Random sparse sampling strategy using stochastic simulation and estimation for a population pharmacokinetic study

Pharmacokinetic-pharmacodynamic modelling and simulation of methotrexate dosing in patients with rheumatoid arthritis

AIMS

To develop a non-linear mixed-effects population pharmacokinetic and pharmacodynamic (PK-PD) model describing the change in the concentration of methotrexate polyglutamates in erythrocytes (ery-MTX-PGn with "n" number of glutamate, representing PK component) and how this relates to modified 28-joint Disease Activity Score incorporating erythrocyte sedimentation rate (DAS-28-3) for rheumatoid arthritis (RA), representing PD component.

METHODS

An existing PK model was fitted to data from a study consisting of 117 RA patients. The estimation of population PK-PD parameters was performed using stochastic approximation expectation maximisation algorithm in Monolix 2021R2. The model was used to perform Monte Carlo simulations of a loading dose regimen (50mg subcutaneous methotrexate as loading doses, then 20mg weekly oral methotrexate) compared to a standard dosing regimen (10mg weekly oral methotrexate for 2 weeks, then 20mg weekly oral methotrexate).

RESULTS

Every 40 nmol/L increase in ery-MTX-PG3-5 total concentration correlated with 1-unit reduction in DAS-28-3. Significant covariate effects on the therapeutic response of methotrexate included the use of prednisolone in the first 4 weeks (positive use correlated with 25% reduction in DAS-28-3 when other variables were constant) and patient age (every 10-year increase in age correlated with 3.4% increase in DAS-28-3 when other variables were constant). 4 methotrexate loading doses led to a higher percentage of patients achieving a good/moderate response compared to the standard regimen (Week 4: 87.6% vs. 39.8%; Week 10: 64.7% vs. 57.0%).

CONCLUSIONS

A loading dose regimen was more likely to achieve higher ery-MTX-PG concentration and better therapeutic response after 4 weeks of methotrexate treatment.

Semi-Automated Therapeutic Drug Monitoring as a Pillar toward Personalized Medicine for Tuberculosis Management

Standard tuberculosis (TB) management has failed to control the growing number of drug-resistant TB cases worldwide. Therefore, innovative approaches are required to eradicate TB. Model-informed precision dosing and therapeutic drug monitoring (TDM) have become promising tools for adjusting anti-TB drug doses corresponding with individual pharmacokinetic profiles. These are crucial to improving the treatment outcome of the patients, particularly for those with complex comorbidity and a high risk of treatment failure. Despite the actual benefits of TDM at the bedside, conventional TDM encounters several hurdles related to laborious, time-consuming, and costly processes. Herein, we review the current practice of TDM and discuss the main obstacles that impede it from successful clinical implementation. Moreover, we propose a semi-automated TDM approach to further enhance precision medicine for TB management.

Development of population pharmacokinetics model of isoniazid in Indonesian patients with tuberculosis

OBJECTIVES

No population pharmacokinetics (PK) model of isoniazid (INH) has been reported for the Indonesian population with tuberculosis (TB). Therefore, we aimed to develop a population PK model to optimize pharmacotherapy of INH on the basis of therapeutic drug monitoring (TDM) implementation in Indonesian patients with TB.

MATERIALS AND METHODS

INH concentrations, N-acetyltransferase 2 (NAT2) genotypes, and clinical data were collected from Dr. Soetomo General Academic Hospital, Indonesia. A nonlinear mixed-effect model was used to develop and validate the population PK model.

RESULTS

A total of 107 patients with TB (with 153 samples) were involved in this study. A one-compartment model with allometric scaling for bodyweight effect described well the PK of INH. The NAT2 acetylator phenotype significantly affected INH clearance. The mean clearance rates for the rapid, intermediate, and slow NAT2 acetylator phenotypes were 55.9, 37.8, and 17.7 L/h, respectively. Our model was well-validated through visual predictive checks and bootstrapping.

CONCLUSIONS

We established the population PK model for INH in Indonesian patients with TB using the NAT2 acetylator phenotype as a significant covariate. Our Bayesian forecasting model should enable optimization of TB treatment for INH in Indonesian patients with TB.

Pharmacokinetic Modeling, Simulation, and Development of a Limited Sampling Strategy of Cycloserine in Patients with Multidrug-/Extensively Drug-Resistant Tuberculosis

BACKGROUND AND OBJECTIVE

Multidrug-resistant tuberculosis has much poorer treatment outcomes compared with drug-susceptible tuberculosis because second-line drugs for treating multidrug resistant tuberculosis are less effective and are frequently associated with side effects. Optimization of drug treatment is urgently needed. Cycloserine is a second-line tuberculosis drug with variable pharmacokinetics and thus variable exposure when programmatic doses are used. The objective of this study was to develop a population pharmacokinetic model of cycloserine to assess drug exposure and to develop a limited sampling strategy for cycloserine exposure monitoring.

