Minutes

ITEM 1: APOLOGIES FOR ABSENCE/ANNOUNCEMENTS

2. Apologies for absence were received from the COC members Professor D Forman and Dr R Roberts and the assessors Dr H Stemplewski (MHRA) and Mrs K Cameron (DEFRA).

Announcements

3. The chairman welcomed D T England and Mr K Okona-Mensah (DH Toxicology unit at Imperial College London), Dr S Strickland (Cabinet Office), Dr S Gupta (DH) and Dr D Russell (National Assembly for Wales).

4. The Chairman congratulated Professor Alan Boobis on his OBE awarded in the recent Queen’s birthday honours.

5. Members were informed that some COC papers were now being placed on the COC website before meetings.

6. Members were reminded of the need to make any relevant declaration of interests before the discussion of items.

ITEM 2: WELCOME TO NEW MEMBERS

7. The Chairman welcomed the new members Professor Alan Boobis, Dr Phil Carthew and Dr Nicola Wallis to their first meeting.

ITEM 3: MINUTES OF THE MEETING OF 6 MARCH 2003

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

ITEM 4: MATTERS ARISING

9. There were no matters arising not on the agenda.

ITEM 5: ALCOHOL AND BREAST CANCER

10. The minutes of this item now published - see paragraphs 49 - 57.

ITEM 6: CARCINOGENIC IMPURITIES PRESENT IN THE PESTICIDE 1-METHYLCYCLOPROPENE (1-MCP) (CC/03/17)

11. Dr N Wallis declared a personal interest in this item due to her employment by a pesticide company, which was potentially a competitor to the data holder company, and was asked not to contribute to the discussion.

12. 1-Methylcyclopropene (1-MCP) is a new pesticide active ingredient being assessed by the Pesticide Safety Directorate (PSD) under EU Directive 91/414/EEC. 1-MCP blocks ethylene release from apples thus preventing over ripening and softening. Ministers recently noted the advice from the ACP on this, but asked for further advice from the COC on the carcinogenic risks posed by potential exposure to the impurities, impurity 1 and impurity 2, present in the active ingredient. In undertaking this evaluation, it has been presumed that there are sufficient data available to consider that these two impurities are genotoxic carcinogens. The COC was not asked to evaluate the active ingredient and data on this compound have not been reviewed by the committee. DEFRA had specifically asked the COC for advice on whether:

a) minimum risk levels for impurity 1 and impurity 2 can be derived.

b) exposures to impurity 1 and impurity 2 present any significant carcinogenic risk to consumers or operators.

c) maximum levels of impurity 1 and impurity 2 should be prescribed.

13. The COC had been provided with published information on the mutagenicity, carcinogenicity and metabolism of impurity 1 and impurity 2. Additional information provided included, details on the manufacture of 1-MCP, information on the formulation and product used in the UK and estimates of the potential exposure to impurities present in 1-MCP.

14. The Committee agreed that from the available evidence it was prudent to consider both impurity 1 and impurity 2 to be genotoxic carcinogens. Members felt that a pragmatic minimum risk level could be established for these two impurities after considering the available carcinogenicity data together with conditions of use and potential exposure anticipated from the fumigant use of 1-MCP for apples. However, it was noted that under likely conditions of use a very small but unquantifiable carcinogenic risk existed and therefore a policy of controlling exposures ‘to as low as reasonably practicable’ (ALARP) should apply.

15. The data holder had provided an estimation of the 5% bench mark dose (BMD) for the most sensitive carcinogenic response for impurity 1 and impurity 2, which was forestomach tumours in rats. The 5% BMD for impurity 1 was reported to be 0.6 mg/kg bw/day and 12.8 mg/kg bw/day for impurity 2. The committee agreed that an application of an uncertainty factor of 10,000 to the estimated 5% BMD resulted in estimated minimum risk levels of 0.06 µg/kg bw/day for impurity 1 and 1.28 µg/kg bw/day for impurity 2. The committee agreed that this was a conservative approach to setting minimum risk levels and that any risk posed at this level would be negligible.

16. The maximum predicted exposures were reported to be below the proposed minimum risk levels for impurity 1 by factors of 3 for operators and about 250 for consumers. For impurity 2 the predicted exposures were reported to be factors of 58 and about 5300 below the proposed minimum risk levels for operators and consumers respectively.

