The BfR considers the approach of EFSA for derivation of a new TDI and the result as inappropriate with respect to several points:
1) Selection of the critical endpoint (adversity and human relevance)
EFSA lowered the daily tolerable intake (TDI) of bisphenol A to 0.2 nanograms per kilogram of body weight per day. This was based on data from studies on mice on the effect of bisphenol A on the immune system: In offspring of dams that were administered bisphenol A during pregnancy and during nursing, an increase in the percentage of certain immune cells (Th17 cells) in the spleen was measured. It should be noted that these (healthy) animals did not show any signs of an adverse effect. Other studies showed that bisphenol A can increase already existing inflammation in allergy models with sick animals. So far, mice are the only species in which the effects identified by EFSA have been studied.
Th17 cells are special T-helper cells that are mainly found in mucosal barriers (e.g. in the intestine) and play an important role there in the immune defence against fungal infections. In addition to this protective function, Th17 cells are also associated with inflammatory reactions such as psoriasis. Furthermore, there are indications that the relative number of Th17 cells is increased in autoimmune diseases such as rheumatoid arthritis. However, it is still unclear what role an increased Th17 cell count plays within the respective clinical pictures, i.e. whether the increased relative Th17 cell count causes the disease or is merely associated with it (association or causality). In clinical trials, drugs that counteract Th17 cells have not been successful in treating various of these diseases.
In healthy mice, the relative increase in Th17 cells had no adverse effect, for example no inflammatory reactions occurred. In an extensive study on rats, carried out within the framework of the US National Toxicology Program (NTP), the animals were administered bisphenol A daily over their entire lifespan - beginning in the womb and ending after two years. A variety of analyses were performed, including testing for over 530 possible effects on the immune system. No adverse (harmful) effects were found. The authors concluded that bisphenol A is “unlikely to alter immune competence in adult rats”. The highest dose of bisphenol A tested was 25 million nanograms per kilogram of body weight per day, which is 125 million times higher than the new TDI derived by EFSA.
It is questionable whether the results from the mouse studies can be transferred to humans. Despite existing similarities between murine (mouse) and human immune systems, some important differences were identified, such as different maturation and regulatory processes in the immune system. Furthermore, the methods used for the measurements of the relative increase in Th17 cells are not standardised nor validated. There is no information on the measurement uncertainty associated with the tests and no positive controls are established. Therefore, a quality assessment of the measurements is not possible.
Epidemiological studies (population studies) conducted so far have not shown a causal link between bisphenol A intake and immunological effects, but some of these studies have methodological shortcomings.
In conclusion, the BfR considers the critical effect "relative increase in Th17 cells" used by EFSA as a basis for TDI derivation, as well as other immunological effects discussed by EFSA, as not suitable for predicting adverse health effects in humans. Therefore, they should not be used to derive a TDI.
After evaluating the available data, the BfR concludes that adverse immunological effects in humans - if present at all - are unlikely to occur at a bisphenol A exposure in the range of the TDI of 200 nanograms per kilogram of body weight per day.
2) Selection and weighting of the literature considered
Another point of criticism of the EFSA re-evaluation by the BfR concerns the selection and quality of the studies on which the EFSA re-evaluation is based. With a few exceptions, only studies from the period between 2013 and 2018 were evaluated. Relevant studies with earlier or later publication dates were not (sufficiently) considered, even if they contained relevant information. This contradicts the internationally recognised principles of risk assessment, to which EFSA also refers.
The BfR also sees partly fundamental differences in the assessment of the quality of the evaluated studies and accordingly the reliability of the data generated in them. For example, the mice from the key study on which the TDI is based were kept in polycarbonate cages. Since bisphenol A is a starting material in the production of polycarbonate, in the opinion of the BfR, a relevant background exposure of the animals is very likely and also documented in the literature for similar experiments. In addition, the animals were fed "standard feed", which was neither tested for its bisphenol A content nor specifically selected with regard to a low bisphenol A content. As comparable studies have shown, this also very likely results in a relevant additional bisphenol A intake of the study animals, which is why the actual dose administered is unclear. The study is therefore not suitable for a quantitative risk assessment in the opinion of the BfR.
3) Derivation of the conversion factor mouse - human
Hazard characterisation is mostly based on experimental data from animal studies. Humans and animals may differ with regard to the uptake of a substance into the blood, its distribution and metabolism in the body, and its excretion. Therefore, the administered critical doses in animal experiments are converted into corresponding intake levels for humans via appropriate conversion and assessment factors.
EFSA bases the calculation of the conversion factor (mouse-human) on a study on the time course of bisphenol A blood levels after a single dose in mice. However, far too few measurement points could be generated on too few animals (1 to 2 animals; no statistical meaningfulness) and only over a very short measurement period (one hour instead of 24 hours after dosing). From the BfR's point of view, the study is therefore unsuitable for determining a conversion factor. This study is contrasted with two other studies that come to clearly different results. These studies are based on sufficient measurement points over a period of 24 hours, in a statistically sufficient number of animals.
In the BfR's view, just the use of the above-mentioned study, which in the BfR's view is unsuitable for determining the mouse-human conversion factor, means that the TDI calculated by EFSA is too low by a factor of 10 to 100.
The large discrepancy between the TDI of the BfR and EFSA results from the choice of a different endpoint for the TDI derivation and a different conversion factor mouse - human as well as from a different methodological, guideline-compliant approach applied by the BfR to determine remaining uncertainties. The scientific divergences in the TDI derivation by BfR and EFSA are summarised in a document and are publicly available. (https://www.efsa.europa.eu/sites/default/files/2023- 04/bfr-efsa-art-30.pdf ).
The European Medicines Agency (EMA) has also drafted a divergence paper with EFSA. The EMA essentially criticises the same points as the BfR.
