(254h) Tailoring Treatment for Individual Patients: Bayesian Modeling and Control of Chemotherapeutics

Tailoring Treatment for Individual Patients: Bayesian Modeling and Control of Chemotherapeutics

Jayachandran Devaraj^a, José Miguel Laínez^a, Doraiswami Ramkrishna^a,*

^a School of Chemical Engineering, Purdue University, 480 Stadium Mall Way, West Lafayette, IN 47907, USA                   

^*Corresponding author at: School of Chemical Engineering, Purdue University, 480 Stadium Mall Way, West Lafayette, IN 47907, USA. Tel: +1 765 494 4066; fax: +1 765 494 0805; email: ramkrish@purdue.edu

Abstract

Inter-patient variation in drug response presents a great challenge in determining the optimum dose for an individual patient. In the present form of chemotherapy administration, patients are started with an average dose of drug and ‘titrated’ to a maximum tolerated dose through a trial and error approach. During this process, patients undergo several dose modifications and encounter severe episodes of acute side-effects. Many retrospective studies show that such endeavors are responsible for the long-term medical complication. A possible alternative would be a system theoretic framework adequately empowered by statistical and optimal control theory. Such a quantitative approach, if implemented carefully, will greatly enhance the decision making capabilities of the treating physicians.

In this work, we present a Bayesian modeling and robust model predictive control based approach to individualize the dosing of an important chemotherapeutic drug, 6-Mercaptopurine (6-MP). 6-MP is one of the key drugs in the treatment of many pediatric cancers, auto immune diseases and inflammatory bowel disease (IBD). 6-MP is a prodrug which is converted to 6-TGN, an active metabolite, through enzymatic conversion. An enzyme known as TPMT plays a key role in this metabolic conversion and determines the amount of 6-TGN produced. A genetic polymorphism observed in the TPMT enzyme produces a huge variation in drug response among patient population. The pharmacological model for 6-MP metabolism accounts for such variations in enzyme activities and predicts the 6-TGN concentration in humans. To circumvent the data scarcity in the clinical settings, a global sensitivity analysis based model reduction approach is used to minimize the parameter space. The complete and reduced models were compared in terms of goodness-of-fit and number of samples required in characterizing a new patient. Furthermore, we show how the availability of additional information on the biomolecular hierarchy can be integrated into Bayesian information and improve the prediction quality. With the final model, a model predictive control algorithm is used to optimize the dose scheduling with the objective of maintaining the 6-TGN concentration within its therapeutic window. The model and the control approach can be utilized in the clinical setting to individualize 6-MP dosing based on the patient’s ability to metabolize the drug instead of the traditional standard-dose-for-all approach.