A familial modeling framework for advancing precision medicine in neuropsychiatric disorders: A study in children with RASopathies

ObjectiveDespite the significant and growing burden of childhood psychiatric disorders, treatment is hindered by lack of evidence-based precision approaches. We utilized parent cognitive and behavioral traits in a predictive framework to provide a more individualized estimate of expected child neuropsychiatric and neuroanatomical outcomes relative to traditional case-control studies. We examined children with Noonan Syndrome, a neurogenetic syndrome affecting the Ras/mitogen-activated protein kinase (Ras/MAPK), as a model for developing precision medicine approaches in childhood neuropsychiatric disorders. MethodsParticipants included 53 families of children with Noonan syndrome (age 4-12.9 years, mean = 8.48, SD = 2.12, 34 female). This cross-sectional study utilized univariate regression to examine the association between non carrier parent traits (cognition and behavior) and corresponding child traits. We also used multivariate machine learning to examine the correspondence between parent cognition and child multivariate neuroanatomical outcomes. Main outcome measures included child and parent cognition, anxiety, depression, attention-deficit hyperactivity (ADHD) and somatic symptoms. We also included child neuroanatomy measured via structural MRI. ResultsParent cognition (especially visuospatial/motor abilities), depression, anxiety and ADHD symptoms were significantly associated with child outcomes in these domains. Parent cognition was also significantly associated with child neuroanatomical variability. Several temporal, parietal and subcortical regions that were weighted most strongly in the multivariate model were previously identified as morphologically different when children with NS were compared to typically developing children. In contrast, temporal regions, and the amygdala, which were also weighted strongly in the model, were not identified in previous work but were correlated with parent cognition in post-hoc analysis suggesting a larger familial effect on these regions. ConclusionsUtilizing parent traits in a predictive framework affords control for familial factors and thus provides a more individualized estimate of expected child cognitive, behavioral, and neuroanatomical outcomes. Understanding how parent traits influence neuroanatomical outcomes helps to further a mechanistic understanding of Ras/MAPKs impact on neurodevelopmental outcomes. Further refinement of predictive modeling to estimate individualized child outcomes will advance a precision medicine approach to treating NS, other neurogenetic syndromes, and neuropsychiatric disorders more broadly.