Minutes

ITEM 1: APOLOGIES FOR ABSENCE/ANNOUNCEMENTS

2. Apologies for absence were received from COC member Professor G Williams and the assessors Mr A Browning (VMD), Dr H Stemplewski (MCA) and Mrs K Cameron (DEFRA).

Announcements

3. The Chairman welcomed Dr T England and Ms I Lindup (DH Toxicology Unit at Imperial College), Dr H Walton (DH), Dr Gupta (DH) and Dr Strickland (Cabinet Office).

4. The Chairman informed the committee that it was the last meeting for Professors Cooper, Renwick and Williams who had come to the end of their term of membership. He thanked them for their hard work and valuable contribution to the COC over the years.

5. Members were reminded of the need to make any relevant declarations of interests before discussion of items.

ITEM 2: MINUTES OF THE MEETING 21 NOVEMBER 2002 (CC/MIN/02/3)

6. The minutes were agreed subject to minor editorial changes.

ITEM 3: MATTERS ARISING

3.1 Malathion

7. The COC was informed that the joint first draft COC/COM statement on malathion had been sent to the data holder. There had been an exchange of views between the COM and the data holder particularly with regard to the adequacy of an in-vivo rat liver UDS assay. The Advisory Committee on Pesticides (ACP) had also been sent the first draft COC/COM statement on malathion and had considered that there was no regulatory need for additional testing. Members were informed that the ACP had given different weighting to the value of the carcinogenicity data in reaching their conclusion.

8. A 2nd revised draft COM statement containing additional comments, in response to the data holder's submission, had been forwarded to the data holder for comment.

3.2 Risks associated with exposure to low levels of air pollutants (CC/03/8)

9. Members suggested a number of changes to the draft statement on risks associated with exposures to low levels of carcinogenic air pollutants.

10. The Committee noted that the one-hit model represents a very conservative approach to genotoxic carcinogen risk assessment. Thus the COC agreed that the upper bound 95% CI risk estimate from such an approach would probably overestimate the actual risk associated with exposure to a genotoxic carcinogen. The Committee could support the use of this approach to carcinogenic air pollutants provided the estimates arising were not quoted as if they were real estimates of risk but were used in the consideration of risk management options. The Committee considered it important to restate that even small exposures to a genotoxic carcinogen carried a small, but unquantifiable, risk of cancer.

11. The COC therefore agreed that the proposal from the Department of Health was a useful approach to considering risk management options for carcinogenic air pollutants that are genotoxic carcinogens. The Committee agreed that it was acceptable, in this situation specific, to use the upper bound (95% CI) estimate of cancer risk at environmental exposure levels based on dose-response modelling derived from adequately performed cancer epidemiology studies and using the one-hit model.

12. The statement would be finalised by chairman's action.

3.3 Genotype-environment: Letter to Dr Zimmern (CC/03/10)

13. The COC heard that the secretariat had written to Dr Zimmern (Director of the Public Health Genetics Unit, Cambridge) with members' comments from the previous meeting on suggestions that the UK Biobank Project and proposed Genetic Research Parks be used for investigating genotype-environmental chemical exposure interactions. In response the Department of Health Research Development Division (who were also sent a copy of the letter) had suggested that members of the DH/Welcome team responsible for setting up the Biobank project should provide a short presentation to the COC at the 26 June meeting.

3.4 COM high dose in-vivo bone marrow mutagenicity

14. The COC was informed that the COM had discussed a paper outlining the evaluation and significance of apparently positive in-vivo mutagenicity results seen in the bone marrow at high doses (associated with severe toxicity and lethality) at its last meeting in February 2003. Two members of the COC had attended this discussion because of the paper's consideration of the role played by carcinogenicity data in assessing such mutagenicity data. It is intended that the COM paper will be discussed by the full COC before finalisation.

ITEM 4: ALCOHOL AND BREAST CANCER

15. The minutes of this item now published - see paragraphs 32 - 40.

ITEM 5: DRAFT COC GUIDELINES ON RISK ASSESSMENT (CC/03/5)

16. The Committee had a generic discussion of the draft COC guidelines on chemical carcinogenic risk assessment at its last meeting in November 2002. The secretariat had made a number of amendments in light of members' comments. Reference had been made to the International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals (ICH) in attaining harmonisation for hazard evaluation of pharmaceuticals. The main use of the 'Threshold of toxicological concern' would be for the prioritisation of chemicals rather than as a method for setting minimum risk levels. Proposals for research listed in the draft were those considered at the November 2002 meeting in accordance with the conclusions reached by ECETOC.

