As stated in our review (Vandenberg et al. 2010
), great concern exists about exposure of human fetuses, infants, and neonates to BPA because of the sensitivity of developing organs and the brain to exogenous hormones (Vandenberg et al. 2009
). However, to translate findings from animal studies to health risks in humans, exposure assessments and biomonitoring of BPA in different populations are essential. Thus, in November 2006, the European Food Safety Authority (EFSA) released its opinion on the plausibility of data regarding levels of BPA in human blood and excretion of BPA and BPA metabolites in environmentally exposed humans. The EFSA panel (2006)
[T]here is very low oral bioavailability of the parent substance, BPA, in humans and other primates. Due to this rapid biotransformation and excretion and plasma protein binding in humans, peak BPA-concentrations after dietary exposures to BPA available for receptor binding are predicted to be very low even in worst case exposure scenarios.
The EFSA panel was asked to reconsider their assessment based on recent studies that suggested the possibility for age-dependent toxicokinetics of BPA. In July 2008, the EFSA released its second opinion in support of their original statement (EFSA 2008
The Panel therefore considers that its previous risk assessment … can be considered as conservative for humans. The Panel concluded that the differences in age-dependent toxicokinetics of BPA in animals and humans would have no implication for the EFSA 2006
risk assessment of BPA.
In stark contrast to these statements, we analyzed > 80 biomonitoring studies and came to the conclusion that measurable levels of BPA and BPA conjugates are present in human blood and urine, as well as in other tissues and fluids (Vandenberg et al. 2010
). These biomonitoring studies examined thousands of individuals from many developed and some developing countries and collectively indicate that humans are internally exposed to unconjugated BPA (Vandenberg et al. 2007
; Welshons et al. 2006
). Biomonitoring studies are crucial for understanding current human exposure levels because, by their very nature, they account for all exposures. This is essential, because all exposure sources for BPA have not yet been identified, and existing data suggest that non-oral exposures may be significant (Gies et al. 2009
; Stahlhut et al. 2009
A comprehensive review of the large number of biomonitoring studies indicates that they are highly consistent and therefore reliable (Vandenberg et al. 2010
). The detection rates and concentrations of BPA in urine and blood of environmentally exposed individuals are remarkably similar in studies performed in many laboratories using a variety of techniques, including highly accurate and sensitive methods [e.g., solid-phase extraction coupled with isotope dilution-HPLC-tandem mass spectrometry, as used by the U.S. Centers for Disease Control and Prevention (Calafat et al. 2005
)]. Further, there is no evidence to suggest that these studies should be invalidated because of poor quality control (e.g., contamination from collection materials, breakdown of conjugates during storage, inadequate blanks) (Gies et al. 2009
; Vandenberg et al. 2010
). In total, the reproducibility of these results indicates that humans are internally exposed to doses of unconjugated BPA, with a central measure of the distribution in the 0.5–3 ng/mL range. In spite of these consistent findings, the EFSA panel came to a completely different conclusion about current human exposures (EFSA 2006
). What is the basis for this disparity?
The pivotal factor we identified in the EFSA report was the bias in the selection of studies used in this evaluation. The EFSA panel (EFSA 2006
) ignored the majority of the biomonitoring studies. Although they reviewed 2 toxicokinetic studies (Volkel et al. 2002
) extensively, only 2 of the 17 urine biomonitoring studies published by 2006 were discussed in any detail. Only a small number of the blood biomonitoring studies were cited in the EFSA report, and none of these studies were discussed in detail at any level. Instead, the EFSA panel identified potential problems with these biomonitoring studies, including the use of ELISA (used only in a few studies), possible contamination of reagents with BPA, and the leaching of BPA from materials used for sample collection, storage, and processing. Without providing any evidence that these are indeed issues in the biomonitoring studies examined, the EFSA (2006)
Due to all these confounders, the reported analytical results on BPA blood concentrations most probably considerably overestimate real blood concentrations actually present.
