Elevated Fetal Steroidogenic Activity in Autism

Background: Autism Spectrum Conditions (ASC) are much more common in males. One hypothesized biological mechanism that could potentially influence this male bias is fetal testosterone (FT) or more generally the broader steroidogenic pathway leading to the synthesis of testosterone (Baron-Cohen et al 2011, PLOS-Biology; Baron-Cohen et al, 2005, Science). Sex steroids are well established as an epigenetic fetal mechanism for modifying gene expression and a host of other molecular/cellular factors in early brain development and may be helpful as predictive markers for those who may be at increased risk for later diagnosis. 

Objectives: To test for the first time the hypothesis that fetal steroidogenic activity is elevated in individuals who later receive a diagnosis of ASC. 

Methods: 62 male cases of classic autism (without a comorbid diagnosis of ’mental retardation’) or Asperger Syndrome and 231 typical male controls were selected from the Historic Birth Cohort, a biobank of amniocentesis samples taken from the Danish population since 1993. Using mass spectrometry we assessed the concentration of 4 hormones in the Δ4 steroidogenic pathway tied explicitly to CYP17 enzymatic pathway (i.e. progesterone, 17α-hydroxyprogesterone, androstenedione, and testosterone) in amniotic fluid sampled during weeks 10-20 of gestation. Cortisol was also measured as a control hormone that is not within the main Δ4 sex steroid biosynthesis pathway. Analysis of the main hypothesis (that Δ4 pathway hormones are elevated in ASC) consisted of computing the multivariate Wilk’s lambda statistic within a permutation test (re-computed over 1,000,001 iterations). Further classification analyses were implemented using logistic regression and classification performance measures were compared to null distributions estimated under chance conditions via permutation tests. 

Results: A permutation test (1,000,001 iterations) on the multivariate Wilk’s lambda statistic showed that when testing all hormones there was an overall group difference in the direction of ASC>Control (p=0.01). Following up this multivariate result with tests on each hormone individually, we found that concentration of all 4 steroidogenic hormones in the Δ4 pathway were elevated in the ASC group, but there was no between-group difference in cortisol concentration. Logistic regression was then used to classify diagnostic status using all hormones. Classification accuracy, specificity, PPV, and NPV were all significantly higher than chance values estimated by permutation tests (all p<0.02). Sensitivity approached statistical significance (p=0.06).

Conclusions: This work represents the first direct verification that fetal exposure to sex steroids is elevated in those who later receive a diagnosis of autism. Given the role of sex steroids in a host of interactions at the genetic and molecular/cellular level, this finding represents an important breakthrough in understanding early factors that contribute to the pathophysiology of ASC. Classification analyses show that while these markers are statistically significant in predicting later diagnosis status, such markers should not be used as a fetal test. Future work comparing autism to other neurodevelopmental conditions will be important in clarifying the specificity of such markers to autism, and how such fetal hormones impact the neurodevelopment of autism.