Abstract
Introduction
Obstructive sleep apnea (OSA) is a sleep disorder caused by repetitive obstruction of the upper airway, resulting in hypopnea (reduced airflow during sleep) or apnea (complete airflow cessation during sleep) [1]. It leads to increased respiratory effort, sleep fragmentation, and alterations of gas exchange and is associated with numerous sequelae, primarily involving the cardiovascular, metabolic, and neurocognitive systems [2]. Although there are many features shared by adult and pediatric OSA patients, it has been recognized that the disease prevalence, pathopysiology and treatment of pediatric OSA are quite different from adult OSA [3–5], indicating that more studies should be conducted to confirm the influence of adult OSA’s risk factors in children. For example, regional deficiencies in brain areas were recently reported to have significant impact on OSA syndrome in adults, as the brain mediates breathing during sleep [6]. Particularly, structural alterations have been detected in the brain of adult patients with OSA, with the gray matter loss in cortex (frontal, temporal and parietal lobe), basal ganglia (hippocampus, thalamus, caudate and putamen), and cerebellar regions, based on voxel-based morphometry (VBM) analysis [6–11]. In addition, with the help of fiber integrity analysis based on diffusion tensor imaging (DTI), white matter integrity were found to be affected in OSA adults within multiple brain sites [12, 13]. Children with OSA may also involve the aforementioned structural changes in the brain, but research on this topic is still lacking.
In such background, we proposed a study to investigate the differences in brain structure between children with OSA and healthy age- and gender-matched controls. In particular, we aimed to examine the pediatric patients for evidence of shape changes in the subcortical structures of basal ganglia. A surface-based shape analysis was performed to compare the brain morphology between children with OSA and the normal controls, especially in lateral thalamus, hippocampus, amygdala, caudate, putamen, and pallidum.
Materials and methods
Subjects
We recruited 25 children (aged 10.3±1.5 years, 20 boys) with OSA and 30 age-matched children (aged 10.1±1.8 years, 24 boys) without OSA. All of these children were invited to undergo overnight polysomnography (PSG) at a dedicated sleep laboratory with CNS 1000P polygraph (CNS, Inc., Chanhassen MN) [14]. The computerized sleep data were further manually edited by experienced polysomnography technologists and clinicians according to standardized criteria [15, 16]. An obstructive apnoea was defined as the absence of airflow with persistent respiratory effort lasting longer than two baseline breaths, irrespective of arterial oxygen saturation changes. An obstructive hypopnoea was defined as a reduction of 50% or more in the amplitude of the airflow signal with persistent respiratory effort. It was only quantified if it was longer than two baseline breaths and was associated with oxygen desaturation of at least 3% and/or arousals [17]. The deficit in OSA was measured by the obstructive apnea hypopnoea index (OAHI), which is identified as the total number of obstructive apnoeas and obstructive hypopnoeas per hour of total sleep.
Subjects were classified as healthy control group (OAHI <1 and history of snoring <3 nights per week, n = 30), mild OSA (OAHI 1 to 5, n = 12), and moderate-to-severe OSA (OAHI ≥5, n = 13). Subjects with primary snoring (OAHI <1 and history of snoring 3 nights or more per week) were not included as there was a potential of misclassifying subjects with upper airway resistance syndrome (UARS) as primary snoring [17]. In the study cohort, the OAHI was 8.09±7.78 for the OSA group, and 0.06±0.23 for the control group. In addition, the body max index (BMI) was well-matched between of the OSA group (19.92±3.89) and the control group (20.03±3.89).
MRI acquisition
The recruited subjects were examined using a 1.5T MRI scanner (Sonata, Siemens, Erlanger, Germany). High resolution iso-voxel T1-weighted images covering the whole brain were obtained using the magnetization prepared rapid acquisition gradient echo (MPRAGE) sequence with the following parameters: TR = 2070 ms, TE = 3.9 ms, TI = 1110 ms, flip angle = 15°, field of view = 230×230 mm, slice thickness = 0.9 mm (no gap), matrix = 256×256. The total scanning time for each subject is about 8 minutes.
