| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
(Stroke. 1997;28:2518-2527.)
© 1997 American Heart Association, Inc.
Articles |
From the Department of Neurology, Massachusetts General Hospital, Harvard Medical School, Boston, Mass (S.C.C., G.N., J.D.K., D.N.K., S.P.F.); The Neurorecovery Program, Massachusetts General Hospital and Spaulding Rehabilitation Hospital, Harvard Medical School, Boston, Mass (S.C.C., G.N., J.D.K., S.A.F.); The Clinical Investigator Training Program, Harvard-MIT Division of Health Sciences and Technology and Beth Israel-Deaconess Medical Center in collaboration with Pfizer Inc. (S.C.C.); The MGH-NMR Center, Department of Radiology, Massachusetts General Hospital, Harvard Medical School, Charlestown, Mass (S.C.C., R.R.B., K.K.K., D.N.K., B.R.R.); Biometrics Center, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, Mass (R.A.P.); The Center for Morphometric Analysis, Massachusetts General Hospital, Harvard Medical School, Charlestown, Mass (D.N.K.); and Department of Neurology, Spaulding Rehabilitation Hospital, Harvard Medical School, Boston, Mass (G.N., J.D.K.).
Correspondence to Steven C. Cramer, MD, VA Medical Center, Department of Neurology (127), 1660 S Columbian Way, Seattle, WA 98108. E-mail cramers{at}uwashington.edu
| Abstract |
|---|
|
|
|---|
Methods Functional magnetic resonance imaging was performed in ten stroke subjects with good recovery, five with deep, and five with cortical infarcts. Brain activation was achieved by index finger-tapping. Statistical parametric activation maps were obtained using a t test and a threshold of P<.001. In five bilateral motor regions, the volume of activated brain for each stroke subject was compared with the distribution of activation volumes among nine controls.
Results Control subjects activated several motor regions.
During recovered hand finger-tapping, stroke subjects activated
the same regions as controls, often in a larger brain volume. In the
unaffected hemisphere, sensorimotor cortex activation was increased in
six of nine stroke subjects compared with controls. Cerebellar
hemisphere contralateral and premotor cortex ipsilateral to this
region, as well as supplementary motor areas, also had increased
activation. In the stroke hemisphere, activation exceeding controls was
uncommon, except that three of five cortical strokes showed
peri-infarct activation foci. During unaffected hand finger-tapping,
increased activation by stroke subjects compared with controls was
uncommon; however, decreased activation was seen in unaffected
sensorimotor cortex, suggesting that this region's responsiveness
increased to the ipsilateral hand and decreased to contralateral hand
movements. Use of a different threshold for defining activation
(P<.01) did not change the overall findings (
=.75).
Conclusions Recovered finger-tapping by stroke subjects activated the same motor regions as controls but to a larger extent, particularly in the unaffected hemisphere. Increased reliance on these motor areas may represent an important component of motor recovery. Functional magnetic resonance imaging studies of subjects who recovered from stroke provide evidence for several processes that may be related to restoration of neurologic function.
