Does astroglial protein S100B contribute to West Nile neuro-invasive syndrome?

The clinical spectrum of West Nile Virus (WNV) infection ranges from a flu-like febrile condition to a more severe neuro-invasive disease that can cause death. The exact mechanism of neurodegeneration in neuro-invasive form of WNV infection has not been elucidated; however, a destructive role played by glial cells in promoting WNV mediated neurotoxicity has widely been speculated. The clinical studies revealed that the astroglial protein S100B is significantly elevated in the blood and CSF of patients with WNV infection, even in the absence of neuro-invasive disease. Therefore, the present study was designed to explore the potential role of S100B in the pathophysiology of WNV infection. The overarching hypothesis was that WNV primes astroglia to release S100B protein, which leads to a cascade of events that may have deleterious effects in both acute and chronic stages of WNV disease. To justify our hypothesis, we first ascertained increased levels of S100B in post-mortem tissue samples from WNV patients. Next, we looked at the effects of UV-inactivated WNV particles on astroglia using astroglial cell lines or primary cultures. Astroglial activation was measured as an increase in the expression of S100B and was analyzed by immunofluorescence and real-time PCR. Further, the in vitro effects of purified S100B protein on neutrophil migration and glutamate uptake were also determined in astroglial cell lines or primary cultures. We found that incubation of cultured astroglial cells with UV-inactivated WNV particles caused induction of S100B both at the mRNA and protein levels. Varying concentrations of S100B stimulated neutrophil migration in vitro. In addition, varying amounts of S100B caused inhibition of glutamate uptake in astroglia in a dose-dependent manner. Our data suggest that inactivated WNV particles are capable of inducing S100B synthesis in astroglia in vitro. We speculate that S100B release by activated astroglia may have multiple roles in the pathophysiology of WNV neuro-invasive disease, including induction of neutrophil migration to the sites where blood brain barrier is disrupted as well as glutamate neurotoxicity. To further elucidate the WNV-S100B neurotoxic pathway, in vivo studies using mouse models are warranted.

Variability, frequency composition, and complexity of submaximal isometric knee extension force from subacute to chronic stroke.

We examined changes in the variability, frequency composition, and complexity of force signal from subacute to chronic stage of stroke during maintenance of isometric knee extension and compared these parameters between chronic stroke and healthy subjects. The sample included 15 healthy (65±8 years) and 23 chronic stroke subjects (65±14 years, 6-112 months post-stroke) of whom 10 (64±15 years) were also examined 11-22 days post-stroke (subacute stage). The subjects performed isometric knee extension at 10%, 20%, 30%, and 50% of peak torque for 10s (two trials each). Coefficient of variation (CV) was used as a measure of force variability. The median frequency and relative power in the 0-3, 4-6, and 8-12 Hz bands were obtained through a power spectrum analysis of the force signal. The signal complexity was quantified using the sample entropy (SampEn). The longitudinal analysis revealed a significant decrease in CV from subacute to chronic stage across all contraction levels (P<0.001) but no significant changes in the frequency and entropy parameters. Comparison between the chronic stroke and control subjects revealed no significant difference in CV across the force levels (P>0.05) but significantly decreased median frequency (P<0.01), with the relative power increased in 0-3 Hz band and decreased in 4-6 and 8-12 Hz bands in both paretic and non-paretic legs (P<0.001). SampEn was also significantly decreased in chronic stroke, bilaterally (P<0.001). These results indicate a shift toward lower frequencies and a less complex physiological process underlying force control in chronic stroke. The overall results suggest the improvement in force variability from subacute to chronic stroke but without normalization in the frequency composition and complexity of the force signal. Thus, disordered structure of the force signal remains a marker of impaired motor control long after stroke occurrence despite apparent recovery in force variability.

Impaired force steadiness is associated with changes in force frequency composition in subacute stroke.

