Optimizing Predictive Performance of Bayesian Forecasting for Vancomycin Concentration in Intensive Care Patients

Dose Individualisation of Antimicrobials from a Pharmacometric Standpoint: The Current Landscape

Successful antimicrobial therapy depends on achieving optimal drug concentrations within individual patients. Inter-patient variability in pharmacokinetics (PK) and differences in pathogen susceptibility (reflected in the minimum inhibitory concentration, [MIC]) necessitate personalised approaches. Dose individualisation strategies aim to address this challenge, improving treatment outcomes and minimising the risk of toxicity and antimicrobial resistance. Therapeutic drug monitoring (TDM), with the application of population pharmacokinetic (popPK) models, enables model-informed precision dosing (MIPD). PopPK models mathematically describe drug behaviour across populations and can be combined with patient-specific TDM data to optimise dosing regimens. The integration of machine learning (ML) techniques promises to further enhance dose individualisation by identifying complex patterns within extensive datasets. Implementing these approaches involves challenges, including rigorous model selection and validation to ensure suitability for target populations. Understanding the relationship between drug exposure and clinical outcomes is crucial, as is striking a balance between model complexity and clinical usability. Additionally, regulatory compliance, outcome measurement, and practical considerations for software implementation will be addressed. Emerging technologies, such as real-time biosensors, hold the potential for revolutionising TDM by enabling continuous monitoring, immediate and frequent dose adjustments, and near patient testing. The ongoing integration of TDM, advanced modelling techniques, and ML within the evolving digital health care landscape offers a potential for enhancing antimicrobial therapy. Careful attention to model development, validation, and ethical considerations of the applied techniques is paramount for successfully optimising antimicrobial treatment for the individual patient.

Model-informed precision dosing: State of the art and future perspectives

Model-informed precision dosing (MIPD) stands as a significant development in personalized medicine to tailor drug dosing to individual patient characteristics. MIPD moves beyond traditional therapeutic drug monitoring (TDM) by integrating mathematical predictions of dosing, and considering patient-specific factors (patient characteristics, drug measurements) as well as different sources of variability. For this purpose, rigorous model qualification is required for the application of MIPD in patients. This review delves into new methods in model selection and validation, also highlighting the role of machine learning in improving MIPD, the utilization of biosensors for real-time monitoring, as well as the potential of models integrating biomarkers for efficacy or toxicity for precision dosing. The clinical evidence of TDM and MIPD is discussed for various medical fields including infection medicine, oncology, transplant medicine, and inflammatory bowel diseases, thereby underscoring the role of pharmacokinetics/pharmacodynamics and specific biomarkers. Further research, particularly randomized clinical trials, is warranted to corroborate the value of MIPD in enhancing patient outcomes and advancing personalized medicine.

External validation of population pharmacokinetic models of tacrolimus in Thai adult liver transplant recipients

OBJECTIVE

Several population pharmacokinetic models of tacrolimus in liver transplant patients were built, and their predictability was evaluated in their settings. However, the extrapolation in the prediction was unclear. This study aimed to evaluate the predictive performance of published tacrolimus models in adult liver transplant recipients using data from the Thai population as an external dataset.

METHODS

The selected published models were systematically searched and evaluated for their quality. The external dataset of patients who underwent the first liver transplant and received immediate-release tacrolimus was used to assess the predictive performance of each selected model. Trough concentrations between 3 and 6 months were retrospectively collected to evaluate the predictability of each model using prediction-based diagnostics, simulation-based diagnostics, and Bayesian forecasting.

RESULTS

Sixty-seven patients with 360 trough concentrations and eight selected published models were included in this study. None of the models met the predictive precision criteria in prediction-based diagnostics. Meanwhile, four published population pharmacokinetic models showed a normal distribution in NPDE testing. Regarding Bayesian forecasting, all models improved their forecasts with at least one prior information data point.

CONCLUSION

Bayesian forecasting is more accurate and precise than other testing methods for predicting drug concentrations. However, none of the evaluated models provides satisfactory predictive performance for generalization to Thai liver transplant patients. This underscores the need for future research to develop population PK models tailored to the Thai population. Such efforts should consider the inclusion of nonlinear pharmacokinetics and region-specific factors, including genetic variability, to improve model accuracy and applicability.

Dose optimization of β-lactam antibiotics in children: from population pharmacokinetics to individualized therapy

INTRODUCTION

β-Lactams are the most widely used antibiotics in children. Their optimal dosing is essential to maximize their efficacy, while minimizing the risk for toxicity and the further emergence of antimicrobial resistance. However, most β-lactams were developed and licensed long before regulatory changes mandated pharmacokinetic studies in children. As a result, pediatric dosing practices are poorly harmonized and off-label use remains common today.

