Development of a Novel Protocol for Characterizing Dysmorphology to Enhance the Phenotypic Classification of Autism Spectrum Disorders

Background:  Pediatric dysmorphology evaluations assess children for physical features that occur rarely in the general population. For example, short stature, protruding ears, and webbed toes are well-established dysmorphic features. Children with multiple dysmorphic features are more likely to have an underlying genetic condition or teratogenic exposure that affected the typical developmental process. In children with autism spectrum disorders (ASDs), the presence of multiple dysmorphic features might identify a distinctive ASD phenotype and serve as a potential marker for understanding cause and prognosis.

Objectives:  To develop a quantitative method that defines criteria for classifying children as Dysmorphic in a case-control study of ASD, and to apply these criteria to determine if the proportion classified as Dysmorphic differs between children with and without ASDs.

Methods:  This analysis included children enrolled in the Study to Explore Early Development (SEED), a multisite population-based case-control study of children aged 2-5 years with ASD (n=145) and two control groups, one drawn from the general population (POP, n=129) and one with non-ASD developmental problems (DD; n=90). Study participants received a physical examination including basic anthropometry and standardized body region photographs. Seven clinical geneticists were each assigned a body region—Head/Hair/Face/Neck, Ears, Eyes/Eyebrows, Nose/Philtrum, Mouth/Lips/Teeth, Hands/Feet, Growth/Skin—for which they conducted on every child a standardized dysmorphology review of potential dysmorphic features within just that body region. Each feature was scored: 0=normal/absent; 1=possible/questionable; 2=mild; 3=moderate; 4=severe. Geneticists were blinded to each child’s study group assignment. For each reviewed feature, the scores that occurred in ≤5% of POP children (e.g., 4 only; 3 and 4 together; 2, 3, and 4 together) were designated as dysmorphic. To obtain a total dysmorphology score, the number of features (across body regions) designated as dysmorphic was divided by the number of all features scored for the child. The total dysmorphology scores for POP children were fit to the log normal distribution to delineate dysmorphology classifications: <90th percentile=Non-dysmorphic; ≥90th and <95th percentile=Equivocal; ≥95th percentile=Dysmorphic. Based on each feature’s definition of dysmorphic in the POP children, total dysmorphology scores were similarly calculated for ASD and DD children and the proportions within each dysmorphology classification (based on the distribution for POP children) were compared between groups by chi-squared or Fisher’s exact tests.

Results:  These preliminary analyses included only non-Hispanic white children. Significantly higher proportions of children with ASD or DD were classified as Dysmorphic, compared with POP children (χ2=14.14, p=0.0008 and χ2=10.99, p=0.0041, respectively). When children with known chromosome anomalies or genetic syndromes were removed from each group, the significant differences in dysmorphology classifications persisted between POP children and the other two groups. Within each group there was no significant difference in dysmorphology classification between boys and girls. Future analyses will evaluate racial/ethnic differences among ~1100 children in SEED.

Conclusions:  More ASD group children met criteria to be classified as Dysmorphic than children in the POP group. This novel protocol defines a quantitative dysmorphology classification and identifies categories of Dysmorphic and Non-dysmorphic children with ASDs. These children can be further assessed according to etiologic risk factors.