Neonatal Thyroid Stimulating Hormone and Subsequent Diagnosis of Autism Spectrum Disorders and Intellectual Disability

Background: Thyroid abnormalities in early life, if left untreated, are associated with adverse neurodevelopmental outcomes, including intellectual disability (ID). However, evidence addressing the role of neonatal thyroid hormones in the altered neurobiology underlying autism spectrum disorders (ASD), particularly among its subphenotypes, is limited.

Objectives: To determine the association between neonatal thyroid hormone levels and subsequent diagnosis of ASD, ID, and subphenotypes of ASD.

Methods: Utilizing data from the Early Markers for Autism (EMA) Study, we conducted a population-based, case-control study among a sample of 4.5-9-year-old children born during 2000-2003 in Southern California. We examined neonatal thyroid-stimulating hormone (TSH) levels measured during routine newborn screening among children later diagnosed with ASD (n=524), ID (n=124), and general population (GP) controls (n=402). GP controls were randomly sampled from birth certificate files and frequency matched to cases by sex, birth month, and birth year. TSH was further analyzed in relation to ASD subgroups defined by intellectual ability and onset type (early onset ASD vs. ASD with regression) ascertained by expert review of records from the California Department of Developmental Services. ASD cases with composite scores on standardized cognitive tests less than or equal to 70 were classified as ASD with intellectual disability (ASD+ID) and those with scores greater than 70 were classified as ASD without intellectual disability (ASD-noID). We modeled differences in TSH levels between ASD or ID status vs. GP controls using odds ratios (ORs) obtained from multivariate unconditional logistic regression. We examined neonatal TSH both as a continuous (ln-transformed) and as a categorical variable based on percentiles. In sensitivity analyses, we examined relationships excluding preterm (gestational age <37 weeks) and multiple births.

Results: In models adjusting for matching factors (sex, birth month and year), birthweight, gestational age, and age at newborn bloodspot collection, we found no association between continuous neonatal TSH levels and ASD (adj-OR: 0.96, 95%CI: 0.76-1.21) nor ID (adj-OR: 0.95, 95%CI: 0.69-1.30). Results were similarly null for models examining percentiles of TSH. Among ASD subphenotypes, we observed an inverse trend between neonatal TSH and odds of ASD with regression; ASD with regression was marginally associated with continuous TSH (adj-OR: 0.76, 95%CI: 0.55-1.06) and significantly associated with the highest quartile of TSH (adj-OR: 0.48, 95%CI: 0.26-0.92). The odds of ASD-noID were also lower with increasing TSH but this relationship was more modest and not significant in models of continuous TSH (adj-OR: 0.83 95%CI: 0.63-1.11) and models comparing the highest to lowest TSH quartiles (adj-OR: 0.68, 95%CI: 0.40-1.14). Neonatal TSH was not associated with early onset ASD nor ASD+ID. Sensitivity models excluding preterm and multiple births yielded similar findings.

Conclusions: While there was little evidence that neonatal TSH is related to overall risk of ASD, our findings suggest that neonatal TSH levels may be associated with particular subtypes of ASD defined by onset type and co-occurrence of intellectual disability. Given that thyroid hormone deficiencies at birth are amenable to therapy, further scrutiny of the intersection of thyroid hormones and ASD is warranted, with attention to ASD subphenotypes.