Key Words: stroke outcome magnetic resonance imaging motor activity
| Introduction |
|---|
|
|
|---|
Studies in animals have provided several insights into the basis for recovery. First, electrophysiological studies in monkeys demonstrated that recovery from an experimentally induced focal lesion in primary sensory3 or motor4 cortex may be associated with peri-infarct reorganization of the hand representational map. Second, increased activation in the supplementary motor area (SMA) has also been identified in association with motor recovery.5 Third, a number of cellular and histological changes have been documented in the unaffected hemisphere during recovery from stroke. For example, rats given a focal cortical lesion develop cortical dendrite expansion in sensorimotor cortex of the unaffected hemisphere, which peaks at day 18 postlesion; interventions reducing this dendritic response are associated with less recovery.6
These three processes have also been found in humans both normally and during stroke recovery. For example, plasticity of the motor cortex representational map has been described with motor learning in normal subjects.7 After unilateral brain insult, rearrangement of motor maps has been identified in the unaffected8 and stroke-affected9 hemisphere using transcranial magnetic stimulation. Recruitment of SMA has been described as a normal accompaniment to anticipation or sequencing of a motor task, as well as performance of a complex task.10 Clinical studies suggest that SMA may contribute to return of motor function after stroke.11 Increased task complexity is associated in normal subjects with increased activity in motor cortex ipsilateral to the active hand.12 13 Weiller et al,14 using positron emission tomography (PET), studied subjects who recovered from deep capsular infarct and found individual patterns of increased cerebral blood flow (CBF) in multiple brain regions, including bilateral sensorimotor cortex and SMAs. Although these studies have provided insight into the cerebral responses to stroke, additional functional neuroimaging studies are needed to further characterize these processes, such as examination of a wider range of stroke subtypes.15
In the current study, functional magnetic resonance imaging (fMRI) was used to determine if increased activation in peri-infarct foci, SMA, or unaffected hemisphere motor regions was present as evidence for operation of the above processes in stroke recovery. fMRI with blood oxygenation level-dependent contrast uses the intrinsic paramagnetic signal of deoxyhemoglobin, rather than exogenous or radioactive tracers, to detect the local changes in blood flow, which occur in association with neuronal activation.16 17 18 Its excellent spatial and temporal resolution makes it ideal for the study of functional brain anatomy. Activation patterns in healthy subjects were compared with those found in subjects who recovered from stroke, including subjects who recovered from a cortical infarct, a group not previously examined with PET or fMRI. We found patterns of motor activation in normal subjects that were similar to those described previously. Most stroke subjects activated the same motor regions, but to a greater extent within unaffected hemisphere motor areas, SMAs, and foci along the rim of a cortical infarct, suggesting a contribution by all three of the above processes. Subjects with different stroke topographies, but equal degrees of motor recovery, showed similar activation patterns.
| Subjects and Methods |
|---|
|
|
|---|
4+ on the Medical Research Council scale19
in both wrist extensor and hand interossei muscles for at least 48
hours; (2) good motor recovery defined by resolution of upper extremity
synkinesis, improvement in wrist extensor and interossei strength by at
least one level on the Medical Research Council scale, and final
strength at least 4+ in either wrist extensors or interossei; (3) no
prior stroke with sensorimotor deficits. All stroke subjects received
postinfarct motor rehabilitation therapy. Stroke topography was
classified20 as either deep (lenticulostriate or anterior
choroidal artery territory) or cortical infarction. Nine elderly
controls (4 males) older than 60 years were recruited through local
advertisements; each had a normal neurologic examination, no history of
a stroke, and no significant active neurological problems. Neurological
examinations were conducted within 2 days of MRI scanning in all but
one case. Handedness was evaluated using the Edinburgh
Inventory.21 The presence of stroke risk factors was
examined, including hypertension, diabetes mellitus,
hypercholesterolemia, cigarette smoking,
coronary artery disease, and atrial fibrillation. Subjects were
trained to tap each index finger at the metacarpophalangeal joint
before imaging and further practiced inside the MRI scanner.
MRI Image Acquisition
Echo planar images and conventional images were obtained using a
high-speed, whole-body scanner (1.5 Tesla General Electric Signa
modified by Advanced NMR Systems) and a quadrature head coil. Head
movement was minimized with foam rubber pads and a restraining Velcro
band across the forehead, and body movement was limited by placement of
bilateral proximal arm straps. Each scanning session included the
following: (1) high resolution volumetric gradient echo images, 2.8-mm
thickness; (2) an automated shim procedure to improve magnetic field
homogeneity;22 (3) flow-compensated images in plane with
functional images; (4) high-resolution echo planar anatomic images in
plane with functional images; (5) in the stroke subjects, a measurement
of relative CBF using the flow-sensitive alternating inversion recovery
(FAIR) method23 24 25 in a single supra-Sylvian axial slice;
(6) blood oxygen level-dependent contrast functional images, consisting
of asymmetric spin echo images for T2* signal change, with TR of 2.5
seconds, TE of 70 mseconds, effective field of view of 20
cm2, and in-plane resolution of 3.1 mm2.