17. Thus the COC agreed that the answers to the above questions from DEFRA were as follows:

a) The COC concluded that pragmatic minimum risk levels equivalent to oral doses of 0.06 µg/kg bw/day and 1.28 µg/kg bw/day could be established for impurity 1 and impurity 2 respectively.

b) Maximum predicted exposures for operators and consumers would be below the proposed minimum risk levels for impurity 1 and impurity 2 by factors of about 3 to 5300. The carcinogenic risk posed by such exposure to impurity 1 and impurity 2 is considered to be negligible.

c) That maximum impurity levels for impurity 1 and impurity 2 in the pesticide 1-MCP should not result in doses exceeding the proposed minimum risk level. The COC restated that there was always a need to apply ‘as low as reasonably practicable’ (ALARP) to genotoxic carcinogens.

[Post meeting note: The data holder requested that the identity of the two impurities should be confidential after the COC meeting of the 26 June 2003. This request was agreed since it was consistent with the guidelines for openness of COT/COC/COM and had also been agreed with the pesticides Safety Directorate with regard to any PSD published information on 1-MCP].

ITEM 7: EPA DRAFT RISK ASSESSMENT GUIDELINES AND SUPPLEMENTAL DATA FOR ASSESSING SUSCEPTIBILITY FROM EARLY LIFE EXPOSURE (CC/03/20)

18. The Committee was informed that the EPA Guidelines for Carcinogen Risk Assessment had been circulated for information. These were the final version and had been discussed on a number of occasions by the COC.

19. The EPA Supplemental Guidance for assessing cancer susceptibility from early life exposure to carcinogens was a new draft document that the EPA had issued for comment and had been circulated to members after the March meeting. The USEPA had argued that conventional bioassays underestimated risks (from the dose response slopes) arising from early life exposures for genotoxic carcinogens, but not for non-genotoxic carcinogens. These data led the USEPA to propose adjustment factors of 10 (for below 2 years of age) and 3 (for ages 3-15 years). The committee’s view was requested on this and the implications, if any, for the UK approach to risk assessment for chemical carcinogens and the methodology used.

20. Members felt that it was very difficult to draw any definite conclusions from the comprehensive EPA comparisons between results with conventional carcinogen bioassays and the studies involving neonatal/ perinatal exposure. This was because of the great variability in the study designs used for the neonatal/perinatal studies. Rarely were any pharmacokinetic data available to allow a comparison to be made between the systemic doses achieved by oral exposure in the conventional assays and the parental routes frequently used in the studies involving exposure of neonates. Furthermore there were only limited consideration of the mechanism involved and the target tissues (tamoxifen was given as an example of when the approach adopted was inappropriate due to differing tissue specificity).

21. It was agreed that there was some biological plausibility to the argument of increased sensitivity in the early life stages (due to factors such as differing metabolism and cell turnover), and the analysis did provide some limited data to support this. However, this was not always the case.

22. It was pointed out that the document was concerned with the risk of cancer in later life, and whether early life stage exposure made this more likely. It did not address childhood cancer, only the lifetime risk of cancer following early life exposure.

23. As the COC does not recommend the use of slope of the dose response from animal bioassays to calculate human cancer risks (and the estimation of the tolerable exposure levels) it was agreed that the adjustment factors being proposed by the US EPA were not relevant to the UK. In the UK a risk management approach of reducing exposures to as low as reasonably practical for all age groups is adopted for genotoxic carcinogens.

24. The Committee considered that in some instances the data suggested that animal models that include perinatal exposure may be more sensitive; these difference were quantitative, and there was no evidence that the use of conventional cancer bioassays in animal would fail to detect chemical carcinogens.

25. Members agreed that these data supplied by the USEPA had very limited implications for the way carcinogenicity assessment was carried out in the UK.

ITEM 8: COC DRAFT GUDELINES ON CARCINOGEN RISK ASSESSMENT (CC/03/21 & 23)

8.1 Evaluation of dose-response at low dose levels (ED 01)

26. At the March 2003 meeting, members briefly considered data on dose response for carcinogens at low doses. The secretariat agreed to provide a further paper on the subject. Paper CC/03/23 examined the available data reported from the ED01 study using 2-acetylaminofluroene (2-AAF) and subsequent reanalyses. Papers documenting dose response data for other carcinogens were also provided to members to provide further information.

