemon 1983). During the Fitt's task, the FDI is likely active during precision grip of the stylus while the wrist flexor and extensor activity rapidly alternates to produce a tapping motion. Although the Fitt's task is partly dependent upon the FDI, it is also highly dependent upon control of the wrist extensor and flexor muscles. Surprisingly, we did not find improvements on the Fitt's to be directly associated with changes in the EDC and FCR motor representations. The correlation between Fitt's tapping performance and FDI map area and volume changes may simply be related to the role of this muscle in maintaining a precision grip on the stylus during the task, which would allow the subject to increase the rapidity of tapping. The most salient relationship between improved function and neurophysiological change is the correlation between grip strength and alterations in the EDC motor representation found in this study. Although the hand dynamometer directly tests the functional strength of intrinsic and extrinsic finger flexors, it also indirectly assesses the wrist flexors and extensors, which are required to stabilize the wrist during power grip (Basmajian 1964). To our knowledge, a parallel change in muscle strength and cortical representations has not been investigated. Several investigations, however, have established that

PAGE 138

128 steady contraction of the target muscle during TIMS increases the size of MEPs and lowers the threshold for stimulation, as compared to the relaxed muscle (Carroll et al. 2001, Devanne et al. 1997, Hess et al. 1987). The underlying mechanism for muscle facilitation is not completely understood, but likely includes increased excitability of the corticospinal system (Thompson et al. 1991, Wilson et al. 1993). Although muscle activation produces immediate results during the TMS session, a similar process might occur during more long-term increases in muscle strength. Improvements in grip strength (and thus wrist stabilization strength) are partly dependent upon a subject's ability to actively recruit motor units during maximal contraction, as well the ability to increase the rate of firing of these motor units. Individuals with hemiparesis typically demonstrate abnormally low and more variable motor unit discharge rates, which contribute to muscular weakness (Rothwell 1994). Improvements in strength partly reflect the person's ability to increase the rate of firing and recruitment of motor units; factors that are dependent upon supraspinal control (Rothwell 1994). At the cortical level, motor unit control may be reflected as an increase in the muscle representation's size and excitability, as seen in the present investigation. To this end, improvements in grip strength, which is dependent upon wrist stabilization, are partly related to alterations in the EDC cortical representation. We found fair correlations between EDC map volume and MAL amount scores, and between EDC area and mean task completion time on the WMFT. In

PAGE 139

129 combination witli the grip strength-EDC map relationship, these findings indicate that changes in the size and excitability of the EDC representations may serve as a neural correlate to improved strength, movement capacity and the amount of real-world use of the affected limb. Interestingly, the size and excitability of the EDC representation did not increase to a statistically significant degree. There are two potential factors that may have influenced this finding, both of which are important to establishing the relationship between neurophysiological and behavioral changes. First, there was high between-subjects variability in the EDC map, volume and recruitment curve data. Second, a small change in the EDC representation may be associated with a significant change in upper extremity function. In either case, a larger sample size would help to reduce the variability and would increase the power of the relationship. Out of a possible 162 behavioral-TMS cross-correlations, we only found seven instances of a fair to good correlation. The majority of these associations occurred between a behavioral measure that significantly improved and a TMS measure that did not. This finding indicates that despite finding significant changes in both TMS and behavioral measures, the relationship between them is not always clear. Three potential factors may be involved. First, the behavioral tests employed may not assess upper extremity functions that are directly related to alterations in motor representations. Second, CI therapy may result in therapy-induced changes in motor performance and neurophysiological function, but the amount of change differs between these two spectrums. Third, the

PAGE 140

130 improvements in function could be related to activity-dependent neuroplasticity in areas not assessed by TMS. In the final analysis, a strong direct correlation did not exist between cortical reorganization and improvements in both specific and global measures of upper extremity motor function. The parallelism between functional and neurophysiological changes, however, suggests that CI therapy drives plasticity and recovery on some level. Additionally, alterations in motor cortex representations likely serve as one of the neural correlates to recovery of movement following stroke. Resolution of Third Hypothesis: Effect of Level of Function In general, individuals with a less severe motor deficit tended to demonstrate more robust alterations in the motor representation of the hand, and higher levels of function as determined by the behavioral assessments. We found motor representations to be larger and more plastic in the high functioning group, with no change in the APB and EDC representation size in the low functioning group. There are three potential and inter-related factors that explain this disparity between higher and lower functioning individuals. First, individuals with a more severe motor deficit may have a much lower number of intact corticospinal fibers, which is evidenced by typically smaller (or absent) MEPs and a higher motor threshold than persons with better functional recovery (Cicinelli et al. 1997, Escudero et al. 1998, Trompetto et al. 2000). Experimental and clinical studies have shown that only 20% of the corticospinal fibers need be

PAGE 141

131 intact for the production of fractionated hand movement (Rossini, 2000). Subcortical strokes affecting the internal capsule could leave individuals with less than this critical amount. Two reports indicate that the degree of motor cortex plasticity is reduced in subjects with sub-cortical strokes (Cicinelli et al. 1997, Traversa et al. 1998). Information about the type and location of stroke was not available to the investigators in the present study, so we can only speculate that location of the stroke was a factor relating to neurophysiological and behavioral findings. A second potential factor influencing the outcomes relates to the nature of the treatment intervention. The key therapeutic factor in CI therapy is massed practice of functional tasks using the affected limb (Taub et al. 1999). This practice comes both during the 6-hour therapy day and during the time away from the clinic, when the subject is wearing the constraining mitt on the unaffected hand. Even after 2 weeks of CI therapy, lower functioning individuals in the present study reported very rare use of the affected limb during common daily activities, indicating that they still predominantly relied upon the unaffected arm and hand for performing daily activities. Because of their continued reliance on the unaffected limb, these individuals had a less intensive therapeutic experience, which likely affected the degree of activity-dependent changes in the motor cortex. The neurophysiological findings in the low function group were also paralleled by smaller improvements in fine-motor skill as compared to the high function group. The third and final factor relates to the location of plasticity. One theme for recovery of motor function following stroke is the

PAGE 142

recruitment of ipsilateral pathways. In the human, some 10-15% of the corticospinal fibers descend uncrossed (Lee 1995), which implicates the ipsilateral unaffected motor cortex as a potential contributor to recovery of function. Although these ipsilateral pathways play an apparent role in posthemispherectomy recovery (Cohen et al. 1991) and in recovery from perinatal lesions (Carr et al. 1993), studies with TMS have not confirmed the utility of these pathways for recovery of limb function in adult stroke. In fact, Turton et al. (1996) found that 9 subjects with MEPs evoked during stimulation of the ipsilateral (unaffected) hemisphere, typically had poor recovery. In the case of adult stroke, ipsilateral pathways may be non-functional and only predominate when the pathways from the contralateral (affected) hemisphere cannot recover (Hallett 2001). Lower functioning individuals in the present study might have incurred activity-dependent changes in the ipsilateral hemisphere, however, we did not assess the function of these ipsilateral pathways. Interestingly, we found that the low function group demonstrated larger absolute shifts in the CoG of each motor representation from preto post-CI therapy. This result could be indicative of the recruitment of motor areas adjacent to the original map location, as seen in the Nudo et al. (1996) study in squirrel monkeys. However, this finding also may be related to the diminished MEP size and slightly higher motor threshold found in the lower functioning subjects. Small MEPs are more variable and are more likely to be affected by rapid and spontaneous fluctuations in corticospinal and segmental motor neuron

PAGE 143

133 excitability levels (Kiers et al. 1993). This Is especially true for responses evoked on the outer perimeter of a motor representation (Mortifee et al. 1994). The effect of these fluctuations is that stimulation sites may produce a positive response on one occasion and a null response on another (Weber & Eisen 2002). This variability would affect the distribution of the map, and therefore, the location of the CoG. Mapping at a higher intensity would alleviate some of the variability in MEPs (Brasil-Neto et al. 1992), however, in many of the lower functioning subjects, the maximum stimulation intensity was already being used during mapping. Contrary to our expectations, recruitment curve slope was higher in the lower functioning individuals. We expected that slope of the recruitment curve would be higher in HF subjects, as more recovered stroke survivors typically have an increased level of motor cortex excitability than poorly recovered individuals (Trompetto et al. 2000). The steeper slopes in the LF group could be related to the location of the stroke and plasticity in areas other than the primary motor cortex. The large circular coil used to generate MEPs during the recruitment curve procedure produces fields that penetrate more deeply and over a larger volume of tissue (Weber & Eisen 2002). When a large portion of the hand area of the motor cortex is destroyed, activity in secondary or higher order areas occurs. For example, Weiller et al. (1992) demonstrated increased activation in the lateral premotor cortex and supplementary motor area during motor tasks in stroke patients compared to normal subjects. Increased levels of

