Position paper on Margins of Safety (MOS) in human health risk assessment expressed at the 22nd CSTEE plenary meeting, Brussels, 06/07 March 2001
Background
In the framework of Council Regulation (EEC) 793/93 on the evaluation and control of the risks of existing substances, the CSTEE is requested to give its opinion on the results of the risk assessments. The terms of reference are the following:
1. Does the CSTEE agree with the conclusions of the Risk Assessment Report?
2. If the CSTEE disagrees with such conclusions, the CSTEE is invited to elaborate on the reasons for this divergence in opinion.
In developing the present Position paper The CSTEE has focussed its attention on human health. The issues concerned with ecological risks are somewhat different, but as we shall indicate there are analogous problems that will need separate consideration.
The conclusions concerning human health consider workers, consumers and man indirectly exposed via the environment, and are based on a comparison (setting the Margin Of Safety - MOS) between the estimated exposure for each of the three mentioned population groups and the identified N(L)OAEL for those toxic effects for which information is available. According to the Technical Guidance Document (TGD) on risk assessment for existing chemicals, there are three options for the conclusions:
i. there is need for further information and/or testing
ii. there is at present no need for further information and/or testing and no need for risk reduction measures beyond those which are being applied already
iii. there is a need for limiting the risks; risk reduction measures that are already applied shall be taken into account
However, the TGD offers very little advice on how to arrive at each of the different conclusions (page 143-144):
"Where the exposure estimate is higher than or equal to the N(L)OAEL, this indicates that the substance is "of concern" with regard to the exposure of the human population considered.... Where the exposure estimate is less than the N(L)OAEL, the risk assessor will need to decide which of the possible results applies. For this step, the magnitude by which the N(L)OAEL exceeds the estimated exposure (i.e. the "margin of safety") needs to be considered taking account of the following parameters:
- the uncertainty arising, among other factors, from the variability in the experimental data and intra-and interspecies variation;
- the nature and severity of the effect;
- the differences in exposure (route, duration, frequency and pattern);
- the dose-response relationship observed;
- the overall confidence in the database
Expert judgement is required to weigh these individual parameters on a case-by-case basis. The approach used should be transparent and a justification should be provided by the assessor for the conclusion reached."
So far (7 March 2001, date of adoption of the Position Paper by the CSTEE), the CSTEE has given its opinion on 17 Risk Assessment Reports. As the calculated margins of safety (MOS values) vary from compound to compound, a principal question has been raised within the CSTEE on how the Committee should look upon low MOS values with regard to the conclusions, especially concerning the choice between conclusions ii) and iii). A working group was set up in order to identify the issues. Two short meetings were held before the plenary meetings on October 9 and November 9 in 2000, and a more extended meeting took place on January 9, 2001. There were also short discussions at plenary meetings.
The views and recommendations expressed in this Position paper need to be set in the context of the Scientific Steering Committee's (SSC) report on "the harmonisation of risk assessment procedures in the EU" and the new WHO/IPCS document "Guidance document for the use of data in development of chemical-specific adjustment factors (CSAF) for inter-species differences and human variability in dose/concentration response assessment" (CSTEE/20000/19-Add.2).
Summary of the discussions
The main question has been on how far the responsibility of the CSTEE goes in relation to the conclusions in the risk assessment reports. The CSTEE should clearly concentrate on scientific issues, and be less concerned with risk management. The wording of the conclusions, especially iii), can be interpreted as being close to a risk management issue.
There is agreement that the CSTEE should comment on the confidence in the MOS with regard to the reliability of the NOAEL and the reliability of the exposure assessment. According to the terms of reference, the CSTEE should also give an opinion on the conclusions of the risk assessment reports. The critical question then appears when there is a low calculated MOS. If the CSTEE finds the MOS to be too low, it should provide an explanation of its opinion. This should be expressed in terms of scientific concerns, but in a way that can be understood and interpreted for risk management. The CSTEE was of the view that this should be described in terms of "cause for concern", meaning "scientific concern". There should be an attempt to standardise terminology based on an expanding base of case histories in the CSTEE. Guidance will also be available in a report from the SSC on harmonisation of terms in risk assessment.
In only one of the risk assessment documents, the one on 1,4-dichlorobenzene, did the CSTEE disagree with the conclusion ii) in favour of conclusion iii), although a need for further information and testing (conclusion i) instead of ii) was recognised for some of the other substances (see the attached table). Below are citations from the CSTEE opinion on 1,4-dichlorobenzene:
GENERAL COMMENTS:
Due to a limited or not adequately conducted discussion of available toxicological data together with the fact that the calculated MOS are to small to be acceptable to this committee, conclusion (ii) of the risk assessment for consumer health is deemed not acceptable by the CSTEE.
Human Health:
Margins of safety of 13, 35, 14, 14 and 43 that are discussed on page 94 are lower than those that are usually thought to provide full protection to the public. As a consequence, it is the CSTEE understanding that they all indicate a conclusion (iii) there is a need for limiting the risk. These low values for MOS should not be accepted unless there are special, well documented and discussed reasons to do so.
