Application of Bayesian Population Physiologically Based Pharmacokinetic (PBPK) Modeling and Markov Chain Monte Carlo Simulations to Pesticide Kinetics Studies in Protected Marine Mammals: DDT, DDE, and DDD in Harbor Porpoises

Abstract

Physiologically based pharmacokinetic (PBPK) modeling in marine mammals is a challenge because of the lack of parameter information and the ban on exposure experiments. To minimize uncertainty and variability, parameter estimation methods are required for the development of reliable PBPK models. The present study is the first to develop PBPK models for the lifetime bioaccumulation of p,p'-DDT, p,p'-DDE, and p,p'-DDD in harbor porpoises. In addition, this study is also the first to apply the Bayesian approach executed with Markov chain Monte Carlo simulations using two data sets of harbor porpoises from the Black and North Seas. Parameters from the literature were used as priors for the first "model update" using the Black Sea data set, the resulting posterior parameters were then used as priors for the second "model update" using the North Sea data set. As such, PBPK models with parameters specific for harbor porpoises could be strengthened with more robust probability distributions. As the science and biomonitoring effort progress in this area, more data sets will become available to further strengthen and update the parameters in the PBPK models for harbor porpoises as a species anywhere in the world. Further, such an approach could very well be extended to other protected marine mammals.

Overview of the Current State-of-the-Art for Bioaccumulation Models in Marine Mammals

“…Berkeley Madonna was the software used for model development of selected PCBs and PBDEs in harbour porpoises [17][18][19], AcslX/Libero was used for the models of p,p'-DDE, p,p'-DDT and p,p'-DDD in harbour porpoises [20] and AcslX/Libero and MCSim were used for the model of PCB 153 in long-finned pilot whales [21]. Consequently, available marine mammal models to date were built with a mixture of open source, as well as commercially available software.…”

“…In terms of modeling, metabolic biotransformation can be seen as a clearance or elimination pathway of the chemical of interest as it has been transformed into something else. That "something else" is invisible for the model as long as the model excludes the kinetics of the chemical"s metabolites which is the case in most models to date, except for the recent DDX-models in harbour porpoises [20] which include the kinetics of p,p'-DDE and of p,p'-DDD as metabolites of p,p'-DDT.…”

“…If sufficient information is available, which is hardly true for marine mammals, equations describing first and/or zero order kinetics or Michaelis-Menten kinetics can be used. Without sufficient information regarding biotransformation or for compounds that convert into multiple metabolites like PCBs, metabolic biotransformation rates can be deduced from first order kinetics [39] or from elimination half-life values of the chemicals [12,[16][17][18][19][20][21]40]. Elimination half-lives do not have to be fixed constants.…”

“…For terrestrial mammals and humans, these losses can be described by the contaminant concentrations detected in a fecal/urinary sample multiplied by the rate of fecal/urine release. Urinary and fecal losses are often lumped together with the metabolic biotransformation (which is described by a whole-body elimination half-life value) for marine mammals [16][17][18][19][20][21] or as discussed above sometimes one of the two pathways may be completely ignored [12].…”

“…Assuming that datasets are different populations from the same animal species as in this work, the parameter estimates will be representative for the entire population. This Bayesian approach with MCMC simulations has been used so far in bioaccumulation models for harbor porpoises and long-finned pilot whales (AcslX/Libero software and MCSim; [20,21]). …”