Population TK/TD for chemical risk assessment and drug safety: the cadmium example
Amzal B., Dorne J.-L.
European Food Safety Authority (Parma, Italy)
Background: Population pharmacokinetic/pharmacodynamic (PK/PD) models have been largely developed and used in the context of drug development for a number of purposes,including the optimisation of dose regimens and the risk analysis of acute toxicity. These models are often based on dedicated clinical trials. On the other hand, very few models have been used to evaluate chronic or long term adverse effects such as population toxicokinetic/toxicodynamic (TK/TD) models particularly in the context of quantitative risk assessment in humans. In contrast to PK/PD models, TK/TD models developed for chronic exposure in humans can often only be based on epidemiological data.
Objectives: In this work, we illustrate how similar modelling approaches can be used in pharmacological and toxicological contexts, emphasizing their specificities and similarities. In the context of toxicological risk assessment of chemicals, population TK and dose-response models (TD) can be developed and linked together in order to evaluate a "safe dose" ("health-based reference value") for a given population. For this purpose, assessment and integration of population variability based on aggregated epidemiological data is a key element to perform a sound quantitative risk assessment.
Methods: The recent health risk assessment of Cadmium in food performed by the European Food Safety Authority (EFSA), illustrates the whole process of TK/TD model-based risk assessment. From a systematic review and Bayesian meta-analysis of scientific literature on cadmium renal effects, a populated Hill dose-effect model was built and the benchmark dose method implemented. The TK assessment involved the comparison of a 8-compartment toxicokinetic model and a one-compartment population TK model, based on a cohort study of 680 Swedish women, over a 20-year-long period.
Results: For cadmium renal effects, β2-microglobulinuria was the most commonly reported biomarker. The dose-effect model showed that a urinary cadmium level above 1 μg/g creatinine leads to an excess risk of 5% for the Caucasian population of being with β2-microglobulinuria above the critical cut-off of 300 μg/g creatinine. The TK model showed that a food intake of about 2.5ug/kg body weight and per week would prevent 95% of the Caucasian population from being above the threshold of 1 μg/g creatinine of urinary cadmium.
Conclusion: The modelling tools used in the pharmacological context could be applied to a toxicological context for the risk characterization of a contaminant for human health. Similarly, some toxicological risk assessment tools such as benchmark dose models or meta-analysis of epidemiological data could also be used for pre- or post-market drug safety evaluation in a pharmacological or pharmaceutical set-up.