Speaker
Description
This study investigates autism as a highly heterogeneous neurodevelopmental condition, and proposes a neuroanatomical stratification as a biologically grounded approach to better capture individual variability beyond behavioural measures.
We analysed MRI-derived cortical thickness data from 993 participants, including 512 autistic individuals aged 3–31 years, across three large European datasets. Using normative modelling, we generated individual “neuroanatomical fingerprints” reflecting deviations from typical brain development. Principal component analysis identified 15 components explaining 72% of variability, and unsupervised clustering revealed three distinct neuroanatomical subgroups among autistic individuals.
Each subgroup showed unique patterns of cortical thickness differences that were associated with divergent clinical profiles, mapped onto different gene sets, and correlated with distinct core functional systems.
Overall, the findings demonstrate that autism can be meaningfully stratified based on highly individualised neuroanatomical pattern, supporting more personalised research into underlying biological mechanisms and future targeted interventions.