| Front page | | Contents | | Previous | | Next |

The Advisory list for self-classification of dangerous substances

3 Discussion & Conclusions

• 3.1 Chemicals on AL2010 that were not on AL2001
• 3.2 Chemicals on AL2001 that are not on the current list
• 3.3 Conclusion

This consolidated report contains documentation for all of the current advisory classifications on the DK EPA Advisory self-classification List (AL), i.e.:

• Carcinogenicity (update, from 2009)
• Mutagenicity (update, from 2009)
• Reproductive toxicity/harm to the unborn child (new, from 2009)
• Acute oral toxicity (update, from 2010)
• Sensitisation by skin contact (not updated, from 2001)
• Skin irritation (new, from 2010)
• Danger to the aquatic environment (update from 2009)

The 2009 update of the advisory classifications for cancer and mutagenicity was made using entirely new models; i.e. none of the models used to make the advisory classifications for mutagenicity and carcinogenicity on AL2001 were used in the update.

Annex 2 contains examples of how further structural analyses of substances belonging to various chemical classes can be made on top of the predicted properties from this project to visualize and gain further insight into relations between sub-structures and, in this case, the carcinogenicity properties of chemicals.

For the environmental advisory classifications some of the models used for AL2001 were used again for AL2009 (BCFWIN and model for aquatic toxicity to Fathead minnow), and new models were applied for biodegradation and aquatic toxicity to Dahpnia and Algae.

Comparisons between AL2001 and AL2009 are made in the following for the individual advisory classifications represented in both lists.

The following text originates from the AL2009 report /62/ and has only partly been updated to the 2010 update of acute oral toxicity and the addition of skin irritation.

3.1 Chemicals on AL2010 that were not on AL2001

As shown in figure 9, a larger number of chemicals have been assigned advisory classifications for the individual advisory classifications in the current advisory list than in the former. This is due primarily to the application of entirely different models with in many cases larger chemical domains than the models applied for AL2001. Also, a little more substances were included in the start list for AL2009 and AL2010 than for AL2001 (49,292 for AL2009/AL2010 and approximately 47,000 for AL2001)

For the advisory classifications for danger to the aquatic environment the reasons for the differences more specifically relate to the addition of aquatic toxicity models for Daphnia and Algae, plus the use of the non-linear BIOWIN 2 model instead of the linear BIOWIN 1 model, which was used for AL2001. BIOWIN 1 has a lower sensitivity than BIOWIN 2.

For the carcinogenicity and mutagenicity endpoints the increased number of predictions on AL2009 as compared with AL2001 is generally due to the use of new and improved (Q)SAR-models with larger applicability domains.

Figure 9 presents an overview of the number of advisory classifications for individual endpoints on AL2001 and the current consolidated AL2010. Reproduction and skin irritation are included although these endpoints were not addressed in AL2001.

Figure 9: Overview of the number of substances for each advisory classification in the current version compared to the 2001 version of the Advisory self-classification list. (Note: Reproductive toxicity and skin irritation were not included in AL2001. The advisory classifications sensitisation by skin contact, R43, have not been updated, and the number in the current version is therefore the same as in 2001.)

3.2 Chemicals on AL2001 that are not on the current list

There are also substances that were assigned advisory classifications on AL2001 that are not on AL2009/2010. It is for the individual endpoints seen that between 11 and 14% of the advisory classifications from AL2001 are not on AL2009/2010. An exception is acute oral toxicity, where around 45% of the 2001 advisory classifications are not on AL2010, primarily due to different applicability domains for the TOPKAT model applied for AL2001 and the Pharma ToxBoxes model applied for the current update. The differences for the other endpoints are also primarily due to the use of new models for AL2009/2010.

Chemicals on AL2001 may not have been included in updates made for AL2009/2010 for one or more of the following reasons:

• they have entered the List of Dangerous substances in the EU (EU harmonised classifications)
• they were not included on the new starting list for technical reasons (e.g. errors in the structural information or structure information not accepted by the (Q)SAR software)
• they were not within the applicability domain of some or all of the models applied for AL2009/2010
• they do not fulfil the new (Q)SAR model algorithms established for advisory classifications in AL2009/2010

For the mutagenicity endpoint, for example, where five models were used, many of the chemicals that were included on AL 2001 but not on AL 2009 did not have robust predictions (within applicability domain) in two or three models, but often with flags in one or more of these models showing that a possible active fragment was identified. Additionally, many have positive predictions in models for in vitro genotoxicity endpoints (which were not included in the evaluation). In total, the majority of the chemicals that were not identified this time appear to be borderline mutagens.

As there were mixed results (negative / out-of-domain / positive) from the battery of models applied within an endpoint, it is not possible to separate the chemicals strictly into groups of chemicals that were not identified this time because they could not be predicted (i.e. outside domain) or because the models applied in the new selection algorithm for AL2009 predict them to be negative for the effect.

A detailed comparison between numbers of chemicals with advisory classifications for carcinogenicity, mutagenicity and danger to the aquatic environment on AL2001 and AL2009 is given in table 12. VIOS det

* Due to overlap; some chemicals have advisory classifications for more than one CMR endpoint
** with one of the classifications for danger to the aquatic environment

Table 12. Overview of the occurrence of substances on AL2001 and AL2009

3.3 Conclusion

Due primarily to the application of combinations of new (Q)SAR models, in many cases with larger applicability domains, the number of substances with advisory classifications has increased considerably for individual classifications as compared to AL2001. Moreover, reproductive toxicity (possible harm to the unborn child), skin irritation, and differentiated advisory classifications for acute toxicity (harmful, toxic and very toxic) were included for the first time.

| Front page | | Contents | | Previous | | Next | | Top |

Version 1.0 March 2010, © Danish Environmental Protection Agency