| Growing controversy swirls around research article key to FDA's decision on bisphenol A safety. | ||
In 2008, Rochelle Tyl along with multiple co-authors, including Steven Hentges, a spokesperson for the American Chemistry Council on BPA safety, published a research article in the peer-reviewed journal Toxicology which reported to find no adverse effects of bisphenol A at low doses.
Even before publication, these data were playing a prominent role in agency decisions on BPA safety and in testimony before legislative bodies by Hentges and others. The results were specifically cited in the FDA's draft 2008 finding that BPA was safe as the key data that trumped many other studies that had found adverse effects. The principal reasons given for leaning principally on this study were two: it employed a large sample size and it followed Good Laboratory Practices (GLP).
In March 2009 Tyl et al. was heavily criticized in a peer-reviewed commentary published in Environmental Health Perspectives, coauthored by 36 scientific experts on BPA from 6 countries on 3 continents. Although GLP requirements ensure that laboratory procedures and practices are well documented, they provide no assurance of the quality of the science. This became very clear as a careful analysis of Tyl et al (2008) in the commentary revealed flaws so severe as to render it useless in regulatory proceedings. Indeed, the commentary concluded that 4 of the four industry-funded GLP studies of BPA have fatal scientific errors.
Tyl et al.'s problems actually took a turn for the worse when, at the September FDA hearing on BPA, she first asserted that the animals in her study were 6 months old but it was then pointed out at the hearing that the paper itself said they were 3 months old. She had offered the 6 month age because one of the key measurements in the study was prostate size. Prostates grow as an animal ages, so if she could argue that they were much older then it would be at least plausible to argue that her control animals' prostates were larger than those of all other scientists using this strain. But it turns out her control prostates weren't just a little larger. They were over 70% larger than prostates of 3 month old mice from virtually every other lab that uses this strain of mouse. This would make it very difficult to detect the 35% increase in prostate size that had been reported in two other previous studies. And for a prostate of an adult mouse of this strain to be that large it would have to be visibly diseased.
Tyl compounded this problem at a recent meeting in Germany, reported on in the Milwaukee Journal Sentinel on 12 April 2009. According to scientists present at the meeting, when asked about the age of the mice she said they 5 months old. When asked directly, "Are you now saying that what you reported in the paper was not accurate." She said yes.
Her conflicting answers raise serious concerns about the entire paper. What is the correct age? What other errors are in the published paper? Why would she offer two different ages in testimony before two separate federal hearings, first in the US and then in Germany? If GLP procedures ensure that lab practices are well documented, why was the original paper flawed?
Tyl asserted at the meeting that many papers have shown CD-1 mice (the strain used) to have prostates averaging 72 milligrams, as did hers. She has been asked to provide those references.
Moreover, the Journal Sentinel goes on to report on a polycarbonate fire that occurred in Tyl's lab approximately 3 years prior to initiating these experiments. This fire could have led to widespread bisphenol A contamination throughout the laboratory, leading to all animals being hit with doses of BPA that, according to most research on BPA (but not hers) would lead to an enlarged prostate (along with other effects). With all animals, including control animals, affected by BPA, it would be impossible to detect the consequences of additional BPA exposure.
According to the Journal Sentinel:
"Tyl's company, RTI International, investigated the fire - which appeared to be arson. An independent firm analyzed the lab and reported that it had been sufficiently cleaned. The group declared that the animals were not compromised and that Tyl's lab "was, and continues to be, an excellent facility." But scientists at the conference, when told of the fire by the Journal Sentinel, said they would not trust that assessment. BPA is a sensitive chemical that acts at extremely low levels, and the group that analyzed the lab did not test for the chemical at those levels, they said. Vandenbergh, of North Carolina State, who examined the reports on the fire and the laboratory's investigation for the Journal Sentinel, said he would have conducted the experiments at a different lab. "In hindsight, this is too important of an issue to leave open these questions," Vandenbergh said." |
Based on the scientific flaws in the publication and on the contradictory statements from Tyl about key details, two steps would appear appropriate:
- The U.S. House Commerce Committee should extend its current investigation of FDA's review of BPA to include a thorough examination of Tyl et al.'s original data; and
- The journal in which the paper appeared originally, Toxicology, should launch an immediate review of the paper to determine whether or not it should be retracted.