Age-Related Differences in Optic Nerve Geometry in Autism Spectrum – a Potential Imaging Marker?

Background: Brain maturation in autism is aberrant. In early life, brain volumes enlarge, particularly in amygdala. However, by adolescence there is generally no overall volume difference relative to age-matched controls, although complex regional anatomical changes persist.  

Objectives:  With an eye to early diagnosis and multi-centre application, we explored MRI measurements which would not rely upon scanner hardware or intensity based image processing methods. We developed a routine to map the optic chiasm and its branches in T1 scans to extract measurements of: optic angle, anterior extension of the optic nerve (a), and posterior extension of optic nerve (b).

Methods:  A balanced sample of 30 boys with autism spectrum disorders and 30 typically developing age- and verbalIQ- matched controls had T1 scans available for analysis. Parents gave informed consent for the scanning protocol approved by the Hong Kong West Cluster Institutional Review Board. T1 images were screened for rotational tilt due to ‘roll’ or ‘pitch’ in sagittal and coronal planes.  A transformation matrix quantifying the tilt angle for each image was constructed and realignment achieved using SPM2 software (Wellcome Department of Imaging Neuroscience, London (http://www.fil.ion.ucl.ac.uk). A reverse transformation was applied to the image to adjust tilt.  The re-aligned image was resliced to generate isotropic 1x1x1 mm voxel dimensions. Using ITKsnap software the optic nerves were traced from the tip of the orbit where the optic nerves are not surrounded by the extraocular muscles to the last slice where the optic tracts are visible before entering the brain thalamic substance. Data were imported into Matlab and a measurement kernal established to calculate the anterior distance from optic chiasm forward (a), back (b) and optic angle.

Results: There were no group differences in cranial volumes. In a multivariate general linear model with age as a covariate, for dependent variable ‘a’, there was a significant main effect of age (F = 5.22, p = 0.026) and Group and Group x age interaction approached significance (F = 3.71, p = 0.058), (F = 5.22, p = 0.026) respectively. Correlation analysis revealed a strong positive correlation between age and anterior distance in the autism group only (r = 0.54, p = 0.002) and this correlation was significantly different from controls (Z = -2.03, p = 0.02). Thus, the anterior distance from the optic chiasm in the autism group is shortened compared to controls prior to 12years, then expands beyond controls in adolescence. In contrast the posterior distance from optic chiasm back towards thalamus tended to be greater in young children with asd but smaller in adolescents compared to controls but this was not statistically significant.

Conclusions:  We are currently examining replication datasets. If this finding proves robust, we will extend our study to examine optic nerve geometry in younger children with autism and test the hypothesis that anterior distance from optic chiasm is markedly shorter in younger children with autism. The hope is that this observation may hold potential as an early imaging marker to help identify at risk infants.