Letter to the Editor and Response Jan 2013

Letter: A New Paradigm for Risk Assessment of Endocrine Disruptors in Food Contact Materials

Dear Editor,

In their recent CRFSFS article “Human Risk Assessment of Endocrine-Disrupting Chemicals Derived from Plastic Food Containers” Bang and others (Vol. 11, iss. 5, pp. 453–470; September 2012) reached the conclusion that “synthetic resins used for the manufacture of plastic food containers are mostly non-endocrine-disrupting chemicals (EDCs) ingredients, and should be considered safe.” This view is based on a comparative risk assessment exercise relying upon dietary exposure and acceptable daily intake/reference dose estimates for the compounds bisphenol A (BPA), styrene, and several phthalates. Further, Bang and others stated that “PE, PET, and PP… are non-EDC-related ingredients” despite of scientific evidence in the peer-reviewed literature showing the contrary (Wagner and Oehlmann 2009, Yang and others 2011, Andra and others 2012).

We agree with Bang and others that their conclusion on the safety of BPA, styrene, and phthalates can be reached when applying the traditional risk assessment scheme. However, this approach may be misleading, because it fails to consider recent advances in EDC scientific research, such as low dose effects, non-monotonic dose response curves, absence of threshold values, and chemical mixture effects. Therefore, in our view, the conclusions reached by Bang and others are too broad, and may not be extrapolated to a multitude of compounds present in packaging materials that are known or suspected EDCs.

In our opinion, the greater issue at hand is whether food contact materials (FCM) are a relevant source of EDCs, and if their migration into food poses a risk to human health. Answering this question requires consideration of several aspects.

First of all, it is important to understand why health-related concerns for EDCs have arisen. The Endocrine Society, a major scientific society for clinical practitioners and basic endocrinologists, has expressed serious concerns for EDCs due to their interfering action with the endocrine system on several levels, ranging from hormone production, to receptor occupancy, to protein functioning, and eventually to detrimental response(s) (Diamanti-Kandarakis and others 2009, Zoeller and others 2012). This concern is based on a large body of high-quality scientific research linking EDC exposure to adverse effects in animal models. Furthermore, epidemiological studies have shown an association between various chronic diseases and exposure to EDCs (Birnbaum 2012).

Second, there is sufficient scientific evidence to warrant questioning the currently practiced risk assessment for EDCs (Birnbaum 2012). With the notable exception of mutagens, chemical risk is linked to exposure levels. Safe exposure levels are derived by testing high concentrations of single chemicals in animal models and then applying safety factors for extrapolation to doses thought to be safe for humans. For EDCs, this approach seems to lack scientific basis, because hormone-mimicking chemicals can be active at low doses, and because their dose-response curves can be non-monotonic (Vandenberg and others 2012). As a result, extrapolation from high dose testing may miss health-relevant effects occurring at low doses for both natural hormones and EDCs (Myers and others 2009). This questions the relevance and reliability of current RfD and ADI for EDCs.

Third, it is well-documented that mixtures of EDCs can cause “cocktail effects”, even if the individual components of the mixture are present at levels below their no observed adverse effect level (Kortenkamp 2007, Kortenkamp and others 2007). Plastic FCM are a complex mixture of individual substances varying in reactivity and chemical structures (Bradley and Coulier 2007).

Fourth, knowledge on the presence of EDCs in FCM is very limited, with a few notable exceptions. Extracts of various plastic types of FCM have been shown to exhibit EDC properties (Yang and others 2011). Snails cultured in plastic (polyethylene terephthalate, PET) bottles showed a significantly increased reproduction compared to controls housed in glass vessels (Wagner and Oehlmann 2009). Since this endpoint is sensitive to EDC exposure, these findings demonstrate that compounds migrating from the FCM can interfere with the reproduction of an animal model.

