The pharmacokinetics of i.v busulfan vary significantly in pediatric population, and the optimal therapeutic range (TR) (expressed as AUC) for children has not been accurately determined. Since higher systemic exposure is associated with an increased risk of toxic adverse effects, such as mucositis, GvHD, VOD, or hepatic venous obstruction, whereas low systemic exposure to busulfan has been associated with a higher risk of graft rejection or recurrence, individualization of the dose of iv busulfan is necessary in children undergoing allogeneic HSCT, as described in the SPC of the commercially available formulation. In children, Busulfan is administered by i.v. infusion at a very low flow rate ("microinfusion") which may affect the calculation of the area under the curve (AUC), since a lag-time of drug entry into the general bloodstream. is generated due to the dead volume that must be removed from the injection device assembly until the drug enters the general bloodstream.
In general, infusion of drugs at a low rate results in an increased lagtime (delay time of drug release into the bloodstream) which may significantly affect the calculation of AUC if it is not taken into account. This is of particular importance for narrow-therapeutic range drugs for which the individualization of the dose is performed using the calculated AUC assuming linear pharmacokinetics. Therefore, in cases where AUC-based dose adjustment is needed, lagtime should be taken into account.
The objectives of this study were: 1) The measurement and interpretation of concentration-time profiles following intravenous BU administration and the individualization of dosage in pediatric patients which underwent HSCT at the "Bone Marrow Transplantation Center" in “Agia Sofia” Children Hospital of Athens, as described in the SPC of the administered drug, 2) In vitro evaluation of the lagtime of BU entrance in the bloodstream during the syringe pump-assisted microinfusion of BU in pediatric patients, using the patient's body weight based dosage categories described in the SPC of the drug (Busilvex®), 3) The incorporation of the in vitro determined lag-time in the AUC calculation methodology and dosage individualization as well as the evaluation of the possible effect on the predicted dose correction, 4) The development of a population pharmacokinetic model for BU in pediatric patients by incorporating the lag-time of BU entrance in the blood during i.v. syringe-pump-assisted microinfusion of the drug in pediatric patients, application of Bayesian individualization of dosage, and comparison with the conventional methodology.
This thesis is structured in 5 chapters. The first introductory chapter describes basic theoretical concepts, while the remaining four describe the methodologies that were followed and / or developed to achieve the objectives of the study, and discuss the results achieved. More specifically:
The second chapter of the dissertation deals with the development and validation of a HPLC method with a UV detector to measure the levels of Bu levels in blood plasma, which was necessary for the reliable and effective application of therapeutic monitoring and individualization of the BU dose. The analytical method which was developed requires low plasma volume (250 µL) and low analytical time (13 min) and is characterized by low cost, accuracy, repeatability and sensitivity (Range: 0.048 - 4.8 µg / mL, LOD: 0.048 µg / mL, LOQ: 0.013 µg / mL). The method is selective and accurate for within- and between-laboratory precision and accuracy not exceeding 6.5%. Thus, it is fully applicable to TDM in pediatric patients, where quantification in small plasma volumes with a lower quantification threshold is required.
The third chapter deals with the determination of the infusion lagtime when BU is administered to pediatric patients. For this purpose, various in vitro simulations were performed for the 5 different dosage categories described in the Summary of Product Specifications (SPC) of BU according to the actual body weight ABW (Category 1: 1.0mg / kg for MS <9kg, Category 2: 1.2 mg / kg for MS 9-16kg, Category 3: 1.1 mg / kg for MS 16-23 kg, Category 4: 0.95 mg / kg for MS 23-34 kg and Category 5: 0.8 mg / MS kg for MS> 34kg). Dose individualization was based on the calculated AUC value, using the trapezoidal method. For the calculation of AUC samples are taken at times: before infusion (0 h), immediately after the end of infusion, and 30 min, 1 h, and ~ 2-3 h after the end of infusion. Indicative time-series simulations were performed to study lag-time, corresponding to an infusion flow range of 15 mL / h to 30 mL / h. Given that the lag-time depends on the rate of transfer of the drug from the pump to the patient, the preparation of the infusion simulation solutions was based on the choice of the appropriate flow in order to achieve to the required volume of the drug to be administered according to the body weight per drug dosage category as described in the SPC of the drug. The sampling schedule consisted of 3 periods. During period A, samples were collected every 5 min for the first 30 min. During period B samples were collected every 10 min until the end of the infusion (~ 2h) and during period C samples were collected every 5 min until the infusion was rinsed. Thus, it was found that in patients with low body weight we have a high lagtime of 40 min, and as the body weight increases the lagtime is significantly reduced. More specifically, for patients with BW <9kg and BW 9-16 kg the estimated lag-time is 40 min. In patients with 16-23 kg BW, lag-time ranging between 35-25 min is observed. For BW 23-34 kg the lag-time is calculated at 20 min. While for BW> 34 the estimated lag-time ranges from 20 to 5 min.