The BfR considers the approach of EFSA for derivation of a new TDI and the result as inappropriate with respect to several points:
1) Selection of the critical endpoint (adversity and human relevance)
EFSA lowered the daily tolerable intake (TDI) of bisphenol A to 0.2 nanograms per kilogram of body weight per day. This was based on data from studies on mice on the effect of bisphenol A on the immune system: In offspring of dams that were administered bisphenol A during pregnancy and during nursing, an increase in the percentage of certain immune cells (Th17 cells) in the spleen was measured. It should be noted that these (healthy) animals did not show any signs of an adverse effect. Other studies showed that bisphenol A can increase already existing inflammation in allergy models with sick animals. So far, mice are the only species in which the effects identified by EFSA have been studied.
Th17 cells are special T-helper cells that are mainly found in mucosal barriers (e.g. in the intestine) and play an important role there in the immune defence against fungal infections. In addition to this protective function, Th17 cells are also associated with inflammatory reactions such as psoriasis. Furthermore, there are indications that the relative number of Th17 cells is increased in autoimmune diseases such as rheumatoid arthritis. However, it is still unclear what role an increased Th17 cell count plays within the respective clinical pictures, i.e. whether the increased relative Th17 cell count causes the disease or is merely associated with it (association or causality). In clinical trials, drugs that counteract Th17 cells have not been successful in treating various of these diseases.
In healthy mice, the relative increase in Th17 cells had no adverse effect, for example no inflammatory reactions occurred. In an extensive study on rats, carried out within the framework of the US National Toxicology Program (NTP), the animals were administered bisphenol A daily over their entire lifespan - beginning in the womb and ending after two years. A variety of analyses were performed, including testing for over 530 possible effects on the immune system. No adverse (harmful) effects were found. The authors concluded that bisphenol A is “unlikely to alter immune competence in adult rats”. The highest dose of bisphenol A tested was 25 million nanograms per kilogram of body weight per day, which is 125 million times higher than the new TDI derived by EFSA.
It is questionable whether the results from the mouse studies can be transferred to humans. Despite existing similarities between murine (mouse) and human immune systems, some important differences were identified, such as different maturation and regulatory processes in the immune system. Furthermore, the methods used for the measurements of the relative increase in Th17 cells are not standardised nor validated. There is no information on the measurement uncertainty associated with the tests and no positive controls are established. Therefore, a quality assessment of the measurements is not possible.
Epidemiological studies (population studies) conducted so far have not shown a causal link between bisphenol A intake and immunological effects, but some of these studies have methodological shortcomings.
In conclusion, the BfR considers the critical effect "relative increase in Th17 cells" used by EFSA as a basis for TDI derivation, as well as other immunological effects discussed by EFSA, as not suitable for predicting adverse health effects in humans. Therefore, they should not be used to derive a TDI.
After evaluating the available data, the BfR concludes that adverse immunological effects in humans - if present at all - are unlikely to occur at a bisphenol A exposure in the range of the TDI of 200 nanograms per kilogram of body weight per day.
2) Selection and weighting of the literature considered
Another point of criticism of the EFSA re-evaluation by the BfR concerns the selection and quality of the studies on which the EFSA re-evaluation is based. With a few exceptions, only studies from the period between 2013 and 2018 were evaluated. Relevant studies with earlier or later publication dates were not (sufficiently) considered, even if they contained relevant information. This contradicts the internationally recognised principles of risk assessment, to which EFSA also refers.
The BfR also sees partly fundamental differences in the assessment of the quality of the evaluated studies and accordingly the reliability of the data generated in them. For example, the mice from the key study on which the TDI is based were kept in polycarbonate cages. Since bisphenol A is a starting material in the production of polycarbonate, in the opinion of the BfR, a relevant background exposure of the animals is very likely and also documented in the literature for similar experiments. In addition, the animals were fed "standard feed", which was neither tested for its bisphenol A content nor specifically selected with regard to a low bisphenol A content. As comparable studies have shown, this also very likely results in a relevant additional bisphenol A intake of the study animals, which is why the actual dose administered is unclear. The study is therefore not suitable for a quantitative risk assessment in the opinion of the BfR.
3) Derivation of the conversion factor mouse - human
Hazard characterisation is mostly based on experimental data from animal studies. Humans and animals may differ with regard to the uptake of a substance into the blood, its distribution and metabolism in the body, and its excretion. Therefore, the administered critical doses in animal experiments are converted into corresponding intake levels for humans via appropriate conversion and assessment factors.
EFSA bases the calculation of the conversion factor (mouse-human) on a study on the time course of bisphenol A blood levels after a single dose in mice. However, far too few measurement points could be generated on too few animals (1 to 2 animals; no statistical meaningfulness) and only over a very short measurement period (one hour instead of 24 hours after dosing). From the BfR's point of view, the study is therefore unsuitable for determining a conversion factor. This study is contrasted with two other studies that come to clearly different results. These studies are based on sufficient measurement points over a period of 24 hours, in a statistically sufficient number of animals.
In the BfR's view, just the use of the above-mentioned study, which in the BfR's view is unsuitable for determining the mouse-human conversion factor, means that the TDI calculated by EFSA is too low by a factor of 10 to 100.
The large discrepancy between the TDI of the BfR and EFSA results from the choice of a different endpoint for the TDI derivation and a different conversion factor mouse - human as well as from a different methodological, guideline-compliant approach applied by the BfR to determine remaining uncertainties. The scientific divergences in the TDI derivation by BfR and EFSA are summarised in a document and are publicly available. (https://www.efsa.europa.eu/sites/default/files/2023- 04/bfr-efsa-art-30.pdf ).
The European Medicines Agency (EMA) has also drafted a divergence paper with EFSA. The EMA essentially criticises the same points as the BfR.