MATERIAL AND METHODS

Patients with multidrug-/extensively drug-resistant tuberculosis who were treated for > 7 days with cycloserine were eligible for inclusion. Patients received cycloserine 500 mg (body weight ≤ 50 kg) or 750 mg (body weight > 50 kg) once daily. MW/Pharm 3.83 (Mediware, Groningen, The Netherlands) was used to parameterize the population pharmacokinetic model. The model was compared with pharmacokinetic values from the literature and evaluated with a bootstrap analysis, Monte Carlo simulation, and an external dataset. Monte Carlo simulations were used to develop a limited sampling strategy.

RESULTS

Cycloserine plasma concentration vs time curves were obtained from 15 hospitalized patients (nine male, six female, median age 35 years). Mean dose/kg body weight was 11.5 mg/kg (standard deviation 2.04 mg/kg). Median area under the concentration-time curve over 24 h (AUC_0-24 h) of cycloserine was 888 h mg/L (interquartile range 728-1252 h mg/L) and median maximum concentration of cycloserine was 23.31 mg/L (interquartile range 20.14-33.30 mg/L). The final population pharmacokinetic model consisted of the following pharmacokinetic parameters [mean (standard deviation)]: absorption constant K_{a_po} of 0.39 (0.31) h^-1, distribution over the central compartment (V_d) of 0.54 (0.26) L/kg LBM, renal clearance as fraction of the estimated glomerular filtration rate of 0.092 (0.038), and metabolic clearance of 1.05 (0.75) L/h. The population pharmacokinetic model was successfully evaluated with a bootstrap analysis, Monte Carlo simulation, and an external dataset of Chinese patients (difference of 14.6% and 19.5% in measured and calculated concentrations and AUC_0-24 h, respectively). Root-mean-squared-errors found in predicting the AUC_0-24 h using a one- (4 h) and a two- (2 h and 7 h) limited sampling strategy were 1.60% and 0.14%, respectively.

CONCLUSIONS

This developed population pharmacokinetic model can be used to calculate cycloserine concentrations and exposure in patients with multidrug-/extensively drug-resistant tuberculosis. This model was successfully validated by internal and external validation methods. This study showed that the AUC_0-24 h of cycloserine can be estimated in patients with multidrug-/extensively drug-resistant tuberculosis using a 1- or 2-point limited sampling strategy in combination with the developed population pharmacokinetic model. This strategy can be used in studies to correlate drug exposure with clinical outcome. This study also showed that good target attainment rates, expressed by time above the minimal inhibitory concentration, were obtained for cycloserine with a minimal inhibitory concentration of 5 and 10 mg/L, but low rates with a minimal inhibitory concentration of 20 and 32.5 mg/L.

Is One Sample Enough? β-Lactam Target Attainment and Penetration into Epithelial Lining Fluid Based on Multiple Bronchoalveolar Lavage Sampling Time Points in a Swine Pneumonia Model

Describing the disposition of antimicrobial agents at the site of infection is crucial to guide optimal dosing for investigational agents. For antibiotics in development for the treatment of nosocomial pneumonia, concentrations in the epithelial lining fluid (ELF) of the lung are frequently determined from a bronchoscopy at a single time point. The influence of profiles constructed from a single ELF concentration point for each subject has never been reported. This study compares the pharmacokinetics of two β-lactams, ceftolozane and piperacillin, among different ELF sampling approaches using simulated human regimens in a swine pneumonia model. Plasma and ELF concentration-time profiles were characterized in two-compartment models by the use of robustly sampled ELF concentrations and by the random selection of one or two ELF concentrations from each swine. A 5,000-subject Monte Carlo simulation was performed for each model to define the ELF penetration, as described by the ratio of the area under the concentration curve (AUC) for ELF to the AUC for free drug in plasma (AUC_ELF/fAUC_plasma) and the probability of target attainment (PTA). Given the intersubject variability of the ELF penetrations observed, differences between the models developed using robust numbers of ELF samples versus one or two ELF samples per swine were minimal for both drugs (maximum dispersion < 20%). Using a threshold exposure target of 60% of the time that the free drug concentration remains above the MIC target, the ceftolozane and piperacillin regimens achieved PTAs of ≥90% at MICs of up to 4 and 1 μg/ml, respectively, among the different ELF sampling strategies. These models suggest that the ELF models constructed with concentrations from sparse ELF sampling time points result in exposure estimates similar to those constructed from robustly sampled ELF profiles.

Hybrid Invasive Weed Optimization Algorithm for Parameter Estimation of Pharmacokinetic Model

Given that the traditional methods of estimating pharmacokinetic parameters are constrained by the sensitivity of their initial value and incapability of evolutionary algorithm to determine search range, this paper proposes a hybrid invasive weed optimization (HIWO) algorithm by combining the Hooke–Jeeves (HJ) and invasive weed optimization (IWO) algorithm. Using the HIWO algorithm for the parameter optimization by the experiment of extravascular administration two-compartment model, we can see that the proposed method is not only better than traditional feathering method (FM) in terms of numerical stability, but also better than HJ and IWO in terms of error minimization. The experimental results show that HIWO algorithm is a feasible method to optimize the pharmacokinetic parameters, which has higher precision and stronger robustness than the other techniques.