Finding Individual Developmental Brain Circuitry and Brain-Behavior Associations in Autism By a New Multivariate Crossmatch Method

Saturday, May 14, 2016: 11:30 AM-1:30 PM
Hall A (Baltimore Convention Center)
N. Lange1, D. C. Dean2, B. G. Travers3, B. A. Zielinski4, A. L. Alexander2 and J. E. Lainhart2, (1)Psychiatry, Harvard Medical School, Boston, MA, (2)Waisman Center, University of Wisconsin-Madison, Madison, WI, (3)Occupational Therapy Program in Kinesiology, University of Wisconsin Madison, Madison, WI, (4)Pediatrics and Neurology, University of Utah, Salt Lake City, UT

There is an ongoing worldwide search for clinically meaningful associations between an individual’s cognitive-behavioral functioning and brain structure and function in autism. For this pursuit, we developed a new case-by-case method, being a longitudinal extension of an existing crossmatch method, to compare one individual’s longitudinal Mahalanobis distances from another’s based on any type and number of individual brain and/or other measurements. 

Objectives:   To employ the new method to identify individual associations between circuitry network distances and scores on ADOS, SRS, and VABSII Receptive and Expressive Language assessments, if any.

Methods:   Our sample consisted of repeated and co-temporal brain image measurements and cognitive-behavioral assessments contributed by 92 male participants with autism ages 3-35 years followed for 12 years, and 56 age-matched typically developing male participants also followed for 12 years. Study design and scanner settings have been described elsewhere. Based on current literature on the corpus callosum (CC) and arcuate fasciculus (AF) in autism, we examined means and growth rates of fractional anisotropy (FA) in the CC genu, CC body, CC splenium, right AF, left AF, right AF longitudinal segment, and left AF longitudinal segment in order to find the best DTI brain measurements that separate the participants with autism from typically developing participants, if any.

Results:   The new multivariate method found that the best subset of the regional FA measures in the corpus callosum and arcuate fasciculus included the body, splenium, left AF, right AF, right AF longitudinal segment, left AF longitudinal segment, and the growth rate of the genu (distribution separation p-value < 0.000000005). Two other measurement subsets performed equally well, having identical distribution separation p-value. These included, separately and in addition, the genu, splenium, left AF longitudinal segment, and the growth rates of the body, splenium, left AF, and the left AF longitudinal segment. The single measurement included in all three measurement subsets was the growth rate of the genu. These aggregated FA means and growth rates showed significant associations with the selected cognitive-behavioral measures.

Conclusions:  Developmental circuitry differences in autism, as measured in the corpus callosum and arcuate fasciculus, are complex. Our findings suggest that the growth rates of the brain circuitry structures examined may play a more salient role than the mean sizes of regional DTI coefficients in distinguishing differences in the brains and cognitive-behaviors of individuals with autism.