PAGE 144

134 excitation in these other cortical areas, and their direct inputs into the corticospinal tract, may enhance recruitment curves in lower functioning subjects. Higher functioning individuals might also be less dependent upon these regions during the performance of rote motor skills. Nonetheless, this finding was predictive of more impaired upper extremity function, as increased recruitment curve slope was paralleled by smaller improvements on functional measures. Potential Mechanisms of ActivityDependent Plasticity Rapid alterations in cortical representations are the result of short-term changes in the connectivity of neuronal networks (Johansson 2000). Such alterations are not, per se, the result of anatomical changes, but rather shortterm adaptations in the existing circuitry. These represent the brain's attempt to make rapid adaptations in the face of a behavioral demand. Unmasking of 'silent' connections, and changes in overall excitation and inhibition represent such short-term adaptations. The alterations in motor representations (i.e., increased size and excitability and changes in location) in our subjects suggests that unmasking of these silent connections are a result of intensive and concentrate use of the hemiparetic limb. Short-term alterations in the functioning of neural circuitry also involve changes in the inhibitory and excitatory characteristics of the brain. For example, the excitability of the neuronal membrane may be altered as a function of the behavior of sodium ion channels (Halter et al. 1995). Such a change

PAGE 145

135 accounts for a lower excitation threshold for a given neuron (Chen 2000). Unmasking or changes in membrane excitability occur, in part, due to the removal of tonic inhibition (Jacobs & Donoghue 1991). In the case of brain damage or stroke, increased inhibition is noted (Liepert et al. 2000). Plasticity is enhanced with the release of tonic inhibition, which may be evidence for the cortex's attempt to repair itself (Hallett 2001). We observed decreased motor threshold and increased map volumes following CI therapy, indicating an increase in the overall excitability and perhaps disinhibition of the motor cortex. Limitations of the Study The investigators went to great lengths to elucidate the relationship between intensive therapy, neurophysiological changes, and the recovery of motor skill. Nonetheless, certain aspects of this study limit the power of the results. First, no control situation was employed. Although our findings in Experiment 1 indicate thatTMS is generally reliable in healthy individuals, the same may not be true for individuals with stroke or normal age-matched peers. To this end, some of the findings were possibly due to variability that is unrelated to therapy-induced changes. On the other hand, even though we did not always find statistically significant changes in neurophysiological parameters, the results of each TMS assessments almost always occurred in the predicted direction. We did not adjust the level of statistical significance for multiple t-tests or correlations. Correcting the alpha level in this manner reduces the probability of

PAGE 146

136 making a Type I error when multiple contrasts are made. Because this is a clinical study on a number of different and complex neurophysiological and behavioral mechanisms, an alpha less than 0.05 is probably too strict for the hypotheses being tested. Nonetheless, one could argue that because multiple ttests were used, some of the significant findings might have occurred by chance. We submit that future TMS studies of this nature should employ a larger sample size and then should correct for multiple comparisons. Some of the TMS results were obtained from a reduced number of subjects. We found that one or more of the studied muscles did not respond to TMS in a number of subjects. In such cases, a data point was essentially removed from the analysis, reducing the sample size and the generalizability of the results. The power analysis revealed that moderate increases in sample size would have been required to achieve 70% power at an alpha level of 0.05. Based on this issue and the high MEP variability in low functioning individuals, future studies may need to employ a larger sample size. Finally, although neurophysiological changes were paralleled by an apparent improvement in motor skill, there was a limited direct correlation between TMS and behavioral assessment results. The behavioral measures employed in this study may, in fact, lack the ability or sensitivity to assess motor functions that are highly dependent upon corticospinal function. In addition, part of the improvement of in the motor skills may have come from activity-

PAGE 147

137 dependent plasticity in areas not directly evaluated (i.e., association cortices or unaffected hemisphere). Summary The present investigation found changes in the organization and excitability of the primary motor cortex following CI therapy. This finding, coupled with the parallel improvements in motor skill, indicate that an intensive therapy program drives activity-dependent plasticity elemental to post-stroke recovery of motor skill. CI therapy appears to have a greater therapeutic and neurophysiological effect in higher functioning individuals, although further investigations with larger sample sizes are needed to properly address this issue. We generally found limited direct correlations between both specific and global indicators of motor skill and motor cortex reorganization. Functional measures that more directly assess corticospinal function should be employed to establish a more direct relationship. Finally, multiple baseline and follow-up testing sessions would add a very strong control situation, thereby, separating variability in testing techniques and random 'noise' in the central nervous system from true changes in the organization and excitability of the motor nervous system.

PAGE 148

CHAPTER 4 GENERAL SUMMARY AND CONCLUSIONS Specific and detailed discussion sections liave been provided concerning the two study experiments of TMS reliability and activity-dependent plasticity following stroke. The purpose of this final chapter is to summarize and integrate the key findings for each experiment. Experiment I Summary TMS may be a useful tool to study the organization, excitability and plasticity of the human primary motor cortex (Ml). As with any sophisticated neurophysiological technique, test-retest reliability should be established to determine any technical limitations in the equipment, general experimental procedure or data analysis. Reproducibility studies also serve to establish the limits of normal variation within the human central nervous system. The reliability of TMS assessments of cortical organization and excitability, however, has received very limited attention. The purpose of this experiment was to assess the test-retest reliability of several TMS measures that have been extensively used in neurological and rehabilitation research. The investigation of motor map area demonstrated moderate to good test-retest reliability. The stability of these motor maps was better for forearm muscles than for intrinsic hand muscles. Test-retest reliability of APB and FDI 138

PAGE 149

139 motor maps may be lower for three interrelated reasons. First, because of the large number of corticospinal neurons devoted to these muscles, the APB and FDI are presumably more affected by spontaneous fluctuations in Ml cortical outflow than the EDC and FCR. A second factor that probably influences the reliability of hand muscle maps is changes in daily activity patterns. For example, Pascual-Leone et al. (1995) noted that the FDI motor maps of Braille proofreaders were larger on workdays, i.e., following concentrated finger use, than on days off from work. Similar variations in APB and FDI maps may have occurred in the present study, as we found that half of the participants spent a significant portion of their day typing. Finally, because of the small muscle belly size, high motor unit density and complex cortical representation of the APB and FDI, the MEPs from these muscles are presumably more affected by slight intersession variations in EMG electrode placement than the forearm muscles. Future investigations of intrinsic hand motor maps should consider these factors and take initial measures to establish the limits of normality for the stability of their muscle representations. Despite finding only moderate test-retest reliability of FDI map area, we found the volumetric measurement of the FDI map to be highly reliable. This finding indicates that although the area measurement was somewhat unstable across testing sessions, the relative size of MEPs produced within the map was highly reproducible. Unlike the mapping data, however, responses used to generate a volumetric representation of the FDI map were normalized to a

PAGE 150

140 peripherally generated CMAP. Normalizing data in this manner corrects for variations in the size of MEPs, which may occur due to differences in electrode placement or subject arousal across testing sessions. Because map volume is normalized to a known muscle response, this measure provides a good assessment of corticospinal excitability, which is less subject to errors in electrode placement than the assessment of map area. The relative locations of the map CoGs followed a predicted somatotopic orientation and remained stable in the medial-lateral direction across testing sessions, with the exception of the FCR. Coil position during TMS mapping can influence the size of MEPs and location of the map (Mills et al. 1992, Miranda et al. 1997, Pascual-Leone et al. 1994). We mapped all 4 muscles simultaneously; with the figure-of-eight coil handle positioned parallel to the midsagittal and with a backward flowing inducing current. This position has been shown to produce optimal responses in the APB and FDI, while allowing for easier replication of coil position between stimulus sites (Pascual-Leone et al. 1994). The optimal coil position for the FCR, however, was previously shown to be at a 45-degree angle to the midsagital (Pascual-Leone et al. 1994). Based upon our findings and those of Pascual-Leone et al. (1994), future investigations that assess CoG of the FCR should employ separate coil positioning and testing in order to obtain optimal stimulation and accurate location of this muscles cortical representation. With the exception of the APB, recruitment curves for these muscles were highly reliable across testing sessions. In general, these data indicate that the

PAGE 151

141 excitability of neurons, other than those in the core region of activation, remains stable. As with motor mapping, recruitment curves may be likewise affected by between-session differences in electrode placement, as was probably the case with the APB curves. Skin marking or needle electrodes would help with electrode placement issues. The very low reliability of the APB recruitment curve supports the importance of normalizing data as a percent of the maximum response or to a known muscle response, i.e., a CMAP. Experiment II Summary Activity-dependent neuroplasticity in Ml is one neural correlate to recovery of upper extremity function following stroke. To provide a comprehensive assessment of Ml plasticity and any accompanying improvements in arm and hand function, we employed numerous neurophysiological and behavioral measures to assess the impact of CI therapy. The results of the study demonstrate that engaging the affected limb in intensive practice of motor skills results in alterations of the organization and excitability of muscle representations in the stroke affected Ml. These neurophysiological changes were also paralleled by improvements in upper extremity function. Of the muscle representations studied, the APB and FCR motor maps demonstrated the largest changes in size, location and excitability. Smaller changes were noted in the FDI and EDC representations, which were likely influenced by high variability in these muscles' responses to TMS. In addition, we presume that less significant alterations in FDI and EDC motor maps were

PAGE 152

142 related to the activity level of these muscles during CI therapy. Individuals with hemiparesis often rely upon gross hand grasp with wrist flexion to manipulate objects. The APB and FCR are primarily involved in this type of movement. Stroke survivors typically have difficulty performing fine-motor manipulations, a function in which the FDI is involved. Likewise, wrist and finger extension (functions of the EDC) are commonly difficult to perform for this population, and are often substituted by compensatory shoulder and elbow movements. To this end, our findings suggest that CI therapy preferentially trained more gross patterns of manipulation, and that the activity level was thus greater for the APB and FCR than the FDI and EDC. With a less intensive dose of practice, the FDI and EDC representations are less likely to show activity-dependent changes. The behavioral results also support this notion, as the participants did not demonstrate significant improvements in the more demanding fine-motor tasks. As predicted, motor threshold at the optimal stimulating position was significantly lower following CI therapy. Like map volume, motor threshold is intimately related to the level of excitability within the area of motor cortex being stimulated during TMS. In a relaxed muscle, the motor threshold reflects the global excitability of the corticospinal pathway, including the large pyramidal cells, excitatory and inhibitory interneurons, and spinal motoneurons (Weber & Eisen 2002). Our finding of a lower motor threshold, therefore, suggests that an activity-induced increase in corticospinal excitability occurred following CI therapy.

