Cardiac Comorbidities and Their Effect on Brain Structure

Background: Autism Spectrum Disorder (ASD) is a neurodevelopmental disorder with an occurrence rate greater than 1%, characterized by high heterogeneity in etiology and symptomology (Kim et al., 2014). It is associated primarily with neurodevelopmental and behavioural deficits, but has various associated comorbidities, amongst which cardiac comorbidities are common (Kohane et al., 2012). Cardiac abnormalities range from mild cardiac defects to severe congenital heart disease (CHD) (Doshi-Velez, et al., 2014). Accumulating evidence suggests that severe cardiac abnormalities are present and also contribute to the observed neurodevelopment (Ortinau, et al., 2012).

Objectives: Here we explore the association between anatomical brain characteristics and cardiac abnormalities, in a diverse set of prominent ASD mouse models.

Methods: Fixed (ex-vivo) brains of 66 prominent ASD mouse models were scanned using a 3D T2-weighted MRI protocol on a 7 Tesla Agilent scanner (Santa Clara, CA). MRI data were analyzed using image registration and deformation-based morphometry to obtain anatomical volumetric information on 336 brain structures. Data from 19 models (224 females, 497 males; 378 wild-types, 340 mutants) for which a cardiac annotation was found in literature were used. Each model was assigned a normal (7 models) or abnormal (12 models) cardiac phenotype. For example, Arid1b gene is associated with ASD and CHD (Homsy, et al., 2015). So, the Arib1b knockout mouse model was assigned an abnormal cardiac phenotype. We further examined 12 of those models having either normal or CHD-related abnormal cardiac phenotype (134 females, 290 males; 223 wild-types, 201 mutants). Appropriate controls were used, all assigned a normal cardiac phenotype.

Statistical analysis using a linear mixed effects model was performed to test for the effect of cardiac phenotype on the volumetric differences between mutant and wild-type mice, for each mouse model across 336 brain regions. FDR correction for multiple comparisons was performed.

Results: At 5% FDR there was no significant effect of abnormal cardiac phenotype on the volumetric differences between mutants and wild-types. At 20% FDR, however, there was a significant effect of CHD-related cardiac abnormalities in the left trabenular commissure (FDR = 20%, T=3.54).

Conclusions: Cardiac dysfunction is an ASD comorbidity (Memari, et al., 2012). Additionally, CHD newborns show abnormal brain development (Limperopoulos et al., 2010). For a set of prominent ASD mouse models, an effect of abnormal cardiac phenotype on neurodevelopment would be expected. Preliminary results did not show a significant effect at an FDR of 5%. The inability to observe an effect may be a result of two factors, both related to cardiac phenotype being a binary variable. Firstly, the type of cardiac abnormality may influence the strength and direction of the neurodevelopmental effect. Then, with a group-wise analysis these patterns would disappear or cancel out. Secondly, the lack of measurement-based cardiac phenotyping may have resulted in low sensitivity. However, at an FDR of 20%, a trending effect of CHD-related cardiac abnormalities was observed for the left trabenular commissure. To address the above limitations, we will obtain measurements assessing cardiac function for each model and perform further analysis possibly using a more powerful model.