Report on the public consultation on the draft EFSA Bisphenol A (BPA) hazard assessment protocol

TECHNICAL REPORT APPROVED: 13 December 2017 doi:10.2903/sp.efsa.2017.en-1355 Report on the public consultation on the draft EFSA Bisphenol A (BPA) hazard assessment protocol European Food Safety Authority (EFSA), Ursula Gundert-Remy, Fulvio Barizzone, Cristina Croera, Claudio Putzu and Anna F. Castoldi Abstract The hazard assessment protocol to be used in the upcoming EFSA re-evaluation of Bisphenol A (BPA) has undergone a web-based public consultation from 30 June to 3 September 2017. Overall EFSA has received 151 comments from various interested parties including national agencies, non-governmental organisations, industry, academia and private individuals. The contributors of the consultation and other relevant parties were invited to a Workshop on BPA hazard assessment protocol held in Brussels on 14 September 2017. This report summarises the outcome of the public consultation, lists and offers a brief summary of the comments received and how they were taken into account when finalising the protocol for endorsement by the Panel on Food Contact Materials, Enzymes, Flavourings and Processing Aids (CEF). This document also includes in an Appendix the summary report on the above mentioned Workshop. European Food Safety Authority, 2017 Key words: BPA, hazard assessment methodology, public consultation, BPA workshop, Prometheus Requestor: EFSA Question number: EFSA-Q-2017-00491 Correspondence: FIP@efsa.europa.eu www.efsa.europa.eu/publications 1 EFSA Supporting publication 2017:EN-1355

Acknowledgements: EFSA wishes to thank Johanna Bodin, Cristina Bosetti, Rex FitzGerald, Annika Hanberg, Ulla Hass, Carlijn Hooijmans, Andrew A. Rooney, Christophe Rousselle, Henk van Loveren, Detlef Wölfle, and Julia Cara Carmona for the support provided to this scientific output. Suggested citation: EFSA (European Food Safety Authority), Gundert-Remy U, Barizzone F, Croera C, Putzu C and Castoldi AF, 2017. Report on the public consultation on the draft EFSA Bisphenol A (BPA) hazard assessment protocol. EFSA supporting publication 2017:EN-1355. 78 pp. doi:10.2903/sp.efsa.20yy.en-1355 ISSN: 2397-8325 European Food Safety Authority, 2017 Reproduction is authorised provided the source is acknowledged. www.efsa.europa.eu/publications 2 EFSA Supporting publication 2017:EN-1355

Summary The current report summarises the outcome of the web-based public consultation which was held on the draft bisphenol A (BPA) hazard assessment protocol prepared by an ad-hoc EFSA Working Group (WG) composed of experts nominated by national bodies and independent experts. This protocol was endorsed for public consultation by the EFSA Panel on Food Contact Materials, Enzymes, Flavourings and Processing Aids (CEF) on 14 June 2017. The consultation took place from 30 June to 3 September 2017. To complete the consultation process, EFSA also held a Workshop on BPA hazard assessment protocol in Brussels on 14 September 2017 to engage in discussion with all the stakeholders and other relevant parties. The present report also includes a summary of the outcome of this public scientific event. As a result of the web-based consultation, 151 comments were received from 16 different parties, including national agencies and governmental bodies, industry and industry associations, nongovernmental organisations, academia and private citizens. This report lists all the comments received, includes a brief summary of the most relevant ones and explains how they were addressed by the WG on BPA hazard assessment protocol. The Workshop on BPA hazard assessment protocol had a two-fold goal: (i) to discuss with contributors of the consultation and other relevant parties, scientific experts and EFSA staff the comments received during the public consultation, to ensure a full understanding of the submissions; (ii) to gather new suggestions from meeting attendees directed at improving the protocol. The protocol was revised according to the suggestions received from both processes. The updated version was endorsed by the CEF Panel during its 72nd Plenary meeting on 30 November 2017 and published on the EFSA website concomitantly with this report. EFSA thanks all stakeholders for their valuable contributions. www.efsa.europa.eu/publications 3 EFSA Supporting publication 2017:EN-1355

Table of contents Abstract... 1 Summary... 3 1. Introduction... 5 1.1. Background and Terms of Reference as provided by EFSA... 5 1.2. Background and Terms of Reference as provided by the EC... 5 2. Major points addressed by the comments... 6 2.1. Problem formulation and Inclusion/Exclusion criteria... 6 2.2. Appraisal of the internal validity and relevance of the studies... 7 2.3. Weight of evidence, assignment of likelihood to each effect and hazard characterisation... 8 2.4. General comments... 8 References... 10 Abbreviations... 11 Appendix A Contributors to the public consultation... 12 Appendix B Comments submitted to EFSA on its draft BPA hazard assessment protocol... 13 Appendix C Summary Report on the Workshop on bisphenol A (BPA) hazard assessment protocol (Brussels, 14/09/2017)... 77 www.efsa.europa.eu/publications 4 EFSA Supporting publication 2017:EN-1355