Finally, knowledge gaps about the presence and leaching of EDCs from FCM exist and hinder evaluating FCM safety with sufficient certainty. It is well established that known or suspected EDCs are widely used for the production of FCM, exhibiting a range of migration rates under a plethora of environmental conditions, such as high temperature, frequent re-use, and prolonged UV exposure (Muncke 2009, Andra and others 2011). Other EDCs – known or so far unidentified – may potentially be present in FCM, either as monomer, additive, adhesives, or as non-intentionally added substances (NIAS), like byproducts.

Such NIAS of styrene-based polymers have displayed EDC properties in vitro and in vivo (Ohyama and others 2007, Yanagiba and others 2008), though findings for styrene oligomers varied as pointed out by Bang and others (2012). Furthermore, certain brominated organic compounds with known endocrine disrupting activity in vivo have been shown to migrate from FCM, raising further questions as to their source(s), for example, recycled plastics use (Andra and others 2012). We therefore question whether “PE, PET, and PP… are non-EDC-related ingredients and are generally considered safe.” As long as there is no general requirement for testing EDC properties of FCM and FCM substances, we consider this statement as premature, and the absence of such data should promote research efforts towards an improved FCM safety assessment.

Reassessing health risk quantification protocols for EDCs requires dedicated collaboration between relevant disciplines (exposure science, toxicology, epidemiology, risk assessment). Much awaited updates in existing regulations for FCMs (and other products) based on current scientific knowledge is necessary, and will strengthen the preventive nature of public health policies. In conclusion, improving our scientific understanding of the magnitude and uncertainty associated with EDC exposures from FCM should be a priority because of our society's dependence on FCM.


Jane Muncke

Food Packaging Forum

Zurich, Switzerland

Martin Wagner

Goethe Univ. Frankfurt a.M., Germany

Konstantinos C. Makris

Cyprus International Inst. for Environmental and

Public Health in association with Harvard School of

Public Health

Cyprus University of Technology, Limassol, Cyprus.

Disclosure: Jane Muncke is an employee of the charitable non-profit Food Packaging Forum foundation. The view expressed in this letter is J. Muncke's personal professional opinion and not the opinion or position of the Food Packaging Forum.

Response Letter: Do We Need a New Risk Assessment Guideline for Endocrine Disrupting Chemicals (EDCs)?

Dear Editor,

In their letter, Muncke, Wagner, and Makris agreed with our conclusions that when applying traditional risk assessment methodology based upon sound scientific data for bisphenol A (BPA), styrene, and phthalates, these chemicals in the concentrations in food contact materials (FCM) are safe. They argue that this approach is misleading based upon recent advances in endocrine disrupting chemicals (EDCs) scientific research on low dose effects, non-monotonic dose response curves, absence of threshold values and chemical mixtures effects. However, if one considers that the recent research referred to (1) is not based upon actual concentrations of the FCM found in humans, (2) has not shown that these so-called EDC effects actually occur in humans, (3) is extrapolated in large part from in vitro findings where excessive concentrations are used, (4) where FCM at levels encountered in the environment have not fully been examined in vivo, and (5) where mixture effects remain to be proven, the use of traditional risk assessment methodology still remains scientifically sound especially for compounds present in packaging materials as this “recent advances” research has not supplanted the existing data. In fact, Muncke, Wagner, and Makris clearly stated that they were in agreement with our paper.

Are there limitations of current classical risk assessment for EDCs?