17. Members agreed that it should be made clear that the current revision would concentrate on only one aspect of the 1991 COC guidelines, namely carcinogenic risk assessment and new approaches such as 'minimum risk levels'. Therefore, there was no consideration of some other aspects of the 1991 COC guidelines such as the use of epidemiology. It was agreed that a review of the use of epidemiology in carcinogen risk assessment would be undertaken as a separate exercise. The COC considered that risk assessment was usually divided in to four stages, hazard identification (ie identification of hazard to humans), hazard characterisation (dose-response assessment), exposure assessment and risk characterisation. These stages should be referred to throughout the guidelines.

18. It was agreed that well-conducted positive carcinogenicity epidemiology studies were valuable and the most appropriate studies for the identification of human carcinogens. However, it was noted that epidemiological studies might not have sufficient power to identify the absence of a clinically relevant carcinogenic hazard. When good quality epidemiology studies were not available, carcinogens could be identified by animal bioassays. The committee recognised that there were difficulties in the dose response evaluation when extrapolating to likely human exposure levels.

19. Members considered that it should be made clear that not all carcinogens act by a genotoxic mechanism, and therefore short-term mutagenicity tests could not be expected to detect all carcinogens. The COC asked for further consideration to be given to a recent study on ED01 and the existence of thresholds. Members agreed that a distinction needed to be made between 'mode' and 'mechanism' of carcinogenic action. The latter implied a more detailed knowledge of the carcinogenic process at the molecular level.

20. The COC considered that the draft guidelines needed to make a clearer distinction between biomarkers of exposure, effect and susceptibility. The section on the risk assessment approach to non-genotoxic carcinogens should make reference to the Interdepartmental Group on Health Risks of Chemicals (IGHRC) document on uncertainty factors. Additionally, for non-genotoxic carcinogens a full toxicological review should be carried out to identify the lowest NOAEL for the most sensitive endpoint, which may not be carcinogenicity.

21. Regarding mathematical modelling of animal data, members considered that some revision was required to the draft document and that it should be made clear that the models attempted to estimate the risk of developing cancer at low doses by interpolation from the effect observed at the lowest dose to zero. Members agreed that the 'threshold of toxicological concern' used as an example of a minimum risk level had been applied only to flavours in the food additive area.

22. The secretariat would use members' comments to revise the draft COC guidelines for consideration at the next meeting.

23. Members were informed that the final draft of the EPA guidelines for the risk assessment was available and would be circulated to members for discussion at the COC meeting in June. The EPA had also published additional guidance for assessing cancer susceptibility from early life exposure to carcinogens. Members would be asked to specifically comment on this document at the June COC meeting.

ITEM 6: ORGANOCHLORINE INSECTICIDES AND BREAST CANCER; UPDATE 1999-2002 (CC/03/4)

24. There was not sufficient time to discuss this item, it would be considered at a later date.

ITEM 7: INTRAHEPATIC CHOLANGIOCARCINOMA; UPDATE FOR MEMBERS (CC/03/6)

25. Members had a very brief discussion of the submitted data and asked for more details of the research to be provided before further consideration.

ITEM 8: CARCINOGENIC POTENCY RANKING FOR AIR POLLUTANTS (CC/03/9)

26. This item was brought forward and considered during matters arising along with the statement on risks associated with exposure to low levels of air pollutants (CC/03/8).

27. The COC discussed a generic approach to ranking environmental carcinogens at the November 2000 meeting. A simplistic approach based on assigning potency to one of four bands, and scoring exposure by route (air, water, food), and by percentage of the population exposure was proposed. Members cautioned against the development of ranking schemes giving a single numerical value of ranking, which might give an unduly simplistic assessment. The committee undertook a further review of the applicability of the T25 approach to ranking carcinogenic potency at its November 2001 meeting. The COC considered that the use of the T25 in potency ranking of genotoxic carcinogens was an acceptable pragmatic approach, but that the parameter should not be overinterpreted.

28. Members were informed that Dr Schuhmacher-Wolz and colleagues (Regulatory Toxicology and Pharmacology 36, 221-233 (2002)) had evaluated the acceptability of unit risk estimates derived from epidemiology studies or animal bioassays to rank the potency of carcinogenic air pollutants. Members agreed that the proposed approach involved many assumptions and overall was not appropriate for ranking carcinogenic air pollutants.

