Pregnancy and child

Testosterone linked to autism

“A prenatal screening test for autism comes closer today,” says The Guardian. It reports that scientists have found links between high testosterone levels in the womb and autistic traits in children. It says this could lead to tests that can identify autistic children before birth.

The findings are based on a scientific study of 235 children aged between eight and 10, whose mothers had amniocentesis, a test analysing fluid taken from around a foetus. None of these children were autistic, but those exposed to higher testosterone levels showed higher levels of “autistic traits”, such as poor verbal and social skills.

While this research gives us further insight into the biology behind autistic-like traits, it is important to remember that none of the children in this study were autistic. The researchers must now confirm that their findings apply to children with the condition. Should this prove to be the case, the ethical issues surrounding prenatal screening for risk of autism would need to be debated before any testing could be introduced.

Where did the story come from?

This research was conducted by Dr Bonnie Auyeung and colleagues from University of Cambridge, two Cambridge hospitals, and a university in the US. It was funded by the Nancy Lurie-Marks Family Foundation and the Medical Research Council. The study was published in the peer-reviewed British Journal of Psychology.

What kind of scientific study was this?

This was a cohort study looking at the relationship between levels of the male hormone testosterone in the womb and levels of autistic traits in children.

Studies have suggested that exposure to testosterone in the womb may affect some aspects of cognition and behaviour that differ between males and females. Autism is more common among males, and some people have suggested that the condition is an extreme form of typical male traits.

The researchers identified records from 950 women who had routine amniocentesis in the Cambridge region between 1996 and 2001. The children from these pregnancies would have been aged six to 10 years old at the time of the study.

The researchers excluded certain types of pregnancy from the study. These included pregnancies in which a chromosomal abnormality was identified, pregnancies that ended in termination or miscarriage, pregnancies where there were significant medical problems after birth, or the mother was carrying twins. Cases were also excluded where there was incomplete information, or if medical practitioners felt that contacting the family would be inappropriate.

The remaining 452 women were sent two standard questionnaires, which assessed their children’s levels of autistic traits. These were the Childhood Autism Spectrum Quotient (AQ-Child) and the Childhood Autistic Spectrum Test (CAST).

Of the 452 women contacted, 235 completed and returned both questionnaires and were included in this study. The researchers measured IQ using a standard test in a subgroup of 74 children whose mothers agreed to bring them in for cognitive testing.

The researchers then looked at the levels of testosterone found in the amniotic fluid taken during amniocentesis. The researchers used statistical tests to assess whether there was any relationship between testosterone levels in the womb, and the children’s IQ and levels of autistic traits.

The researchers also looked at girls and boys separately to see if gender had any effect. The researchers also took into account various factors that might affect their results, such as the mother's age, duration of the pregnancies when the amniocentesis was carried out (usually between 14 and 22 weeks), parental education, having an older sibling, and the child's age at the time of the questionnaire.

What were the results of the study?

The researchers found that, as expected, the amniotic fluid in pregnancies carrying males had higher testosterone levels than in pregnancies carrying girls. At age six to 10 boys, showed higher levels of autistic traits than girls.

Children whose amniotic fluid contained higher levels of testosterone had stronger autistic traits, as indicated by higher scores on the CAST and AQ-Child questionnaires. The researchers found similar results if they looked at boys and girls separately on the AQ-Child measure, but on the CAST measure, foetal testosterone levels were only associated with levels of autistic traits in boys, not girls.

There was no relationship between IQ and testosterone levels or level of autistic traits in the subset of children who were tested for IQ.

What interpretations did the researchers draw from these results?

The researchers concluded that their findings fit with the theory that exposure to testosterone in the womb is related to higher levels of autistic traits.

They add that they need to repeat their study in a much larger sample to see if these findings extend to children with autism.

What does the NHS Knowledge Service make of this study?

This study indicates an association between higher levels of testosterone in the womb and levels of autistic traits at age six to 10 years.

There are a number of points to consider:

  • As the authors acknowledge, the association between testosterone levels and autistic traits does not necessarily mean that high levels of testosterone in the womb directly “cause” an increase in autistic traits. Other factors could have an effect. For example, genetic variations might affect both the levels of testosterone in the womb and levels of autistic traits.
  • The samples of amniotic fluid tested were taken at different points in pregnancies, and at different times of day. As testosterone levels are likely to fluctuate over time, it is unclear whether one measurement of testosterone is representative of the foetus’ overall exposure to testosterone.
  • Women who undergo routine amniocentesis are often older than the general childbearing population. The average age of women in this study was 35 years old. Although the researchers took maternal age into account, these results may not be representative of younger pregnant women.
  • None of the children in this study had autism, therefore the authors note that “caution needs to be taken when extrapolating these results to individuals with a formal diagnosis of [autistic spectrum conditions]”. They report that they are currently working on obtaining a larger sample so that they can determine whether their results apply to children with autistic spectrum conditions.
  • The current sample of 235 children was still relatively small. When considering that there was only a 52% response rate among those who were sent the questionnaires, the children may not be representative of the whole group. For example, some parents who had concerns about their child’s development may have felt less inclined to answer a questionnaire about it than those who were happy with their child’s level of development.

Although many newspapers describe the potential for a prenatal test for autism, the authors did not aim to develop such a test. Instead, their aim was to further understand how testosterone may affect development of autistic traits.

Even if such a test were possible, it is important to note that this would be a screening test and not a definitive diagnostic test, i.e. it would identify foetuses more or less likely to develop autism rather than identify those who would definitely go on to develop autism.

Screening tests are rarely 100% accurate, and the many ethical issues surrounding prenatal screening for risk of autism would need to be debated before any test could be offered. Also, there are currently no ways to prevent a child from developing autism. Therefore, it is unclear whether identifying children at greater risk of autism would benefit the child or the parents.

Sir Muir Gray adds...

Screening looks a long, long way away to me.

NHS Attribution