By Lewis Perdue (Corresponding author: firstname.lastname@example.org) and Becca Yeamans
A new manuscript accepted for publication in the journal Toxicological Sciences (1) is fatally flawed in numerous ways and should be withdrawn from publication.
That study, "Toxicity Evaluation of Bisphenol A Administered by Gavage to SPRAGUE-DAWLEY Rats from Gestation Day 6 through Postnatal Day 90" concluded that low-dose BPA is not harmful.
However, it suffers from numerous and massive errors including one of the most basic scientific mistakes: all the test animals -- including controls -- were contaminated by BPA, something that invalidates any conclusions and taints the equally contaminated data.
Despite that massive error the authors buried their admission of the issue deep within the 37-page paper.
And, while the study authors claim to have no idea where the BPA contamination came from, an examination of the paper indicates numerous avenues of contamination that are well-established and easy to locate, and should have been considered in any careful experiment design and conduct.
In addition to the contamination error, the study fails on an accumulation of other serious -- yet easily avoided -- self-inflicted wounds:
· Inappropriate Rat Strain
· Poor choice of outcome measures
· Failure to use current science in BPA and EDC research
· Erroneous claim that BPA is “weak” estrogen
· Failure to identify well-known exogenous source(s) of BPA contamination
· Failure to identify well-known exogenous source(s) and estrogenic acitivty of equipment and supplies
· Failure to measure BPA levels in any of the animals except in the test cohort at the end of the trial
· Failure to measure other exogenous estrogenic compounds in test animals
· Failure to consider Non-Monotonic Dose Response
· Use of oral gavage
· Wrongly examining toxicity when timing is a bigger issue
· Failure to include disclosure of conflicts
A negative or naïve control is defined as a group of experimental units that have not had any experimental treatment applied, nor possess any quality that would have an effect on the results of the study. In the study under the microscope, in order to be a true negative control, all animals would need to be devoid of any BPA or other endocrine disruptors (EDC) in their body that interact with or act similar to BPA. Unfortunately, this was not done, thereby rendering all further experimental results invalid. The paper itself stated that:
“BPA-glucuronide was detected in the serum of vehicle and naïve control animals at levels similar to those detected in animals dosed with 2.5 ug BPA/kg bw/day. The presence of BPA-glucuronide, which is formed by metabolism of aglycone BPA in the body, in the serum of control animals implies unintentional exposure of the animals to BPA rather than post-sampling contamination of the serum samples. The source of this exposure has not been definitively identified. Thus, the 2.5 μg BPA/bw/day dose group is not distinguishable from the two negative control groups.” (1).
This finding automatically violates one of the most important functions of a control, in that no longer can the researchers determine if the low-dose BPA has an effect on the specific endpoints described in the paper (2), since ALL rats, even the supposed negative controls were contaminated with BPA prior to experimentation.
If there are no differences in the low-dose BPA group and the control group, it is impossible to make the conclusion that low-doses of BPA do not have any effect on the rats when there is no real naïve control to compare to in the first place (2). Additionally, without the comparison of a positive control (i.e. a control treated with BPA, in this case, that is known to promote adverse effects), it is even more difficult to determine if the negative result is real, or if it is a false negative (3).
Inappropriate Rat Strain
It has been found that there are clear species and strain-specific differences in sensitivity to various EDCs (4,5), so using an animal model that has been shown to respond to the compound(s) in question is crucial for accurate interpretations of results. Specifically, depending upon the strain of rat used, results for similar experiments have been found to be markedly different.
Different strains of rats appear to have varying sensitivities to compounds with various estrogenic activities, with Charles River Sprague-Dawley rats, studies have found that they are very insensitive to BPA and other estrogenic compounds, thus using them in low-dose studies is inappropriate and not representative of how the compounds act in more sensitive strains or species (4,5,6) In the FDA study under review, they, too, utilized Sprague-Dawley rats as their animal model.
According to the results of one meta-analysis of all BPA literature through the end of 2004, 90% of government-funded peer-reviewed research articles found adverse effects of BPA exposure. A side note: the same meta-analysis discovered that of the industry-funded research articles on BPA, 0% found any adverse effects of BPA exposure, which raises more red flags.