27. Members were aware that an ED01 chronic study carried out by the US National Centre for Toxicological Research (NCTR) to determine the shape of the dose-tumour response curve down to a 1% increase in tumour incidence, had been recently re-evaluated. The initial study used 24,192 female BALB/c StCrlfC3Hf/Nctr mice, chosen for their suspected low level of naturally occurring bladder and hepatic tumours. The mice were exposed to low doses for up to 33 months. The study included Interim sacrifices, life span and discontinued dosing.

28. The initial summary and conclusions reported that the incidence of bladder tumours dropped off sharply as the dosage of 2-AAF was reduced, suggestive of a threshold. However, the author noted that the ‘compatibility of these data with a threshold model does not constitute proof as non-threshold models also fit the data.’ In contrast, liver tumours showed a nearly linear response over the experimental dose range, which was indicative of no-threshold.

29. The committee was aware of the continuing debate and reanalysis of dose response curves from animal bioassays for genotoxic carcinogens at low doses (in particular the ED01 study). A number of investigators have attempted to evaluate the tumour incidence data by a variety of mathematical models. Most recently Waddell and colleagues (from the Department of Pharmacology and Toxicology, University of Louisville, Kentucky, USA) have used the ‘Rozman’ scale (which plots percentage of animals developing tumours against a logarithmic scale of dose in molecules/kg bw/day). Wadell and colleagues have argued that such an approach demonstrates the existence of a threshold, however other investigators have refuted this by noting that even a linear dose-response passing through the origin would appear to have a threshold using the ‘Rozman’ scale. Wadell and colleagues had replied by stating that the method of plotting a linear scale with a forced fit through the origin had the disadvantage of truncating and compressing at very low doses. Therefore, they argued, the shape of the curve at very low doses could not be evaluated using such an approach.

30. Members considered that these data and various mathematical models could not prove or disprove the existence of a threshold. This was because of the assumptions that had been made in the mathematical analyses and that there were insufficient animal numbers to provide the necessary statistical power to identify a threshold, even if one existed. Additionally, it was noted that there was evidence that DNA adduct formation with genotoxic carcinogens was linear down to the lowest measurable dose levels. Therefore, members concluded that it was prudent to assume no threshold for genotoxic carcinogens unless there is sufficient mechanistic data in specific cases to demonstrate otherwise.

8.2 2nd Draft COC guidelines (CC/03/21)

32. A major revision concerned a section on biomarkers of exposure, effect and susceptibility. Some initial comments had been received which suggested that the new section on biomarkers was too long and that the section should be split, as the sections on biomarkers of effect and susceptibility did not fit well in the exposure section of the guidelines.

33. The risk characterisation section had been extensively revised. The section reconfirmed the different approaches to genotoxic and non-genotoxic carcinogens.

34. Members agreed with the approach set out in the flow diagrams. The committee also endorsed the approach advocated for the risk assessment of genotoxic carcinogen impurities/contaminants. This involved a requirement to set the level as low as reasonably practicable (ALARP), with the possibility of identifying minimum risk levels in certain specific situations, with the use of expert judgement, but the ALARP principle would always apply.

35. Members would provide the secretariat with any comments they might have after the meeting. It was hoped that after further revision, taking in to account member’s comments, the secretariat would be in a position to seek wider consultation on the draft document.

ITEM 9: INTRAHEPATIC CHOLANGIOCARCINOMA; ADDITIONAL DATA FROM PROFESSOR HOWARD THOMAS (IMPERIAL COLLEGE) (CC/03/19)

36. The COC had previously evaluated an observed increase in the incidence of intrahepatic cholangiocarcinoma in both England and Wales and the US over the last 30 years. It was concluded that diagnostic transfer could have contributed to this increase.

