2.1 Mode of action as a guidance parameter for risk quantification
(1) Information on the predominant mode of action or the predominant modes of action of the observed carcinogenic effect of a substance is useful both for determining the point of departure (Section 3) and for ex-trapolation into the low risk range (Section 5). For this purpose, the fol-lowing factors must be characterised: a) the type of possible genotoxic effects, b) the type of non-genotoxic events as impact parameters on the multifactorial process of carcinogenicity, and c) the respective impor-tance of these factors for the mode of action of carcinogenicity and the uncertainty of the relevant conclusion. The results must be documented in an appropriate way (Section 8).
2.2 Primary and secondary genotoxicity
(1) It must be examined whether direct interaction of the substance with the genetic material is substantiated or to be assumed based on other infor-mation. Secondary genotoxicity (e.g. via oxidative stress, interference with the mitotic process, inhibition of topoisomerase, inhibition of the DNA repair enzymes, etc.) is to be distinguished from primary genotoxi-city (direct/indirect: DNA interaction, adduct formation and mutations caused by the parent substance or metabolites).
(2) The quality and verification of the assessment of genotoxic properties must be characterised (differentiation according to in vivo/in vitro fin-dings, compatibility of the available study results, impact of the dose range in the available test and information about gaps).
(3) Information on genotoxicity (type of genotoxicity and quality and
verifi-cation of the findings) can be essential for the specificity on the target
organ in which tumourigenicity was observed. For some forms of
genotoxicity (e.g. aneuploidies), minimum concentrations of dangerous
substances that are required to cause cancer can be assumed.
In the assessment of genotoxicity tests, it must be considered that up to 80% of the substances that are negative in carcinogenicity tests in rodents are positive in one or several in vitro tests. This applies particularly to chromosome aberration tests, micro-nucleus tests and the mouse lymphoma test. Depending on the test system used and the class of substances, there are numerous reasons why in vitro results cannot be transferred to the in vivo situation; some of them are listed below by way of example:
- the use of high concentrations that overload metabolic detoxification mechanisms, - absence of phase II enzymes and their cofactors in the test system,
- test system with DNA repair deficiency (all Salmonella strains and E. coli),
- test system without or with abnormal expression of p53 protein (CHO cells, L5178Y cells and V79 cells), and
- effects with a threshold that is not reached in vivo: aneuploidy, inhibition of DNA po-lymerase, of topoisomerases or kinases, cytotoxicity or pH change.
Transferability to humans is furthermore restricted through a use of rat-specific meta-bolic activation that does not reflect the pattern of activating enzymes metabolising xenobiotics in humans (Kirkland et al. 2007a). However, it is possible that activation in the organism is not reproduced in standard in vitro tests, e.g. if the substance is acti-vated via sulfotransferases and false negative results are therefore obtained (Kirkland et al. 2007b).
The relevance of in vitro genotoxicity test results must therefore be examined on the basis of the conditions used in the tests (e.g. comparison of the dose-response rela-tionships of genotoxicity and cytotoxicity and high dose effects) and of the structure of the tested substance to decide whether a carcinogenic substance is primarily genotoxic. If necessary, structure-effect relationships should be included. In unclear cases, the results of valid in vivo tests are decisive for systemically acting carcino-gens. For locally acting carcinogens, negative in vivo tests are conclusive only if it has been demonstrated that the target organ was reached.
2.3 Non-genotoxic events
(1) Information on non-genotoxic effects with a potentially causal impact on the process of carcinogenicity must be recorded and described and the dose range determined must be compared with the carcinogenic doses.
This mainly includes cytotoxicity (e.g. irritation, inflammation and necro-sis), induced cell proliferation, toxicokinetic information (e.g. enzyme in-duction, saturation or new metabolites typical of high doses), receptor-mediated processes, protein binding, direct hormonal effect, indirect im-pact on hormonal feedback systems, organ specificity and sex specific-ity.
(2) The quality and reliability of the assessment of non-genotoxic properties must be characterised (differentiation according to in vivo/in vitro find-ings, compatibility of the available study results, impact of the dose range in the available test or information about gaps).
(3) Information on non-genotoxic events (type of effect and quality and
reli-ability of the findings) must be specified particularly for its relevance in
the target organ in which tumourigenicity was observed.
2.4 Relevance of different impacts in a multifactorial process
(1) According to a weight-of-evidence approach, the relevance of primary and/or secondary genotoxicity (see Section 2.2) and of non-genotoxic events (see Section 2.3) to the process of carcinogenicity must be as-sessed. The central factor(s) of impact on cancer is (are) to be described and its (their) assumed relevance to humans substantiated.
(2) A distinction of the assumed modes of actions differentiated according to tumour localisation and/or dose range may also be a result. The exis-tence of several (possible) modes of action must be identified.
(3) The occurrence of pre-malignant effects (like the formation of foci in the liver) must be examined and their dose-response relationship described, if possible.
(4) Background rates and the occurrence of spontaneous tumours in the control group are to be assigned to the discussion of the mode of action.
2.5 Targeted conclusion
(1) After all the information has been recorded, the following statements can be made:
•
Postulated mode of action
•
Key events (observed; agreement with mode of action)
•
Dose-response relationship
•
Time-related association
•
Intensity of the association; consistency of the data for this conclusion;
specificity of the association
•
Biological plausibility
•
Other possible modes of action
•
Confidence in the assessment
•
Data gaps; uncertainties
(2) The following questions must specifically be answered:
•
Is the weight of evidence sufficient to identify a mode of action in an animal study?
•
Can human relevance of the mode of action be ruled out with sufficient likelihood on the basis of fundamental qualitative differences in key events between animals and humans?
•
Can human relevance of the mode of action be ruled out with sufficient likelihood on the basis of quantitative toxicokinetics and/or toxicody-namic differences between animals and humans?
•
What is the confidence of a generated assessment (relevance)?
There may also be a threshold for genotoxic events. Genotoxic events must be differ-entiated from this point of view (see TGD, Risk Characterisation, Section 4.14.3.4;
Butterworth, 2006).
Non-genotoxic events cannot regularly be associated with a threshold either; for ex-ample, such a threshold cannot always be identified for some receptor-mediated processes (see TGD, Risk Characterisation, Section 4.14.3.3; Butterworth, 2006).
As far as data for an exposure-risk relationship in the experimental range are required to determine the relevance of the different statements, there is an interdependence between tasks according to Section 3 and tasks according to Section 2 of this Guide (in particular 2.4 and 2.5: Exposure-risk relationship). Accordingly, the items of this Guide cannot be dealt with in a strict chronological order.
The items mentioned under 2.5 are based on considerations by WHO (IPCS) and are explained in detail in Boobis et al. (2006). Examples of the procedure in the discus-sion of the mode of action can be found in Kirman et al. (2004), Cohen et al. (2003) and Preston and Williams (2005). The basic method for recording the mode of action is explained in Meek et al. (2003) and Seed et al. (2005).
In various publications (e.g. Streffer et al., 2004, Hengstler et al., 2006, Bolt and Hu-ici-Montagud, 2007 and Foth et al., 2005), the differentiations of the mode of action postulated were similar to those used as a basis for the procedure described here.
They lead to a differentiation, as is shown in Section 5.1 of this Guide.
Neumann (2006a,b,c) substantiates why it is impossible to find a definite threshold for a carcinogenic effect and recommends avoiding the term completely. However, since there are no alternatives that can be communicated better, the term will continue to be used in the present Guide with the above restrictions of its meaning.