|Home | About | Journals | Submit | Contact Us | Français|
Cerebral palsy (CP) is frequently linked to white matter injury in children born preterm. Diffusion tensor imaging (DTI) is a powerful technique providing precise identification of white matter microstructure. We investigated the relationship between DTI-observed thalamocortical (posterior thalamic radiation) injury, motor (corticospinal tract) injury, and sensorimotor function.
Twenty-eight children born preterm(16 males, 12 females; mean age 5y 10mo, SD 2y 6mo, range 16mo–13y; mean gestational age at birth 28wks, SD 2.7wks, range 23–34wks) were included in this case–control study. Twenty-one children had spastic diplegia, four had spastic quadriplegia, two had hemiplegia, and one had ataxic hypotonic CP; 15 of the participants walked independently. Normative comparison data were obtained from 35 healthy age-matched children born at term(19 males, 16 females; mean age 5y 9mo, SD 4y 4mo, range 15mo–15y). Two-dimensional DTI color maps were created to evaluate 26 central white matter tracts, which were graded by a neuroradiologist masked to clinical status. Quantitative measures of touch, proprioception, strength (dynamometer), and spasticity (modified Ashworth scale) were obtained from a subset of participants.
All 28 participants with CP had periventricular white-matter injury on magnetic resonance imaging. Using DTI color maps, there was more severe injury in the posterior thalamic radiation pathways than in the descending corticospinal tracts. Posterior thalamic radiation injury correlated with reduced contralateral touch threshold, proprioception, and motor severity, whereas corticospinal tract injury did not correlate with motor or sensory outcome measures.
These findings extend previous research demonstrating that CP in preterm children reflects disruption of thalamocortical connections as well as descending corticospinal pathways.
Cerebral palsy (CP) describes a group of motor-impairment syndromes secondary to a wide range of recognized and uncharacterized genetic and acquired disorders interfering with early brain development.1 CP can be etiologically grouped into disorders of early brain formation, injury associated with preterm birth, neonatal encephalopathies, and a heterogeneous group of postnatal disorders, which can be associated with specific risk factors.2 Etiological identification has implications for treatment, prognosis, and risk of recurrence.3
Studies using conventional magnetic resonance imaging (MRI) have shown that 70 to 90%of affected children have structural brain abnormalities.4 Careful interpretation frequently reveals patterns of selective vulnerability characteristic of gestational timing and severity of brain dysgenesis or injury.5,6
Children with spastic diplegia or quadriplegia frequently have white matter injury in association with preterm birth.7–9 Advances in imaging techniques have led to improved understanding of the pathogenesis, from the circumscribed lesions described in the past to the complex patterns of diffuse white matter injury now seen in preterm infants, referred to as periventricular white matter injury or periventricular leukomalacia.10 Other frequently reported findings include decreased cortical gray matter volumes11 and basal ganglia,12 thalamic,13 and cerebellar abnormalities, 14 as well as injury in subplate neurons.15 With advances in perinatal and neonatal care, the incidence of cystic forms of periventricular leukomalacia16 has fallen, whereas the identification of deep white matter injury (non-cystic periventricular leukomalacia) has risen.17,18
The pathogenesis of periventricular white matter injury is related to a combination of factors including pre-oligodendroglial vulnerability, ischemic or infectious activation of toxic reactive oxygen and nitrogen compounds, and the presence of regional vascular end zones or border zones.19,20 Neuropathological studies have shown lesions in corticospinal, thalamocortical, optic radiation, superior occipitofrontal, and superior longitudinal pathways.21,22
Although conventional MRI is of benefit in the identification of periventricular white matter injury, it does not provide information on the extent of injury in specific white matter pathways. Diffusion tensor imaging (DTI) captures restrictions in the random movement of water protons by macromolecules and myelin to visualize brain white matter tracts. It can be used to create two- and three-dimensional atlases of normal white matter architecture and to provide more precise identification of white matter dysgenesis and injury than conventional imaging.23,24
In two previous studies we demonstrated the utility of two- and three-dimensional DTI fiber tracking of sensory and motor pathways in children born preterm with periventricular leukomalacia.25,26 In these children we found more severe injury in posterior white matter fibers connecting the thalamus to the sensory cortex (thalamocortical pathways) than in descending corticospinal tracts – in contrast to the pathways classically linked to CP in preterm children; however, we did not know the clinical significance of these findings.
