g-twyst stakeholder workshop on planning stage issues ......4june"2015! huibdevriend!...
TRANSCRIPT
4 June 2015
Huib de Vriend Armin Spök
First Stakeholder Workshop Report Vienna, 16-‐17 December 2014
G-‐TwYST GMP Two Year Safety Testing
632165
2
“GMP Two Year Safety Testing” (G-‐TwYST) is a Collaborative Project of the Seventh Framework Programme of the European Community for Research, Technological Development and Demonstration Activities. Grant agreement no: 632165 Project duration: 21 April 2014 – 20 April 2018 Project website: www.g-‐twyst.eu
Acknowledgment and Disclaimer The authors of this document thank all project partners and participants in the G-‐TwYST Stakeholder Workshop of December 16-‐17, 2014 in Vienna for their valuable contributions to this draft report. This report must not be used as a resource for further research without prior permission of WP leader Armin Spök. In the whole document, the acronym “G-‐TwYST” has been used to refer to the project.
Content page List of abbreviations 5
1. Introduction to the report 6
2. Welcome and introduction 8 2.1. Introduction 8 2.2. Project partners 8 2.3. Objectives 9 2.4. Structure 9 2.5. Introduction to the study design 10 2.6. Interlinkages between G-‐TwYST and GRACE 11 2.7. Questions and answers 11
3. Stakeholder Involvement and Transparency in G-‐TwYST 13 3.1. Principles and procedures 13 3.2. Questions and answers 14
4. Production of plant material 15
5. Analysis of plant material, storage and diet preparation 16 5.1. General strategy 16 5.2. Analytes to be tested 16 5.3. Storage 17 5.4. Diet preparation 17 5.5. Questions and answers 18
6. The test facility 19 6.1. Accreditation 19 6.2. Experimental rooms and organizational set-‐up 20 6.3. Laboratory of Clinical and Experimental Biochemistry SMU 21 6.4. Haematological Laboratory and surgery 21 6.5. Questions and answers 22
7. Study design 23 7.1. General conditions 23 7.2. Original and alternative study plans 23 7.3. Questions and answers 25
8. Panel discussion day 1 28
9. Histopathology 30 9.1. Necropsy, tissue sampling and trimming of organs 30 9.2. Histology 30 9.3. Histopathology 30 9.4. Questions and answers 30
10. Biostatistics 32 10.1. Blocking 32 10.2. Randomization 32 10.3. Blinding 32 10.4. Replication and power analysis 33 10.5. Equivalence 34
4
10.6. Questions and answers 34
11. Statistics and prospective power analysis 36 11.1. Draft list of relevant endpoints 36 11.2. Draft methods for statistical analysis 37 11.3. Equivalence 38 11.4. Use of external reference data: a pragmatic approach 40 11.5. Questions and answers 42
12. Data storage and access 45 12.1. General principles 45 12.2. CADIMA characteristics 45 12.3. How to use CADIMA 46
13. Interlinkages with other research projects 47 13.1. G-‐TwYST and GRACE 47 13.2. G-‐TwYST and GMO90+ 47 13.3. Questions and answers 49
14. General discussion 50
Annex 1: Workshop agenda 53
Annex 2: List of participants 55
Annex 3: Written comments received from stakeholders 57 A3.1. Dr. Robin Mesnage1*, Dr. Michael Antoniou1, Pr. Gilles-‐Eric Séralini2 57 A3.2. Austrian Agency for Health and Food Safety (AGES) 61 A3.3. EFSA 63 A3.4. Eurogroup for Animals 65 A3.5. Bundesamt für Naturschutz, Germany 67 A3.6. ANSES, Risk Assessment Department, Food Biological Risk Assessment Unit, France 69 A3.7. Crop Life and Europabio 74
5
List of abbreviations
CA Competent Authorities CADIMA Central Access Database for Impact Assessment of Crop Genetic Improvement
Technologies CSO Civil Society Organisation DoW Description of Work EFSA European Food Safety Authority EU European Union GLP Good Laboratory Practice GM Genetically Modified GMO Genetically Modified Organism GRACE GMO Risk Assessment and Communication of Evidence GRAS Generally Recognized as Safe G-‐TwYST GMO Plant Two Year Safety Testing NOAEL No Observed Adverse Effect Level OECD Organisation for Economic Co-‐operation and Development SOP Standard Operating Procedure WP Work Package
6
1. Introduction to the report
This report This report summarises the presentations, comments and discussion in the course of the first G-‐TwYST Stakeholder Workshop held on 16-‐17 December 2014 in Vienna. The structure follows a chronological order and thereby mirrors the workshop flow. Thereby, some discussion topics came up more than one time and can therefore be found in several sections. The process to the workshop A safe-‐the-‐date announcement was sent by email to some 700 stakeholder contacts representing Competent Authorities, both at the EU and the national level of all 28 Member States, industry, farming and professional organisations, civil society organisations (CSOs) and scientists in September 2014 followed by a more detailed invitation with a preliminary agenda in October. Participation in the workshop was open to anyone representing one of the stakeholder categories mentioned above. Journalists that requested participation in the workshop were denied access because the project partners wanted to maintain a protected space in which everybody would feel free to express views and make suggestions without being hampered by the possibility of being quoted and/or framed in the media. Stakeholders who wanted to participate were asked to register through the G-‐TwYST website. Those who registered were given access to the draft study plans well in advance of the workshop. The stakeholder workshop 41 stakeholders from 14 Member States, USA and Norway participated in the workshop. Figure 1 indicates the participation of the different stakeholder groups. The workshop started with a general welcome and introduction by Pablo Steinberg, the project leader, an introduction to G-‐TwYST’s stakeholder engagement and communication strategy and an introduction to the workshop by the two co-‐leaders of the work package in charge of these topics. After a brief round of questions and answers following these talks, more detailed introductions were given on four specific parts of the study plan on the first day: 1. Production of plant material MON810 and NK603; 2. Analysis of plant material, storage and diet preparation; 3. The test facility; 4. Study design. After each introduction participants could ask for clarifications. During a session at the end of the day participants could discuss these elements of the study plans in more detail.
Figure 1: Participation in the G-‐TwYST stakeholder workshop on 16-‐17 December 2014.
7
Another 4 elements of the study plan were presented and discussed in the same manner on the second day: 1. Histopathology; 2. Statistics and prospective power analysis; 3. Data storage and access; 4. Interlinkages with other research projects. All introductions, questions and comments were audio recorded. Next steps Stakeholders were invited to file written comments on the workshop documents and on the workshop discussions. Eight sets of comments were provided by stakeholder organisations or individuals (see Annex 3) before the deadline of 16 January, 2015. Overall a total of 131 written comments were received. The draft workshop report was circulated to participants in May 2015 for feedback in order to make sure that the questions, comments, responses and discussions have been accurately depicted. The final version will also include the written stakeholder comments as well as the responses of the G-‐TwYST project team. Both the workshops slides and the final report are published at the G-‐TwYST website.
8
2. Welcome and introduction
2.1. Introduction After welcoming all participants project leader Pablo Steinberg (TiHo) provided basic information regarding the EC call for the project1. It is a Collaborative Project (small or medium-‐scale focused research project) with a budget of 2,999,890 € and a duration of 4 years, starting date: 21st of April 2014. The European Commission asked to address the following issues: • Execution of at least one rat feeding trial(s) with GM maize NK 603 (and additional GMOs where
scientifically justified) by taking into account EFSA recommendations. Participating institutions should strictly comply with all applicable international standards and norms concerning feeding trials in close collaboration with EFSA;
• Analysing, reporting and providing recommendations, in particular as to the scientific justification and added value of such long-‐term feeding trials with regard to GMO risk assessment.
2.2. Project partners The 8 project partners represent a governmental institution, 2 SMEs and 5 universities and research institutes (see Table 1 below). Table 1: G-‐TwYST participants Participant no Organisation Type Country Work
package 1 (Coordinator)
Tiermedizinische Hochschule Hannover (TiHo)
Uni Germany WP1 WP3 WP5
2 Centre de Recerca Agrigenòmica Consorci CSIC-‐IRTA-‐UAB (CRAG)
Res Spain WP2
3 Stichting Dienst Landbouwkundig Onderzoek (DLO)
Res Netherlands WP4
4 Julius Kühn-‐Institut (JKI) Gov Germany WP1 WP2 WP5 WP6
5 LIS Consult (LIS) SME Netherlands WP5 WP7
6 Roger Alison Ltd. (RA) SME UK WP3 7 Slovenska Zdravotnicka Univerzita v
Bratislave (SZU) Uni Slovakia WP3
8 Universitaet Klagenfurt (UNI-‐KLU) Uni Austria WP5 WP7
1 Call KBBE-‐2013-‐FEEDTRIALS: Two-‐year carcinogenicity rat feeding study with maize NK603
9
2.3. Objectives Steinberg then highlighted the main objectives of G-‐TwYST: • to elaborate a scientifically sound approach to evaluate the potential toxicity of genetically
modified plants in whole feed based on the results of extended feeding studies with the GM maize NK603 and MON810 performed according to current OECD Test Guidelines and EFSA recommendations2
• to define criteria to evaluate the scientific quality of long-‐term feeding studies • to define when long-‐term animal feeding trials are scientifically justified and which is their added
value in the frame of the GMO risk assessment process • to analyse the role/influence of broader societal issues including ethical aspects (normative
dimensions) in/on the on-‐going debate associated with animal feeding studies in GM food/feed risk assessment
• to make accessible the detailed scientific information including raw data via the project website, the open access database CADIMA, open access journal papers and stakeholder consultations
• to communicate the results of the project and their significance for the GMO risk assessment process to risk assessors, risk managers, a broad range of stakeholders and the general public.
2.4. Structure The project has been divided in 7 work packages. Apart from all prerequisites needed to perform the feeding trials -‐ from feed production and plant analyses to the actual trials and
Figure 2: Structure of G-‐TwYST
2 In the original Desdription of Work a 2-‐year carcinogenicity study on the GM maize MON810 had been foreseen. Due to the limited animal housing capacity and budget the 2-‐year feeding trial with MON810, which was heavily criticized in the past due to its inadequate experimental design, will be substituted by an additional 90-‐day feeding trial, in which the GM maize NK603 at an inclusion rate of 50%, as recommended by EFSA, will be tested.
WP1:%Coordina-on%and%management%
WP2:%Feed%produc-on%&%plant%analysis%
WP3:%Feeding%trials,%90%days%&%combined%1@2%year,%
incl.%histopathology%
WP4:%Biosta-s-cs%
WP5:%Evalua-on%of%scien-fic%value%and%norma-ve%issues%
Risk%assessors,%risk%managers%and%other%stakeholders%
WP6:%Central%Access%Database%(CADIMA)%
WP7:%Stakeholder%engagement,%communica-on%
&%dissemina-on%
Advisory%Board%
10
analysis of data -‐ the project has separate work packages dealing with data storage in the Central Access Database CADIMA, the evaluation of the scientific value of the approach and societal issues, and stakeholder engagement and communication activities. The project also has a work package for coordination and management and an Advisory Board. The scheme in Figure 2 presents how the activities in the work packages are related, the blue arrows representing the data flow and the green arrows the communication flow.
2.5. Introduction to the study design Steinberg also briefly presented the proposed approach for the 90-‐day study with GM maize NK603 and the combined chronic toxicity/carcinogenicity feeding (1-‐ and 2-‐ year) trial with GM maize NK603, as well as the 2-‐year carcinogenicity feeding trial with GM maize MON810. These proposals are presented in tables 2-‐4. Table 2: 90-‐day feeding trial with GM maize NK603 (original planning)
Table 3: Combined chronic toxicity/carcinogenicity feeding trial with GM maize NK603 (original planning)
11
Table 4: 2-‐Year carcinogenicity feeding trial with GM maize MON810 (original planning)
2.6. Interlinkages between G-‐TwYST and GRACE Steinberg ended his presentation with a brief explanation of the relationship between G-‐TwYST and GRACE, another EU-‐funded FP7 research project dealing with long-‐term safety assessment of GM maize. These projects are comparable in terms of the type of test guidelines (OECD) and recommendations (EFSA) taken into account when planning the feeding trials. They use the same laboratory for animal testing in Bratislava, they use the same rat strain and GM maize MON810 from the same source. The projects are complementary in terms of events and tests (see Table 5). Table 5: GM maize events and types of tests performed by GRACE and G-‐TwYST (original planning) Event Test GRACE G-‐TwYST MON8102 Subcronic toxicity
(90-‐days)
Chronic toxicity (1-‐year)
Carcinogenicity (2-‐year)
NK603 Subcronic toxicity (90-‐days)
Chronic toxicity (1-‐year)
Carcinogenicity (2-‐year)
Moreover, GRACE does systematic reviews on GMO impacts, and both projects feed their (raw) data into the CADIMA database (see chapter 11).
2.7. Questions and answers Several stakeholders asked about the herbicide treatment of the GM maize, pest infestations and other maize quality-‐related issues: • The project team confirmed that the same product/formulation of glyphosate (Roundup) has
always been applied; • Good Agricultural Practices principles have been applied. Roundup had to be applied only once;
12
• The material that is being shipped will also be checked for residues of other herbicides, such as atrazine;
• Currently no pest infestations have been reported and the visual impression was that there were no pest infestations, but this question will be addressed to the contractors at the production site;
• Treatment of the seeds with nicotinoids has not been reported. This question will also be addressed to the contractors at the production site;
• The moisture content will be below 15% (targeted at 14% for storage reasons).
13
3. Stakeholder Involvement and Transparency in G-‐TwYST
3.1. Principles and procedures Armin Spök, co-‐leader of work package 7, explained the objectives of and principles behind the stakeholder engagement approach chosen in G-‐TwYST. The project’s policy for Stakeholder Engagement, which started with this meeting in Vienna, is meant to inform and fine-‐tune the research process by asking stakeholders to comment on both the study design (at this stage of the project) and the interpretation of results after the feeding trials and analysis of data. Thus, the G-‐TwYST partners want to enhance the relevance of the project results from a broader societal perspective. This will be ensured by transmitting what is being done as well as why, how and by whom research is being done. Spök emphasized that this approach is different from usual approaches to stakeholder involvement, that only include a stakeholder input step at the final stage, where results, conclusions and recommendations are presented and discussed. G-‐TwYST does not only include stakeholder involvement before the actual start of the feeding trials, the project also enables for much more input by stakeholders than usual. After discussion with stakeholders and written comments, draft study plans and draft results, interpretations and conclusions will be revised. All comments will be evaluated regarding their relevance and impact on the study plans and presentation of final results, taking into account constraints in terms of time (a strict limit of 4 years set by the European Commission) and budget, and will be responded to. There will also be an open scientific discussion forum in the journal that will publish the G-‐TwYST results. This process is depicted in Figure 3.
Figure 3: G-‐TwYST Consultation Phases and Topics All steps and procedures will be traceable by means of reports that will be published on the G-‐TwYST public website. These reports will show how stakeholder inputs are shaping the research process and
14
the outcomes: drafts, comments, discussions, team responses, results and raw data will be accessible for stakeholders. 3.2. Questions and answers One of the stakeholders asked how stakeholders for this workshop had been defined and selected. Spök explained that a list of more than 700 stakeholders had been built up during the course of a number of projects. This list includes a wide range of stakeholders: Competent Authorities, Scientists, Industry, Civil Society Organisations, Farmers Associations, and Retailers. They all received an open invitation for this workshop.
15
4. Production of plant material
Ralf Wilhelm of the Julius Kühn-‐Institut (JKI) and on behalf of Maria Plá from the Centre de Recerca Agrigenòmica Consorci CSIC-‐IRTA-‐UAB (CRAG) introduced the procedures for production of the GM maize NK603 and MON810. As cultivation of NK603 maize is not allowed in the EU, maize varieties (harvests) with this event are imported from Canada and the United States. The project will be supplied with untreated NK603, Roundup-‐treated NK603 and a near-‐isogenic conventional counterpart. MON810 is cultivated in the Northeast of Spain and the project will be supplied with MON810, as well as a near-‐isogenic counterpart of MON810 and SY NEPAL as a conventional variety. In all cases Good Agricultural Practices principles have been followed, which means that the crops were planted and provided with fertilizers in accordance with local conditions and treated with herbicides or other chemicals for normal control of weeds and pests. The GM maize NK603 cultivated in the USA is the hybrid Prairie Brand 882RR2. The grower planted 32,000 seeds per acre on May 23, 2014, and applied five gallons per acre of 10-‐34-‐0 fertilizer at planting. Another 150 lbs. of N and 50 lbs. of K was applied at about the 4-‐leaf stage. Roundup was applied to one of the RR2 plots on June 14, 2014. The maize was harvested on November 4, 2014. 2.5 tons of NK603 treated with Roundup, 2.5 tons of untreated NK603 and 4 tons of the near-‐isogenic counterpart were transported to the storing place located in Germany in calendar week 50, 2014. The GM maize NK603 cultivated in Canada was the hybrid Pioneer 8906. The grower planted 84,721 seeds/ha. The non-‐GM seed was planted on May 13, while the 8906R seed was planted on May 14, 2014. 220 kg/ha of 20-‐0-‐30 + 2.4% Sulphur (granular) were pre-‐plant incorporated and 47 L/ha of 7-‐27-‐3 + 5% Zinc liquid starter fertilizer were applied in-‐furrow at planting. 116 kg/ha actual nitrogen were applied as 28% liquid nitrogen fertilizer on June 19, 2014. The grower applied herbicides twice: 3,5L/ha Primextra II Magnum (S-‐metolachlor + atrazine) on all blocks on May 20, 2014, and 2,5L/ha Roundup Transorb HC (540g/L) on the Roundup-‐treated maize field on June 20, 2014 only (= 1.35 kg/ha glyphosate; potassium salt). The crop was harvested in calendar week 48, and 4 t of NK603 with Roundup, 4 t of NK603 without Roundup and 7.8 t of the near-‐isogenic counterpart were transported to the storing place in calendar week 50/51, 2014.
16
5. Analysis of plant material, storage and diet preparation
Ralf Wilhelm also presented the proposed analyses of plants and diets. For the production of the diets, EU protocols require a call for tenders, which could only be further prepared after receiving the comments of the stakeholders on this specific topic. 5.1. General strategy The general strategy planned for G-‐TwYST is: • employ certified labs, • start with limited analyses of maize harvests assessing the quality (presence of mycotoxins, anti-‐
nutrients etcetera), • follow the OECD consensus document, • perform a full set of analyses (as in the GRACE project) on the maize commodity used for diet
preparation in the first batch of diets, • perform a selection of analyses for the follow-‐up batches (due to financial constraints).
