Predicting ready biodegradability of premanufacture notice chemicals.

Chemical substances other than pesticides, drugs, and food additives are regulated by the U.S. Environmental Protection Agency (U.S. EPA) under the Toxic Substances Control Act (TSCA), but the United States does not require that new substances be tested automatically for such critical properties as biodegradability. The resulting lack of submitted data has fostered the development of estimation methods, and the BioWIN models for predicting biodegradability from chemical structure have played a prominent role in premanufacture notice (PMN) review. Until now, validation efforts have used only the Japanese Ministry of International Trade and Industry (MITI) test data and have not included all models. To assess BioWIN performance with PMN substances, we assembled a database of PMNs for which ready biodegradation data had been submitted over the period 1995 through 2001. The 305 PMN structures are highly varied and pose major challenges to chemical property estimation. Despite the variability of ready biodegradation tests, the use of at least six different test methods, and widely varying quality of submitted data, accuracy of four of six BioWIN models (MITI linear, MITI nonlinear, survey ultimate, survey primary) was in the 80+% range for predicting ready biodegradability. Greater accuracy (>90%) can be achieved by using model estimates only when the four models agree (true for 3/4 of the PMNs). The BioWIN linear and nonlinear probability models did not perform as well even when classification criteria were optimized. The results suggest that the MITI and survey BioWIN models are suitable for use in screening-level applications.

Using Biowin, Bayes, and batteries to predict ready biodegradability.

Whether or not a given chemical substance is readily biodegradable is an important piece of information in risk screening for both new and existing chemicals. Despite the relatively low cost of Organization for Economic Cooperation and Development tests, data are often unavailable and biodegradability must be estimated. In this paper, we focus on the predictive value of selected Biowin models and model batteries using Bayesian analysis. Posterior probabilities, calculated based on performance with the model training sets using Bayes' theorem, were closely matched by actual performance with an expanded set of 374 premanufacture notice (PMN) substances. Further analysis suggested that a simple battery consisting of Biowin3 (survey ultimate biodegradation model) and Biowin5 (Ministry of International Trade and Industry [MITI] linear model) would have enhanced predictive power in comparison to individual models. Application of the battery to PMN substances showed that performance matched expectation. This approach significantly reduced both false positives for ready biodegradability and the overall misclassification rate. Similar results were obtained for a set of 63 pharmaceuticals using a battery consisting of Biowin3 and Biowin6 (MITI nonlinear model). Biodegradation data for PMNs tested in multiple ready tests or both inherent and ready biodegradation tests yielded additional insights that may be useful in risk screening.