You are here:
High Throughput pharmacokinetic modeling using computationally predicted parameter values: dissociation constants (TDS)
Strope, C., K. Mansouri, J. Kancherla, C. Stevens, AND J. Wambaugh. High Throughput pharmacokinetic modeling using computationally predicted parameter values: dissociation constants (TDS). Presented at ToxCast Data Summit, Research Triangle Park, NC, September 29 - 30, 2014. https://doi.org/10.23645/epacomptox.5197147
Association and dissociation constants are listed only as a number, and very little consideration is paid to the atom it is associated with or whether it is associated with a protonation or deprotonation event. Here we present a notation for pKas, and further examine several pKa prediction methods to determine the trend of their predictions, specifically relating to pharmaceutical versus environmental compounds.
Estimates of the ionization association and dissociation constant (pKa) are vital to modeling the pharmacokinetic behavior of chemicals in vivo. Methodologies for the prediction of compound sequestration in specific tissues using partition coefficients require a parameter that characterized the charge of the atom type. Current methods for reporting the pKa report only the pH at which the pKa-associated atom will be ionized in 50% of the molecules. Important considerations for in vivo usage of the pKa such as (i) the chemical class (i.e., acid/base), or (ii) the interplay between ionization states at other atoms to determine the fraction of a chemical to exist in a particular ionization state are reduced to “missing” information status. We propose a new method that more fully describes the process associated with of reporting pKa values. Further, this new format is designed to support high-throughput applications. We are comparing the ionizable atom types between 815 pharmaceutical and 2200 environmental compounds, and investigating the performance of several publically and commercially available pKa predictive models on these 3015 chemicals from published sources and the ToxCast library. Finally, the analysis methodology developed herein for efficient estimation of the parameters critical for predicting chemical pharmacokinetics will be publicly accessible as an R package. This abstract does not necessarily represent U.S. EPA policy.