Chapter 4 presents the application of therapeutic drug monitoring (TDM) and dose individualization i.v. BU in 76 pediatric patients with a mean age of 7.6 years (range 0.5-19 years) undergoing bone marrow transplantation at "Agia Sophia" Children's Hospital of Athens, between July 2014 and January 2017. The dose given was 0.8-1.2 mg / kg depending on patient weight. Each infusion lasted about two to three hours and was administered to each patient every six hours for four consecutive days before chemotherapy or two consecutive days. Samples were taken at time 0 (before infusion), at 2-3 h (at the end of infusion), at 2.5 or 3.5 h (half an hour after the end of infusion), at 4 and 6 hours after starting the administration of the first dose. Using real-time sampling and calculated Busulfan concentrations, AUCtotal was calculated using the trapezoidal rule. The percentage of patients found within TR was 59.2%, while 7.9% of the patients were above TR 900.0-1350 µM * min and 32.9% below it. Dose individualization was performed for the patients found out of the TR. The percentage of patients within the TR (900.0-1350.0 µM * min) after the dose modification was 89.04% while 10.96% were above the TR 900.0-1350.0 µM * min.
Taking into account lag-time for the calculation of AUC (AUClag), 42.1% of the 76 patients who participated in the study were within the range 900.0-1350.0 µM * min, 7.9% were above the TR and 50% is below the TR, while the proportion of patients within the TR after dose modification increases to 89.0% with 4.1% below the TR 900.0-1350.0 µM * min and 6.9% laying above it. In each case, after dose individualization, all patients are below the upper toxic level of 1500.0 µM * min based on the extended TR 900.0-1500.0 µM * min, regardless of how the AUC is calculated (with or without lag-time contribution).
The present study showed that when infusion lag-time was taken into account, the calculated AUClag values are lower than those calculated without the lag time, AUCnolag. The use of AUClag is more sensitive to the detection of patients on subtherapeutic levels (AUC <900 µM * min), while the number of patients above the threshold (AUC> 1350 µM * min) is not significantly affected. At the same time, it is necessary to adjust the dose based on the strict TR (900
In fifth chapter of this thesis, the concentration-time data, taking into account the lag-time, were used to develop a population PK (POPPK) model using the nonlinear mixed effect models (NONEMEM 7.3).
The model that best described the data was the two-compartment model with first order elimination kinetics. The estimated parameters were clearance, CL, volume of distribution of the central Vc and peripheralVp compartments and intercompartmental clearance, Q. Inter-individual variability (IIV) was estimated for the first 2 parameters, and a significant correlation was found between CL and Vc. The combined error model was used to describe residual variability. The combination of weight and age as covariates in CL and weight in Vc had a significant effect on the fit of the data. The model was evaluated for the fit to the data (goodness of fit plos), by bootstrap analysis, and its predictive ability (VPC plot) and was considered suitable for describing BU's PK in pediatric patients.
Post hoc estimates of the individualized PK parameters for each patient were used to estimate the AUC parameter which was used for dose-individualization. Subsequently, the percentage of patients within the TR when AUC was calculated by the trapezoidal method (AUCtrapezoidal-lagtime) and the respective percentage of patients using population modeling (AUCBaeysian) were compared. The percentage of patients within TR using the trapezoidal method and taking into account the lag-time differs slightly from that of the Bayesian method, and in particular the Bayesian individualization revealed a higher proportion of patients within the TR (44.7%) after the first infusion, compared to that of trapezoidal AUC containing the lag-time which is 42.1%.
In addition, and in order to investigate the best sampling scheme and the possibility of using fewer sampling points, a comparison was performed for the percentage of patients within the TR (900.0-1350.0μM * min) based on the Bayesian individualization, between the sampling design of the study, and smaller sample sizes by removing 1, 2 and 3sampling times from the original sample scheme. The comparison showed that the population-based PK model developed could correctly estimate the AUC by the posthoc method even if one sampling point was removed immediately after the end of the infusion, as the percentages of patients within the TR did not differ from that calculated using the trapezoidal AUC which takes into account the lag-time and with the sample scheme used in the study, after posthoc estimation.
Overall, this study shows that:
• The bio-analytical method developed is fully selective and accurate with values of within and between laboratory days accuracy and precision not exceeding 6.3%. The method is rapid, repeatable, with a limit of detection and quantification that allows the determination of low concentrations of busulfan in the blood with relatively low cost.
• When applying TDM and individualization of BU dosage in pediatric patients, taking into account the infusion lag-ime, the calculated AUC values are lower than those calculated without the lag- time, and the use of AUClag is more sensitive for the determination of patients at sub-therapeutic levels (AUC <900 µM * min), while the number of patients above AE (AUC> 1350 µM * min) is not significantly affected.
• At the same time, it is necessary to adjust the dose based on the restricted TR (900 • The POPPK model developed correctly estimates patient percentage within, above and below the 900-1350 µM * min range similar to the SPC proposed trapezoidal method.
• Additionally, the POPPK model with the posthoc method can correctly estimate the AUC if one sampling point is removed immediately after the end of the infusion as the percentages of patients within the TR do not differ from those determined using the trapezoidal method which takes into account the infusion lag-time and with that of the original sampling scheme after posthoc individualization.
