Automated Stratification of Cortical Network Reveals Clinically Meaningful Subgroups of ASD children
Poster No:
336
Submission Type:
Abstract Submission
Authors:
Ahmad Mheich1, Sahar Yassine2, Joana Maria Almeida Osório1, Sonia Richetin1, Vincent Junod1, Laura Mendes1, Katherina Gschwend1, Victoria Aeschbach1, Lorène Arnold1, David Romascano1, Paola Yu1, Marine Jequier Gygax1, Anne Maillard1, Mahmoud Hassan3,4, Nadia Chabane1, Borja Rodríguez-Herreros1
Institutions:
1CHUV Lausanne, Lausanne, Vaud, 2University of Oxford, Oxford, United Kingdom, 3MINDIG, Rennes F-35000, Bretagne, 4School of Science and Engineering, Reykjavik University, Reykjavik, Iceland
First Author:
Co-Author(s):
Mahmoud Hassan
MINDIG|School of Science and Engineering, Reykjavik University
Rennes F-35000, Bretagne|Reykjavik, Iceland
MINDIG|School of Science and Engineering, Reykjavik University
Rennes F-35000, Bretagne|Reykjavik, Iceland
Introduction:
Background
Heterogeneity in the causes and phenotypic presentation of autism spectrum disorder (ASD) poses a major challenge to clinical and translational research. Attempts to stratify individuals with ASD have been based primarily on behavioral criteria1, but clinical subtyping is blind to the underlying neurobiological mechanisms and has limited predictive value of the forthcoming developmental path. Yet, it is still unclear whether and how atypical brain functional connectivity accounts for individual differences across ASD-related symptomatology and behaviors.
Objectives
The goal of the study was to identify clinically meaningful subgroups of young children with ASD based on distinctive patterns of functional brain topology, to better understand of the neural substrates underlying ASD heterogeneity.
Heterogeneity in the causes and phenotypic presentation of autism spectrum disorder (ASD) poses a major challenge to clinical and translational research. Attempts to stratify individuals with ASD have been based primarily on behavioral criteria1, but clinical subtyping is blind to the underlying neurobiological mechanisms and has limited predictive value of the forthcoming developmental path. Yet, it is still unclear whether and how atypical brain functional connectivity accounts for individual differences across ASD-related symptomatology and behaviors.
Objectives
The goal of the study was to identify clinically meaningful subgroups of young children with ASD based on distinctive patterns of functional brain topology, to better understand of the neural substrates underlying ASD heterogeneity.
Methods:
We collected resting-state EEG data on 58 children with ASD aged 2-8 years to explore differences in functional brain network topology. We performed an unsupervised clustering analysis based on cortical network connectivity, using data-driven similarity network fusion and source-based spectral analysis2. We replicated the analysis in two independent samples of ASD participants from the NDA repository.
Results:
Results
We identified three subgroups of ASD children with distinct cortical network properties mainly mapped in the temporal and precentral cortices for the delta band, and in the middle frontal cortex for beta and gamma bands. These three clustered dimensions of functional connectivity and the associated ASD subgroups exhibited different clinical symptom profiles, and were reproducible in two independent samples.
We identified three subgroups of ASD children with distinct cortical network properties mainly mapped in the temporal and precentral cortices for the delta band, and in the middle frontal cortex for beta and gamma bands. These three clustered dimensions of functional connectivity and the associated ASD subgroups exhibited different clinical symptom profiles, and were reproducible in two independent samples.
Conclusions:
Conclusions
Our findings shed light on atypical brain network topology conferring risk for specific phenotypic manifestations of ASD, which may implicate unique underlying neurobiological mechanisms. Cross-validation stability hints at a solid stratification model to challenge ASD heterogeneity. Collectively, the stratification of well-defined neural signatures that give rise to the clinical heterogeneity of ASD has potential to provide more accurate prognosis and help to select the optimal therapeutic intervention strategy.
Our findings shed light on atypical brain network topology conferring risk for specific phenotypic manifestations of ASD, which may implicate unique underlying neurobiological mechanisms. Cross-validation stability hints at a solid stratification model to challenge ASD heterogeneity. Collectively, the stratification of well-defined neural signatures that give rise to the clinical heterogeneity of ASD has potential to provide more accurate prognosis and help to select the optimal therapeutic intervention strategy.
Disorders of the Nervous System:
Neurodevelopmental/ Early Life (eg. ADHD, autism) 1
Modeling and Analysis Methods:
Classification and Predictive Modeling
Connectivity (eg. functional, effective, structural)
EEG/MEG Modeling and Analysis 2
Keywords:
Autism
Computational Neuroscience
Data analysis
Development
Electroencephaolography (EEG)
Psychiatric Disorders
Statistical Methods
1|2Indicates the priority used for review
Provide references using author date format
2. Wang, Bo, Aziz M. Mezlini, Feyyaz Demir, Marc Fiume, Zhuowen Tu, Michael Brudno, Benjamin Haibe-Kains, et Anna Goldenberg. 2014. « Similarity Network Fusion for Aggregating Data Types on a Genomic Scale ». Nature Methods 11 (3): 333‑37. https://doi.org/10.1038/nmeth.2810.