In the present study, we analysed DTI findings in 28 children in relation to neurological examination and quantitative sensorimotor assessment. Our primary hypothesis was that central sensory white matter injury would correlate with measures of sensorimotor function.
Participants for this case–control study were recruited from the Kennedy Krieger Institute’s Phelps Center for Cerebral Palsy, Baltimore, MD, USA. As part of a series of DTI studies of childhood CP, 41 consecutive patients with CP were scanned. Enrollment criteria were as follows: children from birth to 18 years of age; diagnosis of CP; clinically indicated brain scan for diagnosis or follow-up. Informed consent was obtained from the parents or guardians, and the protocol was approved by the institutional review board.
DTI research sequences were preceded by conventional MRI with a standard imaging protocol. A neuroradiologist not involved in the study interpreted the conventional images, which were subsequently reviewed with each family.
The present study was focused on children born preterm (<37wks’ gestation). Five children from the larger sample were excluded because they were born at term, and eight were excluded from analyses because the DTI sequences were incomplete or degraded by motion artifact, which precluded identification of the tracts of interest. The final study sample consisted of 28 children with periventricular white matter injury, diagnosed by neuroradiological review of conventional MRI, and 35 healthy age-matched children as controls.
Normative comparison data were obtained from our pediatric DTI database and comprised anonymized data from children scanned as healthy volunteers or for other unrelated pathologies, who were screened by a pediatric neurologist to rule out developmental abnormalities. All of the children in the control group were born at term.27 Children in the control group were distributed in the age ranges of 12 to 23 months (n=5), 2 to 3 years (n=11), 4 to 5 years (n=5), 6 to 8 years (n=6), 10 years (n=2), and 12 to 15 years (n=6). The control group comprised 16 females and 19 males with a mean age of 5 years 9 months (SD 4y 4mo, range 15mo–15y).
Clinical information, including a detailed history, was abstracted from medical record review, parental interview, and neurological examination (by AH), and was recorded on a structured data form. Motor findings included the type of CP (diplegia, quadriplegia, hemiplegia, extrapyramidal) and primary neurological findings (spasticity, dystonia, athetosis, chorea, hypotonia). Ambulatory status or best motor skill attained (rolls over, sits independently, crawls, stands with support, stands independently, walks with parent assistance, walks with assistive devices, walks independently) was assessed on clinical examination.
A subsample of 19 children, aged 4 to 8 years who could obey simple commands, received quantitative assessment of sensory and motor function. Sensory testing included measures of fine touch and proprioception and was performed with the child’s eyes closed. Semmes–Weinstein monofilaments (Smith and Nephew Rolyan Inc., German-town, WI, USA) were used to evaluate threshold values for touch. Proprioception was assessed as the number of times out of six test trials that a participant could accurately identify the direction of a joint motion of 10° or less at the great toe and index finger.