PAGE 153

143 We found highly significant differences in the medial-lateral location of the CoGs for each of the four muscle representations following CI therapy. This finding implies that the observed changes in motor map area are not only due to the increase in corticomotor excitability, as exhibited by the increases in volume of the maps and decrease in motor threshold. The shift in CoG indicates that areas adjacent to the original map location were recruited, as was found in an analogous study in squirrel monkeys (Nudo & Milliken 1996). Although we found absolute differences in CoG location post-therapy, the shift did not preferentially occur in either the medial or lateral direction. The present study, therefore, does not indicate whether a lateral shift in a motor cortex representation is more or less advantageous than a medial shift. One key behavioral finding might explain the minimal plastic changes found in some of the neurophysiological outcomes. We noted that, although the MAL scores significantly increased on the amount of use scale, these participants still reported using their affected arm and hand rarely and that they relied upon the unaffected arm to perform common daily activities. We presume that the greater severity of motor deficits in half of our subjects influenced how much they used their hemiparetic limb in the real-world setting, as the MAL results clearly show that these individuals were primarily relying on the unaffected arm following CI therapy. Since intense and frequent use of the affected limb appears to be critical to activity-dependent cortical reorganization, smaller plastic

PAGE 154

144 changes likely reflects to the subjects' Inability to use the affected arm and hand in their daily routine outside of the therapy sessions. Out of a possible 162 behavioral-TMS cross-correlations, we only found seven instances of a fair to good correlation. The nnajority of these associations occurred between a behavioral measure that significantly improved and a TMS measure that did not. This finding indicates that despite finding significant changes in both TMS and behavioral measures, the relationship between them is not always clear. Three potential factors may be involved. First, the behavioral tests employed may not assess upper extremity functions that are directly related to alterations in motor representations. Second, CI therapy may result in therapy-induced changes in motor performance and neurophysiological function, but the amount of change differs between these two spectrums. Third, the improvements in function could be related to activity-dependent neuroplasticity in areas not assessed by TMS. Nonetheless, the parallelism between functional and neurophysiological changes suggests that CI therapy drives plasticity and recovery on some level. In general, individuals with a less severe motor deficit tended to demonstrate more robust alterations in the motor representation of the hand, and higher levels of function as determined by the behavioral assessments. We found motor representations to be larger and more plastic in the high functioning group, with no change in the APB and EDC representation size in the low functioning group. There are three potential and inter-related factors that

PAGE 155

explain this disparity between higher and lower functioning individuals. First, individuals with a more severe motor deficit may have a much lower number of intact corticospinal fibers, which is evidenced by typically smaller or absent MEPs and a higher motor threshold than persons with better functional recovery (Cicinelli et al. 1997, Escudero et al. 1998, Trompetto et al. 2000). Second, lower functioning individuals may have relied more heavily on their unaffected limb to perform daily activities away from the laboratory. Because of their continued dependence on the unaffected limb, these individuals had a less intensive therapeutic experience, which likely affected the degree of activitydependent changes in the motor cortex. Finally, one theme for recovery of motor function following stroke is the recruitment of ipsilateral pathways. Although these ipsilateral pathways play an apparent role in posthemispherectomy recovery (Cohen et al. 1991) and in recovery from perinatal lesions (Carr et al. 1993), studies with TMS have not confirmed the utility of these pathways for recovery of limb function in adult stroke. In fact, Turton et al. (1996) found that 9 subjects with MEPs evoked during stimulation of the ipsilateral (unaffected) hemisphere, typically had poor recovery. In the case of adult stroke, ipsilateral pathways may be non-functional and only predominate when the pathways from the contralateral (affected) hemisphere cannot recover (Hallett 2001). Lower functioning individuals in the present study might have incurred activity-dependent changes in the ipsilateral hemisphere; however, we did not assess the function of these ipsilateral pathways.

PAGE 156

146 General Conclusions These experiments provide the most comprehensive investigation on the reliability and utility of TMS to investigate activity-dependent plasticity to date. By employing numerous neurophysiological measures, the results demonstrate which TMS measures are most reliable. The findings also detail which muscle representations are most stable across testing sessions, as well as those which are most influenced by CI therapy. Indeed, CI therapy may preferentially train muscles involved in gross manipulation, as evidenced by the limited neurophysiological and behavioral changes observed in muscles involved in finemotor control. Despite the reasonably high reliability of TMS assessments of Ml organization and excitability, some of these measures are subject to high variability in the evoked responses. This variability is probably the result multiple factors, many of which may be corrected with adjustments in the technical procedures. These factors include: coil orientation, replication of electrode placement and multiple baseline assessments. The latter consideration is probably most important to establish, as normal variations in Ml representations should be determined prior to the assessment of training-related changes. Taken together, these studies indicate that TMS is a reasonably reliable means to investigate activity-dependent plasticity in normal and stroke populations. By employing both neurophysiological and behavioral measures, this study provides mechanistic evidence for the neural correlates to recovery

PAGE 157

from stroke. Such information is critical to the development of new and effective therapies for the management of upper extremity hemiparesis.

PAGE 158

APPENDIX A PILOT DATA A pilot study was conducted specifically for the purposes of this investigation. The general aims of the pilot work was to demonstrate the feasibility of the transcranial magnetic stimulation (TIMS) mapping procedure, to identify descriptive trends in the mapping and behavioral data before and after constraint-induced (CI) movement therapy, and to guide the methodology and theoretical framework of the previously described project. Six volunteers from the Florida Biomedical Grant (Dr. Kathye Light, PI) were enrolled in this pilot study. Three males and three females, ranging from 51 to 80 years of age, were studied. Three were right CVAs and three were left CV As. Three of these subjects had their dominant limb affected. Demographic details are presented in Table A1. All subjects met the inclusion and exclusion criteria described in the methods section of the previously described Experiment 2. These 6 individuals participated with two-weeks of CI therapy and pre-treatment and post-treatment TMS mapping sessions. One subject (P-6) was excluded from the analysis due to excessive spontaneous muscle activity during preand post-treatment mapping sessions. Four TMS variables were investigated: (1) motor map area, (2) motor map volume, (3) location of the optimal stimulating position, and (4) motor 148

PAGE 159

149 threshold. Behavioral measures used were: Box and Block Test (BBT); can, pencil, paperclip pick-up and checker stacking tasks of the Wolf Motor Function Test (WMFT); and grip, pinch, and wrist strength. Descriptive statistics were used to characterize changes on all outcome measures. Pilot Study Outcomes Refer to Tables A-2 to A-5 for details on TMS and behavioral test outcomes. Subject P-1 P-1 was a 75-year-old male who sustained a right CVA. For the FDI, APB and FCR, this subject showed a decrease in the area of the motor map from preto post-treatment mapping sessions. The EDC motor map, however, demonstrated a marked increase in the motor map area: from 0 at pre-treatment to 8 at post-treatment. MT for this subject was 5% lower at the post-treatment session. The overall volume of the FDI and APB maps were markedly decreased at the post-treatment session, with little change for the EDC and FCR. P-1 demonstrated improvements on most of the preand post-treatment behavioral tests. He increased the number of blocks moved in the BBT from Ito 6. P-1 was unable to pick up the pencil and paperclip and stack the checkers on pre-testing, but was able to do so upon post-treatment testing. Strength measures showed minimal changes from session to session. Subject P-2 P-2 was a 76-year-old male who sustained a right CVA. This subject demonstrated a marked decrease in the area of the FDI motor map from preto

PAGE 160

150 post-treatment mapping sessions. The FCR area increased slightly from 2 to 4 active scalp positions. The APB motor map showed little change from session to session, and no active scalp positions were found for the EDC. MT remained constant across TMS mapping sessions. The overall volume of the FDI was markedly decreased at the post-treatment session, with a smaller decrease in the APB volume. The FCR volume was slightly increased from preto post-treatment sessions. P-2 increased the number of blocks moved in the BBT from 1 to 3 blocks. He was able to pick up the paper clip on the post-treatment evaluation, when he was unable to do so on the pre-test. He was unable to pick-up the pencil or to stack the checkers in either of the testing sessions. Pinch and grip strength increased slightly following treatment. Wrist flexion and extension strength was not tested following treatment due to equipment problems. Subject P-3 P-3 was a 62-year-old male who sustained a right CVA. The size of the area of the motor maps was relatively stable from preto post-treatment sessions, with the exception of the FCR, which showed a marked decrease in area from 25 active scalp positions at pre-treatment to 14 at post-treatment. MT remained constant across TMS mapping sessions. The volume of the FDI, APB and FCR motor maps decreased from preto post-treatment while the EDC map volume did not change. P-3's behavioral performance showed little change on the measures used. He was unable to move any blocks on the BBT, to pick up the pencil or paperclip, and to stack the checkers on either the preor post-

