LMC experience in applying read-across/trend analysis approach for predictive purposes shows that best guidance of forming categories is to follow three simple steps:
Step 1: Defining primary group. Primary category should be broad, non-endpoint specific. Primary category depends on the number of identified analogues. The aim is to have broader group of analogues which are structurally similar (to the max extent) with the target chemical. For this purpose for defining the primary group the following broader grouping profilers (structure-based) could be used:
- US-EPA New Chemical Categories
- Organic functional groups (including OFG USEPA and OFG Norbert Haider)
- ECOSAR
Step 2: Second step is subcategorization (refining) the primary group based on mechanistic (endpoint-specific) level (e.g endpoint specific profilers such as DNA alerts for AMES by OASIS or Protein binding alerts for SS, etc.). The latter is not mandatory. For example for assessing Carcinogenicity endpoint as a primary group could be used:
- DNA binding alerts + metabolism
- Protein binding alerts + metabolism
- Carcinogenicity ISS profiler + metabolism
-
etc.
Step 3. As the last step is applied again Step 1. Structure based profilers. This step is applied if need to eliminate dissimilar chemicals (to increase the consistency of category) (e.g. chemical elements).
Simple graphical illustration of the above steps is illustrated below:
