Developmental Trajectories of Social Vocal Behavior As Biomarkers for Autism During the First 24 Months of Life: Risk Status Vs. Diagnosis

Background: Although autism is difficult to diagnose reliably before two years of age, prodromal symptoms may emerge over time during the first twelve months of life. Due to natural variability in developmental timescales, cross-sectional measures of behavior usually fail to capture significant differences over this period. Current research suggests that early biomarkers of risk may be found instead by examining longitudinal trajectories of individual development. Experimental research has shown the importance of contingent interaction between infant and caregiver in scaffolding vocal development, which suggests that developmental profiles of early vocal behavior and vocal interaction may be candidate biomarkers for autism. In previous investigations, we found evidence for a developmental cascade in vocal development discriminating high-risk and low-risk siblings, beginning at 12 months, with deficits in vocal contingency that lead to later deficits in adult and infant volubility, all of which were predictive of speech and language outcome at 24 months. In this follow-up study, we re-analyzed the developmental profiles of the same children according to diagnostic outcome at 24 and 36 months, to determine whether deficits in social vocal engagement appear earlier in infants who later receive a diagnosis, and whether high-risk siblings who do not go on to receive a diagnosis exhibit patterns that more closely resemble autism or typical development.

Objectives: The goal of this study is to determine the relationship between developmental profiles of social vocal engagement from birth to two years of age, risk status and diagnosis, to test whether early vocal development is a categorical indicator of outcome.

Methods: As part of an NIH Autism Center of Excellence, we tracked vocal development among 37 high-risk infant siblings with a family history of ASD and 35 low-risk controls. Each child wore a recording device (LENA) for one day every month from 0-24 months to provide audio recordings of their vocal behavior and natural language environment. Utilizing automatic speech recognition technology developed by our laboratory, we identified the number of vocalizations per hour for infant and caregiver, and calculated the rate of contingent interactions based on timing statistics. Using Functional Data Analysis, we determined developmental trajectories for each child and mean developmental trajectories by group, based on risk status and diagnostic outcome: 10 children were diagnosed with ASD, 27 were DD/BAP, and 35 were typically developing.

Results: The developmental cascade linking early declines in vocal interaction to later declines in adult volubility and subsequent declines in infant volubility is exaggerated in infants diagnosed with ASD. Differences that only became apparent at 12 months between high-risk and low-risk infants are present between infants with ASD and typically developing peers within the first year of life. Developmentally delayed or BAP infants resemble typically developing infants at birth, but transition towards developmental trajectories resembling infants with ASD by the second year of life.

Conclusions: Developmental trajectories of social vocal engagement over the first two years of life differentially predict diagnostic outcome at two years of age in children with ASD and developmentally delayed and typically developing controls.