We tested the hypothesis that impaired force steadiness early after stroke is associated with changes in frequency composition of the force signal during constant-force task. The power spectra and the relationship between power spectra and force variability during isometric knee extension (10%, 20%, 30%, and 50% of peak torque for 10s) were studied in the paretic and non-paretic legs of 34 stroke patients (64±14years, 8-25days post-injury) and the dominant leg of 20 controls (62±10years). Power spectrum analysis of the force signal included the median frequency, peak power frequency, relative peak power, and relative power in 0-3, 4-6, and 8-12Hz bands. Force variability, quantified by coefficient of variation (CV), was increased in patients at 3 of the 4 contraction levels (P⩽0.001). Median frequency across all force levels was decreased and the relative peak power was increased in the paretic and non-paretic legs compared to controls (P⩽0.001). The relative power was increased in 0-3Hz band and decreased in both 4-6 and 8-12Hz bands in the paretic leg only (P⩽0.001). Progressively stronger contractions brought about a significant decrease in relative power in the 0-3Hz band and increase in 8-12Hz band in controls but not in stroke subjects. The hypothesis was confirmed by significant non-linear correlations between CV and each relative spectral power found in the paretic leg at most contraction levels (0.22⩽R(2)⩽0.72, P⩽0.0004) and in the non-paretic leg at 10% only (0.35⩽R(2)⩽0.52, P⩽0.0002), but not in controls. Fugl-Meyer lower extremity motor and sensory scores were not related to the frequency measures in stroke subjects (P>0.05). Limited modulation of frequency spectra and the emergence of non-linear relation between power spectra and force variability suggest that less broadband force output may account in part for impaired force steadiness in paretic and non-paretic legs early after stroke.

Predictors of oral health after spinal cord injury.

STUDY DESIGN: Cross-sectional study. OBJECTIVES: To examine predictors of oral health in people with spinal cord injury (SCI). METHODS: Ninety-two people with SCI (> or =6 months, 44% cervical level) completed questionnaires and underwent oral examination. Socio-economic, injury-related and oral habits variables were used for predicting oral health score (OHS); Decayed, missing and filled teeth (DMFT) score; and periodontal screen and recording index (PSR). RESULTS: Most people with SCI were able to bring at least one hand to the mouth (82%) and brush teeth independently (65%). Regarding daily oral habits, 84% reported brushing teeth, 48% rinsing mouth, 14% flossing, 33% tobacco use and 13% mouthstick use. Only 32% had teeth cleaned within the past year. Oral examination revealed three decayed and eight missing teeth on average, with prominent periodontal disease (64%). Employment before SCI and more risky oral habits were significant predictors of worse OHS (P=0.005 and P=0.014, respectively) and PSR score (P=0.010 and P=0.035, respectively). Older age was the only predictor of worse DMFT score (P<0.001). CONCLUSION: Oral health appears compromised in people with SCI. Identification of modifiable risk factors warrants examination whether intervention with focus on behavioral changes may improve oral health in this population.

Neuronal and glial cerebrospinal fluid protein biomarkers are elevated after West Nile virus infection.

Neurotrophic West Nile virus (WNV) disease is a severe arbovirus infection in which neuronal loss is the likely anatomical substrate for the high morbidity and mortality. We investigated whether cerebrospinal fluid (CSF) protein biomarkers were elevated in vivo and related to disease severity in patients with WNV infection. This exploratory study included 114 patients (24 acute WNV, 77 noninflammatory controls, six peripheral neuropathies, seven aseptic meningoencephalitis). CSF levels of neuronal (neurofilaments, NfH-SMI35) and glial (glial fibrillary acidic protein, GFAP, S100B) biomarkers were measured by enzyme-linked immunosorbent assay (ELISA). Immunocytochemistry was performed in two fatal WNV cases. A significant proportion of patients with WNV had pathological CSF levels for NfH-SMI35 (58%, median concentration 1.01 ng/mL), GFAP (58%, 10 pg/mL), and S100B (90%, 1.29 ng/mL). The results were consistent with postmortem evidence for neuronal death and astrogliosis. Surprisingly, CSF protein biomarker levels were also found to be pathological in a considerable proportion of patients who presented with WNV fever only (100% for GFAP and S100B and 43% for NfH-SMI35). Elevated CSF protein biomarker levels are suggestive of neuronal death and glial pathology in human WNV infection. The results indicate the presence of neuroinvasive disease across the spectrum of WNV disease, including WNV fever.

Neurophysiological basis and clinical applications of the H-reflex as an adjunct for evaluating response to intrathecal baclofen for spasticity.

