Eleni Pefani

Chemotherapy Drug Scheduling for the Induction Treatment of Patients With Acute Myeloid Leukemia

Leukemia is an immediately life-threatening cancer wherein immature blood cells are overproduced, accumulate in the bone marrow (BM) and blood and causes immune and blood system failure. Treatment with chemotherapy can be intensive or nonintensive and can also be life-threatening since only relatively few patient-specific and leukemia-specific factors are considered in current protocols. We have already presented a mathematical model for one intensive chemotherapy cycle with intravenous (IV) daunorubicin (DNR), and cytarabine (Ara-C) [1]. This model is now extended to nonintensive subcutaneous (SC) Ara-C and for a standard intensive chemotherapy course (four cycles), consistent with clinical practice. Model parameters mainly consist of physiological patient data, indicators of tumor burden and characteristics of cell cycle kinetics. A sensitivity analysis problem is solved and cell cycle parameters are identified to control treatment outcome. Simulation results using published cell cycle data from two acute myeloid leukemia patients [2] are presented for a course of standard treatment using intensive and nonintensive protocols. The aim of remission-induction therapy is to debulk the tumor and achieve normal BM function; by treatment completion, the total leukemic population should be reduced to at most 10

Design of optimal patient-specific chemotherapy protocols for the treatment of acute myeloid leukemia (AML)

^{9}

A framework for the design, modeling and optimization of biomedical systems

cells, at which point BM hypoplasia is achieved. The normal cell number should be higher than that of the leukemic, and a 3-log reduction is the maximum permissible level of population reduction. This optimization problem is formulated and solved for the two patient case studies. The results clearly present the benefits from the use of optimization as an advisory tool for treatment design.

Towards a high-fidelity model for model based optimisation of drug delivery systems in acute myeloid leukemia

Abstract AML is a cancer of the blood and bone marrow which results from the combined effects of genetic mutations, aberrant interactions in the microenvironment and altered networks of complex chemical reactions at the molecular and cellular level, some of which can be targeted with anti-neoplastic drugs called chemotherapy. AML can be treated with chemotherapy, the types and doses of which are dependent on characteristics of the patient, the sub-type of the tumor and the use of other, often synergistic, anti-cancer drugs, and the doses for which are limited by toxic adverse effects of treatment. Current treatment protocols are designed based on pre-clinical animal experiments and on empirical clinical trials as well as the acquired experience of subspecialist physicians. Mathematical modeling can assist in improving chemotherapy effectiveness and limiting toxicity through a systematic approach in designing treatment protocols. Specifically, these mathematical models should enable a description of the normal and the leukemic cell populations as dependent on disease characteristics (cell cycle distribution into phases, proliferation rate, initial disease and normal population state) and on physiological characteristics of the patient such as age, sex, body surface area that control and define the drug kinetics (concentration profile in tumor site). Such a model can then lead to an optimal management of the available drug kinetics in order to effectively eradicate the maximum possible tumor volume while limiting toxicity of the normal cell population and that will be maintained within certain defined limits. Herein, a model is presented for the first cycle of chemotherapy induction treatment for AML using daunorubicin (DNR) and cytarabine (Ara-C) anti-leukemic agents, a standard intensive treatment protocol for AML. The proposed model combines critical targets of drug actions on the cell cycle, together with pharmacokinetic (PK) and pharmacodynamic (PD) aspects providing a complete description of drug diffusion and action after administration. Tumor-specific characteristics, such as tumor burden and cell cycle times, as well as patient-specific characteristics, such as gender, age, weight and height, are incorporated into the model in an attempt to gain insights into the personalized cell dynamics during treatment. Moreover, an optimal control problem is formulated and solved so as to obtain the chemotherapeutic schedule which would maximize leukemic cell kill (therapeutic efficacy) while minimizing death of the normal cell population, thereby reducing toxicities. Simulation results for a standard treatment protocol are obtained for a patient case study; an optimized treatment schedule is also obtained and the cell populations are analyzed and compared in detail for both the standard and the optimized treatment protocols.

