A Model Averaging/Selection Approach Improves the Predictive Performance of Model‐Informed Precision Dosing: Vancomycin as a Case Study

Abstract

Many important drugs exhibit substantial variability in pharmacokinetics and pharmacodynamics leading to a loss of the desired clinical outcomes or significant adverse effects. Forecasting drug exposures using pharmacometric models can improve individual target attainment when compared with conventional therapeutic drug monitoring (TDM). However, selecting the 'correct' model for this model-informed precision dosing (MIPD) is challenging. We derived and evaluated a model selection algorithm (MSA) and a model averaging algorithm (MAA), which automates model selection and finds the best model or combination of models for each patient using vancomycin as a case study, and implemented both algorithms in the MIPD software 'TDMx'. The predictive performance (based on accuracy and precision) of the two algorithms was assessed in (i) a simulation study of six distinct populations and (ii) a clinical dataset of 180 patients undergoing TDM during vancomycin treatment and compared with the performance obtained using a single model. Throughout the six virtual populations the MSA and MAA (imprecision: 9.9-24.2%, inaccuracy: less than ±8.2%) displayed more accurate predictions than the single models (imprecision: 8.9-51.1%; inaccuracy: up to 28.9%).In the clinical dataset the predictive performance of the single models applying at least one plasma concentration varied substantially (imprecision: 28-62%, inaccuracy:-16-25%), while the MSA or MAA utilizing these models simultaneously, resulted in unbiased and precise predictions (imprecision: 29% and 30%, inaccuracy:-5% and 0%, respectively). MSA and MAA approaches implemented in 'TDMx' might thereby lower the burden of fit-for-purpose validation of individual models and streamline MIPD.

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

“…Using the example of vancomycin dosing in a pediatric ICU population, the authors demonstrate that such an approach can reduce the prediction error significantly by up to 13% compared with the original model. Uster and colleagues proposed an automated model averaging/selection approach 50 . This algorithm uses a number of candidate models, some of which may be mis‐specified for a specific patient.…”

“…The algorithm selects the best‐fitting model (model selection) or a combination of models (model averaging) for an individual patient when TDM data become available in the course of therapy. When using this approach for heterogenous data sets for vancomycin in general ward and ICU populations, 47,51 the predictive performance was better after model averaging than for the best single model in previous external evaluations 47,50,51 …”

Key Challenges in Providing Effective Antibiotic Therapy for Critically Ill Patients with Bacterial Sepsis and Septic Shock

“…A number of dose optimization software programs based on traditional PK/PD targets and population PK models are currently available for clinical use and comparisons of their performance in informing individualized antibiotic dosing have been published. [113][114][115][116][117] Currently, mCL CR (i.e., using urinary excretion and a mid-point serum concentration) is the suggested method for daily determinations of renal function in critically ill patients. 28 However, as discussed above, that approach, like all methods that incorporate serum creatinine concentration in their determination, is affected by the lag in serum creatinine concentrations after a change in the actual GFR.…”

“…Those parameter estimates are then utilized to compute personalized modifications to the dosing regimen to optimize antibiotic exposure in the treated patient. A number of dose optimization software programs based on traditional PK/PD targets and population PK models are currently available for clinical use and comparisons of their performance in informing individualized antibiotic dosing have been published 113–117 …”

Precision Dosing: The Clinical Pharmacology of Goldilocks

“…Plasma exposure‐based dose adjustments are widely used to maximise the chance of vancomycin efficacy and avoid its serious adverse effects. Automated tools to select the most suitable pharmacokinetic model improve the clinical benefit from exposure guided vancomycin dosing 14 . Setting up an advisory service to guide prescribers 15 or the use of modelling tools integrated seamlessly into the electronic medical record and electronic prescribing system 16 removes much of the burden from the treating clinician to make implementation much easier.…”