Strength was quantified in all participants using a MicroFET2 handheld digital dynamometer (Hoggan Health Industries, West Jordan, UT, USA). Strength was measured in seven upper- and lower-extremity large-muscle groups (quadriceps, hamstrings, hip flexors, wrist flexors, wrist extensors, biceps, and triceps) using isometric tests identical to those described by Bohannon.28 These muscles accounted for strength at two joints as well as for agonist and antagonist groups of a single limb (both upper and lower extremity). For this test the dynamometer was placed on the portion of the limb furthest from the joint (e.g. just proximal to the knee for a measure of hip flexion). Participants were instructed to increase their force to a maximum over 2 seconds, hold it for 4 to 5 seconds, and then relax. The two sides of the body were tested independently, and two determinations were made. To achieve the best representation of each upper and lower limb, we used the average force across the respective extremity muscle groups in this analysis. Pinch force was also measured. Spasticity was assessed in the arm (sum of wrist and elbow) and leg (sum of ankle and knee) using the modified Ashworth scale.29
An MRI study with standard imaging sequences preceded the DTI research protocol. Participants were imaged on a 1.5T scanner (ASC-NT, Philips Medical Systems, Best, the Netherlands). All but two children required sedation for the clinical images. Chloral hydrate was used under a standard sedation protocol with direct supervision by a research nurse until each child had returned to baseline state. Routine clinical pulse sequences were obtained including sagittal and axial T1-weighted sequences (4mm slice thickness, no interslice gap, repetition time 297–599ms, echo time 10.5–13ms), fat-saturated axial T2-weighted sequences (repetition time 3992–4525ms, echo time 110ms), and fluid-attenuated inversion recovery sequences (repetition time 6000ms, inversion time 2000ms, echo time 120ms).
DTI was acquired after all of the routine clinical sequences and consisted of a diffusion-weighted spin-echo pulse sequence with a single-shot echo-planar imaging readout, with repetition time ranging from 6.2 to 9.4 seconds and echo time of 80ms. Fifty axial slices parallel to the anterior–posterior commissure line were acquired, covering the entire brain. The maximum b value was 700s mm2, used in a scheme of 30 different gradient directions along with five reference images with minimal diffusion weighting. Sensitivity encoding was used, employing an 8-element phased-array coil, converted to a 6-channel coil to be compatible with a 6-channel receiver, with a sensitivity encoding reduction factor of 2.5. The field of view was adjusted to the brain size, and the imaging matrix was changed within a range of 80×80 to 96×96, resulting in an in-plane imaging resolution of 2.0–2.5mm. All images were zero-filled to a 256×256 matrix. Slice thickness was set to approximately the same as the in-plane resolution. Scan time varied from 4 minutes 18 seconds to 6 minutes 34 seconds per sequence. Three repetitions were performed to increase signal-to-noise ratio. Magnetization-prepared rapid-gradient echo images were also obtained with the same slice localization, number, and thickness.
Imaging data sets were transferred to a workstation, corrected for bulk motion using the automated image registration software (Woods RP, University of California, Los Angeles, CA, USA; http://bishopw.loni.ucla.edu/AIR5/index.html), and processed using DtiStudio (Johns Hopkins University and Kennedy Krieger Institute, Baltimore, MD, USA; http://www.mri.kennedykrieger.org). Fractional anisotropy, vector maps, and color-coded maps were generated. Orientation-based color-coding, a two-dimensional visualization approach, was used to identify specific white matter tracts. In this approach, image brightness represents diffusion anisotropy, with a red–green–blue color scheme indicating tract orientation (red, revealing fibers with lateral orientation; green, anterior–posterior; and blue, cranio–caudal).
Twenty-six white matter tracts were graded bilaterally on an ordinal scale (0=normal; 1=abnormal; 2=severely abnormal or absent) by the primary study rater (LMN), an experienced neuroradiologist, who was masked to clinical status. Abnormalities of the white matter tracts were based on size reduction on visual inspection in comparison with unaffected age-matched children without CP, in whom white matter tracts were all scored 0. A total injury score from 0 to 4 was calculated for each white matter tract by adding scores from the left and right side. A more detailed description of the imaging methods and white matter classification system has been published by Wakana et al.23 and Nagae et al.26 who established intra- and interrater reliability for this tract scoring scale within the reference population.
For the present study, the analysis was focused on thalamocortical pathways connected to the sensory cortex and descending corticospinal tracts. We hypothesized that the posterior thalamic radiation, which connects the thalamus with the parietal and occipital lobes, would be most related to sensory function, whereas the corticospinal tracts are pyramidal-pathway fibers related to motor function.