One functional study examined right index finger-tapping, and a
subsequent study examined left index finger-tapping. Index
finger-tapping was selected because such fractionated finger movements
represent the highest motor recovered state after
stroke.26 The whole brain was examined using 20 contiguous
horizontal slices of 7-mm thickness; for stroke subject 4, only the
cerebrum (12 slices) was imaged because of use of an alternate imaging
protocol. One hundred images were obtained per slice over a 4-minute
period, during which subjects alternated between 30-second periods of
rest and activity. The first two subjects studied (subjects 1 and 2)
tapped at 2 Hz ad libitum, whereas a 2-Hz metronome beep was
transmitted through plastic headphones (continuously during rest and
active periods) in subsequent studies. The cue to begin and to cease
movements was a light tap on the knee. Subjects kept eyes closed at all
times. All movements were monitored for accurate performance by
one of the experimenters standing in the scanner room at the subject's
side.
Image Analysis
Image analysis was performed on Sun SPARC workstations.
Head motion was detected and corrected using image registration
software adapted for fMRI by Jiang et al27 from Woods et
al.28 The algorithm automatically removes the top and
bottom slices, because even small amounts of movement through the plane
can cause voxels to leave/enter the image volume in these slices. To
account for this, imaging was performed so the top slice was superior
to the brain and, thus, contained no relevant information. Statistical
parametric maps were generated voxel-by-voxel using a
t test, contrasting images taken during finger-tapping with
those taken at rest. Identification of images as tapping or rest was
shifted 5 seconds to account for the delay normally seen in the rise of
CBF after neuronal activation.29 Statistical maps were
mildly smoothed with a Hanning filter to improve the signal-to-noise
ratio, halving the effective in-plane resolution. Maps showing
significant motion artifact,30 defined as continuous
activation of voxels with P<.001 around the rim of
one-fourth or more of the brain circumference on any analyzed
slice, were excluded from further analysis.
For each subject, the volume of activation was measured in five
bilateral motor regions. By applying region of interest (ROI) template
methods described previously,31 32 33 each subject's echo
planar high-resolution images were used to generate ROIs for
cerebellum, basal ganglia, SMA, premotor cortex (PMC), and sensorimotor
cortex (SMC). The SMC was further divided into precentral gyrus and
postcentral gyrus for secondary analyses. By using an
adaptation of methods described previously,34 parcellation
of ROIs was guided by reliable sulcal landmarks35 and a
multiplanar radiological brain atlas,36 while viewing the
high-resolution gradient echo images reformatted in the three cardinal
planes simultaneously. Infarcted tissue was excluded from
ROIs. Basal ganglia contained caudate head, globus pallidus, and
putamen. Cortical ROI definitions are presented in Fig 1
. ROIs were generated from anatomic
scans, without knowledge of brain activation patterns; the number of
significantly activated voxels within each was then counted
using a threshold of P<.001.
|
An ipsilateral index was calculated for each task and group, defined as (C-I)÷(C+I), where C=contralateral SMC activation volume and I=ipsilateral SMC activation volume. The Wilcoxon test was used to compare values for control-right finger tap with stroke subjects-recovered finger tap (recovered finger was the right for all but 2 stroke subjects) and control-left finger tap with stroke subjects-unaffected finger tap.
Values for relative CBF were obtained by subtracting flow-insensitive from flow-sensitive images for 64 pairs of images.23 24 25 The mean value for relative CBF, obtained from the 64 pairs of subtractions, was measured in each voxel of noninfarcted cortex within the territory of the middle cerebral artery.20 These results were then compared with an identically sized, symmetrical set of voxels from the unaffected hemisphere.
Statistics
Within each ROI, increased activation in stroke subjects was
identified by comparing the number of activated voxels for each
stroke subject against a distribution generated from all control
subjects' activations for the same task. To accomplish this, the mean
number of activated pixels among control subjects was treated
as a Poisson-distributed variable for each of the ROIs, separately
for each task. A one-sided upper 95% confidence limit was then
calculated. In the majority of cases, a conventional two-sided
confidence interval included zero, leading to use of a one-sided limit.
Increased activation in a stroke subject was considered to be
present in an ROI when the number of activated voxels
exceeded the 95% upper limit established in controls for that ROI.
Activation data from stroke subjects right finger-tapping were compared
with data from control subjects right finger-tapping; left
finger-tapping data between the two groups were also compared. Results
of these comparisons were then classified as related to paretic or to
unaffected finger-tapping. Decreased activation in stroke subjects was
identified by comparing activation in controls with a new Poisson
distribution generated from stroke subjects. To simplify this
comparison, available data from only the eight subjects with left-sided
infarcts were used in this analysis. Control right
finger-tapping results were then compared with recovered (right)
finger-tapping and control left tapping with unaffected (left)
finger-tapping, separately for each task and ROI. An ROI with more than
four control subjects exceeding the 95% upper limit derived from
stroke subjects was considered to be indicative of decreased activation
in stroke subjects.