Implanted programmable pumps that infuse intrathecal baclofen (ITB) markedly enhance the ability of clinicians to manage severe spasticity in appropriately selected patients. Studies addressing the efficacy of this treatment modality have primarily used clinical outcome measures of impairment, particularly reduction in stiffness as measured by the Ashworth scale. Several recent studies, however, highlight comparalively higher sensitivity of neurophysiologic techniques, especially the H-reflex, as an objective index of spinal cord response to ITB administration. We review the conceptual, physiological, and methodological hases for use of the H-reflex as an adjunct to clinical evaluation among patients receiving ITB infusion, including published reports and selected case studies that address the potential advantages and limitations of such techniques when applied to dose titration and system "troubleshooting" scenarios, We also address the implications of such findings in the context of reported complications such as "tolerance" to ITB administration and catheter "microfracture". The accumulated knowledge suggests that H-reflex is a sensitive method for documenting altered spinal cord responsiveness in the presence of ITB delivery. We therefore recommend using H-reflex as an adjunct to clinical evaluation when judging the overall effectiveness of ITB administration.

Clinical neurophysiological assessment of residual motor control in post-spinal cord injury paralysis.

OBJECTIVE: This study was designed to characterize the rudimentary residual lower-limb motor control that can exist in clinically paralyzed spinal-cord-injured individuals. METHODS: Sixty-seven paralyzed spinal-cord-injured subjects were studied using surface electromyography recorded from muscles of the lower limbs and analyzed for responses to a rigidly administered protocol of reinforcement maneuvers, voluntary movement attempts, vibration, or the ability to volitionally suppress withdrawal evoked by plantar surface stimulation. RESULTS: Markers for the subclinical discomplete motor syndrome were found in 64% of the subjects. The tonic vibration response was recorded in 37%, volitional plantar surface stimulation response suppression in 27%, and reinforcement maneuver responses in 6% of the subjects. Three subjects, 4%, produced reliable but very low amplitude surface electromyography during the voluntary movement segment of the protocol. Surface electromyography recorded during passive leg movement was related to Ashworth scores as was the tonic vibration response marker (P < 0.05). CONCLUSIONS: Multimuscle surface electromyography patterns recorded during a rigidly administered protocol of motor tasks can be used to differentiate between clinically paralyzed spinal-cord-injured individuals using subclinical motor output to identify the translesional neural connections that remain available for intervention testing and treatment planning after spinal cord injury.

Nociceptive fingertip stimulation inhibits synergistic motoneuron pools in the human upper limb.

BACKGROUND: Activation of distinct muscle groups organized in a stereotyped manner ("muscle synergies") is thought to underlie the production of movement by the vertebrate spinal cord. This results in movement with minimum effort and maximum efficiency. The question of how the vertebrate nervous system inhibits ongoing muscle activity is central to the study of the neural control of movement. OBJECTIVE: To investigate the strategy used by the human spinal cord to rapidly inhibit muscle activation in the upper limb. METHODS: The authors performed a series of experiments in 10 healthy subjects to assess the effect of nociceptive cutaneous stimulation on voluntarily contracting upper limb muscles. They recorded the electromyogram (EMG) with surface electrodes placed over various upper limb muscles. RESULTS: The authors found evidence of a simple inhibitory strategy that 1) was dependent on the intensity of the stimulus, 2) was maximally evoked when stimulation was applied to the fingertips, 3) preceded the earliest onset of voluntary muscle relaxation, and 4) produced inhibition of EMG activity in specific upper limb muscle groups. Nociceptive fingertip stimulation preferentially inhibited contraction of synergistic muscles involved in reaching and grasping (intrinsic hand muscles, forearm flexors, triceps) while having little effect on biceps or deltoid. CONCLUSIONS: Neural circuitry within the human spinal cord is organized to inhibit movement by rapidly deactivating muscles that constitute distinct muscle synergies. This strategy of selective and concurrent deactivation of the same basic elements that produce synergistic movement greatly simplifies motor control.

Sources of movement-related cortical potentials derived from foot, finger, and mouth movements.