1. Introduction The development of a protocol detailing the strategy for the hazard assessment of BPA (hazard identification and characterisation) was initiated as an EFSA self-task, as described in mandate M- 2016-0207 (EFSA-Q-2016-00673). This was triggered by the need to ensure that the EFSA Panel on Food Contact Materials, Enzymes, Flavourings and Processing Aids (CEF Panel) is prepared for the upcoming re-evaluation of the safety for consumers of BPA, once the results of the two-year CLARITY- BPA project (Consortium Linking Academic and Regulatory Insights on BPA Toxicity) developed by the US National Institute of Environmental Health Sciences (NIEHS) and the US Food and Drugs Administration (FDA) and run under the auspices of the US National Toxicology Programme (NTP) become available in 2017/2018. After the initiation of this work, EFSA received an additional mandate from the European Commission (EC) (EFSA-Q-2016-00635) to re-evaluate the safety for consumers of BPA, which required setting-up a BPA hazard assessment protocol as a first step. Both these two mandates from EFSA and the European Commission mention the need to organise a public consultation on the draft protocol prior to publication. A web-based consultation was therefore run on the draft protocol from 30 June to 3 September 2017, with a total of 151 comments received from 16 different parties (see Appendix A), including national agencies and governmental bodies, scientific and industry associations, non-governmental organisations, academia and private citizens. The written comments are listed and addressed individually in Appendix B of this report. They are tabulated by the section of the draft protocol they refer to, and in alphabetical order according to the submitter s name. Comments submitted formally on behalf of an organisation appear with the name of the affiliation. Private submitters are instead listed anonymously. This document also summarises the main comments and criticism received and explains how they were addressed in the protocol by the WG. Comments suggesting editorial changes have been directly addressed in the text of the protocol, if they were considered appropriate/applicable. To complete the public consultation process, EFSA also held a public scientific event, i.e. Workshop on BPA hazard assessment protocol, to engage in discussion with all the stakeholders and other relevant parties, and to gather additional feedback on the draft protocol. This event took place in Brussels on 14 September 2017. This report also includes a summary of the Workshop (see Appendix C). The WG on BPA hazard assessment protocol carefully examined all the comments and feedback received and took them into account when finalising the protocol before its endorsement by the CEF Panel in November 2017. 1.1. Background and Terms of Reference as provided by EFSA The need to have a public consultation is described in the terms of reference of the internal mandate (work package 1), as shown below. Terms of reference of Work Package 1 Communication plan Given the sensitivity of the topic it is foreseen to launch a web-based public consultation in 2017. A dedicated meeting with all the Member States representatives and stakeholders should be considered to present and discuss this draft protocol prior to its finalisation. The protocol, with subsequent amendments (if any) will be published prior to the start of the evaluation, after endorsement by the CEF Panel, and as a part of the final scientific opinion. 1.2. Background and Terms of Reference as provided by the EC Below is an extract of the background to the mandate received from the EC mentioning the need to have a public consultation. It is essential that well-defined and transparent scientific criteria concerning the selection of the new scientific studies are laid down in advance of the re-evaluation. This would enable a www.efsa.europa.eu/publications 5 EFSA Supporting publication 2017:EN-1355

comprehensive assessment of all relevant and adequate studies, and avoid the need to react to adhoc requests concerning individual scientific studies. The efficiency of work would thus be maximised. My services have taken due note of the work that you have already undertaken in this respect and welcome the establishment of an ad hoc Working Group of experts including those from EFSA, external experts and those from Member States to set clear review criteria for the scientific evidence on BPA. Therefore, taking into account the timing for the activities involved in this work as foreseen by EFSA, including a public consultation, as the first part of this mandate the Commission therefore kindly requests EFSA: - To establish a protocol detailing the criteria for new study inclusion and for toxicological evidence appraisal for the re-evaluation of BPA as soon as possible, to ensure an efficient and transparent re-assessment of BPA. Once this work is complete, the Commission will kindly request EFSA the second part of this mandate: - To re-evaluate the risks to public health related to the presence of BPA in foodstuffs, taking into account the results of all relevant scientific data insofar as it meets the criteria laid down in the protocol mentioned above and in line with the terms of reference set out in the annex to this letter.... 2. Major points addressed by the comments 2.1. Problem formulation and Inclusion/Exclusion criteria 2.1.1. It was pointed out that the use of narrative approaches should be limited as much as possible and that justifications should be made when these are used EFSA agreed to provide better justifications in the protocol on the rationale followed when deciding which sub-questions are to be addressed narratively or systematically (please consult section 2.6 of the revised protocol). Those studies that have the potential to provide a reference value for setting a Tolerable Daily Intake (TDI) will be addressed systematically. Other evidence, including crosssectional studies, genotoxicity studies, toxicokinetic and Mode of Action (MoA) studies will be addressed narratively. EFSA acknowledges that the use of narrative approaches is not as rigorous and methodologically sound as the use of systematic methodologies. The current procedures followed by EFSA allow, however, making use of narrative approaches, especially when evaluations need to be done in a timely manner and resources are limited (EFSA, 2015). 2.1.2. The choice to evaluate studies published from 01/01/2013 onwards not already evaluated by EFSA was criticised EFSA has thoroughly discussed this issue and has decided to maintain its previous decision to evaluate only BPA studies published from 2013 onwards that have never been appraised by EFSA (i.e. BPA risk assessment opinion (EFSA CEF Panel, 2015) and CEF Panel statement on BPA immunotoxicity (2016)). EFSA considers that the review methodology previously employed by the CEF Panel is robust and that, although less structured, it implicitly fulfils the criteria specified in the new methodology. In order to prove that this assumption is correct, the ad hoc WG, which will be appointed to carry out the next BPA evaluation, will test at a very early stage the new methodology (inclusion/exclusion criteria and critical appraisal) on a selection of papers previously appraised by EFSA (e.g. previously concluded to be of high, medium or low reliability). If a significant divergence is observed between the new and old methodology, then both EFSA hierarchy and the EC will be informed of this outcome and a decision on the way forward will be agreed. When selecting the cut-off date of the literature search, EFSA considered the points below: i. The mandate received from the EC specifically asks to review new BPA publications not previously appraised by EFSA which may challenge the t-tdi of BPA. www.efsa.europa.eu/publications 6 EFSA Supporting publication 2017:EN-1355