SPECIFIC COMMENTS.
Human Health:
A realistic worst case scenario for inhalation exposure is set at 3,3 mg/m 3, corresponding to an oral dose of 0.69 mg/kg bw/d with a ventilation rate of 0.7 m 3/h, bodyweight of 60 kg and 75% absorption from the lungs. The lowest NOAEL for the critical effect (hepatotoxicity) is 10 mg/kg bw/d in the dog. In the mouse, hepatoxicity is also seen (but at higher doses), as well as liver carcinogenicity which could have liver toxicity and secondary proliferation as an underlying mechanism. No sound arguments are presented for discounting the endpoint liver toxicity as being relevant to man. Since a MOS value of at least 100 is generally considered to be protective, a MOS value of 14 (10/0.69) for 1,4-dichlorobenzene raises some concern.
Interestingly the MOS approach is not used in ecological risk assessment 1 .
How large should the MOS be, then, in order not to "raise concern"? The CSTEE has just started to discuss this. As already mentioned, risk assessment must be done on a case-by-case basis using expert judgement based on the scientific information. The commonly used uncertainty factor of 100 (based on the NOAEL from long-term animal studies) is thought to be a useful default value. However, it is difficult to say what is scientifically justified, both generally and in the specific case. It is agreed that a MOS of at least 100 should be used as a starting point, taking into account factors such as the type, incidence, magnitude and severity of the toxicological effect/response, the duration and administration schemes in the animal experiments, dose-response relationships, sensitive subgroups within the human population, the quality of the underlying database and the quality of the exposure assessment. As the future direction in human health risk assessment will probably involve chemical-specific and/or default assessment factors to a larger extent, this development must be followed and discussed further.
Examples of factors that would require a higher MOS than 100 are:
- the NOAEL is based on less-than-chronic exposure, while exposure has a chronic duration
- only a LOAEL but not a NOAEL was identified
- uncertainties regarding the quality of the available data. This is particularly the case in situations where the most sensitive effect detected is a lesion regarded as serious, e.g. teratogenicity or carcinogenicity, or where the statistical information is inadequate.
- there are sensitive subgroups identified within the exposed population not properly addressed in the risk assessment
- in situations where co-exposure to other chemicals with similar mechanisms of toxicity is likely
- probability for underestimation of exposure
Examples of factors that would justify a lower MOS than 100 are:
- there are documented and validated scientific data showing that for mechanistic reasons, humans are less sensitive than the experimental animals used in the critical study
- there is strong evidence that the internal target dose(s) in humans is markedly lower than in experimental animals due to e.g. metabolic or toxicokinetic differences
- the N(L)OAEL is based on human data, and these are adequate
- the human population considered is less heterogeneous, e.g. occupationally exposed workers
- the critical effect is due to local irritation
- the exposure estimate is clearly an unrealistic worst case scenario.
Consistency of the CSTEE opinions
As the CSTEE agreed that it should concentrate on the scientific data in its opinions on the risk assessment documents, and comment on the confidence in the MOS with regard to the reliability of the NOAEL and the reliability of the exposure assessment, it was decided to check the consistency between the different opinions that have been delivered. In the attached table there is a compilation of which parts of the risk assessment documents that the CSTEE has commented upon in their opinions.
Although there are relatively few compounds that have been covered this far, some inconsistencies can bee seen from the table. The exposure estimates are only seldom questioned, whereas the toxicological effects are commented upon in most cases. Sometimes a comment is given on the reliability of the NOAEL. The magnitude of the MOS has been mentioned mostly in the later opinions. The CSTEE has commented on the conclusions in all opinions. Only in one case (1,4-dichlorobenzene) was it explicitly said that the MOS was regarded as too small for conclusion ii).
It should be noted that the CSTEE, in two early opinions (on methoxy- and buthoxyethoxyethanol), did not agree on the use of assessment factors in the risk assessment report for establishing health based occupational recommended values (HBORV), which is the responsibility of the SCOEL. The CSTEE has also pointed out the limitations in the scientific basis for developing minimal margins of safety (minimal MOS) for occupational exposure that were calculated in the documents on acrylaldehyde, hydrogen fluoride and dioxane
In only two of the opinions did the CSTEE discuss possible sensitive subgroups of the population:
- dioxane : the preferential uptake of 1,4-dioxane in the milk of lactating women should lead, at least in the workplace, to measures limiting exposure of the breast feeding women
- anisidine : because young children are of concern, the CSTEE would recommend further testing regarding reproductive and developmental toxicity.
The outline of the CSTEE opinions is somewhat different, depending on the responsible CSTEE WG rapporteur.
In conclusion, the consistency of the CSTEE opinions could be improved. A difficult but important improvement would be to more carefully scrutinize the exposure assessments, which in turn would necessitate to weigh the relative importance of the exposure part of the MOS in each risk characterisation.
A further exercise to check on consistency would be to go through the risk assessment documents to check which conclusions that have been drawn in the documents from the MOS values for different population groups and different toxicological effects. There is a need to review how the MOS are being used, if they are being used consistently, the rationale behind them, and if they can be improved. However, this is a fairly big task that has been done only once by the ECB (« Margins of safety for human health. Overview of conclusions agreed at Technical meetings in 1997 ». Doc. ECB 4/34/97).