To insure that results were not dependent on the threshold used to
define significance, all calculations for recovered finger-tapping were
repeated using a threshold of P<.01 rather than
P<.001. A
statistic was then used to assess the degree
of agreement between the two different thresholds in identifying
increased activation in stroke subject ROIs.
A FAIR study of relative cortical CBF was performed in seven of the stroke subjects; two other subjects were studied before FAIR was added to the protocol, whereas in a third, FAIR was omitted. For the CBF measurements, the distribution of values of relative cortical CBF from the stroke-affected hemisphere was compared with the distribution from the unaffected hemisphere using a Wilcoxon test; significance level was set at P<.05.
| Results |
|---|
|
|
|---|
|
Brain Activations (Fig 2
)
Normal Subjects
One study was excluded because of excess motion (right finger
tap). One subject had small amounts of middle finger movement during
right finger-tapping, and another had some thumb flexion during left
finger-tapping. The principal ROIs activated (mean number of
activated voxels,
4) during right (dominant) index
finger-tapping were contralateral SMC (mean of 15.1 voxels
activated), ipsilateral cerebellum (9.6), ipsilateral SMA
(4.5), and contralateral SMA (4); the mean number of activated
voxels was 3.1 for contralateral cerebellum, 1.4 for ipsilateral SMC,
and .25 for ipsilateral PMC. During left (nondominant) index
finger-tapping, principal ROIs activated (mean number of
activated voxels,
4) were contralateral SMC (29.2),
ipsilateral cerebellum (25.2), ipsilateral SMC (9.8), contralateral SMA
(8.4), and contralateral cerebellum (8.2); the mean number of
activated voxels for ipsilateral cerebellum was 3.8. Basal
ganglia activation was inconsistently seen; bilateral
activation of basal ganglia during motor task performance is
more apparent with higher magnetic fields and alternate imaging
strategies.38 In most cases, control subjects had larger
ipsilateral and contralateral SMC activations with tapping of the left,
nondominant finger compared with the right, dominant finger. The values
for SMC activation are presented in Table 2
. A typical control study is
presented in Fig 2
.
|
|
Stroke Subjects, Unaffected Hand Index Finger Tap
Subject 10 was excluded because of excess motion, and data were
not acquired for subject 2. The principal ROIs activated during
unaffected finger tap were the same compared with left finger tap by
controls (unaffected finger was on the left in eight of ten stroke
subjects), except that SMA activation was ipsilateral to the tapping
finger, not contralateral. Unaffected finger-tapping by stroke subjects
exceeded control values in very few cases. In five of eight evaluable
studies, no ROI exceeded the 95% upper limit defined in controls.
Increased activation was seen in unaffected hemisphere SMC and SMA for
subject 3, stroke-side cerebellum for subject 7, and stroke-side SMC
plus PMC for subject 5. Decreased activation was identified in one ROI,
the contralateral SMC, where six control subjects (left finger tap)
exceeded the 95% upper limit derived from data on stroke subjects with
left-sided infarct (left finger tap).
Stroke Subjects, Recovered Hand Index Finger Tap (Fig 2
)
Subject 4 was excluded because of excess motion. Mirror movements
in the unaffected hand were visible during scanning in only one case,
subject 1, in whom they were very small. Subjects 3 and 5 showed very
small simultaneous movement of recovered hand thumb,
whereas subject 10 had minimal middle finger movement; these movements
were present during <25% of index finger taps. The principal ROIs
activated during recovered finger tap included all of the
regions activated by controls during right finger tap
(recovered finger was on the right in eight of ten stroke subjects). In
addition, three other regions exceeded a mean of four activated
voxels: SMC and PMC ipsilateral to the tapping finger and cerebellum
contralateral to the tapping finger. The volume of brain
activated during recovered finger-tapping frequently exceeded
the 95% upper limit defined in controls, with one or more ROIs showing
increased activation in seven of nine studies.
