Effect of antenatal growth on brain white matter maturation in preterm infants at term using tract-based spatial statistics
- First Online:
- Cite this article as:
- Lepomäki, V., Matomäki, J., Lapinleimu, H. et al. Pediatr Radiol (2013) 43: 80. doi:10.1007/s00247-012-2509-9
- 436 Views
White matter maturation of infants can be studied using diffusion tensor imaging (DTI). DTI of the white matter of the infant brain provides the best available clinical measures of brain tissue organisation and integrity.
The purpose of this study was to compare white matter maturation between preterm infants born small for gestational age (SGA) and preterms with weight appropriate for gestational age (AGA) at birth.
Materials and methods
A total of 36 preterm infants were enrolled in the study (SGA, n = 9). A rater-independent method called tract-based spatial statistics (TBSS) was used to assess white matter maturation.
When measured by TBSS, the AGA infants showed higher fractional anisotrophy values in several white matter tracts than the SGA infants. Areas with significant differences included anterior thalamic radiation, corticospinal tract, forceps major and minor, inferior fronto-occipital fasciculus, inferior longitudinal fasciculus, superior longitudinal fasciculus, uncinate fasciculus, and superior longitudinal fasciculus (temporal part). No significant difference was found for mean diffusivity.
As an objective and user-independent method, TBSS confirmed that preterm infants with impaired antenatal growth have impaired white matter maturation compared to preterm infants with normal antenatal growth. The differences were mainly detected in radiations that are myelinated first.
KeywordsDiffusion tensor imagingTract-based spatial statisticsBrainPreterm infant
Diffusion tensor imaging (DTI) provides quantitative measures of the diffusion of water. The fact that water diffusivity reflects processes related to brain maturation such as myelination, axonal integrity, and the general organisation and alignment of groups of axons in white matter tissue means that DTI can be used to study white matter maturation [1–6]. These quantitative measures can even be analysed for the pre-myelinated neonatal brain .
The diffusion parameters that are often used to characterise white matter maturation are fractional anisotropy (FA) and mean diffusivity (MD). The methods typically used to analyse DT images are classic region-of-interest (ROI) analysis, voxel-based morphometry (VBM), and TBSS. ROI analysis is a feasible method for FA and MD measurement in known white matter areas or tracts of infant brain in clinical praxis. However, ROI analysis is time-consuming and requires a priori knowledge about white matter tracts that should be studied. The ROI size and shape are known to affect the FA and MD measurement results. TBSS and VBM are observer-independent computer-based methods and analysis of the whole brain is easily performed.
Normal white matter maturation is characterised by increasing FA and decreasing MD values [8–10]. However, brain maturation is a complex process and disturbance in these processes could affect the white matter pathway [11, 12]. Therefore, it is clear that DTI does not only measure one single characteristic of white matter development or integrity. Instead, DTI reflects underlying aspects of tissue organisation at numerous levels . Changes in FA values are generally associated with changes in microstructure of directional organisation of white matter, and changes in MD are associated with changes in cell density and volume of the extracellular space of the white matter [13, 14].
The published results regarding the effect of prematurity on white matter maturation have been contradictory. Huppi et al.  showed that preterm infants have delayed white matter maturation compared to infants born at term, using ROI method in DTI analysis. Hasegawa et al.  and Thompson et al.  showed that the degree of prematurity affects the white matter maturation at term; they used tractography-based ROI selection. In both studies, very premature infants showed more immature white matter in the corpus callosum than infants born at later GA. Also, Anjari et al.  and Rose et al.  reported an association between premature birth and more immature white matter compared to infants born at term using tract-based spatial statistics (TBSS). On the other hand, a recently published work by Bonifacio et al.  showed that extreme premature birth is not associated with impaired development of brain microstructure. This work was performed using manual ROI selection in DT image analysis.
SGA infants are either constitutionally small or have had impaired intrauterine growth (intrauterine-growth-restriction-infants or IUGR-infants). IUGR infants are reported to have more immature cortical microstructure and reduced volumes in cerebral cortical GM and occipital volumes [20–22]. IUGR infants have shown higher ADC values in the internal capsule, which indicates more immature white matter than the term controls . Using ROI methodology, an earlier study showed that poor antenatal growth is associated with delayed maturation in the corpus callosum .