5.2. Analytes to be tested It will depend on the stage in which the analytes will be tested (see Table 6). Table 6a: Analytes planned to be tested at different stages of harvest and diet preparation in the frame of G-‐TwYST
Analyte preTest harvest
Maize Diet (1st batch) Diet (further batches)
Proximates, fibres YES YES YES (YES) Fatty acids No YES YES (selected) Amino acids No YES YES (selected) Carbohydrates raffinose YES YES (selected) Minerals No YES YES (selected) Vitamins, carotenoids
No YES YES (selected)
Anti-‐nutrients Phytic acid YES YES YES (selected) Trypsin-‐Inhibitor YES YES YES (selected) Lectins No YES YES (selected) Secondary compounds Sterols No YES YES (selected) Phenolics No YES YES (selected) Furfural No YES YES (selected) Isoflavones No No YES (selected)
17
Table 6b: Analytes planned to be tested at different stages of harvest and diet preparation in the frame of G-‐TwYST
Analyte preTest harvest
Maize Diet (1st batch) Diet (further batches)
Contaminants Heavy metals YES YES YES (selected) Dioxins, PCBs YES YES YES (selected) PAHs No YES YES (selected) Nitrosamines No YES YES (selected) Nitrate, nitrite No YES YES (selected)
Pesticides broad screen; glyphosate and AMPA
YES YES (selected)
Mycotoxins DON YES YES YES (YES) Aflatoxins YES YES YES (YES) Fumonisine (B1-‐3) YES YES YES (YES) Ochratoxin YES YES YES (YES) HT-‐2 toxin YES YES YES (YES) T2-‐toxin YES YES YES (YES) Zearalenone YES YES YES (YES) Microbiology No No YES (selected) GMOs quant: NK603, cp4 epsps
YES YES YES (YES)
quant: MON810, cry1Ab
YES YES YES (YES)
Nos, S35 ... YES YES YES (selected) Event screen GMmaize YES YES (selected) plant DNA, CMV No YES YES (selected) 5.3. Storage Maize storage is performed in accordance with the HACCP concept. The commodity is stored in big bags at 7° C and 69% relative humidity in a rented facility. The maize is transported to the feed producer batch-‐wise when needed. 5.4. Diet preparation For the feed production the harvest of best quality will be selected. The diets will be produced batch-‐wise by a certified feed supplier and shipped to the test facility in Bratislava (SZU). Preparation and composition of the diets will be comparable to those in GRACE. The diets will contain no compounds from animals, no GMO except for the test material and a minimized amount of soy meal to avoid
18
oestrogenic effects. To avoid degradation of compounds, processing will be gentle; batches will be irradiated for disinfection. 5.5. Questions and answers One of the participants asked why the GMO material is not bought on the market instead of being cropped specifically for this study. Wilhelm replied that it is important to be sure that all the varieties to be used are cultivated under the same controlled conditions. Storage conditions One of the questions concerned the storage conditions. Is the temperature controlled? What is the shelf life of the diets? Wilhelm replied that at the central storage temperature is regulated >7 °C and conditions in summer to be monitored. The diet producer will be requested to apply regulated storage conditions and at SZU temperature is controlled. The shelf life is likely 6 month after irradiation. Omics Two remarks were made regarding the omics technologies to be applied: • There is no randomization in the maize fields. If you grow identical varieties in adjacent fields
you'll see differences in omics results. There is a risk of attributing differences to the GM plant, which may be actually caused by other factors. Wilhelm replied that G-‐TwYST will not only take into account data from the field material, but also compare data with other varieties from the (CADIMA) database, which was set up in the course of GRACE;
• One of the participants asked how the parameters for omics were selected. Wilhelm explained that in the first instance the omics-‐methods are used in an untargeted approach generating a high number of parameter (values) for each sampled variety/site. The omics data from many varieties (and sites) (including but not limited to samples provided through GRACE, G-‐TwYST) will further be analysed whether potential “unintended effects” can be identified against a background of (conventional) varietal effects.
• GMO contamination Another question concerned the possibility of GMO contamination in the diet for the control group. Will that be a commercially available diet and will that be certified GMO-‐free, thereby demanding a certificate? The diet will be tailored/adjusted to the purposes of G-‐TwYST. We will check whether it is GMO-‐free. An analysis will be made and the data will be available. One of the participants asked whether the chemical and microbial analysis will be done on the plants and the grains? Wilhelm answered that there is no data regarding the bacterial strains associated with plants in the field. A comparison with available commercial varieties will be done in the French GMO90+ project. Nutritional value A remark was made concerning the lack of historical data from long term-‐feeding studies. A high protein level in the diet may pose a problem when you are working with standard diets (see later chapter for further discussion on this topic). A final question concerned the timing of the analyses. These will be performed after import, and after storage.
19
6. The test facility
Dagmar Zeljenková and Jana Tulinská presented the facilities at the Slovak Medical University (SZU) in Bratislava, where the feeding trials will be performed. The animal rooms are situated in the Department of Toxicology, Faculty of Public Health, SZU. The complex with the experimental animal rooms was built in 1992 and consists of 4 parts: 1. A Specific Pathogen Free (SPF) unit; 2. A conventional unit and a surgery room; 3. A virology animal room, in which animals fed GM plant material can be kept; 4. An ichthyologic laboratory. In 1995, this animal housing facility obtained a „ Statement of Good Laboratory Praxis Compliance“ certified by the Slovak National Accreditation Service and an authorization for using laboratory animals in experiments certified by the State Veterinary and Food Administration of the Slovak Republic. Studies in the SPF unit are carried out following: • The revision of EU Directive EC 2010/63 on the Protection of Animals used for Experimental and
other Scientific Purposes which determines the minimum Requirements for the Regulation of Animal Experimentation by Member States of the EU;
• The European Convention for the Protection of Vertebrate Animals used for Experimental and other Scientific Purposes (Strasbourg, 18.03 1986);
• The implementation of the principle of the 3R’s (Replacement, Reduction, and Refinement) requires that all personnel involved in laboratory animal care and use have the proper education, training and experience when performing animal studies.
6.1. Accreditation Accreditation is in compliance with the Direction of Government of the Slovak Republic No. 377 from November 14, 2012. The animal experiments can be only carried out in an experimental facility, which is certified to perform animal experiments by the State Veterinary and Food Administration of the Slovak Republic. Accreditation records include: • A photocopy of the acquisition register; • References of the veterinary organization in charge; • Operating instructions of experimental facilities 1. Written delegation of responsibility for experimental animal keeping; 2. Approval of the Ethical Committee and Advisory Board; 3. Animal source records; 4. Regional Veterinary Administration Inspection (once a year); 5. `Project of experiment’3 -‐ a description of each individual experiment 6. Report on the number of animals used and the type of animal experiments performed (once a
year).
3 ‘Project of experiment’ is a definition used by the state veterinary administration.
20
6.2. Experimental rooms and organizational set-‐up The facility uses different animals. Mice, rats, guinea pigs, rabbits, poultry and pigs (Sus scrofa) for conventional breeding experiments, laboratory mice and rats for breeding experiments under SPF conditions and zebrafish (Danio rerio) in the ichthyologic laboratory. The maximal number of animals that can be housed is 1,600 rats, 500 mice and a lower number of guinea pigs, poultry, rabbits, turkeys and pigs. Maximally 800 rats can be handled in parallel in one part of the SPF unit.
Figure 4: Animal cages in the SPF unit.
Figure 5: Cryobanking at the Toxicology Unit.
21
The type of toxicity studies that can be performed in these facilities are: • Acute toxicity studies • Short-‐term toxicity studies (repeated dose) • Sub-‐chronic toxicity studies • Long-‐term toxicity studies • Chronic toxicity studies • Carcinogenicity studies • Chronic toxicity/carcinogenicity studies • Reproductive toxicity studies • Developmental (teratology) studies • Genotoxicity (mutagenicity) studies • Aquatic toxicity The documentation in the experimental animal rooms accepted by the Slovak National Accreditation Service includes: • Principles of Good Laboratory Practice; • Operating procedures carried out in experimental animal rooms -‐ Study plans; • Methodical Standard Operating Procedures; • All records (e.g. records on animal health status, records on food consumption, records on body
weight).
6.3. Laboratory of Clinical and Experimental Biochemistry SMU The Laboratory of Clinical and Experimental Biochemistry of the Slovak Medical University holds an accreditation certificate (M-‐013) from the Slovak National Accreditation Service, is subjected to the national quality control program for clinical biochemistry and is controlled by the quality assurance unit (QAU) of the Slovak Medical University. The Vitros 250 Chemistry System (Ortho-‐Clinical Diagnostics, No. 219037234, USA) performs clinical tests on serum, urine and cerebral spinal fluid specimens. Methodologies include colorimetric, and potentiometric tests using multi-‐layered Vitros Slides. Tests are performed in accordance with the Norm STN EN ISO/IEC 17025. Methods of clinical chemistry (ŠPP/LEKB/M001) are proceeded according to international norms of IFCC (International Federation of Clinical Chemistry and Laboratory Medicine), CLSI (Clinical and Laboratory Standards Institute) and NCCLS (National Committee on Clinical Laboratory Standards). The Laboratory takes part in the external quality control twice a year following a control cycle by INSTAND e.V. (Düsseldorf, Germany). The calibration is performed with certificated reference materials from the manufacturer of the apparatus (accreditation No.: 061/M-‐013). 6.4. Haematological Laboratory and surgery The haematological laboratory is part of the immunotoxicological laboratory. In this lab blood samples from the tail vein will be taken for haematological examination. EDTA will be used as anticoagulant. Blood samples will be stored under room temperature (17-‐25° C) maximally up to 4 hours until measurement. Haematology analysis will be performed in accordance with SOP ŠPP/IMU/M002 using the haematological analyser Sysmex K-‐4500 (SYSMEX TOA Medical Electronics Co. LTD, Japan, No. VČ F-‐1466).
22
A complete necropsy will be performed on all animals at study termination. The weight of organs will be recorded, and organs will be examined macroscopically for any deviations from normal. Tissues will be sampled and organs will be trimmed. Formalin-‐fixed and trimmed tissues will be shipped to the Department of Pathology at the University of Veterinary Medicine Hannover. These steps will be performed in accordance with the SOP ŠPP/TOX/005. 6.5. Questions and answers One of the participants asked about the authorization procedure by the ethical committee. Zeljenková explained that a full protocol with detailed numbers of animals has to be drafted first. Moreover, it is possible to ask a permit for the whole project. Another question concerned the organization of the trials: Does the Bratislava facility have GLP compliance? Zeljenková explained that the clinical chemistry analyses, which will be performed in Bratislava, do not require GLP, and the facility has an accreditation from the Slovak National Accreditation Commission. The clinical pathology analyses, which will be presented later, are performed according to GLP principles. Necropsy will be done in Bratislava. A participant recommended to clarify this in the protocol. Can trials be done in parallel? Zeljenková replied that it is not possible to do 3 studies at a time. Since all available personnel will be needed for necropsy the project will start with the 2-‐year study, and the other studies will follow later. The choice of the Wistar rat strain was commented by one of the participants, wondering whether there will be sufficient historical data available and whether this strain is sensitive to tumours. Roger Alison replied that Wistar rats develop mammary gland tumours. This strain of rats has typically been used for toxicity studies, and we can use historical data regarding toxicity studies from Harlan on 20,000 animals.
23
7. Study design
Project leader Pablo Steinberg presented the study design, general conditions, the original study plans and alternative study plans in more detail. 7.1. General conditions Male and female Wistar Rcc Han rats will be purchased from Harlan and will only be a few days apart in age. The animals will be acclimatised to the animal housing conditions 4-‐6 days prior to the start of the feeding trials. A detailed examination of all animals will be carried out on study day 1, prior to the start of the treatment. In order to perform the feeding trials, a randomised block design will be followed. The route of administration will be the oral route. Sentinels will be fed the standard rat diet Teklad Global Diet. The different diets will be coded and labelled by the supply company; feed containers and scoops will be colour-‐coded and animal house staff will be “blind” with respect to the identity of the diets. The dose groups will be unblinded for the histopathological evaluation of the tissues after necropsy and the weighing of organs. 7.2. Original and alternative study plans In accordance with the original planning, 5 groups of 32 animals each (16 males and 16 females) and a sentinel group of 12 animals (6 males and 6 females) will be used in the 90-‐day feeding trial with NK603. In each group there will be a control group fed a diet containing near-‐isogenic non-‐GM maize, a group fed untreated NK603 at an incorporation rate of 11%, a group fed NK603 at an incorporation rate of 33% NK603, a group fed NK603 treated with Roundup at an incorporation rate of 11% and a group fed NK603 treated with Roundup at an incorporation rate of 33% (see Table 7a). Table 7a: 90-‐day feeding trial with GM maize NK603 (original planning)
24
For methodological reasons one can argue that in toxicological research the level of the compound you are testing should be pushed to the limit, as recently recommended by EFSA in an explanatory statement on the applicability of the Guidance of the EFSA Scientific Committee on conducting repeated-‐dose 90-‐day oral toxicity study in rodents on whole food/feed for GMO risk assessment (EFSA J. 12: 3871 [2014]). Therefore, in the alternative proposal the maize content of the diets is increased to what is considered the amount of maize that can be fed without causing a nutritional imbalance. Therefore, in a second 90-‐day feeding trial diets containing 11, 33 or 50% NK603 will be tested (see Table 7b). Table 7b: 90-‐day feeding trial with GM maize NK603 (alternative)
The combined chronic toxicity/carcinogenicity feeding trial has 5 groups with 40 animals (20 males and 20 females) each for the chronic toxicity testing phase, 100 animals (50 females and 50 males) each for the carcinogenicity testing phase and 20 sentinels. There will be a control group fed a diet containing near-‐isogenic non-‐GM maize a group fed a diet containing 11% untreated NK603, a group fed a diet containing 33% untreated NK603, a group fed a diet containing 11% Roundup-‐treated NK603 and a group fed a diet containing 33% Roundup-‐treated NK603 (see Table 8a). The alternative combined chronic toxicity/carcinogenicity feeding trial would include 7 groups of 20 animals (10 females, 10 males) for the chronic toxicity phase, 100 animals (50 males, 50 females) for the carcinogenicity testing phase and 20 sentinels (see Table 8b). In the case of the alternative plans more animals will be needed for the feeding trials with NK603 maize, which can be realized within the limits of the animal housing capacity and budget by deleting the 2-‐year feeding trial with MON810 (Table 4), which was heavily criticized in the past due to its inadequate experimental design.
25
Table 8a: Combined chronic toxicity/carcinogenicity feeding trial with GM maize NK603 (original planning)
Table 8b. Combined chronic toxicity/carcinogenicity feeding trial with GM maize NK603 (alternative proposal)
7.3. Questions and answers This part of the workshop triggered considerable discussion on issues like the percentage of GM maize in the diet, the lack of historical data, the risk of loosing too many animals in the 2-‐year study and the parameters to be measured. Parameters One of the participants asked whether G-‐TwYST is planning to measure only some standard parameters. He emphasized the importance of including immunological parameters. The project might otherwise be criticised for not having taking this type of parameters into consideration.
26
Steinberg replied that this point will be taken into consideration and is also a matter of laboratory and budget capacity. Another participant recommended additional parameters that reflect immunotoxicity in organs/tissues such as spleen and bone marrow. Reducing the number of animals Animal Welfare organizations would like to reduce the number of animals as much as possible. Against this backdrop one of the participants wondered about the relevance of having a group of animals fed with a diet without Roundup; the maize is on the market in the treated form. Steinberg explained that the Roundup formulation could be problematic. The point is: if toxic effects were detected in rats fed NK603 treated with Roundup, the effects could be due to Roundup or to the GM plant. Therefore, both untreated and Roundup-‐treated NK603 needs to be tested. Steinberg reminded that this aspect was also part of the EU call. The 50% incorporation rate of NK603 in the diet “In carcinogenicity studies the highest dose should not trigger a toxicological effect”, one of the participants noted. “We have to prove that there is no difference. When applying equivalence (as regulation requires) you may be missing differences/effects”. On the other hand, a 50% diet is a risk in terms of nutritional balance. You will need a sufficient number of animals by the end of the 2 years. The project team replied that one consequence of raising the incorporation rate of NK603 in the diet to 50% is that we will have to increase the soy content, which may in turn cause oestrogenic effects. One of the participants asked whether there is data available about the survival of rats fed with high rates of maize during 2 years. What is the upper limit of the rate that we can add to the diet? Another participant replied that 33% is probably approaching the nutritional limit because it stems from a practice where maize is used as a filler. A point to take into consideration is that the EFSA recommendation on 50% diet was done for the 90-‐day study because it is good practice in toxicology to push the incorporation of the test material to the limit. EFSA looked at many studies, but did not find a clear rationale or justification for doses in diets. It is based on papers with caveats. In the context of a chronic toxicity study, 50% is risky because there is no historical data available. Another participant commented that deviating from diet formulation normally used in Harlan rats is a point of concern. It could cause effects in clinical chemistry and haematology parameters. This participant’s advice was to adhere as closely as possible to the usual diets. The proposal to include a 50% incorporation in diets also raised the question whether the G-‐TwYST data would still be comparable with the GMO90+ data. Lack of historical data One of the participants wondered about the implications of the lack of historical data for the quality of the findings of the study. Risk of loosing too many animals It was also commented that the issues of having only one control group and the lack of sufficient historical data are problematic, particularly in view of the fact that in the course of a 2-‐year feeding trial one could loose a lot of animals.
27
Two control groups After this discussion Steinberg asked whether a design with two doses of say 20% and 40% with 2 control groups –the usual approach in pharmaceutical feeding trials-‐ larger than 50 animals would be on the safer side. Yet, another participant argued that an additional control group is not a very good idea because one would widen the statistical variance and therefore weaken the statistical result. And how would we interpret the results if we see an effect at the high dose and no effect at the low dose? A participant answered: “If something is toxic, it is usually more toxic at a higher dose. You should take that into account”.
28
8. Panel discussion day 1
After the first round of presentations the speakers were invited to participate in a panel discussion, and participants were invited to raise issues for further discussion. General objectives and added value One of the participants asked for more clarity about the precise objectives of the study: What are the precise objectives of the study? What is then the most appropriate treatment structure? Is there a need to include the NK603 treated with Roundup group? We should then be looking for the minimum of treatments that are needed to meet those objectives, reducing the number of animals. From this, the number of replicate cages will follow. One of the participants said that the added value of the G-‐TwYST project is the search for unintended effects that have been overlooked by other experiments, but another participant argued, by referring to EFSA’s review of the 90-‐day studies (2008), that one would not expect to see effects after 2 years if you have seen no effects after 90 days. Steinberg replied that G-‐TwYST is meant to analyse whether the short-‐ and long-‐term feeding trials are an adequate test system to assess the risks arising from GM plants, and to answer this question, we need data from feeding trials of different duration using the same feed. This was understood by several participants as a statement to perform a study on each type of feed, to which they objected, both because of the use of large numbers of animals and the use of a study that has no added value in terms of a general conclusion concerning the need of long-‐term studies: “If you need to do another 90-‐day study because you have to use exactly the same feed as in the long term studies and if you want to be able to say something about the relationship between 90-‐day and long-‐term studies, how can you then extrapolate the results of this study to more general conclusions?” One of the participants argued that the added value will depend on whether you can introduce some criteria or measurements that are targeted to certain effects such immunotoxicity or hormonal effects. Steinberg replied that the study will take into account immunotoxicity and endocrine disruption, if the assays needed to analyse the above-‐mentioned forms of toxicity do not interfere with the study design of the subchronic toxicity test as well as with the combined chronic toxicity/carcinogenicity test. The study design One of the participants thought that it does not make sense to do a 90-‐day study if you are assessing the long-‐term effects because a 90-‐day study will not show effects of the treatment or compounds that appear at very low levels. Another participant argued that it would make sense to perform a 90-‐day study first, see what comes out of it and then design the combined 1/2-‐year studies, so you can focus on particular toxicity issues. Steinberg agreed that this would be normal way to go, but we cannot do it because we would need more than the 4 years we got from the European Commission, and there is no chance of getting an extension, not even if it is cost-‐neutral. Wilhelm warned not to draw conclusions before we run the experiments. Some say that 90-‐day studies are not sensitive enough; others say we need the results of 90-‐studies first to know what we should be looking for in the long-‐term studies. He asked participants to keep two things in mind: • We will have to start with the questions fully open;
29
• It is not possible to run a full evaluation due to time, budgetary and animal housing capacity constraints.
One of the participants suggested to integrate some methods that can replace animal studies, such as cell cultures. Schiemann agreed with this suggestion, but pointed to the lack of capacity in the G-‐TwYST project and to the fact that this is being looked at in the GRACE project with in vitro/omics studies. Another participant warned against non-‐adequate exposure studies, e.g. exposure to monolayers of cells, which do not reflect what happens in the human gastrointestinal tract. A participant noted: “If you include more endpoints, these should be validated.” GM/maize dosage in the diets and mortality rate Steinberg raised again the issue of the GM maize content of the diets. The previous discussion clearly indicated that 50% maize could cause a high mortality rate in the long-‐term studies. We should also keep the protein level below 14%. He proposed an adapted design with two doses and a maximum maize content of 40%. It was reiterated by participants that EFSA only considered the 50% diet for 90-‐day studies and 33% could be close to the nutritional limit. One might also consider different levels for the chronic toxicity study and the carcinogenicity study. Alison raised the question what to do if you end up with an insufficient number of animals in the high dose group after 2 years to reach sufficient statistical power. One of the participants referred to an OECD Test Guideline saying that 25% of the animals have to survive and there is a fair chance that one will manage it with Wistar rats. Effects of Roundup treatment One of the participants suggested to go for a Maximum Residue Level (MRL) based on agricultural practices. Another participant replied that a MRL would be so low that you would see no effects. He referred to a NK603 study from 2004: He had doubts about the necessity of repeating such a test. Yet another participant commented that you want to respond to the criticism that says that Roundup treatment could have an effect on the plant metabolism. It was also suggested that it would make sense to test the effects of Roundup separately over longer time, e.g. by adding it to the drinking water. It could help to establish whether an effect only comes from the use of the herbicide. Van der Voet replied that it would be a good idea to disentangle effects if there is a clearly identified effect, but in this case we do not know if there is an effect. Animal welfare In the study plans it says that Standard Operating Procedures (SOP) will be available. It is important to ensure proper registration of the cause of mortality. G-‐TwYST should also make clear the added value of the planned 90-‐day study (NK603) in the light of the two 90-‐day studies performed in the GRACE project (MON810).