ii. iii. It has been estimated that selecting, including and appraising all the evidence published on BPA so far (without any time limit) as opposed to only that published after 2013 would require almost 3-fold more time (roughly 539 vs 190 FTE days) just for animal and human evidence appraisal, without including time for discussions, opinion drafting, narrative review of in vitro/moa studies, weighing and integrating the evidence, etc.). The WG involved in the next BPA evaluation will test the new methodology on a selection of papers previously appraised in the 2015 BPA opinion and will compare the outcomes. This pilot phase should ensure that the methodology used for the 2015 BPA opinion and 2016 statement on immunotoxicity is robust, even though it is not as structured as the new one. 2.1.3. Routes of exposure different than oral, sub-cutaneous or dermal are potentially relevant and should not be excluded from the assessment of animal studies EFSA has agreed to include other routes of exposure in the assessment (i.e. intraperitoneal, intravenous and inhalation); the protocol was updated accordingly. Different cut-off doses will be applied for inclusion of the studies considering the specific route of exposure: for subcutaneous, intravenous and intraperitoneal exposure, the cut-off will be 0.5 mg/kg bw day, while for inhalation and dermal no cut-off value will be applied as a result of difficulties to standardise the dose levels (please consult Section 2.2 of the revised protocol for more information). 2.1.4. Cross-sectional studies should not be excluded from the assessment. While these studies taken singularly are highly unreliable, if there is a great number of them, aggregating them could improve the precision. Excluding studies on the basis of study design a priori is not in line with systematic methodologies; all studies should be included and only then, if justifiable, not further evaluated EFSA has agreed to evaluate and address cross-sectional studies in a narrative manner, and not to exclude them from the assessment. A systematic approach will not be used since EFSA considers studies with cross-sectional design unsuitable to establish a causal dose-response relationship and to identify a reference point for setting a full TDI for BPA, which is the scope of the BPA evaluation. 2.1.5. Non-English studies are potential source of relevant information and should not be excluded from the assessment EFSA acknowledges that non-english language studies may potentially contain relevant information and has therefore decided to take this into account in the uncertainty analysis. In support of this decision are the results of a pilot exercise that indicates that a significant majority of the outputs of the planned literature search (over 85%) would be in English in any case. Furthermore, when considering the sample of studies reaching full-text screening, over 95% were found to be in English. For more information please consult Section 4.2.1 and Appendix E of the revised protocol. 2.2. Appraisal of the internal validity and relevance of the studies 2.2.1. Differences between quality and Risk of Bias (RoB) were unclear. The methodology was criticised since it lacked validation and was too complicated. In the previous version of the protocol, the study internal validity was appraised in two separate steps: the risk of bias appraisal considered only the elements introducing a systematic difference between the control and the exposed group whereas through the quality appraisal tool elements introducing systematic differences in all (control and exposed) groups were evaluated. www.efsa.europa.eu/publications 7 EFSA Supporting publication 2017:EN-1355

Following the suggestions received, the EFSA WG on BPA hazard assessment protocol decided to simplify the overall process and to assess the study internal validity in a single step. This proposal will be supported by the use of an adapted version of the well-established NTP RoB tool complemented with aspects taken from the previous quality tool (Section 6 of the revised protocol). 2.2.2. The two-step approach described in the draft protocol to assess relevance received criticism for its unclear purpose In the previous version of the protocol two aspects of relevance ( Relevance of the study to a hazard sub-question and Relevance of the effect to humans ) were proposed to be considered separately and assessed at a different step of the appraisal process. EFSA acknowledges the criticism received and has amended the protocol accordingly. Relevance of an effect to humans (i.e. external validity) will only be evaluated for animal data (for human data the effect is considered relevant by default). For animal studies, the external validity of the animal model used and of the endpoint examined will be assessed after the evaluation of the internal validity (see Section 7 of the revised protocol). The external validity judgement (directly relevant, indirectly relevant or not relevant) will be one of the criteria used to grade the confidence in the body of evidence (Section 8 of the revised protocol). 2.3. Weight of evidence, assignment of likelihood to each effect and hazard characterisation 2.3.1. The weight of evidence was commented as not being transparent enough and too strongly reliant on expert judgement. In order to provide more structure and transparency to the process, EFSA has updated the protocol (see Section 8 in the revised version) by proposing to use an approach to rating evidence similar to GRADE (Acronym for Grading of Recommendations Assessment, Development and Evaluation) developed by the NTP Office of Health Assessment and Translation (NTP-OHAT, 2015). 2.3.2. The approach described in the protocol not to consider studies of effects categorised as As Likely as Not or dose-response analysis was considered inappropriate. EFSA holds the opinion that studies supporting effects that are classified as As Likely as Not (ALAN) are not suitable for a dose-response analysis. However, these studies will be considered in the analysis of uncertainties and could have a direct impact on the TDI as additional uncertainty factors could be applied on the point of departure (EFSA Scientific Committee, 2017) 2.4. General comments 2.4.1. The added value of performing the hazard identification of BPA for reproductive toxicity and endocrine disruption (ED) was questioned, given the recent classification of BPA as a Substance of Very High Concern (SVHC) by the European Chemicals Agency (ECHA). The purpose of EFSA s new hazard assessment is not to conclude on a potential ED-mediated mechanism for BPA, as was the case of the ECHA report (ECHA, 2017). Instead EFSA s next opinion will aim at identifying the dose at which no potential adverse effects of BPA occur. Benchmark dose modelling of the dose-response relationship from suitable studies will result in the reference point, based on which a TDI can be derived. Thus, EFSA will base the TDI on the most sensitive effect, irrespective of whether or not the underlying mechanism or MoA is known. The assessment will therefore not be limited to endpoints for which an ED-mediated mechanism can be anticipated. Of course, if a clear dose-response relationship cannot be established for a certain adverse effect, in the absence of a clear BMDL, appropriate considerations will be made. www.efsa.europa.eu/publications 8 EFSA Supporting publication 2017:EN-1355