PAGE 161

151 treatment assessments. Wrist flexor and extensor strength were decreased on the post-treatment session, while pinch strength increased slightly. No change in grip strength was noted. Subject P-4 P-4 was a 77-year-old female who sustained a left CVA. This subject demonstrated no MEP activation with maximal stimulation during the pretreatment mapping session. Subsequent to therapy, this subject demonstrated robust patterns of activation for all muscles. P-4 demonstrated the second largest map volumes for the FDI, APB and EDC, and the largest FCR volume at the post-treatment mapping session. MT was determined to occur at a stimulation intensity of 80%. P-4 also showed marked improvements on behavioral measures. During the pre-treatment assessment, she had been unable to move any blocks on the BBT, or to pick up the pencil or paper clip. At post-test, P-4 was able to move 4 blocks, and was able to pick up the pencil and paper clip. She was unable to stack the checkers in either session. P-4's wrist extensor and flexor strength and grip strength decreased slightly from preto post-treatment, while pinch strength remained unchanged. Subject P-5 P-5 was an 80-year-old female who sustained a left CVA. This subject's APB, EDC and FCR motor maps increased in size from preto post-treatment, while the there was little change in the size of the FDI map area. The APB map was particularly enlarged subsequent to treatment. Additionally, no MEPs were

PAGE 162

152 elicited in the EDC and FDR on the pre-treatment mapping session, while I^EPs were elicited from 4 scalp positions on the post-treatment session for these muscles. MT was slightly elevated on the post-test session. The APB, EDC and FCR map volumes markedly increased while the FDI volume slightly decreased. P-5 showed no improvement on the BBT, pencil and paper clip picl< up, and checker stacking tasks, as she was unable to perform any of these at preor post-treatment. However, she did demonstrate increases on all strength measurements. Overall Results and Conclusions From Pilot Study The first three subjects showed a general decrease in the motor map size following treatment, while the fourth and fifth subjects demonstrated an increase in map size. The decrease in the first three subjects was an unexpected finding, as It was predicted that the motor maps would increase with use of the affected limb (which was found to occur in P-4 and P-5). This finding may have been, in part, the result of different experimenters performing the TMS technique. In the first subjects, Dr. William Triggs applied the EMG electrodes and performed the TMS mapping for the pre-test sessions (while the investigator assisted). For these same subjects, the investigator performed said procedures while Dr. Triggs assisted. Differences in the application of EMG electrodes (such as interelectrode distance, and placement over the muscle) could affect signal-to-noise ratio and thus influence presence and size of MEPs. Also, differences in the orientation of the stimulating coil may have lead to differences in how cortico-

PAGE 163

cortical interneurons in the Ml were excited. In subject P-4 and P-5, the investigator performed both pre and post-test mapping, thereby reducing the experimenter variability between sessions. In the previously described experiments, the investigator performed all preand post-treatment mappings. I^T threshold generally remained stable across testing sessions, with the exception of subject P-4, who had no observable MEPs during pre-testing and robust response on post-testing. MT is a relatively general measure of cortical excitability. Changes in the MEP (motor) map volume paralleled changes in the motor map area. This analysis is subject to the same experimenter limitations that occurred in determining the motor map. The relationship between behavioral changes and neural changes was inconsistent between subjects. While some subjects improved in behavioral measures, the expected neural changes in these subjects did not occur or went in the opposite direction (i.e. a reduction of the motor map area). Other subjects showed no improvement on behavioral measures, yet they had no change on the neurological measures. Again, the experimenter limitations outlined above may play a role in these findings. Those subjects with consistent pre and post-test mapping procedures (P-5 and P-4) did demonstrate increased map size and volume, which was paralleled by functional improvements in manipulation (P-4) or strength (P-5). Reducing experimenter-centered variability should help to elucidate the changes induced by the independent variables.

PAGE 164

154 Table A-1. Pilot subject demographics. SLtjel age gender side cf stroke domnancE P-1 75 M R brain noTKlom ?-2 76 M R brain noTKlom P-3 62 M R brain ncrKbn P4 77 F L brain doninant F5 80 F L brain donrinant 51 F L brain doninant Cemogaphc irforrration fbr pilot siijjels. DoninancE refes to \Artdi hanci/afTTi\Aas aflfe^ Note(*) ttiat SLi3jeGt P6 \Aas exduded from all anal\es because of e)
PAGE 165

155 Table A-3. Comparison of pre-and post-CI therapy map volume in 5 pilot subjects. VOLUME OF MOTOR MAPS FDI EDC subject pre post change subject pre post change P-1 13,3 3 1 -10.2 P-1 0 1 1 P-2 18.6 9.9 -8.7 P-2 0 0 0 P-3 24.2 16.6 -7.6 P-3 4.5 4.4 -0.1 P-4 0 126 12.6 P-4 0 2.9 2.9 P-5 2 9 1.9 -1 P-5 0 0.9 0.9 mean 11,8 8.82 -2.98 mean 0.9 1.84 0.94 SD 9.4 SD 1.2 APB PGR subject pre post change subject pre post change P-1 7.9 1.5 -6.4 P-1 1.1 0.8 -0.3 P-2 1.3 0.8 -0.5 P-2 0.2 0.4 0.2 P-3 6,1 3.9 -2.2 P-3 5.1 1.7 -3.4 P^ 0 9.7 9.7 P-4 0 2.3 2.3 P-5 0,9 10.1 9.2 P-5 0 0.5 0.5 mean 3,24 5.2 1.96 mean 1.28 1.14 -0.14 SD 7.2 SD 2.1 The number of scalp positions that elicited an MEP area 10% or greater of the largest MEP area are listed for each testing session. The absolute change in excitable spots is listed for each subject, with the group mean chanc and standard deviation for each muscle. Table A-4. Comparison of preand post-CI therapy motor threshold (MT) in 5 pilot subjects. su bject MOTOR p re THRESHOLD post change P-1 70 65 -5 P-2 85 85 0 P-3 75 75 0 P-4 > 100* 80 -20 P-5 45 50 5 mean -4 SD 9.6 Motor threshold is expressed as the minimum stimulus intensity (percent output) to produce a discernable MEP on 5 of 10 trials. *Note that we were unable to elicit any MEP's during maximal stimulation of the affected hemisphere in this subject. MT is therefore stated as >100% of stimulator output. Also, note that we were able to elicit MEP's following CI therapy, with a minimum stimulus intensity of 80%.

PAGE 166

Table A-5. Comparison of preand post-CI therapy results of behavioral assessments in 5 pilot subjects. BEHAVIORAL ASSESSMENTS Wrist Flexion (lbs) BBT (# blocks moved) 5 u b j c ct pre post subject pre post change p-1 5.5 5 -0.5 P-1 1 6 5 r 4. 0 NT N/A P-2 1 3 2 P-3 10.3 9 -1.3 P-3 0 0 0 P-4 6 4.3 -1.7 P-4 0 4 4 P-5 4.5 6.5 2 P-5 0 0 0 Wrist Extension (lbs) Pencil pickup (sec)* subject pre post rhannp subject pre post change P-1 1.4 0 -1.4 P-1 120 3.63 -116.37 P-2 7 NT N/A P-2 120 120 0 p.3 6.9 4 -2.9 P-3 120 120 0 p-4 4 2.6 -1.4 P-4 120 17.75 -102.25 p-5 3.6 4 0.4 P-5 120 120 0 Grip Strength (kg) Paperclip pick-up (sec)* subject pre post change subject pre post change P-1 7 8 1 P-1 120 9.9 -IlO.l P-2 2 3.5 1.5 P-2 120 10.18 -109.82 P-3 11 11 0 P-3 120 120 0 P-4 7 5 -2 P-4 120 39.6 -80.4 P-5 1 4.5 3.5 P-5 120 120 0 Pinch Strength (kg) Stacking Checkers (sec)* subject pre post change subject pre post change P-1 5 5 0 P-1 120 19.53 -100.47 P-2 0 0.5 0.5 P-2 120 120 0 P-3 8.5 12 3.5 P-3 120 120 0 P-4 3.5 3.5 0 P-4 120 120 0 P-5 2 2.5 0.5 P-5 120 120 0 •Note that a time score of "120" indicates that subject was unable to perform task in allotted time of 120 seconds

PAGE 167

APPENDIX B BEHAVIORAL TESTS FORMS FITTS TAPPING TASK P re -te s t / P o s t-te s t (circle one) S ubject Code D ate : A ffe cte d Side *10 seconds trial w ith 17 cm ta rg e t d ista n ce F ITTS una ffe c te d tap a ffe c te d tap Trial 1 Trial 2 T ria 1 3 Average Comments Additional Comments Tapping : "You have 10 seconds to tap as many times as you can. You want to move the probe up and down as rapidly as possible. Keep going until I say stop. I will give you one practice trial and then 3 more trials. Use the unaffected hand first. Any questions?" Figure B-1. Fitts Tapping Task score sheet. 157

PAGE 168

158 BQXA^DBLJOCKTES^ Pre-test / Post-test (dnde one) atjed Code / Deof Test I (affectedsicte: R/ q Ben* Uhaffeded Mected(R^L£ft) # OrrrrErts # Cdrrrrerts NiiTter of Bocks stanoe frcm CamEra to Box (suggested dstanoe is 10 feet) Mftional GomTGrts: Instnictions: 'IVbve the Uode, one at a time from one side to Hie ofherin one ninute. You have one niniite to move as many as you caa" Start with the unaffected side. Figure B-2. Box and Block Test form.