Movement-related cortical potentials (MRCPs) register brain electrical activity before and during movement execution. In an attempt to delineate the components of MRCPs that reflect common sources to various movements and that are movement-specific, simple self-paced voluntary foot, finger, and mouth movements were studied. MRCPs were recorded in eight healthy volunteers with 30 electrodes placed on the scalp. Data were analyzed using Brain Electric Source Analysis software, and multiple equivalent dipole models were developed to separate spatial and temporal aspects of brain activity related to the execution of voluntary movements. Independent models were separately developed for the grand average data and for the individual subjects' data for each movement type. MRCPs derived from foot movements were accounted for using a 5-dipole model, finger movements using an 8-dipole model, and mouth movements with a 7-dipole model, yielding the grand average residual variances of 3%, 2%, and 6%, respectively. Based on individual models, intersubject variability of dipole locations was less than 10 mm (+/- SD). Overlaying the mean dipole coordinates onto the stereotaxic atlas provided proof that the sensorimotor cortical areas, supplementary motor area, and also cerebellum and thalamus were active in all three movements. Locations of the dipoles in the contralateral sensorimotor area clearly implied well-known medial to lateral somatotopic organization of foot, finger, and mouth movements. Temporal separation of the activity spread over different brain areas was demonstrated by evolution in the moments of dipole source potentials. The authors' models support the view of simultaneous activation of the primary motor cortex and supplementary motor area at the time of movement execution. Multiple equivalent dipole models developed in this study implied the activity originating in corresponding brain areas as previously detected by positron emission tomography or functional magnetic resonance imaging. However, MRCPs provided additional information regarding the temporal evolution of the brain activity related to the execution of voluntary movements. Thus, the concurrent use of MRCPs and other imaging techniques may provide complementary information not easily obtained by the other imaging techniques themselves.

Relating clinical and neurophysiological assessment of spasticity by machine learning.

Spasticity following spinal cord injury (SCI) is most often assessed clinically using a five-point Ashworth score (AS). A more objective assessment of altered motor control may be achieved by using a comprehensive protocol based on a surface electromyographic (sEMG) activity recorded from thigh and leg muscles. However, the relationship between the clinical and neurophysiological assessments is still unknown. In this paper we employ three different classification methods to investigate this relationship. The experimental results indicate that, if the appropriate set of sEMG features is used, the neurophysiological assessment is related to clinical findings and can be used to predict the AS. A comprehensive sEMG assessment may be proven useful as an objective method of evaluating the effectiveness of various interventions and for follow-up of SCI patients.

Source localization of P300 from oddball, single stimulus, and omitted-stimulus paradigms.

Three different auditory stimulus paradigms were used to elicit P300 potentials. Normal subjects were tested on the classical rare target stimulus, single-stimulus and omitted-stimulus conditions. Noninvasive identification of the cerebral sources of the event-related potentials (ERPs) was performed using spatio-temporal multiple dipole modeling (BESA software) with individually sized spherical head models. The grand average data of each condition was first independently modeled and these models were used as starting values for modeling each individual subject's data. Models for the rare-stimulus condition and single-stimulus condition both consisted of 6 dipoles. Models for the omitted-stimulus condition consisted of 2 dipoles. The dipole locations of the final individual 6-dipole models for the rare and single-stimulus conditions did not differ significantly from each other or from one previous result obtained from a another group of subjects (Tarkka et al. 1995). Super-imposition of the dipole coordinates on the sterotaxic brain atlas suggests that bilateral deep medial temporal lobe structures are the major contributors to rare and single-stimulus P300s. Because both the wave form morphology and the source model of the P300 elicited by single stimulus were close to those of the rare-stimulus P300 it may be that the underlying neural mechanisms eliciting these P300 potentials are essentially the same.

Assessment of corticospinal function in spinal cord injury using transcranial motor cortex stimulation: a review.

Other than clinical examination, few methods exist for assessing the functional condition of descending long tracts of the spinal cord in humans. This review covers neurophysiological examination of the corticospinal system using transcranial electrical and magnetic motor cortex stimulation. The neurophysiological basis for the motor evoked potentials (MEPs) and the differences between the two methods are discussed followed by a review of their use in individuals with spinal cord injury (SCI). Transcranial motor cortex stimulation is used to monitor descending spinal cord tract condition during spinal surgeries and could be useful for assessing central nervous system trauma, especially in the unconscious multitrauma patient. In the chronic phase of SCI, recordings of MEPs have enabled the estimation of central conduction times that relate to the condition of axons passing through the injured segment of the spinal cord. They were found to correlate well with clinical examination scores but as predictors of outcome, the reports have been mixed. The use of transcranial motor cortex stimulation to modify segmental reflexes and in combination with volitional attempts have also provided evidence of conduction across the lesion in paralyzed SCI subjects. However, MEPs can be absent in some SCI individuals who may be able to volitionally activate muscles below the level of the spinal cord lesion. Such findings are useful in elucidating the neural mechanisms underlying the performance of a volitional movement and may serve to guide and monitor the effects of future treatments for paralysis in SCI and other neurological disorders.