The review submitted to ECHA by the French Agency for Food, Environmental and Occupational Health & Safety (ANSES) (ECHA, 2017) does not use the same specifications as defined in the EFSA BPA protocol, such as inclusion/exclusion criteria and appraisal tools. It would be methodologically inconsistent and not scientifically sound to apply different inclusion and appraisal criteria to the evidence published after 2013, depending on whether ANSES has already reviewed part of it or not. As explained in Section 2.1.2 above, the WG involved in the next BPA evaluation will test the new methodology on a selection of papers previously appraised in the 2015 BPA opinion EFSA (e.g. previously concluded to be of high, medium or low reliability) and will compare the outcomes to verify the robustness of the previous applied EFSA methodology. 2.4.2. Sometimes studies examining multiple chemicals do not discuss the chemicals with null results in the abstract. These null results represent an important aspect of a true WoE approach. It is in our point of view therefore important that the Panel does not exclude studies if BPA is not mentioned in the study s title or abstract While EFSA is aware that studies reporting null results may be missed during the Title/Abstract screening procedures, the only way to avoid excluding these studies would be to read the full-text of all the outputs of the search. Given well over 10,000 outputs are expected to be retrieved by the search, the current time and resource constraints would not allow doing so. The WG assessed the approximate impact of such a decision though a pilot tests (see Appendix F for details). On the basis of the results, it seems reasonable to infer that the first screening step of papers on the basis of title and abstract only would not lead to an inappropriate exclusion of relevant null studies, compromising the overall assessment. This issue will be considered in the uncertainty analysis (see section 4.1 and Appendix F of the revised protocol for more details). 2.4.3. The approach used by EFSA to consider financial conflicts of interest was questioned by many comments EFSA would like to clarify that information regarding the source of funding of the studies will be collected during the stage of data extraction. However, while this information will be reported in a transparent manner in the evaluation, it will not be recognised as an element to be considered when appraising the internal validity of the studies. The latter will be based on scientific grounds. www.efsa.europa.eu/publications 9 EFSA Supporting publication 2017:EN-1355

References Bauer, S. M., Roy, A., Emo, J., Chapman, T. J., Georas, S. N., & Lawrence, B. P. (2012). The effects of maternal exposure to bisphenol A on allergic lung inflammation into adulthood. Toxicological Sciences, 130(1), 82-93. ECHA, ANNEX XV Identification of 4,4'-isopropylidenediphenol (bisphenol a) as SVHC. Available online: https://echa.europa.eu/documents/10162/93bf4be3-9af6-d7ca-8b07-4e8fb42bad11 EFSA (European Food Safety Authority), 2015. Scientific report on Principles and process for dealing with data and evidence in scientific assessments. EFSA Journal 2015;13(5):4121, 35 pp. doi:10.2903/j.efsa.2015.4121 EFSA CEF Panel (EFSA Panel on Food Contact Materials, Enzymes, Flavourings and Processing Aids) A statement on the developmental immunotoxicity of bisphenol A (BPA): answer to the question from the Dutch Ministry of Health, Welfare and Sport. EFSA Journal 2016;14(10):4580, 22 pp. doi:10.2903/j.efsa.2016.4580 Tyl RW, Myers CB, Marr MC, Sloan CS, Castillo NP, Veselica MM, Seely JC, Dimond SS, Van Miller JP, Shiotsuka RN, Beyer D, Hentges SG and Waechter JM, 2008. Two-generation reproductive toxicity study of dietary bisphenol a in CD-1 (Swiss) mice. Toxicological Sciences, 104, 362-384. NTP-OHAT, 2015. Handbook for conducting a literature-based health assessment using OHAT approach for systematic review and evidence integration. Available online: http://ntp.niehs.nih.gov/ntp/ohat/pubs/handbookjan2015_508. pdf www.efsa.europa.eu/publications 10 EFSA Supporting publication 2017:EN-1355

