Children with autism can be grouped into four clinically meaningful subtypes with distinct genetic mutations and developmental trajectories that align with specific clinical outcomes, according to a recent study.
Researchers applied a person-centered phenotypic-genetic framework to uncover biologically and clinically relevant subtypes of autism spectrum disorder (ASD). The researchers, led by Aviya Litman of the Quantitative and Computational Biology Program, Princeton University, Princeton, NJ, and colleagues, used a general finite mixture modeling approach to analyze 239 item-level and composite features in 5,392 children with autism from the SPARK cohort, integrating data from standardized instruments such as the Social Communication Questionnaire, Repetitive Behavior Scale-Revised, and the Child Behavior Checklist.
The model highlighted four phenotypically distinct ASD classes: Social/behavioral (n = 1,976), Mixed ASD with developmental delay (DD) (n = 1,002), Moderate challenges (n = 1,860), and Broadly affected (n = 554). These classes exhibited differential enrichment across seven phenotype categories, including attention deficit, anxiety, disruptive behavior, and DD. The class assignments were externally validated through co-occurring diagnoses, including ADHD, language delays, and intellectual disability, and replicated in an independent cohort from the Simons Simplex Collection (n = 861; Pearson r = 0.927, P < .0001).
Genetic analyses demonstrated class-specific patterns of common and rare variation. Polygenic scores (PGS) for ADHD and depression were significantly elevated in the Social/behavioral and Broadly affected classes compared with siblings without autism (FDR < .01; Cohen d > 0.22). The Broadly affected class exhibited reduced PGS for intelligence and educational attainment (FDR < .1). Rare variant analysis showed that high-impact de novo loss-of-function variants were significantly enriched in the Broadly affected class (FDR = .01; FE = 1.66 vs. siblings), while the Mixed ASD with DD class showed enrichment for both de novo and rare inherited variants (FDR = .016; FE = 2.55).
Genes disrupted in each class mapped to distinct biological pathways. For example, the Social/behavioral class showed enrichment in chromatin regulation (FE = 3.5; FDR = .0019), while the Mixed ASD with DD class was associated with voltage-gated sodium channel activity (FE = 28.8; FDR = .0035). Gene expression trajectories also differed; the Social/behavioral class was enriched for postnatally expressed genes, whereas the Mixed ASD with DD class had prenatal expression profiles. These expression patterns aligned with clinical milestones: the Mixed ASD with DD class had significantly later developmental milestones and earlier diagnosis (Cohen d > 0.38; FDR < 1.9 × 10⁻¹⁹) compared to the Social/behavioral class.
According to the researchers, these findings define reproducible ASD subtypes with distinct genetic and developmental profiles that may inform future studies on diagnostic stratification and individualized treatment approaches.
The authors reported no competing interests.
Source: Nature Genetics