Taking only the government-funded peer-reviewed research articles in consideration, when those studies that used Charles River Sprague-Dawley rats were removed from the meta-analysis, the number of studies finding adverse effects of BPA exposure increased from 90% to 96%. This jump indicates that the use of the appropriate species be used in BPA experiments, as some strains (including Charles River Sprague-Dawley rats) are not sensitive enough to accurately reflect what could be occurring in the human model (6).
Poor choice of outcome measures
Another flaw with the FDA study is that they chose to focus primarily on the toxic effects of BPA exposure on a set of anatomical outcome measures that has been shown to be unacceptable and unrepresentative of BPA or other EDC exposure over the years. A major part of the problem is that studies by federal agencies are required to follow “Good Laboratory Practices”(GLP) guidelines, which often necessitate the use of multi-generational tests in order to stand up in a court of law if needed (6).
The problem with GLP multi-generational testing regimes is that they use out-of-date testing methods and outcome measures which do not fully capture the effects of BPA exposure, and completely ignore the more contemporary methods and outcome measures like gene expression, DNA methylation, cellular pathways, or epigenetics (6).
Specifically, the FDA study ignored current science that indicates that DNA methylation and other epigenetic effects along with interference with cellular signaling pathways are responsible for more relevant biological phenomena that affect obesity, cancer, Alzheimer's Disease and other maladies promoted by BPA and other Endocrine Disrupting Compounds (EDCs).
Including the FDA study current under review, these types of studies focus mainly on anatomical measurements, gross pathological observations, and other macroscopic outcome measures (1). Since the 1990s, there have been vast improvements in BPA and EDC research methods, with the understanding that out-of-date GLP methods established as early as the 1950s are not sensitive enough to detect effects of low-dose exposure of BPA and other EDCs. However, federal regulatory agencies still insist on following out-of-date protocols that were created prior to the improvements in our understanding of BPA and EDC exposure on more sensitive measures (DNA methylation, cellular signaling pathways, etc.) (7).
Failure to use current science in BPA and EDC research
Even though GLP-based testing regimes do not accurate reflect the harmful public health hazards of BPA or other EDC exposures, federal regulatory outcomes and decisions are still based on those studies using GLP methods and not based on the many hundreds of academic studies that use more modern techniques and have overwhelmingly found harmful effects of BPA and other EDCs on various human health aspects (6,7).
Both the United States FDA and European Food Safety Authority have in recent years published information showing that their conclusions regarding the safety of BPA has been based solely on the results of a small number of studies following GLP guidelines, while ignoring the expert advice and research of over 700 peer-reviewed academic articles as of 2006.
Since that time, there have been many hundreds more articles demonstrating harmful public health effects of BPA and other EDCs, while the FDA and other federal regulatory agencies continue to ignore this mountainous evidence and respond with their own poorly-designed experiments and out-of-touch science(7).
Erroneous claim that BPA is “weak” estrogen
In the FDA investigation, the authors stated that “BPA-induced effects partially overlapped those induced by EE2 estradiol, consistent with the known weak estrogenic activity of BPA” (1). While this may have been the understanding very early on in BPA-related research, a significant amount of research has been done since then that refutes this “weak” claim.
In fact, many studies are now finding that BPA can elicit estrogen-like activity that has been shown to be as strong as, if not stronger, than EE2, the mechanisms by which are different than “traditional” EE2 pathways (8,9).
Specifically, more recent studies have found that BPA promotes very fast responses, even at low doses, via “non-classical estrogen pathways” and the activation of many different transduction signaling pathways in different cell types which when combined results in greater effects related to gene expression triggered by estrogenic action (8,9,10).
Failure to identify well-known exogenous source(s) of BPA contamination
Feed: One potential source of BPA contamination in the rats in this and many other EDC research studies is the specific diet or feed that is given. The FDA paper itself acknowledged this piece of information and stated that “since many of the effects of BPA are thought to occur through interference with estrogen signaling pathways, a soy- and alfalfa-free diet was used and dietary levels of the phytoestrogens genistein and daidzein were monitored, in addition to BPA.”