It may be generally recognized that the classical risk assessment guidelines are not always ideal to evaluate the actual risks for all kinds of toxicants including chemical, biological, and physical agents, or exposure to mixtures or to interactions between 2 or more agents. Currently, one of the most controversial issues regarding the adverse effects of EDCs is the presence of low dose effects, hormesis, threshold responses, as well as non-monotonous dose-response relationships (Weltje and others 2005, van der Woudeand others 2005, Phillips and others 2008). It has been suggested that there is an association between exposure to low dose of EDCs and adverse health effects obtained from both animal and epidemiological data (Kortenkamp 2008, Vandenberg and others 2012). On the other hand, it is still controversial and has not been shown conclusively that low dose effects of EDCs are likely in humans (Witorsch 2002, Rhomberg and Goodman 2012). However, it needs to be emphasized that the effects reported were primarily derived from chemicals that are not present in FCM. In addition, a large portion of the animal data was derived from in vitro investigations with extrapolation to humans. One should also be apprised of the fact that the epidemiologic data has severe limitations as exposure has not been clearly established to correlate an effect with a response. Because of that, Birnbaum (2012) recently proposed that collaboratory efforts between scientists in academia, government, and industry should develop more sophisticated study designs to facilitate regulatory decisions. Further, Birnbaum did not indicate that the traditional risk assessment should be discarded but needs to complement the current directions. It would appear that Muncke, Wagner, and Makris are in agreement with our conclusions that the risk assessment approach taken in our paper is justifiable in light of the current knowledge available. It should be reiterated that the letter authors did not provide any concrete basis or data for the speculativelow dose effects for FCM, while our paper did utilize current available data.

First, it should be noted that at present it is not possible to draw the causal relationship between low dose exposure to EDCs and adverse health outcomes in humans for a variety of reasons including numerous risk or confounding factors that are involved in the etiology of endocrine disruption or endocrine-related diseases (Golden and others 2012). One needs to be cognizant of the fact that exposure is extremely difficult to monitor and correlate to a specific EDC outcome and the question of mixtures further compounds the situation. It should also be made clear that the concentration to which humans are environmentally exposed is minute and in experimental in vivo settings, not sufficient to produce an effect. In addition, most animal or epidemiological studies have not fully or comparatively assessed causation of diseases in terms of ranking relative weight of evidence of various risk factors at environmental exposure levels. Therefore, we did not dismiss the presence of low dose effects under conditions that are dramatically different than exposure to FCM in humans and this is in agreement with Muncke's letter. However, based upon the “best” science available, it is clear that the risk approach taken here did demonstrate that EDC use was safe and the low dose effect was not applicable as data were not available. Muncke, Wagner, and Makris’ letter speculated a role for a mixture effect with respect to FCM but failed to provide any concrete foundation or actual data for this statement.

Second, in the classical risk assessment guidelines, one has to rely on numerous default values, which include uncertainty factors and modifying factors, reflecting interspecies and intraspecies differences. As long as there is a dependence on animal toxicity data, it may not be possible to overcome the limitations of uncertainties. In an Endocrine Society Scientific Statement, a review by Diamanti-Kandarakis and others (2009) on EDCs concluded that direct causal links between exposures to EDCs and disease states in humans are difficult to draw, and therefore more research on screening for exposures and targeting at-risk groups is a high priority. It is worthwhile noting that this review described EDCs in general and did not consider FCM at environmentally-relevant concentrations.

Third, toxicological data are obtained predominantly from single agent and not from mixtures (2 or more), which may or may not produce additive or synergistic effects and may also counteract each other be antagonistic with no effect seen. It needs to be emphasized that humans and animals are environmentally exposed to mixtures, but inferences of EDC effects are usually to one compound such as the low dose effect propounded by the letter writers as being crucial, but this is not realistic as in the case of FCM examined in our paper where such information was not available. Therefore, risk assessment based on no observed adverse effect levels (NOAEL) derived from a single chemical agent might underestimate the synergistic (or cocktail) or antagonistic effects of mixtures of EDCs in real exposure situations (Kortenkamp 2007, Kortenkamp and others 2007). As proposed by Muncke, Wagner, and Makris, FCM are a mixture of individual substances where effects can be synergistic but they can also be antagonistic. Based upon the “best” current data available risk assessment was carried out in individual chemicals but the presence of mixtures was not discounted. Current regulatory limits of individual chemical for specific migration limits of plastic monomers and other alternatives are available and were thus used in our evaluation. The use of low dose effects was not discounted by us, but simply these data were not available. Our findings for FCM should not be negated because traditional risk assessment was conducted as Muncke's letter did not provide substantiative evidence based upon scientific observations to counter the conclusions.

Is an in vitro assay sufficient to predict EDCs?