30
9. Histopathology
Roger Alison introduced the histopathology analysis, which involves three steps: 1. necropsy, tissue sampling and trimming of organs, 2. histology, and 3. histopathology.
9.1. Necropsy, tissue sampling and trimming of organs In the first step, necropsy, tissue sampling and trimming of organs will be carried out at SZU. The samples will be formalin-‐fixed and trimmed tissues will be shipped to the Department of Pathology at the University of Veterinary Medicine Hannover. 9.2. Histology The Department of Pathology at the University of Veterinary Medicine Hannover will be in charge of the histology procedure, which includes: • Automatic embedding of the tissues • Production of paraffin blocks • Slide production by experienced technicians • Automatic haematoxylin-‐eosin staining • Automatic application of cover glasses Paraffin bocks will be stored until the end of the study and stained slides will be shipped to Roger Alison in the UK. 9.3. Histopathology The histopathological evaluation will be performed by Roger Alison Ltd. The slides received from Hannover will be stored at a fire-‐safe secure UK government approved storage place. Evaluation will be conducted using the market leading PathData pathology software. Necropsy data will be entered in PathData to allow correlation of necropsy findings with histopathology. The histopathological findings will be directly entered in PathData. All procedures will be audited by an in-‐house UK government approved Quality Assurance Auditor. The results will be reported in a GLP compliant “Final Pathology Report”. This will include a description of materials and methods, summary of results, discussion, conclusion (approx. 30 pages), a computer generated PathData Appendix (approx. 2500 pages) with incidence tables, individual animal data in tabular (AOFT) and text format, as well as specialist statistical software in PathData producing industry standardised “Peto” test age-‐adjusted analysis of lethal and non-‐lethal neoplasms. 9.4. Questions and answers One of the participants asked whether the histopathologist will give advice to Bratislava on how to perform the necropsies/prepare the organs. Alison told he will do so in a very detailed and completely standardized way. There will be some practice sessions before the study starts. Another participant emphasised the need for consistency in the description of what you see and asked whether the histopathologist will be present in Bratislava to check the necropsy? Alison confirmed: “Yes, we will go there for the final necropsy. The system we use, PathData, allows to enter necropsy data in English, French, German, and Slovak, predefined. In Bratislava all data will be entered in English/Slovak.” Alison was also asked about the use of tissue from animals that have
31
suffered. He replied that euthanasia of animals will be done before suffering, both on human grounds and animal data quality criteria.
32
10. Biostatistics Hilko van der Voet of Biometris, Wageningen UR, introduced six statistical aspects of the study design: blocking, randomization, blinding, replication and power analysis, and equivalence limits. 10.1. Blocking G-‐TwYST uses randomised block designs. Blocking is recommended by OECD Guidance Document 116 (OECD 2012) and two EFSA documents (EFSA 2011, 2013). When designing blocks, one has to take into account that not all animals can be handled at the same time and not all analyses can be performed at the same time. The principle applied in G-‐TwYST looks as follows: • Combine one cage (2 animals) of each dose
group into a block; • Perform all work block by block (as far as
practical): o Starting the experiment; o Feeding and cleaning; o Observations during the experiment: feed consumption, weighing, blood and urine sampling; o Necropsy, weighing of organs; o Analysis of samples.
Blocking minimizes the variation between dose groups within blocks. The statistical analysis can be based on a model that corrects for accidental differences between blocks. 10.2. Randomization Randomisation is important to prevent confounding of treatment effects with other sources of variation. For example, an effect on the cages in the leftmost position of each row is expected to be the same for all treatments.
G-‐TwYST will work with the random assignment of: • rats to cages, • dose groups to codes 1-‐5, • coded dose groups to cages in each block (see Table 9).
10.3. Blinding The use of nonblinded outcome assessors of subjective outcomes may cause considerable observer bias in animal model experiments: nonblinded assessors exaggerate odds ratios by approximately 59% in 10 animal model experiments including 2,450 animals (Bello et al. 2014). Therefore, codes will be randomly assigned to feed groups by the feed supplier4. Numbers and colours will be used for coded groups, and the code will be only given to two employees of JKI. The dose groups will be unblinded after the analyses have been performed. An exception will be made for histopathology. The histopathologist needs to know the identity of the codes because the within-‐group ‘normal variation’ has to be assessed. Therefore codes will be 4 At the time of the workshop the tender for feed production was not published and the feed supplier was therefore not decided.
33
unblinded for the histopathologist after necropsy and the weighing of organs.
Table 9: Example of randomization of the position of dose groups in the racks.
10.4. Replication and power analysis Replication is the basis to estimate variation and therefore allows statistical testing. The more replication, the higher the precision. According to the COMMISSION IMPLEMENTING REGULATION (EU) No 503/2013: “A power analysis to estimate a sample size capable of detecting a pre-‐specified biologically relevant effect size with a specified power and significance level should be used.“ Also EFSA recommends the use of a power analysis: “The use of a power analysis to estimate a sample size capable of detecting a pre-‐specified biologically relevant effect size with a specified power and significance level should be done to determine an appropriate sample size.” (EFSA 2011). “The OECD Test Guidelines indicate the appropriate sample sizes for each group. In the carcinogenicity study, the sample size is usually at least 50 animals of each sex at each dose level. This group size reflects a trade-‐off between the statistical power of the design and economic practicalities of the design. In practice, the carcinogenicity study has low power in the sense that treatment effects that might be considered biologically important cannot be detected routinely as statistically significant.” (OECD Guideline Document 116, 2012).
34
A power analysis is in principle easy, at least for simple tests. It includes four quantities: • A significance level 𝛼 • A power=1−𝛽 • An effect size • A sample size N and provides three options: • N = f(𝛼, effect size, power) • Power = f(𝛼, effect size, N) • Effect size = f (𝛼, power, N) An important question to consider in this context is: What is a relevant effect size? 10.5. Equivalence Relevant effect sizes can be set as equivalence limits for the difference between GM and control groups. By applying equivalence limits you set upper and lower values for each parameter (see Figure 6). Both the EFSA Food/Feed Guidance (2011) and the European Commission (2013) prescribe both the difference testing and equivalence testing. Equivalence limits can be set by study-‐internal data, external data, expert knowledge or combinations of those. In his presentation on statistical and prospective power analysis (chapter 10.3) van der Voet elaborated further on the issue of equivalence.
Figure 6: A theoretical example of equivalence limits with lower and upper limits 10.6. Questions and answers Blocking One of the participants noticed that blocking is a formalisation of standards. Another participant added that you should only deviate from the blocking structure if it is absolutely essential.
35
Unblinding during necropsy One of the issues that triggered some comments was blinding during the necropsy process. Alison told it is better to unblind during the necropsy process, but this has not been decided yet. Alison referred to a 2004 publication on best practices, and explained that a blind study is much slower (takes 3 times as long) and you have a higher chance of missing effects. In the G-‐TwYST project the problem of non-‐blinding is considered by having a 2nd (peer) pathologist that looks at every neoplasm and at about 10% of the animals. Equivalence The concept of equivalence was criticised by one of the partners: “In safety tests we want to prove that there is no difference and when applying equivalence you may be missing effects”. In EFSA the discussion on equivalence is presently going on. Number of animals per group One of the participants commented that the first goal in carcinogenicity studies is to detect tumours. For this purpose we have the Peto test. You have to focus on a number of parameters. You will have to end up with sufficient animals per group at the end of the carcinogenicity study, so that it may be necessary to increase the number of animals. Less groups with more animals per group will increase the statistical power. It would be better to have 7 groups of 50 or 6 groups of 60 animals. A final comment concerned the difference in weight of the animals. Two animals with very different weights in the same cage may have an impact on the parameters to be measured in the two animals. To counter this potential problem, animals will be assigned to cages after sorting on body weight.
36
11. Statistics and prospective power analysis On the second day of the workshop van der Voet elaborated further on the prospective power analysis. He repeated that such a power analysis is required by the EFSA 90-‐day study guidance document (2011) and the European Commission (2013) and that both also require difference and equivalence testing. The G-‐TwYST project follows the OECD Guidance Document 116 (2012) for sample sizes and power analyses and will be used to evaluate the power of difference and equivalence tests for relevant endpoints. He noted that there may be a discrepancy between the desired power and the expected power for some endpoints, which should be known before the experiment.
The power analysis (to evaluate the power) requires prior specification of: 1. A List of relevant endpoints; 2. Statistical methods to be used; 3. A significance level (𝛼=0.05 is proposed); 4. A sample size (cf. OECD); 5. A relevant effect = equivalence limit. Van der Voet suggested further discussion of points 1, 2 and 5. 11.1. Draft list of relevant endpoints Van der Voet mentioned the following preliminary list of endpoints and noted that a finalisation of this list should be made before starting the experiments: 1. Survival 2. Body weight (time series) 3. Feed consumption (time series) 4. Haematology (month 3, 6, 12, end of study)
• erythrocyte count (RBC) • haematocrit (HT) • haemoglobin (Hb) • leukocyte count (WBC) • differential leukocyte count • platelet count (PLT) • mean corpuscular volume (MCV) • mean corpuscular haemoglobin (MCH) • mean corpuscular haemoglobin concentration (MCHC)
5. Clinical biochemistry (month 3, 6, 12, end of study) • total protein • albumin • aspartate aminotransferase • alanine aminotransferase • alkaline phosphatase • creatinine • glucose • urea • total cholesterol • gamma-‐glutamyl transpeptidase • Na, K, Cl, Ca, P
37
6. Urinalysis (month 3, 6, 12, end of study) • volume • osmolality • pH • total protein • glucose • occult blood • ketone • urobilinogen
7. Organ weights (at necropsy) • kidneys • spleen • liver • adrenals • lungs • heart • thymus • brain • testes • epididymides • uterus • ovaries
8. Histopathology (at necropsy) Due to the specific necessities of the procedures the histopathology data cannot be analysed according to standard statistical methods. An interpretation will be given by the histopathologist.
Van der Voet emphasized that we know in advance that a power analysis will not be possible for all these endpoints due to lack of available data. 11.2. Draft methods for statistical analysis Van der Voet mentioned the following methods for statistical analysis: 1. The Cox Proportional hazards model will be applied for survival data; 2. Linear models with repeated measurements will be applied for body weight and feed
consumption; 3. For haematology, clinical biochemistry, urinalysis and organ weight data: Linear models per time
point. Finalised proposals will be described in the statistical analysis plan. Data will be transformed (e.g. log) where appropriate. Outliers will be identified by graphical means. Sensitivity of including/excluding outliers will be investigated, and results will be reported as effect sizes (difference between GMO and control groups) scaled to a relevant effect size and 95% confidence intervals. OECD Draft Guidance Document N° 116 on “The design and conduct of chronic toxicity and carcinogenicity studies” supporting the OECD Test Guidelines 451, 452 and 4535 describes traditional methods for statistical analysis (see selection of methods in the OECD flowchart in Figure 7).
5 http://www.oecd.org/chemicalsafety/testing/44076587.pdf
38
However, the OECD also warns against automatically following the flowchart. G-‐TwYST will adjust the OECD methods to more modern statistical approaches. Comparisons between traditional and modern approaches will be made.
Figure 7: A statistical decision tree, summarising common statistical procedures for analysis of data in long-‐term toxicology studies (OECD, 2010). Time series will be analysed as a whole. We will check for treatment by time interaction. If found (or assumed a priori), then an analysis per point in time will be performed. Data on males and females will be analysed together, unless there is a prior indication that data should be analysed separately, or there is an identified treatment by sex interaction (EFSA 90-‐day Guidance Document 2011)6 and data of the low and high dose groups will be analysed simultaneously.
11.3. Equivalence It is assumed that a GMO is equivalent to a control group if the difference between the groups is not biologically relevant. Significance in a difference test (𝐻0 ∶ Δ = 0) can show that a difference (𝐻1 ∶ Δ = 0) is not directly related to biological relevance. If we introduce biological relevance quantified in equivalence limit L, then significance in an equivalence test (𝐻0 ∶ Δ ≥ 𝐿) can show equivalence (𝐻1 ∶ Δ < 𝐿). The results of a difference test and an equivalence test can be jointly presented in a graph (Figure 8).
Equivalence limits can represent safety limits (ideal, but often difficult to obtain) or observed variation of references (more restricted, “history of safe use”). 6 http://www.efsa.europa.eu/en/efsajournal/pub/2438.htm
39
Options to set equivalence limits L are: • Expert knowledge, as written in the G-‐TwYST grant application; • Use of study-‐internal data, which was applied in the GRACE project; • Use of external reference data: this is ideal and in principle feasible; • Use of external reference and study-‐internal data. This is a pragmatic alternative that will be
used for the power analysis in the G-‐TwYST project.
Figure 8: The results of a hypothetical difference (red line) and equivalence test (blue lines) with (i) equivalence, (ii) equivalence more likely than not, (iii) non-‐equivalence more likely than not, (iv) non-‐equivalence.
Van der Voet illustrated how equivalence testing was applied in the GRACE project. He quoted the EFSA 90-‐day study Guidance Document, which presents an example using the standardized effect size (SES) approach: “An alternative approach described here, is to base sample size on a pre-‐specified effect size measured in SD units. This is known as the “standardised effect size” (SES). It is the difference between treatment groups divided by the standard deviation (SD) among experimental units, and can be regarded as a signal/noise ratio. Power analysis can then be used to estimate the required sample size needed to be able to detect a specified effect size in these units. If experience from previous toxicity tests shows that an effect size of, say, one SD or less is of little toxicological relevance then this can be used to determine sample size in new situations.“7 The assumption is that 1 SD unit represents an effect of toxicological relevance. Confidence intervals that do not touch the line at 0 represent a statistically significant difference. In this case that is the case for liver, left adrenal gland, pancreas, GLU, TP, CHOL, Cl, K, Na, P, TRG, lymphocytes, and eosinophils. If we look at point estimates outside ±1 that represent non-‐equivalence more likely than not, we should add creatinine to this list. However, internal and external reference data show more spread than 1 SD.
7 http://www.efsa.europa.eu/de/efsajournal/doc/2438.pdf
40
Zeljenková et al. (2014)8 presented graphs of SES for the GRACE 90-‐day study (Figure 9).
Figure 9: Standard Effect Sizes in the GRACE 90-‐day study (Zeljenková et al., 2014). 11.4. Use of external reference data: a pragmatic approach Although we have access to background data on variations in endpoint values for Wistar Han Rcc rats fed a standard lab chow from Harlan Inc.9 (see Table 10 for examples), the statistics are available, but not the primary data, and between-‐ and within-‐study are confounded or only the between-‐study is
8 http://www.grace-‐fp7.eu/sites/default/files/GRACE-‐FeedingTrials_AB_ArchToxicol_2014.pdf 9 http://webapps.harlan.com/wistarhannover
41
characterized. Table 10: Two examples of Harlan data
Body weight [g, %]
Haemoglobin (HB) [mmol/L]
For G-‐TwYST we will therefore apply a pragmatic approach: • Harlan historical data will be used to characterise between-‐study or total variation; • GRACE data will be used to characterise within-‐study variation. Ideally, this should be replaced by
other external data (such as the Harlan data). Between-‐study variation will be reconstructed from total and within-‐study variation;
• The difference between GMO and control in the current study will be compared to the typically observed variation between references;
• Equivalence limits will be set corresponding to a population equivalence criterion. The expected mean square comparing GMO to control divided by the expected mean square comparing two references should not exceed a threshold (based on a 95% inclusion).
The method is still under investigation, so the results of an exercise shown at the workshop are provisional. How this turns out, can be illustrated if we apply this approach to the GRACE data to estimate the difference between the animals fed the GMO 33% diet and the control diet. If real data would be like this, one could conclude that (see Figure 10): • All endpoints are equivalent or equivalent more likely than not; • Power of design seems sufficient for these endpoints.
42
Figure 10: GRACE study-‐corrected PE equivalence test, male. 11.5. Questions and answers Van der Voet suggested a number of issues for further discussion: 1. For a prospective power analysis the list of endpoints should be known; 2. Difference and equivalence testing can be effectively performed using Scaled Effect Sizes; 3. SES can be based on traditional or modern statistical analysis methods; 4. However, the example set by EFSA (SD=1) seems not realistic; 5. External reference data are ideal to define equivalence limits based on observed variation; 6. Unfortunately, external reference data are only available as summary statistics with insufficient
detail; 7. Pragmatically, study-‐internal estimates of within-‐study variation can be used as well but a larger
within-‐study variation then leads to wider equivalence limits, so that this is not ideal. Parameters One of the comments concerned the parameters. Although they are in line with the guidelines, the last measurement at the end of the study for haematology, organ weight and biochemical data is unusual because we have to expect geriatric changes. Therefore, for month 24 the haematology,
43
clinical chemistry as well as urinalysis and organ weight should be excluded. These are only to be determined after one year as the latest point in time. It is stated correctly in the Study Plan. It was also noted that in the GRACE study the parameter “pancreas weight” was included. Steinberg replied that the pancreas is hard to analyse because it is a diffuse organ and is intermingled with fat tissue. For this reason, it was decided not to include pancreas weight in the G-‐TwYST study and is not listed in the OECD Test Guidelines on subchronic testing and combined chronic toxicity/carcinogenicity testing. One of the participants commented that it is important to argue very precisely why you exclude certain data from G-‐TwYST (based on the GRACE experience). Use of historical data (Harlan diets) Several comments concerned the use of Harlan data as a reference: • This would raise questions concerning the comparability, as these have been collected in the
context of pharmaceutical testing. It is therefore important to make sure that the historical data are transparent, i.e. where the data come from. Van der Voet replied that you want the full range of variation where you are sure there is no concern, if you want to translate biological relevance into equivalence limits;
• Another participant warned to be careful in using data from studies with a duration of 104 weeks;
• It was argued that one has to know the maize incorporation rate in the Harlan diets as well as the experimental conditions of the feeding trials, i.e. they should be comparable;
• The feed in the control group should not contain GM material and it is not clear whether this was required in the studies which originated the Harlan data;
• You cannot use historical Harlan data because of the huge range in tumour incidents (0 -‐ 70%) and the unreliability of the studies.
Different approaches in power analysis for 90-‐day, 1-‐year and 2-‐year studies In response to a question concerning different approaches to the power analyses per study, van der Voet explained that this still has to be discussed and worked out. Standardised effect sizes Industry statisticians and toxicologists remain sceptical regarding the value of performing statistical analysis based on standardized effect sizes. There are several reasons for this. Firstly, there is no relationship between SES and toxicological relevance. Secondly, SES does not provide a basis for comparison with historical ranges. Thirdly, all endpoints arising from the same experimental design will appear equally sensitive regardless of the differences in underlying levels of variability from one lab to another. Industry toxicologists did provide EFSA with a list of proposed effect sizes for 90-‐day studies based on an extensive literature search in 2012. This list might serve as a starting point for the choice of effect sizes to be used in the G-‐TwYST power calculations. Another participant commented that EFSA is aware of those data, but they are not adequate for this kind of study because the data are derived from MRD or MTD studies, single dose or short-‐term administration. So that the provided effect sizes are clearly toxicological effects and the effect size is really large. Here we are talking about biological effects, i.e. treatment-‐related findings that do not necessarily represent an adverse (toxic) effect. Measuring body weight
44
G-‐TwYST is proposing to measure body weight at certain time points. You may see differences in body weight gains during the course of the study, but final body weight is particularly relevant. The sex of animals One of the participants wondered whether it is not dangerous to mix the data of male and female rats. You may miss differences in sexual organs as a result of hormonal effects. Van der Voet replied that G-‐TwYST will go through the list of endpoints one-‐by-‐one and have statements from the toxicologists as to where the sex of the animal will be relevant and where not. Apply safety factors or ‘margin of exposure’ One of the participants suggested that the G-‐TwYST team does not necessarily have to define risk factors or limits, but could apply the concept of ‘margins of exposure’, leaving it up to the risk managers. Van der Voet replied that then we need estimates of exposure, which may not be available for not yet authorised GMOs. Overall F-‐test and/or specific comparisons For those studies that will have more than 2 entries the protocol that has been issued does not explain what specific comparisons will be made. You may decide to do an F-‐test, testing for differences amongst all entries, and then go on to make specific comparisons if that F-‐test is significant. Alternatively you may decide to do only those specific comparisons without doing the overall F-‐test. Van der Voet answered that the intention is to further discuss this aspect with the project partners in order to come up with very clear comparisons to be tested. The meaning of EFSA’s example (yellow marking) in practice As for the power analysis, EFSA just provided an example. This should be put in practice in the context of a specific background or available data.
