Ankle dorsiflexion fMRI in children with cerebral palsy undergoing intensive body-weight-supported treadmill training: a pilot study John P Phillips* MD, Department of Neurology and Pediatrics, University of New Mexico Health Science Center, Albuquerque, NM; Katherine J Sullivan PhD, Department of Biokinesiology and Physical Therapy, University of Southern California, CA; Patricia A Burtner PhD, Department of Pediatrics, University of New Mexico Health Science Center; Arvind Caprihan PhD, New Mexico Resonance and The MIND Institute; Beth Provost PT PhD, Department of Orthopedics, University of New Mexico Health Science Center; Ann Bernitsky-Beddingfield PT, Department of Neurology, University of New Mexico Health Science Center, Albuquerque, NM, USA. *Correspondence to first author at Department of Neurology, UNM MIND Imaging Center, 1 University of New Mexico, 1101 Yale Boulevard NE, Albuquerque, NM 87131-0001, USA. E-mail:
[email protected] This pilot study investigated the feasibility of using functional magnetic resonance imaging (fMRI) as a physiological marker of brain plasticity before and after an intensive body-weightsupported treadmill training (BWSTT) program in children with cerebral palsy (CP). Six ambulatory children (four males, two females; mean age 10y 6mo, age range 6–14y) with spastic CP (four hemiplegia, two asymmetric diplegia, all Gross Motor Function Classification System Level I) received BWSTT twice daily for 2 weeks. All children tolerated therapy; only one therapy session was aborted due to fatigue. With training, overground mean walking speed increased from 1.47 to 1.66m/s (p=0.035). There was no change in distance walked for 6 minutes (pre-: 451m; post-: 458m; p=0.851). In three children, reliable fMRIs were taken of cortical activation pre- and postintervention. Post-intervention increases in cortical activation during ankle dorsiflexion were observed in all three children. This study demonstrates that children with CP between 6 and 14 years of age can tolerate intensive locomotor training and, with appropriate modifications, can complete an fMRI series. This study supports further studies designed to investigate trainingdependent plasticity in children with CP. See end of paper for list of abbreviations.
Task dependent neuroplasticity is an important mechanism underlying motor recovery after brain injury. Active participation in task performance is critical to such recovery, which in animal models leads to an expansion of the representative motor cortex and increased synaptogenesis.1 Body-weightsupported treadmill training (BWSTT) is task-dependent training that has been used successfully in stroke rehabilitation.2 Several reports suggest that BWSTT may also improve gait in children with cerebral palsy (CP),3 although the mechanism of action has not been studied in this population; it is unknown whether cortical reorganization occurs in children with CP undergoing BWSTT as it does in adults after stroke.4 Functional magnetic resonance imaging (fMRI) has been used to identify patterns of cortical activation of the upper extremities in adults5 and children6 with brain lesions. Technical challenges make fMRI of the lower extremity difficult,7 thus, to our knowledge, there are few reports on the use of fMRI to understand neuroplasticity in gait rehabilitation, and there is no report using fMRI during lower extremity movement in children. The aim of the present pilot study was to determine whether children with CP could undergo a series of fMRI using ankle dorsiflexion as a motor stimulus. Another objective was to investigate whether children with CP could tolerate an intensive twice-daily BWSTT program. fMR images were analyzed to determine whether changes in cortical activation occurred as a result of BWSTT in children with CP. Method PARTICIPANTS