In fact, it is true that most of commercial rodent feed contains the phytoestrogens genistein and daidzein, which show a strong affinity for estrogen receptor beta and have been shown to be, at times, more potent than the effects of environmental estrogens such as DDT and BPA (11,12). Studies have confirmed that it is the feed itself that can activate estrogen receptors as a result of estrogenic compounds in the feed, and that it is not due to the metabolism of the food in the gut (21).
While they were careful to choose an appropriate diet for minimizing phytoestrogen content in the rodent diets during the experiment, they failed to take into consideration the diet the rats received prior to starting the experiment. In the FDA paper, it was stated that “while in the breeding colony, [the breeders] were maintained with their dams on NIH-41 irradiated feed pellets” from Harlan Laboratories.
Coming directly from the Harlan Laboratories website, it does not appear that the NIH-41 diet is phytoestrogen-free. Specifically “the diets 2014, 2016, 2019, and 2020X while not completely “free” are best described as minimal phytoestrogen diets.
Quarterly testing of these diets for isoflavones (phytoestrogens from soybean) typically reveals levels ranging from less than detectable to 20 ppm. For some perspective, most traditional rodent diets contain from 200 to 500 ppm isoflavones. Phytoestrogen minimal diets such as 2014, 2016, 2019, and 2020X are generally acceptable for studies where phytoestrogens are of concern.” (13)
In other words, NIH-41 from Harlan Laboratories does not appear to be one of their phytoestrogen minimal diets, indicating that prior to experimentation, the rats at the facility used by the FDA study were, in fact, exposed to a range of phytoestrogens on a daily basis, thus indicating a possible source for the pre-experimental contamination of the rats. After getting in touch directly with Harlan Laboratories about NIH-41, they confirmed that this particular feed contains soy-based products (5% soybean meal as well as 2% soy oil), thus confirming the potential for exogenous phytoestrogen exposure prior to testing(25).
In regards to the experimental food, while it was reported to be free of the phytoestrogens genistein and daidzein, it was also reported by the FDA researchers to contain “BPA levels below the average analytical blanks.” (1) Even though they controlled for certain phytoestrogens in the animal feed, BPA itself was found in the diet and thus a possible contributor to the BPA contamination of the rats during experimentation (3).
Finally, since recent research has indicated that the half-life of BPA may be longer than initially thought (6,14), the low levels of BPA in the feed could be more of a problem than the FDA researchers assumed.
Failure to identify well-known exogenous source(s) of BPA and estrogenic acitivty of equipment and supplies
Use of polysulfone cages: In addition to the feed, there are many other exogenous sources of estrogens that the rats were exposed to both prior to and during experimentation by the FDA researchers.
Specifically, laboratory animals are in other words inundated with possible sources of BPA and other EDC exposure, including from bedding, cages, water bottles, water, and disinfectant cleaners (6,11). According to the FDA researchers, their rats were housed in polysulfone cages. While they indicated that equipment used was “low BPA” and that levels were monitored regularly, it wasn’t clear what those values actually were.
Polysulfone pellets are commonly used in many types of equipment for both research and medicine. In fact, one study focusing on kidney hemodialyzers found that polysulfone pellets, which are major components of the hemodialyzer machines, leach a significant amount of BPA into the environment (15).
When combined to create the final machine, the polysulfone does not appear to leach as much BPA as it did when in the pellet form; however, it is possible that after long term exposure, patients requiring prolonged use of this polysulfone-based equipment could be adversely affected (15).
Finally, there was a failure on the part of the FDA researchers to test the estrogenic activity of all these aforementioned possible exogenous sources of estrogens. It is clear they were not in touch with the current state of EDC research, as they would have been more careful testing the estrogen content of everything the rats came in contact with prior to and during experimentation, since many studies have found vast variability in the sources of possible EDC contamination that must be minimalized as much as possible in BPA and other EDC experiments.
Failure to measure BPA levels in any of the animals except in the test cohort at the very end of the trial
By not measuring BPA levels of any of the animals from the very beginning, and only measuring BPA levels at the very end of the trial, effectively no consideration was given to actual levels of BPA in the animals. Without having a starting point to compare to, it's near impossible to draw any conclusions based on what was seen at the end. If the researchers had tested the animals at the very beginning, they would have found that all animals were contaminated with BPA, even the controls, and at that point some amendments to the protocol could be made.
