TKPlate: a web-based open source platform for Toxicokinetic and Toxicodynamic Modelling.

The TKPlate 1.0 application implements models for Toxicokinetic modelling (i.e. one compartment kinetic model and physiologically-based kinetic (PBK) models and toxicodynamic modelling (i.e. benchmark dose modelling) to integrate New approach methodologies (NAMs) in human health, animal health and environmental risk assessment (RA) of chemicals into a graphical interface. It is structured within a workflow of seven modules: 1) Input module to select the PBK models in humans, test species (rat, mice, rabbit, dog) and farm animals (pig, sheep, chicken and cattle), the chemical-specific data, exposure patterns and related time scales, 2) Forward dosimetry module to predict kinetic parameters and concentrations in blood plasma and organs of interests, 3) Reverse dosimetry module to back-calculate exposure from an internal dose profile using, for example, blood and urine biomonitoring data, 4) Toxicodynamic module for benchmark dose modelling on an internal dose basis using model averaging, 5) Dynamic energy budget module to assess the impact of chemicals on the life cycle of individuals and populations of species of ecological relevance, 6) MIXTOX module for deterministic risk characterisation of combined exposure to multiple chemicals in human health, animal health and ecological RA using the component-based approach and default assumption of dose addition allowing risk characterisation of multiple chemicals, 7) An automated report summarising inputs provided by the user and outputs, graphs and datasets. The technical development of TKplate 1.0 is described in an external scientific report associated with an EFSA editorial and two technical reports as a suer guide and case studies to illustrate applications of TKplate 1.0 (Bossier et al., 2020; Bossier et al., 2023a, b, c. Dorne et al., 2023). 

The material associated with this publication includes:

• The TKPlate 1.0 stand-alone application (tktd_1.0-22.tar.gz). The application can in theory be installed in R using install.package(). However, note that a working version of MCSim (https://www.gnu.org/software/mcsim/) should be installed first in order to install 'tktd'.
• Model codes implemented in MCSim: 
  • pbtk1cpt.model: 1-compartment kinetic model
  • generic_pbk.model as a generic PBK model for humans and test species (rat, mice, rabbit, dog)
  • farmanimalModel.model as generic PBK models for farm animals (cattle, pig, sheep, chicken)
  • cosmos_pbk.model as a human multi-route 6 compartment PBK model

These files contain the differential equations that are being solved. Input files from module 1 containing all input parameters are created dynamically within the application and depend on user input.

• QSAR.R as a model code for a QSAR model predicting partition coefficients for a given polar or neutral chemical in PBK model compartments (blood and organs) from input n-Octanol/Water partition coefficient.
• The BMD R source code/ user manual and web application can be found in the following zenodo publication: https://zenodo.org/record/3760370 (Varewyck et al., 2017) and https://r4eu.efsa.europa.eu/app/bmd respectively. In TKplate the R source code has been slightly modified in TKPlate to allow BMD modelling based on internal dose instead of external dose. In addition, the user can also access the new Bayesian BMD application is described under https://zenodo.org/record/7986184 (Kremer et al., 2022) and is accessible under : https://r4eu.efsa.europa.eu/app/bmdbayesian
• R package for the Dynamic Energy Budget model
• Several Excel and csv files used throughout the application have also been uploaded:
  • speciesData.xlsx with physiological parameters used to build the generic PBK models for humans, laboratory animals (rat, mice, rabbit, dog) and farm animals (pig, sheep, chicken and cattle) 
  • data_chem.csv and data_TK.csv as templates to upload chemical-specific parameters and TK data respectively
  • PopGen.csv as a pre-made population used in the human generic PBK model (McNally et al., 2014). PopGen contains relevant physiological information and represent 1000 individuals, 500 per sex from a Caucasian population. BMI range from 21 to 24 and height is in between 150 and 190 for women and 160 and 200 for men. Since PopGen outputs are derived from non-truncated distributions, variability in parameter value is limited to the 95th percentile for each parameter to avoid the creating of unrealistic individuals.
  • Variability distributions in human metabolism from previously published meta-analysis. These include Phase I enzymes (CYP3A4, CYP2D6, CYP2C9, carboxyl-esterases) and Phase II enyzmes (UGTs).