Role of the AOSI in Predicting Autism Spectrum Disorder in Tuberous Sclerosis Complex

Background:  Historically, autism spectrum disorder (ASD) has been reported in approximately 50% of individuals with Tuberous Sclerosis Complex (TSC). The severity and underlying causes of ASD are complex and highly variable, which presents a major barrier to identifying at-risk infants.

Objectives:  Evaluate the ability of the Autism Observation Scale for Infants (AOSI) to identify at-risk infants with TSC who will later get diagnosed with ASD.

Methods:  Analysis was performed on 130 patients ages 0-36 months with TSC participating in the TSC Autism Center of Excellence Network, a large multicenter, prospective observational study evaluating biomarkers predictive of ASD. Patients were evaluated longitudinally at ages 3, 6, 9, 12, 18, 24 and 36 months with standardized evaluations using cognitive, adaptive, and behavioral measures, as well as autism-specific testing at specific time points. At 12 months, the AOSI was administered to look for specific behavioral risk markers for ASD. The Autism Diagnostic Observation Schedule-Second Edition (ADOS-2) and Autism Diagnostic Interview-Revised (ADI-R) were administered at 24 and 36 months at which point a diagnosis of ASD or not ASD was assigned based on a combination of above stated measures, clinical impression, and consideration of comorbid diagnoses.

Results:  At 24 months, 14 patients out of 65 were given a clinical diagnosis of ASD (21.5%). At 36 months, 4 patients out of 21 were given a clinical diagnosis of ASD (19%). The Mean AOSI total score in patients diagnosed with ASD at 24 months was 13.5 (SD 6.7) versus mean AOSI total score of 7 (SD 5.3) in patients not diagnosed with ASD (p=0.003). The odds ratio for the AOSI predicting ASD at 24 months was 1.19 (p=0.003 [CI 1.06-1.33]). In other words, for every 1 point increase in total score on the AOSI, the odds of going from non-ASD to ASD increases by 19%. The AOSI total score also significantly predicted meeting cut-off scores on the individual domains on the ADI-R in patients diagnosed with ASD at 24 months (p<0.05). There were also several individual items on the AOSI that were found to significantly predict ASD diagnosis at both 24 and 36 months. Additionally, diagnosis of ASD based on ADOS-2 classification alone was compared to clinical diagnosis. At 24 months the ADOS and clinical diagnosis agreed on all non-ASD diagnoses. In patients who were diagnosed with ASD, the ADOS-2 classification and clinical diagnosis agreed 46% of the time (17 out of 37 patients). However, 54% of patients (20 out of 37) met ASD classification on the ADOS-2 but did not receive a clinical diagnosis (p<0.0001). The study is ongoing and many patients have not completed 36-month assessments. This may help to explain the lack of significance for this age group.

Conclusions:  The AOSI performed at 12 months predicts later diagnosis of ASD in patients with TSC at 24 months. Clinical diagnosis of ASD in individuals with TSC differs from classification based on the ADOS-2. This difference is likely due to other medical comorbidities seen in TSC. Further analysis will help to clarify these discrepancies.