ASD-related spatial covariance gray matter pattern revealed with a large-scale multi-center dataset
Poster No:
384
Submission Type:
Abstract Submission
Authors:
Shengzhi Ma1, Xing-Ke Wang2, Chen Yang3, Wen-Qiang Dong3, Qiu-Rong Zhang3, Yu-Feng ZANG4, Li-Xia Yuan5
Institutions:
1Hangzhou Normal University, Hangzhou, China, 2Beijing Normal University, Beijing, Beijing, 3Hangzhou Normal University, Hangzhou, Zhejiang Province, 4The Affiliated Hospital of Hangzhou Normal University, Hangzhou, Zhejiang Province, 5Zhejiang University, Hangzhou, Zhejiang Province
First Author:
Co-Author(s):
Introduction:
Autism Spectrum Disorder (ASD) is a severe neurodevelopmental disorder, and the underlying neuroanatomy mechanism of ASD remains unclear. Many researches have used gray matter (GM) volume to investigate structural abnormalities in ASD. Until now, the majority of GM studies have applied univariate analysis approaches, including region-of-interest-wise and voxel-wise analysis (Riddle et al., 2017; Wang et al., 2022). The scaled subprofile model of principal component analysis (SSM-PCA) is a multivariate method for exploring the disease related pattern based on the spatial covariance across different brain regions and able to identify subtle changes caused by the disease (Alexander, 1994; Yuan et al., 2018). This study aimed to obtain the ASD-related GM volume pattern with SSM-PCA to reveal the neuroanatomical mechanisms of ASD.
Methods:
We utilized the Autism Brain Imaging Data Exchange (ABIDE), an open-access dataset for ASD, and performed preprocessing on the T1-weighted structural MRI images (sMRI). Anatomical segmentation was performed on sMRI from ABIDE II dataset to get the GM, and then the GM volume of each brain region defined by the automated anatomical labeling 3 (Rolls et al., 2020) template was computed by Computational Anatomy Toolbox 12. The SSM-PCA algorithm was then used for the GM volume matrix of subject-brain region to obtain the spatial covariance GM patterns (Wang et al., 2022). The sign of ASD-related GM pattern is determined by keeping higher average expression in the ASD group compared to the typically developed (TD) group. Within the 15 patterns explaining the top higher proportion of variance, two sample t-test was conducted for the expression of each GM pattern to identify those displaying significant inter-group expression differences. Then, the ASD-related GM pattern was obtained by the linear combination of these patterns. We further investigated the relationship between the ASD-related GM pattern and clinical symptoms. Next, we verified the reproducibility of the ASD-related pattern by projecting it on to ABIDE I dataset and checked the expression difference between ASD and TD and its relationship with clinical scores. In addition, we divided the cohort into three age groups, namely primary school students with [7, 12) years old, adolescents with [12, 18) years old, and adults with [18, 28) years old, to explore the influence of age on the ASD-related pattern.
Results:
The expressions of the second and fourteenth pattern showed significant group difference between ASD and TD, which were linearly combined to form the ASD-related GM pattern. The pattern included thalamus, right parahippocampal gyrus, left locus coeruleus, basal ganglia, and cerebellum, which mainly concerned with cognitive functions of visual imagery, auditory, theory of mind, and Îperception. Furthermore, the expression of this patterns is correlated with scores of Social Response Scale (SRS, r = 0.18, p = 3.4×10-4) and Social Communication Questionnaire (SCQ, r = 0.17, p = 7×10-4). For reproducibility validation, the expression of the ASD-related pattern in ABIDE I revealed significantly difference in the ASD relative to the TD (Cohen's d = 0.21, p = 0.016), which is also positively correlated with the scores of SRS (r = 0.20, p = 0.0015). For age effect, the primary school group (Cohen's d = 0.37, p = 0.0013) and adult group (Cohen's d = 0.44, p = 0.038) showed significant differences, while the adolescent group illustrated no significant difference (Cohen's d = 0.21, p = 0.19) of the expression of the ASD-related pattern between the ASD and TD.
Conclusions:
We captured a reproducible ASD-related GM pattern, which is related to the social deficit, and revealed its heterogeneity among different age groups. Our findings facilitate the understanding of the underlying neural mechanisms of ASD and provide therapeutic targets for individuals with ASD.
Disorders of the Nervous System:
Neurodevelopmental/ Early Life (eg. ADHD, autism) 1
Modeling and Analysis Methods:
Multivariate Approaches 2
Neuroanatomy, Physiology, Metabolism and Neurotransmission:
Cortical Anatomy and Brain Mapping
Novel Imaging Acquisition Methods:
Anatomical MRI
Keywords:
Autism
STRUCTURAL MRI
Other - autism spectrum disorder (ASD); gray matter (GM) volume; scaled subprofile model of principal component analysis (SSM-PCA); Autism Brain Imaging Data Exchange (ABIDE)
1|2Indicates the priority used for review
Provide references using author date format
Riddle, K. (2017). Brain structure in autism: A voxel-based morphometry analysis of the Autism Brain Imaging Database Exchange (ABIDE). Brain Imaging and Behavior, 11(2), 541–551.
Rolls, E. (2020). Automated anatomical labelling atlas 3. NeuroImage, 206, 116189.
Wang, H. (2022). Developmental brain structural atypicalities in autism: A voxel-based morphometry analysis. Child and Adolescent Psychiatry and Mental Health, 16, 7.
Wang, X.-K. (2022). Gray Matter Network Associated With Attention in Children With Attention Deficit Hyperactivity Disorder. Frontiers in Psychiatry, 13, 922720.
Yuan, L.-X. (2018). Intra- and Inter-scanner Reliability of Scaled Subprofile Model of Principal Component Analysis on ALFF in Resting-State fMRI Under Eyes Open and Closed Conditions. Frontiers in Neuroscience, 12, 311.