Six children with spastic CP (two females, four males; mean age 10y 6mo; age range 6–14y), recruited through local rehabilitation centers in Albuquerque, New Mexico, participated in this study. Four children had spastic hemiplegia and two had asymmetric spastic diplegia. Inclusion criteria were the ability to: (1) ambulate independently without an assistive device; (2) actively dorsiflex the most involved ankle at least 10˚; and (3) follow verbal directions for standardized testing and fMRI procedures. Children were excluded if they had undergone orthopedic surgery or neurosurgery in the past 12 months, had received botulinum toxin injections within the past 6 months, or had changed oral antispasticity medications in the month before testing. Informed consent and assent were obtained from the parents and children before enrollment. All children were classified as Level I of the Gross Motor Function Classification System, i.e. all children could walk and run independently, despite reduced balance, speed, and coordination.8 Table I summarizes participants’ demographic and clinical characteristics. Ethical approval for this study was obtained from the University of New Mexico Human Subjects Research Review Committee. CLINICAL ASSESSMENT
Clinical measures were taken in the 2-week period before and following BWSTT. Primary clinical outcomes, measured by an experienced pediatric physical therapist, included walking speed, walking endurance, and the Gross Motor Function Measure (GMFM)9 Dimension E which was used to determine changes in walking, running, and jumping. Walking speed was determined by measuring the time taken to walk along a 10-meter walkway. Children were encouraged to
Developmental Medicine & Child Neurology 2007, 49: 39–44 39
hydraulic weight-support system (LiteGait I system; Mobility Research, AZ, USA) and positioned over a motor-driven treadmill with variable speed control. The 12-session training protocol occurred twice daily (one morning, one afternoon session), 6 days a week for 2 consecutive weeks. Each training session consisted of a total walking time of 30 minutes (three 10-min walks interspersed with 5-min rest breaks). Body support decreased from 30% initially to 0% by the
walk as quickly as possible without running. Walking endurance was determined by measuring the distance walked in 6 minutes. Walking velocity and the 6-minute walk are valid and reliable measures of walking ability in children with or without neuromuscular disability.10,11 TRAINING PROTOCOL
Participants were fitted in a harness suspended from a
Table I: Demographic and clinical characteristics of study participants (n=6) No.
Sex
Age (y)
GMFCS level
Diagnosis
Clinical description
Medications (current)
Prior procedures
MRI
1
F
14
I
Left hemiplegia
None
Left UE tendon transfers, BTX-A injections to UE
Right MCA stroke
2
F
14
I
Right hemiplegia
None
M
6
I
Right hemiplegia
4
M
12
I
Diplegia right > left
Right UE tendon transfers, BTX-A and stretch casting to LE Right UE and LE BTX-A and stretch casting Right LE BTX-A and stretch casting
Left MCA stroke
3
5
M
9
I
Diplegia right > left
6
M
8
I
Right hemiplegia
Term delivery, regular school, language-based LD Term delivery, regular school, language-based LD Term delivery, special education 34wk delivery, twin, regular school, ‘A’ student 33wk delivery, special education 27wk delivery with IVH, regular school, ADHD
None None
None None
Bilateral derotational osteotomies Right LE BTX-A and stretch casting
Left MCA stroke PVL, left > right, posterior fossa cyst Normal Left PVL
GMFCS, Gross Motor Function Classification System; LD, learning difficulty; IVH, intraventricular hemorrhage; ADHD, attentiondeficit–hyperactivity disorder; UE, upper extremity; LE, lower extremity; BTX-A, botulinum toxin A; MCA, middle cerebral artery; PVL, periventricular leukomalacia.
Table II: Comparison of group means on pre- and post-intervention gait measures
10-meter walking velocity (m/s) Pre-intervention Post-intervention 6-minute distance (m) Pre-intervention Post-intervention
Mean
SD
Range
p
1.47 1.66
0.32 0.41
1.10–2.00 1.20–2.26
0.035
451 458
96 84
320–566 364–585
0.851
Table III: Ankle dorsiflexion and finger tapping cortical activation in three participants Participant no.
