Characteristic cortical thickness patterns in adolescents with autism spectrum disorders: interactions with age and intellectual ability revealed by canonical correlation analysis.

Published

Journal Article

To investigate patterns and correlates of cortical thickness in adolescent males with autism spectrum disorders (ASD) versus matched typically developing controls, we applied kernel canonical correlation analysis to whole brain cortical thickness with the explaining variables of diagnosis, age, full-scale IQ, and their interactions. The analysis found that canonical variates (patterns of cortical thickness) correlated with each of these variables. The diagnosis- and age-by-diagnosis-related canonical variates showed thinner cortex for participants with ASD, which is consistent with previous studies using a univariate analysis. In addition, the multivariate statistics found larger affected regions with higher sensitivity than those found using univariate analysis. An IQ-related effect was also found with the multivariate analysis. The effects of IQ and age-by-IQ interaction on cortical thickness differed between the diagnostic groups. For typically developing adolescents, IQ was positively correlated with cortical thickness in orbitofrontal, postcentral and superior temporal regions, and greater thinning with age was seen in dorsal frontal areas in the superior IQ (>120) group. These associations between IQ and cortical thickness were not seen in the ASD group. Differing relationships between IQ and cortical thickness implies independent associations between measures of intelligence and brain structure in ASD versus typically developing controls. We discuss these findings vis-à-vis prior results obtained utilizing univariate methods.

Full Text

Duke Authors

Cited Authors

  • Misaki, M; Wallace, GL; Dankner, N; Martin, A; Bandettini, PA

Published Date

  • April 15, 2012

Published In

Volume / Issue

  • 60 / 3

Start / End Page

  • 1890 - 1901

PubMed ID

  • 22326986

Pubmed Central ID

  • 22326986

Electronic International Standard Serial Number (EISSN)

  • 1095-9572

Digital Object Identifier (DOI)

  • 10.1016/j.neuroimage.2012.01.120

Language

  • eng

Conference Location

  • United States