(376h) Development of a Generic Physiology Based Biokinetic Model for Predicting Internal Dose and Assimilation of Biomonitoring Data for Industrial Chemicals

Physiology Based BioKinetic (PBBK) models are continuously gaining ground in regulatory toxicology, describing in quantitative terms the absorption, metabolism, distribution and elimination (ADME) processes in the human body, with a focus on the effective dose at the expected target site. This trend is further amplified by the continuously increasing scientific and regulatory interest about aggregate and cumulative exposure. PBBK models translate external exposures from multiple routes   into internal exposure metrics, addressing the effects of exposure route in the overall bioavailability, or the dependence on critical developmental windows characterized by enhanced susceptibility, such as pregnancy, lactation and infancy. Recently, efforts have shifted also towards the integration of whole-body physiology, disease biology, and molecular reaction networks, as well as integration of cellular metabolism into multi-scale whole-body models.

The current study aims at the development of (a) a lifetime PBBK mode that takes stock of gestation and breastfeeding while covering a large chemical space and (b) a framework for biomonitoring data assimilation. The model captures multi-route exposure (including inhalation, skin absorption and gastro-intestinal absorption), and incorporates detailed compartmental description, plasma protein/red cell binding and estimates the concentration of parent compounds and metabolites in several biological fluids including breast milk and the contribution of lactation to breast feeding exposure, as well as the potential of bioaccumulation. The latter is very important when dealing with persistent/bioaccumulative compounds. Assessment of internal dose through PBBK modeling allows the estimation of internal doses of xenobiotics that exceed levels associated with biological pathway alterations and, eventually, induction of biological perturbations that may lead to health risk. The biological pathway altering dose (BPAD) outperforms current metrics of risk in that it combines dose−response data with analysis of uncertainty and population variability so as to derive exposure limits. BPADs, which can be  derived from high-throughput screening (HTS) in vitro data, are more pertinent in the case of pathways the perturbation of which is a key event in the mode of action (MOA) leading to an apical of systemic adverse outcome.

In order to expand the applicability of the generic PBBK model to cover as much as possible the chemical space, parameterization of the model for data poor chemicals is done using advanced quantitative structure-activity relationships (QSARs). This pertains to all necessary parameters such as blood-tissue partition coefficients, clearance and elimination kinetics. Several QSAR modeling approaches have been investigated, including (a) the Peyret, Poulin and Krishnan algorithm, which is based on the fractional content of cells, interstitial fluid in tissue, plasma in blood, erythrocyte in blood, tissue lipids and the lipophilicity of the compound of interest; (b) the molecular fractions algorithm proposed by Béliveau et al. that takes into account the frequency of occurrence of the several molecular fragments of the compounds and (c) Abraham’s solvation equation for estimating biological properties, which takes into account the excess molar refraction that can be determined simply from knowledge of the compound refractive index, the compound dipolarity/polarizability, the solute effective or summation hydrogen-bond acidity, the solute effective or summation hydrogen-bond basicity and the McGowan characteristic volume that can trivially be calculated for any solute simply from a knowledge of its molecular structure. Up to now, these QSARs seem to perform well for a significant number of chemical families. A major breakthrough came from the use of Artificial Neural Networks coupled to Abraham’s solvation equation parameters for predicting chemical-specific biological/biochemical properties such as blood-tissue partition coefficients, maximal velocity (Vmax) and the Michaelis - Menten constant. This was a remarkable advance, since the prediction capability of the Michaelis - Menten constant in the existing studies to date was rather poor (R² up to 0.35). With our coupled ANN - Abraham’s solvation equation method for the investigated group of 55 chemicals, R² rose to 0.88. For the rest of the parameters (partition coefficients, Vmax), performance of prediction against experimental values was consistently high (R² always above 0.9).