37. The secretariat reminded members that it had previously held a meeting with Professor Thomas’ group at St Mary’s, which discussed a possible case-notes study of intrahepatic cholangiocarcinoma (IHCC) cases from the early 1990s. This had been proposed to help address the question of whether the recorded increase in IHCC cases in recent years was due to an increasing incidence of IHCC or to diagnostic transfer and/or better ascertainment. Professor Thomas had informed the secretariat of other work which his group had undertaken to investigate the hypothesis that the increase in IHCC was due to chemical exposure i.e. measurements of DNA adducts and genotyping of P53 genes from cancer tissue for a mutational signature. Further details of this work were included in the present paper (CC/03/19). Members’ comments on this work could be fed back to Professor Thomas. However, the key question for the committee was whether it would support the proposed case-notes study.

38. Members considered that the evidence for an increase in the incidence of IHCC was becoming more convincing, in particular, the evidence on trend now available from several countries. The proposed case-notes study would not be useful in that it would only examine the accuracy of diagnosis as given on death certificates in the early 1990s and would not provide an answer to the possibility of diagnostic transfer. The committee concluded that it did not support the study.

39. Members commented on the published paper on DNA adducts (Khan et al, Gut (2003): 52, 586-591). This was considered to be a well-conducted study, although there was only a small number of controls. However, it was not possible to determine the cause of the adducts. They could be due to an exogenous genotoxic agent or an endogenous process, or they could be a marker of tumour development. Looking at tumour tissue and tumour-adjacent tissue from people with IHCC caused by liver fluke might help to clarify this. The Committee was unable to draw any conclusions about the relevance of the high levels of DNA adducts in tumour-adjacent tissue. It was commented that the use of cystic duct as a control tissue for comparative purposes was not appropriate. A member offered to contact the authors to obtain clarification about whether the pattern of adducts seen in the tumour-adjacent tissue were similar to those seen in the tumour tissue.

40. The unpublished paper on P53 mutations was not considered to provide any support for causation by an exogenous genotoxic agent because the mutational spectra recorded showed no reproducible pattern. The Committee considered that the possibility of an infectious aetiology as an explanation for the rise in IHCC should also be considered. A member suggested that the work of Dr Richard Thompson at King’s College Hospital on the bile duct transporter protein could be of relevance in consideration of the aetiology of IHCC.

ITEM 10: HORIZON SCANNING (CC/03/22)

41. The COC (and sister committees COT/COM) noted that as part of the agreement to follow the Office for Science and Technology (OST) guidance for scientific advisory committees it was important to undertake regular horizon scanning exercises. The covering paper (from the secretariat) outlined ways in which COC could achieve this. It is suggested that a considerable amount is already being done but that more suggestions from members would be valuable. In addition the DH Toxicology Unit had provided a paper which updated the research priorities discussion and conclusions reached by COC in 1996 and also reviewed potential areas for further work such as cancers where there was evidence for an increasing trend in incidence.

42. The chair thought that it would be valuable if the COC revisited this item for a short period at each future meeting. Members considered that other topics which might usefully be considered included dose-response at low dose levels. Members noted the appended paper to CC/03/22 from Calabrese EJ and Baldwin LA (Nature, vol 421, 691-692, 2003.) provided an argument for the occurrence of hormesis (ie the occurrence of a "U" shaped dose-response curve at low dose level). However, although it was stated that there were up to 5000 examples of the hormetic effect in the published literature, it was not possible to assess this claim on the evidence available. It was noted that in the few studies on genotoxic carcinogens using group sizes sufficient to detect effects at 1% levels or below, the evidence was generally consistent with the absence of a threshold. Also, DNA adduct formation with genotoxic carcinogens was linear down to the lowest measurable dose levels. The dose-response data for non-genotoxic carcinogens was consistent with the occurrence of a threshold. Members considered that there was no evidence available to justify the use of a hormetic approach to risk assessment for chemical carcinogens. It was agreed that the evidence for DNA repair following exposure to very low doses of genotoxic carcinogens warranted further review. Overall members felt that the arguments presented by Calabrese and Baldwin should be considered further in the future. It was agreed that there might theoretically be a point of departure in the dose-response for a genotoxic carcinogen but it was not possible to identify any potential threshold with methods available. Members considered it prudent to reaffirm that for practical purposes genotoxic carcinogens should be presumed to have no threshold.

