Is Common Food Packaging Giving Children Autism?
The new study on BPA and why the evidence is "less than compelling"
Bisphenol A (BPA) is a perennial source of concern. It’s a common chemical that is used to manufacture a variety of plastic containers, some of which are used for food. There’s been a range of worries over the last few decades that BPA could be causing a range of negative health outcomes, because some portion of the chemical may leach into food, and that can get into your body when you eat the food.
This week, a new paper has come out that seems to confirm the worst fears about BPA. According to the headlines, the new study shows that boys exposed to BPA may be six times more likely to develop autism. Unsurprisingly, parents around the world are now terrified that their children may be getting poisoned by the BPA in their snack boxes and water bottles.
But the reality of the research is far more complex than the story you might’ve heard in the headlines. While this was a careful and interesting study, it’s not clear that the data shows a direct link between BPA and autism, and even if it does this is only for a very small subgroup of the population.
Let’s look at the science.
The Science
The study - or, technically, studies - in question is a recently published paper looking at a range of experiments that review the association between BPA and autism. There are a total of 25 experiments in the series, many of which were really interesting pieces of scientific work.
It’s hard to summarize all of these experiments in a few paragraphs, but the basic points are:
The authors showed that there was an association between higher BPA exposure and autism in a specific subgroup of boys. These boys had a combination of genes leading to a lower level of an enzyme called aromatase, which impacts hormone activity in human brains and may be linked to autism.
Next, the authors looked at cells in petri dishes. In this experiment, the authors found that BPA exposure reduced aromatase expression in these human cells.
Then, the authors did a series of experiments in mice. They showed that exposing pregnant mice to BPA reduced the number of brain cells in their male offspring that produce aromatase. They also showed that male mice whose mothers had been exposed to BPA during pregnancy had slightly different social behaviours, and that this was also the case if they studied male mice with a genetic lack of aromatase.
Finally, the authors looked at a specific series of mechanisms that could explain this. They found that the relationship between BPA and brain cell activity was possibly mediated by a chemical called 10HDA, and that giving mice whose mothers had been exposed to BPA while pregnant high doses of this chemical caused some of the difference in behaviour seen in the BPA group to disappear.
Basically, these experiments show that BPA may be associated with autism in humans, and lay out a really interesting series of experiments that show what might be causing this association.
However, there are some BIG caveats that have been almost entirely unmentioned in the reporting of this study.
Firstly, the associations seen in this paper are only for a very specific group of children. The human studies looked at a population of 605 children whose mothers had been tested while pregnant for BPA. There was no association at all between BPA and autism symptoms at age 2, or autism diagnoses at age 9, in the group as a whole. There was also no association between BPA exposure and autism in boys or girls as overall groups.
The only group that showed an association between autism and BPA was boys who had low aromatase. Specifically, they showed that boys in the top 25% of BPA exposure who also had at least 3 of the 5 genes that were associated with aromatase expression had an increased risk of autism symptoms at age 2 (p=0.03) and diagnosis at age 9 (p=0.05).
While it’s true that this relationship was statistically significant, it’s also quite hard to infer meaning from these estimates because the authors did not have enough data to really model the relationship properly. For one thing, rather than running a complete model with all of the potential confounders adjusted for, they ran each confounder in their model separately. They also dichotomized most of these potential confounders (i.e. they divided parental income into above/below $100k), which limits what we can take home from these associations further.
As a standalone epidemiological study, the data is extremely weak. At best, I’d describe this as a very vague association that requires a great deal of further research to evaluate.
The other main findings were related to mice. Giving pregnant mice BPA definitely did impact their offspring’s development, which is very interesting. There are also a few issues with interpreting these results as well.
For one thing, the levels of BPA exposure were extremely high. The authors gave the mice a dose of 50 micrograms/kilogram as an injection while they were pregnant. This was based on the highest allowable dose of BPA that humans could be exposed to legally in Europe in the mid-00s (when the authors note that the women in their human studies were mostly pregnant).
But this isn’t the average dose of BPA that people were getting in that period - it’s the highest legal dose that people could get at all. In fact, the same assessment that I linked above - which was cited by the authors as the reason for the 50 microgram/kilogram dose - notes that the average exposure to BPA in Europe at this time was likely about 50x lower than this dose.
Moreover, the assessment also notes that this dose is only for oral exposure. In other words, this is for BPA in foods that we eat. The studies on mice did not feed them BPA, they injected it. While almost all injected BPA goes into the bloodstream, only about 5% of the BPA that rodents eat eventually makes it into their blood. In other words, not only was this dose about 50x higher than the average amount people are exposed to, due to the method of exposure it’s probably more like 1,000x higher than what most people actually get in their daily life.
In addition, while the mice in the study whose mothers had been exposed to BPA did have different behaviours than the other mice, they weren’t that different. There was a lot of overlap between the BPA and non-BPA groups of mice on all of the things that the studies measured.
It’s also hard to draw a comparison between the mouse behaviours and human autism. Autism is a very complex human diagnosis which you can’t really test for in mice. Instead, the authors looked at how mice responded to new stimuli, and also how often they engaged in repetitive behaviours. It’s unclear how much these measurements might be related to human autism, if at all.
Bottom Line
The point here is not to dunk on the study - it’s a really interesting and generally thorough series of investigations that is definitely more rigorous than the average paper you see on autism and environmental pollutants. It’s possible that the series of mechanisms proposed by the authors does actually happen, and that this is indeed causing autism in some boys.
However, the data is at the moment still fairly slim. We’ve got a series of possible mechanistic triggers, most of which at least need to be confirmed by further research for us to have confidence that they are real effects.
It’s also important to remember that the association seen in real humans here is very weak, and was only evident for a really small population. Remember, the only group that had any increased risk of autism at age 9 was boys (not girls) who had both high BPA exposure AND low aromatase activity. Even if the study is entirely correct, and BPA can cause autism in some boys, it would be a very small minority of the population who would even be susceptible to this issue.
Ultimately, it’s possible that BPA is causing autism, but to quote Dr. Ian Musgrave of the University of Adelaide the evidence at this point is “less than compelling”. What we’ve got is a weak association and a series of studies in lab mice, but that may not mean all that much to real humans worried about BPA in their food and drink.
Personally, I won’t be changing anything in my life based on these findings. If you’re really worried, the main source of exposure for BPA is plastic packaging, particularly for things that are heated up (like stuff you put in the microwave), so check to make sure that those things are BPA-free when you buy them.
This is so problematic, and offensive to the autism community that this is even being studied. It is as though autism is something dire health problem that requires finding some sort of "cause", rather than just a natural variation of how people are.
But nothing can "give people autism" since it's a genetic condition. I'm autistic, I don't like seeing misinformation spread about us.