20328
From Experimental to Investigational: The Potential for Eye Tracking As a Biomarker for Outcome in Clinical Trials
Objectives: To characterize discriminant validity, test-retest reliability, and relationships with functioning/clinical dimension of eye tracking measures in adults with ASD across a wide range of eye-tracking paradigms.
Methods: Participants were high functioning adult males with ASD from 2 different studies, study 1, (n=19, mean age = 23years, median FSIQ=104) and study 2 (n=19, mean age= 25 years, median FSIQ = 102) and typically developing (TD) adult males (n=19, mean age = 26 years, median FSIQ = 118). Eye tracking tasks included biological motion preference (BioPref), social versus geometric movies preference (SocGeo), gaze direction discrimination (GazeDiscrim), gender discrimination (GenderDiscrim), activity monitoring (ActMon), and complex social movies (ComplexSoc). Overall saccade rates and pupil diameters were also computed as was an a priori defined manual composite score (Composite; computed as a linear combination of other eye-tracking variables). Participants saw the 30-minute eye-tracking battery twice, separated by two hours. Additional assessments included an affective speech recognition (ASR) task, the state component of State-Trait Anxiety Inventory (STAI), the Reading the Mind in the Eyes Test (RMET), a Smell Test (Smell), and Autism Quotient (AQ).
Results: The AQ was significantly different between groups (ES=2.6, p<0.001). In eye tracking tasks the largest differences between groups were observed in Activity Monitoring and Biomotion paradigms. The ASD group looked less at people (p=.011, ES=-0.9) and heads (p<.01, ES=-1.0) during Activity Monitoring and showed diminished looking preference at biological motion task (p=.0.035, ES=-.7). Pupil size was significantly larger in the ASD group across all eye tracking tasks (ES=.7 to .9)). Test-retest reliability within ASD for most eye-tracking measures and all pupil size measures was acceptable to excellent (eye tracking: ICC~0.7 to ~ 0.8; pupil size: ICC ~ 0.8 to ~ 0.9. Patients scored higher on the STAI and showed worse performance on RMET, ASR and Smell Test, but none of these differences reached statistical significance. Overall, correlations with clinical dimensions were low.
Conclusions: Comparisons suggest that the eye tracking battery shows comparable or better ASD-TD differentiation than other tasks meant to examine non-core features associated with ASD (STAI, ASR, Smell, RMET), but inferior differentiation compared to autism-specific questionnaires (AQ). Between group findings validate many prior research results. ICCs showed eye tracking variables exhibit good to excellent test-retest reliability, a critical feature for use in monitoring of treatment effects. In summary, this study suggests eye tracking may have good properties for monitoring clinical outcomes.