Wednesday, February 27, 2013

Bisphenol A: Forgotten lessons from Science 101

Bisphenol A (BPA) in the plastics that come into contact with our food is responsible for a lot of what ails us, from infertility to diabetes.  Isn't it? 

BPA
BPA is a chemical that has properties that resemble estrogen, and as such is thought to be a risk factor for "diseases of westernization" such as heart disease, diabetes, hypertension and stroke, although it must be said that study results are equivocal, some showing an association and some not.  BPA as a component of baby bottles was thought to be particularly worrisome, as early exposure has been suggested to be responsible for neurological disorders, with other who-knows-what longterm effects.

But now a speaker at the American Association for the Advancement of Science (AAAS) meetings that took place in Boston this month suggests that while there may indeed be an association between blood levels of BPA and illness, it may not be causal.

The Independent reports:
A more likely explanation is that obese people who overeat - and who develop chronic illnesses as a result - are also more likely to have raised levels of BPA in their bloodstream from plastic food packaging, said Professor Richard Sharpe of the Medical Research Council’s Centre for Reproductive Health in Edinburgh.
According to Sharpe:
“None of these studies actually shows that bisphenol A exposures causes the disorders, the exposure is simply associated with occurrence of the disorders,” he said.
“If this association was due to cause and effect, it would mean that bisphenol A was incredibly potent and toxic, and this does not agree with published studies. This possibility therefore seems illogical,” he added.
So, according to Richard Sharpe, the correlation is nothing more than an indicator that obese people, who are at higher risk of chronic diet-related diseases, simply have higher BPA blood levels because they eat more food that comes into contact with plastic.  But it's the food itself that's causal, responsible for obesity, which then causes the diseases for which BPA is being blamed.  And, Sharpe says, studies show that the BPA levels in blood are far too low for the chemical to be responsible for disease. 

Sharpe has been arguing his case for several years now, including in a piece, in The Independent again, in 2010 in which he said that the original animal studies that showed an effect were not replicable, and that implanting BPA under the skin and injecting it into the blood, which is how many animal studies have tested its effect, isn't representative of human exposure.  We ingest it, and it is thus broken down in our gut.  

But, on another public platform, Nicholas Kristof argued strongly in his widely-read column in the New York Times last August ("Big Chem Big Harm") that BPA is known to be responsible for many diseases and disorders, including autism, hyperactivity, aggression and depression in children, and breast cancer and diabetes in adults. He concluded that regulation of chemicals must be much stricter than it is.  


We aren't here to adjudicate.  But, this is rather astonishing, really.  One thing that the reported correlation has unquestionably caused, and Kristof's op/ed is an example, is a lot of fear, and BPA has been removed from baby bottles, and a lot of BPA-free choices are now available for other containers, like water bottles and the like.  How could the "science" be so backward here that apparently no one thought to consider the "Correlation doesn't equal causation" adage, something everyone should have learned in Science 101? 

Yes, determining causation can be problematic. But still, it's rather astonishing that simply questioning the direction of the association is enough to get a speaker a platform at a major scientific conference. 

Causation is a problematic concept, and its nature is far from obvious
Causation is a very complex subject that has been debated since, well, since there were  debates.  The classical idea was that some 'thing' is responsible for another 'thing' in the sense that if the first is present the second will ineluctably follow.  That's causal determinism and is what people generally have in mind.  Some argue, and have for a long time, that this is an oversimplification and truly deterministic properties, or laws, of nature are an illusion.  More accurate, from that point of view, would be a more probabilistic notion of such antecedent-consequent relationships.

One thing has been essential from the beginning, and that is a time ordering: cause must come before effect.  When and how time has such a clear-cut arrow is a subject for other times and isn't trivial.  However, here, the issue can be seen by the common idea of causation that when A always precedes B, and whenever A's around, B is sure to come, and thus A 'causes' B.

But it is easy to see why correlation between two factors--even if one can order them--clearly doesn't mean causation.  In the above example, plastic and over-eating both precede the subsequent diseases.  If all we observe are the plastic, we can make that the A in A+B.  But once we know that there is also a C, that also precedes B, it becomes more challenging to assign actual cause....whatever that is.

Unfortunately, until a mechanism is discovered we can't always decide whether A or C, or even some other correlate, are causal in regard to B.  This is a quandary that affects many areas of science, but especially observational ones like epidemiology, where or if we can't set up a controlled experiment to investigate the question.

5 comments:

  1. Judea Pearl's work on causality helps blow away a lot of the mental cobwebs on the construct. Although his book (Causality, 2nd Edition) is a challenge, even for the most mathematically sophisticated, his earlier essays and presentations are not. Check out: http://bayes.cs.ucla.edu/IJCAI99/ijcai-99.pdf, http://singapore.cs.ucla.edu/LECTURE/lecture_sec1.htm and, my favourite, http://bayes.cs.ucla.edu/R264.pdf

    ReplyDelete
  2. On the topic of BPA, while it's an excellent point that the chemical nature of a substance can change in the processes of ingestion, there have been animal studies that look at effects following BPA ingestion:

    http://onlinelibrary.wiley.com/doi/10.1034/j.1600-0722.2002.11201.x/full (for example)

    Also, the studies which measure BPA levels in the urine or in the blood for that matter-- i.e., post-digestion-- would suggest that it or a very close metabolite is present and not degraded by the digestion process. Meaning that those studies which bipass the digestive tract may still hold relevance for human health.

    I only have to think of 5-6 legged or two-headed frogs, shifts in sex ratios, weak-shelled birds eggs, etc., which are known to occur following environmental exposure to various xenosteroids in order to entertain the idea that BPA, or a metabolite, is targeting the estrogen receptors in our own bodies.

    Sharpe has a point that obese people may simply be consuming more BPA-packaged products and the obesity itself is the true hazard. However, that is only a single possibility and, based upon my knowledge of the relevant research, it's probably a combination of the two, not one nor the other.

    I think I would trust Sharpe more (and Kristof for that matter) if they both didn't have blatant agendas and hence overwhelming potential bias.

    I would also probably trust researchers from the fields of cancer research who've been studying environmental xenosteroids for some time in the hopes of solving and preventing cases of cancer. Even in reference to my own field of autism research, cancer research is streets ahead in terms of molecular and cellular biology and epigenetics.

    ReplyDelete
    Replies
    1. Thank you for these very valid points. My intention was not to get into arguing for or against BPA as a harmful substance, however -- as you point out, the evidence can be, and has been argued either way, often based on political bias, and I didn't mean to wade into that.

      My primary aim here was to highlight the fact that something as simple as knowing whether A causes C, or B causes C and also causes A is often not really so simple. And probably if that were acknowledged by both the A -> C people and the B -> C and A people, this might lead to at least the acknowledgement from both sides that what seems so simple is in fact complex.

      Delete
  3. Yes, I managed to not comment on the actual point of the post, a particular talent of mine but got distracted by the details, lol. Inevitably happens when a favorite topic is touched upon. Digression. ;)

    ReplyDelete
    Replies
    1. You know much more about the actual topic than I; I'd love to see you post on it!

      Delete