43. Regarding other potential topics, members noted that there was to be a presentation on the Biobank research project at the September 22 meeting and that a paper on prostate cancer had been drafted for the November meeting. Members also considered a review of oesophageal cancer would be valuable. It was noted that there were a number of recent publications regarding hair dyes and potential for bladder cancer. The secretariat noted this area is being actively considered by European Commission’s scientific Committee on non- food consumer products. Members considered evaluation of potential risks of chemical induced cancer in children had received considerable interest in the scientific literature and it would be appropriate for COC to form a view on the subject.

44. The Chair asked members to retain CC/03/22 for future use and asked the secretariat to consider how the proposals could be taken forward.

ITEM 11: ANY OTHER BUSINESS

11.1 Environmental Tobacco smoke: paper by Enstrom JE and Kabat, BMJ, vol 326, 1057-1066. (CC/03/18)

45. The COC recalled that its statement on environmental tobacco smoke (ETS) and lung cancer was one of the main pieces of evidence used by the Scientific Committee on Tobacco and Health (SCOTH) in its 1997 evaluation of ETS. The conclusions of SCOTH have been important in formulating policy towards ETS. The paper from Enstrom JE and Kabat GC received widespread publicity including articles suggesting risks associated with ETS might be lower than previously thought.

46. Members considered that the design of this particular investigation using the American Cancer Society Prevention study 1 cohort (CPS1) was similar to many epidemiological investigations of passive smoking and lung cancer undertaken in the 1980s. Had this study been available in 1997 then it would have been included in the meta-analysis studies evaluated by COC on behalf of the Scientific Committee for Tobacco and Health (SCOTH). Members felt the study would not have affected the overall summary or relative risk estimate. In addition members noted the new information appended to CC/03/18 which related to a recent systematic review (Taylor R et al, Australian and New Zealand Journal of Public Health, vol 25, 203-211, 2001) and the results a study investigating metabolites of tobacco-specific lung carcinogens in non-smoking women exposed to Environmental Tobacco Smoke (Anderson K et al Journal National Cancer Institute, vol 93, 378-381, 2001). These data provided supporting evidence for the association between exposure to ETS and increased risk of lung cancer.

47. The Committee agreed a short statement regarding the data considered at the meeting subject to agreement by chairman’s action with regard to a number of minor editorial comments. It was agreed the results of the Enstrom and Kabat study reviewed at this meeting of the Committee should be used with caution. Members were concerned that some of the letters received by the British medical Journal following the publication of the Enstrom and Kabat study did not provide valid scientific criticisms of the paper. The Committee concluded that overall there was no need to undertake a further full review of the published literature or to amend the conclusion reached by COC in 1997 that passive smoking in non-smokers exposed over a substantial part of their life is associated with a 10-30% increase in risk of lung cancer.

ITEM 12: DATE OF NEXT MEETING

48. Date of next meeting 22 September 2003.

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

ITEM 5: ALCOHOL AND BREAST CANCER

2nd draft statement (with members comments), 3rd draft statement. Draft Lay summary (CC/03/15), Additional papers (CC/03/16)

49. The Committee discussed the 1st draft statement at the March 2003 meeting. Members had considered that it was important to ask the Imperial College London research group to undertake additional analysis of the best quality studies to provide a narrower confidence interval for the estimate of Population Attributable Risk and to compare the PAR estimates from the Imperial Group and the Oxford Collaborative group. The additional requests from COC were discussed with the Imperial group shortly after the March 2003 COC meeting. Some further data were provided by the Imperial group, which were circulated with the 2nd draft statement. A further revision of the statement (3rd draft Annex 1 to CC/03/15) was undertaken when the revised draft report from the Imperial group was received. This new draft includes a graphical presentation of cumulative risk (as requested by members at the March 2003 meeting).

50. Members heard that the revised PAR for the UK based on the best quality studies using 1999 cancer registration data for the UK and alcohol consumption data derived predominantly from the Health Survey for England suggested 3,190 cases/year (95% CI 2,000-4,380) could be prevented. The further analyses undertaken by the Imperial group suggested that differences in the proportion of non-drinkers in UK and level of daily alcohol consumption compared to other western countries explained most of the differences between the PAR estimates derived by the Imperial group and the Oxford Collaborative group.

