A framework for prioritizing fragrance materials for aquatic risk assessment.

More than 2,100 chemically defined organic chemicals are listed in the Research Institute of Fragrance Materials/Flavor and Extract Manufacturers' Association (RIFM/FEMA) Database that are used as ingredients of fragrances for consumer products. An approach was developed for prioritizing these fragrance materials for aquatic risk assessment by first estimating the predicted environmental concentration (PEC) of these fragrance materials in the aquatic environment based upon their physicochemical properties and annual volume of use. Subsequently, an effect level was predicted with a general quantitative structure-activity relationship (QSAR) for aquatic toxicity, and a predicted no-effect concentration (PNEC) was calculated from this effect level by using an assessment factor (AF) that accounts for uncertainty in the toxicity QSAR prediction. A conservative AF of 10(6) was applied to the endpoint predicted by the QSAR to provide an adequate margin of safety in the calculation of the PNEC. The PEC was compared to the PNEC to characterize the risk to freshwater aquatic organisms (e.g., Daphnia magna and Pimephales promelas). If the ratio of PEC to PNEC was below one, the material was considered to have negligible environmental risk and to be acceptable for the aquatic environment at current use levels. If this ratio exceeded one, the PNEC was refined by using more specific QSAR models (Ecological Structure-Activity Relationships [ECOSAR]). If the ratio continued to exceed one, the material became a candidate for further aquatic risk assessment procedures, which involve iterative steps to refine the PEC, the PNEC, or both by using measured ecotoxicological endpoints. Prioritization for this latter process can be based upon the magnitudes of the estimated PEC:PNEC ratios. When using the first tier of this approach, only 568 of 2,141 fragrance materials (26.5%) in the RIFM/FEMA Database had PEC:PNEC ratios greater than one. This percentage decreased to only 164 materials (7.7%) when PNECs were derived with ECOSAR. Comparison of predicted PECs and PNECs with those based upon measured data confirmed the conservatism and low risk for type I errors associated with the framework. These combined exercises demonstrated the ability of this highly precautionary risk-based screening approach to quickly prioritize a large number of materials without benefit of experimental ecotoxicological or fate data.