Shape analysis
Automated segmentation and surface-based shape analysis were performed to compare the subcortical structures especially in basal ganglia between OSA group and control group, using the FMRIB’s Integrated Registration and Segmentation Tool (FIRST) [18] developed by Oxford Centre for Functional MRI of the Brain (FMRIB). The shape and appearance models used in FIRST were constructed from manually segmented images provided by the Center for Morphometric Analysis (CMA), MGH, Boston. FIRST utilizes deformable surfaces to automatically parameterize the volumetric labels in terms of meshes, and the deformable surfaces preserves vertex correspondence across the training data. Vertex correspondence determined by FIRST facilitates the investigation of localized shape differences through the examination of group differences in the spatial location of each vertex [19].
The T1-weighted images of the subjects were first registered to the nonlinear MNI152 template (T1 standard brain averaged over 152 subjects; Montreal Neurological Institute, Montreal, QC, Canada) [20] using a 12 DOF affine registration. The second registration was a 12 DOF registration with a subcortical mask to exclude voxels outside subcortical regions. Pose (rotation and translation) was removed by minimizing the sum-of-squares difference between the corresponding vertices of an individual’s surface and the mean surface [21]. The vertices between the two groups were compared in MNI152 space with F-statistics, and the statistics were rendered on the shape surface, providing a map of the regions where there were significant displacements of the mean vertex location between groups. The aforementioned analyses were applied to investigate local shape differences of all the basal ganglia structures (including lateral thalamus, hippocampus, amygdala, caudate, putamen, and pallidum) between the OSA group and the control group.
Permutation inference was performed using randomize in FIRST [22]. Threshold-free cluster enhancement (TFCE) based multiple comparison correction was applied to acquire corrected p-values with f-test results for each vertex of the segmented basal ganglia structures. The clusters with significant regional shape difference (p < 0.05) were rendered on the surface map of the test structures.
Results
TFCE based multiple comparison correction was performed 1000 times to generate the vertex-wise statistics in shape difference of the 12 ROIs (both sides of thalamus, hippocampus, amygdala, caudate, putamen, and pallidum) between the two groups. Only the left thalamus (minimum corrected p = 0.023) and the left pallidum (minimum corrected p = 0.003) presented statistical significance at the level of p < 0.05 (Table 1). Supplementary analyses were made with higher permutation times, but the results did not varied significantly from this analysis with 1000 permutations.
To determine the direction of the shape changes, the visualization of shape difference generated from fslview (a built-in tool in FIRST software that helps visualize the results of shape difference) was improved, based on the corrected vertex-wise F statistics and the sign of shape difference among individuals. The distribution map of shape change (Fig. 1) showed significant difference in the ventral posterior (VP) nucleus and medial dorsal (MD) nucleus of the left thalamus (Fig. 1A), and both internal and external segments of the left palladium (Fig. 1B). The vertices with significant displacement were labeled in yellow to red on the light blue mask of the left thalamus and pallidum, with p-values from 0.05 to 0, and the direction vector points from the control group to the OSA group. The VP and MD nucleus of the left thalamus presented atrophy while both internal and external segments of the left pallidum dilated in the OSA group compared to the control group.
Discussion
We performed a surface-based shape analysis on the basal ganglia structures to associate brain morphometry and OSA syndrome in children. To our knowledge, it was the first study to investigate the brain structural alterations of pediatric OSA with MRI using shape analysis. We demonstrated that when compared with controls, there were significant shape changes in the VP and MD nucleus of the left thalamus, and both internal and external segments of the left pallidum in the brains of children with OSA.