Abbreviations BPA WG CEF Panel NTP FDA EC ECHA MoA GRADE OHAT ANSES RoB WoE PROMETHEUS Bisphenol A Working Group Panel on Food Contact Materials, Enzymes, Flavourings and Processing Aids National Toxicology Programme Food and Drugs Administration European Commission European Chemical Agency Mode of Action Grading of Recommendations Assessment, Development and Evaluation Office of Health Assessment and Translation French Agency for Food, Environmental and Occupational Health & Safety Risk of Bias Weight of Evidence Promoting Methods for Evidence Use in Scientific Assessments www.efsa.europa.eu/publications 11 EFSA Supporting publication 2017:EN-1355

Appendix A Contributors to the public consultation Organisation Country Number of comments ANSES FRA 50 Breast Cancer UK UK 1 Endocrine Society USA 10 European Commission - 1 Evidence-Based Toxicology Collaboration (EBTC) USA 4 Fera UK 11 Food safety systems GmbH CHE 2 Lancaster Environment Centre UK 8 Metal Packaging Europe BEL 15 National Institute of Public Health and Environment (RIVM) NED 1 PlasticsEurope BEL 6 R.I.S.K. Consultancy BEL 10 Technical University of Denmark DNK 5 The Endocrine Disruption Exchange USA 19 University of Melbourne AUS 7 University of Sussex UK 1 www.efsa.europa.eu/publications 12 EFSA Supporting publication 2017:EN-1355

Appendix B Comments submitted to EFSA on its draft BPA hazard assessment protocol # Contributor Country Section Comments Answer to the comments 1. ANSES FR Generic Comments ANSES welcomes the development of this protocol detailing the strategy for hazard assessment of bisphenol A (BPA). The document introduction gives EFSA s interpretation of the Terms of Reference. It clearly indicates that the development of the protocol detailing the strategy for the hazard assessment of BPA (hazard identification and characterization) was initiated by EFSA as a self-task triggered by the need to re evaluate the safety of BPA for consumers, once the results of the two-year US NTP/FDA toxicity study will become available. The section on problem formulation includes a specific paragraph on the terms of reference mentioning that EFSA will: establish the criteria for new studies inclusion and for toxicological evidence appraisal for the re-evaluation of BPA in order to re-evaluate the risks to public health related to the presence of bisphenol A (BPA) in foodstuff. (...) In particular, the re-evaluation of BPA should take into consideration new data available from the results of the US NTP/ FDA study due in 2017 as well as all other new available information not previously evaluated by EFSA and which fulfil the criteria laid down in an established protocol. However, if the risk evaluation related to the presence of BPA in foodstuff has necessarily to be reconsidered to integrate the upcoming US NTP/ FDA data, ANSES questions the added value of re-assessing the hazard identification of BPA in the context of the recent SVHC identification for reprotoxicity and endocrine disruption for human health. This statement has been approved unanimously by the Member States Committee at ECHA level on 14 June 2017 in the context of the REACH regulation : France presented a proposal in accordance with Article 59(3) and Annex XV of the REACH Regulation (2 March 2017, submission number a) ANSES questions the added value of reassessing the hazard identification of BPA in the context of the recent SVHC identification for reprotoxicity and endocrine disruption for human health. This issue is addressed in Section 2.4.1 of this report. b) In general, experts from ANSES were not in favour of this two-step evaluation: the division of tasks as described in the document (selection and extraction/scoring of the database by scientists and evaluation of the weight of evidence by experts, the two sets of persons being different) could be an obstacle difficult to overcome. EFSA will give clear and effective instructions to the scientists (working in couples) performing the title and abstract and Full Text screening of the papers and will closely monitor the process quality, via continuous communication and testing the procedure. The members of the WG on BPA review will be in charge of the appraisal of the studies internal and external validity as well as all the subsequent steps critical for the hazard assessment. c) Eventually the panel opinion will be based in part on expert s judgment which requires that the basis and criteria be better explained. www.efsa.europa.eu/publications 13 EFSA Supporting publication 2017:EN-1355

SPS-013185-17) on the identification of BPA as a substance of very high concern due to its endocrine disruptive properties for which there is scientific evidence of probable serious effects to human health which give rise to an equivalent level of concern to those of other substances listed in paragraphs (a) to (e) of Article 57 of REACH. The Annex XV dossier was circulated to Member States on 9 March 2017 and the Annex XV report was made available to interested parties on the ECHA website on the same day according to Articles 59(3) and 59(4). Comments were received from both Member States and interested parties on the proposal. The dossier was referred to the Member State Committee on 22 May 2017 and discussed in the meeting on 12-16 June 2017 of the Member State Committee. In this dossier, adverse effects of BPA on the female reproductive function, on the mammary gland, on the cognitive function, and on the metabolism were confirmed based on a literature review, including recent publications published before May 2016. Modes of action were evaluated and the plausibility of a causal link between endocrine properties of BPA through pathways that commonly involve disruption of estrogenic regulation and adverse effects was considered sufficient. It was not excluded that BPA may also alter other physiological functions, e.g. the immune function, through a similar ED MoA but the level of evidence was considered insufficient at the moment for this effect to be presented. To enhance the clarity and the transparency of the process of integrating and weighing the evidence, EFSA has updated the protocol: the internal validity evaluation has been simplified, the evaluation of external validity has been made clearer, and an approach similar to GRADE from OHAT NTP has now been proposed (see Sections 2.2.1, 2.2.2, 2.3.1 above). d) The methodology will be applied only to new studies that were not assessed in the previous evaluation leading to the t-tdi which could create inconsistencies in the way some key studies would have been assessed (depending if they have been published before or after the t-tdi). This point is addressed in Section 2.1.2 of the current report. ANSES considers that the reevaluation process should rely on these effects that were acknowledged at the European level to derive a new TDi. Therefore the protocol should describe the methodology for hazard characterization and not for hazard identification. When the US NTP/FDA study results become available and if considered appropriate for modifying the SVHC identification, then ECHA should be mandated to update the hazard identification. This would avoid discrepancies between ECHA and EFSA conclusions which could lead to misunderstanding. ANSES also recommends modifying the title of the document as Methodology to take into account new data for BPA risk assessment www.efsa.europa.eu/publications 14 EFSA Supporting publication 2017:EN-1355

