Cranial Width (not Macrocephaly) Is Markedly Increased in Multiplex Autism and Predicts the Extent of Craniofacial Dysmorphology

Background: Several laboratories have reported a statistical overrepresentation of macrocephaly in autism. This begs the question of how cranial enlargement is manifested. Also, earlier studies point to an excess of dysmorphic features in autism, but the phenotypic relationship between cranial dimensions and other forms of craniofacial dysmorphology has not yet been explored.

Objectives: We now ask whether these dimensions are proportionately increased or, rather, are disproportionately increased along the principal axes of the cranium: width and length. Further, we determined the extent to which increased dimensions predict craniofacial dysmorphology. This was measured by quantitative morphometry of 3D craniofacial surface images, providing objective and reliable measures of anomalies.

Methods: Subjects were recruited from the Autism Genetic Resource Exchange (AGRE) and the testing was conducted at Children's Hospital, Los Angeles. Craniofacial surface images were obtained by a stereo-photogrammetric image capture system (3dMD, Atlanta), and quantitative analyses were performed using 3D landmark-based morphometry. Spreading calipers (GPM, Zürich) were used to directly measure cranial length and width. The study was limited to subjects who were Caucasian (N=40), for whom normative data are most extensive. Measurements were converted to standardized (z-) scores, conditioned on demographics using a new digitized version of the Farkas norms (FaceValue; N = 1312).

Results: There was a substantial increase in cranial width (biparietal diameter) among subjects with autism. In contrast, cranial length (anterior-posterior distance) was decreased. Width approached a full standard deviation (mean/SD z-score = + 0.80/ 1.43) above the population mean (p < .0001), yet length was approximately half of a standard deviation in the opposite direction (mean/SD z-score = - 0.49/ 1.10; p < .004). In contrast, head circumference was only slightly increased (mean/SD z-score = + 0.18 / 1.06; ns) above population norms. The proportion of head width to length (expressed as the cephalic index), was correspondingly high (mean/SD z-score = + 0.89/ 1.49; p < .0001); this disproportionate pattern is classically described as brachycephaly. Moreover, increased width was predictive of increased craniofacial dysmorphology (full-scale anomaly score; r = 0.46, p < .006).

Conclusions: Cranial width was markedly increased in autism, but there was a corresponding decrease in cranial length. This excessive head width (termed eurycephaly) was seen in 23% of probands using a 3^rd percentile cut-off point, constituting over a 7-fold increase above population values. This finding --- combined with the fact that eurycephaly predicted the extent of craniofacial dysmorphology --- suggests that eurycephaly may present a distinct subtype of individuals with ASD and this may help to resolve some of the heterogeneity in studies of genetics and neurobiology.