Sunday, December 1, 2013

One Study Shows BPA Safety

Bisphenol A (BPA) is a chemical that is used to soften plastics. It has been used in many flexible plastics (such as water bottles, baby bottles, pacifiers, and the plastic inner liners on some food cans).

Over the last few years, I've seen repeated questions about BPA triggering type-1 diabetes. These worries were not fueled by direct evidence, but rather by indirect evidence. Some researchers consider BPA to be an endocrine disruptor and since insulin is part of the endocrine system, it might be possible that BPA was triggering type-1, or making it more likely to be triggered.

Also, until now, there were no intervention studies that looked specifically at BPA's effect on type-1 diabetes. There had been some general safety testing, but nothing specifically targeting type-1 diabetes.

Recently, however Bodin and team reported on their intervention safety study. This one study found that BPA does not effect type-1 diabetes diagnosis in humans in the amounts that people are exposed to. Although, by itself, this is not proof of safety, it is very reassuring.

The Study

The basic format was this: NOD mice were divided into four groups of 20. One group got no BPA, another got the equivalent of 20 times the amount that humans consume (more on this later), the third group got 200 times, and the forth group got 2000 times as much as people do. Each group was studied both for overall type-1 diabetes rate and the timing of type-1 onset, and also for a variety of internal immune changes.

Obviously, this was a mouse study, and mice are not people. However, when it comes to safety studies, they are the best we have. No one is going to test a chemical on people to see if it is safe. ("Parent, we'd like to give your children a chemical to see if it causes them to get type-1 diabetes, please sign here!") That's unethical. So we are stuck with animals; and NOD mice are the best animal model for type-1 diabetes that we have. [d1] Obviously, I pay little attention to cure results in animals, but for safety results, for intervention studies, there really is no choice.

I discuss the dose levels more below, but for now, the important thing to remember is that the only two dose levels that matter in the real world, are the zero dose, and the 20x normal level. Obviously, if a chemical shows no bad effects, even when 20x the normal dose is consumed, then it's pretty safe.


When measuring rates of type-1, the zero group, the 20x group and the 200x group all had the same overall levels of type-1 diabetes [d2]. The 2000x group was slightly higher.

Also, in terms of when the NOD mice got type-1 diabetes, the zero, 20x, and 200x groups all were diagnosed at about the same time [d2], while the 2000x group was a little earlier.

For the various immunological markers, the results were all over the place. Some markers showed a dose-response change, while other showed no change, and others had peaks in the middle of the dosing range. There might be some work here for immunologists, but I think a more likely results is going to be "if you take a massive overdose of BPA, strange things happen". The same is true of aspirin, salt, water, insulin, meat, water, indeed just about every substance there is. [d3]


How much BPA are real people exposed to?
When I described the tested doses, I described them in comparison to the normal human consumption. However, there is some controversy about exactly what is the normal human consumption. The European equivalent to the EPA has estimated this at 1.5 ug/kg, and that's the number I used. However, this is an estimated exposure, not a measured exposure, and not everyone agrees with it. Some researchers think this estimate is too low to explain the levels commonly seen in blood tests.

So, the most important thing to remember is this: my opinion on this test is based on the European EPA's estimate of human exposure being accurate (or at least within 10x of the right number). If future research shows that the actual human exposure is wildly different, then this research will need to be reevaluated, based on the real dose. But in the meantime, this study shows safety even well above the currently estimated dose.

Is More Research Warranted?
I am fond of saying that one clinical trial never answers a question, because even if the results are very clear, it is still just one study, and more are needed to be truly persuasive. And I think that is true for this research. So in the perfect world, another group would do a similar study, and that second study would confirm these results, and that would be the end of it.

But we live in a world with limited resources, and a huge number of artificial chemicals. So, if someone has money to study chemicals that might cause type-1 diabetes, should they say "well, these other guys suggest that BPA does not cause type-1 (at the exposures that people actually get) so I'll spend that money to do a confirmation study" [d4] Or should they say "With one study showing BPA safety, I'll investigate some other chemical".

