Towards model-informed precision dosing of piperacillin: multicenter systematic external evaluation of pharmacokinetic models in critically ill adults with a focus on Bayesian forecasting

Predictive performance of multi-model approaches for model-informed precision dosing of piperacillin in critically ill patients

OBJECTIVES

Piperacillin (PIP)/tazobactam is a frequently prescribed antibiotic; however, over- or underdosing may contribute to toxicity, therapeutic failure, and development of antimicrobial resistance. An external evaluation of 24 published PIP-models demonstrated that model-informed precision dosing (MIPD) can enhance target attainment. Employing various candidate models, this study aimed to assess the predictive performance of different MIPD-approaches comparing (i) a single-model approach, (ii) a model selection algorithm (MSA) and (iii) a model averaging algorithm (MAA).

METHODS

Precision, accuracy and expected target attainment, considering either initial (B1) or initial and secondary (B2) therapeutic drug monitoring (TDM)-samples per patient, were assessed in a multicentre dataset (561 patients, 11 German centres, 3654 TDM-samples).

RESULTS

The results demonstrated a slight superiority in predictive performance using MAA in B1, regardless of the candidate models, compared to MSA and the best single models (MAA, MSA, best single models: inaccuracy ±3%, ±10%, ±8%; imprecision: <25%, <31%, <28%; expected target attainment >77%, >71%, >73%). The inclusion of a second TDM-sample notably improved precision and target attainment for all MIPD-approaches, particularly within the context of MSA and most of the single models. The expected target attainment is maximized (up to >90%) when the TDM-sample is integrated within 24 h.

CONCLUSIONS

In conclusion, MAA streamlines MIPD by reducing the risk of selecting an inappropriate model for specific patients. Therefore, MIPD of PIP using MAA implicates further optimisation of antibiotic exposure in critically ill patients, by improving predictive performance with only one sample available for Bayesian forecasting, safety, and usability in clinical practice.

Pharmacokinetics of Piperacillin-Tazobactam in Critically Ill Patients with Open Abdomen and Vacuum-Assisted Wound Closure: Dosing Considerations Using Monte Carlo Simulation

BACKGROUND

Open abdomen with vacuum-assisted wound closure therapy (OA/VAC) is frequently used in critically ill patients although the impact of OA/VAC on antibiotics pharmacokinetics (PK) remains unknown. We thus aimed to characterize the PK of piperacillin-tazobactam (PTZ) in critically ill patients with OA/VAC and assess the optimal dosing regimens based on pharmacodynamics (PD) target attainment.

METHODS

Over a 15-month study period, 45 patients with OA/VAC treated with PTZ administered continuously and adapted to 24 h creatinine clearance (CLCR) underwent measurements of free concentrations in their plasma, urine, VAC exudate, and peritoneal fluid. Population PK modeling was performed considering the effect of covariates, and Monte Carlo simulations were employed to determine the probability of target attainment (PTA) for the PK/PD targets (100% fT > 16 mg/L) in the plasma and at the peritoneal site at steady state.

RESULTS

Piperacillin concentrations were described using a two-compartment model, with age and total body weight as significant covariates for central volume of distribution (V1) and estimated renal function for clearance (CL). Tazobactam concentrations were described using a two-compartment model with estimated renal function as a significant covariate. The central volume of distributions V1 of piperacillin and tazobactam were 21.2 and 23.2 L, respectively. The VAC-induced peritoneal clearance was negligible compared to renal clearance. Most patients achieved the desirable PK/PD target when using a CLCR-pondered PTZ dosing regimen from 12 g/1.5 g/day to 20 g/2.5 g/day.

CONCLUSIONS

Despite a wide inter-individual variability, the influence of OA/VAC on piperacillin and tazobactam PK parameters is not straightforward. The use of a CLCR-pondered PTZ dosing regimen from 12 g/1.5 g/day to 20 g/2.5 g/day is needed to reach a PTA > 85%.

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.

Is it time to recommend AUC-based vancomycin therapeutic drug monitoring only? A cross-sectional survey in China

Background

The latest published therapeutic drug monitoring (TDM) guidelines for vancomycin recommend changing trough-based monitoring to area under the concentration-to-time curve (AUC)-based monitoring. This study aimed to evaluate the implementation status and perceptions of vancomycin AUC-based TDM in China and to determine the challenges in performing AUC-based TDM.

Methods

A nationwide cross-sectional survey was conducted in China using an online questionnaire. The questionnaire comprised a total of 25 questions with open- and closed-ended answers to collect information about the current implementation of vancomycin TDM and the participants' perceptions of these practices. The questionnaire responses were collected via the Questionnaire Star platform and analyzed.

