J. Ellegood, A. Beauchamp, Y. Yee, G. Devenyi, J. Ziolkowski, L. Qiu, R. Askalan, M. Ayub, P. Suetterlin, A. Donovan, M. A. Basson, K. M. Quesnel, N. G. Berube, T. Woo, D. Beversdorf, H. Bjornsson, R. Blakely, F. Calderon de Anda, J. Crawley, J. Crosbie, B. O. Orr, G. W. Davis, M. Genestine, E. DiCicco-Bloom, S. Egan, K. D. Fink, S. Asbury, J. Lai, K. Rilett, J. A. Foster, J. B. Vincent, P. Frankland, S. Georgiades, O. Penagarikano, D. Geschwind, R. J. Giger, S. Markx, J. Gogos, C. Golzio, R. Goveia, A. Muotri, M. Pagani, A. Gozzi, L. K. Pacey, Hamps
The heterogeneity of autism and related neurodevelopmental conditions has impeded accurate prognoses and treatment discovery. Using translational neuroimaging across 135 mouse models (3,515 mice) and two human MRI datasets (n = 1,234 and n = 1,015), we derived participant subgroups from shared neuroanatomical features. These subgroups did not distinguish autism, ADHD, or OCD diagnoses and only modestly differentiated cognitive and behavioural phenotypes. Instead, they mapped onto four molecular pathways: (1) synaptic function; (2) MAPK and Wnt signalling; (3) chromatin modification and cellular stress responses; and (4) broader chromatin, immune, and second-messenger signalling pathways. This framework bridges preclinical models and idiopathic human neurodevelopmental conditions, linking patients to biologically relevant molecular mechanisms.