AREAS COVERED

β-Lactam pharmacokinetics and dose optimization strategies in pediatrics, including fixed dose regimens, therapeutic drug monitoring, and model-informed precision dosing are reviewed.

EXPERT OPINION/COMMENTARY

Standard pediatric doses can result in subtherapeutic exposure and non-target attainment for specific patient subpopulations (neonates, critically ill children, e.g.). Such patients could benefit greatly from more individualized approaches to dose optimization, beyond a relatively simple dose adaptation based on weight, age, or renal function. In this context, Therapeutic Drug Monitoring (TDM) and Model-Informed Precision Dosing (MIPD) emerge as particularly promising avenues. Obstacles to their implementation include the lack of strong evidence of clinical benefit due to the paucity of randomized clinical trials, of standardized assays for monitoring concentrations, or of adequate markers for renal function. The development of precision medicine tools is urgently needed to individualize therapy in vulnerable pediatric subpopulations.

Efficiency of dosing software using Bayesian forecasting in achieving target antibiotic exposures in critically ill patients, a prospective cohort study

INTRODUCTION

Broad-spectrum antibiotics such as beta-lactams and vancomycin are frequently used to treat critically ill patients, however, a significant number do not achieve target exposures. Therapeutic drug monitoring (TDM) combined with Bayesian forecasting dosing software may improve target attainment in these patients. This study aims to describe the efficiency of dosing software for achieving target exposures of selected beta-lactam antibiotics and vancomycin in critically ill patients.

METHODS

A prospective cohort study was undertaken in an adult intensive care unit (ICU). Patients prescribed vancomycin, piperacillin-tazobactam and meropenem were included if they exhibited a subtherapeutic or supratherapeutic exposure informed by TDM. The dosing software, ID-ODS™, was used to generate dosing recommendations which could be either accepted or rejected by the treating team. Repeat antibiotic TDM were requested to determine if target exposures were achieved.

RESULTS

Between March 2020 and December 2021, 70 were included in the analysis. Software recommendations were accepted for 56 patients (80%) with 50 having repeated antibiotic measurements. Forty-three of the 50 patients (86%) achieved target exposures after one software recommendation, with 3 of the remaining 7 patients achieving target exposures after 2. Forty-seven patients out of the 50 patients (94%) achieved the secondary outcome of clinical cure. There were no antibiotic exposure-related adverse events reported.

CONCLUSION

The use of TDM combined with Bayesian forecasting dosing software increases the efficiency for achieving target antibiotic exposures in the ICU. Clinical trials comparing this approach with other dosing strategies are required to further validate these findings.

Unraveling the Effects of Acute Inflammation on Pharmacokinetics: A Model-Based Analysis Focusing on Renal Glomerular Filtration Rate and Cytochrome P450 3A4-Mediated Metabolism

BACKGROUND AND OBJECTIVES: Acute inflammation caused by infections or sepsis can impact pharmacokinetics. We used a model-based analysis to evaluate the effect of acute inflammation as represented by interleukin-6 (IL-6) levels on drug clearance, focusing on renal glomerular filtration rate (GFR) and cytochrome P450 3A4 (CYP3A4)-mediated metabolism.

METHODS

A physiologically based model incorporating renal and hepatic drug clearance was implemented. Functions correlating IL-6 levels with GFR and in vitro CYP3A4 activity were derived and incorporated into the modeling framework. We then simulated treatment scenarios for hypothetical drugs by varying the IL-6 levels, the contribution of renal and hepatic drug clearance, and protein binding. The relative change in observed area under the concentration-time curve (AUC) was computed for these scenarios.

RESULTS

Inflammation showed opposite effects on drug exposure for drugs eliminated via the liver and kidney, with the effect of inflammation being inversely proportional to the extraction ratio (ER). For renally cleared drugs, the relative decrease in AUC was close to 30% during severe inflammation. For CYP3A4 substrates, the relative increase in AUC could exceed 50% for low-ER drugs. Finally, the impact of inflammation-induced changes in drug clearance is smaller for drugs with a larger unbound fraction.

CONCLUSION

This analysis demonstrates differences in the impact of inflammation on drug clearance for different drug types. The effects of inflammation status on pharmacokinetics may explain the inter-individual variability in pharmacokinetics in critically ill patients. The proposed model-based analysis may be used to further evaluate the effect of inflammation, i.e., by incorporating the effect of inflammation on other drug-metabolizing enzymes or physiological processes.

