Systemic heterogeneity in autism spectrum disorder revealed by individualized structural covariance network analysis

Qiuyue Zhang; Xi Yang; Jianfeng Qiu; Weizhao Lu

doi:10.1016/j.pnpbp.2024.111224

Systemic heterogeneity in autism spectrum disorder revealed by individualized structural covariance network analysis

Prog Neuropsychopharmacol Biol Psychiatry. 2024 Dec 18:136:111224. doi: 10.1016/j.pnpbp.2024.111224. Online ahead of print.

Authors

Qiuyue Zhang¹, Xi Yang², Jianfeng Qiu³, Weizhao Lu⁴

Affiliations

¹ Department of Radiology, The Second Affiliated Hospital of Shandong First Medical University, Taian 271016, China.
² Department of Mathematics & Statistics, Graduate School of Arts & Sciences Boston University, Boston, MA, United States of America.
³ School of Radiology, Shandong First Medical University & Shandong Academy of Medical Sciences, Taian 271016, China.
⁴ Department of Radiology, The Second Affiliated Hospital of Shandong First Medical University, Taian 271016, China. Electronic address: mingming9053@163.com.

PMID: 39706545
DOI: 10.1016/j.pnpbp.2024.111224

Abstract

Background and purpose: Autism spectrum disorder (ASD) is clinically heterogeneous, and resent neuroimaging studies have shown the presence of brain structural heterogeneity in ASD. However, there is currently a lack of evidence for systemic level brain structural heterogeneity. This study aimed to reveal the heterogeneity of brain structural changes at the systemic level in ASD patients through individual differential structural covariance network (IDSCN) analysis.

Materials and methods: We included 803 neurotypical controls (NCs) and 650 ASD patients from 24 sites and the corresponding structural magnetic resonance and clinical data. 516 ASD patients were used as training group, and 134 ASD patients were selected as an independent validation group. In the training group, we constructed IDSCN for each ASD patient, identified differentiated structural covariance edges, and resolved systemic heterogeneity using K-means clustering algorithm. We then conducted statistical analyses on the demographical and clinical data of the ASD subgroups, and performed correlation analyses between structural covariance edges and clinical profiles within each ASD subgroup. We also tested the reliability of the IDSCN in the validation group.

Results: The results of the training and validation groups were similar, revealing two subtypes of ASD, with 17 brain connections showing differences between the two subtypes. There were differences in clinical profiles between the two subgroups in restricted repetitive behavior score from autism diagnostic interview-revised (ADI_RRB_TOTAL), total score of autism diagnostic observation schedule (ADOS), social interaction score from ADOS and stereotyped behaviors score from ADOS. In both datasets, we also found a significant correlation between ADI_RRB_TOTAL scale and the Z-score for edges between the bilateral ventral tegmental area (VTA) and between the left VTA and right substantia nigra pars compacta.

Conclusion: ASD exhibited brain structural heterogeneity at the systemic level, mainly involving nucleus in the subcortical regions and brainstem, which can affect RRB in ASD patients. The two subtypes discovered in this study have the potential to be applied in precise diagnosis and treatment of the disorder.

Keywords: Autism spectrum disorder; Individual differential structural covariance network; Structural magnetic resonance imaging; Systemic heterogeneity.