Microstructural Brain Development After Perinatal Cerebral White Matter Injury Assessed by Diffusion Tensor Magnetic Resonance Imaging
Objective. Brain injury in premature infants is characterized predominantly by perinatally acquired lesions in the cerebral white matter (WM). The impact of such injury on the subsequent development of cerebral WM is not clear. This study uses diffusion tensor magnetic resonance imaging (MRI) to evaluate the effects of cerebral WM injury on subsequent microstructural brain development in different WM areas of the brain.
Methods. Twenty premature infants (gestational age: 29.1 ± 1.9 weeks) were studied by conventional MRI within the first 3 weeks of life and again at term, with the addition at the latter time of diffusion tensor MRI. Ten of the preterm infants had cerebral WM injury identified by the early MRI and were matched with 10 premature infants of similar gestational age and neonatal course but with normal neonatal MRI scans. Diffusion tensor MRI at term was acquired in coronal and axial planes and used to determine the apparent diffusion coefficient, a measure of overall restriction to water diffusion, and the relative anisotropy (RA), a measure of preferred directionality of diffusion, in central WM, anterior frontal WM, occipital WM, temporal WM, and the posterior limb of the internal capsule. Diffusion vector maps were generated from the diffusion tensor analysis to define the microstructural architecture of the cerebral WM regions.
Results. At term, the diffusion tensor MRI revealed no difference in apparent diffusion coefficient among preterm infants with or without perinatal WM lesions. By contrast, RA, the measure of preferred directionality of diffusion and thereby dependent on development of axonal fibers and oligodendroglia, was 25% lower in central WM, the principal site of the original WM injury. However, RA was unaffected in relatively uninjured WM areas, such as temporal, anterior frontal, and occipital regions. Notably, RA values in the internal capsule, which contains fibers that descend from the injured cerebral WM, were 20% lower in the infants with WM injury versus those without. Diffusion vector maps showed striking alterations in the size, orientation, and organization of fiber tracts in central WM and in those descending to the internal capsule.
Conclusions. Perinatal cerebral WM injury seems to have major deleterious effects on subsequent development of fiber tracts both in the cerebral WM and more distally. The ultimate impact of brain injury in the newborn should be considered as a function not only of tissue destruction, but also of impaired subsequent brain development.
Periventricular leukomalacia, or cerebral white matter (WM) injury, is the major form of brain injury in the premature infant.1 Although the exact time of occurrence of the WM injury is not known, the bulk of available data implicates the periods shortly before birth and shortly after birth to varying extents.1 Although the adverse neurologic impact of the tissue loss caused directly by this perinatal WM injury is recognized,1 the impact of such WM injury on subsequent brain development is largely unknown. Previous studies of the brain of seemingly normal premature infants by quantitative volumetric magnetic resonance imaging (MRI) and by diffusion tensor MRI documented dramatic developmental changes from approximately 28 weeks of gestation to term in both cerebral cortex and WM.2–4 These changes include an increase in volume, surface area, and sulcation of cerebral cortex and in volume and microstructural organization of cerebral WM. The potential vulnerability of these rapidly developing systems to WM injury is suggested by our recent demonstration that cerebral cortical gray matter volume is reduced at term in premature infants who previously sustained perinatal WM injury.5 Moreover, myelinated WM volume was found to be reduced at term in the infants with perinatal WM injury.5 This latter quantitative observation is consistent with previous qualitative evidence obtained by conventional imaging for an impairment of myelination after WM injury.6–10
In this study, our focus is cerebral WM development after perinatal WM injury in the premature infant. We used diffusion tensor MRI to investigate fiber tract development in the cerebrum in premature infants who had sustained WM injury documented in the neonatal period by conventional MRI.
Infants for the study were selected from premature infants who had consecutive admissions to the neonatal intensive care unit at the Brigham and Women's Hospital and who had a gestational age of <32 weeks and the ability to undergo initial MRI (no longer required ventilatory support). Informed consent was obtained from the infants' parents. The initial conventional MRI was undertaken for all infants between 8 and 21 days of age. All infants also received cranial ultrasound scans during the first 72 hours of life and again at 7, 28, and 42 days of age.