The most common ROI with increased activation was the SMC of the
ipsilateral, unaffected hemisphere (Fig 2
), where six stroke subjects
(1, 2, 3, 6, 7, and 10) exceeded the 95% upper limit defined by
normals. Use of an ipsilateral index also identified increased
ipsilateral SMC activation in stroke subjects (Table 2
). In contrast,
only three subjects (1, 7, and 10) showed increased activation in the
contralateral stroke-side SMC. The voxel counts for right and for left
SMC are shown for individual subjects in Table 2
. Some SMC activations
were centered anterior to the central sulcus, and some were posterior.
This was equally true in control and stroke subjects. Thus, when
separate Poisson distributions for precentral and postcentral gyri were
created in the unaffected hemisphere, increased activation was found
exclusively in the same six subjects who had increased activation when
these two gyri were combined as SMC. Likewise, the same three subjects
with increased stroke-side SMC activation were those with an increased
activation after SMC subdivision.
Three other areas often showed increased activation, but only among
subjects with increased activation in unaffected SMC. Five of nine
subjects (1, 2, 6, 7, and 10) showed increased activation in either SMA
(unaffected side, three cases; stroke-side, four cases; both, two
cases; see Fig 2
). Four subjects (1, 2, 6, and 10) showed increased
activation in unaffected hemisphere PMC. Stroke-side cerebellum (ie,
the cerebellar hemisphere contralateral to unaffected hemisphere SMC)
showed increased activation in five subjects (1, 2, 6, 7, and 10).
An additional pattern of increased activation was found in three
(subjects 7, 9, and 10) of the five subjects with cortical stroke. In
these three subjects, contralateral SMC activations were present at
the hand area along the central sulcus37 ; however, foci of
increased activation were also present within one voxel of the
infarct rim (Fig 2
). For subject 7, one was located on the central
sulcus, two were on the postcentral sulcus, and one was in premotor
areas. For subject 9, one was located on the precentral gyrus, and one
was on the postcentral gyrus. For subject 10, two were located on the
precentral gyrus, and one was in premotor areas.
Stroke-side PMC showed increased activation in one subject. Unaffected side cerebellum showed increased activation in three subjects, one of whom also showed increased activation in stroke-side SMC. Unaffected hemisphere basal ganglia showed increased activation in two subjects. Stroke-side basal ganglia did not show increased activation in any of the stroke subjects. Recovered finger-tapping was not associated with decreased activation in any ROI.
Values for the ipsilateral index are presented in Table 2
.
Values for recovered finger-tapping (right side in eight of ten
subjects) by stroke subjects ranged from -.5 (predominantly
ipsilateral activation) to +1 (contralateral activation only), with a
mean of +.44. This value is significantly less than that found in
controls-right finger-tapping (range, +.56 to +1, mean, +.84;
P<.05), indicating a relatively increased reliance on
ipsilateral SMC by stroke subjects. A trend toward a lower index for
stroke subjects-unaffected finger tap (+.49) compared with
controls-left finger tap (+.69) did not achieve significance
(P>.5).
Varying the threshold used to define activation in a voxel had little
impact on patterns of increased activation in stroke subjects. Control
and stroke subjects both showed an increased number of
activated voxels when the threshold for significance was
changed from P<.001 to P<.01, but little
changed on comparing the two groups. Substantial39
agreement between the two thresholds was found (
=.75) on comparing
the stroke subject ROIs identified as having increased activation
during recovered finger-tapping.
For subjects with deep stroke (3 and 5), the FAIR slice was located immediately supra-Sylvian and did not include infarct. The FAIR slice was through the infarct, as intended, for cortical stroke subjects 6, 7, 9, and 10; for subject 8, the slice was inferior to the cortical infarct. No significant asymmetry in cortical CBF was present in six of these subjects. In one subject (subject 5), relative CBF in stroke-side cortex was 134% of unaffected side cortex, suggesting either increased flow in the stroke-side hemisphere or decreased flow in the unaffected hemisphere. Although magnetic resonance angiogram was normal in this subject, a mild stenosis of the middle cerebral artery in the unaffected hemisphere was suggested by the observation of increased velocities (40%) in that artery during transcranial Doppler examination. One control subject, incidentally discovered to have an occluded left internal carotid artery during acquisition of flow-compensated images, received a FAIR study that showed no asymmetries.