Probabilistic methods in risk assessment
Some CSTEE members have mentioned the importance of introducing new techniques in risk assessment, especially the use of probabilistic methods that are currently not recommended explicitly in the TGDs. We believe that a fundamental shortcoming of the MOS approach is that it promotes a false deterministic impression of risk assessment; that if we have an exposure concentration (x) and an effect (no-effect) concentration (y) then the MOS is precisely specified by the difference between x and y. Yet in reality both x and y are associated with variability (x +/- v and y +/- w), some of which is due to irreducible stochasticity (spatial and temporal variability in x) and some of which is due to uncertainty (extrapolations between taxa in y). The MOS approach, by using single numbers (x and y), tends to obscure the variability. However, increasingly, techniques are available which take the variability into account and which make use of it to calculate objective probabilities of y (+/- w) being exceeded (or not) by x (+/- v). Moreover, techniques are also being developed that can make explicit the "confidence" that scientists associate with various elements of risk assessment (e.g. endpoint in one species being representative of others) and that can be used in calculating subjective probabilities of x exceeding (or not) y. We believe that these probabilistic techniques need to be given serious consideration for incorporation in the EU's risk assessment procedures, as a refinement and improvement of the MOS approach. Probabilistic risk assessment is no substitute for scientific judgement, but informs that judgement by making the underlying variability more explicit.
The CSTEE procedure for reviewing the Risk assessment documents
Besides appointing a rapporteur for each document, there are « progress chasers » for the following subjects:
1. Terrestrial environment
2. Environmental exposure assessment
3. Carcinogenicity
4. Modelling
5. Water environment
6. Endocrine disruption
7. Allergenicity
Conclusions and recommendations
- In reviewing the existing substances risk assessment documents, the CSTEE should concentrate on scientific issues, including the reliability of the NOAEL and the reliability of the exposure assessment. The latter is less well covered by the CSTEE today. More attention should be given to the exposure assessment element of each report. The committee has to consider whether it has enough expertise in this area.
- In the terms of reference, the CSTEE is asked to give an opinion on the conclusions of the risk assessment documents. This means that a value judgement must be given in some cases whether a MOS is large enough not to raise any concern. The working group agrees that the estimated MOS values should be discussed on a case-by-case basis before an opinion is given. It will often be useful to start with a default value of 100.
- The CSTEE should come to a common interpretation of the standardised phrasing of conclusions ii) and iii), especially iii). The interpretation of the working group is that in case ii) the estimated exposure caused by the existing substance is so low that it raises no concern, but in case iii) the estimated exposure is so high that it raises concern. If one regards the exposure as being too high it is obvious that there is a need for limiting the risks. Which risk reduction measures are or should be applied is a risk management issue that is the responsibility of others. However, the phrase "risk reduction measures that are already applied shall be taken into account" is ambiguous. It could also be interpreted to mean that one should look scientifically at the data that are available in order to assess whether such measures are applicable and effective. The CSTEE recommends that the ambiguity of this phrase be remedied.
- The CSTEE recommends the adoption of a framework for making a judgement on the adequacy or otherwise of a particular MOS value. Parameters for this framework are set out in page 4 - 5 of this paper. The framework should be updated regularly in the light of evolving experience of the CSTEE with the Existing substances Risk assessment reports.
- The CSTEE considers that there would be value in the ECB initiating a review of all the Risk assessment reports that have been prepared to date, in order to assess how the MOS values have been used in developing the conclusions.
- The CSTEE should be proactive in the area of probabilistic methods in human and environmental risk assessment.
- The CSTEE recommends that in order to improve the consistency of its opinions on the Risk assessment reports a common outline should be applied.
- The CSTEE recommends that a procedure is adopted for regular review of the consistency in its opinions on the Risk assessment reports. This could be performed after the review of 10-15 reports.
- The CSTEE recommends that clarification be sought on the role that the committee has in discussing protection of the health of workers, and the relevance of calculating minimal MOS for occupational exposure.
- There would be a virtue in CSTEE members contributing to the revision of the TGD on risk assessment for existing chemicals.
- The CSTEE needs to consider its organisational arrangements for dealing with the risk assessments in the light of its experience and the recent changes in the committee membership.
Specific comments from the CSTEE on different parts of the Existing substances Risk assessment documents
In the table below, √ means that there is a discussion on this specific subject in the CSTEE opinion. Concerning the conclusions of the risk assessment report, a√ means that the conclusion has been discussed by the CSTEE. If the CSTEE has disagreed on the conclusion, the reason is given in the Comments column
----------------------------------------
1 In ecological risk assessment, fixed application factors are applied to endpoints, the size of which depends on the quality and quantity of available data. The application factors give less flexibility in making judgements about the relationship between effect thresholds and environmental concentration. As with the default values used with MOSs, however, the application factors should be subject to scientific scrutiny and should not be applied without question.