In silico modelling is an important aspect of drug discovery and preclinical drug development because it provides low-cost prediction of pharmacokinetic (PK) parameters, which can help focus and de-risk drug discovery pipelines early on.
The kidneys play a vital role in the elimination of xenobiotics and their metabolic products from the body. Renal clearance is particularly important for molecules with low or negligible hepatic clearance. It is a complex process which includes glomerular filtration, tubular secretion and tubular reabsorption with active transport playing a major role in influencing these processes. Very few renal clearance models are available which illustrates the difficulties in developing useful tools for predicting this property.
Cyprotex has developed a novel state-space model for predicting human in vivo renal clearance and urinary elimination of chemicals. This model was optimised using xenobiotic plasma concentration and urine excretion time series data following intravenous (bolus and infusion), oral and subcutaneous dosing. The unknown model parameters were estimated using 5,000 data points drawn from 125 compounds. Three distinct statistical models were generated to predict the renal elimination rate constant, kr, using an in-house statistical pattern recognition system. These three statistical models were based on the following inputs:
- Structural descriptors only
- Structural descriptors and predicted protein binding (a Cyprotex proprietary model)
- Structural descriptors and observed plasma protein binding data.
It was clear from the validation of the models that including plasma protein binding has enhanced the predictive accuracy of the model. However, the lack of any transporter data is a limitation and future development will investigate the impact of these additional in vitro data on model performance. Cyprotex will also evaluate the effect of incorporating the renal clearance model into whole body physiologically-based pharmacokinetic (PBPK) models such as the recently introduced chemPKTM. In addition, further development of the renal clearance model could contribute to nephrotoxicity prediction through modelling local, intra-renal xenobiotic concentrations.
This research was recently presented at the Eurotox 2015 congress in Porto, Portugal by Dr Mohammed Atari, Mathematical Modeller at Cyprotex.