This post, by Sarah Vogel, is cross-posted from The Pump Handle.
If you thought the scientific debate about bisphenol A was over or even quieting down, you haven’t been reading the latest issues of Toxicological Sciences. (What are you doing with your spare time?) Last month in an editorial piece published in the journal, Richard Sharpe queried: “Is It Time to End Concerns over the Estrogenic Effects of Bisphenol A?” His answer was an unequivocal ‘yes’, based on the latest study from Ryan et al. (published in the same issue) that found no reproductive effects from bisphenol A exposure in rats. The study, according to Sharpe, “throws cold water on this controversy.”
Not so fast. On Wednesday, February 17, 2010, the journal published a second letter to the editors, “Flawed Experimental Design Reveals the Need for Guidelines Requiring Appropriate Positive Controls in Endocrine Disruption Research,” by Fred vom Saal and 23 other researchers. In a position quite contrary to Sharpe’s, the letter pointed to an important design flaw in the study.
This latest iteration of the controversy is about a fundamental and persistent challenge in the research on bisphenol A and other endocrine disrupting chemicals—what is the appropriate study design. Issues of animal selection, route of exposure, animal feed and housing, and appropriate use of positive controls all point to the complexity of studying extremely low levels of endocrine disruptors.
These are not trivial issues. Proper study design is essential to conducting good science, and charges of inappropriate design have been used to discredit research findings of adverse effects of as well as no effects of bisphenol A at low-doses.
This most recent letter critical of the Ryan et al. paper points to a flaw in the selection of dosing levels for the positive control. To understand the argument requires a basic understanding of a positive control.
Let’s start with a very different example recently shared with me. Say you have an unknown substance in the garage that you think might be a fertilizer. To test this hypothesis you take two dishes of seeds and treat one with water and the other with the unknown substance and see what happens. When there is no growth in the seeds of either treatment, you could conclude that the substance isn’t a fertilizer. But what if the seeds you had started with were already dead? So, you decide that a better way to test the unknown substance is to take three dishes of seeds and treat one with water, one with the unknown substance and a third with Miracle-Gro. This way if the seeds grow with Miracle-Gro you know they’re alive and reactive to fertilizer. And if the seeds grow with the unknown substance and the ones in the water don’t, you know it’s a fertilizer. In this example, Miracle-Gro is the positive control and water the negative control.
But what happens if it turns out that when testing your seeds, you find that you have to use 10 times the amount of Miracle-Gro that is recommended to make the seeds grow? What if your unknown substance is a fertilizer, but over the years of sitting in your garage has become less potent? Your seeds might grow if you were to use a ton of the substance but because you didn’t have that much you only tested a small amount. So in your study you use water, a small concentration of your fertilizer and a ton of Miracle-Gro. You find that only the seeds doused with Miracle-Gro response positively. This could lead you to incorrectly conclude that the unknown substance is not a fertilizer.
How does this apply to the critique of the Ryan et al. paper? In their study the researchers used a positive control, ethinylestradiol (EE), the hormone used in birth control. However, they had to use extremely high doses to trigger the reproductive endpoints of interest (e.g. sexually dimorphic behavior, age at puberty, reproductive function). They basically had to douse these rats with this estrogen to elicit a positive response—that is, to make the positive control work. If they had used lower concentrations they would have observed no effects. At the lowest concentration of EE used, 5 µg/mg, Ryan et al reported no effects in the animals. The letter by vom Saal et al. noted that the pharmacologically effective dose of EE in oral contraceptives used in humans is less than 0.5 µg/mg. So the rats used in Ryan et al.’s lab were not sensitive to levels of EE above the concentration used to avert pregnancy in women.
Given that it took such high concentrations of the positive control to elicit a positive effect, the researchers should have selected much higher concentrations of BPA. This is because as demonstrated by the positive controls, the animals are insensitive to estrogen.
The details of this scientific debate can be confusing, particularly for the non-scientists. But what it unsettling about Sharpe’s commentary is that he turned what should have been a scientific debate into a theological discussion. Sharpe used the recent findings by Ryan et al. to paint a simplistic picture of the bisphenol A debate as a struggle between rational facts observed by scientists and exemplified in the Ryan et al. article, and “believers” of false hypotheses propagated by “nonscientific” media, blogs and website. According to Sharpe, these “believers” argue that bisphenol A has estrogenic effects at low doses, whereas scientists prove otherwise.
Simplifying a complex scientific debate using theological arguments ironically achieves the exact opposite intent of Sharpe’s editorial. Rather than pulling this debate out of the mire, Sharpe drags it back in, and in the process, pulls it away from the rich scientific debate that needs to occur, particularly in the pages of scientific journals.
It is a credit to the 24 bisphenol A researchers that they did not take offense to this characterization of their work and careers. They chose to stick to the scientific aspects of the debate and provide recommendations for appropriate use of positive controls.
In publishing the recent letter to the editor, Toxicological Sciences pushes scientific progress forward one small step by encouraging healthy and constructive skepticism and scientific debate. Now we must wait for the next installment: a response from Ryan et al.
Sarah Vogel received her PhD from Columbia University in the Department of Sociomedical Sciences’ Center for the History and Ethics of Public Health and Medicine; her dissertation was entitled “Politics of Plastic: the economic, political and scientific history of bisphenol A.” She holds master’s degrees in public health and environmental management from Yale University. She authored the case study “Battles Over Bisphenol A” at DefendingScience.org.