51. The Committee commented on the new cumulative incidence plots and noted that the baseline reported cumulative incidence of breast cancer attributed to average intake of alcohol across the whole population of women and the potential increase in breast cancer cases attributed to drinking 1-3 drinks (assuming each contained on average 9.5 g ethanol/day). Members considered this would present a confusing message. Thus an estimate of the average intake was not reported in the additional analyses and it was therefore open to the reader to determine what was meant by average intake. It was agreed that COC epidemiologists should discuss these analyses with the Imperial group with a view to undertaking further analyses to report one of the following; cumulative incidence for non-drinkers, a specified level of alcohol consumption for average drinking, or cumulative incidence for below average consumption of alcohol by women. Members were aware that the Oxford Collaborative group had access to alcohol consumption data for individual cases of breast cancer. It would be particularly difficult for the Imperial group to calculate cumulative incidence of breast cancer for non-drinkers.

52. The Chair asked members for comments on the third draft COC statement. Members noted that the average drink size for UK consumers used in the Imperial group analysis contained 9.5 g ethanol/drink. However information regarding trends in drinking patterns taken from the General Household Survey and the Health Survey for England reported intakes in terms of units of alcohol consumed according to the definition of unit (ie 1 unit = 8 g of ethanol) used in the Department of Health report on Sensible Drinking published in 1995. Members considered that the size of some standard drinks and alcohol content, particularly wine, resulted in intakes of greater than 1 unit/drink. Members felt that the general public were aware of units of alcohol as a guide to consumption of alcoholic beverages. Members agreed that results reported by the Imperial group should be reported, at least in the Committee statement, in terms used by the authors (ie 1 drink = 9.5 g ethanol) and also units of alcohol consumption. It would be necessary to convert intakes reported by the Imperial group to grams of ethanol consumed/day in order to calculate units of alcohol consumed. The Committee agreed that the introductory sections of the statement needed clarification with regard to the reporting of alcohol consumption in terms of drinks and units. It was agreed that the section should focus on intakes amongst women and omit information on men.

53. Members asked that the sections of the statement reporting the methods used by the Imperial group reflect the terminology used in the revised draft report. Members also asked for further clarification regarding which estimates of intake had been used by the Imperial group in deriving their estimates of Population Attributable Risk (PAR). Members asked that the statement give clearer definitions of “Non-drinker” when summarising the data from both the Oxford Collaborative and Imperial group research reports.

54. The Committee welcomed the cumulative incidence graph provided for the statement but considered it was important to clarify what level of intake was reported as the baseline. The Committee agreed that the section dealing with significance for public health would need to be revised when information on “average” had been included in the graph. Members felt it would also be important to derive a conclusion regarding potential risks, which might be associated with increasing consumption of alcoholic beverages by women aged 16-24 years.

55. The Chair asked members for comments on the first draft lay summary. Members thought it important to clarify the intended audience for the lay statement. The secretariat responded that the primary role of the lay summary was to provide a non-technical explanation of the Committee rationale for coming to a decision. Members felt that although the 1st draft lay summary was easier to follow than the full statement, it still contained considerable sections of technically based information. It was noted that the secretariat had attempted to incorporate the uncertainties in the evaluation into the lay summary, which complicated the presentation of the key conclusions of the review. One suggestion was that the secretariat considers if the critical information could be summarised in one page of A4 text as opposed to the current proposal of 4-5 pages.

56. The Committee agreed to review amended statement and lay summary at the extra COC meeting arranged for 22 September 2003.

57. [Post meeting note Ms Howel and Prof Forman discussed the cumulative incidence graphs with the Imperial group and a number of revisions were made with new graphs to be submitted to the COC at the 22 September 2003 meeting.]

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

ACTIONS

ITEM	ACTION	WHO
5. Alcohol & breast cancer	Revise draft statement	Secretariat
6. Carcinogen impurities in the pesticide 1-MCP.	Finalise statement.	Secretariat
8. COC draft guidelines on carcinogen risk assessment.	Revise draft	DH Toxicology unit
11. Environmental tobacco smoke	Finalise statement	Secretariat

Top

copyright: © | updated: 24 December 2004