The aim of this study was to compare whole-brain white matter maturation using the TBSS method between preterm infants who are born small for gestational age (SGA) and preterm infants whose weight was appropriate for gestational age (AGA). Gestational age (GA) was used as a confounding covariate in the regression analysis.
Materials and methods
This study is part of the Development and Functioning in Very Low Birth Weight Infants from Infancy to School Age (PIPARI) study conducted at Turku University Hospital.
Infants were selected according to the entry criteria of the Vermont-Oxford Network database and local decision (now also the Finnish national register criteria).
Demographic and clinical characteristics of the infants
AGA infants (n = 27)
SGA infants (n = 9)
GA at MRI (weeks), mean ± SD
39.9 ± 0.4
40.0 ± 0.2
Birth weight (g), mean ± SD
1,481 ± 308
1,294 ± 215
Birth weight z-score, mean ± SD
−0.44 ± 1.12
−2.61 ± 0.75
Head circumference, z-score mean ± SD
0.11 ± 1.45
−1.74 ± 1.01
Maternal age (years), mean ± SD
32 ± 5
30 ± 4
GA at birth (weeks), mean ± SD
30.0 ± 1.8
31.6 ± 1.7
Apgar score (5 min), median [lower quartile, upper quartile]
8 [7, 9]
8 [8, 9]
Antenatal steroids (no/yes)
Postnatal steroids (no/yes)
Umbilical artery pH ≥ 7.00 a
Cesarean section (no/yes)
Absent or reversed flow (AREDF) in the umbilical artery (UA) (no/yes)b
The weight of the infants was measured at birth and compared to the expected average weight of infants of the same gender born at the same gestational age; the difference was expressed as standard deviations from the mean of gender- and age-specific Finnish national growth charts (birth weight z-score). If birth weight z-score was less than 2 SDs from the mean, the infant was classified as a small-for-gestational-age (SGA) infant.
Study protocol was approved by the Ethics Review Committee of the Hospital District of the South-West Finland. All families provided informed consent.
MR examination and data analysis
The MR imaging was performed with a 1.5-T MR system (Gyroscan Intera CV Nova Dual; Philips Healthcare, Best, the Netherlands) with a SENSE head coil. MR imaging was conducted during postprandial sleep at term without any pharmacological sedation. The infants were swaddled to calm them and to reduce movement artifacts in the images. A pulse oximeter was routinely used during MR examination. A physician attended the examination if necessary to monitor the infant. Ear protection was used (3M Disposable Ear Plugs 1100; 3M, Brazil and Wurth Hearing protector art. Nr. 899 3000 232; Wurth, Austria).
The sequence used for diffusion-weighted imaging was a single-shot echo planar imaging (EPI) with SENSE. SENSE reduction was 2. The slice thickness was 5 mm with a 1-mm gap between slices. A 200x200-mm field of view (FOV) was used. Imaging matrix was 111 × 89 and the reconstructed voxel size was 0.78 mm x 0.78 mm. The number of signal averages was 2 and the EPI factor was 47. Repetition time (TR) was the shortest (minimum 2,264 ms) and echo time (TE) was 68 ms. The b values were 0, 600 and 1,200 s/mm2 with 15 directions. Fat suppression was achieved using spectral presaturation with inversion recovery (SPIR). In addition to the diffusion tensor data set, imaging protocol included conventional T1, T2 and FLAIR images.
Data processing was performed using FMRIB Software Library (FSL) 4.1.7 . Data were corrected for eddy currents and motion by registering data to b = 0 images. FA and MD maps were calculated using FMRIB Diffusion Toolbox (FDT), which is part of the FSL programme. Tract-based spatial statistics (TBSS) was performed by first co-registered (non-linear registration) images to the study-specified target . Registration was confirmed using visual inspection. The target was transformed into the Montréal Neurological Institute (MNI) 152 space. All transformed FA images were averaged to create a mean FA image. Taking the centre of the white matter tracts, a FA skeleton was created. A threshold value of 0.20 was used for FA when creating the skeleton. The values of each subject’s FA maps were projected onto the skeleton by searching the local maxima along the perpendicular direction from the skeleton. We also performed the voxel level analysis to MD images. Voxelwise analysis was performed using the randomise tool. Gestational age was used as a confounding covariate in the analysis. Also, correlation between gestational age and FA and MD was studied with a separate voxel level analysis. Threshold-free cluster enhancement (TFCE) was used in the analysis. TFCE is used to enhance cluster-like structures without having to define an initial cluster-forming threshold or carry out a large amount of data smoothing . Decided contrasts were tested using 5,000 permutations. Differences with P < 0.05 (corrected for multiple comparisons) were considered statistically significant. Areas were identified using the JHU White-Matter Tractography Atlas.
