Maternal Autoantibody Combinations as Predictors of Autism Risk: Retrospective Discovery and External Validation in a Prospective Cohort at Increased Risk

Background/Objectives: Maternal Autoantibody-Related Autism (MARA) is a recently defined subset of autism characterized by specific combinations of maternal autoantibodies which cross the placenta and bind neurodevelopmentally critical target proteins. A clinical multi-ELISA MARA assay was recently rigorously analytically validated. The primary objective of this study was to assess the potential performance of this MARA autoantibody assay for predicting autism risk in high-risk populations. Secondary objectives were to establish the significant MARA autoantibody combinations and verify clinical discrimination in an external cohort. Methods: Bio-banked plasma samples from mothers of 420 autism and 335 typically developing (TD) children were analyzed by ELISA for MARA autoantibodies in combination. Transcendent autoantibody pairwise combinations were identified in discovery where both autoantibodies in a combination exceeded the values of every control participant. Permutation testing was used to establish that the probability of observed results significantly exceeded chance. Performance was then validated in an independent cohort (mothers of 42 autistic and 90 TD children). Results: Transcendent antibody levels were discovered in 70 mothers (16.7% of autism cases; no typically developing participants) in the discovery cohort. Specificity and positive predictive values for autism were 100% by construct. An external cohort showed similar performance with 14 (33.3%) transcendents. Permutation testing of each cohort demonstrated probabilities of transcendent cases to be significantly higher than expected be chance. Conclusions: The data reported here provides evidence for a high-risk subset of mothers of children with autism identified by the MARA autoantibody assay which enables further clinical validation.