Subtyping Autism Spectrum Disorder With a Population Graph-Based Dual Autoencoder: Revealing Two Distinct Biotypes

Area of Science:

• Neuroscience
• Genetics
• Psychiatry

Background:

• Autism spectrum disorder (ASD) presents significant heterogeneity in clinical symptoms and neurobiology.
• Understanding this heterogeneity is crucial for developing targeted interventions.

Purpose of the Study:

• To identify distinct ASD biotypes using a novel graph-based subtyping approach.
• To uncover the neurobiological underpinnings of these identified ASD biotypes.

Main Methods:

• Analysis of resting-state fMRI and clinical data from 443 males with ASD.
• Utilized a population graph-based dual autoencoder for subtyping (PG-DAS) framework.
• Compared clinical scores and functional connectivity patterns between biotypes; examined brain-behavior associations.

Main Results:

• Identified two distinct ASD biotypes with differing clinical scores and network connectivity.
• ASD1 showed reduced network integration linked to communication symptoms; ASD2 exhibited greater network segregation.
• Biotype-specific brain-behavior associations were revealed, correlating with social and communication scores.

Conclusions:

• Biotype-specific brain network patterns are critical for understanding ASD heterogeneity.
• The PG-DAS framework is effective for ASD subtyping and applicable to other neuroheterogeneous disorders.