Four autism groups in a genetic study: what they mean
A study published in Nature Genetics analysed 239 characteristics in 5,392 children from the SPARK cohort and identified four statistical groups. The groups showed different clinical profiles and genetic associations. They are not new diagnoses, cannot be assigned using a test available in clinics and still need to be replicated in other samples.
What they found
Researchers at Princeton and the Simons Foundation analysed 239 characteristics in 5,392 autistic children from the SPARK cohort. These included development, communication, social relationships, repetitive behaviour and co-occurring conditions. They then looked for clusters in the data and examined their genetic associations.
The analysis identified four groups within this sample. Each group brought together a different combination of traits and genetic associations:
- Social and behavioural challenges (around 37%): they show clear autistic traits but reach developmental milestones — walking, talking — at a pace similar to everyone else. They often co-occur with ADHD, anxiety, depression or OCD.
- Mixed with developmental delay (19%): developmental delays alongside a combination of other traits and conditions.
- Moderate challenges (34%): less pronounced autistic traits on the measures used and fewer co-occurring conditions.
- Broadly affected (10%): more extensive difficulties, including developmental and communication delays, with a higher proportion of large-effect genetic variants.
What it really means
The analysis suggests that combining all participants may conceal genetic associations linked to particular combinations of traits. This can guide further research into development. The study does not yet show that assigning someone to one of these groups improves assessment or treatment.
What it does NOT mean
- It isn't a closed classification. The authors themselves warn: there may be more than four subtypes. It's a starting point, not a final label.
- It cannot be used to diagnose someone at home. The classification came from a research data analysis and is not currently used to assign diagnoses in clinical practice.
- It does not determine an individual's support. Sensory, communication, educational and daily-living needs still require an individual assessment.
Decisions about support still require an individual assessment and information from the person, their family and their professionals. These four groups describe this sample; they do not replace clinical support levels or predict a child’s life on their own.
Frequently asked questions
Can I find out which subtype my child is in?
No, and for now it isn't used in practice either. It's a research tool, not a diagnostic one. What helps you today is still watching how they experience each situation.
Does this mean autism is going to be "cured"?
The study doesn't talk about a cure, but about understanding the biology better to give more personalised support. The aim isn't to change the person, but to understand them well enough to help them better.
Does it change anything I do with my child?
No. What helps in everyday life — lowering the noise, giving warning of changes, respecting their pace — is just as valid today as it was yesterday.
Litman, A., et al. (2025). Decomposition of phenotypic heterogeneity in autism reveals underlying genetic programs. Nature Genetics. doi.org/10.1038/s41588-025-02224-z
Autism in Everyday Life
The book brings together explanations and practical examples for autistic people, families and professionals. Publication is planned for September 2026.
Tell me when it's out →
Writes about neurodivergence: practical guides for families, teachers and adults, and illustrated fiction. About me