There is an "opportunity cost" here. There are 100s or 1000s or even 10000s of chemicals out there. We now have one animal study showing that this one does not affect type-1 diabetes in doses even 100s of times bigger than people actually get. If we get funding for another study, why not put it into a chemical that has zero animals studies done on it?

For BPA specifically, there is an even more pressing question: how much BPA are real people exposed to? Maybe a better place to put funding is to determine the real human exposure level, rather than run another test on a dose which might be way off?

What are the ethics of using mega-overdoses in safety testing?
To give you some idea of the scale of this research, consider that a normal dose of aspirin is 1 pill. Now, if someone did a safety study and told you that aspirin was unsafe, because they had given the equivalent of 2000 pills to a mouse and the mouse had problems, your reaction might be to ask them why they gave the animal such a huge amount [d5]. You probably would not be worried about the safety of Aspirin, nor should you be, especially if they told you Aspirin was safe at 20 and even 200 pill doses. Yet, that is exactly what these researchers did, and many reporters are trumpeting the danger, based on the mega-overdose they gave.

For me, a huge question in this research is, why did the researchers choose such a silly dosing level? I mean 20 times actual dose is high, but reasonable because it gives a big safety margin.  But 200 times and especially 2000 times average exposure seem so high as to be a joke.

What troubles me about this experiment's design is the following: Everyone (including researchers) should know that headline writers ignore dose levels. So a researcher can manipulate a headline writer (and often the reporter, as well), by including a dose level so high that something bad is bound to happen. Then the headlines will scream that there is a danger, even if the researcher knows the danger only exists with exposure levels so high they never happen.

In future, does type-1 diabetes drop in countries that ban BPA?
If using BPA increases the type-1 diabetes rate, then removing BPA would decrease the type-1 rate.

Another type of research which might be interesting here, would be to see if type-1 diabetes cases start to drop, or the average age goes up, in countries were BPA has been banned. Data from those countries could be examined for 10 to 20 years after the ban. This type of study (a population-based study) is not as well controlled as an intervention-based study, but it is done in people, so (if type-1 numbers continued to go up or stayed the same), it would provide some support that BPA is not involved. Comparing different countries' consumption of BPA over different years might provided a range of interesting data.  Conversely, if the numbers go down, that would suggest that maybe BPA was involved, especially if they went down the same number of years after BPA stopped being used in each country.

Summary of BPA Safety

The one intervention study we have, where BPA was directly tested in animals for type-1 diabetes, found no increase in type-1 diabetes, nor any earlier diagnosis of type-1 diabetes, even when given in doses hundreds of times higher than estimated human exposures. This was just one study, and there is some controversy about how much BPA people consume. Of course, it is always possible that BPA might be dangerous in other ways; this discussion is focused on type-1 diabetes. 

Extra Discussion

[d1] This is a general problem with all safety research: we want to know safety in people, but are forced to test safety in animals.

[d1b] The difference in outcome was not statistically significant, in both these cases.  With only 20 NOD mice, the numbers were usually not exactly the same, nor would they be expected to be exactly the same.  However, the differences were never statistically significant, which is what matters in these kinds of tests.

[d1c] Even water would be toxic at 2000x a normal amount.  If a normal amount of water is 1 gallon a day, then 2000 gallons a day could easily cause death.

[d4] For safety testing in general, people who don't like the results, often want to rerun the test in a different animal. (No matter which side they are on.)  But in this case, there are very few animals that get autoimmune-based diabetes (ie. a diabetes similar to human type-1), so there are not a lot of choices, and NOD mice are generally considered the closest.

[d5] For another comparison, the FDA wants us to eat about 2 grams of salt per day. The actual average is 3 grams per day. Would you do safety studies based on 6 kilograms per day? Would that study have any connection to reality at all? 6 kilograms is over 13 pounds.

Joshua Levy --
publicjoshualevy at gmail dot com 
All the views expressed here are those of Joshua Levy, and nothing here is official JDRF, JDCA, or Tidepool news, views, policies or opinions. My daughter has type-1 diabetes and participates in clinical trials, which might be discussed here. My blog contains a more complete non-conflict of interest statement. Thanks to everyone who helps with the blog.