Neonatal free testosterone and head circumference: need for replication


  • This article is a commentary on Whitehouse et al., pp e73-e77 in the March 2010 issue.

Whitehouse et al.1 report that, in a sample of 82 females whose umbilical cord blood was analysed for androgen concentrations, an inverse correlation (−0.24) was found between free testosterone at birth and head circumference growth (measured both at birth and at 1y). The authors state that this is the first report of such an association. They interpret their finding in the context of previous animal research (from sheep) that exposure to testosterone during pregnancy restricts fetal growth (they have smaller offspring) and a previous human study that mothers exposed to elevated maternal testosterone had infants who had reduced height and weight. They measured head circumference because head circumference shows an impressive correlation with brain size in human children (of 0.93).

I applaud these authors for undertaking an interesting and novel study addressing the important topic of the role of fetal androgens in neural development, simply because of the extensive research showing that fetal testosterone ‘masculinizes’ both neuroanatomy and post-natal behaviour.2,3 However, I was surprised by several aspects of their study.

First, why did they only study this relationship in females? One would have expected clearer results in males since males have a larger range of individual differences in testosterone levels, while females can be so low in testosterone as to be at the limits of detectability. It turns out that the authors study females only because this was originally part of a study of the antecedents of polycystic ovary syndrome in adolescents. This suggests that the idea of testing for a relationship between umbilical cord blood testosterone and head circumference might have been a post-hoc analysis, not a main prediction of their study.

Second, their observed correlation was −0.24 and in Table II this is shown as significant at the p<0.05 level. Whilst this suggests nominal significance, given that they carried out a number of comparisons (21 comparisons are reported in Table II), this could also just be a chance association, since if correction for multiple comparisons had been applied, this would not have met significance at more stringent values of p. As such, I would want to wait to see if this result is replicated in an independent hypothesis-driven study (i.e. just testing this relationship and no other) or after conducting permutation testing.

Third, why expect that testosterone will restrict or reduce growth rather than accelerate or promote growth? I would have expected a positive rather than a negative correlation with testosterone. This is not only because males are on average taller and heavier than females, and on average they produce more testosterone (though these may be unrelated), but also because in our own studies of testosterone measured in the amniotic fluid via amniocentesis we have found that the higher a child’s fetal (amniotic) testosterone, the more autistic traits they show.4 Since a proportion of children with autism have macroencephaly (elevated head circumference), I would, therefore, have expected the correlation between testosterone and head circumference – if there was any correlation at all –to be a positive association.

Fourth, whose testosterone was being measured in the umbilical cord blood? Maternal or fetal? There are ways of separating this but this is not mentioned in the Whitehouse et al. study.

Lastly, what is the relationship between testosterone levels measured at different points in human development? Are fetal levels (e.g. measured in amniotic fluid) related to neonatal levels (e.g. measured in umbilical cord blood) or later circulating levels (e.g. measured in saliva)? Did these authors collect 2nd to 4th digit ratio measures, which we found are a proxy index of fetal testosterone?5 And does testosterone have different effects depending on the timing of its action? Some researchers have argued that the prenatal surge in testosterone production (in the second trimester of pregnancy) has organizational (i.e. permanent) effects on brain development, whilst post-natal testosterone production has activational (i.e. transient) effects on brain function and behaviour.2,3 If true, how big is this critical period window, and does umbilical cord blood testosterone fit reflect the prenatal or postnatal pattern?

In conclusion, this study raises many interesting questions and it is hoped the authors’ novel finding will prompt other research groups to test if the effect they report replicates, and to begin to nail down answers to these very tractable and fundamental issues.