Towards Model-Informed Precision Dosing of Voriconazole: Challenging Published Voriconazole Nonlinear Mixed-Effects Models with Real-World Clinical Data

BACKGROUND AND OBJECTIVES

Model-informed precision dosing (MIPD) frequently uses nonlinear mixed-effects (NLME) models to predict and optimize therapy outcomes based on patient characteristics and therapeutic drug monitoring data. MIPD is indicated for compounds with narrow therapeutic range and complex pharmacokinetics (PK), such as voriconazole, a broad-spectrum antifungal drug for prevention and treatment of invasive fungal infections. To provide guidance and recommendations for evidence-based application of MIPD for voriconazole, this work aimed to (i) externally evaluate and compare the predictive performance of a published so-called 'hybrid' model for MIPD (an aggregate model comprising features and prior information from six previously published NLME models) versus two 'standard' NLME models of voriconazole, and (ii) investigate strategies and illustrate the clinical impact of Bayesian forecasting for voriconazole.

METHODS

A workflow for external evaluation and application of MIPD for voriconazole was implemented. Published voriconazole NLME models were externally evaluated using a comprehensive in-house clinical database comprising nine voriconazole studies and prediction-/simulation-based diagnostics. The NLME models were applied using different Bayesian forecasting strategies to assess the influence of prior observations on model predictivity.

RESULTS

The overall best predictive performance was obtained using the aggregate model. However, all NLME models showed only modest predictive performance, suggesting that (i) important PK processes were not sufficiently implemented in the structural submodels, (ii) sources of interindividual variability were not entirely captured, and (iii) interoccasion variability was not adequately accounted for. Predictive performance substantially improved by including the most recent voriconazole observations in MIPD.

CONCLUSION

Our results highlight the potential clinical impact of MIPD for voriconazole and indicate the need for a comprehensive (pre-)clinical database as basis for model development and careful external model evaluation for compounds with complex PK before their successful use in MIPD.

Towards Precision Dosing of Clozapine in Schizophrenia: External Evaluation of Population Pharmacokinetic Models and Bayesian Forecasting

BACKGROUND

Therapeutic drug monitoring and treatment optimization of clozapine are recommended, owing to its narrow therapeutic range and pharmacokinetic (PK) variability. This study aims to assess the clinical applicability of published population PK models by testing their predictive performance in an external data set and to determine the effectiveness of Bayesian forecasting (BF) for clozapine treatment optimization.

METHODS

Available models of clozapine were identified, and their predictive performance was determined using an external data set (53 patients, 151 samples). The median prediction error (PE) and median absolute PE were used to assess bias and inaccuracy. The potential factors influencing model predictability were also investigated. The final concentration was reestimated for all patients using covariates or previously observed concentrations.

RESULTS

The 7 included models presented limited predictive performance. Only 1 model met the acceptability criteria (median PE ≤ ±20% and median absolute PE ≤30%). There was no difference between the data used for building the models (therapeutic drug monitoring or PK study) or the number of compartments in the models. A tendency for higher inaccuracy at low concentrations during treatment initiation was observed. Heterogeneities were observed in the predictive performances between the subpopulations, especially in terms of smoking status and sex. For the models included, BF significantly improved their predictive performance.

CONCLUSIONS

Our study showed that upon external evaluation, clozapine models provide limited predictive performance, especially in subpopulations such as nonsmokers. From the perspective of model-informed prediction dosing, model predictability should be improved using updating or metamodeling methods. Moreover, BF substantially improved model predictability and could be used for clozapine treatment optimization.

External Evaluation of Population Pharmacokinetic Models of Busulfan in Chinese Adult Hematopoietic Stem Cell Transplantation Recipients

Objective: Busulfan (BU) is a bi-functional DNA-alkylating agent used in patients undergoing hematopoietic stem cell transplantation (HSCT). Over the last decades, several population pharmacokinetic (pop PK) models of BU have been established, but external evaluation has not been performed for almost all models. The purpose of the study was to evaluate the predictive performance of published pop PK models of intravenous BU in adults using an independent dataset from Chinese HSCT patients, and to identify the best model to guide personalized dosing.

Methods: The external evaluation methods included prediction-based diagnostics, simulation-based diagnostics, and Bayesian forecasting. In prediction-based diagnostics, the relative prediction error (PE%) was calculated by comparing the population predicted concentration (PRED) with the observations. Simulation-based diagnostics included the prediction- and variability-corrected visual predictive check (pvcVPC) and the normalized prediction distribution error (NPDE). Bayesian forecasting was executed by giving prior one to four observations. The factors influencing the model predictability, including the impact of structural models, were assessed.

Results: A total of 440 concentrations (110 patients) were obtained for analysis. Based on prediction-based diagnostics and Bayesian forecasting, preferable predictive performance was observed in the model developed by Huang et al. The median PE% was -1.44% which was closest to 0, and the maximum F₂₀ of 57.27% and F₃₀ of 72.73% were achieved. Bayesian forecasting demonstrated that prior concentrations remarkably improved the prediction precision and accuracy of all models, even with only one prior concentration.

