(37e) Model-Based Novel Strategy for Individualized Treatment of Sickle Cell Disease with Hydroxyurea

Sickle Cell Disease (SCD) is a hereditary disease caused by a single nucleotide mutation in the β subunit of hemoglobin producing abnormal, rigid, and sickle shaped red blood cells. Current treatment method involves daily oral dosage of hydroxyurea over a patient’s entire lifetime. Hydroxyurea is known to induce fetal hemoglobin synthesis that inhibits hemoglobin polymerization and thus relieves the symptoms associated with the disease. The major side effect of the drug is the inhibition of the enzyme ribonucleotide reductase, which causes bone marrow toxicity.

The main challenges associated with hydroxyurea are substantial interpatient variability in pharmacokinetic (PK) and pharmacodynamic (PD) profiles, cytotoxicity, lack of an effective biomarker to predict treatment efficacy and consequently lack of optimal dosing regimen for individual patients. The current biomarkers in use are fetal hemoglobin (HbF) and mean cell volume (MCV) of red blood cells (RBCs) to monitor treatment progression. A semi-mechanistic compartmental PK model is developed, which includes enzymatic reaction that converts hydroxyurea into nitric oxide and its derivatives. The PD model captures the biochemical and morphological changes the cell populations are undergoing during bone marrow differentiation. A population balance model approach is used where cells are compartmentalized according to their functionalities in hematopoietic process. All the cellular properties are lumped into two internal coordinates, cell maturation age and cell volume. The model predicts the volume distribution of reticulocytes and RBCs, and concentrations of leukocytes, reticulocytes and erythrocytes of SCD patients under treatment with hydroxyurea. Through our studies, we established that MCV of reticulocytes – an immediate precursor of erythrocytes – is a better biomarker as its dynamics changes on a faster timescale compared to HbF and MCV of RBCs. Further, global sensitivity analysis is done using Sobol’s method to identify sensitive parameters for obtaining individual patient model. Once the individual patient PK, PD trajectory is predicted, nonlinear model predictive control (NMPC) is applied to determine optimal dosage that bounds the model predictions within target values.

Another major issue with the lifelong treatment of hydroxyurea is non-compliance. It is difficult to differentiate non-compliance from treatment inefficacy, as non-complying patients might mislead clinicians into believing that treatment at the current dose level is insufficient. In this case, the dose might be increased but once the patient starts complying, the excess dose might prove to be harmful. The PK-PD model incorporates non-compliance to predict response and optimal dose for a non-complying patient. The current model is being tested on data collected from Riley Hospital for Children in Indianapolis.