Cell cycle model selection for leukemia and its impact in chemotherapy outcomes

Abstract We present an overview of the key building blocks of a design framework for modeling and optimization of biomedical systems with main focus on leukemia, that we have been developing in the Biological Systems Engineering Laboratory and the Centre for Process Systems Engineering at Imperial College. The framework features the following areas: (i) a three-dimensional, biomimetic, in vitro platform for culturing both healthy and diseased blood; (ii) a novel, hollow fiber bioreactor that upgrades this in vitro platform to enable expansion and continuous harvesting of healthy and diseased blood; (iii) a global optimization-based approach for the design and operation of the aforementioned bioreactor; (iv) a pharmacokinetic / pharmacodynamic model representing patient response to Acute Myeloid Leukemia treatment; (v) an experimental framework for cell cycle modeling and quantitative analysis of environmental stress. This manuscript recapitulates the progress made in the different areas as well as the way in which these areas are connected, finally leading to a hybrid in vitro/in silico platform which allows the optimization of the ex vivo expansion of healthy and diseased blood.

Key environmental stress biomarker candidates for the optimisation of chemotherapy treatment of leukaemia

Abstract High doses of chemotherapy drugs are required to efficiently eradicate cancer cells during treatment of Acute Myeloid Leukemia (AML). Although effective for debulking of the leukemia tumour burden, the use of these chemotherapy regimens is also highly destructive to normal cell populations, often to a life-threatening extent. Drug optimisation of treatment dose in a patient-specific and leukemia-specific regimen is therefore essential in order to balance the benefits of therapy against the risks of toxicity. This optimisation may be achieved by producing a model of the key biological parameters of the patient, the AML cells and the effects of chemotherapy on both. This model may then be used as a predictive tool of the patient response during treatment, as demonstrated in our previous work pertaining to the optimal model-based control schedule for breast cancer chemotherapy treatment (Dua et al. 2008; Dua et.al. 2005). Cancer, including AML, results from the loss of control of the cell cycle where the cells proliferate abnormally. A complicated network of reactions and cell-signalling pathways are involved in this process of leukemogenesis. The actions and targeting of chemotherapeutic treatments to interfere with this abnormal cell signalling are equally as complicated in vivo . Herein, we attempt to describe this process through the development of a dynamic model for the in vivo actions of a single chemotherapeutic drug, cytosine arabinoside (ARA-C), used routinely for the treatment of AML, where the optimum dose is calculated within tolerable levels of drug toxicity. The proposed model combines the actions on the cell cycle, which is the target of the drug, with pharmacokinetic and pharmacodynamic aspects in order to provide a comprehensive description of drug diffusion and action after administration. The model also takes into account patient factors such as age, sex, weight and height in an attempt to gain insights towards optimisation of individual treatment protocols for effective patient-specific and leukemia-specific therapy that can also minimise toxicity.

Design of optimal disease and patient-specific chemotherapy protocols for the treatment of Acute Myeloid Leukaemia (AML)

Abstract The cell cycle is the biological process used by cells to replicate their genetic material and give birth to new cells that are in turn eligible to proliferate. It is highly regulated by the timed expression of proteins which trigger cell cycle events such as the start of DNA replication or the commencement of mitosis (when the cell physically divides into two daughter cells). Mathematical models of the cell cycle have been widely developed both at the intracellular (protein kinetics) and macroscopic (cell duplication) levels. Due to the cell cycle specificity of most chemother-apeutic drugs, these models are increasingly being used for the study and simulation of cellular kinetics in the area of cancer treatment. In this work, we present a population balance model (PBM) of the cell cycle in leukemia that uses intracellular protein expression as state variable to represent phase progress. Global sensitivity analysis highlighted cell cycle phase durations as the most significant parameters; experiments were performed to extract them and the model was validated. Our model was then tested against other differential cell cycle models (ODEs, delay differential equations (DDEs)) in their ability to fit experimental data and oscillatory behavior. We subsequently coupled each of them with a pharmacokinetic/pharmacodynamic model of chemotherapy delivery that was previously developed by our group. Our results suggest that the particular cell cycle model chosen highly affects the outcome of the simulated treatment, given the same steady-state kinetic parameters and drug dosage/scheduling, with our PBM appearing to be the most sensitive under the same dose.