Statistical analysis was performed using SPSS 13.0 for Windows (SPSS Inc., Chicago, IL, USA). Analysis of ranked variables was performed using the Spearman’s rank correlation test. A significance level of 0.05 was used for the a priori hypotheses described above.
The mean age of participating children with CP was 5 years 10 months (SD 2y 6mo, range 16mo–13y). There were 16 males and 12 females. Gestational age at birth ranged from 23 to 34 weeks (mean 28wks, SD 2.7). Mean birthweight was 1.26kg (SD 0.51, range 0.46–2.3kg). The majority of children (21 out of 28) were diagnosed with spastic diplegia, four had spastic quadriplegia, two had hemiplegia, and one had ataxic/hypotonic CP. Fifty-four percent of the participants (15 out of 28) walked independently.
On conventional MRI, all participants had evidence of periventricular white matter injury. Nine patients presented with moderate ventricular dilatation, including one with a porencephalic cyst. Data on white matter injury in the posterior thalamic radiation and corticospinal tracts by gestational age is shown in Table I. Using two-dimensional visualization of the white matter tracts, 27 of the 28 children had evidence of posterior thalamic radiation injury. Corticospinal tract abnormalities were observed in eight children. DTI was superior to conventional MRI in visualization of both presence and variability of white matter tract injury (Fig. 1). In this sample of children with CP who were born preterm, there was no relationship between gestational age and injury severity in either the posterior thalamic radiation or the corticospinal tracts.
Injury severity in the posterior thalamic radiation tracts was significantly related to the severity of sensory and motor involvement (Table II). Children with abnormal posterior thalamic radiation tract scores had higher touch thresholds on the contralateral side of the body (right-sided injury, left-sided touch: rho=0.78, p=0.003; left-sided injury, right-sided touch: rho=0.72, p=0.008). Right-sided posterior thalamic radiation injury also correlated significantly with reduced contralateral and upper- and lower-extremity proprioception (contralateral: rho=−0.86, p=0.001; upper-extremity: rho=−0.69, p=0.03; lower-extremity rho=−0.67, p=0.02). We also found that children with abnormal posterior thalamic radiation tract scores had reduced contralateral lower-extremity strength scores (right-sided injury, left-sided strength: rho=−0.60, p=0.04; left-sided injury, right-sided strength: rho=−0.51, p=0.087).
There was no significant relationship between corticospinal tract injury and quantitative measures of sensation or strength (Spearman’s rho ranges: sensation 0.29–0.47; proprioception −0.10 to −0.31; strength 0.00–0.31).
Excluding two children less than 2 years of age, for whom final ambulatory abilities could not be determined, in a post-hoc analysis we found that ambulation deficits were positively associated with severity of posterior thalamic radiation injury (right-sided rho=0.57, p=0.001; left-sided rho=0.37, p=0.03; total rho=0.52, p=0.003). There was no association between ambulation status and corticospinal tract injury (rho range −0.01 to 0.01).
DTI has been fundamental to the evolving understanding of the complexity and variability in brain injury in children born preterm with CP. In two previous studies we demonstrated that white matter injury in preterm infants preferentially involves thalamocortical pathways.25,26 This observation is consistent with conventional MRI of children with periventricular white matter injury, most of which is seen posterior and superior to the occipital horns or atria of the ventricles, corresponding to the posterior thalamic radiation location. In the present study we extend our understanding of the pathophysiology of CP in children with periventricular white matter injury by showing that injury severity in thalamocortical pathways can be linked to sensorimotor function. By contrast, corticospinal tract injury was not related to measures of either strength or sensation.
