Autism dysmorphology measure validity study
Judith Miles, M.D., Ph.D.
University of Missouri
The overarching goal of our work is to delineate the clinical and genetic heterogeneity within the Autism Spectrum Disorder diagnosis so that prognoses, treatments and preventive strategies can be implemented which are specific to each autism subgroup. We start by looking for biologically-based phenotypes which can be used to define discrete subgroups. One autism subgroup has been recognized based on findings generalized through often subtle differences in their physical features, indicating an insult to normal morphogenesis. Based on the premise that they would be genetically different from the rest, our goal has been to learn how to distinguish those children, characterize the ways they differ from children without dysmorphology and translate that information into best clinical practices.
We find in autism, dysmorphology usually is not diagnostic of a specific syndrome but rather a phenotypic indicator of different path(s) of embryological development. Children with generalized dysmorphology are more apt to have microcephaly, seizures, brain abnormalities & low IQ scores; they generally don’t present with regression. In addition, they are genetically distinctive with a lower sex ratio is lower (2.9:1), low sib recurrences (<1%) and less autism and neuropsychiatric disorders in their families. These differences indicate this subgroup is distinct; they are said to have complex autism, compared to “essential” autism in the rest.
To aid clinicians in the identification of complex autism, we developed the Autism Dysmorphology Measure (ADM), which directs the clinician, following a brief, clothed examination, to make a judgment of dysmorphic vs. nondysmorphic. The use of this measure in the Simons Simplex Collection (SSC) is providing the first test of the ADM, including comparing ADM reliability when used by practitioners with different levels of training and analyzing the reliability of the ADM compared to a comprehensive dysmorphology examination. So far, comparisons between the larger SSC data set, Missouri data and previous reports have yielded excellent agreement. At Missouri, 12.2% of probands were coded as dysmorphic, similar to 13% across all sites and to 11.7% in the population from which the ADM was developed. Comparing the ADM with a comprehensive dysmorphology examination by a geneticist has provided exceptional correlation (99% sensitivity; 100% specificity). In addition to enhancing previous associations, rigorously controlled SSC data allows the identification of new associations, including the use of the Social Responsiveness Scale (SRS) to study autism family loading in subgroups. Finding lower SRS scores for fathers (p<0.001) and unaffected sibs (p=0.004) of probands with dysmorphology supports our previous assessment that complex autism is more apt to occur in families with less genetic loading for autistic traits.