1 4 6
Region
Ankle Hand Ankle Hand Ankle Hand
Pre-intervention maximal activation Active volume, cm 3 % BOLD Total fMRI (Contra/ipsilateral) signal change activation 2.448 (2.448/0) 4.066 (4.066/0) 1.316 (0.735/0.581) 7.910 (7.910/0) 0.132 (0.132/0) 1.969 (1.969/0)
1.7 1.75 2.1 1.7 2.7 2.0
4.162 7.116 2.764 13.447 0.356 3.938
Post-intervention maximal activation Active volume, cm 3 % BOLD Total fMRI % change (Contra/ipsilateral) signal change activation 3.191 (3.191/0) 5.452 (5.452/0) 6.923 (4.240/0.735) 7.318 (7.318/0) 1.608 (1.594/0.014) 2.843 (2.843/0)
1.9 2.3 1.86 2.9 2.3 1.8
6.063 12.540 12.877 21.222 3.698 5.117
BOLD, blood oxygenation level dependent; fMRI, functional magnetic resonance imaging; S1, primary sensory cortex; M1, primary motor cortex; PMC, premotor cortex; SMA, supplementary motor area; Contra, contralateral; Pre/Post, pre-/post-intervention.
40
Developmental Medicine & Child Neurology 2007, 49: 39–44
46 76 366 58 939 30
end of training. Treadmill speeds ranged from 2.4 to 3.1kmph initially and increased to 3.7 to 5.0kmph with training. A physical therapist and two research assistants provided assistance with stepping on the treadmill during the initial training sessions. As gait patterns improved for each child through the 2week training period, facilitators were able to decrease the amount of assistance provided. fMRI INSTRUMENTATION AND PROCEDURES
fMRI data were acquired using a 1.5 Tesla scanner (Siemens, Germany) with a standard head coil with single-shot gradient echo-planar imaging (FOV=220mm, matrix=64×64, TE=50ms, TR=2600ms) to obtain 28 images, each with a slice thickness of 5mm. High resolution anatomic T1-weighted images were obtained (FOV=220mm, matrix=256×256, TE=10ms, TR=22ms), each with a slice thickness of 1.25mm. Anatomic and fMRI images were obtained in an axial direction parallel to the anterior commissure/posterior commissure line. fMRI studies were performed before and after intervention and included three motor tasks: (1) active ankle dorsiflexion of the involved ankle; (2) finger tapping of the uninvolved hand; and (3) active ankle dorsiflexion of the involved ankle (repeat of task 1). The first ankle fMRI experiment was used for analysis unless significant motion artifact occurred, in which case the second ankle fMRI run was used. Because the BWSTT intervention involved repetitive activity of both lower extremities, finger tapping of the uninvolved hand was used as an internal control which was expected to remain relatively constant compared with ankle dorsiflexion. A block design was used for fMRI studies with 26-second periods of rest alternating with 26-second periods of motor activity (ankle dorsiflexion or finger tapping at 0.5Hz) for a total of six cycles. In addition to keeping frequency constant at 0.5Hz, an MR-compatible ankle brace, set at 20˚ of plantar flexion and 5˚ of dorsiflexion, was worn by participants to standardize ankle range of motion. fMRI data were analyzed using AFNI software (http://afni.nimh.nih. gov). Data processing consisted of: (1) time shifting the data to correct for the time at which the slice was collected; (2) movement artifact correction by three-dimensional (3D) volume registration; (3) baseline drift correction by high-pass filtering; and (4) spatial smoothing with a Gaussian filter with a full-width-half-maximum of 6mm. fMRI activity was analyzed using a voxel-by-voxel linear regression, contrasting active and resting states over time. A cluster of voxels was considered to be active if the t-statistic for the
Table III continued Region of interest analysis: voxels active at t>5, contralateral cortex S1 M1 PMC SMA Pre/Post Pre/Post Pre/Post Pre/Post 653/1090 3216/5307 0/0 4399/4416 62/465 1391/1879
863/1054 3467/3441 0/0 2779/2485 343/1351 994/1483
409/472 1414/1476 359/394 521/238 0/150 0/6
997/1089 0/0 1566/3755 0/0 571/1948 0/0
stimulus function coefficient was greater than a t-value=5.0, and if there were at least three contiguous voxels. This corresponds to p