PAGE 169

159 WOLF MOTOR FUNCTION TEST P articipant D ate Affected side Time to Complete More Less Task Affected A ffected F 0 re a rm to ta b le Forearm to box Extend elbow Extend elbow-push weight Hand to table Hand to box Reach and retrieve weight L ift can Lift pencil L ift paperclip Stack checkers Flip cards Turn key in lock Fold to w e 1 L ift basket Time in seconds Enter score of 121 if subject unable to perform task. Figure B-3. Wolf Motor Function Test score sheet.

PAGE 170

160 MOTOR ACTIVITY LOG ACTIVITY A iVl U U N 1 U r U o t TURN ON A LIGHT SWITCH OPEN DRAWER REMOVE ITEM OF CLOTHING FROM DRAWbK PICK UP PHONE WIPE OFF COUNTER GET OUT OF CAR OPEN REFRIGERATOR OPEN DOOR USING KNOB USE TV REMOTE CONTROL WASH HANDS TURN WATER ON/OFF DRY HANDS PUT ON SOCKS TAKE OFF SOCKS PUT ON SHOES TAKE OFF SHOES GET UP FROM CHAIR WITH ARMRESTb PULL CHAIR AWAY FROM TABLE PULL CHAIR TOWARD TABLE PICK UP GLASS, BOTTLE, CUP OR CAN BRUSH TEETH PUT ON MAKEUP BASE, LOTION, SHAVING CREAM USE KEY TO UNLOCK DOOR WRITE ON PAPER CARRY OBJECT IN HAND USE FORK OR SPOON FOR EATING COMB/BRUSH HAIR PICK UP CUP BY HANDLE BUTTON SHIRT EAT HALF SANDWICH OR FINGER FOODS Motor Activity Los-Amount Scale 0 Did not use my weaker arm for that activity (not used). .5 1 Occasionally tried to use my weaker arm for that activity (very rarely). 1.5 2 Sometimes used my weaker arm for that activity but did most of the activity with my stronger arm (rarely). 2.5 3 Used my weaker arm for that activity about half as much as before the stroke (half pre-stroke). 3.5 4 Used my weaker arm for that activity almost as much as before the stroke (3/4 or 75% pre-stroke). 4.5 5 Used my weaker arm for that activity as much as before the stroke (same as pre-stroke). Figure B-4. Motor Activity Log.

PAGE 171

APPENDIX C INFORMED CONSENT TO PARTICIPATE IN RESEARCH IRB# 555-2002 Informed Consent to Participate in Research You are being asked to take part in a research study. This form provides you with information about the study. The Principal Investigator (the person in charge of this research) or a representative of the Principal Investigator will also describe this study to you and answer all of your questions. Before you decide whether or not to take part, read the information below and ask questions about anything you do not understand. Your participation is entirely voluntary. 1. Name of Participant ("Study Subject") 2. Title of Research Study The Reliability of Transcranial Magnetic Stimulation as a Tool for Measuring Therapy-Induced Neuroplasticity in Stroke 3. Principal Investigator and Telephone Number(s) Matthew P. Malcolm, ABD, OTR Principal Investigator Department of Physical Therapy Phone: (352) 265-0680 extension 891 1 1 161

PAGE 172

162 4. Source of Funding or Other Material Support This study is funded in part by a grant from the Florida Department of Health, and by the Department of Veterans Affairs Rehabilitation Research and Development. 5. What is the purpose of this research study? You are being asked to participate in a research study that records changes in brain activity. Following a stroke, the human brain experiences changes in how it generates movement of the arm and hand. Many survivors of stroke have difficulty using their brains to produce effective and coordinated movement of one of their arms. However, some research studies suggest that intensive occupational or physical therapy helps the brain to develop new strategies to generate improved arm and hand movement on the side affected by a stroke. These new strategies may be reflected as changes in the brain's activity pattern. One technique for measuring changes in brain activation is called transcranial magnetic stimulation (TMS). TMS is a safe and nonpainful method to study the brain. Although TMS has been frequently used to study changes in the brain, it is less well known if this technique will produce similar results when repeated over two or more sessions. The first purpose of this study is to determine if TMS produces repeatable results in 20 people who have sustained a stroke and in 20 people who have NOT had a stroke. As was previously mentioned, the brain's activity patterns may change in survivors of stroke following a course of intensive therapy directed at the affected arm and hand. The second purpose of this study is to use TMS to examine brain activation before and after intensive occupational/physical therapy. 6. What will be done if you take part in this research study? OVERVIEW If you take part in this study, a technique called transcranial magnetic stimulation (TMS) will be used to study one part of your brain that controls arm and hand movement. Twenty individuals who have NOT had a stroke will be studied, and 20 individuals who HAVE had a stroke will be studied. For those who have NOT had a stroke, you will be asked to participate with two separate evaluation sessions. These sessions will be held 2 weeks apart from each other. For those who have had a stroke, you will be asked to participate in three testing sessions, all separated by 2 weeks. TRANSCRANIAL MAGNETIC STIMULATION PROCEDURE TMS will be used to study changes in brain activity from one testing session to another. TMS is a non-painful method that may be used to stimulate your brain to cause a brief activation of the muscles in your forearm and hand.

PAGE 173

163 During the TMS procedure: • You will be seated comfortably in a modified dental chair. • A latex swim cap will be placed on your head. This will allow the Principal Investigator to make marks on the cap, rather than on your scalp. • Surface electrodes will be placed on your forearm and hand. These electrodes are designed to monitor activity in your forearm and hand muscles and do NOT provide an electric shock. The electrodes will be connected to a computer, which will record muscle activity responses. • Two different types of magnetic coils will be used to study your brain. First, a magnetic coil shaped like a "figure eight" will be positioned on your scalp, about three inches above the hairline and to side of the center of your head. This coil will be used to locate the best spot for stimulation and to create a map for part of your brain. During the stimulation, the surface electrodes and computer will record muscle responses in your forearm and hand. Following stimulation with the "figure eight" coil, a larger circular coil will be placed over the center of your head. This circular coil will be used to study how different TMS intensities influence the response of your forearm and hand muscles. As with the "figure eight" coil, muscle responses will be recorded with the surface electrodes and computer. • You will be asked to relax your muscles during the testing. The Principal Investigator will use the "figure eight" coil to deliver five stimuli to each of the designated scalp positions. Following this procedure, the circular coil will be used to deliver 1 0 stimuli over the center of your head. In both cases, you may feel a mild to moderate "tapping" sensation. You should NOT experience any discomfort during this process. PERIPHERAL NERVE STIMULATION Part of the testing procedure will be to determine the muscle response following stimulation to the peripheral nerv e that supplies a particular muscle. To do this, the Principal Investigator and the technician will use a device to deliver a mild electric stimulus to the skin near your wrist. This stimulation will be brief Surface electrodes will be placed on one of your hand muscles and connected to a computer. This set-up will allow the Principal Investigator to monitor your muscle activity. The intensity of the stimulation will be gradually increased until a maximum muscle response is observed. The electrical stimulation may cause mild and brief discomfort, but should not be painful. If the stimulation is too uncomfortable, please alert the Principal Investigator and/or technician so that the stimulation may be stopped. BEHAVIORAL TESTS For those individuals who have had a stroke, we will also be testing your general hand function during each evaluation session. These tests will require you to do things like picking up blocks, a paperclip, a pencil, and stacking checkers. Also, we will test your hand grip strength and wrist strength. In all cases, both of your hands and wrists will be tested. Finally, we will ask you to complete a questionnaire about "how much" and "how well" you think that you use your arm

PAGE 174

164 and hand that was affected by the stroke. For those individuals who have not had a stroke, we will ask you to fill out a brief activity questionnaire at the completion of your final testing session. 7, What are the possible discomforts and risks? Transcranial magnetic stimulation (TMS) may cause a seizure in individuals that have a history of epilepsy or previous seizures. For this reason, any individual who has a history of epilepsy or seizures will be excluded from this study. TMS may also interfere with devices such as a heart pacemaker or deep brain stimulator. For this reason, individuals who have a pacemaker or other metal implants in the head, neck or upper body, will be excluded from this study. The Principal Investigator will ask you if you have a history of seizures or epilepsy, and if you have a pacemaker or other implanted metal device. During the TMS procedure, you will feel a mild to moderate "tapping" on your scalp, which should not be painful. If this becomes uncomfortable for any reason, please alert the Principal Investigator or technician so that we may stop the procedure. During the peripheral nerve stimulation, we will be delivering a mild and brief electric stimulus over your wrist. This electrical stimulation may produce a short, mild discomfort. If this procedure becomes too uncomfortable, please alert the Principal Investigator or technician so that we may stop the procedure. In order to make TMS procedures as safe and effective as possible, we will exclude individuals with certain medical conditions. To protect your safety in this study, we ask that you review the following list of medical conditions. If you think that any of these conditions applies to you, we ask that you notify the Principal Investigator obtaining this consent before beginning the TMS procedures The names and phone numbers of all the investigators involved in this study are listed on the front page of this form. IMPORTANT WARNINGS: I. Persons with any history of neurological illness, including epilepsy or seizures, brain tumors, any neurological disease (other than stroke), any history of head injury with loss of consciousness of any length should not receive TMS. II. Persons with implanted heart pacemakers, deep brain stimulators or medication pumps, intravenous lines, metal plates in the skull, or metal objects in the eyes or skull should not receive TMS. III. Patients with a history of schizophrenia, manic-depression, or alcohol or drug abuse within the past year should not receive TMS.

