Genomic and Developmental Models to Predict Cognitive and Adaptive Outcomes in Autistic Children.

Area of Science:

• Neuroscience
• Genetics
• Developmental Pediatrics

Background:

• Early signs of autism spectrum disorder (ASD) appear in early childhood, but predicting co-occurring intellectual disability (ID) remains difficult.
• Current clinical tools lack the predictive power to identify autistic children at high risk for developing ID.

Purpose of the Study:

• To develop and validate predictive models for identifying intellectual disability (ID) in children diagnosed with autism spectrum disorder (ASD).

Main Methods:

• A prognostic study developed and validated models integrating genetic variants and developmental milestones.
• Models were trained, cross-validated, and tested across three large autism cohorts (SPARK, SSC, MSSNG).
• Predictive performance was assessed using AUROC, PPVs, and NPVs.

Main Results:

• The integrated model achieved an AUROC of 0.653, demonstrating generalizability across cohorts.
• Combinations of genetic variants, not typically clinically relevant, achieved 55% PPVs, identifying 10% of individuals who developed ID.
• Polygenic scores improved negative predictive values (NPVs) when added to developmental milestones.

Conclusions:

• Individual neurodevelopmental condition-associated variants are generally insufficient for predicting ID.
• Models combining genetic variants and developmental milestones offer clinically relevant predictions for early intervention targeting.
• Genetic stratification of ID probability was more effective in individuals with delayed developmental milestones.