Ten premature infants (gestational age: 29.2 ± 1.6 weeks; birth weight: 1318 ± 251 g) had cerebral WM injury identified by conventional MRI at the first examination (Fig 1) and were matched with 10 premature infants of similar gestational age (gestational age: 29.0 ± 2.1 week; birth weight: 1294 ± 310 g) and neonatal course but with normal MRI scans. The abnormalities that were classified as perinatal cerebral WM injury were characterized by diffuse or nodular periventricular hypointensities or hyperintensities on T1- and T2-weighted images, respectively (n = 9), and/or the presence of apparent cystic or cavitary lesions (n = 5). Infants with evidence of cerebral WM injury had lower Apgar scores (WM injury: 7 of 10 with 1-minute Apgar scores <5, and 4 of 10 with 5-minute Apgar scores <6; no WM injury: 9 of 10 with 1-minute and 5-minute Apgar scores >7; P < .05) and greater use of inotropic therapy (WM injury: 6 of 10 had received dopamine; no WM injury: 2 had received dopamine; P = .01). Diffusion tensor MRI was performed in all 20 infants at their expected due date, ie, term. The study was approved by the institutional Human Subjects Research Committee. No sedation was necessary for the MRI studies. Infants were fed immediately before the examination. Ear/hearing protection was used (Ear Muffs, Natus Inc, San Carlos, CA). A vacuum fixation pillow (S&S Ray Products, Inc, Brooklyn, NY) was used as a head cradle. Cardiorespiratory monitoring (MR-Equipment Corp, Bay Shore, NY) was used for all examinations.
Line Scan MR Diffusion Imaging
The diffusion-weighted line scan sequence was implemented on a 1.5T conventional whole-body MRI scanner (SIGNA; General Electric Medical Systems, Milwaukee, WI).11 This robust diffusion imaging method does not require cardiac gating or special gradient and receiver hardware. Previous studies12,13 showed that image distortions and incomplete fat-suppression artifacts, sometimes present in diffusion-weighted echo-planar images, are not found with the diffusion-weighted line scan sequence. The diffusion sequence was run with diffusion tensor encoding (2 low and 6 noncollinear high b-factors interleaved). For a diffusion weighting (b-factor) of 700 s/mm2, as used in our earlier infant studies,3 the resulting echo-time was 105 ms with the repetition time and the effective repetition time at 155 ms and 2480 ms, respectively. Images were acquired at a rectangular field of view of 180 × 165 mm and a matrix size of 128 × 96 columns. The bandwidth was chosen at 4 to 6 kHz to avoid loss of signal-to-noise at higher bandwidth and augmented image distortion caused by field inhomogeneities at lower bandwidth. The effective in-plane resolution was 1.4 × 2.7 mm with an effective section thickness of 7.3 mm. Because the directions of diffusion encoding did not coincide with the gradient coil principal axes, a diffusion gradient strength higher than the maximum gradient strength was selected for a single principal axis alone. Slices were acquired in all infants in coronal and axial planes.
Data Analysis for Diffusion Tensor MRI
Diffusion was measured in terms of the apparent diffusion coefficient (ADC), according to the Stejskal and Tanner equation.14 The matrix describing the directional dependence of the ADC was estimated for each voxel with nonlinear regression, as suggested by Basser and Pierpaoli.15 For each estimate, the 3 orthogonal eigenvectors and their related positive eigenvalues were calculated. The eigenvector with the maximum diffusion describes the direction along which maximum diffusion occurs (along the fiber-tract axis), whereas the eigenvector with the minimum diffusion represents the direction with the least diffusion (transverse to the fiber-tract axis). The average ADC was calculated as the average of the principal diffusions for each voxel and averaged for specific areas of interest (diameter = 2.6 mm). In the same area of interest, the relative anisotropy (RA) is the standard deviation of the eigenvalues of the diffusion tensor divided by the ADC and is given as a percentage. The magnitude of the percentage corresponds to the degree of preference of water movement along one direction. For an isotropic tissue, in which water diffusion is identical in all directions, regardless of the direction of measurement, the RA is 0%. The specific areas of interest were chosen in selected WM areas as shown in Fig 2. Regions of interest were selected carefully according to anatomic criteria, such as adjacency to the roof of the lateral ventricle in the central WM (typical localization of WM injury; coronal slice), the posterior limb of the internal capsule (axial slice), the frontal WM lateral to the frontal horns of the lateral ventricle (axial slice), the occipital WM (avoiding the optic radiation; axial slice), and the temporal WM close to the lateral angle of the temporal horn of the ventricle (avoiding hippocampal structures; coronal slice).
The principal direction of diffusion is given by the eigenvector that corresponds to the largest eigenvalue. Therefore, this eigenvector represents the fiber bundle direction, because the principal diffusion during the MRI diffusion encoding time is along the fiber long axis. For even better delineation of fiber tracts, we multiplied the eigenvector, which has the unit length of 1, by the RA. With this method, vectors that indicate the direction of principal diffusion would appear longer in areas with high anisotropy (fiber tracts) and shorter in areas without a preferred diffusion direction, as in gray matter.3
For the comparison of ADC and RA between premature infants at term with and without perinatal cerebral WM injury, Student's unpairedt test was used. A value of P < .05 was considered significant.