PAGE 175

165 There are no other anticipated discomforts or risks associated with this study. Throughout the study, the researchers will notify you of new information that may become available and might affect your decision to remain in the study. If you wish to discuss the information above or any discomforts you may experience, you may ask questions now or call the Principal Investigator or contact person listed on the front page of this form. 8a. What are the possible benefits to you? The benefit to you for participating in this study is that you will learn about a part of your brain that contributes to hand and arm movement. 8b. What are the possible benefits to others? The information obtained from this study may help in the development of tools to measure brain activity. In addition, this information may be used to develop improved therapies for individuals who have sustained a stroke. 9. If you choose to take part in this research study, will it cost you anything? There is no financial cost involved with participating in this study. 10. Will you receive compensation for taking part in this research study? You will receive no compensation for participating in this study. 11. What if you are injured because of the study? If you experience an injury that is directly caused by this study, only professional medical care that you receive at the University of Florida Health Science Center will be provided without charge. However, hospital expenses will have to be paid by you or your insurance provider. No other compensation is offered. 12. What other options or treatments are available if you do not want to be in this study? The option to taking part in this study is doing nothing. If you do not want to take part in this study, tell the Principal Investigator or his/her assistant and do not sign this Informed Consent Form.

PAGE 176

166 13a. Can you withdraw from this research study? You are free to withdraw your consent and to stop participating in this research study at any time. If you do withdraw your consent, there will be no penalty, and you will not lose any benefits you are entitled to. If you decide to withdraw your consent to participate in this research study for any reason, you should contact Matthew P. Malcolm at (352) 265-0680 extension 891 1 1 If you have any questions regarding your rights as a research subject, you may phone the Institutional Review Board (IRB) office at (352) 846-1494. 13b. If you withdraw, can information about you still be used and/or collected? Only information that is obtained after your consent to participate and before your decision to withdraw will be used. No further information will be collected if you choose to withdraw from the study. 13c. Can the Principal Investigator withdraw you from this research study? You may be withdrawn from the study without your consent for the following reasons: • You have a history of epilepsy or seizures. • You have a heart pacemaker or other metal implanted device in your head, neck, or upper body. • The Principal Investigator decides that participating in the study would be harmful to you. • You require medical or other treatment that is not part of this study. • Study procedures have a bad effect on you. • You are unable to keep appointments as required. 14. How will your privacy and the confidentiality of your research records be protected? Authorized persons from the University of Florida, the hospital or clinic (if any) involved in this research, and the Institutional Review Board have the legal right to review your research records and will protect the confidentiality of them to the extent permitted by law. Otherwise, your research records will not be released without your consent unless required by law or a court order.

PAGE 177

167 If the results of this research are published or presented at scientific meetings, your identity will not be disclosed. 14b. Consent to be Videotaped and to Different Uses of the Videotape(s) With your permission, you will be videotaped during this research. Your name or personal information will not be recorded on the videotape, and confidentiality will be strictly maintained. When these videotapes are shown, however, others may be able to recognize you. The Principal Investigator of this study, Matthew P. Malcolm, or his successor, will keep the videotape(s) in a locked cabinet. These videotapes will be shown under his direction to students, researchers, doctors, or other professionals and persons. Please sign one of the following statements that indicates under what conditions Matthew P. Malcolm has your permission to use the videotape. I give my permission to be videotaped solely for this research project under the conditions described. Signature Date I give my permission to be videotaped for this research project, as described in the Informed Consent Form, and for the purposes of education at the University of Florida Health Science Center Signature Date I give my permission to be videotaped for this research project, as described in the Informed Consent Form; for the purposes of education at the University of Florida Health Science Center; and for presentations at scientific meetings outside the University. Signature Date 15. How will the researcher(s) benefit from your being in this study? In general, presenting research results helps the career of a scientist. Therefore, the Principal Investigator may benefit if the results of this study are presented at scientific meetings or in scientific journals.

PAGE 178

168 16. Signatures As a representative of this study, I have explained to the participant or the participant's legally authorized representative the purpose, the procedures, the possible benefits, and the risks of this research study; the alternatives to being in the study; and how privacy will be protected. Signature of Person Obtaining Consent Date Consenting Adults You have been informed about this study's purpose, procedures, possible benefits, and risks; the alternatives to being in the study; and how your privacy will be protected. You have received a copy of this Informed Consent Form. You have been given the opportunity to ask questions before you sign, and you have been told that you can ask other questions at any time. Adult Consenting for Self. By signing this form, you voluntarily agree to participate in this study. By signing this form, you are not waiving any of your legal rights. Signature of Adult Consenting for Self Date Parent/Adult Legally Representing the Subject. By signing this form, you voluntarily give your permission for the person named below to participate in this study. You are not waiving any legal rights for yourself or the person you are legally representing. After your signature, please print your name and your relationship to the subject. Signature of Parent/Legal Representative Date Print: Name of Legal Representative of and Relationship to Participant: ^

PAGE 179

169 Signature Date Participants Who Cannot Consent But Can Read and/or Understand about the Study Although legally you cannot "consent" to be in this study, we need to know if you want to take part. If you decide to take part in this study, and your parent or the person legally responsible for you gives permission, you both need to sign. Your signing below means that you agree to take part (assent). The signature of your parent/legal representative above means he or she gives permission (consent) for you to take part. Assent Signature of Participant Date

PAGE 180

REFERENCE LIST Aimonetti JM, Nielsen JB, Laboratoire Daeveloppement et Pathologie du Mouvement CNdlRScJAMCF. 2002. Cortical excitability and motor task in man: an investigation of the wrist extensor motor area. Experimental brain researcti. Experimentelle l-lirnforscliung. Experimentation cerebrale. 143(4): 431-9 Barker AT, Jalinous R, Freeston IL. 1985. Non-invasive magnetic stimulation of human motor cortex. Lancet 1: 1106-7 Basmajian J. 1964. Primary Anatomy. Baltimore: The Williams &Wilkins Company Bastings EP, Gage HD, Greenberg JP, Hammond G, Hernandez L, et al. 1998. Coregistration of cortical magnetic stimulation and functional magnetic resonance imaging. l\leuroreport9: 1941-6 Boroojerdi B, Battaglia F, Muellbacher W, Cohen LG. 2001a. Mechanisms influencing stimulus-response properties of the human corticospinal system. Clinical Neuropliysiology \12: 931-7 Boroojerdi B, Ziemann U, Chen R, Buterfisch CM, Cohen LG. 2001b. Mechanisms underlying human motor system plasticity. Muscle & Nerve 24: 602-13 Brasil-Neto JP, Cohen LG, Pascual-Leone A, Jabir FK, Wall RT, Hallett M. 1992a. Rapid reversible modulation of human motor outputs after transient deafferentation of the forearm: a study with transcranial magnetic stimulation. Neurology ^2: 1302-6 Brasil-Neto JP, McShane LM, Fuhr P, Hallett M, Cohen LG. 1992b. Topographic mapping of the human motor cortex with magnetic stimulation: factors affecting accuracy and reproducibility. Electroencephalogr Clin Neurophysiol QS\ 9-16 Brasil-Neto JP, Pascual-Leone A, Valls-Sole J, Cohen LG, Hallett M. 1992c. Focal transcranial magnetic stimulation and response bias in a forced-choice task. J Neurol Neurosurg Psychiatry SS\ 964-6 Byrnes ML, Thickbroom GW, Phillips BA, Wilson SA, Mastaglia FL, Australian Neuromuscular Research Institute DoMUoWAQEIIMCN. 1999. 170