Results

A total of 161 questionnaires were completed by 131 hospitals and were included. Approximately 59.5% (78/131) of the surveyed hospitals conducted vancomycin TDM; however, only 10.7% (14/131) of these hospitals performed AUC-based vancomycin TDM. Of the eligible participants, 58.4% (94/161) had experience with vancomycin TDM, and only 37 participants (37/161, 23.0%) had the ability to estimate the AUC, primarily through Bayesian simulation (33/161, 20.5%). The participants considered the following challenges to implementing AUC-based monitoring: (1) the high cost of AUC-based monitoring; (2) inadequate knowledge among pharmacists and/or physicians; (3) the complexity of AUC calculations; (4) difficulty obtaining AUC software; and (5) unclear benefit of AUC-based monitoring.

Conclusion

The majority of surveyed hospitals have not yet implemented AUC-based vancomycin TDM. Multiple challenges should be addressed before wide implementation of AUC-based monitoring, and guidance for trough-based monitoring is still needed.

Antibiotic combinations prediction based on machine learning to multicentre clinical data and drug interaction correlation

BACKGROUND

With increasing antibiotic resistance and regulation, the issue of antibiotic combination has been emphasised. However, antibiotic combination prescribing lacks a rapid identification of feasibility, while its risk of drug interactions is unclear.

METHODS

We conducted statistical descriptions on 16 101 antibiotic coprescriptions for inpatients with bacterial infections from 2015 to 2023. By integrating the frequency and effectiveness of prescriptions, we formulated recommendations for the feasibility of antibiotic combinations. Initially, a machine learning algorithm was utilised to optimise grading thresholds and habits for antibiotic combinations. A feedforward neural network (FNN) algorithm was employed to develop antibiotic combination recommendation model (ACRM). To enhance interpretability, we combined sequential methods and DrugBank to explore the correlation between antibiotic combinations and drug interactions.

RESULTS

A total of 55 antibiotics, covering 657 empirical clinical antibiotic combinations were used for ACRM construction. Model performance on the test dataset showed AUROCs of 0.589-0.895 for various antibiotic recommendation classes. The ACRM showed satisfactory clinical relevance with 61.54-73.33% prediction accuracy in a new independent retrospective cohort. Antibiotic interaction detection showed that the risk of drug interactions was 29.2% for strongly recommended and 43.5% for not recommended. A positive correlation was identified between the level of clinical recommendation and the risk of drug interactions.

CONCLUSIONS

Machine learning modelling of retrospective antibiotic prescriptions habits has the potential to predict antibiotic combination recommendations. The ACRM plays a supporting role in reducing the incidence of drug interactions. Clinicians are encouraged to adopt such systems to improve the management of antibiotic usage and medication safety.

Achievement of therapeutic antibiotic exposures using Bayesian dosing software in critically unwell children and adults with sepsis

PURPOSE

Early recognition and effective treatment of sepsis improves outcomes in critically ill patients. However, antibiotic exposures are frequently suboptimal in the intensive care unit (ICU) setting. We describe the feasibility of the Bayesian dosing software Individually Designed Optimum Dosing Strategies (ID-ODS™), to reduce time to effective antibiotic exposure in children and adults with sepsis in ICU.

METHODS

A multi-centre prospective, non-randomised interventional trial in three adult ICUs and one paediatric ICU. In a pre-intervention Phase 1, we measured the time to target antibiotic exposure in participants. In Phase 2, antibiotic dosing recommendations were made using ID-ODS™, and time to target antibiotic concentrations were compared to patients in Phase 1 (a pre-post-design).

RESULTS

175 antibiotic courses (Phase 1 = 123, Phase 2 = 52) were analysed from 156 participants. Across all patients, there was no difference in the time to achieve target exposures (8.7 h vs 14.3 h in Phase 1 and Phase 2, respectively, p = 0.45). Sixty-one courses in 54 participants failed to achieve target exposures within 24 h of antibiotic commencement (n = 36 in Phase 1, n = 18 in Phase 2). In these participants, ID-ODS™ was associated with a reduction in time to target antibiotic exposure (96 vs 36.4 h in Phase 1 and Phase 2, respectively, p < 0.01). These patients were less likely to exhibit subtherapeutic antibiotic exposures at 96 h (hazard ratio (HR) 0.02, 95% confidence interval (CI) 0.01-0.05, p < 0.01). There was no difference observed in in-hospital mortality.

CONCLUSIONS

Dosing software may reduce the time to achieve target antibiotic exposures. It should be evaluated further in trials to establish its impact on clinical outcomes.