45
12. Data storage and access 12.1. General principles Christian Kohl from JKI introduced the CADIMA database. Following the project’s transparency obligations, all data are made publically available. The portal facilitating the open-‐access publication has to satisfy specific demands on data storage and access mode. Data must be protected from manipulation by third parties and the system has to allow continuous updating. Depending on the responsibilities of the participants in the project different user rights must be allocated. Access mode has to be determined by the right holder and will include protected access to temporarily confidential documents (via password) and open access publication following e.g. creative common licenses. The date of publication has to be decided by the right holder. 12.2. CADIMA characteristics CADIMA (Central Access Database for the Impact Assessment of crop genetic improvement technologies) is an internet portal for the dissemination of information and data from on-‐going and future projects in biosafety research. CADIMA is designed to store and provide access to data collected in the GRACE project, which elaborates and implements systematic, transparent and inclusive reviews of existing evidence of potential health, environmental, and socio-‐economic impacts (risks and benefits) of GM plants (GMPs) or food and feed derived from GMPs, and considers the design, execution and interpretation of results of animal feeding trials as well as in vitro studies for assessing the safety of GM food and feed.
Figure 11: The CADIMA database requires registration The database is subdivided into three topic areas (services): • Area 1: Central access point for relevant information on the impacts of GM plants, i.e. a list of
links to guidance documents, consensus documents etc.; • Area 2: Evidence synthesis database (related to the GRACE project); • Area 3: Animal feeding trials and alternative in vitro approaches.
46
CADIMA will be permanently established and maintained at JKI, also during and after the time frame of G-‐TwYST. Apart from GRACE, toxicological data will be provided by G-‐TwYST and GMO90+ (France). Different user rights will be assigned to each project. 12.3. How to use CADIMA If a user wants to conduct an evidence synthesis or wants to upload data to CADIMA an account has to be created at www.cadima.info (see Figure 9). After registration you will be contacted by the CADIMA administrator to discuss which of the services you want to use. Depending on the allocated user right it will be possible to upload now data in the animal feeding trials section. You can add a new study by giving a brief description of your study, the project you are working for, the study type or the animal model, the crop, the trait and the year in which the study was performed. You can also specify which license you want to use to make your data available at a later stage. You can also choose the option ‘hidden’, which can be useful at a very preliminary stage when you want the data to be visible for no one else but yourself. You can now start uploading files.
Figure 12: Example of an Excel file with raw data from the GRACE project If you are interested in the data provided by others you can go to the Services page, where you will find a list of all the projects that already provided accessible data (see Figure 9 for an example from GRACE).
47
13. Interlinkages with other research projects In his talk about interlinkages between G-‐TwYST, GRACE and GMO90+ Joachim Schiemann reported that the management teams of G-‐TwYST, GRACE and GMO90+ decided to have a close linkage between the projects because this will increase the comparability of results. Moreover, a close interlinkage will expand the data basis for political decisions and allows joining forces and sharing costs. It will mitigate the occurrence of force majeure, facilitate the exchange of data and material or tissues, and facilitate the exchange of knowledge and experiences. 13.1. G-‐TwYST and GRACE The major goal of GRACE (GMO Risk Assessment and Communication of Evidence) is to increase the transparency and traceability of information dealing with potential risks and benefits associated with the deliberate release of genetically modified plants (GMPs) and their products by: • Providing a systematic and structured quality assessment of existing evidence following a
harmonized framework; • Identifying knowledge gaps regarding impacts of GMPs; • Testing designs and identifying scientific values and limits of rat feeding trials and in vitro studies
with whole food/feed; • Providing an open-‐access database and a central access point of reviewed studies and further
data relevant to GMP impact assessment. GRACE and G-‐TwYST are complementary (see Figure 10). GRACE focuses on MON810 maize only and includes: • two 90-‐day feeding trials for subchronic toxicity; • a 90-‐day longitudinal metabolomics study; • a 1-‐year feeding trial for chronic toxicity; • in vitro assays with primary cell lines; • “omics”. G-‐TwYST complements GRACE with: • Feeding trials with NK603 maize for subchronic toxicity (90-‐day); • Feeding trials with NK603 maize for combined chronic toxicity/carcinogenicity (1-‐ and 2-‐year); • A 2-‐year feeding trial with MON810 maize for carcinogenicity (original planning). 13.2. G-‐TwYST and GMO90+ The major goal of the French project GMO90+ (funded by the Ministry of Ecology and Sustainable Development in France) is to investigate whether feeding rats with genetically modified MON810 or NK603 maize may lead to the identification of biomarkers of effects by: • Performing repeated-‐dose oral toxicity studies with a duration of 3 to 6 months accompanied by:
o blood and urine sampling (T0-‐T90-‐T120-‐T180) to determine biochemical parameters and metabolomic profiles
o trancriptome and micro RNA analyses of gut and liver tissue (T90-‐T180) • Identification of biomarkers by high-‐throughput (omics) techniques in conjunction with
pathophysiological analyses mainly centred on the gut, liver, kidney and reproductive apparatus. G-‐TwYST and GMO90+ are sharing costs by planting, harvest and shipment of NK603 (NK603 was grown at two different locations in Northern America in order to mitigate force majeure) and analysis of the maize material. The projects will also use the same rat strain and the same maize material
48
(NK603). Blood, urine, kidney and liver samples will be exchanged and GMO90+ will perform omics and analytical analyses on blood/urine samples and on liver/kidney extracts from G-‐TwYST rats being fed with GM corn for 1 and 2 years. Moreover, G-‐TwYST’s project coordinator Pablo Steinberg is a member of the scientific advisory board of GMO90+.
Figure 13: Comparability and complementarity of G-‐TwYST, GRACE and GMO90+
Figure 14: G-‐TwYST, GRACE and GMO90+, sharing knowledge, material costs and data Results from the projects will be included in the open access database CADIMA set up by GRACE (see Figure 14). Common dissemination will support synergistic effects between the research projects. The projects have an agreement with the journal Archives of Toxicology, which will facilitate a transparent and tailored scientific discussion on the added value of animal feeding trials as well as in vitro and in silico
GRACE& G'TwYST& GMO90+&
Comparability&
Complementarity&
!!!MON810:&
• Subchronic&(TG408)&
• Chronic&(TG452)&
NK603:&• Subchronic&(TG408)&
• Combined&chronic/
carcinogenic&&
(TG453)&
NK603,&MON810:&&
90&day&&&180&day&
&
AddiQonal&variables&
Toxicity&studies&with&whole&food&feed&following&OECD&Guidelines&
and&EFSA&recommendaQons:&
• Same&laboratory&
• Same&rat&strain&
49
approaches with whole food/feed for the risk assessment of genetically modified plants.10 13.3. Questions and answers Publication strategy G-‐TwYST’s publication strategy was an issue to a number of participants. One participant commented that Archives of Toxicology, the journal chosen for publication of the feeding trial results and a platform for scientific debate, is not the right platform because it is too closely linked to the researchers in the project and because of close links to the tobacco industry. It was therefore suggested to choose a journal with a higher reputation in terms of integrity. Schiemann replied that it will depend on the review process whether Archives of Toxicology will publish results and confirmed that there is an agreement on providing a platform for further discussion. One of the reasons for choosing this particular journal is that it has a high impact factor and is a recognized toxicological journal in the scientific community. He referred to the transparency policy that is also being applied in GRACE: both the comments and the reply from GRACE have been published on the GRACE website. In addition, the chief editor will respond to the issue of integrity, which directly concerns the journal. GRACE prefers an open-‐minded scientific debate on scientific issues, and Schiemann invited all participants to make use of the opportunity to comment on the arguments provided by Testbiotech and GRACE. Referring to some recent examples, another participant suggested to ask the editor to disclose the names of the reviewers. Schiemann replied that the names of the reviewers are not disclosed in the case of Archives of Toxicology and it is up to the chief editor to decide on this topic. Bioinformatics and omics One of the participants agreed that the bioinformatics and omics approach is interesting from a research perspective, but doubted whether we know enough regarding normal range of values for the individual parameters in different organs and biological fluids of the rat. Schiemann confirmed that this is a research topic. The EU call specifically asked for this aspect. Once these data are available, they will be discussed by the GRACE and G-‐TwYST partners. The draft conclusions and recommendations from the GRACE project will be provided to the stakeholders and will be discussed in two back-‐to-‐back stakeholder meetings at the beginning of October 2015. After this discussion with the stakeholders, the GRACE partners will draw final conclusions, which will be presented in the frame of a final GRACE Conference at the beginning of November 2015. The partners are aware of the limitations of this approach and the sensitivity of the different omics technologies.
10 http://www.ncbi.nlm.nih.gov/pmc/articles/PMC4247473/
50
14. General discussion Irradiation of the diets Participants were offered the opportunity to write comments or questions on cards and put them on a board. The only card used concerned irradiation of the diets. Is there any information that irradiation may change the composition of the diets, especially the protein content? Will this be checked? Wilhelm replied that if you use heat there will be a change. Irradiation has been performed in GRACE, and the corresponding analyses have been conducted. In G-‐TwYST we will check whether there is a change, but it should not be the case with the amount of irradiation that will be applied. One of the participants referred to a research study in a Vienna hospital, where they found out that L-‐amino acids were converted to the D-‐isomers, which in turn could lead to mutagenicity. One of the partners and Steinberg added that Hammond et al. as well as further studies on irradiated food in Europe showed no effects due to irradiation. Different varieties of NK603 One of the participants noticed that two different varieties of NK603 are grown in the US and Canada. As you may see effects that may be due to differences between the varieties, he suggested the analysis of these varieties in parallel. Steinberg replied that this would be interesting indeed, but given the budgetary and time constraints he did not see a possibility of doing such studies in parallel. G-‐TwYST will use the harvest that shows the highest quality (i.e. the lowest amounts of chemical contaminants and residues as well as the lowest degree of microbial contamination). Compounds that protect against cancer One of the explanations given by Seralini et al. for the enhanced tumour incidence in NK603-‐fed rats’ tumours was a decrease in the level of phenolic acid, a compound in maize that has an anticarcinogenic effect in rats. Alison replied to this comment by explaining that first you perform the carcinogenicity study as a general screen. If you find an increase in the tumour incidence, then you may use an animal model of the relevant human disease to determine the cause of the increased tumour rate. Moreover, in essence it does not matter which compound causes an increase in tumour incidence, a carcinogenic compound or the lack of a protective component. Glyphosate dosage You will have to use the material that has recently been harvested in USA and Canada, but that material might not be representative for the material normally being harvested. For instance, one has to take into account that spraying(s) are performed to treat glyphosate-‐resistant weeds, a practice that increasingly is being applied in USA. Furthermore, different formulations might also have an impact. In addition, there are reports in the scientific literature saying that there might be changes in the plant metabolism as a result of glyphosate treatment related to the event, and the level of dosage could be relevant. If you find out that the level of residues in the two harvests is different, would you then choose the one with the lower residue levels? Schiemann replied that NK603 has been cultivated according to Good Agricultural Practice, which is dictated by local environments, so Roundup has been applied according to the local experience. In Canada no glyphosate-‐resistant weeds have been observed, so that one treatment with Roundup, quite early in the year, was sufficient. One of the participants reiterated that from a regulatory point of view the MRL cannot be exceeded, and the MRL is 100-‐fold below a known safe dose. Moreover, in the course of the regulatory procedure a plant analysis is performed, and depending on the results obtained the plant may be qualified as ‘substantially equivalent to its conventional comparator’.
51
The company preparing the feed It is relevant to know which company is going to prepare the feed. Wilhelm replied that currently we do not know the company. G-‐TwYST is in the course of raising a tender. The outcome of this workshop, quality specifications and data on the composition will be included in the documents for the call, which will be a public European call. Integrated analysis of the data One of the questions concerned the integrated approach and how the results of the different studies will be combined and interpreted in total. Will there be a final publication on all these results? Schiemann explained that GRACE will do so during the course of the project. Steinberg added that G-‐TwYST will also have to integrate the data from the 90-‐day, the 1-‐year and 2-‐year studies. In work package 5 of G-‐TwYST there are two important tasks: • To describe triggers for long-‐term studies; • To define the added value of such long-‐term studies. This will be done by presenting all the data and discussing the significance of these data with the stakeholders. In addition, G-‐TwYST will not end before May 2018 and will be able to include the results of GRACE and GMO90+ projects. Parameters One of the participants urged to include endocrine effects. Another participant warned to be careful when it comes to steroid hormones. It is technically feasible but difficult to measure for example estradiol. This is also related to the status of the cycle of the female rat, and these parameters will definitely be difficult to correlate if the analyses are performed in a blinded way. Omics: More information needed A participant mentioned the GMO90+ project, which will receive organs from G-‐TwYST and asked for more clarity about what will be done in that project. Alison clarified that the organs from the 2-‐year trials will first be used for the histopathological analyses. If thereafter tissue samples still are available, they will be forwarded to GMO90+ for omics analysis. Bérengere Broux from GMO90+ explained that the protocols for the omics analyses have already been defined and will be on the internet in the next weeks. The omics work includes transcriptomics, metabolomics, and lipidomics. This work will be done by two private partners: Methodomics and Profilomic11. The samples will remain blinded for the omics study. GMO90+ also includes a histopathological analysis, a blood analysis and urinalysis. All analyses and parameters will be the same as in G-‐TwYST, and the details will also be made available on the GMO90+ website. Omics: No golden standard (yet) Steinberg reiterated that the omics data from GRACE and GMO90+ will also be included, but at the present time these cannot be used for regulatory purposes. He agreed with a previous comment concerning the difficulty of having a validated database e.g. on the different metabolite patterns in the blood of different rat strains. In Germany, for example, the Max Rubner-‐Institute has analyzed the metabolome of normal Fischer 344 rats. 11 http://www.methodomics.com/ and http://www.profilomic.com/fr/
52
Dosage of the diets One of the participants asked for the panel’s thoughts on the previous discussion about the dosage of GM maize in the diets. Steinberg replied that the project team had not come to a conclusion yet, but he would consider 2 doses with and without Roundup if one takes into account that in G-‐TwYST we have only one hit because of time and budget limitations. The project will take into account the comments of the stakeholders on the protein content of the diets. Next steps Spök explained the next steps in the stakeholder consultation process. First of all, the G-‐TwYST team will draft a workshop report that presents what has been discussed. The stakeholder consultation process will be presented in a report that will include the written comments and responses. Alongside the workshop report G-‐TwYST will publish a revised study plan that will show how comments have been processed, how the study plan has been modified accordingly or why comments could not be adopted. This will be disseminated in Spring 2015. Participants of the workshop will be notified when this is uploaded to the website. After publication of the final study plans the feeding trials will have to start soon. Therefore, there will be no opportunity to comment the study plans in a second round. Wilhelm added that the modified study plans may not satisfy all the questions and comments, some of which were contradictory. The G-‐TwYST team will try to figure out which study designs could be relevant tools for risk assessment, fit in the time line and meet the financial constraints. We will definitely not be able to include all testable parameters. The proposal of the G-‐TwYST team may not necessarily represent the final standard for conducting a feeding trial. The conclusions will also integrate the results of the experiments, comments that were made during this stakeholder engagement process and desktop studies that have been conducted elsewhere. Being asked for more time to deliver written comments the project team indicated that diet related comments would have to be sent as soon as possible because the tender for the production of the diets will have to go out in January. Closing remarks by the panel The panel thanked the participants for the enormous amount of very useful comments and the very constructive discussion. The information exchange has been very positive for all partners in G-‐TwYST and the panel members assured the audience that the comments and discussions will certainly influence the study plans. Regarding the written comments, G-‐TwYST would appreciate comments on testable hypotheses, which would fit into the experimental structure of G-‐TwYST.
53
Annex 1: Workshop agenda
16 December 2014
12:00 – 13:00 Registration and Welcome Snack
Introduction Chair: Huib de Vriend
13:00 – 13:15 Welcome Armin Spök, Alpen-‐Adria Universität Klagenfurt-‐Wien-‐ Graz, Graz
13:15 – 13:35 Introduction to G-‐TwYST Pablo Steinberg, University of Veterinary Medicine, Hannover
13:35 – 13:55 Stakeholder Involvement and Transparency in G-‐TwYST Armin Spök, Alpen-‐Adria Universität Klagenfurt-‐Wien-‐ Graz, Graz
13:55 – 14:10 Introduction to the Workshop Huib de Vriend, LIS Consult, Driebergen
14:10 – 15:00 Qs & As
15:00 – 15:30 Coffee Break
Animal Feeding Studies Chair: Armin Spök
15:30 – 16:00
Production of Plant Material Ralf Wilhelm, Julius Kühn Institut, Quedlinburg Jose Luis La Paz, Center for Research in Agricultural Genomics, Barcelona
16:00 – 16:30
Analysis of Plant Material, Storage, and Diet Preparations Ralf Wilhelm, Julius Kühn Institut, Quedlinburg Jose Luis La Paz, Center for Research in Agricultural Genomics, Barcelona
16:30 – 16:50 Test Facility Dagmar Zeljenkova, Slovenska Zdravotnicka Univerzita v Bratislave, Bratislava
16:50 – 17:20 Coffee Break
17:20 – 18:10
Study Design Pablo Steinberg, University of Veterinary Medicine, Hannover Dagmar Zeljenkova, Slovenska Zdravotnicka Univerzita v Bratislave, Bratislava Hilko van der Voet, Wageningen University and Research Centre, Wageningen
18:10 – 19:10 Discussion
Ca. 19:10 Adjourn for Day 1
54
17 December 2014
Animal Feeding Studies cont. Chair: Ralf Wilhelm
9:00 – 9:15 Recap Day 1
9:15 – 9:45 Histopathology Roger Alison, Roger Allison Ltd., Lampeter
9:45 – 10:15 Statistics and Prospective Power Analysis Hilko van der Voet, Wageningen University and Research Centre, Wageningen
10:15 – 10:35 Data storage and access Christian Kohl, Julius Kühn Institut, Quedlinburg
10:35 – 11:05 Coffee Break
11:05 – 11:25 Interlinkages of G-‐TwYST With Other Research Projects Joachim Schiemann, Julius Kühn Institut, Quedlinburg
General Discussion Chair: Armin Spök
11:25 – 13:15 Discussion
13:15 – 13:30 Closing Remarks
13:30 End of workshop
Annex 2: List of participants
Family name Name Affiliation Country Adduci Gloria Institute of Science, Technology, and Society
Studies, Alpen-‐Adria Universität Klagenfurt -‐ Wien-‐ Graz
Austria
Alexandrova-‐Stefanova
Nevena Agricultural Innovation Systems and Knowledge-‐Sharing Officer, Food and Agriculture Organization
Alison Clare Roger Alison Ltd. UK Alison Roger Roger Alison Ltd. UK Andersson Hans Christer National Food Agency Sweden Baumgärtner Wolfgang Dept. of Pathology, University of Veterinary
Medicine Germany
Broux Berangère Anses (GMO90+) France Brueller Werner Austrian Agency for Health and Food Safety
AGES Austria
Buesen Roland BASF SE de Vriend Huib LIS Consult Netherlands Engel Josefine Julius Kühn-‐Institut Germany Fouquet Rémi Ministry of Ecology, Sustainable Development
and Energy France
Schorsch Frédéric European Society of Toxicologic Pathology / Bayer
France
Fulgosi Hrvoje Institute Rudjer Boskovic Croatia Georgieva Violeta EuropaBio Belgium Gereková Petra Ministry of Agriculture and Rural Development
of Slovak Republic Slovak Republic
Ghinea Steluta NATIONAL ENVIRONMENTAL PROTECTION AGENCY
Romania
Gietl Eva KWS Saat AG Germany Greiter Anita Umweltbundesamt Austria Hardegger Markus Swiss Federal Office for Agriculture Switzerland Homoki Hajnalka
Homoki Ministry of Agriculture Biodiverzitás-‐ Biodiversity and Gene Conservation Unit
Hungary
Jany Klaus-‐Dieter WADI-‐International University (Syria) Germany Jekkel Zsolt Pioneer Hi-‐Bred Hungary Kleter Gijs A. RIKILT -‐ Wageningen UR Netherlands Koch Michael Monsanto Company USA Kohl Christian Julius Kühn-‐Institut Germany Lanzoni Anna EFSA Italy Leek Janneke Ministry of Health, Welfare and Sport Netherlands Lempp Charlotte Dept. of Pathology, University of Veterinary
Medicine Germany
Lombardo Michelangelo ISPRA Italy Matzeit Volker Federal Ministry of Food and Agriculture Germany Mesnage Robin King's College of London and CRIIGEN
scientific board France
56
Family name Name Affiliation Country Nolan Emma Euro Coop Belgium Racovita Monica Italy Rascle Jean-‐Baptiste BAYER S.A.S. Bulgaria Rastelli Valentina I.S.P.R.A. (National Institute for
Environmental Protection and Research) Italy
Ribarits Alexandra AGES -‐ Austrian Agency for Health and Food Safety
Austria
Scheepers Andrea Federal Office of Consumer Protection and Food Safety
Germany
Schiemann Joachim Julius Kühn-‐Institut Germany Schmidt Kerstin BioMath GmbH Germany Sipinen Ville Erling Norwegian scientific committee for food
safety GMO Panel Norway
Spök Armin Institute of Science, Technology, and Society Studies, Alpen-‐Adria Universität Klagenfurt -‐ Wien-‐ Graz
Austria
Staiano Giovanni I.S.P.R.A. (National Institute for Environmental Protection and Research)
Italy
Steinberg Pablo University of Veterinary Medicine Hannover Germany Stepanek Walter AGES -‐ Austrian Agency for Health and Food
Safety Austria
Sørensen Ilona Institute of Toxicology , Technical University of Denmark
Denmark
Teichmann Hanka Federal Agency for Nature Conservation (BfN), Division GMO Regulation, Biosafety
Germany
Then Christoph Testbiotech Germany Tulinska Jana Slovak Medical University Slovak
Republic van der Voet Hilko Wageningen University and Research Centre Netherlands Van Oers Tanja Ministry of Economic Affairs Netherlands Velten Guido Bayer CropScience Belgium Vikse Rose Norwegian scientific committee for food
safety GMO Panel Norway
Vybiral Dietmar Austrian Federal Ministry of Health, Dep. II/B/15
Austria
Wagner Kristina Eurogroup for Animals / German Animal Welfare Federation
Germany
Ward Keith Syngenta & EuropaBio UK Wilhelm Ralf Julius Kühn-‐Institut Germany Wögerbauer Markus AGES -‐ Austrian Agency for Health and Food
Safety / DSR Austria
Zeljenkova Dagmar Slovak Medical University Bratislava Slovak Republic
In blue are indicated G-‐TwYST partners or members of the Advisory Group.