The correlations that we observed between the severity of posterior thalamic radiation abnormalities on DTI and deficits in measures of touch and proprioception are consistent with the role that thalamocortical pathways play in sensory perception. However, the significant relationship between posterior thalamic radiation abnormalities and deficits in walking and strength were less expected. They suggest that motor impairments in CP may be caused by disruption of sensory connections that are distinct from or in addition to injuries to motor pathways.
Damage to the posterior thalamic radiation may impair motor function because the somatosensory cortex has an important influence on the motor system, as shown in simplified form in Figure 2. Sensory information from the periphery is initially processed in the thalamus, with diffuse projections to the cortex, including the parietal–occipital cortex, by posterior thalamic radiation.30 The parietal cortex connects to premotor, prefrontal areas as well as to the cerebellum through pontine nuclei.31 Integrating this information with basal ganglia and other inputs, pre-rolandic motor centers determine motor activity through the descending corticospinal tracts.
In our theoretical model of motor impairment associated with periventricular white matter injury, we hypothesize that CP is primarily the result of injury in posterior thalamocortical pathways involved in motor control. These pathways are proximal to sensorimotor loops involving movement. Therefore, any injury outcome in descending motor pathways may be attenuated by the more proximal thalamocortical injury.
However, another recently published study has shown that DTI metrics (lower fractional anisotropy and higher transverse diffusivity) in descending pyramidal tracts correlate with more significant motor impairment.32 These metrics indicate microstructural injury in these tracts. In our study, we did not find macroscopic injury in the corticospinal tract at the pons or in the superior corona radiata, but we did not calculate fractional anisotropy values in any tracts.
Furthermore, injuries in other components of the motor circuits can also lead to disorders of motor control and movement. In children with hemiplegia associated with venous infarction, arterial infarction, or polymicrogyria, there is a correlation between injury in the pyramidal tracts and the severity of motor impairment.33 Injury in the basal ganglia in children with glutaric aciduria type 134 and in those with dyskinetic CP of various causes can also lead to CP.
Limitations of our study include the small sample size and the qualitative white matter tract grading system. This was a relatively small sample (n=28), and, because the sensorimotor assessment subsample had a narrower age range (4–8y) and a number of these children could not perform the simple motor tasks, the subgroup with quantitative sensorimotor outcomes available for analysis was even smaller (n=11–12). Therefore, moderate associations observed between variables, such as corticospinal tract and measures of strength (Spearman’s rho range 0.3–0.5), may rise to significance in a larger sample. Additional children need to be studied before we can draw definitive conclusions about the significance of the strength associations and timing of injury patterns in children with periventricular white matter injury.
It is reported that diffusion properties (anisotropy and diffusivity) vary with increasing brain maturation during infancy and childhood.35 Whether these quantitative approaches are more sensitive in detecting lesions in specific motor pathways than this qualitative (visual) tract assessment warrants further study.
Complications of preterm birth pose a continuing challenge to neurodevelopmental clinicians. Our research suggests that DTI provides new information, beyond that captured on conventional MRI, about the nature and extent of patterns of white matter cerebral abnormalities in children with CP born preterm. The visualization of posterior thalamic radiation injury through DTI may improve our ability to identify children at high risk for motor impairment earlier, and distinguish a subset of children with CP whose sensorimotor circuit is predominantly affected proximal to the corticospinal tract pathway. We believe that these findings are provocative and set up interesting questions to be studied in future research, including the relative contributions of macroscopic and microstructural injury in central white matter pathways.
The authors gratefully acknowledge the contributions of Dr Peter van Zijl and our neuroimaging research team, Terri Brawner, Kathleen Kahl, and Carolyn Gillen. Special thanks to the families and children who participated in this study. This study was supported by grants from the Dana Foundation Clinical Hypothesis Program in Imaging, United Cerebral Palsy Research & Educational Foundation, and the Johns Hopkins University School of Medicine General Clinical Research Center, grant no. M01-RR00052 from the National Center of Research Resources/National Institutes of Health, and National Institutes of Health grant RO1 AG20012, P41 R15241.