The MD values of the brain white matter did not differ between the AGA and SGA infants. No significant correlations were found between GA and FA or MD values.
Using TBSS, we found that infants with impaired antenatal growth have impaired white matter maturation compared to infants with normal antenatal growth. Lower FA values were found in SGA infants than in AGA infants when GA was used as a confounding covariate. No significant difference was observed for MD values between the study groups.
No significant correlation was found between GA and white matter maturation at term. This was unexpected as previous studies of very preterm infants had shown more immature white matter compared to infants born at later gestation [16, 17]. However, our results are in line with those published by Bonifacio et al. , who showed that extreme premature birth is not associated with impaired development of brain microstructure using ROI analysis.
In a previous ROI analysis, it was shown that the difference in maturation is specific only in the corpus callosum . Using the TBSS method, the present study showed that white matter maturation was delayed in several additional white matter areas of brain between the AGA and SGA infants. This is not that surprising considering that the ROI analysis is limited to certain parts of the white matter tracts, whereas TBSS analysis covers the whole brain. The partial volume effect caused by the positioning and size of the manually placed ROI on the tracts could also have masked the difference of white matter tracts between the AGA and SGA infants in the ROI-based analysis.
Our finding on FA values is in line with those reported earlier using the ROI measurement method. In the previous study, FA values correlated positively with antenatal growth in the genu and splenium of corpus callosum . With TBSS, lower FA values were found in SGA children at term age than in AGA children. The differences were mainly detected in radiations among those that are myelinated first. The differences between groups could have been caused by impaired white matter maturation in the SGA group.
In the previous study, MD values correlated negatively with antenatal growth in the splenium of corpus callosum . In the TBSS non-significant difference was found between AGA and SGA for MD values. It is unclear why MD showed a non-significant difference between groups in TBSS and a significant difference in the ROI analysis. One possible explanation is that ROI-based analysis is more sensitive than TBSS to non local and diffuse changes in the brain maturation process.
The present study has certain limitations. Infants born at early gestational weeks are known to have significantly more brain pathologies than infants at term age. In order to provide a more accurate picture of the preterm infant population, infants with major brain abnormalities should be included. However, severe brain pathologies bring technical problems in terms of the reliability of the registration process of TBSS. Another limitation was that the number of SGA infants eligible for this study was quite small. It is also possible that image resolution, especially slice thickness, had an effect on the results.
It was not possible in this study to separate constitutionally small SGA infants from IUGR infants. There are several possible reasons for intrauterine growth restriction such as placental insufficiency related to pre-eclampsia and nicotine-induced vasoconstriction of the uteroplacental vessels. Intrauterine growth, typically restriction, affects 3–10% of all pregnancies . In our study population, a third of the mothers of SGA infants had mild or severe pre-eclampsia, which is a known cause for growth restriction in utero. Also, seven of the nine infants showed asymmetrical growth, which is typical in pre-eclampsia. Therefore, it is likely that a large proportion of the infants in the study were small due to impaired antenatal growth. For future studies, separating IUGR infants from constitutionally small infants would further enlighten the mechanisms behind white matter maturation. The correlation between imaging findings and outcome should also be studied.
Future studies are needed to show prognostic value of DT measurements on child outcome. It would be interesting to see whether specific functional tests correlate with changes in physiologically plausible white matter tracts. Only a few studies to date have correlated diffusion values to clinical outcome. For example, correlation between DTI data of auditory pathways and auditory BAEP measurements has been reported . Recently, Van Kooij et al.  reported associations between white matter microstructure and cognitive, fine-motor and gross-motor performance in preterm infants at 2-year corrected age. However, correlating DTI measures to outcome is challenging because outcome measures are quite general and there are several confounding factors, as brain maturation is a rather complex process that includes multiple processes (including germinal matrix distribution, cortical folding, development of the cortical projections to cortical and deep brain structures, and glial cell migration).
The results of this study show that TBSS can be used as an objective, user-independent method of confirming that antenatal growth affects the white matter maturation. These results increase our understanding of white matter maturation and the influence of antenatal growth.