Identification and multidimensional characterization of two distinct autism spectrum disorder subtypes

Autism spectrum disorder (ASD) shows high variability in brain structure and behavior, complicating identification of biologically meaningful subtypes and their neuroanatomical and molecular bases. To address this, we analyzed structural MRI data from 1055 ASD and 1164 typically developing (TD) individuals. Gray matter volume (GMV) values from regions showing significant ASD-TD differences were extracted in ASD participants and used for supervised affinity propagation (AP) clustering. Subtypes were then characterized via individual differential structural covariance networks (IDSCN), followed by gene set enrichment analysis (GSEA) based on spatial transcriptomic data from the Allen Human Brain Atlas (AHBA) to identify molecular pathways associated with subtype-specific spatial GMV patterns. This integrative analysis identified two ASD subtypes with distinct neuroanatomical and molecular features. Subtype 1 showed widespread GMV reductions in social, limbic, and salience networks, reduced structural covariance, and enrichment in genes linked to neurodevelopment, neuronal differentiation, and synaptic organization. In contrast, Subtype 2 exhibited GMV increases in similar regions, stronger covariance, and enrichment in pathways involving cytoplasmic translation, mitochondrial organization, and synaptic signaling. These structural and molecular differences were also associated with variations in clinical symptom severity. Our findings reveal two ASD subtypes with distinct cortical morphometry, network patterns, and associated molecular signatures, emphasizing multiple neurodevelopmental paths and the value of biologically informed subtyping for precise diagnosis and personalized treatment.