The document content seems useful and is relatively complete. It offers better transparency on the process. It is well written and gives clear orientations on how to proceed. ANSES approves the choice made by EFSA s working group to rely on the OHAT and Sci-Rap approaches and to adapt them to BPA. ANSES considers that the approach taken in the document could be applied to other compounds. The methodology seemingly requires a lot of resources especially if there are many new papers. EFSA should therefore clearly indicate the level of resources needed to do the evaluation. It is also likely to increase the workload and make difficult for the experts to have a complete view on the entire database and therefore minimize an integrated evaluation. In general, experts from ANSES were not in favor of this two-step evaluation: the division of tasks as described in the document (selection and extraction/scoring of the database by scientists and evaluation of the weight of evidence by experts, the two sets of persons being different) could be an obstacle difficult to overcome. Moreover, some sections require further clarification particularly with regards to the methodology for the follow up of the steps. It would be useful to explain in more details how the protocol was elaborated (test-phase s performed to check if the protocol is comprehensive and applicable and if it leads to consistent results). It would be also beneficial to clarify by whom and how EFSA anticipates that it will be used. Eventually the panel opinion will be based in part on expert s judgment which requires that the basis and criteria be better explained. The methodology will be applied only to new studies that were not assessed in the previous evaluation leading to the t-tdi which could create inconsistencies in the way some key studies will been assessed (depending if they have been published before or after the t-tdi). This could apply In particular to some www.efsa.europa.eu/publications 15 EFSA Supporting publication 2017:EN-1355

2. Breast Cancer UK GBR Generic Comments endpoints such as effects of BPA on metabolism or on neurodevelopment, since most of the dataset were published during the last 2-3 years. Therefore, it would be better to assess the whole database with this new protocol, irrespectively of the year of publication of the studies. It should be reminded that in the case of BPA, the weight of evidence is very important for subtle effects such as metabolism. Each study, even the oldest, may help to understand the global picture. In this case it will not dramatically increase the burden of work. Breast Cancer UK welcomes the opportunity to provide feedback on the protocol EFSA plans to use to reassess the toxicology of bisphenol A. Breast Cancer UK is a charity which aims to prevent breast cancer by reducing public exposure to hazardous chemicals in the environment. We are especially concerned about the potential role of exposures to environmental chemicals, such as bisphenol A, in increasing breast cancer risk. Overall, the protocol presented by EFSA will provide an appropriate assessment of the hazards associated with BPA, although we do have some concerns with details of some of the methods proposed. We agree with the use of a systematic review structure (line 182 and line 231), and believe the search strings described (Appendix A) are comprehensive and the approach is robust. However, the proposed methodology appears to lack clarity in certain areas. Specifically, the assessment of the reliability of the evidence, in particular the quality of assessment (line 441), is not clearly defined and appears subjective, and the weight of evidence assessment (lines 628-683) is insufficiently clear. We also have concerns regarding the date cut-off proposed for the studies to be considered. One of the terms of reference (line 143) is to re-evaluate the risks to public health related to the presence of bisphenol A in foodstuffs, and in particular this reevaluation should take into consideration new data. This doesn t mean previously appraised data cannot be reconsidered. The protocol goes on to state that few studies published after 31/12/2012 and already appraised by the CEF panel in its 2015 opinion or in its 2016 statement on immunotoxicity (EFSA CEF panel 2016) will not be reappraised a. Specifically, the assessment of the reliability of the evidence, in particular the quality of assessment (line 441), is not clearly defined and appears subjective, and the weight of evidence assessment (lines 628-683) is insufficiently clear. The appraisal of the internal validity of the studies was simplified; see Section 2.2.1 of the current report. The WoE approach was also modified to be clearer, better structured and transparent, see Section 2.3.1 of the current report. b. One of the terms of reference (line 143) is to re-evaluate the risks to public health related to the presence of bisphenol A in foodstuffs, and in particular this reevaluation should take into consideration new data. This doesn t mean previously appraised data cannot be re-considered. This point is addressed in Section 2.1.2 of the current report. EFSA would like to clarify that the Terms of Reference quoted in line 143 also mention that evidence not previously evaluated by EFSA should be considered. www.efsa.europa.eu/publications 16 EFSA Supporting publication 2017:EN-1355

