Our health isn't just affected by the things we do after we're born - the conditions we face inside our mother's womb can have a lasting impact on our wellbeing, much later in life. This message comes from a growing number of studies that compare a mother's behaviour during pregnancy to the subsequent health of her child.
But all of these studies have a problem. Mothers also pass on half of their genes to their children, and it's very difficult to say which aspects of the child's health are affected by conditions in the womb, and which are influenced by mum's genetic legacy.
Take the case of smoking. Doing it while pregnant is bad news for the foetus, and studies have suggested that children whose mothers smoke during pregnancy are more likely to be born prematurely, be born lighter, have poorer lung function, and be more likely to die suddenly before their first birthday. More controversially, they may even show higher levels of behavioural problems including autistic disorders and antisocial tendencies.
Biologically, these results make sense, but many of these risks can be inherited too. For example, genetic factors can strongly influence both a person's susceptibility to nicotine addiction and their propensity for violent behaviour. A mother's genes could also affect the birth weight of her child.
To untangle these influences, the ideal experiment would involve randomly implanting foetuses either in the wombs of their own mothers, or those of unrelated women. That's possible in animals but deliberately doing so in humans would be both unethical and impractical. Nonetheless, Frances Rice from Cardiff University realised that this experiment was actually well underway.
Since the advent of in vitro fertilisation (IVF) technology in the late 1970s, many mothers have nourished babies in their womb, who weren't genetically related to them. Here was an ideal chance to study the effects of conditions in the womb, without any confusion caused by shared genes.
Among the many possible detriments of smoking during pregnancy, Rice focused her attention on two - birth weight and antisocial behaviour. Antisocial behaviour has obvious consequences for the child's later life and infants who are very light at the point of birth have higher risks of poor health and chronic conditions as adults.
Rice examined the fates of 774 "test-tube babies", 204 of whom were born to mothers who were genetically unrelated to them. The children are currently between 5-8 years old. Their medical records revealed how heavy they were at birth, and Rice gave their mothers a series of questionnaires to work out if they smoked while pregnant.
She found that babies born to smokers weighed about a pound less than those born to non-smokers. The difference was statistically significant, and it held no matter whether their mothers were genetically related to them or not. Any one of the numerous poisons in tobacco smoke could be responsible for this effect; previous studies have pointed the finger at carbon monoxide, a major part of tobacco smoke, and nicotine, which constricts blood vessels and is a neurotoxin in its own right. Either way, this result provides very strong evidence that smoking directly affects the weight of an infant by changing conditions in the womb.
Rice found a different pattern when she considered antisocial behaviour. Using another questionnaire given to both parents, she catalogued each child's portfolio of disruptive behaviour, from tantrums to stealing to bullying. Overall, the children were more likely to behave antisocially if their mothers smoked than if they didn't, as many other studies have found. But this only held true for related pregnancies - among mums who were unrelated to their children, smoking had no impact on their inclination for troublemaking.
Rice was also careful to adjust her results for many other factors that could have biased them. She took the child's gender and number of siblings into account, as well as the mother's educational level, age, degree of stress in the third trimester, and her own level of antisocial behaviour. Having accounted for this wide assortment of biases, Rice believes that the link between smoking during pregnancy and antisocial behaviour has little to do with conditions in the womb, and is probably due to other factors inherited from biological parent to child.
There is a big problem with this data, which weakens its conclusions. The number of mothers who were genetically unrelated to their babies and smoked during pregnancy was very low - just nine women out of a total of 204. That's a pretty small sample to base rigorous conclusions on but admittedly, it does help matters that studies looking at naturally conceived children found similar links between smoking and birth weight.
Other studies disagree on the subject of prenatal smoking and disruptive behaviour. Some reviews have concluded that the existing evidence supports a true link. Others have agreed with Rice's study, and found that the effect is almost entirely due to a combination of inherited genes and inherited environments. Take away the influence of genetics, the child's social background, their parents' antisocial behaviour, and the connection either disappears or is severely reduced.
Rice's results certainly support this second conclusion, but they aren't strong enough to seal the debate. Future studies would benefit from larger samples, but the method is very interesting and potentially useful. It could be used to answer wider questions about how much of a child's life is influenced by its time in the womb.
Some scientists have tried to separate the effects of nature and nurture by looking at the bodies and behaviours of adopted children, who are genetically unrelated to their parents and even come from different races or nationalities. Rice's method is very similar; the main difference is that the child is "adopted" into a womb while it's still an embryo, rather than into a home while it's a toddler.
Reference:F. Rice, G. T. Harold, J. Boivin, D. F. Hay, M. van den Bree, A. Thapar (2009). Disentangling prenatal and inherited influences in humans with an experimental design Proceedings of the National Academy of Sciences DOI: 10.1073/pnas.0808798106
More on reproduction:
This is really interesting. I wonder if this adds to the argument that epigenetics play a part in autism? Smoking can alter your epigenetic code, and there's evidence that epigenetic mutations contribute to the development of autism.
In trying to separate out the environmental effects of smoking during pregnancy from genetic influences, has it ever occurred to anyone to compare differences between the children of smoking and non-smoking fathers with those between the children of smoking and non-smoking mothers?
If a father smoking has no effect on his offspring but a mother smoking does, then it is clearly directly environmental, whereas equal effects from both parents would show it to be genetic.
Wouldn't this be a far simpler way to clarify the issue than trying to extrapolate from a tiny sample of people whose fertility problems mark them out as so different from the wider population that we cannot be sure the results can be generalised.
The problem with studying smoking, marijuana, suspected environmental toxins, and other harmful and/or addictive substances appears to be the ethical issues in actually giving these substances to study participants. This seems to have precluded RCTs and limited researchers to observational studies.
One strategy for getting around this could be to start with a population of heavy substance users, randomly assign them into treatment and control groups and make substance use cessation the treatment. The study could probably be designed to account for spontaneous dosage changes by subjects in the control arm, but at least for addictive substances, they are unlikely to quit of their own accord. The treatment group could be encouraged, perhaps by the promise of monetary reward, or a reduced sentence to keep to the cessation program, possibly under heavy supervision and frequent blood tests.
This could help clarify things like - whether heavy smoking leads to Parkinson's disease/Schizophrenia, merely alleviates their symptoms or neither. Similarly for the link between marijuana and Schizophrenia. It could also help to distinguish between the genetic and environmental effects in children exposed to tobacco smoke by their parents.