Geometric Responders: A Clearly Definable Subgroup of Toddlers with ASD

Background: Eye tracking technology holds promise as an objective methodology for characterizing early features of autism. Given the choice, typically developing (TD) toddlers prefer to observe social over non-social images. It is unclear if this preference exists in toddlers with autism (ASD), particularly during early stages of clinical onset. Using a preferential looking paradigm, we recently discovered (Pierce et al., 2010) that a subset of toddlers as young as 14-months with later confirmed autism spent considerably greater time visually fixating on dynamic geometric shapes versus social images, a pattern not found in TD or developmentally delayed (DD) toddlers.  Specifically, when using a 67% fixation time cut off, the positive predictive value for accurately classifying toddlers as ASD was 100%. In this first study, 40% of ASD toddlers displayed a preference for geometric repetition and 60% displayed the typical pattern-an interest in social images. Thus, the sensitivity of the geometric test for autism is low, but the specificity is exceptionally high.

Objectives: To determine, in a new sample, if toddlers at-risk for ASD prefer to look at dynamic geometric images (DGI) over dynamic social images (DSI) and the degree to which preferential looking patterns can discriminate toddlers at-risk for an ASD from those at-risk for a DD as well as TD controls.

Methods: Using a population-based screening method, toddlers with later confirmed ASD or DD as young as 12-months were recruited and longitudinally tracked.  One hundred and thirty-nine toddlers ranging between 12-42 months participated (51 ASD; 65 TD; 23 DD; mean age=27 months) and only those with later confirmed diagnoses were included.  As in the original study, toddlers viewed a 1-minute movie consisting of the simultaneous and adjacent presentation of DGI and DSI. Fixation length was determined using a TOBII eye tracker and preference was defined as looking time >50% of total time looking towards one movie type.

Results: Overall, toddlers at-risk for ASD spent significantly more time looking at DGI than TD, t(114)=5.5, p <.001 and DD t(72)=1.9, p <.05 toddlers. Thirty-seven percent of the ASD group displayed a DGI preference in contrast to only 1.5% of typical and 21% of DD toddlers. Several of the ASD toddlers exceeded 67% DGI viewing time, a pattern not found in any other group in either study. Thus, as in the first study, when 67% DGI viewing time is used as a cut off, the positive predictive value for accurately classifying that toddler as ASD is 100%. Out of the >100 TD and >100 ASD toddlers that have participated in either the original or current study, only 3 TD have preferred DGI slightly (<60% viewing time) in contrast to 34 geometric responders from the ASD groups.

Conclusions: Results of the present study replicate the original findings. Both studies strongly suggest that a preference for geometric repetition is a clearly definable phenotype of early autism and thus can be considered an early marker of autism. The degree to which brain, blood, and other behavioral correlates exist with this unique subgroup will be explored.