Quantitative Measures at Term of Water Diffusion in Cerebral WM Regions in Preterm Infants With or Without Cerebral WM Injury
In premature infants without cerebral WM injury, the measure of the ADC was similar in all areas of cerebral WM but lower in the internal capsule (Table 1). This finding confirms previous observations made by ourselves3 and others.4 The lower ADC in internal capsule than in cerebral WM may reflect in considerable part the restriction in water diffusion associated with the development of this tightly-packed fiber tract, including the occurrence of myelination, the latter essentially absent in the cerebral WM regions at this maturational stage (see “Discussion”). In the infants without WM injury, the most notable regional difference in RA, the measure of preferred directionality of diffusion, is the occurrence of markedly higher values in the internal capsule than in cerebral WM regions (Table 1). This change also is consistent with the developmental changes of this fiber tract (see “Discussion”).
It is notable that in premature infants with WM injury, there was no difference in ADC values in all WM regions when compared with ADC values in infants without WM injury (Table 1). By contrast, RA values were significantly lower both in central WM, the principal site of the original WM injury, and in the internal capsule, the site of the fibers descending from the area of the original WM injury (Table 1). In both central WM and the internal capsule, the mean RA was reduced by 25% in the infants with WM injury. RA in the other areas of cerebral WM was not significantly different among infants with and without WM injury.
Imaging of Microstructural Development at Term of Cerebral WM in Preterm Infants With or Without Perinatal WM Injury
The quantitative changes in RA in infants with perinatal WM injury just described suggested a disturbance in microstructural development of the central WM and its descending fibers. We next asked whether the diffusion tensor methodology could provide insight into the nature of the developmental disturbance. The geometric nature of the diffusion tensor can be used to display the architecture of the developing cerebral WM. Using the RA and its multiplication by the eigenvector, we generated vector maps of fiber orientation and density and tract size (Fig 3). The orientation and the density of the fibers are shown by the indicated vectors, and the degree of anisotropy is symbolized by the vector length. Fibers that are perpendicular to the image plane are distinguished by the use of colored dots.
Diffusion vector images obtained in the coronal plane of a premature infant without perinatal WM injury at term (Fig 3A) show a discrete dense array of in-plane fiber bundles throughout the internal capsule, around the horn of the lateral ventricle, extending into the corona radiata, and reaching cortical areas. Colored dots (yellow and green), representing fibers with out-of-image plane orientation, were identified superior to the lateral ventricle and corresponded most probably to anteroposterior fibers of the cingulate bundle. A similar bundle lateral to the lateral ventricle corresponded most probably to the fibers of the superior longitudinal fasciculus. By contrast to the image obtained in the premature infant without WM injury, a coronal image at the same site in a premature infant with perinatal WM lesions at term clearly is different. Thus, disruption of the orientation and the density of fibers descending around the lateral ventricle toward the internal capsule is apparent. Moreover, there is a notable absence of anteroposterior fibers in the central WM, the general region of the previous WM injury. The only anteroposterior fiber tract present to a similar extent in the infants with and without perinatal WM injury is the cingulate bundle. Moreover, in the posterior limb of the internal capsule itself (Fig 3B), the infant with perinatal WM injury shows fewer homologous directed vectors, and these are shorter and less densely packed than those in the posterior limb of the infant without perinatal WM injury (Fig 3A). This finding provides a structural correlate for the reduced RA measured in the posterior limb of the internal capsule after perinatal WM injury (Table 1). All of the vector maps of the infants with perinatal cerebral WM lesions at term showed fewer defined fiber tracts in the central WM and diminished directed fibers in the posterior limb of the internal capsule.
This study of premature infants at term by diffusion tensor MRI provides new information concerning normal WM development and, more important, the effects of perinatal WM injury on this development. The data also have important clinical implications.
Quantitative Measures at Term of Water Diffusion in Cerebral WM Regions in Premature Infants Without Perinatal WM Injury
Previous diffusion tensor MR studies3,4 of cerebral WM development in human premature and term infants demonstrated that, in general, the ADC decreases while the RA increases with brain maturation. The most prominent regional difference at term is the lower ADC and the higher RA in the internal capsule relative to the values in the cerebral WM regions. The posterior limb of the internal capsule, unlike the cerebral WM areas, exhibits active myelination at term, and the presence of myelin could underlie in part the lower overall diffusion of water, ie, the lower ADC, in the capsule. The increased RA in the internal capsule, indicative of high directionality of diffusion, also could be related in part to myelination, with its restriction of water diffusion perpendicular to the fiber axis. However, axonal diameter increases before and during myelination, and this diameter change also could contribute importantly to the diminished water diffusion perpendicular to the orientation of the fiber and thereby the increase of RA. In our previous study of the developmental changes of RA in cerebral WM between 28 and 40 weeks' postconceptional age, we documented a pronounced increase in RA even though cerebral myelination had not yet begun.3 This observation indicated that increasing directionality of diffusion can be related to developmental events other than myelination. Anisotropic diffusion before the onset of myelination has been shown clearly in the developing rat brain.16 This so-called premyelination anisotropy seems to be related to some combination of premyelinative changes, including an increase in fiber diameter, axonal membrane changes, and enstheathment of axones by differentiating oligodendroglia.