| Discussion |
|---|
|
|
|---|
In recovered stroke subjects, the region most frequently showing an increased volume of activation over controls was the unaffected hemisphere SMC during recovered finger-tapping. Although most controls showed some activation in ipsilateral SMC, an increased volume of brain was activated in this region in six of nine stroke subjects during recovered finger-tapping. The ipsilateral index mitigates the impact of intersubject differences that may exist in the blood oxygen level-dependent response to neuronal activation because it gauges ipsilateral activation relative to that found contralaterally within a given subject. The mean value in controls (.84) is approximately that found by Kim et al (.9),40 whereas the mean value in stroke subjects was significantly lower (.44), consistent with increased reliance on ipsilateral SMC. Recruitment of ipsilateral, unaffected SMC after brain injury has been identified using single photon emission computed tomography,44 PET,14 42 43 fMRI31 and transcranial magnetic stimulation.8 For some stroke and control subjects, SMC activation was on both sides of the central sulcus; in some, it was centered on the precentral gyrus, whereas for others it was on the postcentral gyrus. This variation was similar between stroke and controls, as dividing the SMC into separate precentral gyrus and postcentral gyrus ROIs did not alter the classification of any stroke subject's activation as being increased or not relative to controls. The observations may reflect the normal rostral-caudal variability in motor activation site45 or result from the contribution to corticospinal tract by both the precentral and the postcentral gyri.46 47
The rarity of mirror movements in the current study may shed light on the significance of increased activation in unaffected hemisphere SMC. In a prior PET study,14 all of the subjects showing increased CBF in unaffected hemisphere SMC during use of the recovered hand were also noted to have simultaneous unintended (mirror) movements in the unaffected hand. As a result, the authors could not determine whether the increased activation of unaffected SMC was merely a reflection of mirror movements in the unaffected hand or if such activation genuinely reflected a restorative role by ipsilateral SMC. The occurrence of mirror movements during scanning in only one subject in the current study supports the view that increased activation in unaffected SMC is not merely a marker for simultaneous movement of the unaffected hand. It is conceivable that minute degrees of mirror movement activity might have been found with sensitive methods such as electromyography. Observed activations in unaffected SMC were relatively large, however, and minute muscle forces would be expected to produce relatively small activations.48 Variance in mirror movement incidence between the two studies might be explained by differences in the stroke subject characteristics or by the nature of the intended task. Indeed, Armatas et al49 showed that the relative degree of mirror movements is increased with use of the pinky compared with the index finger. Pinky movements were part of the activating task used by Weiller et al, but not in the current study.
Other evidence supports a restorative role for unaffected SMC. Divergent methods indicate that a relationship normally exists between SMC and ipsilateral hand motor function. Approximately 15% of the corticospinal tract is known to descend undecussated.50 Moving the ipsilateral hand is associated with increased activity in motor cortex neurons, as measured electrophysiologically,51 with fMRI,40 and with magnetoencephalography.12 Cortical stimulation in monkeys52 and in humans using transcranial magnetic stimulation53 has been associated with motor responses in ipsilateral fingers. The relationship between SMC and ipsilateral hand movements may be intensified after injury. Studies of rats with motor recovery after an experimental lesion have identified an expansion of dendritic arborization in pyramidal cells of unaffected SMC. Blocking this histological reaction was associated with a lesser degree of motor recovery in the affected limb,6 54 suggesting that neural growth in unaffected hemisphere SMC is part of an effective brain response to injury.
The stroke subjects showing increased activation of unaffected SMC also showed increased activation of motor regions known to have rich connections with this area, suggesting recruitment of a larger motor network. Thus, a subset of the six stroke subjects with increased activation in unaffected SMCand no other stroke subjectsshowed increased activation in the cerebellar hemisphere contralateral (five subjects) and the PMC ipsilateral (four subjects) to unaffected hemisphere SMC, as well as either SMA (five subjects). Although most control subjects showed some activation in SMC ipsilateral and cerebellum contralateral to the tapping finger, ipsilateral PMC activation occurred in only two of eight controls. These are part of a distributed motor network.55 The existence of a distributed motor network has been suggested in normal subjects at rest,56 during imagined movement,57 during actual finger-tapping,41 in stroke subjects at rest,58 and in stroke subjects during recovered finger-tapping.42 Increased reliance on multiple nodes in this network has been measured with imposition of a more complex task in normal subjects,10 12 13 as well as in the current study during performance of a simple task by stroke subjects. The analogy between performance of a complex task in normal subjects and a simple task in stroke subjects is highlighted by the self-observations of Brodal,59 that "very great mental effort" was needed to use a paretic muscle. The activation results lend support to the notion that motor recovery after stroke is accompanied by increased activity in an unaffected hemisphere motor network, which is normally present, but less active during performance of a simple task.
