Utilizing maternal autoantibody patterns to predict risk of autism and intellectual disability in offspring

Maternal autoantibody-related autism (MARA) is a subset of autism in which specific patterns of maternal autoantibodies (aABs) in the circulation of pregnant women have been associated with increased autism risk in offspring. While initial MARA studies identified patterns consisting of two maternal aABs that predicted increased risk of autism, other multi-aAB patterns predictive of autism and other neurodevelopmental disorders have not yet been fully assessed. In this study, we aimed to determine if additional patterns of MARA aABs can be used to predict the risk of autism and intellectual disability (ID). We tested maternal plasma samples from the Early Markers for Autism (EMA) study for reactivity to eight proteins with clinical relevance in our initial MARA studies. Least Absolute Shrinkage and Selection Operator (LASSO) statistical modeling was used to identify patterns of three maternal aABs that were predictive of offspring autism and ID risk. We identified novel patterns consisting of three aABs associated with increased risk of autism or ID compared to general population controls (GP). Additionally, we found that specific patterns of three maternal aABs differentially predicted risk of autism with intellectual disability (AU ID) and autism without intellectual disability (AU noID), compared to GP. Overall, novel patterns consisting of three maternal aABs have been identified and can be used to predict child clinical risk of autism, ID, and autism subgroups, AU ID and AU noID.