29. The COC agreed that in the absence of good human epidemiological data that T25 could be used as a general guide to ranking the carcinogenic potency of air pollutants that were genotoxic carcinogens. When appropriate human epidemiology data were available, this should be used in preference to provide information on ranking and setting priorities for risk management purposes.

ITEM 9: ANY OTHER BUSINESS

30. Members were informed that due to the overrunning of items of business from this meeting and a number projects awaiting COC consideration that it was possible there would be a need for an additional meeting later this year.

ITEM 10: DATE OF NEXT MEETING

31. Date of next meeting: 26 June 2003.

******************************

ITEM 4: ALCOHOL AND BREAST CANCER
(FURTHER CONSIDERATION OF META-ANALYSIS UNDERTAKEN BY IMPERIAL COLLEGE)

32. The Chairman reminded members that a request for additional information had been forwarded to the Oxford Collaborative Group. A response had been provided. In addition some further analyses had been provided by the Imperial College group in response to members’ queries at the November 2002 meeting. The secretariat had prepared an update of epidemiology literature retrieved since the November 2002 meeting and a 1st draft statement. Members noted it was important to time the publication of the COC statement to occur at the same time as the publication of the results of the Imperial College meta-analysis study.

4.1 Additional Analyses from Imperial group: Meta-Analysis report (CC/03/01)

33. Members noted the additional analysis undertaken by the Imperial group to evaluate the effect of duration of follow-up in prospective studies. The odds ratio for least adjusted analyses including all studies was slightly higher for less than 10 years follow-up compared to 10 years or more of follow-up. However the opposite finding had been documented in analyses using the most adjusted analyses from studies having a quality score of two or three. Members noted that the initial finding of increased risk with a follow-up period of less that 10 years in all studies was consistent with a similar analysis reported by Ellison et al (American J of Epidemiology, vol 154, (8), 740-745, 2001). However the Committee agreed that since this observation had not been reproduced in the Imperial meta-analysis when quality scoring of studies was included, and since the ratio of odds ratios (<10 years follow-up versus >10 years follow-up) was not found to be significantly different from 1 in either less or most adjusted analyses, it was not possible to draw any conclusions on duration of follow-up.

4.2 Additional epidemiology/mechanism studies (CC/03/2)

34. Members were informed that all of the studies appended to CC/03/2 would be referenced in the statement and were asked to discuss which studies should be specifically commented upon in the statement.

35. Members considered the study by McDonald PAG et al (Cancer Causes and Control, 13, 543-548, 2002) provided limited evidence regarding an effect of drinking alcohol on mortality from breast cancer in African American women. It was possible that other life-style factors could also affect mortality as well as stage of disease at diagnosis, compliance with treatment. It was noted that there was a small published literature on the potential for drinking alcohol to modify risk of death from breast cancer. Members agreed that a review of available published data for the June COC meeting might assist in discussing the relevance of this study for the COC statement on alcohol and breast cancer, although the general view was that it was unlikely that any conclusion could be drawn. Members also raised the potential for the variation in risk of breast cancer with drinking between different ethnic groups and asked whether there was any relevant literature. The study by Bryne et al (Cancer Epidemiology, Biomarkers and Prevention, 11, 1369-1374, 2002) provided some limited evidence that early onset of drinking was associated with a higher risk of Benign Breast Disease (BBD) in 18-22 year old women. The significance of the finding by these authors was unclear as the potential for BBD to be a precursor marker for breast cancer in these women was unknown. In addition the literature on the association between drinking alcohol in premenopausal women stratified by age was inconsistent.

36. Members recognised the limitations of the studies by Britton J et al (American J of Epidemiology, 156, 507-516, 2002) and Sellers T et al (Cancer Epidemiology, Biomarkers and Prevention, 11, 1104-1107, 2002) on receptor status but agreed they should be cited as they provided evidence to show an effect of alcohol on induction of breast tumours that were Estrogen Receptor negative. It was noted that the recent update on Nurses Health Cohort study (Chen et al Annals of Internal Medicine, 137, 7798-804) provided evidence that HRT and alcohol acted independently with regard to risk of breast cancer.

37. Members agreed that the study investigating ethanol modulation of growth in human breast cancer cells in-vitro (Izevbrgie et al Experimental Biology and Medicine, vol 227, 260-265, 2002) provided no relevant information to the assessment of potential mechanisms.