Intracortical inhibition of lower limb motor-evoked potentials after paired transcranial magnetic stimulation.

The aim of the present study was to determine the characteristics of intracortical inhibition in the motor cortex areas representing lower limb muscles using paired transcranial magnetic (TMS) and transcranial electrical stimulation (TES) in healthy subjects. In the first paradigm (n=8), paired magnetic stimuli were delivered through a double cone coil and motor evoked potentials (MEPs) were recorded from quadriceps (Q) and tibialis anterior (TA) muscles during relaxation. The conditioning stimulus strength was 5% of the maximum stimulator output below the threshold MEP evoked during weak voluntary contraction of TA (33+/-5%). The test stimulus (67+/-2%) was 10% of the stimulator output above the MEP threshold in the relaxed TA. Interstimulus intervals (ISIs) from 1-15 ms were examined. Conditioned TA MEPs were significantly suppressed (P<0.01) at ISIs of less than 5 ms (relative amplitude from 20-50% of the control). TA MEPs tended to be only slightly facilitated at 9-ms and 10-ms ISIs. The degree of MEP suppression was not different between right and left TA muscles despite the significant difference in size of the control responses (P<0.001). Also, conditioned MEPs were not significantly different between Q and TA. The time course of TA MEP suppression, using electrical test stimuli, was similar to that found using TMS. In the second paradigm (n=2), the suppression of TA MEPs at 2, 3, and 4 ms ISIs was examined at three conditioning intensities with the test stimulation kept constant. For the pooled 2- to 4-ms ISI data, relative amplitudes were 34+/-6%, 61+/-5%, and 98+/-9% for conditioning intensities of 0.95, 0.90, and 0.85x active threshold, respectively (P<0.01). In conclusion, the suppression of lower limb MEPs following paired TMS showed similar characteristics to the intracortical inhibition previously described for the hand motor area.

Generators for human P300 elicited by somatosensory stimuli using multiple dipole source analysis.

Cognitive event-related potentials, such as P300, are sensitive to manipulations of psychological variables and may provide evidence to support theories of brain mechanisms involved in cognition. However, the relationship between event-related potentials and the active neural structures is not yet understood. Electrical stimulation of the index and little fingers of the left hand in the context of a somatosensory target discrimination task, performed by healthy human subjects, elicited the middle-latency component of somatosensory evoked potentials, N60, the long-latency component, N140, and the P300 component. Identification of the generators for both the earlier components and P300, using equivalent electrical dipole modeling, was performed. Individual spatiotemporal seven-dipole models were developed in order to suggest locations of the sources generating each subject's scalp-recorded wave forms. Three dipoles with fairly weak moments, located in the primary and secondary sensory areas, explained the middle- and long-latency somatosensory evoked potential components, and the remaining four dipoles (4-7), with stronger dipole moments, were active during P300. There was a clear temporal separation of dipole activity between the somatosensory evoked potential components and the P300 component. Dipoles 4 and 5 were found quite symmetrically in the parahippocampal areas of the two hemispheres, while dipoles 6 and 7 were slightly asymmetrical. Dipole 7 was found in the left hippocampal area. Dipole 6 appeared in the right insular cortex. The locations of the four dipoles implicated in the generation of the somatosensory P300 were compared with the locations of four dipoles accounting for the auditory evoked P300 described in our previous paper [Tarkka et al. (1995) Electroenceph. clin. Neurophysiol. 96, 538-545]. No substantial difference in source locations of the P300 was found between auditory and somatosensory modality other than an asymmetrical activity in the somatosensory modality contralateral to the stimulated hand.

Effect of fatiguing maximal voluntary contraction on excitatory and inhibitory responses elicited by transcranial magnetic motor cortex stimulation.