There are various in vitro and in vivo test methods to screen the potential of EDCs at different levels (1–5) or Tier 1 or 2, which implies that 1 single test method is not sufficient to predict endocrine disrupting properties of potential EDCs. Yang and others (2011) used a roboticized MCF-7 cell proliferation assay, an in vitro screening assay to identify only estrogenic activity (EA) of chemicals extracted from plastic FCM. These screening results may need further validation using in vivo assays such as uterotrophic assay for estrogenicity and Hershberger assay for androgenic activities because an in vitro test system may produce false positive or false negative results (Baker 2001). It is also difficult to extrapolate in vitro findings to environmental human exposure as the EDCs concentration needed to produce an effect is exceedingly high and not likely to be encountered. It is worthwhile noting that a recent study on plastic FCM demonstrated that there were no estrogenic or androgenic alterations induced by the chemicals both in vitro (for example, E-screen assay, transactivation) or in vivo (for example, uterotrophic, Hershberger assay) assay systems (Osimitz and others 2012). In addition, it should also be noted that detection of EA by an in vitro test need not necessarily be considered a determining factor of risk or safety of plastic FCM in humans (Mantovani and others 1999).

Are the leaching or migration levels of chemicals from FCM unsafe?

There have been controversial issues for analysis of not only EDCs, but other chemicals measured by different analytical methods. Bach and others (2012) reported that inappropriate extraction methods and sample treatment may result in false-negative or false-positive responses when testing water extracts of polyethylene terephthalate (PET) bottles in bioassays. Therefore, they suggested a combinational analysis of chemicals with bioassays to be carried out for hazard assessments. As noted by Muncke (2009), sources of contaminants, impurities, NIAS, or recycled plastic ingredients can provide unexpected measurements of chemicals in plastic FCM; however, it needs to be emphasized that little is known about these compounds, whether they are directly related to FCM or not, and the concentrations present (Leivadara and others 2008, Skjevrak and others 2005, Grob and others 2006). Leaching or migration of chemicals from FCM is dependent on environmental conditions, but specific migration limits (SML) from FCM have been regulated in terms of human exposure limits set by regulatory agents (EFSA 2006, Westerhoff and others 2008; Lim and others 2009; IRIS 2012). At even extreme use conditions (for example, brushing, microwaving, and so on) of FCM, the migration levels of EDCs from FCM are within SML for each EDC or provisional tolerable daily intake, and the use of plastic food containers is considered as safe under normal use guidelines (Lim and others 2009, Welle and Franz 2011). Although the issue of mixtures still remains to be resolved, one needs to be cognizant that the environmental concentrations of each FCM within the mixture to which a human is exposed is extremely small. It should also be noted that FCM within mixtures need not necessarily be additive or synergistic but can indeed act in antagonistic fashion.


In the classical risk assessment process, there are a variety of uncertainties including low dose effects, hormesis, chemical mixture effects, threshold, and interactions. In addition, the acceptable daily intake, tolerable daily intake, or reference dose is derived from NOAEL based upon animal data using uncertainty factors. Therefore, the estimates obtained from the classical risk assessment guidelines need to be interpreted with consideration of uncertainty. When other risk assessment guidelines are available, the limitations of classical risk assessment may be overcome or utlized in concert with the new approaches. More data on human studies as well as well-designed toxicological data for risk assessment on EDCs derived from FCM are needed to better understand the uncertainties for human risks. Currently the approach taken in this review may not be perfect but the “best” science data available was utilized and should not be discounted due to some limitations. One also needs to be aware that the crtitisms raised by Muncke, Wagner, and Makris were not based upon concrete scientific findings for FCM but from inferences that may in reality not be appropriate.


Byung-Mu Lee

Division of Toxicology,

College of Pharmacy,

Sungkyunkwan University,

440–746, Gyeonggi-do, Korea

Sam Kacew

Inst. of Population Health Risk Assessment

Univ. of Ottawa,

Ontario, Canada

Hyung Sik Kim

College of Pharmacy

Pusan National Univ.,

San 30, Jangjeon-dong, Geumjeung-gu, Busan,

609–735, South Korea