Annex 3: Written comments received from stakeholders
A3.1. Dr. Robin Mesnage1*, Dr. Michael Antoniou1, Pr. Gilles-‐Eric Séralini2 Comments on the protocol used to assess the long-‐term toxicity of the NK603 Roundup tolerant GM maize in the G-‐TwYST project 1. Department of Medical and Molecular Genetics, King's College London School of Medicine, Guys Hospital, London, United Kingdom 2. Institute of Biology, CRIIGEN, EA 2608 and Risk Pole, MRSH-‐CNRS, University of Caen, Caen, France 1. Contamination of the feed: The contamination of the feed should be extensively measured. Generally, rat feeding studies only measure a few contaminants, such as banned organochlorine pesticides. This limited assessment of the tested substances can question the reliability of the conclusions (Mesnage et al., 2014). Recent analytical methods can measure hundreds of pesticides (including the most recently introduced ones) at a reasonable cost with a good sensitivity. For instance, the QuEChERS method (Eitzer et al., 2014) measuring more than 300 pesticides is available in accredited laboratories (ISO 17025 and ISO 9001:2008) (European and French Standard NF EN 15662 from January 2009 for foods of plant origin). For instance, the information at this link is from the company SGS: (http://www.sgs.com/~/media/Global/Documents/Flyers%20and%20Leaflets/SGS-‐AGRI-‐Pesticide-‐Residue-‐A4-‐EN-‐10.pdf) (http://www.sgs.com/Trade/Commodity-‐Trading/Agricultural-‐Goods/Softs/Pesticide-‐Residue.aspx) These methods should be applied to ensure a good assessment of the tested substances. 2. Rat stain The choice of the Wistar rat from Harlan is not inappropriate, but it makes the results difficult to compare with those of previous studies with the same GMO performed with Sprague-‐Dawleys. 3. Dose levels: During the stakeholder meeting, the question of the number of dose levels was raised (two or three?). OECD protocol 453 relative to the combined chronic toxicity/carcinogenicity studies recommends the use of at least 3 dose levels, plus a concurrent control, to ensure a good assessment of the dose-‐response relationship. To avoid imbalances in important nutrients and protein levels that can occur with large concentrations of the tested substances, we suggest the use of 11, 22 and 33% concentration of NK603 in a standard diet. 4. Cultivation of corn, Roundup residues and controls
58
The cultivation of the corn was supposed to be performed according to good agricultural practices, sprayed once before the summer and harvested in November. However, it does not correspond to classic farming management, which generally sprays Roundup at least twice, before and during the summer, adding in some cases an additional pre-‐harvest desiccation spray to facilitate the harvest. To ensure a relevant risk assessment, the exposure should have used the worst case scenario with additional sprays. Roundup residues are classically found in commercialized corn grain and are the potential source of health risks. The major concern about GMO toxicity is relative to their accumulation of residues from Roundup sprayed during the cultivation (Mesnage and Séralini, 2014). The toxicity of Roundup residues has never been tested for more than 28 days in rodents, and even without blood testing (ANSES, 2013). 5. Lack of potential toxicity of Roundup residues If the corn analysis does not reveal Roundup residues in the diet incorporating the NK603 sprayed with Roundup, an additional group of rats fed the control diet supplemented with Roundup residues should be added to ensure the capacity of the experimental design to cover all the possible health risk sources. 6. Use of historical data: For the statistical analysis, it is planned to use the historical data of the Harlan rat strain as a secondary control. These are not relevant because the rat feed may have been contaminated with different pesticides residues, heavy metals, and dioxins, and the rats may have been raised in different conditions. The bibliography on this topic is extensive.
- Contaminations by dioxins (Schecter, 1996) or heavy metals (Kozul et al., 2008) have been detected at levels that may confound the toxicity analysis.
- The background of pathologies is not stable over time: this is generally attributed to genetic variations because outbred rats are mostly used. However, the pathology background is not stable even in inbred rats, which cannot be affected by polymorphism, suggesting the influence of environmental factors (Kacew and Festing, 1999).
- The use of historical controls increases the variance and reduces the statistical power (Cuffe, 2011)
- The historical data pertaining to RccHan™: WIST rats have extremely wide ranges (Harlan, 2014). For instance, the incidence of some pituitary adenomas in supposedly non-‐treated female controls varied from 0 to 83.33% across studies. Obviously, some uncontrolled environmental factors affect the health of the rats and a differential effect in tumour induction will not be detectable with such an amplitude of so-‐called “spontaneous” neoplasms.
While a verbal assurance was made by the panel at the GM-‐Twyst workshop in Vienna (Dec 2014) that historical control data would only be used as an additional guide and not in the primary comparison, it should be written into the protocol that the researchers recognize that historical control data must not be used to dismiss statistically significant differences in the GM-‐fed group on the grounds that they are within the range of the historical control data.
59
7. Sex differences Females and males should not be analysed in the same statistical tests, but separately. Even liver and kidneys are sex-‐differentiated organs (Chang et al., 2013). General principles in endocrinology have taught us that we cannot expect similar responses in males and females after lifetime exposures (Diamanti-‐Kandarakis, 2009). 8. Biochemical analysis Even if the OECD guidelines recommend the measurement of biochemical parameters up to month 12, we recommend an additional measurement at the end of the study to cover the impact of late life exposures. Indeed, the protocol does not include the major windows of sensitivity to toxic effects, which are the prenatal period, puberty, and late life (Diamanti-‐Kandarakis et al., 2009). References ANSES, 2013. http://www.criigen.org/SiteFr//images//anses_letter.pdf Chang KA, Lin IC, Sheen JM, Chen YC, Chen CC, Tain YL, Hsieh CS, Huang LT. (2013) Sex differences of oxidative stress to cholestatic liver and kidney injury in young rats. Pediatr Neonatol. 54(2):95-‐101. Cuffe RL. (2011) The inclusion of historical control data may reduce the power of a confirmatory study. Stat Med. 30(12):1329-‐38. Diamanti-‐Kandarakis E, Bourguignon JP, Giudice LC, Hauser R, Prins GS, Soto AM, Zoeller RT, Gore AC. Endocrine-‐disrupting chemicals: an Endocrine Society scientific statement. Endocr Rev. 2009 Jun;30(4):293-‐342. doi: 10.1210/er.2009-‐0002. Eitzer, B. D., et al., 2014. Interlaboratory Comparison of a General Method To Screen Foods for Pesticides Using QuEChERs Extraction with High Performance Liquid Chromatography and High Resolution Mass Spectrometry. J Agric Food Chem. 62 (1), 80–87. Harlan (2014) Historical Control Data on Neoplastic Findings in RccHan™: WIST Rats Compiled data from 54 studies. Available at http://webapps.harlan.com/wistarhannover/ Kacew S and Festing MFW (1999) Role of Rat Strain in the Differential Sensitivity to Pharmaceutical Agents and Naturally Occurring Substances. J Toxicol Environ Health. 1996 Jan;47(1):1-‐30. Kozul, C., et al., 2008. Laboratory diet profoundly alters gene expression and confounds genomic analysis in mouse liver and lung. Chem Biol Interact. 173, 129 -‐ 140. Mesnage, R., et al., 2014. Letter to the Editor regarding " Delaney et al., 2014": Uncontrolled GMOs and their associated pesticides make the conclusions unreliable. Food Chem Toxicol. doi: 10.1016/j.fct.2014.07.003.
60
Mesnage, R. and Séralini, G.-‐É. (2014) The Need for a Closer Look at Pesticide Toxicity during GMO Assessment, in Practical Food Safety: Contemporary Issues and Future Directions (eds R. Bhat and V. M. Gómez-‐López), John Wiley & Sons, Ltd, Chichester, UK. doi: 10.1002/9781118474563.ch10 Schecter, A. J., et al., 1996. Exposure of laboratory animals to polychlorinated dibenzodioxins and polychlorinated dibenzofurans from commerical rodent chow. Chemosphere. 32, 501-‐8.
61
A3.2. Austrian Agency for Health and Food Safety (AGES) Dear Project Team, With regard to your request the Austrian Agency for Health and Food Safety would like to submit the following statement on the G-‐TwYST draft study plans. This statement was prepared by Austrian experts in a coordinated approach of the Federal Ministry of Health and the Austrian Agency for Health and Food Safety.
Comments on the G-‐TwYST stakeholder workshop and the study drafts presented there
The Austrian experts would like to submit written comments and questions in addition to the remarks already made at the stakeholder workshop held on 16th and 17th December 2014 at the Diplomatic Academy Vienna. We would kindly request to consider them in the planning and realisation of the animal toxicity studies within the G-‐TwYST project. 1) An important point for the feeding trials is dosing and the number of the test groups.
This was reflected by a broad discussion on this topic during the December workshop. Amongst other things, the inclusion of a 50% GMmaize/isogenic maize group was discussed, as well as the potential use of three dosage groups. We would recommend a 40% and a 20% dose group, because – from the experience with shorter-‐term studies which showed in the past not very substantial GMO caused effects – it would be advisable to go with the upper dose as high as possible without reaching imbalance and therefore potential toxicity levels (except for perhaps less weight gain or other minimal effects derived from shorter-‐term studies as lege artis requested for the highest dose in long-‐term/carcinogenicity studies). In this context, 50% dose groups could already pose a risk, and thus we would not recommend to got to that level in rat studies. The 20% dose group would be a usual gradation (high dose divided by two – geometric series) in toxicity tests. With respect to possible GMO contaminations of the control group (isogenic non-‐GM) we would like to ask for some clarifications: a) Has a contamination threshold been established? (0.9% would be appropriate from our
perspective; this is the threshold for adventitious or technically unavoidable traces of GM in food and feed.)
b) Will the charges be checked with respect to GMO content?
c) At which times contamination tests will be executed? (after harvest; before processing; before feeding?)
d) Would certain levels of contamination lead to elimination of the charge?
2) In relation to the endpoints it was already suggested at the workshop that in addition to standard parameters, parameters especially indicating immunotoxicity should be included. For this aim, immunoglobulin concentrations (for instance IgE, IgM, IgG) in blood or other endogenous fluids of the test animals could be measured. Appropriate "omics" analysis could help to evaluate potential edocrine disrupting effects (see for example "Genomic approaches for cross-‐species extrapolation in toxicology" (Benson and Di Giulio 2007)).
62
3) According to the presentations at the workshop, the evaluation and statistical analysis of the animal feeding studies will make use of equivalence limits instead of NOAEL or benchmark dose (BMD) approaches which are mentioned by OECD test guidelines. Even though we understand the problems with background variability and normal biological variation in the current studies, however, the usual approach for establishing safety criteria for human exposure would be to calculate NOAEL and reference dose. For exposure assessment a worst case scenario can be used. This is regularly done by GMO applicants yet, and is based on the assumption that any conventional food product will be replaced by a food derived from the GMO. By doing this, even a "margin of exposure" (MOE) could be calculated.
4) According to the workshop presentations, the production of the plant material for the animal
feeding tests is performed in Canada, the USA and Spain. It would be justified to provide more details on the actual test design and the underlying concept used at the production sites; in particular, whether or not a fully randomised plot design, as outlined in current EFSA guidance (EFSA 2011), has been used. If not, the reasons behind the presented concept should be explained and the approach should be set forth conclusively.
5) With respect to the irradiation of the feed material it should be properly documented that no
chemical changes of important nutrients (e.g. protein isomerisation) occur.
In addition, there are some more or less substantial drawbacks existing which could hamper the outcome of the studies.
• A fully blinded evaluation of the tissue samples should be expected-‐ not a mixture of blinded and non-‐blinded steps during the evaluation process.
• The Slovakian institute which performs the animal experiments is not accredited according to international certificates (e.g. ISO certification) and does not work according to good laboratory practices (i.e. not GLP conform)
• The animal sample processing steps are separated between several laboratories dispersed over Europe, and thus require an immense load of coordination, documentation and management. The execution of the experiments at a single renowned institution with a long and honoured record of expertise would have been preferable.
• The quality control at the tissue sample analysis institution/pathology should be executed by an external expert (not by a related party of the owner of the company).
References Benson WH, Di Giulio RT, 2007. Genomic approaches for cross-‐species extrapolation in toxicology: proceedings from the Workshop on emerging molecular and computational approaches for cross-‐species extrapolations, 18-‐22 July 2004, Portland, Oregon, USA. Pensacola, FL [etc.], SETAC [etc.].
EFSA, 2011. Guidance of the GMO Panel for risk assessment of food and feed from genetically modified plants. The EFSA Journal 9(5):2150: 1-‐37.
63
A3.3. EFSA
EFSA considerations on dose selection in GTwYST feeding studies in rodents (NK603: subchronic and combined chronic/carcinogenicity; MON810: carcinogenicity)
• The dose levels in feeding studies in rodents should be selected in accordance to the
appropriate guidance (OECD Technical Guidances , EFSA documents -‐ e.g. EFSA Scientific Committee Guidance on 90-‐day studies, 2011; EFSA Scientific Report on Considerations on the applicability of OECD TG 453 to whole food/feed testing, 2013). In the case of whole food feed, in particular, the well-‐known limitations in high dose selection should be considered.
• In any case, a clear justification of the dose levels selected should be provided for all protocols of the GTwYST feeding studies in rodents.
• Dose levels might differ in different types of studies. In particular, it is highlighted that in a comprehensive project aimed to support the safety assessment of a test substance, making use of several animal studies, the studies schedule is pivotal to achieve sound results and in the respect of the 3Rs principles. Typically, subchronic studies provide information relevant to set the dose for chronic and, in particular, for carcinogeniciy studies. Unfortunately, during the Stakeholder meeting it was explained that, due to time/budget constraints, this is not possible in the context of GTwYST (in particular for NK 603), with obvious limitations for a sound rationale for dose selection.
Subchronic study on NK 603 maize • EFSA recently provided indications on the incorporation rate to be used as high dose for some
crops, including maize, in subchronic feeding studies on whole GM food/feed(EFSA Explanatory Statement, 2014). These are based on a literature review on 90-‐day studies on whole food/feed. As discussed in the GTwYST Stakeholder meeting , the key message from EFSA 2014 is that the high dose should be properly justified, taking into account the best knowledge on the incorporation rate not associated with nutritional imbalance (e.g. from literature) and, if appropriate, on the basis of preliminary studies demonstrating its adequacy in the specific test conditions.
• The high dose (33% inclusion rate) used/proposed respectively in the subchronic study on MON810 (conducted in GRACE) or on NK603 (to be conducted in GTwYST) is not supported by any sound justification. Actually 33% was the incorporation rate previously tested by the applicant (again with no clear justification).
• Considering the issues/criticisms to the recently published paper on the results of GRACE 90-‐day study, it would be advisable to select the high dose level to ensure a more robust toxicological study.
• In this specific project no information is available to select an appropriate inclusion rate under the selected test conditions (HCC Wistar rats, that age, that facility).
• A further complication is represented by the fact that the rodent feeding studies performed in the context of G-‐TwYST will be interlinked with other research projects (GRACE, GMO90+). This interlinkage is claimed to increase comparability of results, expand data basis for political decisions, join forced and share costs (Schiemann -‐ Interlinkages of G-‐TwYST with other Research Projects).
• Against this background, it is recommended to select doses allowing such interlinkage, as appropriate. This particularly refers to subchronic studies on NK603 (G-‐TwYST and GMO 90Plus).
• Discussion with the diet provider on the experience/possibility of a high dose >33%maize incorporation rate is recommended.
• The low dose (currently 22%) could be modified accordingly.
64
Combined chronic/ carcinogenicity on NK 603; Carcinogenicity on MON810 • A literature search on the incorporation rate of maize in a carcinogenicity study is
recommended. • Same considerations as above • More than 1 dose levels should be selected.
A3.4. Eurogroup for Animals We would like to thank the G-‐TwYST team who organised the stakeholder workshop, which took place in Vienna on 16 and 17 December 2014, for the opportunity to attend and to express our views. As requested of us, we would like to put forward some comments on subjects highlighted during the discussions at the workshop.
The G-‐TwYST project is relevant to Eurogroup for Animals due to involvement of animals used in feeding studies with genetically modified (GM) maize. While there were also several scientific issues identified at the stakeholder workshop, we would like to focus our comments on the animal welfare issues we identified. Comments by Eurogroup for Animals:
• First of all, we would like to voice our concern that more feeding trials are conducted in the frame of an EU research project at all. For GRACE there was an obligatory call for the animal test study laid down in point 1.4.4.1 of Annex II of the Commission Implementing Regulation 503/2013/EU, on applications for authorisation of genetically modified food and feed. Here it specifically calls on the Commission, in Article 12, to monitor the outcome of the GRACE research project and to review the requirement to perform 90-‐day feeding studies in rodents with whole genetically modified food/feed on the basis of new scientific information. However, there is no similar call to conduct a 2-‐year feeding trial in Regulation 503/2013/EU or any other relevant legislation. At the stakeholder workshop, it was claimed several times that G-‐TwYST was initiated for political reasons in answer to the Séralini study rather than due to a concrete scientific concern. We would like to remind the project coordinators and the European Commission (EC) that the rules for project evaluation laid down in Article 38 of Directive 2010/63/EU also apply for research projects initiated by the EC. Thus, we believe that the 2-‐year feeding trials as they are planned now for G-‐TwYST do not fulfil any of the requirements or comply with Article 38, nor those laid down in Article 4, 5, 6 or 13 of the before mentioned Directive.