In addition, failure to measure BPA levels in test animals makes it impossible to determine serum concentrations and their possible correlation to effects or whether the method used for administering the BPA (oral gavage) may have been flawed.
Use of oral gavage which has been found inadequate
The FDA study used oral gavage dosing of pregnant rats during the experiment (1). Recent studies in BPA research have found that the oral gavage method may be an inappropriate technique for BPA and other EDC experiments, as this method does not accurately represent the actual ingestion process by rodents and humans (16).
Specifically, “most previous studies have relied on the gavage method, where BPA is distributed evenly in the gut. This is not how food actually enters our bodies. We chew it, move it around in our mouths, and it interacts with numerous surfaces--our tongue, cheeks, etc--before it enters the stomach.” (16). In other words, using the gavage method would suggest that there shouldn’t be any unmetabolized BPA in the blood after exposure, however many studies find that, in fact, there is (16).
In a study using a dog model, researchers compared gavage methods of BPA delivery versus sublingual (under the tongue) exposure methods, and found that BPA absorption into the body was significantly higher using sublingual methods than gavage techniques (17) Specifically, the levels of BPA in the dogs’ bodies were significantly higher in those dogs’ receiving the sublingual BPA treatment than the levels of BPA in those dogs’ receiving BPA through gavage. The study concluded that BPA is more efficiently and rapidly absorbed through the mucosa of the mouth, resulting in much higher BPA levels in the body than using gavage methods (17).
This means that a study using gavage results in BPA levels in the bodies of their test subjects that are significantly lower than they would be in a more “real world” scenario such as sublingual absorption and therefore the conclusions from which would be inaccurate. Similar studies in the mice model have found similar results, in that exposure of BPA through consumption of a diet treated with BPA results in a more natural route of exposure than through oral gavage (18).
Wrongly examining toxicity when timing is a bigger issue
One problem that had plagued early BPA and EDC research was examining these compounds in terms of their toxicity, while advancements in this type of research has demonstrated that the timing of the exposure is markedly more critical than simple toxicity alone.
In other words, a certain level of BPA exposure to an adult may not have much of an effect, while if that same level of BPA is exposed to a developing embryo at a very specific point in its development, a cascading series of events can be "programmed" into the embryo that could develop into significant health problems later in life (4,22,23,24).
There is also evidence in the literature that the adult endocrine system is able to maintain a sort of homeostasis, or regulation of a constant balanced environment, thereby may not be as sensitive to the effects of EDC exposure as say if it were going through puberty, a state of marked hormonal flux, or embryonic development, both which have been shown to be significantly more sensitive to BPA and other EDC exposure at those specific times (4,24).
Choosing an out-dated model such as "toxicity" is inappropriate when timing is what is most crucial in determining adverse effects of BPA and other EDC exposure.
Failure to disclose conflicts of interest
Conflicts of interest, while seemingly short and often times an after-thought for some researchers, is a critical part of the publication process. Even if there are no conflicts of interest, it is in the researcher's best interest to provide this information. The existence of conflicts of interest may negatively influence the results of the study in question, essentially invalidating the conclusions presented (19).
By disclosing any conflicts, or indicating "none" if that happens to be the case, important information is presented to the reader so that they may be able to cast their own judgment on the legitimacy and objectiveness of the researcher's results and conclusions (20) . A failure to disclose conflicts of interest casts a skeptical eye on the validity of the entire study.
Conclusion: This Paper Fails On So Many Levels It Should Be Withdrawn
This paper fails in so many fundamental areas of basic science, methodology, appropriate outcomes, and laboratory practice that it struggles to ascend to the level of an "F" in a sophomore biology class. It is, however, a detailed object lesson in how not to conduct a study.
This paper's greatest success lies in demonstrating the disgraceful state of research underpinning regulatory decisions regarding Endocrine Disrupting Compound.
The FDA especially needs to raise its standards to match those of the National Institutes of Health which has three stringent levels of scrutiny involving peer review: Before grants are awarded, the competency of the investigators is assessed as are, experiment design, appropriate outcomes and use of the most current science and lab techniques(7). This is followed by post-investigation peer-review, publication in peer-reviewed journals and efforts to follow up the study with replication and updated results.