Unaffected SMC was also the only region in which decreased activation was frequently seen during unaffected finger-tapping. In rats subjected to a unilateral SMC lesion, immobilization of the unaffected forelimb for the first 15 days prevented dendritic growth and produced severe, lasting deficits in the impaired forelimb.54 Together, these observations suggest that during recovery from an infarct, unaffected limb activity drives neuronal changes in unaffected SMC, which are associated with recovery of the affected limb; but after recovery has reached a plateau, unaffected limb activity produces an attenuated response in unaffected SMC, in association with an increased response to affected limb activity.
A second recovery process suggested by the results is increased reliance on motor areas whose efferent tracts descend in parallel with those originating in SMC.11 60 This anatomic arrangement may underlie the increased activation in SMA and PMC, although activity in these areas might also relate to their connections with SMC. Alternatively, SMA activity increases with increased task complexity10 ; increased SMA activation in stroke subjects may, therefore, be further evidence for perceived increased complexity of finger-tapping for this group. The relative increase in SMA activity by subjects 1 and 2 need to be interpreted with caution, as in these two cases tapping rate was internally generated, which is known to produce larger SMA activation compared with externally guided tasks.10
A third recovery process, based on observations in three of five subjects with cortical stroke, relates to peri-infarct activation of foci; this may reflect cortical map reorganization, reliance on alternative motor representation sites, or both. Reorganization of representational maps within the surviving cortex has been identified during cortical stimulation studies in monkeys after experimental infarct in sensory3 or in motor cortex.4 In two recent studies examining adult monkeys with experimental cortical infarcts, hand motor recovery was accompanied by functional reorganization of viable regions of the motor cortex. When monkeys received motor retraining for several weeks postinfarct, the volume of brain representing hand movement expanded and did so at the expense of proximal arm representation.4 When monkeys spontaneously recovered without rehabilitative intervention, the volume of brain representing hand movements diminished, with expansion of proximal arm representation.61 Some of the observed peri-infarct activations may reflect cortical motor map reorganizations similar to those described in the monkeys. A second explanation for activation in peri-infarct foci is that these may represent recruitment of alternative motor representation sites, as have been described histologically62 63 using transcranial magnetic stimulation53 and using fMRI.64 Peri-infarct activation may represent SMC plasticity, recruitment of alternative motor sites, or a completely different phenomenon. In any case, the observation65 that the volume of surviving peri-infarct penumbra correlates with the degree of neurologic recovery suggests that preservation of this rim of tissue provides an increased substrate for recovery processes after stroke.
No differences were apparent between the subjects with deep and those with cortical strokes, with a nearly identical fraction of either subgroup showing increased activation in each ROI; however, differences between the two groups in the time post-stroke and in the incidence of persistent deficits complicate this comparison. Brion et al,66 using Xe CBF methods in paretic stroke subjects during slow squeezing, found that subjects with a cortical infarct had increased CBF within stroke-side SMC and parietotemporal areas, as well as unaffected side parietal, temporal, and occipital areas. Subjects with deep lesions showed CBF increases only in an unaffected side parietal region. Differences in degree of recovery, activation task, and resolution of the imaging technique may account for the divergent findings. Given that a cortical infarct at sites throughout the hemisphere may produce hemiparesis,67 further studies examining larger numbers of subjects with cortical infarcts are needed to better understand recovery processes relevant to this subgroup.
Results of prior studies using divergent methodologies compare
favorably with the current study. Lesion studies in monkeys have
highlighted potential motor contributions of the cortex outside of SMC.