• It is not clear to us why the several EU projects that deal with the assessment of the validity of
feeding studies and with potential risks posed by GM plants in food and feed, where not planned so in a way that all testing data and project outcomes can be compared, i.e. there is now “need” to conduct a new 90-‐day feeding trial even though a 90-‐day feeding trial was already conducted in the frame of the GRACE project. The “need” to perform a new 90-‐day feeding trial is justified by project coordinators of G-‐TwYST that they use different GM maize breeds than those in GRACE. This is of very great concern to us because this inconsistency in planning results in more and new animal tests/feeding trials that could have been avoided if the experimental set-‐up of all respective EU projects would have been designed to build on from each other (i.e. use of the same GM breeds, same animal strains, treatment of plants, etc.).
• However, we welcome the fact that the project coordinators have made an effort to revise the
experimental set-‐up and provided an alternative set-‐up that refrains from conducting the 2-‐year carcinogenicity feeding trial with MON810 GM Maize. We would like to encourage the project coordinators to stick to this commitment in order to reduce the number of animals used in the GTwYST programme.
• There were numerous attempts during the stakeholder workshop to reason that the new 90-‐day
feeding trial was “needed” because where with GRACE, the GM maize MON810 was used while in
66
GTwYST, it is planned to assess GM maize NK603 only. It was also explained by the project coordinators that therefore, data and results could not be compared due to the different characteristics of the two GM breeds. In our opinion this is a very alarming issue. At least the feeding trials conducted in the frame of GRACE were supposed to deliver results to help in the assessment of the validity of feeding trials with GM plants for risk assessment (e.g. for human health) in general, not for one particular type of breed of GM maize. We see the danger that due to this proposed position, there will be a further need identified to create projects where each and every new or existing breed of GM plant is tested in feeding trials. From an animal welfare point of view, this cannot be justified.
• We call on the EC to act immediately and initiate a research project to develop and/or identify
suitable animal-‐free testing methods to replace feeding trials for testing of GM food and feed or integrate these type of methods into the methodology of the existing projects to at least balance the use of animals in GRACE and G-‐TwYST and to fulfil the requirements of Directive 2010/63/EU. The methods that are developed and/or identified in such a project would be used for risk assessment in other regulatory areas, like for Novel Foods, pharmaceuticals, etc. This would be a great chance to show where in silico, in vitro and other animal-‐free testing methods have a great potential and are superior to animal testing in terms of validity, significance reliability of data. It is the opportunity to send out a strong signal in favour of animal-‐free testing methods to the scientific and regulatory communities. The GRACE team should follow this through.
We call on the G-‐TwYST project coordinators to make all necessary changes to ensure that all reduction and refinement measures are fully exhausted.
67
A3.5. Bundesamt für Naturschutz, Germany “GM plants Two Year safety Testing” (G-‐TwYST) Comments of the Federal Agency for Nature Conservation (BfN), Germany, on the G-‐TwYST project Relating to the Stakeholder Workshop on 2-‐Year Animal Feeding Studies with GM Maize that took place from 16th – 17th December 2014 in Vienna we would like to submit the following comments.
1. General comments
The BfN appreciates projects to enhance the risk assessment of long-‐term effects of genetically modified (GM) food/feed. We are of the opinion that especially the approach of comparing different feeding studies for the assessment of potential medium and long-‐term toxicity of GM food/feed is an important task. Furthermore the simultaneous assessment of omics-‐techniques and toxicity/adverse effects with the same plant material increases the value of the project. The designated transparency and the early and continues stakeholder involvement is valued.
While the stakeholder workshop was open to discuss plans for the feeding studies the concept of the project has already been set. The objectives and the working program of the G-‐TwYST project are ambitious concerning the goal, the time frame and the budget. From our point of view some of the questions that are aimed to be investigated within this project should rather have been solved in advance or should at least be examined step by step now. Main objectives of this project are i) to define criteria for the scientific quality of long-‐term feeding studies, ii) to define when long-‐term animal feeding trails are scientifically justified and iii) to determine their added value for GMO risk assessment. The conclusions will be based on the results of the long-‐term toxicity studies that are performed within this project combined with the results of the related projects GRACE and GMO90+. However, at the present time there are no standardized protocols to study the potential short, medium-‐ and long-‐term toxicity of GM plants and derived products. The study plans for the three feeding studies that will be performed within the G-‐TwYST project are based on OECD Test Guidelines (TG) that have been developed and standardized for chemicals. Although EFSA provided guidance for subchronic toxicity studies as well as considerations on the applicability of OECD TG for chronic toxicity/carcinogenicity on whole GM food/feed the proposed adaptions have not been validated so far. For example, during the stakeholder workshop it was discussed which incorporation rate of maize in rat diet can be used as highest dose without providing an unbalanced diet and malnutrition. This is an example for a question that should be solved during validation of the test method and fits to the aim i) (quality criteria), but should have been solved in advance of any test with GMO used to answer the questions ii) (scientific justification) and iii) (added value). The determination of the maximum applicable dose is just an example for adaptions of OECD TG that are necessary for GMO testing and needed to be validated first. As no validated toxicity test guidelines for GMO exist, it will hardly be possible to answer the questions of scientific justification and added value of long-‐term feeding studies for GM food/feed in general. Nevertheless the scheduled long-‐term toxicity studies are appreciated because they will improve the knowledge on long-‐term toxicity of GM maize MON810 and NK603.
68
2. Specific comments
Production of plant material:
The GM maize NK603 was cultivated in Canada and the USA in 2014. Glyphosate was applied in a Roundup formulated product according to the agricultural practice that is common at the specific sites of cultivation. In Canada Roundup Transorb HC (540 g/L Glyphosate) was sprayed with a rate of 2.5 L/ha once, i.e. 1.35 kg/ha Glyphosate in total. The application rate of Glyphosate and the type of Roundup formulation used in the USA was not given at the stakeholder workshop but was announced to be provided later. In order to test the worst case, the GM maize NK603 used in the feeding studies (treatment type: NK603 + Roundup) should have been sprayed with the maximum number and highest rates of Roundup application, for which authorization is given. In the long-‐term feeding study with GM Maize NK 603 performed by Séralini et al. (2012) Roundup Weather-‐MAX (540 g/L Glyphosate) was applied with a rate of 3 L/ha, i.e. 1.62 kg/ha Glyphosate.
As the Séralini study was one trigger for the G-‐TwYST project, the use of the same Roundup formulation at least with same or even higher application rate is advisable for the production of test material for the G-‐TwYST project.
Diet:
As it is not expected that GM maize NK603 and MON810 will be highly toxic, the GM maize content in the highest dose group should be as high as possible. Hence, we support EFSAs proposal to use a higher incorporation than 33% maize in the highest dose group.
It should be verified by analysis that all diet components except from the GM test material are free from GMO, i.e. the control maize as well as all other diet ingredients (soy meal etc.). GMO contamination in diet may mask effects of the test item and therefore have to be excluded. Also the test item itself has to be free from contamination with any other GM material.
Reference:
Séralini, G.-‐E., Clair, E., Mesngae, R., Gress, S., Defarge, N., Malatesta, M., Hennequin, D., de Vendômois, J.S. (2012) Long term toxicity of a Roundup herbicide and a Roundup-‐tolerant genetically modified maize. Food Chem. Toxicol. 4221-‐4231.
69
A3.6. ANSES, Risk Assessment Department, Food Biological Risk Assessment Unit, France
Combined chronic toxicity and carcinogenicity study in rats fed GM maize NK603 -‐ Study
Plan, Study No: 632165 A/2015/GLP Comments about the Study Plan General comment 1 : The study plan doesn't highlight the specificities of the chronic toxicity and the carcinogenicity parts of the study, like the OECD Guideline 453 does. Therefore, the following elements of the OECD Guideline 453 are missing in the G-‐TwYST study plan : 1) Historical data : "OBSERVATIONS (CHRONIC TOXICITY PHASE) Haematology and Clinical Biochemistry 46. It is generally considered that baseline haematological and clinical biochemistry variables need be determined before treatment for dog studies, but need not be determined in rodent studies (38). However, if historical baseline data (see paragraph 58) are inadequate, consideration should be given to generating such data." "DATA AND REPORTING (CARCINOGENICITY AND CHRONIC TOXICITY) 58. Historical control data may be valuable in the interpretation of the results of the study, e.g, in the case when there are indications that the data provided by the concurrent controls are substantially out of line when compared to recent data from control animals from the same test facility/colony. Historical control data, if evaluated, should be submitted from the same laboratory, relate to animals of the same age and strain, generated during the five years preceding the study in question." 2) Organ weights : "OBSERVATIONS (CHRONIC TOXICITY PHASE) Pathology Gross necropsy 48. Organ weights should be collected from all animals, other than those excluded by the latter part of paragraph 47. The adrenals, brain, epididymides, heart, kidneys, liver, ovaries, spleen, testes, thyroid (weighed post-‐fixation, with parathyroids), and uterus of all animals (apart from those found moribund and/or intercurrently killed) should be trimmed of any adherent tissue, as appropriate, and their wet weight taken as soon as possible after dissection to prevent drying." 3) Gross necropsy : "OBSERVATIONS (CHRONIC TOXICITY PHASE) Pathology Gross necropsy 47. All animals in the study shall be normally subjected to a full, detailed gross necropsy which includes careful examination of the external surface of the body, all orifices, and the cranial, thoracic and abdominal cavities and their contents. However provision may also be made (in the interim kill or satellite groups) for measurements to be restricted to specific, key measures such as neurotoxicity or immunotoxicity (see paragraph 21). These animals need not be subjected to necropsy and the subsequent procedures described in the following paragraphs. Sentinel animals may require necropsy on a case-‐by-‐case basis, at the discretion of the study director."
70
4) Tumours : "OBSERVATIONS (CARCINOGENICITY PHASE) 51. All animals should be checked for morbidity or mortality, usually at the beginning and the end of each day, including at weekends and holidays. Animals should additionally be checked once a day for specific signs of toxicological relevance. In the case of gavage studies, animals should be checked in the period immediately following dosing. Particular attention should be paid to tumour development; and the time of tumour onset, location, dimensions, appearance, and progression of each grossly visible or palpable tumour should be recorded." General comment 2 : The 90 days study with the 50 % group should be continued with a 1 year and a 2 years studies. 1-‐ NATIONAL REGULATIONS, GUIDELINES AND STANDARDS Good Laboratory Practice "The study will be conducted in accordance with the OECD Principles of Good Laboratory Practice, as revised in 1997 (ENV/MC/CHEM(98)17), and the EU Commission Directive 2004/10/EC (adopted on the 11th of February 2004; Official Journal No L 50/44)".
Comments : Regarding the Multisite study, attention must be underlined about the communication between test sites "OECD SERIES ON PRINCIPLES OF GOOD LABORATORY PRACTICE AND COMPLIANCE MONITORING Number 13 Consensus Document of the Working Group on Good Laboratory Practice The Application of the OECD Principles of GLP to the Organisation and Management of Multi-‐Site Studies".
"Communication For a multi-‐site study to be conducted successfully, it is imperative that all parties involved are aware of their responsibilities. In order to discharge these responsibilities, and to deal with any events that may need to be addressed during the conduct of the study, the flow of information and effective communication among the sponsor, management at sites, the Study Director, Principal Investigator(s), Quality Assurance and study personnel is of paramount importance. The mechanism for communication of study-‐related information among these parties should be agreed in advance and documented. The Study Director should be kept informed of the progress of the study at all sites". 2-‐ GENERAL INFORMATION Archiving "The following documents will be archived under code number 632165A/2015/GLP at the Registry of accredited laboratories and laboratories with GLP certificate of SZU until the year 2026" Comments : The SOP used is not indicated.
71
What is planned after the 10-‐years archiving ? Proposed Time Schedule Comments : The early treatment is different according to gender. Consequently, haematological and biochemical analyzes will be shifted, with a possible effect time analysis between gender. 3-‐ TEST AND CONTROL ITEMS Test item Comments : It would be interesting to know the residual level of glyphosate and AMPA in the GM maize crop. 4-‐ MATERIALS AND METHODS Test item preparation – Diet formulation Comments : It must be noted that "Analyse for nutritive components and relevant possible contaminants are performed regularly. Certificates of analysis are retained".
Storage conditions Comments : Close storage room protected against rodents, insects ? Water Comments : What is the composition of the bottles ? Why are the bottles autoclaved and not cleaned with detergent and disinfectant the same way as cages ? Animal housing "The animals will be subjected to a 12-‐hour light/12-‐hour dark cycle" Comments : What is the light period (which hours in the day) ? Experimental Design Comments : All animal will be housed in 4 separate rooms. Is there a possible room housed effect ? General experimental design with NK603 maize, start March-‐April 2015 "The dose groups will be unblinded at the time of necropsy."
72
Comments : What is the interest to perform the study blinded until this time of the experiment ? Periodical Health Status Observations Clinical signs Detailed physical examination and functional assessment "The outcome of this examination will be recorded for each animal in accordance with the SOP ŠPP/TOX/V003 (Origin of score system: Ország A. et al. [1985] Veterinárna ortopédia a rontgenológia, Bratislava: Príroda, 243 p.". Comments : The score system has been validated compared to Irwin or FOB test ? Ophthalmologic examination Comments : SOP not indicated. Procedures For Sample Collection "At the end of the first and second year 500 μl plasma/rat and 2-‐3 ml urine/rat from 16 rats per experimental group and at the end of the second year liver and kidney samples from 16 rats per experimental group will be sent by SZU to the French consortium GMO90+, which will analyze the expression of a number of biomarkers of effects in the above-‐mentioned samples". Comments : How will the 16 rats (compared to 20 or 50 by group) be chosen ? What is the rational to test only 16 animals ? In a recent study on dairy cows performed with MON 810, gene expression pattern of markers for apoptosis, inflammation and cell cycle were assessed from liver and gastrointestinal tract. It would be interesting to collect and store gastrointestinal tissue for possible additional studies. Urine and blood collection and processing Sample collection and tissue processing at the end of the study Comments : Only one person for blood taking from the abdominal aorta (laparotomy and blood taking) and for decapitation and necropsy of the head including brain ? Haematology "Blood samples will be stored at room temperature (17-‐25°C), maximally up to 4 hours, until measurement". Comments : Were the storage conditions and temperature until measurement validated ? There is no shift between first and last blood sample collection ?
73
Clinical chemistry Comments : Bilirubin (total) ? Will calculated globulin and albumin/globulin ratio be tabulated ? Urinalysis Comments : Why only 8 animals by group ? The list of parameters tested is short : creatinine, N-‐acetyl-‐β-‐D-‐glucosaminidase (NAG), gamma-‐glutamyl transferase (GGT), sodium, potassium, chloride. Why ? Necropsy and Histopathology Gross necropsy Histology processing Comments : SOP not indicated.
74
A3.7. Crop Life and Europabio
Crop Life International Toxicology Working Group and EuropaBio
Comments to G-‐TwYST on Study Plans The Crop Life International (CLI) Toxicology Working Group and EuropaBio would like to thank the G-‐TwYST Project and its organizers for the opportunity to engage in the stakeholder session and provide feedback on the conduct of the two-‐year feeding study. As previously stated, CLI maintains that long term feeding studies are unnecessary to demonstrate the safety of GM crops as food and feed; particularly, when the safety of the gene, its transcription product, and the transgenic event itself have all been thoroughly established as part of its commercial registration. Furthermore, the inserted genes DNA and/or the expressed traits will not remain biologically active after dietary consumption and therefore cannot be expected to cause any toxicity including carcinogenicity. The Weight of Evidence from the safety data of these crops consists of thorough scientific assessment of (i) molecular characterization of the inserted DNA and expression of target proteins, (ii) comparative analysis of agronomic traits and composition, and (iii) the safety of the new protein and the whole food/feed were for potential differences in toxicity, allergenicity and nutritional quality. Consequently, independent regulatory reviewers have repeatedly concluded GMs are as safe as their conventional counterpart. As was the case for NK603, MON 810, and many other GM crops, the feeding studies with the GM crop confirmed the absence of toxicity predicted by the pre-‐existing Weight of Evidence (Hammond et al., 2004; Hammond et al., 2006; Zeljenkova et al., 2014; Snell et al., 2012). General Comments on the Suitability of Two-‐year Studies for GM Crops Two-‐year toxicity studies are designed to evaluate the carcinogenic potential of chemicals, pesticides and drugs over the lifespan of the test subjects. This is a scientifically valid endeavor because chemicals themselves may be carcinogenic or they may be metabolically activated to a carcinogen (e.g., arylamines such as benzidine must undergo P450-‐dependent activation and glucuronidation to become carcinogens) (Kemper et al., 2008). However, the scientific validity of two-‐year toxicity studies from GM crops is questionable because crops such as maize are not known to be carcinogenic and neither are the nucleic acids comprising transgenes or the proteins expressed by these transgenes. Indeed, proteins and nucleic acids are a natural part of foods, and are consumed in gram quantities daily for proper nutrition (Jonas et al, 2001). It is known from long experience that dietary proteins consumed orally are typically broken down into peptide fragments, and are used for energy production or reconstituted into new proteins. DNA and RNA are also readily digested in the gastrointestinal tract by nucleases in saliva, denaturation and depurination by acid in the stomach, intestinal nucleases, pancreatic secretions of bile salts, and degradative enzymes such as hydrolases. Apart from nucleic acid breakdown, gastrointestinal uptake of intact DNA or RNA is further prevented by polysaccharide coating of the intestinal epithelium (which acts as a molecular sieve), cell membrane barriers, and for any nucleic acid remnants, intracellular degradation via endosomes and/or lysosomes (Libonati and Sorrentino, 1992; 2001; O'Neill et al., 2011; Park, et al., 2006; Parrott et al., 2010; Yakovlev, et al., 1995). Thus, systemic exposures to intact transgenes and their protein products are anticipated to be negligible. There is also a well-‐documented relationship between the structural and functional integrity of both proteins and nucleic acids (Lodish et al., 2004; Alberts et al., 2002); and a recent absorption, distribution, metabolism, and excretion (ADME) study demonstrates that nucleic acids rapidly
75
degrade in serum (Christensen et al., 2013). Thus, digestion, even if it is only partial, ensures that DNA, RNA and proteins will not be active once systemically absorbed. In vitro digestion assays with simulated gastrointestinal fluids demonstrate that proteins introduced into GM crops are rapidly degraded, including CP4 EPSPS from NK603 and Cry1Ab from MON810 (EFSA, 2003; EFSA, 2009a; EFSA 2009b). Consequently, there is no scientific basis to believe that digestible dietary protein or DNA/RNA from GM crops will be systemically absorbed intact, or will remain biologically active when absorbed from the gastrointestinal tract. Therefore, their potential to exert adverse systemic effects is also negligible. Given the information above, there is no scientific basis to suspect digestible proteins and DNA/RNA will contribute to the initiation (e.g., DNA modification or mutation), promotion (e.g., increase in cell proliferation of decrease in cell death), and/or progression (e.g., chromosome disarangement) of cancer (Klaunig and Kamendulis, 2008) after oral consumption. A recent review publication (Snell et al., 2012), which examined both “long-‐term” studies (90 days-‐2 years) and multigenerational studies (2-‐5 generations), comes to a similar conclusion stating, “…it is clear that GM food is not revealed to be harmful when the duration of feeding is increased to well over 90 days”, and “…the multigenerational studies on animals fed GM plants do not reveal signs of toxicity or other macroscopic effects on health” (Snell at al., 2012). Indeed, there is evidence that an absence of adverse findings with GM traits and crops on shorter term tests provides insight into their safety when administered on a chronic basis. For example, studies feeding separate groups of rats either GM or non-‐GM soybeans for six months, one year, or two years found no meaningful differences among common toxicological endpoints in rats fed GM or non-‐GM soybeans (Sakamoto et al., 2007; Sakamoto et al., 2008). These results were consistent with the results of shorter term testing conducted in mice, rats, broiler chickens, catfish, and dairy cows that demonstrated the GM soybeans were as safe and nutritious as their traditional comparator (Harrison et al., 1996; Hammond et al., 1996). Lastly, a review conducted by the United States National Toxicology Program (NTP) demonstrated that 70% of the test substances evaluated in the review had results from 2-‐year studies that could be seen or would be predicted by the results of the 90-‐day study (Betton et al., 1994). Based on this extensive review, and other similar findings from other researchers (Munro et al., 1996; Kroes et al., 2004), EFSA is on record as stating, “Rodent feeding studies of 90-‐days duration appear to be sufficient to pick up adverse effects of diverse compounds that would also give adverse effects after chronic exposure, and therefore in general, chronic toxicity testing of GM food and feed does not seem to generate additional valuable information to the safety assessment” (EFSA, 2008). Thus, the available literature suggests there is not a strong scientific basis for anticipating adverse effects after a 2-‐year feeding study with GM maize when the initial 90-‐day study with the crop found no adverse effects. CLI recognizes that the political environment in the European Union is such that investigations of the utility of longer term testing as part of the safety evalauation of GM crops have become a reality despite a clear scientific rationale. From this position, CLI offers the feedback below based on what we have learned in our collective experience with rodent feeding studies (>100 in total). Implementing these recommendations should maximize the chances of producing interpretable data while minimizing the chances that the study will be confounded by nutritional imbalance or other variables.