Quantitative Measures at Term of Water Diffusion in Cerebral WM Regions in Premature Infants With Perinatal WM Injury
The most striking change in quantitative measures of diffusion at term among premature infants with perinatal WM injury, when compared with infants without WM injury, was a lower RA in the area of the previous injury, ie, central WM, and importantly in the underlying posterior limb of the internal capsule. This observation indicates less directionality of diffusion in both of these brain areas and suggests an abnormality in 1 or more of the developmental processes just discussed as the likely substrates for anisotropic diffusion. The neuropathology of cerebral WM injury of the premature infant involves both focal periventricular necrosis of all cellular elements and more diffuse but less severe surrounding WM injury.1 The more diffuse injury involves oligodendroglial destruction and perhaps axonal injury, without overt axonal loss.1 This more diffuse WM injury may occur without the focal necroses, and 5 of our 10 cases exhibited only diffuse injury by MRI.
The lower RA in the cerebral WM site of injury, ie, central WM, suggests that central fiber tracts were destroyed or their subsequent development was impaired or both. The RA measure cannot distinguish among these possibilities, because an impairment of axonal development or in oligodendroglial enstheathment or both could impair the acquisition of anisotropic diffusion. The vector imaging findings (see next section) suggest that both destructive events and impairment of WM development may have occurred.
Finding the lower RA in the posterior limb of the internal capsule that receives descending corticospinal tract fibers from the site of the central cerebral WM injury was unexpected. Overt axonal injury in the capsule is not a common neuropathologic feature of periventricular WM injury of the premature infant. The lower directionality of diffusion in the internal capsule suggests a disturbance in the development of these descending fibers. The vector imaging findings (see later) provide additional insight into this issue.
Imaging at Term of Microstructural Development of Cerebral WM in Premature Infants With or Without Perinatal WM Injury
To define further the disturbances of microstructural development suggested by the RA values just described, we generated diffusion vector maps of fiber orientation and density and tract size in cerebral WM regions and internal capsule. The central cerebral WM at term in premature infants with previous WM injury exhibited shorter vector lengths, more diffuse vector orientation, and therefore altered fiber orientation when compared with the findings in premature infants without WM injury. Moreover, vector distribution in the central WM also was characterized by fewer anteroposterior-oriented fiber bundles in the superior longitudinal fasciculus. The findings indicate major disturbances in fiber tract size and development in the area of previous WM injury. Importantly, vector distribution in the posterior limb of the internal capsule showed a strikingly reduced density when compared with the fiber tract bundles present in preterm infants without central WM injury. Such a disturbance could reflect Wallerian degeneration of the fibers secondary to central WM axonal destruction or involve maldevelopment of the axonal-oligodendroglial unit, such as impairment of fiber growth or of oligodendroglial enstheathment. The functional correlates of our findings will be important to define on long-term follow-up. It is noteworthy in this regard that the asymmetrical hypomyelination, previously reported by conventional MRI, in the posterior limb of the internal capsule in infants who had earlier sustained intraventricular hemorrhage and unilateral cerebral WM involvement has been associated with later hemiplegia.17
Studies of adult stroke patients18–20 have shown that functional abnormalities can be predicted from changes detected in diffusion tensor maps, which contain information about the integrity of fiber structures, information otherwise not obtainable by any other in vivo imaging technique. Neurologic follow-up of our infants will determine the prognostic value of this new technique. However, the potential value of diffusion tensor MRI in understanding the basis of the subsequent neurologic deficits and the likelihood of their occurrence seems substantial.
- Received March 8, 2000.
- Accepted July 26, 2000.
Reprint requests to (J.J.V.) Department of Neurology, Children's Hospital, Fegan 1103, 300 Longwood Ave, Boston, MA 02115.
- WM =
- white matter •
- MRI =
- magnetic resonance imaging •
- ADC =
- apparent diffusion coefficient •
- RA =
- relative anisotropy
- Welch RJ,
- Byrne P
- Mattiello J,
- Basser PJ,
- Le Bihan D
- Basser P,
- Pierpaoli C
- De Vries LS,
- Groenendaal F,
- van Haastert IC,
- Eken P,
- Rademacher KJ,
- Meiners LC
- Jones D,
- Lythgoe D,
- Horsfield M,
- Simmons A,
- Williams S,
- Markus H
- Werring D,
- Clark C,
- Barker G,
- et al.
- Copyright © 2001 American Academy of Pediatrics