For example, Aizawa et al5 found increased SMA activity
during performance of a learned motor task after removal of the
primary motor cortex. A potential role for PMC was suggested by the
studies of Bucy and Fulton,52 in which removal of the
bilateral motor cortex and unilateral premotor cortex produced moderate
motor deficits, but removal of the bilateral motor and premotor cortex
produced complete paralysis. Cao et al,31 using fMRI to
study subjects with unilateral perinatal brain injury, found that
movement of paretic fingers was associated with increased activity in
ipsilateral SMC. Sabatini et al,44 using single-photon
emission computed tomography to study CBF in a subject with a large
hemispheric lesion, found that movement of the recovered hand was
associated with substantial activation of the ipsilateral SMC and PMC.
Weiller et al,14 in a PET study of eight subjects who
recovered from deep capsular infarct, found individual patterns of
increased activation. Precise comparison with the current study is made
difficult by differences in patient population and activation task, but
use of the recovered hand was associated with increased CBF in SMC,
PMC, and a midline SMA region in many cases. The PET study identified
regions with significant increases in CBF using exogenous
C15O2 as a tracer, whereas the current study
identified regions with significant increases in activated
brain volume as determined by endogenous blood oxygen
level-dependent contrast. Despite differing methodologies, there is a
high level of agreement between the results of Weiller et al and those
of the current study. The similarities were preserved when an alternate
threshold was used to define significance in the current study, as
there was good agreement in the current findings between the two
thresholds (
=.75).
The most important finding in this study is that the motor task that produced large SMA and unaffected hemisphere motor region activations in stroke subjects was also associated with significant activations in these regions in most controls. This is consistent with the observations of Denny-Brown,68 who suggested on the basis of cortical ablation studies that movement return relies on mechanisms present as components of normal function. Most stroke subjects showed increased activation in unaffected hemisphere motor regions, suggesting reliance on the motor network related to this hemisphere, whereas few examples of increased activation in the stroke hemisphere were found. The difference between hemispheric activations was not likely caused by abnormalities in CBF. The FAIR studies, done to obtain an estimate of perfusion in noninfarcted cortical regions, showed symmetric perfusion in all but one case examined. Nevertheless, the current study represents application of a technique reliant on vasoreactivity in a population that by definition has some cerebrovascular pathology. Future fMRI studies of stroke recovery may, therefore, benefit from a measure of CBF in activated voxels as well as in voxels bordering a cortical infarct. In addition, serial fMRI studies commencing before completion of stroke recovery will shed greater light on the specificity of altered brain activation for neurologic recovery. As improved treatments targeting the stroke recovery period emerge,69 70 71 a more precise understanding of stroke recovery mechanisms may be directive. Functional MRI may be valuable in studying these processes.
| Selected Abbreviations and Acronyms |
|---|
|
| Acknowledgments |
|---|
Received June 30, 1997; revision received August 26, 1997; accepted September 30, 1997.
| References |
|---|
|
|
|---|
This article has been cited by other articles:
![]() |
J. D. Schaechter and K. L. Perdue Enhanced Cortical Activation in the Contralesional Hemisphere of Chronic Stroke Patients in Response to Motor Skill Challenge Cereb Cortex, March 1, 2008; 18(3): 638 - 647. [Abstract] [Full Text] [PDF] |
||||
![]() |
Y. Murata, N. Higo, T. Oishi, A. Yamashita, K. Matsuda, M. Hayashi, and S. Yamane Effects of Motor Training on the Recovery of Manual Dexterity After Primary Motor Cortex Lesion in Macaque Monkeys J Neurophysiol, February 1, 2008; 99(2): 773 - 786. [Abstract] [Full Text] [PDF] |
||||
![]() |
C. D. Takahashi, L. Der-Yeghiaian, V. Le, R. R. Motiwala, and S. C. Cramer Robot-based hand motor therapy after stroke Brain, February 1, 2008; 131(2): 425 - 437. [Abstract] [Full Text] [PDF] |
||||
![]() |
D. Lule, V. Diekmann, J. Kassubek, A. Kurt, N. Birbaumer, A. C. Ludolph, and E. Kraft Cortical Plasticity in Amyotrophic Lateral Sclerosis: Motor Imagery and Function Neurorehabil Neural Repair, December 1, 2007; 21(6): 518 - 526. [Abstract] [PDF] |
||||