The PBBK model is geared with reverse modeling algorithms in order to reconstruct exposure from human biomonitoring (HBM) data. A tiered approach is followed as a function of data availability (periodicity and size of sampling, specimen type) and requirements of the exposure reconstruction analysis (temporal analysis of exposure, contribution from different routes), ranging from Exposure Conversion Factors (ECFs), up to Markov Chain Monte Carlo analysis. Probabilistic techniques are used to validate the exposure outcome on the basis of actual environmental and population biomonitoring data. Assimilation of human biomonitoring data and their translation into intake distribution amounts to a computational inversion problem, where the objective is to identify the specific input distributions that best explain the observed outputs while minimizing the residual error. Inputs involve spatial and temporal information on micro-environmental media concentrations of xenobiotics and corresponding information on human activities, food intake patterns or consumer product use that result in intakes; outputs are the observed biomarkers. The error metric can be defined in terms of population variation (the latter has to be lower than the intra-individual variation, which may be associated to measurement or other random error source). At the individual level, the PBBK model is combined with multimedia models and survey questionnaires to identify exposure sources, used as ancillary information, aiming to predict exposure magnitude and eventually the timing of exposure events. A computational framework was developed based on Bayesian Markov Chain Monte Carlo (MCMC) combined with the generic Physiological Based Pharmacokinetic (PBPK) model to perform accurate exposure reconstruction (ER). The ER framework developed consists of 3 basic steps:

At first the prior parameter distribution, the joint probability distribution, the population model and the determination of the measurement model have to be specified. At the next step exposure is calculated using MCMC simulation considering the observed biomonitoring data. Finally, the evaluation of the results is realized using MC simulation, with emphasis to the comparison of prior and posterior distribution as well as parameter independence. MCMC simulation refers to a class of iterative simulations in which the random variables of interest are drawn from a sequence, or chain, of distributions that eventually converge to a stable posterior distribution. Moreover, Differential Evolution (DE) and MCMC algorithms have been combined to this problem for the first time. Differential Evolution Markov Chain is a population MCMC algorithm, in which multiple chains run in parallel. In fact DE is a simple genetic algorithm for numerical optimization in real parameter spaces. As a result, this combined computational framework speeds up the calculation and convergence, even for nearly collinear parameters and multimodal densities.

The generic model was applied for the assessment of a highly controversial industrial chemical with widespread applicability in consumer goods, namely bisphenol-A (BPA).  Exposure scenarios were built based on an extensive literature review of BPA exposure data. For the majority of scenarios, the estimated internal dose was close to 0.002 μg/L and only in the case of bottle fed infants, internal exposure concentrations were up to 0.023 μg/L. This is partially explained by the neonates immaturity of the detoxification pathway, resulting to higher internal doses for the same bodyweight normalized dose compared to children older than 1 year old or adults. The biologically effective dose of the developing fetus during gestation is highly linked to the one in maternal blood. According to our model and based on a conservative exposure scenario for the mother (e.g. 5 μg/kg_bw/d), free plasma BPA in maternal blood is almost 0.006-0.007 μg/L, which is slightly higher than what expected for a non-pregnant woman (0.005 μg/L). Placental concentration is 0.0013 μg/L and the corresponding fetal concentration is 0.004-0.005 μg/L. Maternal BPA-Glu bioavailability is also very important in the case of breast-fed infants. Transfer of BPA through milk is not sufficient enough to explain exposure of breast-fed infants; the overall BPA exposure through breast-feeding can only be explained by BPA-Glu cleavage in the gastrointestinal tract. Even when the worst-case scenario is taken into account, breast fed infants seem to be significantly less exposed compared to the bottle fed infants and neonates. In addition, exposure to BPA was reconstructed based on real-life HBM data across Europe, covering different age groups. Average urine BPA-Glu was 2.8 μg/L, resulting in very low overall daily intake, approximately 1 μg/kg_bw/d, which is far below the Tolerable Daily Intake (TDI) of 50 μg/kg_bw/d proposed by the European Food Safety Authority (EFSA). Running the model in a forward mode (based on the exposure levels estimated by exposure reconstruction) and estimating the biologically effective dose through the target tissue, the maximum derived internal exposure values of the worst-case exposure scenarios are 400 times lower to the BPAD, indicating that there is no reason for concern for individual or aggregate scenarios of BPA exposure. 

Overall, our comprehensive and generic life stage PBBK model supports the association of a variety of environmental, exposure and biomonitoring data, as well as the incorporation of recent advances of in vitro toxicology using high-throughput systems in the risk assessment process enhancing thus significantly the artillery of environmental health science and chemical safety regulators.