PAGE 181

171 Physiological studies of the cortlcomotor projection to the hand after subcortical stroke. Clinical neurophysiology : official journal of the International Federation of Clinical Neurophysiology. 110(3): 487-98 Carr U, Harrison LM, Evans AL, Stephens JA. 1993. Patterns of central motor reorganization in hemiplegic cerebral palsy. Brain 116 ( Pt 5): 1223-47 Carroll TJ, Riek S, Carson RG. 2001. Reliability of the input-output properties of the cortico-spinal pathway obtained from transcranial magnetic and electrical stimulation. J Neurosci Methods 111: 193-202 Chen R. 2000. Studies of human motor physiology with transcranial magnetic stimulation. Muscle Nerve Suppl^: S26-32 Cicinelli P, Traversa R, Rossini PM. 1997. Post-stroke reorganization of brain motor output to the hand: a 2-4 month follow-up with focal magnetic transcranial stimulation. Electroencephalogr Clin Neurophysiol l^S: 438-50 Cohen LG, Roth BJ, Wassermann EM, Topka H, Fuhr P, et al. 1991. Magnetic stimulation of the human cerebral cortex, an indicator of reorganization in motor pathways in certain pathological conditions. Journal of Clinical Neurophysiology: 8{iy. 56-6S Cramer SC, Bastings EP. 2000. Mapping clinically relevant plasticity after stroke. Neuropharmacology 39 : 842-5 1 Cramer SC, Moore CI, Finklestein SP, Rosen BR. 2000. A pilot study of somatotopic mapping after cortical infarct. Stroke3\: 668-71 Cramer SC, Nelles G, Benson RR, Kaplan JD, Parker RA, et al. 1997. A functional MRI study of subjects recovered from hemiparetic stroke. Stroke 28: 2518-27 Cramer SC, Nelles G, Schaechter JD, Kaplan JD, Finklestein SR, Rosen BR. 2001a. A functional MRI study of three motor tasks in the evaluation of stroke recovery. Neurorehabilitation and Neural Repair 15: 1-8 Cramer SC, Stegbauer KC, Mark A, Price R, Barquist K, et al. 2001b. Motor cortex activation in hemiparetic stroke patients. Stroke 32: 100 Devanne H, Lavoie BA, Capaday C. 1997. Input-output properties and gain changes in the human corticospinal pathway. Exp Brain Res 114: 329-38 Di Lazzaro V, Oliviero A, Profice P, Saturno E, Pilato F, et al. 1998. Comparison of descending volleys evoked by transcranial magnetic and electric

PAGE 182

172 stimulation in conscious humans. Electroencephalogr Clin Neurophysiol 109: 397-401 Dobl\\ 14-27 Fischer M, Kaech S, Knutti D, Matus A. 1998. Rapid actin-based plasticity in dendritic spines. Neuronic): 847-54 Friel KM, Heddings AA, Nudo RJ. 2000. Effects of postlesion experience on behavioral recovery and neurophysiologic reorganization after cortical injury in primates. Neurorehabilitation and Neural Repair!^: 187-98 Friel KM, Nudo RJ. 1998. Recovery of motor function after focal cortical injury in primates: compensatory movement patterns used during rehabilitative tra i n i ng Somatosensory and Motor Research 15: 1 73 -89 Georgopoulos AP, Kalaska JF, Caminiti R, Massey JT. 1982. On the relations between the direction of two-dimensional arm movements and cell discharge in primate motor cortex. J Neuroscll: 1527-37 Haines DE. 2002. Fundamental Neuroscience. New York: Churchill Livingstone Hallett M. 2000. Transcranial magnetic stimulation and the human brain. Nature 406: 147-50 Hallett M. 2001. Functional reorganization after lesions of the human brain: studies with transcranial magnetic stimulation. Rev Neurol (Paris) 157: 822-6

PAGE 183

173 Halter JA, Carp JS, Wolpaw JR. 1995. Operantly Conditioned i^otoneuron Plasticity: Possible Role of Sodium-Channels. Journal of Neurophysiology 73: 867-71 Hebb DO. 1949. The Organization of Behavior: A Neurophysiological Theory.: John Wiley and Sons Held J. 2000. Recovery of function after brain damage:Theoretical implications for therapeutic intervention. In Movement Science: Foundations for Physical Therapy in Rehabilitation., ed. SR Carr J. Gaithesburg, MD: Aspen Hermens HJ, Freriks B, Disselhorst-Klug C, Rau G. 2000. Development of recommendations for SEMG sensors and sensor placement procedures. J Electromyogr Kinesiol 10: 361-74 Hess CW, Mills KR, Murray NM, Department of Clinical Neurophysiology NHQSL. 1987. Responses in small hand muscles from magnetic stimulation of the human brain. The Journal of physiology. 388: 397-419 Hess G, Donoghue JP. 1996. Long-term depression of horizontal connections in rat motor cortex. Eur J Neurosci Q\ 658-65 Hikosaka 0, Sakai K, Miyauchi S, Takino R, Sasaki Y, Putz B. 1996. Activation of human presupplementary motor area in learning of sequential procedures: A functional MRI study. Journal of Neurophysiology IW. 617-21 Jacobs KM, Donoghue JP. 1991. Reshaping the Cortical Motor Map by Unmasking Latent Intracortical Connections. ScienceTbV. 944-7 Jagodzinski P, Hess G. 2001. Tetraethylammonium-induced long-term potentiation in layer V horizontal connections of rat motor cortex. Neurosci Lettl'^Q: 37-40 Jalinous R. 1991. Technical and practical aspects of magnetic nerve stimulation. J Clin NeurophysiolQ: 10-25 Johansson BB. 2000. Brain plasticity and stroke rehabilitation. The Willis lecture. Stroke 31: 223-30 Kandel ER, Schwartz JH, Jessell TM. 2000. Principles of Neural Science. New York: McGraw-Hill Kami A, Meyer G, Jezzard P, Adams MM, Turner R, Ungerleider LG. 1995. Functional MRI evidence for adult motor cortex plasticity during motor skill learning. Nature J77: 155-8

PAGE 184

174 Kiers L, Cros D, Chiappa KH, Fang J. 1993. Variability of motor potentials evoked by transcranial magnetic stimulation. Electroencephalogr Clin Neurophysiol 89: 415-23 Kopp B, Kunkel A, Muhlnickel W, Villringer K, Taub E, Flor H. 1999. Plasticity in the motor system related to therapy-induced improvement of movement after stroke. Neuroreport \Q\ 807-10 Kunkel A, Kopp B, Muller G, Villringer K, Villringer A, et al. 1999. Constraintinduced movement therapy for motor recovery in chronic stroke patients. Arch Phys Med Rehab// 80 : 624-8 Lee D. 1995. Mechanisms underlying functional recovery following stroke. Canad/an Journa/ of Neuro/og/ca/ Sc/ence 22: 257-63 Leonard C. 1998. The neuroscience of motor learning. In The Neurosc/ence of Human (Movement. St. Louis: IMosby-Year Book, Inc. Levy CE, Nichols DS, Schmalbrock PM, Keller P, Chakeres DW. 2001. Functional MRI evidence of cortical reorganization in upperlimb stroke hemiplegia treated with constraint-induced movement therapy. American Journa/ of Phys/ca/ l^ed/c/ne & Rehabi/itat/on 80: 4-12 Liepert J, Bauder H, Miltner WHR, Taub E, Weiller C. 2000. Treatment-induced cortical reorganization after stroke in humans. Stro/
PAGE 185

175 Maren S, Baudry M. 1995. Properties and mechanisms of long-term synaptic plasticity in the mammalian brain: relationships to learning and memory. Neurobiol Learn Mem 63: 1-18 Mathiowetz V, Volland G, Kashman N, Weber K. 1985. Adult norms for the Box and Block Test of manual dexterity. Am J Occup Ther39: 386-91 McMillan AS, Watson C, Walshaw D, Taylor JP. 1998. Improved reproducibility of magnetic stimulation-evoked motor potentials in the human masseter by a new method for locating stimulation sites on the scalp. Arch Oral Biol ^2>\ 665-8 Mills KR, Boniface SJ, Schubert M. 1992. Magnetic brain stimulation with a double coil: the importance of coil orientation. Electroencephalogr Clin Neurophysiol QS\ 17-21 Miltner WH, Bauder H, Sommer M, Dettmers C, Taub E. 1999. Effects of constraint-induced movement therapy on patients with chronic motor deficits after stroke: a replication. Strol
PAGE 186

176 Nudo RJ. 1996. Neural substrates for the effects of rehabilitative training on motor recovery after ischemic infarct (vol 272, pg 1791, 1996). Science 273: 564Nudo RJ, Milliken GW. 1996. Reorganization of movement representations in primary motor cortex following focal ischemic infarcts in adult squirrel monkeys. Journal of Neurophysiology 7S\ 2144-9 Nudo RJ, Plautz EJ, Frost SB. 2001. Role of adaptive plasticity in recovery of function after damage to motor cortex. Muscle & Nerve 24: 1000-19 Nudo RJ, Plautz EJ, Milliken GW. 1997. Adaptive plasticity in primate motor cortex as a consequence of behavioral experience and neuronal injury. Seminars in Neuroscience9\ 13-23 Pascual-Leone A, Cammarota A, Wassermann EM, Brasilneto JP, Cohen LG, Hallett M. 1993. Modulation of Motor Cortical Outputs to the Reading Hand of Braille Readers. Annals of Neurology 3^: 33-7 Pascual-Leone A, Cohen LG, Brasil-Neto JP, Hallett M. 1994. Non-invasive differentiation of motor cortical representation of hand muscles by mapping of optimal current directions. Electroencephalogr Clin Neurophysiol93\ 42-8 Pascualleone A, Dang N, Cohen LG, Brasilneto JP, Cammarota A, Hallett M. 1995. Modulation of Muscle Responses Evoked by Transcranial Magnetic Stimulation During the Acquisition of New Fine Motor-Skills. Journal of Neurophysiology!^: 1037-45 Pascual-Leone A, Dang N, Cohen LG, Brasilneto JP, Cammarota A, Hallett M. 1995a. Modulation of Muscle Responses Evoked by Transcranial Magnetic Stimulation During the Acquisition of New Fine Motor-Skills. Journal of Neurophysiology 7'^: 1037-45 Pascual-Leone A, Tarazona F, Keenan J, Tormos JM, Hamilton R, Catala MD. 1999. Transcranial magnetic stimulation and neuroplasticity. Neuropsychologia 3,7: 207-17 Pascual-Leone A, Wassermann EM, Sadato N, Hallett M, Human Motor Control Section NIoND, Stroke NIoHBMDUSA. 1995b. The role of reading activity on the modulation of motor cortical outputs to the reading hand in Braille readers. Annals of neurology. 38(6): 910-5 Penfield W, Rasmussen T. 1950. The Cerebral Cortex of Man. A Clinical Study of Localization of Function. New York: Macmillan

