287731 Selection of Individualized Dosage Regimens by Using a Stochastic Optimization Approach
Specific techniques for individualizing and optimizing the dosage regimen and clinically monitoring each patient are necessary to offer her/him the proper doses of medicine promptly. This is of special concern for medicines that are costly or whose toxic effects are severe (e.g., oncological agents). The optimal dosage regimen for an individual is a combination of dose amount and dosing interval (i.e., time between doses) which minimize the risk (1-α) that the drug exposure deviates from the desired therapeutic window. The therapeutic window is defined as the range of drug exposure (e.g., blood concentration, area under the curve concentration-time) [TWL,TWU], which is below a threshold defining an acceptable risk of toxic side effect TWU and above a threshold defining a minimum acceptable level of therapeutic efficacy TWL. Laínez et al. (2011) show how a dose feasible range can be obtained provided an acceptable risk level. Such an approach can be applied for those pharmacokinetic (PK) models for which the dose variable can be isolated. In this work, the dosage regimen optimization problem for general pharmacokinetic models (i.e., those described by differential-algebraic equations) is presented by using a scenario based stochastic NLP which optimizes a downside risk metric. The scenarios are derived from the posterior joint distribution of the individual’s PK parameters which is obtained following an approximate Bayesian inference approach. A Smolyak rule is used for (i) the selection of the scenarios (i.e., combination of PK parameters) to be considered and (ii) the approximation of the downside risk metric. Two case studies, Gabapentin and Cyclophosphamide, are presented to emphasize the advantages and limitations of the proposed approach.
Laínez, J.M., G. Blau, L. Mockus, S. Orcun and G. Reklaitis, “Pharmacokinetic Based Design of Individualized Dosage Regimens Using a Bayesian Approach”, Ind. Eng. Chem. Res, 50, 5114-5130 (2011)
Support from the United States National Science Foundation (Grant NSF-CBET-0941302) is gratefully acknowledged.