There was no regulatory requirement for ecotoxicological testing of human pharmaceuticals authorized before 2006, and many of these have little or no data available to assess their environmental risk. Motivated by animal welfare considerations, we developed a decision tree to minimize in vivo fish testing for such legacy active pharmaceutical ingredients (APIs). The minimum no observed effect concentration (NOECmin, the lowest NOEC from chronic Daphnia and algal toxicity studies), the theoretical therapeutic water concentration (TWC, calculated using the fish plasma model), and the predicted environmental concentration (PEC) were used to derive API risk quotients (PEC/NOECmin and PEC/TWC). Based on a verification data set of 96 APIs, we show that by setting a threshold value of 0.001 for both risk quotients, the need for in vivo fish testing could potentially be reduced by around 35% without lowering the level of environmental protection. Hence, for most APIs, applying an assessment factor of 1000 (equivalent to the threshold of 0.001) to NOECmin substituted reliably for NOECfish, and TWC acted as an effective safety net for the others. In silico and in vitro data and mammalian toxicity data may further support the final decision on the need for fish testing.
Keywords: animal welfare; ecotoxicity; environmental risk assessment; fish plasma model; vertebrate testing.