The determination of the transfer of undesirable substances and contaminants from oral exposure of farm animals into food of animal origin is essential to human risk assessment and management. The aim of this thesis was to develop physiologically-based toxicokinetic (PBTK) models that can be used to predict the concentration of specific contaminants in food of animal origin, based on the concentration in feed. Three groups of toxins were considered in the modelling approaches: polychlorinated dibenzo-p-dioxins (PCDDs) and polychlorinated dibenzofurans (PCDFs), for short “dioxins”; polychlorinated biphenyls (PCBs); and quinolizidine alkaloids (QAs). As the first two groups, dioxins and PCBs, are quite similar from a kinetic point of view, they have been evaluated together and separately from the QAs. To gain a better understanding of the current state of research on the transfer of dioxins and PCBs to cow’s milk, a review of the literature was first compiled and published in two parts. This showed that although many modelling approaches have been developed, sufficiently complex models are often based on limited data sets. Therefore, there is still an urgent need for PBTK models that have been validated with sufficient data to accurately predict the transfer of dioxins and PCBs into milk and tissue. Within the framework of this thesis, two well parameterised PBTK models were successfully developed on a broad basis of experimental measurements. The first published model is based on data derived from a PCB contamination incident and describes the distribution of three non-dioxin-like (ndl) PCBs in various tissues and milk of an adult cow during two lactations and one dry period as well as in their calves. This also includes the transfer of the investigated ndl-PCBs from the adult cow to her calf via placenta and milk. Of particular note is the ability of the model to distinguish between placental transfer and milk transfer through suckling with 10-14 % of the amount of congeners in calves at slaughter was due to placental transfer. Several parameters were derived for risk assessment quantifying theses transfer and distribution processes. The second published model uses data from a feeding study in which dairy cows were fed a mixture of numerous dioxins and PCBs at different time intervals during their negative energy balance (NEB) and positive energy balance (PEB) phases to assess the effects of different metabolic states on the transfer of contaminants into milk. This was done with a three-compartment model that was parameterised separately for the NEB and PEB phases and reliably predicts the concentration-time (ct) profile of the investigated contaminants in milk and blood. It is particularly worth mentioning that some of the parameters derived here were reported for the first time. Comparison of the parameters obtained in the different energy balance phases revealed that the transfer rates of the investigated contaminants are significant higher in the PEB phase than in the NEB phase. Significant differences were also found for other parameters such as β half-lives, but to a much lesser extent. In the second part of the thesis, the knowledge gained from modelling approaches with dioxins and PCBs in cows was used to develop a model describing for the first time the transfer of three ndl-PCBs from feed into tissues of fattening pigs. The PBTK model was based on a feeding study in which fattening pigs were fed with ndl-PCBs contaminated feed at different stages of their fattening period. The model consists of a liver and a fat compartment, which allows to describe the concentration of ndl-PCBs in the most relevant tissues for the consumer, i.e., muscle (based on its fat content) and liver. In addition, various transfer parameters were derived, which allows the quantification of the extent of transfer of these contaminants to these tissues under realistically changing feeding conditions and animal growth. The lowest transfer rate was obtained for PCB-28 with 9.57 % and the highest transfer rate was obtained for PCB-153 with 77.2 %. In the final part of this thesis, QAs in dairy cows were investigated, which are kinetically completely different in their behaviour in comparison to dioxins and PCBs. QAs are potentially toxic plant secondary metabolites from lupins, a high-protein crop. Based on a feeding study in which dairy cows were fed subsequently with different amounts of QA-containing lupins, a PBTK model was developed to investigate the transfer of QAs into milk and its dose-dependency. As QAs are eliminated much faster from the body of the cows than dioxins and PCBs, aspects such as milking time and feeding time had to be taken into account. Since the ct-profile of QAs in milk during the depuration showed a biphasic behaviour, a three-compartment model was developed that induces a biphasic depuration phase in milk. This made it possible to describe the measured data quite well despite the high variability of the data points. The size of the experiment and the limited difference between the doses applies allowed no conclusion regarding a possible dose-dependent transfer. Several risk assessment parameters were derived, including transfer rates ranging from 1.05 % for angustifoline to 2.92 % for isolupanine and α-half-lives ranging from 0.26 d lupanine to 0.28 d hydroxylupanine. However, the very low concentrations covered by the β-half-lives are not expected to be relevant for risk assessment. In summary, in the scope of this work, several PBTK models were successfully developed that describe and quantify the transfer of various contaminants to dairy cows, calves and fattening pigs. To make these models accessible to risk assessors, they will be implemented in the web tool BfR ConTrans.
