Clincal Validation of a Test for Maternal Antibody Related (MAR) Autism

Background: It has been reported based on Western Blot analysis (Braunschweig, et al., 2013), that approximately 23% of mothers of children with ASD have specific patterns of autoantibodies to brain proteins. These patterns are found in less than 1% of mothers of typically developing children. Thus, these autoantibodies are potential biomarkers for a newly identified subtype of autism, Maternal Antibody Related (MAR) autism, and may be risk factors for ASD. Validated biomarkers for ASD can potentially contribute to earlier diagnosis and intervention. We have converted the Western Blot to an ELISA method that is more suitable for clinical laboratory use.

Objectives:   We set out to confirm that the ELISA method replicates the Western Blot findings and to determine the potential value of the test based on its sensitivity, specificity, and Positive Predictive Value [PPV]) in a large prospective-retrospective clinical study.  

Methods:   This study includes over 450 mothers enrolled in the CHARGE (Childhood Autism Risk from Genetics and the Environment) Study whose children (2-5 years old) have had a diagnosis of ASD confirmed on both ADI-R and ADOS, as well as case controls from mothers of typically developing children. Maternal plasma specimens collected at study enrollment were analyzed for the presence of multiple autoantibodies on a quantitative, high-throughput ELISA platform. These results are evaluated to provide a qualitative result for the presence or absence of the MAR autism subtype.

Results:   The study is ongoing and results from the full study will be presented. Preliminary results on 123 samples with 6 of 7 known antigens show that the transfer of the MAR Autism Test to a high-throughput ELISA platform has been successful. We observed a 15% sensitivity (percent of ASD samples that tested positive) with 98% specificity (percent of TD samples that tested negative), consistent with previous results reported using the Western blot method. We will present the results of the decision algorithm training study of about 250 samples and a blinded validation study on about 200 samples, each set containing approximately 63% samples from mothers who have children with ASD, and 37% mothers of typically developing children.

Conclusions:   These preliminary results suggest that the presence of MAR antibodies can be used to identify a substantial subtype of children with ASD with an actionable positive predictive value due to a low false positive rate. A positive result in a mother of a child where developmental delay is observed indicates a high probability of a future ASD diagnosis based on behavioral criteria. In such a case, behavioral therapy could be started immediately for the most effective outcome for the child.  Samples from pregnant women are not included in the study, and the test is not validated for use during pregnancy.