PAGE 187

177 Pennisi G, Alagona G, Rapisarda G, Nicoletti F, Costanzo E, et al. 2002. Transcranial magnetic stimulation after pure motor stroke. Clin Neurophysiol 113:1 536-43 Phillips CG, Porter R. 1977. Corticospinal neurones. Their role in movement. Monogr Physiol Soc. v-xii, 1-450 Portney LG, Watkins MP. 2000. Foundations of Clinical Research: Application to Practice. Norwalk, Connecticut: Appleton & Lange Ramachandran VS. 1993. Behavioral and Magnetoencephalographic Correlates of Plasticity in the Adult Human Brain. Proceedings of the National Academy of Sciences of the United States of America 90: 10413-20 Ray J, IMcNamara B, Boniface S. 2002. Acquisition and expression of proximal and distal upper limb stimulus-response curves to transcranial magnetic stimulation. Muscle Nerve 25: 202-6 Ridding MC, Rothwell JC. 1997. Stimulus/response curves as a method of measuring motor cortical excitability in man. Electroencephalogr Clin Neurophysiol 1^5'. 340-4 Rioult-Pedotti MS, Friedman D, Hess G, Donoghue JP. 1998. Strengthening of horizontal cortical connections following skill learning. Nature Neuroscience 1 : 230-4 Roricht S, Meyer BU, Niehaus L, Brandt SA. 1999. Long-term reorganization of motor cortex outputs after arm amputation. Neurology SI: 106-11 Rossini PM, Caltagirone C, Castriota-Scanderbeg A, Cicinelli P, Del Gratta C, et al. 1998. Hand motor cortical area reorganization in stroke: a study with fMRI, MEG and TCS maps. Neuroreport^: 2141-6 Rossini PM, Pauri F. 2000. Neuromagnetic integrated methods tracking human brain mechanisms of sensorimotor areas 'plastic' reorganisation. Brain Research Reviews 33 : 131-54 Rothwell J. 1994. Control of Human Voluntary Movement, 2nd edition. London: Chapman & Hall Publishers Sanes JN, Suner S, Lando JF, Donoghue JP. 1988. Rapid Reorganization of AdultRat Motor Cortex Somatic Representation Patterns after Motor-Nerve Injury. Proceedings of the National Academy of Sciences of the United States of America QS: 2003-7

PAGE 188

178 Schmidt RA, LeeTD. 1999. Motor Control and Learning: A Behavioral Emphasis. Champaign, IL: Human Kinetics Taub E, Crago J, Uswatte G. 1998. Constraint-induced movement therapy: a new approach to treatment in physical rehabilitation. Rehabilitation Psychology 43: 152-70 Taub E, Miller NE, NovackTA, Cook EW, 3rd, Fleming WC, et al. 1993. Technique to improve chronic motor deficit after stroke. Arch Rhys Med Rehabill^'. 347-54 Taub E, Uswatte G, Pidikiti R. 1999. Constraint-Induced Movement Therapy: a new family of techniques with broad application to physical rehabilitationa clinical review. J Rehabil Res Dev 36: 237-51 Thickbroom GW, Sammut R, Mastaglia FL. 1998. Magnetic stimulation mapping of motor cortex: factors contributing to map area. Electroencephalogr Clin Neurophysiol 109: 79-84 Thompson PD, Day BL, Rothwell JC, Dressier D, Maertens de Noordhout A, et al. 1991. Further observations on the facilitation of muscle responses to cortical stimulation by voluntary contraction. Electroencephalography and clinical neurophysiology. 81(5): 397-402 Traversa R, Cicinelli P, Bassi A, Rossini PM, Bernardi G. 1997. Mapping of motor cortical reorganization after stroke A brain stimulation study with focal magnetic pulses. Stroke28: 110-7 Traversa R, Cicinelli P, Pasqualetti P, Filippi M, Rossini PM. 1998. Follow-up of interhemispheric differences of motor evoked potentials from the 'affected' and 'unaffected' hemispheres in human stroke. Brain Research 803: 1-8 Triggs WJ, Subramanium B, Rossi F. 1999. Hand preference and transcranial magnetic stimulation asymmetry of cortical motor representation. Brain Res 835: 324-9 Trompetto C, Assini A, Buccolieri A, Marchese R, Abbruzzese G. 2000. Motor recovery following stroke: a transcranial magnetic stimulation study. Clin Neurophysiol 111:1 860-7 Turker KS. 1993. Electromyography: some methodological problems and issues. Rhys Ther 73: 698-710 Turrigiano GG. 1999. Homeostatic plasticity in neuronal networks: the more things change, the more they stay the same. Trends Neurosci 22: 221-7

PAGE 189

179 Turton A, Wroe S, Trepte N, Fraser C, Lemon RN. 1996. Contralateral and ipsilateral EMG responses to transcranial magnetic stimulation during recovery of arm and hand function after stroke. Electroencephalogr Clin Neurophysiol 101 : 316-28 Uy J, Ridding MC, Miles TS. 2002. Stability of maps of human motor cortex made with transcranial magnetic stimulation. Brain TopogrlA: 293-7 Wassermann EM, McShane LM, Hallett M, Cohen LG. 1992. Noninvasive mapping of muscle representations in human motor cortex. Electroencephalogr Clin Neurophysiol ^S: 1-8 Weber M, Eisen AA. 2002. Magnetic stimulation of the central and peripheral nervous systems. Muscle Nerve 2S: 160-75 Weiller C. 1998. Imaging recovery from stroke. Exp Brain Res 123: 13-7 Weiller C, Friston KJ, Wise RJS, Frackowiak RSJ. 1992. Functional reorganization of the brain in recovery from striatocapsular infarction in man. Annals of Neurology 31'. 463-72 Wilson SA, Thickbroom GW, Mastaglia FL. 1993. Topography of excitatory and inhibitory muscle responses evoked by transcranial magnetic stimulation in the human motor cortex. Neurosci Lettish: 52-6 Zemke AC, Cramer SC. 2002. A functional MRI study of motor recovery patterns after stroke. Stroke 23: 51 Ziemann U, Lonnecker S, Steinhoff BJ, Paulus W. 1996. Effects of antiepileptic drugs on motor cortex excitability in humans: a transcranial magnetic stimulation study. Ann Neurol ^Q: 367-78

PAGE 190

BIOGRAPHICAL SKETCH Matthew Paul Malcolm was raised in upstate New York and graduated from tine Canandaigua Academy High School in 1991. He then went on to study at the State University of New York at Buffalo, where he learned much about life, friendship and personal freedom. Matthew graduated from UB with a Bachelors of Science in occupational therapy in 1996. Following graduation, he moved to Fayetteville, Arkansas and began working at the Health South Rehabilitation Hospital as a member of the brain injury team. Matthew's interest in neuroscience and stroke rehabilitation brought him back to academia, and in 1999 he entered the Rehabilitation Science Doctoral program at the University of Florida. During his doctoral studies, Matthew worked as the site project coordinator for the EXCUe stroke clinical trial. As a pre-doctoral fellow at the VA Brain Rehabilitation Research Center in Gainesville, Florida, Matthew investigated neurological and behavioral factors that underlie recovery from stroke. Matthew presently lives with his wife April and two dogs in Gainesville, Florida. In his time away from academia, he enjoys outdoor adventure, percussion and spending time with friends and family. In October 2003 Matthew will join the Occupational Therapy faculty at Colorado State University in Fort Collins, Colorado. 180

PAGE 191

I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Kathye E Light, Char Associate Professor of Physical Therapy I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Orit Shechtman, Cochair Associate Professor of Occupational Therapy I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. 7 ^//{l ^^^^^-^^^ William J. "mggs Associate Professor of Neurology I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Leslie Gon^lez^othj; Professor of Communication Sciences and Disorders This dissertation was submitted to the Graduate Faculty of the College of Health Professions and to the Graduate School and was accepted as partial fulfillment of the requirements for the degree of Doctor of Philosophy. Month & Year of Graduation ""^ ^y^^^l^clS /-.^^WC^ Dean, College of Health Professions December 2003 Dean, Graduate School