Dose recommendations for specific populations are not always provided and, when available, typically rely on empirical derivation from a small fraction of the general population. In this study, a prediction/confirmation framework was applied to 2 model-based methods, physiologically based pharmacokinetics (PBPK) and a static model, to evaluate their ability to predict clearance in mild, moderate, and severe renal impairment populations and to inform dosing recommendations in these populations. Simulated renal impairment/healthy subject AUC ratios (AUCRs) from PBPK and static models were compared with observed AUCRs from dedicated clinical studies in renal impairment subjects for 7 drugs eliminated primarily by renal clearance. Both PBPK and static model predictions were within 2-fold of observed AUCRs for most compounds across all renal impairment categories. Predictions were generally more accurate for the mild and moderate renal impairment populations, with the majority of AUCR predictions within 80% to 125% of observed values for both methods. However, the accuracy of predictions was lower for the severe renal impairment population using the PBPK method. Given the accuracy observed, both methods may be suitable for prospective predictions for early decision-making, but are likely not sufficient sole justification for dose recommendations. There is a need to assess a larger database of compounds to enhance the predictive power of currently available tools.
Keywords: labeling; modeling and simulation; pharmacokinetics and drug metabolism; physiologically based pharmacokinetics (PBPK); renal disease.
© 2017, The American College of Clinical Pharmacology.