Comparison of Cramer classification between Toxtree, the OECD QSAR Toolbox and expert judgment

Area of Science:

• Toxicology
• Computational Chemistry
• Risk Assessment

Background:

• The Threshold of Toxicological Concern (TTC) provides exposure limits for chemicals lacking safety data.
• The Cramer decision tree categorizes chemicals into classes to establish TTC values.
• Accurate Cramer class assignment is essential for reliable risk assessment.

Purpose of the Study:

• To evaluate the accuracy of computational tools in assigning Cramer classes for fragrance materials.
• To compare the performance of Toxtree and the OECD QSAR Toolbox against expert judgment.
• To identify challenges and provide guidance for Cramer classification of diverse chemical structures.

Main Methods:

• Cramer classification of over 1000 fragrance materials using Toxtree, OECD QSAR Toolbox, and expert judgment.
• Quantitative comparison of classification results between computational tools and expert assessment.
• Analysis of concordance rates within specific chemical classes, including amines, lactones, and heterocycles.

Main Results:

• Significant discordance observed between Toxtree and OECD QSAR Toolbox (16% of materials).
• Overall concordance with expert judgment: 83% for Toxtree and 77% for OECD QSAR Toolbox.
• Lowest agreement with expert judgment for amines, lactones, and heterocycles, particularly with Toxtree.

Conclusions:

• Computational tools show variability in Cramer classification, necessitating expert oversight.
• Specific chemical classes like amines, lactones, and heterocycles require careful assessment.
• Refined strategies are needed for accurate Cramer classification to ensure risk assessment integrity.