The FDA and the old-line toxicology establishment must bring its last-century knowledge base and paradigms up to date and begin to conduct itself in the public interest. Otherwise the FDA regulation of EDCs will continue to follow its long-established line of failures that include DES, and tobacco.
This paper should be rejected if it is not withdrawn.
(1) Delclos, K.B., Camacho, L., Lewis, S.M., Vanlandingham, M.M., Latendresse, J.R., Olson, G.R., Davis, K.J., Patton, R.E., da Costa, G.G., Woodling, K.A., Bryant, M.S., Chidambaram, M., Trbojevich, R., Juliar, B.E., Felton, R.P., and Thorn, B.T. 2014. Toxicity evaluation of Bisphenol A administered by gavage to Sprague-Dawley rats from gestation day 6 through postnatal day 90. Toxicological Sciences doi:
(2) Johnson, P.D., and Besselsen, D.G. 2002. Practical aspects of experimental design in animal research. ILAR Journal (Institute for Laboratory Animal Research) 43(4): 202-206.
(3) Vom Saal, F.S., Richter, C.A., Ruhlen, R.R., Nagel, S.C., Timms, B.G., and Welshons, W.V. 2005. The importance of appropriate controls, animal feed, and animal models in interpreting results from low-dose studies of Bisphenol A. Birth Defects Research 73: 140-145.
(4) Vandenberg, L.N., Colborn, T., Hayes, T.B., Heindel, J.J., Jacobs, Jr., D.R., Lee, D-H., Shioda, T., Soto, A.M., vom Saal, F.S., Welshons, W.V., Zoeller, R.T., and Myers, J.P. 2012. Hormones and Endocrine-Disrupting Chemicals: Low-Dose Effects and Nonmonotonic Dose Responses. Endocrine Reviews 33: 0000-0000.
(5) vom Saal, F.S., and Hughes, C. 2005. An extensive new literature concerning low-dose effects of Bisphenol A shows the need for a new risk assessment. Environmental Health Perspectives 113(8): 926-933.
(6) Hunt, P.A., Susiarjo, M., Rubio, C., and Hassold, T.J. 2009. The BPA Experience: A primer for the analysis of environmental effects on mammalian reproduction. Biology of Reproduction 81: 807-813.
(7) Myers, J.P., vom Saal, F.S., Akingbemi, B.T., Arizono, K., Belcher, S., Colborn, T., Chahoud, I., Crain, D.A., Farabollini, F., Guillette, Jr., L.J., Hassold, T., Ho, S., Hunt, P.A., Iguchi, T., Jobling, S., Kanno, J., Laufer, H., Marcus, M., McLachlan, J.A., Nadal, A., Oehlmann, J., Olea, N., Palanza, P., Parmigiani, S., Rubin, B.S., Schoenfelder, G., Sonnenschein, C., Soto, A.M., Talsness, C.E., Taylor, J.A., Vandenberg, L.N., Vandenbergh, J.G., Vogel, S., Watson, C.S., Welschons, W.V., and Zoeller, R.T. 2009. Why public health agencies cannot depend on Good Laboratory Practices as a criterion for selecting data: The case of Bisphenol A. Environmental Health Perspectives 117(3): 309-315.
(8) Alonso-Magdalena, P., Ropero, A.B., Soriano, S., García-Arévalo, M., Ripoll, C., Fuentes, E., Quesada, I., and Nadal, A. 2012. Bisphenol-A acts as a potent estrogen via non-classical estrogen triggered pathways. Molecular and Cellular Endocrinology 355: 201-207.
(9) Hugo, E., Brandebourg, T.D., Woo, J.G., Loftus, J., Alexander, J.W., and Ben-Jonathan, N. 2008. Bisphenol A at environmentally relevant doses inhibits adiponectin release from human adipose tissue explants and adipocytes. Environmental Health Perspectives 116 (12): 1642-1647.
(10) Watson, C.S., Bulayeva, N.N., Wozniak, A.L., and Finnerty, C.C. 2005. Signaling from the membrane via membrane estrogen receptor-α: Estrogens, xenoestrogens, and phytoestrogens. Steroids 70: 364-371.
(11) Thigpen, J.E., Setchell, K.D.R., Saunders, H.E., Haseman, J.K., Grant, M.G., and Forsythe, D.B. 2004. Selecting the appropriate rodent diet for endocrine disruptor research and testing studies. ILAR Journal 45(4): 401-416.
(12) Shelby, M.D., Newbold, R.R., Tully, D.B., Chae, K., and Davis, V.L. 1996. Assessing environmental chemicals for estrogenicity using a combination of in vitro and in vivo assays. Environmental Health Perspectives 104(12): 1296-1300.
(13) Harlan Laboratories Diets FAQ page: http://www.harlan.com/products_and_services/research_models_and_services/laboratory_animal_diets/diets_faqs.hl#11 (accessed 02/22/2014).
(14) Stahlhut, R.W., Welshons, W.V., and Swan, S.H. 2009. Bisphenol A data in NHANES suggest longer than expected half-life, substantial nonfood exposure, or both. Environmental Health Perspectives 117(5): 784-789.
(15) Haishima, Y., Hayashi, Y., Yagami, T., and Nakamura, A. 2001. Elution of Bisphenol-A from hemodialyzers consisting of polycarbonate and polysulfone resins. Journal of Biomedical Materials Research 58(2): 209-215.
(16) Barrett, J.R. 2013. Oral Argument: Sublingual findings challenge key assumptions about BPA exposure. Environmental Health Perspectives 121(8): A257.
(17) Gayrard, V., Lacroix, M.Z., Collet, S.H., Viguié, C., Bousquet-Melou, A., Toutain, P-L., and Picard-Hagen, N. 2013. High bioavailability of Bisphenol A from sublingual exposure. Environmental Health Perspectives 121(8): 951-956.
(18) Sieli, P.T., Jašarević, E., Warzak, D.A., Mao, J., Ellersieck, M.R., Liao, C., Kannan, K., Collet, S.H., Toutain, P-L., vom Saal, F.S., and Rosenfeld, C.S. 2011. Comparison of serum Bisphenol A concentrations in mice exposed to Bisphenol A through the diet versus oral bolus exposure. Environmental Health Perspectives 119(9): 1260-1265.
(19) Van McCrary, S., Anderson, C.B., Jakovljevic, J., Khan, T., McCullough, L.B., Wray, N.P., and Brody, B.A. 2000. A national survey of policies on disclosure of conflicts of interest in biomedical research. The New England Journal of Medicine 343(22): 1621-1626.
(20) Sass, J. 2009. Effective and practical disclosure policies: NRDC paper on workshop to identify key elements of disclosure policies for health science journals. National Resources Defense Council Report. https://www.nrdc.org/health/disclosure/files/disclosure.pdf?origin=publication_detail Accessed 02/23/2014.
(21) Ciana, P., Brena, A., Sparaciari, P., Bonetti, E., Di Lorenzo, D., and Maggi, A. 2005. Estrogenic activities in rodent estrogen-free diets. Endocrinology 146(12): 5144-5150.
(22) Markey, C.M., Coombs, M.A., Sonnenschein, C., and Soto, A.M. 2003. Mammalian development in a changing environment: exposure to endocrine disruptors reveals the developmental plasticity of steroid-hormone target organs. Evolution & Development 5(1): 67-75.
(23) Markey, C.M., Wadia, P.R., Rubin, B.S., Sonnenschein, C., and Soto, A.M. 2005. Long-term effects of fetal exposure to low doses of the xenoestrogen Bisphenol-A in the female mouse genital tract. Biology of Reproduction 72: 1344-1351.
(24) Marty, M.S., Carney, E.W., and Rowlands, J.C. 2011. Endocrine disruption: Historical perspectives and its impact on the future of toxicological testing. Toxicological Sciences 120(S1): S93-S108.
(25) NIH Specification: NIH 41 Open Formula Rodent Irradiated Diet specs. 2000. http://www.ors.od.nih.gov/sr/dvr/drs/nutrition/Documents/SpecsDiets/41.pdf Accessed 02/24/2014.