4.3 Background incidence of mammary tumours in ICR mice (CC/03/7)

38. The COC considered a paper by Watabiki et al (Alcoholism, Clinical Experimental Research, vol 24, (4), 117S-122S) at the November 2002 meeting. The authors reported the first evidence that prolonged administration of ethanol to rodents (specifically ICR mice) via the drinking water could induce a tumourigenic response. All previous studies reviewed by COC had reported negative results. Members asked for data on background incidence of tumours in ICR mice. Information had been obtained from Charles River Laboratories. These data showed ICR mice had a low background incidence of mammary tumours (0.785-8.33% from 51 long-term carcinogenicity studies).

39. Members agreed that a tumourigenic response had been documented. Members expressed concerns that the high doses used (ca 10% v/v ethanol for 2 months followed by 15% v/v for up to 25 months) had resulted in nutritional imbalance and had exceeded the Maximum Tolerated Dose. It was noted that no correction for ethanol induced calorific imbalance had been undertaken in this study in contrast to other studies, previously reviewed by the COC, where ethanol had been administered in the drinking water to rodents, over most of their life span. Members agreed it was impossible to interpret the significance of this study with regard to any carcinogenic effects of ethanol.

4.4 First draft statement on alcohol drinking and breast cancer (CC/03/3)

40. Members were asked to comment on the first draft statement (CC/03/3). Members agreed that a glossary and a Lay Summary should be drafted. The Committee agreed that the most important additional work involved requesting further information from the Imperial College group regarding the reasons for the range of estimates reported and to identify the most appropriate estimate of PAR (and measure of precision or uncertainty) to cite in the COC statement. Members thanked the Oxford Collaborative Group for their written response to questions raised at the November 2002 meeting. However members felt that it was still not possible to discern the reasons for the differences between the Oxford and Imperial groups with regard to estimate of PAR. There was a need for additional analysis to compare the PAR estimates from the two research groups. It was agreed that a member of COC would contact the Imperial Research group to take these matters forward. The COC agreed it important to consider the estimate of PAR in detail, as this would have considerable implications with regard to whether a review of the sensible drinking message was required. It was hoped to finalise the COC conclusions at the June 2003 meeting. The main suggested areas for change or further consideration are given below:

i. Introduction: Clarify risk factors for breast cancer. Cite ONS data for breast cancer registrations for all UK.

ii. Background: Define what constitutes a drink and unit of alcohol in UK. Reduce repetitive references to breast cancer risk factors. Clarify what is meant by “modest relative risk of 3 which was rarely exceeded in most studies even by heavy drinkers” (particularly for lay summary).

iii. Introduction to current review: Note date for literature inclusion in COC statement will post date that of the meta-analyses reviewed.

iv. Review of new information: Clarify what was meant by “variable intercept modelling”, the dose-response results reported by the Imperial College group, specify country/region of origin in analyses of heterogeneity, and report Population Attributable Risk Estimate for UK. Clarify the range of values reported by Imperial Group and identify most appropriate result to cite in the COC statement. Note uncertainties in measuring alcohol intakes and potential for effects on PAR estimates.

v. Further DH Tox Unit paper on mechanisms: Clarify the weight of evidence regarding evidence for effects on oestrogen levels and other suggested mechanisms particularly evidence for effect on Insulin-like Growth Factor levels.

vi. Consideration of causality: Reconsider entries to Bradford-Hill criteria in light of new information. Do not interpret the case for causality solely in terms of numbers of criteria that have been met. Note dose levels of ethanol given to ICR mice in drinking water in a carcinogenicity bioassay were considerably greater than the Maximum Tolerated Dose and there was no attempt to address calorific imbalance associated with administration of very high doses of ethanol via the drinking water, and that results were uninterpretable.

vii. Significance to Public Health: Clarify which PAR estimate should be used and report estimates of number of cases potentially associated with drinking alcohol.

viii. Conclusions: Reorder conclusions and include a further conclusion on potential interaction between oral Contraceptive use or Hormone Replacement Therapy and risk of breast cancer associated with drinking alcohol.

[Post meeting note: A member of COC discussed the estimation of PAR for the UK with the Imperial College group and a number of requests for additional analyses focussing on the highest quality study group were made. A comparison between the Imperial Group and Oxford Group analyses was also requested.]

****************************

ACTIONS

ITEM	ACTION	WHO
4. Alcohol & breast cancer	Redraft statement	Secretariat
5. Revision of COC Guidelines	Redraft guidelines	Secretariat

Top

copyright: © | updated: 5 January 2005