Of particular concern relative to this stance is that although the biomonitoring studies have produced reliable, consistent results, the two toxicokinetic studies (Volkel et al. 2002
) the EFSA (2006)
relied upon heavily for their risk assessment have significant inconsistencies and are yet to be replicated. Most concerning is the fact that the methods used in these two toxicokinetic studies were much less sensitive than those used in almost all biomonitoring studies. The toxicokinetic studies had limits of detection (LODs) as high as 2.28 ng/mL for unconjugated BPA and 10.1 ng/mL for conjugated BPA, compared with LODs of 0.0063–0.4 ng/mL in other studies using similar analytical methods (Vandenberg et al. 2010
). These two toxicokinetic studies examined only a small number of adult subjects administered BPA (15 adults total) compared with the thousands of individuals (including infants, children, adolescents, and pregnant women) sampled for biomonitoring purposes. Only one of these studies (Volkel et al. 2002
) examined concentrations of BPA in both blood and urine samples, whereas the other study (Volkel et al. 2005
) reported conjugated BPA concentrations in urine but provided no information about BPA concentrations in the plasma samples collected by the authors. Yet both studies were used by the EFSA to discount the presence of BPA in plasma and blood samples reported in numerous other studies (EFSA 2006
). Additional problems with data analysis and interpretation in the Volkel et al. studies (2002
) are discussed in greater detail in our review (Vandenberg et al. 2010
The EFSA panel (2006)
speculated that the repeated detection of unconjugated BPA in human blood was due to poor sample processing conditions and/or unreliable methods, stating,
The studies reporting detection of BPA in human blood in concentrations higher than 1 [μg] BPA/L have usually determined [unconjugated] BPA, without prior enzymatic cleavage of BPA-glucuronide.… The fate of BPA-glucuronide under the conditions of the diverse sample processing conditions and a possible cross-reactivity of the [ELISA] antibodies with BPA-glucuronide is not reported, leaving the possibility that reported BPA levels actually reflect BPA-glucuronide levels.
Consistent results from a large number of biomonitoring studies cannot be disregarded based only on the speculation that they overestimated unconjugated BPA levels because of hypothetical poor analytical controls. The deficiencies speculated by the EFSA were addressed and invalidated by one or more appropriate controls within each of the individual biomonitoring studies in question; most studies contained numerous controls to counter speculations of contamination or cross-reactivity of ELISA antibodies. For example, blanks reported in these studies would show measurable BPA if cross-contamination occurred at any step in the sample-handling process or analysis—yet they did not, leaving the speculations made by the EFSA without any scientific basis.
The EFSA panel (EFSA 2006
) continued to rationalize their dismissal or lack of attention to biomonitoring studies by referencing the results of toxicokinetic studies:
[O]rally administered BPA is rapidly absorbed from the gastrointestinal tract and undergoes intensive first-pass metabolism to BPA-glucuronide in the gut wall and in the liver.… Concentrations of [unconjugated] BPA were below the limit of detection both in urine … and blood samples….
Further reasoning provided to reject the findings from biomonitoring studies was that the levels measured in environmentally exposed humans are “higher than the peak BPA concentrations determined in blood of monkeys after oral administration of a dose of 100 μg BPA/kg bw [body weight].” The panel concluded that
[T]hese reported concentrations of BPA in blood of unintentionally exposed human subjects of up to 10 [μg] BPA/L are orders of magnitude above the maximal concentrations of BPA predicted in blood by PBPK [physiologically-based pharmacokinetic] models on the basis of human BPA toxicokinetics after oral administration.
In science, if data contradict the hypothesis (i.e., the model), the hypothesis, not the data, must be rejected. It is unexpected, and perhaps unprecedented, for a scientific body to reject studies because their findings did not match a model, rather than to reconsider the model or reassess the findings from the extremely limited toxicokinetic studies that were used to generate the model. This reasoning is simply not founded in logic and is not how science-based regulatory decisions should be made. Considering the size of the biomonitoring literature, the consistency of the results from biomonitoring studies, and the significant problems in the toxicokinetic studies, conclusions drawn primarily from the two toxicokinetic studies (Volkel et al. 2002
) cannot be valid. Therefore the EFSA conclusion that there is negligible internal exposure to unconjugated BPA has no scientific basis.