(lines 184-186). As a different methodology will be used, and older studies are few in number, it would be preferable to include these in the new assessment, to avoid any potential bias. Finally, we would welcome inclusion of a meta-analysis, as this would provide a summary result which is clear and easy to interpret. As such, we propose the inclusion of a protocol for meta-analysis in the final hazard assessment protocol. c. The protocol goes on to state that few studies published after 31/12/2012 and already appraised by the CEF panel in its 2015 opinion or in its 2016 statement on immunotoxicity (EFSA CEF panel 2016) will not be reappraised (lines 184-186). This point is addressed in Section 2.1.2 of the current report d. Finally, we would welcome inclusion of a meta-analysis, as this would provide a summary result which is clear and easy to interpret. As such, we propose the inclusion of a protocol for meta-analysis in the final hazard assessment protocol. EFSA acknowledges that meta-analysis was not mentioned in the draft version of this protocol. The new protocol has been revised in this respect. The process will envisage the use of a meta-analysis for each endpoint, if the homogeneity of the key characteristics of the studies allows it (see Section 8.1 of revised protocol). Human studies will have to show homogeneity for variables such as study design (e.g. cross-sectional, cohort), population, period and duration of the exposure, source of exposure data, level of exposure (e.g. expressed as quantiles), outcome measured, confounders and degree of internal validity. Experimental animal studies will have to be comparable for variables such as animal model (species, strain, sex, and genetic background), life stage of the animals at treatment onset and at outcome assessment, exposure route, dose levels, duration and frequency of the treatment, sampling time at measurements, health outcome tested, availability of raw data, degree of individual study internal validity. www.efsa.europa.eu/publications 17 EFSA Supporting publication 2017:EN-1355

3. Citizen under private capacity GBR Generic Comments Expert selection. The draft protocol does not specify how experts will be selected to participate in the Working Group to implement the protocol (step 1), nor how experts will be selected to participate in judgements on particular issues within the protocol (step 2). EFSA has existing procedures for step 1. Structured procedures for step 2 are included in EFSA s 2014 Guidance on Expert Knowledge Elicitation, involving first an analysis of what expertise is needed, then searching for a long list of suitable experts, and selecting those who will participate. Questions to consider include whether only the 2 experts who review the studies participate in judgements on them, or whether other members of the working group with relevant expertise also do, and whether experts from outside the working group (e.g. from the Panel) participate in judgements for which they have relevant expertise. Some of the selection procedures in the elicitation guidance might also be helpful when selecting the working group as a whole (step 1). a) The draft protocol does not specify how experts will be selected to participate in the Working Group to implement the protocol (step 1), nor how experts will be selected to participate in judgements on particular issues within the protocol (step 2). Step 1: More information was provided on the selection of the experts for the evaluation (see section 6 of the revised protocol). Step 2: Expert knowledge elicitation is mentioned in Section 10, the EFSA s 2014 Guidance on Expert Knowledge Elicitation has been referenced. b) Questions to consider include whether only the 2 experts who review the studies participate in judgements on them, or whether other members of the working group with relevant expertise also do The other members of the WG with relevant expertise will be participating in the decision process. c) and whether experts from outside the working group (e.g. from the Panel) participate in judgements for which they have relevant expertise The members of the EFSA CEF Panel with relevant expertise will be participating in judgements for which they have relevant expertise. 4. Citizen under private capacity CHE Generic Comments Dear Sirs, I have a general comment: First and foremost, I consider a public consultation on a scientific question totally inappropriate. Science is based on This comment does not address specific aspects of the protocol. The public consultation and the Workshop on the BPA protocol were addressed to a wide variety of EFSA www.efsa.europa.eu/publications 18 EFSA Supporting publication 2017:EN-1355

scientific findings and not on public opinions. Fact-finding by democracy?! - You will end up with a political opinion, based on a few participants of unknown bias. Secondly, EFSA is employed and paid to provide scientific information, based on which politicians may or may not pass legislation. Stick to that brief. The course you choose now will undermine if not destroy your credibility! My third point: renew your expert panel by choosing unbiased scientists and leave politics to politicians. audience such as national agencies, governmental bodies, industry and industry associations, nongovernmental organisations, academia and private citizens. As a result of these consultations, most of the comments and constructive feedback received did indeed address scientific aspects of the protocol and we felt that they contributed to improving our document. EFSA strongly believes in the added value of sharing and publicly consulting on its work. 5. Endocrine Society USA Generic Comments I wish you success Dear Members of the EFSA Unit on Food Ingredients and Packaging, The Endocrine Society appreciates the opportunity to participate in the public consultation on the draft Bisphenol A (BPA) hazard assessment protocol. Founded in 1916, the Endocrine Society is the world s oldest, largest, and most active organization devoted to research on hormones and the clinical practice of endocrinology. The Endocrine Society s membership consists of over 18,000 scientists, physicians, educators, nurses, and students in more than 100 countries. Society members represent all basic, applied and clinical interests in endocrinology. Included among the Society s members are the world s leading experts on the health effects of endocrine-disrupting chemicals (EDCs) including BPA. This guidance is a significant improvement over prior evaluations of BPA proposed or completed by the European Food Safety Authority (EFSA). We are particularly impressed with the transparent and systematic methodology to evaluate the literature, grounded in knowledge from clinical and environmental health. However, there are important issues that the Endocrine Society recommends addressing prior to finalizing the protocol. Our detailed comments are provided in the webbased form; to summarize, we recommend: Recommendations listed at the end of the comment a) Replacing breastfed in line 190 to include infants exposed through e.g., consumer products The wording was changed to infants. b) Removing the classification scheme in Table 1. The table only reported the outcome of the previous assessment and would not have been used when performing the assessment. However, the table was removed to avoid potential misunderstandings. c) Cautiously interpret toxicokinetic studies that are unable to know or control BPA levels. A member of the WG with expertise in toxicokinetics will have the knowledge to ensure an appropriate interpretation of these studies. www.efsa.europa.eu/publications 19 EFSA Supporting publication 2017:EN-1355