Model-Informed Precision Dosing of Imipenem in an Obese Adolescent Patient with Augmented Renal Clearance and History of Schizophrenia

Imipenem is a broad-spectrum antibiotic that has been used in treating severe infections and exhibits a time-dependent PK/PD profile. Its dose should be adjusted based on renal function. However, there is little experience with imipenem dosing in obese adolescent patients with augmented renal clearance (ARC) and history of schizophrenia. This case reported successful dosing of imipenem in an obese adolescent patient with ARC based on therapeutic drug monitoring (TDM) and model-informed precision dosing (MIPD). A 15-year-old male adolescent patient with history of schizophrenia was diagnosed with ventilator-associated pneumonia due to carbapenem-susceptible Klebsiella pneumoniae and received imipenem treatment (0.5 g every 8 hours with a 1-hour infusion). However, the exposure of imipenem was suboptimal due to ARC, and there is no available model for MIPD in this patient. Thus, we utilized prediction error to find a population pharmacokinetic model that fit this patient and ran Maximum a posteriori Bayesian estimation and Monte Carlo simulation based on screened models to predict changes in drug concentrations. The dose of imipenem was adjusted to 0.5 g every 6 hours with a 2-hour infusion, and subsequent TDM revealed that dosing adjustment was accurate and successful. Finally, the patient's status of infection improved. This study will be beneficial to imipenem dosing in similar cases in the future, thereby improving the safety and effectiveness of imipenem or other antibiotics.

Covariates in population pharmacokinetic studies of critically ill adults receiving β-lactam antimicrobials: a systematic review and narrative synthesis

Introduction

Population pharmacokinetic studies of β-lactam antimicrobials in critically ill patients derive models that inform their dosing. In non-linear mixed-effects modelling, covariates are often used to improve model fit and explain variability. We aimed to investigate which covariates are most commonly assessed and which are found to be significant, along with global patterns of publication.

Methods

We conducted a systematic review, searching MEDLINE, Embase, CENTRAL and Web of Science on 01 March 2023, including studies of critically ill adults receiving β-lactam antimicrobials who underwent blood sampling for population pharmacokinetic studies. We extracted and categorized all reported covariates and assessed reporting quality using the ClinPK checklist.

Results

Our search identified 151 studies with 6018 participants. Most studies reported observational cohorts (120 studies, 80%), with the majority conducted in high-income settings (136 studies, 90%). Of the 1083 identified covariate instances, 237 were unique; the most common categories were patient characteristics (n = 404), biomarkers (n = 206) and physiological parameters (n = 163). Only seven distinct commonly reported covariates (CLCR, weight, glomerular filtration rate, diuresis, need for renal replacement, serum albumin and C-reactive protein) were significant more than 20% of the time.

Conclusions

Covariates are most commonly chosen based on biological plausibility, with patient characteristics and biomarkers the most frequently investigated. We developed an openly accessible database of reported covariates to aid investigators with covariate selection when designing population pharmacokinetic studies. Novel covariates, such as sepsis subphenotypes, have not been explored yet, leaving a research gap for future work.

Anti-infective prescribing practices in critically ill children on continuous renal replacement therapy: a multicenter survey of French-speaking countries

BACKGROUND

Use of continuous renal replacement therapy in children receiving anti-infective drugs may lead to inappropriate concentrations with risks related to treatment failure, toxicity and emergence of multidrug-resistant bacteria. We aimed to describe anti-infective prescribing practices in critically ill children undergoing continuous renal replacement therapy.

METHODS

An online survey to assess continuous renal replacement therapy, anti-infective prescribing and therapeutic drug monitoring practices was sent by e-mail to physicians working in pediatric intensive care units through the French-speaking Group of Pediatric Intensive Care and Emergency medicine (GFRUP).

RESULTS

From April 1st, 2021 to May 1st, 2021, 26/40 pediatric intensive care units participated in the survey, corresponding to a response rate of 65%. Twenty-one were located in France and five abroad. All pediatric intensive care units administered continuous renal replacement therapy, primarily with Prismaflex™ System. Anti-infective prescriptions were adjusted to the presence of continuous renal replacement therapy in 23 (88%) pediatric intensive care units mainly according to molecular weight in 6 (23%), molecule protein binding in 6 (23%) and elimination routes in 15 (58%) including residual diuresis in 9 (35%), to the continuous renal replacement therapy flow in 6 (23%) and to the modality of continuous renal replacement therapy used in 15 (58%), pediatric intensive care units. There was broad variability among pediatric intensive care units and among physicians within the same unit. Barriers to therapeutic drug monitoring were mainly an excessive delay in obtaining results in 11 (42%) and the lack of an on-site laboratory in 8 (31%) pediatric intensive care units.

CONCLUSIONS

Our survey reported wide variability in anti-infective prescribing practices in children undergoing continuous renal replacement therapy, thus highlighting a gap in knowledge and the need for education and recommendations.