Many studies have reported the functional deficits [23, 24] and structural alterations (in gray matter loss [7, 10] and deficits of tissue integrity [12, 13]) of the thalamus in adult patients with OSA, supported by the fact that thalamus plays a fundamental role in the arousal mechanism and sleep–wake transition process [23]. Our results further revealed the thalamic atrophy within the VP and MD nucleus in children with OSA, indicating that the structural deficits of the thalamus were shared by adult and pediatric OSA syndrome. The VP and MD nucleus were reported to facilitate the coordination and planning of movement and target efferents including the motor cortex, premotor cortex, and supplementary motor cortex [24]. These two nucleus, as part of the thalamus, receive information from the cerebellum. Then they transform information to the premotor cortex and supplementary motor cortex, which are important substructures in respiratory regulation. In this way, the alterations of these two thalamic nucleus may contribute to OSA syndrome due to the deficit caused in respiratory regulation. The shared structural alterations in the left thalamus of OSA children and adults further confirmed the significant role of thalamus in OSA syndrome. In addition, the atrophy of the VP and MD nucleus found in the left thalamus of OSA children facilitates to understand the function of thalamic nucleus in the complex circuit causing pediatric OSA.
There were two studies that respectively reported the functional deficits [24] and gray matter loss [10] in the pallidum of adult patients with OSA. In addition, the role of the pallidum has been recognized to control the voluntary movement [25] and the movement of respiratory musculature [26], which are probably involved in OSA syndrome. The present study confirmed the association between the structural deficits of pallidum and pediatric OSA, and for the first time we refined this alteration into microstructures of pallidum (the dilation of the internal and external segments) with the surface-based shape analysis. In the cortical-basal ganglia-thalamic-cortical loop, the pallidum receives input information from the striatum (including the caudate nucleus and the putamen), and deliver the data to the thalamus, either directly or indirectly. The internal segment of the pallidum feeds directly to the thalamus, while the external segment delivers information to the internal segment and then to the thalamus. In this case, the information delivery in this loop would be affected if there is a deficit in the internal and external segments of pallidum, as well as in the thalamus. Moreover, the pattern of structural change in pallidum might be different between pediatric OSA and adult OSA. The gray matter loss, or the regional atrophy, was identified in a previous study in adult patients with OSA, while in our study regional dilation was revealed in the pallidum for pediatric OSA. This unique pattern might result from the fact that pallidum is still in development during childhood [27]. However, further studies are essential to validate our hypothesis by carefully comparing the structural changes of pallidum between adult and pediatric OSA with larger sample size and broader age range.
A potential limitation of this study is that the recruited subjects are predominately boys in our study, and our results might fail to reflect the corresponding gender-specific brain structural changes in female pediatric OSA. In fact, we had no preference in gender during recruitment of the subjects, and the larger number of boys in our cohort might result from the fact that pediatric OSA is more common in males than females [28]. The small number of moderate-to-severe OSA subjects is another limitation of this cohort, as more subjects with severe OSA would probably contribute to higher sensitivity in detection of structural alterations in the brain. However, this is an intrinsic factor that few children suffer from more severe OSA. In addition, although we applied a widely-used surface-based shape analysis in the brain structures, considerations should be taken when comparing our results with the other studies when using very different MRI scanning protocols, since some of the scanning parameters (e.g. field strength, slice thickness) might influence the sensitivity in detection of shape difference.
Conclusion
This study revealed the structural changes of the thalamus and pallidum between the OSA children and the healthy children. The surface-based shape analysis identified the atrophy in VP and MD nucleus of the left thalamus in OSA children, which provided further evidence to shared structural changes of the thalamus in pediatric OSA and adult OSA. In addition, we demonstrated the dilation of both internal and external segments of the left pallidum in OSA children, and revealed the distinctive structural change pattern of pallidum in pediatric OSA compared to adult OSA. These findings contributed to our knowledge about the shared and distinct structural alterations within the basal ganglia in children and adults with OSA.
Footnotes
Acknowledgments
The work described in this paper was supported by grants from the Research Grants Council of the Hong Kong Special Administrative Region, China (Project No.: CUHK 473012, CUHK 416712, CUHK 14113214), a grant from The Science, Technology and Innovation Commission of Shenzhen Municipality (No. CXZZ20140606164105361), a grant from the National Natural Science Foundation of China (Project No. 81271653), and a grant from Lui Che Woo Foundation.