Conclusion: This is the first study to comprehensively evaluate published pop PK models of BU. The model built by Huang et al. had satisfactory predictive performance, which can be used to guide individualized dosage adjustment of BU in Chinese patients.

External evaluation of population pharmacokinetic models of imatinib in adults diagnosed with chronic myeloid leukaemia

AIMS

Imatinib is considered the standard first-line treatment in newly diagnosed patients with chronic-phase myeloid leukaemia (CML). Several imatinib population pharmacokinetic (popPK) models have been developed. However, their predictive performance has not been well established when extrapolated to different populations. Therefore, this study aimed to perform an external evaluation of available imatinib popPK models developed mainly in adult patients, and to evaluate the improvement in individual model-based predictions through Bayesian forecasting computed by each model at different treatment occasions.

METHODS

A literature review was conducted through PubMed and Scopus to identify popPK models. Therapeutic drug monitoring data collected in adult CML patients treated with imatinib was used for external evaluation, including prediction- and simulated-based diagnostics together with Bayesian forecasting analysis.

RESULTS

Fourteen imatinib popPK studies were included for model-performance evaluation. A total of 99 imatinib samples were collected from 48 adult CML patients undergoing imatinib treatment with a minimum of one plasma concentration measured at steady-state between January 2016 and December 2020. The model proposed by Petain et al showed the best performance concerning prediction-based diagnostics in the studied population. Bayesian forecasting demonstrated a significant improvement in predictive performance at the second visit. Inter-occasion variability contributed to reducing bias and improving individual model-based predictions.

CONCLUSIONS

Imatinib popPK studies developed in Caucasian subjects including α1-acid glycoprotein showed the best model performance in terms of overall bias and precision. Moreover, two imatinib samples from different visits appear sufficient to reach an adequate model-based individual prediction performance trough Bayesian forecasting.

Can Published Population Pharmacokinetic Models of Busulfan Be Used for Individualized Dosing in Chinese Pediatric Patients Undergoing Hematopoietic Stem Cell Transplantation? An External Evaluation

Busulfan is a bifunctional alkylating agent that is widely used before hematopoietic stem cell transplantation (HSCT), in combination with other chemotherapeutic drugs. As of 2020, there is no population pharmacokinetic (popPK) model for busulfan in Chinese pediatric patients. A systemic external evaluation of 11 published popPK models was conducted in Chinese pediatric patients undergoing HSCT. Forty pediatric patients were enrolled in this study, with a total of 183 blood concentrations. The relative prediction error (PE%), median PE%, median absolute PE%, and percentage of PE% within ±20% and ±30% were calculated in prediction-based diagnostics. Simulation-based diagnostics were conducted through a prediction- and variability-corrected visual predictive check and the normalized prediction distribution error. The relative individual prediction error was calculated using Bayesian forecasting with 1 to 3 concentration points. The 1-compartment open linear popPK model, which was built by Su-jin Rhee et al (model H), incorporating the patient's body surface area, age, dosing day, and aspartate aminotransferase as significant covariates had preferable predictability than other popPK models. In prediction-based diagnostics, the median PE%, percentage of PE% within ±20%, and percentage of PE% within ±30% of model H were 8.48%, 45.35%, and 59.56%, respectively. The normalized prediction distribution error of model H showed that it followed the normal distribution. Based on Bayesian forecasting, model H showed good predictive performance. Thus, model H was the most appropriate model that can be used clinically for individualized dosage adjustments in Chinese pediatric HSCT patients.

From Therapeutic Drug Monitoring to Model-Informed Precision Dosing for Antibiotics

Therapeutic drug monitoring (TDM) and model-informed precision dosing (MIPD) have evolved as important tools to inform rational dosing of antibiotics in individual patients with infections. In particular, critically ill patients display altered, highly variable pharmacokinetics and often suffer from infections caused by less susceptible bacteria. Consequently, TDM has been used to individualize dosing in this patient group for many years. More recently, there has been increasing research on the use of MIPD software to streamline the TDM process, which can increase the flexibility and precision of dose individualization but also requires adequate model validation and re-evaluation of existing workflows. In parallel, new minimally invasive and noninvasive technologies such as microneedle-based sensors are being developed, which-together with MIPD software-have the potential to revolutionize how patients are dosed with antibiotics. Nonetheless, carefully designed clinical trials to evaluate the benefit of TDM and MIPD approaches are still sparse, but are critically needed to justify the implementation of TDM and MIPD in clinical practice. The present review summarizes the clinical pharmacology of antibiotics, conventional TDM and MIPD approaches, and evidence of the value of TDM/MIPD for aminoglycosides, beta-lactams, glycopeptides, and linezolid, for which precision dosing approaches have been recommended.