Modelling and Simulation of Drug Delivery Systems for the Treatment of Acute Myeloid Leukemia

The impact of fluctuations of environmental parameters such as oxygen and starvation on the evolution of leukaemia is analysed in the current review. These fluctuations may occur within a specific patient (in different organs) or across patients (individual cases of hypoglycaemia and hyperglycaemia). They can be experienced as stress stimuli by the cancerous population, leading to an alteration of cellular growth kinetics, metabolism and further resistance to chemotherapy. Therefore, it is of high importance to elucidate key mechanisms that affect the evolution of leukaemia under stress. Potential stress response mechanisms are discussed in this review. Moreover, appropriate cell biomarker candidates related to the environmental stress response and/or further resistance to chemotherapy are proposed. Quantification of these biomarkers can enable the combination of macroscopic kinetics with microscopic information, which is specific to individual patients and leads to the construction of detailed mathematical models for the optimisation of chemotherapy. Due to their nature, these models will be more accurate and precise (in comparison to available macroscopic/black box models) in the prediction of responses of individual patients to treatment, as they will incorporate microscopic genetic and/or metabolic information which is patient-specific.

A systematic framework for the design, simulation and optimization of personalized healthcare: Making and healing blood

Abstract The current project focuses on the design and optimization of chemotherapy protocols for Acute Myeloid Leukaemia (AML). AML is a type of blood cancer in which patients are characterized by a weakened blood and immune system due to abnormalities in the bone marrow where the tumour is located. In that respect, there is a high risk that the patient will not withstand the treatment due to life-threatening toxicities of the chemotherapeutic drug mix. Therefore the individualization and optimization of treatment dose and schedule is essential to balance the benefits of higher dose therapy against the tumour with the toxicity to normal tissue. This balance can be achieved by modelling key biological mechanisms as a means to gain insight into the effects of chemotherapy, which can then be used as a predictive tool for patient response during treatment ( Dua et al., 2005 , Dua et al., 2008 ).In this work, we extend our previous model ( Pefani et al., 2011 ) for the first cycle of chemotherapy for the treatment of AML to include the chemotherapeutic drug action of both anticancer drugs used in current treatment protocols: cytarabine (Ara-C) and daunorubicin (DNR). The simulation and optimization results demonstrate the need for optimal treatment schedule in order to limit the life-threatening toxicities of chemotherapy-induced cytopenia in patients with AML undergoing treatment.

In vitro Studies: Acute Myeloid Leukaemia

Cancer, including Acute Myeloid Leukemia (AML), results from the loss of control of the cell cycle where the cells proliferate abnormally. A complicated network of reactions and cell-signaling pathways are involved in this process of leukemogenesis. The actions and targeting of chemotherapeutic treatments to interfere with this abnormal cell signaling are equally as complicated in vivo. Herein, we attempt to describe this process through the development of a dynamic model for the in vivo actions of a single chemotherapeutic drug, cytarabine (ARA-C), used routinely for the treatment of AML. The proposed model combines the actions on the cell cycle, which is the target of the drug, with pharmacokinetic and pharmacodynamic aspects in order to provide a comprehensive description of drug diffusion and action after administration. The model also takes into account patient factors such as age, sex, weight and height in an attempt to gain insights towards optimisation of individual treatment protocols for effective patient-specific and leukemia-specific therapy that can also minimise toxicity. This model may then be used as a means to obtain insight to the chemotherapy procedure and used as a predictive tool of the patient response during treatment, similar to what we have previously demonstrated for breast cancer treatment (Dua et al. 2008; Dua et.al. 2005).