Replacing breastfed in line 190 to include infants exposed through e.g., consumer products. Removing the classification scheme in Table 1. Cautiously interpret toxicokinetic studies that are unable to know or control BPA levels. Cautiously interpret (or avoid interpreting) extrapolations from toxicokinetic studies to biomonitoring studies. Review important literature prior to 2013 for many endpoints. Assess the activity of BPA when co-exposure occurs with endogenous hormones. Include cross-sectional study designs in the inclusionary criteria. Identify sources of funding in studies, to reduce risk of bias. Ensure that expert judgment includes scientists with expertise in hormonal systems and endocrinology in all steps, but particularly the completion of table 10. Clarify the term sufficient number of animals and provide additional explanation throughout section 9. Thank you for considering the Endocrine Society s comments. d) Review important literature prior to 2013 for many endpoints. This point is addressed in Section 2.1.2 of the current report. e) Assess the activity of BPA when coexposure occurs with endogenous hormones. Only those studies having at least one arm with exposure to BPA alone will be considered. Studies reporting on mixtures only cannot be used for doseresponse analysis. f) Identify sources of funding in studies, to reduce risk of bias. This point is addressed in Section 2.4.3 of this report. g) Ensure that expert judgment includes scientists with expertise in hormonal systems and endocrinology in all steps, but particularly the completion of table 10. The expertise considered critical for the evaluation includes endocrinology/metabolism as described in Section 6 of the revised protocol. h) Clarify the term sufficient number of animals and provide additional explanation throughout section 9. Sample size should be large enough to ensure sufficient statistical power to detect any effects on the endpoints measured. This includes considerations of the background incidence and variability of the measured effects, as well as the method of analysis. OECD test guidelines provide recommendations for www.efsa.europa.eu/publications 20 EFSA Supporting publication 2017:EN-1355

number of animals per treatment group for different study types and endpoint measurements. However, primary consideration should be given to justifications for sample size provided by study authors, if stated. 6. Lancaster Environment Centre GBR Generic Comments Overall / general comments ================= A. The publication of a protocol for public comment prior to conduct of the hazard assessment is welcome, given the potential to afford greater transparency, allow scrutiny and therefore potential improvement of methods by the broader community, and potentially limit expectation bias from influencing the interpretation of the evidence during conduct of the review. B. This transparency and minimisation of potential bias can only be achieved if there is sufficient detail in the methods described in the protocol such that each decision can be anticipated and compared to the original plan. C. The ambition to use systematic methods, where possible, is EFSA acknowledges that Section 9 of the protocol was not sufficiently clear and therefore has modified it extensively. Relevance is now covered by Section 7 of the revised protocol entitled external validity. In this context it has been clarified that animal models are considered not relevant to humans if they differ in terms of target anatomical or patho-physiological features for the chemical under investigation. The criteria that have been introduced in the revised protocol for evaluating the relevance of the animal model and of the endpoint to human were extracted from SciRAP to provide more transparency to the judgement (www.scirap.org; Beronius et al., 2014). Adversity will be considered by expert judgement (with transparent justifications), given the broad number of endpoints examined. a) Points A, B, C and H EFSA is grateful for the comments b) Reply to point D The methodology described in the protocol was made more structured and transparent. See revised protocol and Sections 2.2.1, 2.2.2, 2.3.1 of the current report. c) Reply to points E and F This point is addressed in Section 2.2.1 and 2.3.1 of the current report. d) Reply to point G www.efsa.europa.eu/publications 21 EFSA Supporting publication 2017:EN-1355

laudable and if successful can be expected to greatly advance the validity and transparency of the findings of the hazard assessment. This point is addressed in Section 2.1.1 of the current report. D. The challenge with the protocol, as currently written, is that while there is an ambition to employ systematic methods and there has been an attempt to describe the methods which will be used, the methods fall short of being a sufficiently detailed, pre-planned protocol. This is because the proposed methodology too often describes when subjective processes will be employed rather than laying out transparent, reproducible processes which should be followed in a systematic assessment. E. There are particular challenges with the quality assessment of the evidence which functions as the first step in assessing the reliability of the evidence base. The second step, the risk of bias component, is close to the standard method in systematic reviews but the first quality assessment step is not. How this first step relates to the risk of bias assessment, why it is not redundant, how it is transparent and how it enhances rather than undermines the validity or scientific quality of the assessment is unclear. F. Similar concerns apply to the weight of evidence assessment, which is too difficult to follow to permit critical appraisal. Prespecification of methods is only transparent if the methods themselves are transparent, and these are not. The individual steps of the assessment, the fundamental questions being asked, the processes for answering those questions, and how the answers can be expected to be valid all needs clarifying. G. Overall, the protocol gives the impression of falling between two stools, whereby an attempt has been made to compromise between systematic and traditional narrative methods. The compromise is to attempt to do both. Unfortunately, taking methods which are accepted as being systematic and adding to them the narrative processes used in previous assessments seems unlikely to improve the validity of the assessment overall, diluting rather than concentrating the validity of the assessment. www.efsa.europa.eu/publications 22 EFSA Supporting publication 2017:EN-1355