Development of a Decision Support System for the Introduction of Alternative Methods into Local Irritancy/Corrosivity Testing Strategies. Development of a Relational Database
Ingrid Gerner, Gabriele Graetschel, Jürgen Kahl and Eva Schlede
For new chemical substances that are notified within the European Union, data sets have to be submitted to the National Competent Authorities. The data submitted have to demonstrate the physicochemical and toxic properties of the new chemical, such as solubility, partition coefficients and spectra, as well as acute toxic properties and the potential to cause local irritant or corrosive effects. In order to minimise testing for notification purposes (for example, animal testing), it is necessary to develop stepwise assessment procedures, including structure-activity considerations, alternative methods (for example, in vitro tests), and computerised structure-activity relationship (SAR) models. An electronic database was developed which contains physicochemical and toxicological data on approximately 1300 chemical substances. It is used for regulatory structure-property relationship (SPR) and SAR considerations, and for the development of rules for a decision support system (DSS) for the introduction of alternative methods into local irritancy/corrosivity testing strategies. The information stored in the database is derived from proprietary data, so it is not possible to publish the data directly. Therefore, the database is evaluated by regulators, and the information derived from the data is used for the development of scientific information about SARs. This information can be published, for example, by means of tables correlating measured physicochemical values and specific toxic effects caused by the measured chemical. This information is introduced to the public by means of a DSS that predicts local irritant/corrosive potential of a chemical by listing socalled exception rules of the kind IF (physicochemical property) A THEN not (toxic) Effect B and so-called structural rules of the kind IF Substructure A THEN Effect B. These DSS rules “translate” proprietary data into scientific knowledge that can be published.