76
General Comments on the Two-‐year Study with MON 810 (proposed Study No. 632165 B/2015/GLP) In addition to the points made above, a GLP-‐compliant 90-‐day study with MON 810 has been conducted, and the results have been published in the peer-‐reviewed scientific literature (Hammond et al., 2006) and reviewed by regulatory authorities around the world (MON 810 is approved as food and/or feed in Argentina, Australia, Brazil, Canada, China, Colombia, the European Union, Japan, Korea, Mexico, the Philippines, South Africa, Switzerland, Taiwan, the United States, and Uruguay)12. All reviewers have deduced that the data presented in these documents support the conclusion that MON 810 is as safe as conventional maize when fed to rats for 90 days. Likewise, the results of the GRACE Project’s 90-‐day feeding studies, “show that the MON810 maize at a level of up to 33 % in the diet did not induce adverse effects in male and female Wistar Han RCC rats after subchronic exposure, independently of the two different genetic backgrounds of the event” (Zeljenkova et al., 2014). In light of the results of three 90-‐day studies with MON 810 being published in the scientific literature, the NTP analysis, and EFSA’s opinion on longer term studies, the scientific value of conducting longer term testing with MON 810 is questionable. For all of the reasons stated above, CLI believes the conduct of the 2-‐year study with MON 810 is unnecessary. Consequently, given that the conduct of a long term test is a requirement of the funding received from the EU 7th Framework Programme (European Commission call KBBE.2013.3.5-‐03), we recommend shifting the financial resources planned for the conduct of this study to one combined chronic toxicity/carcinogenicity study that is sufficiently robust and thus has the most potential for producing interpretable results. General Comments on the herbicide tolerant maize 90-‐day Study (proposed Study No: 632165/2016/GLP) A GLP-‐compliant 90-‐day study with NK603 has been conducted, and the results have been published in the peer-‐reviewed scientific literature (Hammond et al., 2004) and reviewed by regulatory authorities around the world. Reviewers have deduced that the data supports the conclusion that NK603 is as safe as conventional maize when fed to rats for 90 days. Consequently, the scientific rationale, and Animal Care and Use justification, for repeating this study is not apparent. Furthermore, the current 90-‐day study is to be run concurrently with the longer term study; this defeats an important part of its purpose. For chemicals, a 90-‐day toxicity study is normally run as a precursor to longer term studies (e.g., a combined chronic toxicity/carcinogenicity study) to ensure that the high dose selected for the chronic study does not exceed the Maximum Tolerated Dose – a dose anticipated to produce limited toxicity when administered for the longer period of the chronic study13 . When the 90-‐day and chronic/carcinogenicity studies are run concurrently, at similar inclusion levels, the utility of the 90-‐day data is diminished. Especially, due to the availability of the 90-‐day NK603 study results from Hammond et al. (2004). For these reasons CLI believes the repeated conduct of the 90-‐day study with NK603 is unnecessary. The financial resources planned for the conduct of this study could be shifted to the combined chronic toxicity/carcinogenicity study planned with herbicide tolerant maize to enhance its robustness. A more robust chronic/carcinogenicity study will maximize its potential for producing interpretable results.
12 http://www.cera-‐gmc.org/GmCropDatabaseEvent/MON810 13 http://sis.nlm.nih.gov/enviro/iupacglossary/glossarym.html
77
Comments on the Combined Chronic Toxicity/Carcinogenicity Two-‐year Study with herbicide tolerant maize NK603 (Study No. 632165 A/2015/GLP) In the event that G-‐TwYST proceeds with the conduct of the two-‐year study with MON 810 (proposed study No. 632165 B/2015/GLP), the comments in this section will apply to the corresponding sections of that Study Plan as well. For MON 810, two separate publications support pre-‐empting inclusion of a 3-‐month time point for clinical pathology due to an absence if findings during 90-‐day studies (Hammond et al., 2006; Zeljenkova et al., 2014). General comments: 1. The precise objectives of the study currently remain unclear, particularly with regard to specific
testable hypotheses. These need to be decided upon and clearly indicated in order to be able to select the most appropriate treatment structure and level of replication. Treatment structure also has the potential to impact on the method of statistical analysis.
2. Given the overall aim of this study is to inform on the design and value of future studies rather than to assess the safety of the chosen test materials, replication can be set pragmatically rather than strictly on the basis of statistical power. This is especially relevant given that the accuracy of power calculations is dependent on having a reliable estimate of the expected experimental error, which we do not have due to the lack of relevant data.
3. In some sections the Study Plan lacks sufficient information and direction to the study staff. Per
OECD’s GLP guidelines, the Study Plan or Protocol should contain, “Detailed information on the experimental design, including a description of the chronological procedure of the study, all methods, materials and conditions, type and frequency of analysis, measurements, observations and examinations to be performed, and statistical methods to be used (if any).” Some of the Specific Comments below attempt to address sections where insufficient detail may be observed.
4. The current Study Plan contains assessment intervals and endpoints not indicated in OECD Test Guideline (TG) 453. Reverting to the recommended assessment paradigm in OECD TG 453 would enable a substantial reduction in the number of ophthalmic exams (1000 exams; see Specific Comment 11 for details) and clinical pathology assessments (1200 hematology, clinical chemistry, and urinalysis evaluations; See Specific Comments 14, 15, and 18 for details). The funding currently intended to support these unnecessary assessments could be used to design a more robust study with multiple Reference Groups to expand the Historical Control (HC) data and analyses to confirm that each batch of diet prepared is suitable for administration prior to feeding. These are important considerations particularly if a 2 year study is conducted with higher doses of the grain than what is available in the normal routine rodent chow, and considering that (i) the testing facility does not have HC data for chronic toxicity/carcinogenicity studies, and (ii) that contaminants or nutritional imbalances in the diet could hopelessly confound the results of the feeding study.
5. The method of euthanasia is never distinctly mentioned in the protocol. To ensure uniform
termination between treatment groups and genders a distinct method (i.e., CO2 inhalation, cervical dislocation, etc.) should be included. This will minimize the potential for introducing variability into the study design.
6. All activities conducted outside of Slovak Medical University (SMU) should be clearly identified as
GLP or non-‐GLP and Principal Investigators assigned as appropriate. While this has been done for
78
certain study components (i.e., histology, histopathology, and biostatistics), other elements of study design are less certain (-‐omics analyses as part of the French consortium GMO90+).
7. GLP statement: For the purposes of transparency, please clearly identify the components of the
study which are not GLP-‐compliant. 8. SOPs are referenced throughout the document. To increase transparency and facilitate
understanding among stakeholders, CLI requests that these SOP documents be included on the website and be accessible to stakeholders.
Specific Comments: 1. Line 88, Additional responsibilities Please indicate that peer review means histopathological examination and assessment. 2. Lines 158-‐163, Objectives
In Lines 161-‐163 it is stated that the studies will provide a “comparative assessment” of the results of shorter term subchronic toxicity studies versus extended chronic toxicity and carcogenicity studies. The use of the term “comparative assessment” is perhaps misleading because the term is usually reserved for cases in which formal statistically-‐based comparisons can be made (as in a traditional regulatory safety study, for example). Here, any comparison between studies will be more qualitative in nature and not a true comparative assessment in the statistical sense.
3. Lines 165-‐173, Test Item In its present form the level of detail provided for identification of the test item is insufficient. According to the stakeholder meetings, four varieties (2 transgenic and 2 conventional) were grown for the study. Each transgenic variety (P8906R or Prairie Brand 882RR2) had a near isogenic, conventional variety (P8906 or Prairie Brand 882) grown in a similar geographic region under similar environmental conditions. Consequently, they would be a suitable concurrent control item for their respective transgenic variety. However, the transgenic varieties are not identical to each other, nor are the conventional varieties the near isogenic control of the other transgenic variety (i.e., P8906 is not the near isogen of Prairie Brand 882RR2). Thus, in accordance with national (EFSA, 2011) and international (Codex, 2009) guidance on GM feeding studies they should not be used interchangeably. CLI recommends revision to clearly indicate the variety chosen for testing and provide GLP-‐compliant identification, descriptive, and characterization information. Lastly, the variety chosen for testing should also replace “NK603” throughout the text of the document. 4. Lines 174-‐175, Control Item Per the comment above, please clearly indicate the variety chosen as the control and provide GLP-‐compliant identification, descriptive, and characterization information. 5. Lines 189-‐199, Justification for the selection and number of animals
79
Justification should be given on the high number of animals used for the chronic phase of the study. It is true that 20 animals per sex and gender are typically utilized for chronic studies (OECD TG 452) for statistical reasons. However, OECD TG 453 allows for a reduced animal number (10 per sex/group): “Each dose group (as outlined in paragraph 22) and concurrent control group intended for the chronic toxicity phase of the study should contain at least 10 animals of each sex, in the case of rodents. It should be noted that this number is lower than in the chronic toxicity study TG 452. The interpretation of the data from the reduced number of animals per group in the chronic toxicity phase of this combined study will however be supported by the data from the larger number of animals in the carcinogenicity phase of the study.” Thus, the current design indicates that 100 animals more than is necessary are being used. Additional justification for increased animal usage should be provided. Also, the fate of the sentinel animals is not described in sufficient detail. Any procedures or endpoints to be investigated for these animals should be clearly defined in the document. 6. Lines 204-‐210, Test item preparation -‐ Diet formulation The current level of detail in this section is insufficient for this critical study design element. Please clearly indicate the dietary formulation to be administered to the animals and whether or not it is known to be nutritionally balanced and equivalent between test and control groups. Per comments on diet production provided separately to G-‐TwYST on 9 January 2015, a nutritionally-‐balanced diet similar to the diet fed to animals in Harlan’s Historical Control (HC) database should be fed to the animals. This will facilitate use of the Harlan HC data to help determine whether or not differences observed are outside the normal range for the test species. This is an important way of determining the toxicological relevance of differences on all toxicology studies, including GM crop feeding studies. If the diets being fed on this study are not similar to the HC database, then comparisons to HC may not be valid. It is anticipated that Harlan and the company nutritionist will be able to provide diet formulation information for the diets prepared for this study. Also, comparisons between treatment groups on the study should only be made if the diets fed to the groups are nutritionally balanced and equivalent. CLI recommends adding specific information regarding dietary analysis for environmental contaminants such as aflatoxins, pesticide residues, and heavy metals; and nutritional components should be included (please refer to the detailed comments submitted on 9 January, 2015 for reference). This information, and the formulation records, will demonstrate that the diets are nutritionally balanced and free of contaminants that could confound interpretation of the study results. Specifics regarding dietary sample size, number, identification, and storage conditions; shipping contact information; shipping conditions; and the analyses to be conducted on the samples should be included in this section. A physical description of each diet should be included in the study record (recorded prior to treatment) to minimize the likelihood that diets with grossly different appearances would be administered to different treatment groups on the same study. Gross differences in appearance may be an indicator of substantial differences in content or composition (two factors which could confound study results), and thus should be avoided. Please indicate whether or not the listed activities will be GLP-‐compliant and the individuals responsible for the conduct of each activity.
80
7. Lines 211-‐214, Storage conditions To ensure proper storage and Study Plan compliance, please indicate the controlled temperature and humidity range (e.g., 22 ± 3°C and relative humidity 50 ± 20%). 8. Lines 215-‐220, Water CLI recommends utilizing filters that will minimize the potential for all environmental contaminants (microorganisms, metals, chemicals, etc.) in the drinking water. 9. Lines 221-‐248, Animal housing Five animal rooms are indicated for 14 racks of animals, thus an even number of racks cannot be distributed among the rooms. In either this section or the Randomization section please indicate how the racks will be distributed among the rooms to control for the impact of slightly different environmental conditions in each room. For example, the racks could be distributed as indicated below if two larger (arbitrarily designated 308 and 309) and three smaller (arbitrarily designated 310, 311, and 312) rooms were available:
Racks Room Ca1, Ca2 (males), Ca6, and Ca7 (females) 308 Ca3, Ca4 (males), Ca8, and Ca9 (females) 309
Ca5 (males) and Ca10 (females) 310 Ch1 (males) and Ch3 (females) 311 Ch2 (males) and Ch4 (females) 312
10. Lines 250-‐263, Animal receipt and acclimation An acclimation period of 4-‐6 days is unusually short for a longer term feeding study; ≥2 weeks is more common. Morover, OECD TG 453 requires healthy animals be acclimated for at least 7days. Either way, these acclimation timelines are inconsistent with the study schedule which indicated animal delivery in February 2015 and the beginning of treatment in March-‐April 2015. Please revise for consistency between text of the document and study schedule. 11. Lines 264-‐298, Randomization The level of detail provided in this section seems inconsistent with that provided in the other sections (i.e., it is far more detailed). For the sake of continuity, and utility of the document to the technical staff conducting the study, please consider moving some of the details (e.g., tables indicating randomization schemes and ANOVA details) to an appendix. Given that the animals are reported to fall within a reasonably wide initial weight range (+/-‐ 20% of the mean), blocking by initial weight should be seriously considered (a) to avoid the possibility of having two rats of very different sizes in the same cage, and (b) to help reduce the residual error. While the rationale for staggering the start date between males and females is understood, this effectively invalidates any test of significance of the sex effect, as does housing males and females in separate rooms. This is of no real consequence as there is no interest in the comparison of sexes per se, but this limitation should be made clear in the protocol.
81
12. Lines 299 – 310, Route of administration To avoid the potential for cross-‐contamination of diets, CLI recommends filling feeders outside of the animal room. Color coding diets, and cage cards, for treatment groups should also minimize the potential for errors in diet administration. 13. Lines 311-‐321, General experimental design with NK603 maize, start March-‐April 2015 The robustness of the study can be improved with slight modifications of the indicated study design. The current Groups 2 and 3 (unsprayed, low and high GM maize incorporation rates, respectively), could be omitted and the animals utilized to include two reference groups fed conventional maize varieties in the study. In the absence of HC data for 2-‐year feeding studies at the testing facility these reference groups would provide a better understanding of the normal range of variability under the conditions of testing. While HC data from the animal supplier may partially address this need, a more robust study would have a better understanding of normal variability at the lab’s site and with their equipment. Moreover, the utility of including groups fed unsprayed grain is called into question by recent EFSA guidance on the conduct of rodent feeding studies with GM crops which indicates the test items should be grain treated with the intended herbicide when the test item is an herbicide-‐tolerant variety (EFSA, 2014). This makes inherent sense because a farmer is highly unlikely to pay more for seed containing an herbicide tolerance trait, and then not utilize that technology to improve agronomic performance. To maximize the potential of the study to reach definitive conclusions, more than 2 reference groups should be added to the design; perhaps as many as 6 reference groups. CLI understands that a finite amount of funding has been granted. However, the 90-‐day NK603 study and 2-‐year MON 810 study could be halted based on the existing safety data. Likewise, other cost-‐saving measures have been identified in the comments on this Study Plan (please refer to Specific Comments 11, 14, 15, and 18). These revisions would allow G-‐TwYST to redeploy the financial resources from these unnecessary studies and endpoints to the current chronic toxicity/carcinogenicity study to improve its design, robustness, and ultimately the interpretability of its results. 14. Lines 346-‐351, Clinical signs, Detailed physical examinations and functional assessment The outcome of e.g. gait changes will be recorded according to a specific SOP. In addition, “… animals will also be assessed for gait disturbances using Accuplacer treadmill equipment.” Is this examination procedure often performed in the test facility? Were positive control studies ever been done to assure sensitivity and reliability of these examinations? How often during the course of the study will this be done? Is it part of the DCO? Please specify. 15. Lines 352-‐359, Ophthalmologic examination (plus Appendix, attachment 1) Under the heading “OBSERVATIONS (CHRONIC TOXICITY PHASE)” paragraph 38 of OECD TG 453 indicates these exams, “…should be carried out on all animals prior to the first administration of the test substance”, and, “At the termination of the study…” (OECD, 2009). Please clearly indicate in the Study Plan when the initial and final exams will be conducted for the chronic toxicity phase animals. OECD TG 453 does not indicate that ophthalmologic examination is a requirement for the carcinogenicity phase of the study. Therefore, such evaluations are unnecessary. As currently written, this section of the Study Plan is unclear and could be misinterpreted by readers to indicate
82
all animals will receive an ophthalmologic exam at the end of the study. However, if it is G-‐TwYST’s intention to evaluate all animals at all endpoints, this is unnecessary. By eliminating the ophthalmic exams on the carcinogenicity phase 1000 unnecessary ophthalmic exams (500 animals evaluated twice) could be omitted to provide substantial savings, and those funds could be used to enhance the robustness of other aspects of the study design. 16. Lines 360-‐364, Body weight Recommend adding a body weight collection interval “immediately prior to randomization” to have a better understanding of body weight at the study start, and refining the “at the end of the study” interval to indicate unfasted body weights the day prior to necropsy and fasted body weights the day of necropsy. Unfasted body weights are a better comparator for other body weight intervals on the study and fasted body weights are necessary for the proper calculation of organ-‐to-‐body-‐weight ratios if desired (or applicable, Page 17 of OECD TG 453). 17. Lines 365-‐373, Procedures For Sample Collection This section indicates a single blood sample will be divided for hematology and clinical chemistry analyses (Lines 367-‐368). This is highly unusual as optimal samples for hematological and clinical chemistry analyses are conducted on fundamentally different biological matrices (whole blood and serum, typically). However, it could be a misstatement because the document later indicates hematology samples will be collected in the presence of EDTA (Lines 401-‐402) and clinical chemistry samples will be collected in the absence of an anticoagulant (Lines 420-‐426). The collection methods indicated in Lines 401-‐402 and 420-‐426 are consistent with typical sample collections for hematology and clinical chemistry of serum samples on other rodent feeding studies. CLI recommends clarifying Lines 367-‐368 to state that separate blood samples will be collected for hematological and clinical chemistry analysis. The information in this section also begs the question of how a plasma sample will be obtained for the sample indicated in Line 370. Plasma is a distinct biological matrix from serum (the presumed matrix of the clinical chemistry analysis), and should be collected in the presence of anticoagulants. The EDTA-‐treated sample collected for hematology analysis may be suitable for this purpose, but the handling and processing of the samples for each endpoint should be clearly described in the Study Plan to minimize the potential for confusion among technical staff and stakeholders. 18. Lines 399-‐400, Hematology Per OECD TG 453, “Measurements at 3 months, either in rodents or non-‐rodents, need not be conducted if no effect was seen on haematological parameters in a previous 90 day study carried out at comparable dose levels.” As Hammond et al., (2004) do not report any adverse effects on hematological endpoints from the feeding of NK603 it is not necessary to conduct these evaluations at this interval. Similarly, hematology analysis at the end of the carcinogenicity phase is likely to be confounded by geriatric changes and tumor formation. Consequently, this endpoint is frequently considered optional. Furthermore, the latest endpoint in the HC data available from Harlan is >70 weeks. As the carcinogenicity phase of this study is scheduled to run 104 weeks, and this precise interval is not given in the HC database, it is uncertain that suitable hematology HC data will be available to evaluate differences in these endpoints. For the reasons indicated above G-‐TwYST should carefully consider the conduct of hematology analyses at two unnecessary intervals (3 month and the end of the carcinogenicity phase).
