P2.86: Modeling early brain growth in autism using MRI

During normal development the infant brain undergoes highly regulated periods of rapid growth and expansion. In autistic children, deviations from normal growth curves have been reported at time points preceding any symptom onset. Previous studies using head circumference have reported early brain overgrowth followed by developmental arrest as an early indicator for autism in infants. It is thought that this atypical growth contributes to the final neurobiology of autism and can be used for the identification and classification of subgroups aligned to clinical phenotypes or genotypes. To compare developmental trajectories of autistic versus typically developing infants, appropriate models must first be built that account for the time-dependent developmental profile. Here we present a nonlinear modelling approach to compare head circumference and total brain volume developmental trajectories of autistic infants (n = 51, age 12–48 months) to typically developing controls (n = 48, age 12–48 months). Longitudinal head circumference data were parsed retrospectively from medical reports. Total brain volumes were obtained from MR images acquired while the infants were asleep using an automated image processing pipeline. Data from the NIH Pediatric MRI Data Repository were queried and added to produce an external control cohort. Normative curves were fit using a robust nonlinear regression to a standard CDC growth model and a multivariate sigmoidal form. Zscores were then calculated to address age effects in the model. Preliminary results indicate better model-fit with the sigmoidal approach. We will report on the methodological implementation and MRI-driven findings to facilitate a better understanding of the aberrant neurobiology and provide a means to identify and characterize subgroups within autism based on developmental trajectory.