Vertex transcranial magnetic stimulation (TMS) elicited tibialis anterior motor evoked potentials (MEPs) and silent periods (SPs) that were recorded during and following isometric maximal volitional contraction (MVC). During MVC in 6 healthy subjects, MEP amplitudes in the exercised muscle showed an increasing trend from an initial value of 4539 +/- 809 muV (mean +/- SE) to 550 +/- 908 muV (P < 0.13) while force and EMG decreased (P < 0.01). Also, SP duration increased from 165 +/- 37 ms to 231 +/- 32 ms (P < 0.01). Thus, during a fatiguing MVC both excitatory and inhibitory TMS-induced responses increased. TMS delivered during repeated brief 10% MVC contractions before and after a fatiguing MVC in 5 subjects, showed no change in MEP amplitude but SP duration was prolonged after MVC. This SP prolongation was focal to the exercised muscle. Silent periods recorded after pyramidal tract stimulation were unchanged following the MVC. These results suggest that MEP and SP might have common sources of facilitation during an MVC and that inhibitory mechanisms remain focally augmented following a fatiguing MVC.

Spinal motoneuron excitability after acute spinal cord injury in humans.

BACKGROUND: Few studies in humans have assessed the ability of Ia afferent and antidromic motor volleys to activate motoneurons during spinal shock. Hence, little is known about the excitability state of the spinal motoneuron pool after acute spinal cord injury (SCI) in humans. METHODS: In 14 patients with acute SCI involving anatomic levels T10 and above, we performed clinical and electrophysiologic studies early after injury (within 24 hours in seven subjects) and on day 10, 20, and 30 postinjury. Maximal H:M ratios, F-wave persistence, and tendon tap T-reflexes were recorded. Sixteen normal subjects and eight chronic SCI patients served as control subjects. RESULTS: Ten of 14 patients had spinal shock (complete paralysis, loss of sensation, absent reflexes, and muscle hypotonia below the injury) at the time of initial evaluation. F-waves were absent in patients with spinal shock, reduced in persistence in patients with acute SCI without spinal shock, and normal in persistence in patients with chronic SCI. H-reflexes were absent or markedly suppressed in patients with spinal shock within 24 hours of injury but recovered to normal amplitudes within several days postinjury. This recovery occurred despite absence of F-waves that persisted for several weeks postinjury. Deep tendon reflexes were proportionally more depressed in spinal shock than were H-reflexes. All patients had elicitable H-reflexes for days or weeks before the development of clinical reflexes. CONCLUSIONS: Rostral cord injury causes postsynaptic changes (hyperpolarization) in caudal motoneurons. This hyperpolarization is a major physiologic derangement in spinal shock. The rise in H-reflex amplitude despite evidence of persistent hyperpolarization is due to enhanced transmission at Ia fiber-motoneuron connections below the SCI. Finally, the observation that the stretch reflex is proportionally more depressed than the H-reflex is consistent with fusimotor drive also being depressed after SCI.

Modification of motor control of wrist extension by mesh-glove electrical afferent stimulation in stroke patients.

OBJECTIVE: To study the effect of mesh-glove afferent stimulation on motor control of voluntary wrist movement in stroke patients who have chronic neurological deficits. DESIGN: Case series. Motor control was evaluated by surface EMG of the arm muscles and kinematics of voluntary wrist movements on 3 occasions: before and immediately after the initial session of mesh-glove stimulation, and then after a daily mesh-glove stimulation program conducted over several months. SETTING: Tertiary care center. PATIENTS: The inclusion criteria were: a history of stroke lasting longer than 6 months; completion of a rehabilitation program during early recovery; and preserved cognitive and communicative ability. Fourteen referred patients (age 63 +/- 9yr; time since stroke 31 +/- 22mo) fulfilled the criteria and completed the daily stimulation program. INTERVENTION: A single initial and then daily mesh-glove electrical afferent stimulation was applied to the hand of the involved upper limb for 20 to 30min. MAIN OUTCOME MEASURES: Surface EMGs from the affected biceps brachii and wrist extensor muscles and amplitudes of wrist movements were analyzed. RESULTS: The single, initial mesh-glove application had no effect on outcome measures. Following a daily mesh-glove stimulation program, however, both the amplitude of wrist extension movement and wrist extensor integrated EMG were significantly increased while coactivation of biceps brachii decreased. These findings were most prominent in subjects with partially preserved voluntary wrist movements. CONCLUSION: We conclude that daily mesh-glove stimulation can modify altered motor control and improve voluntary wrist extension movement in stroke subjects with chronic neurological deficits.