83
Eliminating 1200 unnecessary hematology evaluations (700 evaluations at 3 months and 500 evaluations at 2 years) could provide substantial savings and those funds could be used to enhance the robustness of other aspects of the study design. Generally, hematology, clinical chemistry and urinalysis should not be performed at the very end of the 2-‐year study. These parameters are as critical as the organ weights which will not be determined "...since geriatric changes and the development of tumours will confound the usefulness of organ weight data..." (stated in line 487-‐489). Geriatric changes and tumors certainly confound not only organ weights, but clinical pathology endpoints as well. Therefore, applicable OECD TGs do not mandatorily require the examination of these parameters, EPA-‐ and EU-‐specific guidelines as well as Japanese guidelines do generally not require these examinations after 2 years either. Several publications dealing with this matter have also come to the same conclusion regarding the usefulness of organ weight and clinical pathology data at the end of a 2-‐year study (Weingand et al., 1992; Long and Symanowski, 1998; Young et al., 2011). 19. Lines 420-‐422, Clinical chemistry Per OECD TG 453, “Measurements at 3 months, either in rodents or nonrodents, need not be conducted if no effect was seen on clinical biochemistry parameters in a previous 90 day study carried out at comparable dose levels.” As Hammond et al., (2004) do not report any adverse effects on clinical chemistry endpoints from the feeding of NK603 it is not necessary to conduct these evaluations at this interval. Likewise, clinical chemistry analysis at the end of the carcinogenicity phase is likely to be confounded by geriatric changes and tumor formation. Consequently, this endpoint is frequently considered optional. Furthermore, the latest endpoint in the HC data available from Harlan is >70 weeks. As the carcinogenicity phase of this study is scheduled to run 104 weeks, and this precise interval is not given in the HC database, it is uncertain that suitable clinical chemistry HC data will be available to evaluate differences in these endpoints. Also, as mentioned in the hematology section, most regulatory agencies and subject matter experts recommend that clinical pathology endpoints not be included at the end of a 2-‐year study due to the confounding affects of geriatric changes and tumors. For these reasons G-‐TwYST should examine whether the financial resources committed to the conduct of clinical chemistry at the end of the carcinogenicity phase could be better utilized to improve the robustness of other study parameters or design elements. For the reasons indicated above G-‐TwYST should carefully consider whether conducting clinical chemistry analyses at two unnecessary intervals (3 month and the end of the carcinogenicity phase) is appropriate. Eliminating 1200 unnecessary clinical chemistry evaluations (700 evaluations at 3 months and 500 evaluations at 2 years) could provide substantial savings and those funds could be used to enhance the robustness of other aspects of the study design. 20. Line 436, Clinical chemistry The diagnostic utility of gamma-‐glutamyl transpeptidase (GGT) in rat studies has been characterized as “limited” (Ennulat et al., 2010), and as a result CLI recommends that another biomarker suitable for hepatobiliary evaluation be used instead. OECD TG 453 indicates other biomarkers such as 5'-‐nucleotidase, total bilirubin, total bile acids are suitable for this purpose. 21. Lines 443-‐445, In addition The analysis of 17β-‐estradiol, testosterone, T3, and T4 are also indicated as endpoints. These are not requirements of OECD TG 453, and as such are unlikely to be evaluated on other toxicology studies
84
conducted according to this guideline. Accordingly, it is unlikely that there is robust HC data available for these endpoints in studies of this length, particularly at the later time points. Additionally, because hormone levels are known to fluctuate with the female cycle, evaluating potential differences in hormone levels between groups is further complicated and may be prone to misinterpretation without sufficient HC data. Accurately interpreting the relevance of any differences detected will be difficult at best. Presuming the availability of suitable HC data for these endpoints at earlier time points (i.e., perhaps the 6 month interval) CLI recommends limiting the analysis of these endpoints to those intervals with suitable HC data. 22. Lines 446-‐448, Urinalysis Per OECD TG 453, “Measurements at 3 months need not be conducted if no effect was seen on urinalysis in a previous 90 day study carried out at comparable dose levels.” Because Hammond et al., (2004) did not report any adverse effects on urinalysis endpoints from the feeding of NK603 it is not necessary to conduct these evaluations at this interval. Correspondingly, urinalysis at the end of the carcinogenicity phase is likely to be confounded by geriatric changes and tumor formation. Consequently, this endpoint is frequently considered optional. Furthermore, the latest endpoint in the HC data available from Harlan is >70 weeks. As the carcinogenicity phase of this study is scheduled to run 104 weeks, and this precise interval is not given in the HC database, it is uncertain that suitable urinalysis HC data will be available to evaluate differences in these endpoints. Also, as mentioned in the hematology and clinical chemistry sections, most regulatory agencies and subject matter experts recommend that clinical pathology endpoints not be included at the end of a 2-‐year study due to the confounding affects of geriatric changes and tumors. For the reasons indicated above G-‐TwYST should carefully consider whether conducting urinalysis at two unnecessary intervals (3 month and the end of the carcinogenicity phase) is appropriate. Removing 1200 unnecessary urinalysis evaluations (700 evaluations at 3 months and 500 evaluations at 2 years) could provide substantial savings and those funds could be used to enhance the robustness of other aspects of the study design. 23. Lines 470-‐471, Gross necropsy CLI recommends that a supervising pathologist be at both scheduled necropsy intervals (i.e., chronic toxicity and Carcinogenicity phases). Perhaps this was the intent, but as currently written this could be interpreted to mean only the last necropsy interval will include a pathologist. 24. Lines 482 and 484, Gross necropsy Weights of the sternum with bone marrow and the thymus (likely to be involuted and hardly traceable at the indicated time points) are not indicated in OECD TG 453, and as such are unlikely to be evaluated on other toxicology studies conducted according to this guideline. Accordingly, it is unlikely that there is robust HC data available for these endpoints in studies of this length. Thus, interpreting the relevance of any differences detected in these endpoints will be difficult. For these reasons CLI recommends omitting these endpoints. 25. Lines 495-‐543, Tissue for histopathological examination Certain tissues are not optimally preserved by immediate preservation in 10% neutral buffered formalin. For example, testes, epididymides, and, eyes are commonly preserved in Davidson’s solution first and subsequently transferred to 10% neutral buffered formalin after the initial fixation
85
is achieved. This enables the production of optimal sections for evaluation of these tissues and CLI recommends following these procedures as well. 26. Lines 573-‐597, Data Evaluation and Statistical Analysis The design as currently proposed does not provide a basis for estimating the level of natural variation across the population of crop varieties. This being the case, it is difficult to see the value of including some form of formal equivalence test based on pre-‐specified effect sizes as this would not tell us anything more than can be deduced from the proposed difference test for which results are expressed as point estimates and confidence intervals. That is, if the effect size of interest is x, and the upper confidence interval for the difference is less than x, then equivalence can be concluded. (While there is arguably an issue of “sidedness” here between difference tests and equivalence tests, this should not give cause for concern provided results are interpreted pragmatically, especially bearing in mind that, in feeding studies, toxicologists routinely consider the potential relevance of all results regardless of their statistical significance). If equivalence testing is to be included then the pre-‐specified effect sizes should be stated in the protocol. Although the treatment list has yet to be finalized, it is highly likely that there will be more than two treatments. This being the case, there is a need to consider the issue of multiplicity and whether, for example, some form of multiple comparison procedure should be considered in place of standard t-‐tests. There is also a need to consider whether testing the significance of specific treatment differences should be dependent on first obtaining a statistically significant F-‐test for the overall treatment effect. Provided that the comparisons of interest are specified a priori, then arguably the most appropriate course of action would be to disregard the overall F-‐test and focus directly on the comparisons of interest. Industry statisticians and toxicologists remain skeptical of the value of performing statistical analysis based on standardized effect sizes in addition to the analyses performed on the natural scale. There are several reasons for this. Firstly, there is no relationship between SES and toxicological relevance. Secondly, SES does not provide a basis for comparison with historical ranges. Thirdly, all endpoints arising from the same experimental design will appear equally sensitive regardless of the differences in underlying levels of variability. 27. Lines 583-‐584, DATA EVALUATION AND STATISTICAL ANALYSIS It is established convention among toxicologists to evaluate males and females separately due to known differences in body weights, organ weights, reproductive and accessory organ anatomy, and some clinical pathology parameter differences (e.g., hormone levels). 28. Lines 595-‐597, DATA EVALUATION AND STATISTICAL ANALYSIS In the event that the Study Pathologist requests statistical analyses of the histopathology results CLI recommends a formal amendment to the Study Plan as detailed in Lines 102-‐108. 29. Supplimentary Information –Omics
While growing the GM and comparator crops in separate (albeit adjacent) fields may be an acceptable compromise as far as generating the feed material is concerned, it does not provide an appropriate basis for comparative –omics analysis. The absence of randomization means that there is no guarantee that any differences observed would actually be due to the genetic modification and not to the different growing conditions. Moreover, the fact that –omics responses are known to be
86
sensitive to small differences in growing conditions means that the risk of such bias is high. Indeed, many researchers, including Ricroch et al., (2013) have discussed that –omics comparisons have revealed that genetic modifications have a lesser impact on gene expression and composition than do conventional plant breeding. Likewise, at this point in time, the proposed –omics technologies are not useful for understanding potential health effects in humans or other animals; primarily, because –omics technologies have not been validated for diagnostic purposes. Relying on relatively new, and un-‐validated, –omics technologies to determine the absence or presence of unintended adverse effects will not substantially improve hazard identification or risk assessment. Indeed, we must first understand the normal natural variation around –omics endpoints. Once this is understood they might be applied to the routine safety evaluation of biotech crops if they add meaningfully to the hazard identification process. At the present time –omics profiling studies are highly heterogeneous (Ricroch, 2013), and should be standardized and independently validated to reach sound conclusions regarding their ability to detect relevant effects (Blankenburg et al., 2009). Lastly, a fundamental flaw with using –omics technologies for diagnostic purposes is that there is no specific hypothesis to test, which ultimately leads to a bias toward false positive results (Chassy 2010). References Alberts, B., Johnson, A., Lewis, J., Raff, M., Roberts, K., and Walter, P. 2002. DNA and Chromosomes. In Molecular Biology of the Cell. 4th edition. New York: Garland Science. Available from: http://www.ncbi.nlm.nih.gov/books/NBK21054/. Betton, G., Cockburn, A., Harper, E., Hopkins, J., Illing, P., Lumley, C., Connors, T. 1994. A critical review of the optimum duration of chronic rodent testing for the determination of non-‐tumourigenic toxic potential: a report by the BTS working party on duration of toxicity testing. Human Exp. Toxicol. 13, 221–232. Christensen, J., Litherland, K., Faller, T., van de Kerkhof, E., Natt, F., Hunziker, J., Krauser, J., Swart, P. 2013. ADME-‐properties of unformulated [3H]-‐siRNAs. Drug Metab Dispos; published ahead of print March 22, 2013. http://dmd.aspetjournals.org/content/early/2013/03/22/dmd.112.050666.abstract.html Codex. 2009. Codex Alimentarius Commission of the Joint FAO/WHO Food Standard Program. Foods Derived from Biotechnology, 2nd Edition. EFSA. 2003. Opinion of the Scientific Panel on Genetically Modified Organisms on a request from the Commission related to the safety of foods and food ingredients derived from herbicide-‐tolerant genetically modified maize NK603, for which a request for placing on the market was submitted under Article 4 of the Novel Food Regulation (EC) No 258/97 by Monsanto (QUESTION NO EFSA-‐Q-‐2003-‐002). EFSA J. 9:1-‐14. EFSA. 2008. Safety and nutritional assessment of GE plants and derived food and feed: The role or animal feeding trials. Report of the EFSA GMO Panel Working Group on Animal Feeding Trials. Food Chem. Tox. 46, S2-‐S70.
87
EFSA. 2009a. Applications (references EFSA-‐GMO-‐NL-‐2005-‐22, EFSA-‐GMO-‐RX-‐NK603) for the placing on the market of the genetically modified glyphosate tolerant maize NK603 for cultivation, food and feed uses, import and processing and for renewal of the authorisation of maize NK603 as existing products, both under Regulation (EC) No 1829/2003 from Monsanto. Scientific Opinion of the Panel on Genetically Modified Organisms. EFSA J 1137: 1-‐50. EFSA. 2009b. Applications (EFSA-‐GMO-‐RX-‐MON810) for renewal of authorisation for the continued marketing of (1) existing food and food ingredients produced from genetically modified insect resistant maize MON810; (2) feed consisting of and/or containing maize MON810, including the use of seed for cultivation; and of (3) food and feed additives, and feed materials produced from maize MON810, all under Regulation (EC) No 1829/2003 from Monsanto. EFSA J 1149: 1-‐84. EFSA. 2011. Guidance for risk assessment of food and feed from genetically modified plants. EFSA J 9:2150. EFSA. 2014. Explanatory statement for the applicability of the Guidance of the EFSA Scientific Committee on conducting repeated-‐dose 90-‐day oral toxicity study in rodents on whole food/feed for GMO risk assessment. EFSA J 12:3871 Ennulat, D., Magid-‐Slav, M., Rehm, S., and Tatsuoka, K.S. 2010. Diagnostic Performance of Traditional Hepatobiliary Biomarkers of Drug-‐Induced Liver Injury in the Rat. Tox Sci 116: 397-‐412. Hammond, B.G., Vicini, J.L. , Hartnell, G.F., Naylor, M.W., Knight, C.D., Robinson, E.H., Fuchs, R.L., and Padgette, S.R. 1996. The feeding value of soybeans fed to rats, chicken, catfish and dairy cattle is not altered by genetic incorporation of glyphosate tolerance. J Nutr 126:717-‐727. Hammond, B.G., Dudek, R., Lemen, J., and Nemeth, M. 2004. Results of a 13 week safety assurance study with rats fed grain from glyphosate tolerant corn. Food Chem Tox 42:1003-‐1014. Hammond, B.G., Dudek, R., Lemen, J., and Nemeth, M. 2006. Results of a 90-‐day safety assurance study with rats fed grain from corn borer-‐protected corn. Food Chem Tox 44:1092-‐1099. Harrison, L.A., Bailey, M.R., Naylor, M.W., Ream, J.E., Hammond, B.G., Nida, D.L., Burnette, B.L., Nickson, T.E., Mitsky, T.A., Taylor, M.L., Fuchs, R.L., and Padgette, S.R. 1996. The expressed protein in glyphosate-‐tolerant soybean, 5-‐enolpyruvylshikimate-‐3-‐phosphate synthase from Agrobacterium sp. strain CP4, is rapidly digested in vitro and is not toxic to acutely gavaged mice. Journal of Nutrition 126:728-‐740. Jonas, D.A., Elmfada, I., Engel, K.H., Heller, K.J., Kozianowski, G., Konig, A., Muller, D., Narbonne, J.F., Wackernagel, W., and Kleiner J. 2001. Safety Considerations of DNA in Food, Ann Nutr Metab 45: 235-‐54 Kemper, R.A., Hayes, J.R., and Bogdanffy, M.S. 2008. Metabolism: A Determinant of Toxicity. In C.D. Klaassen (ed.), Casarett and Doull’s Toxicology (pp 103-‐168). New York: McGraw Hill Medical. Klaunig, J.E. and Kamendulis, L.M. 2008. Chemical Carciogenesis. In C.D. Klaassen (ed.), Casarett and Doull’s Toxicology (pp 329-‐379). New York: McGraw Hill Medical.
88
Kroes, R., Renwick, A.G., Cheeseman, M., Kleiner, J., Mangelsdorf, I., Piersma, A., Schilter, B., Schlatter, J., van Schothorst, F., Vos, J.G., Wurtzen, G., 2004. Structure-‐based thresholds of toxicological concern (TTC): guidance for application to substances present at low levels in the diet. Food Chem. Toxicol. 42, 65–83. Libonati, M. and Sorrentino, S. 1992. Revisiting the action of bovine ribonuclease A and pancreatic-‐type ribonucleases on double-‐stranded RNA. Mol Cell Biochem 117: 139-‐151. Libonati, M. and Sorrentino, S. 2001. Degradation of double-‐stranded RNA by mammalian pancreatic-‐type ribonucleases. Methods Enzymol 341: 234-‐248. Lodish H., Berk A., Zipursky S.L., et al. 2004. Protein Structure and Function. In Molecular Cell Biology. 5th edition (pp 59-‐72). New York: W. H. Freeman. Long GG, Symanowski JT. 1998. Appropriate parameters to be tested in rodent oncogenicity studies. Toxicol Pathol 26:319-‐20 Munro, I.C., Ford, R.A., Kennepohl, E., Sprenger, J.G., 1996. Correlation of structural class with no-‐observed-‐effect levels: a proposal for establishing a threshold of concern. Food Chem. Toxicol. 34, 829–867. OECD. 2009. Chronic Toxicity Studies. Test No. 452, Organization of Economic Cooperation and Development. Paris, France OECD. 2009. Combined Chronic Toxicity\Carcinogenicity Studies. Test No. 453, Organization of Economic Cooperation and Development. Paris, France O'neill, M.J., Bourre, L., Melgar, S., and O'driscoll, C.M.. 2011. Intestinal delivery of non-‐viral gene therapeutics: physiological barriers and preclinical models. Drug Discov Today 16: 203-‐218. Park, N.J., Li, Y., Yu, T., Brinkman, B.M., and Wong, D.T. 2006. Characterization of RNA in saliva. Clin Chem 52: 988-‐994. Parrott, W., Chassy, B. , Ligon, J., Meyer, L. , Petrick, J., Zhou, J., Herman, R., Delaney, B., and Levine, M. 2010. Application of food and feed safety assessment principles to evaluate transgenic approaches to gene modulation in crops. Food Chem Toxicol 48: 1773-‐1790. Ricroch, A.E. 2013. Assessment of GE food safety using ‘-‐omics’ techniques and long-‐term animal feeding studies. New Biotechnol 30: 349-‐354. Sakamoto, Y., Tada, Y. , Fukumori, N., Tayama, K., Ando, H., Takahashi, H., Kubo, Y. Nagasawa, A. Yano, N., Yuzawa, K. Ogata, A., and Kamimura, H. 2007. [A 52-‐week feeding study of genetically modified soybeans in F344 rats]. Shokuhin Eiseigaku Zasshi 48:41-‐50. Sakamoto, Y., Tada, Y. , Fukumori, N., Tayama, K., Ando, H., Takahashi, H., Kubo, Y. Nagasawa, A. Yano, N., Yuzawa, K. Ogata, A.2008. [A 104-‐week feeding study of genetically modified soybeans in F344 rats]. Shokuhin Eiseigaku Zasshi 49:272-‐282.
89
Snell, C., Bernheim, A., Bergé, J.-‐B., Kuntz, M., Pascal, G., Paris, A., Ricroch, A.E., 2012. Assessment of the health impact of GM plant diets in long-‐term and multigenerational animal feeding trials: A literature review. Food Chem. Toxicol. 50, 1134-‐1148. Weingand K, Bloom J, Carakostas M, Hall R, Helfrich M, Latimer K, Levine B, Neptun D, Rebar A, Stitzel K, et al. 1992. Clinical pathology testing recommendations for nonclinical toxicity and safety studies. AACC-‐DACC/ASVCP Joint Task Force. Toxicol Pathol.20:539-‐43. Yakovlev, G.I., Sorrentino, S., Moiseyev, G.P., and Libonati, M. 1995. Double-‐stranded RNA: the variables controlling its degradation by RNases. Nucleic Acids Symp Ser: 106-‐108. Young JK, Hall RL, O'Brien P, Strauss V, Vahle JL. 2011. Best practices for clinical pathology testing in carcinogenicity studies. Toxicol Pathol. 39:429-‐34. Zeljenková, D., Ambrušová, K., Bartušová, M., Kebis, A., Kovrižnych, J., Krivošíková, Z., Kuricová, M., Líšková, A., Rollerová, E., Spustová, V., Szabová, E., Tulinská, J., Wimmerová, S., Levkut, M., Révajová, V., Ševcˇíková, Z., Schmidt, K., Schmidtke, J., La Paz, J.L., Corujo, M., Pla, M., Kleter, G.A., Kok, E.J., Sharbati, J., Hanisch, C., Einspanier, R., Adel-‐Patient, K., Wal, J.M., Spök, A., Pöting, A., Kohl, C., Wilhelm, R., Schiemann, J., and Steinberg, P. 2014. Ninety‑day oral toxicity studies on two genetically modified maize MON810 varieties in Wistar Han RCC rats (EU 7th Framework Programme project GRACE). Arch Toxicol 88:2289–2314.