ABSTRACT Tobramycin dosing in patients with cystic fibrosis (CF) is challenged by its high pharmacokinetic (PK) variability and narrow therapeutic window. Doses are typically individualized using two-sample log-linear regression (LLR) to quantify the area under the concentration-time curve (AUC). Bayesian model-informed precision dosing (MIPD) may allow dose individualization with fewer samples; however, the relative performance of these methods is unknown. This single-center retrospective analysis included adult patients with CF receiving tobramycin from 2015 to 2022. Tobramycin concentrations were predicted using LLR or Bayesian estimation with two population PK models (Hennig and Alghanem). Then, both methods were used to estimate the AUC for simulated patients. For Bayesian estimation, AUC estimation with flattened priors and limited sampling strategies were also assessed. Predictions were evaluated using normalized root mean square error (nRMSE), mean percent error (MPE), and accuracy. The data set included 70 treatment courses, with 32 not evaluable by LLR due to detection limits or timing issues. Bayesian estimation demonstrated worse accuracy (47.1%–50.7% vs 75.7%), higher MPE (24.2%–32.4% vs −2.4%), and higher nRMSE (35.0%–39.4% vs 24.8%) than LLR for peak concentrations but performed better on troughs (accuracy: 92.0%–92.9% vs 84.6%). Bayesian estimation with flattened priors and a single sample at 4 h was comparable to LLR performance, with better accuracy (42.9%–68.0% vs 41.1% LLR), comparable MPE (−2.3% to −3.7% vs −0.5%) and nRMSE (11.3%–21.6% vs 17.3%). Bayesian estimation with one concentration and flattened priors can match LLR prediction accuracy. However, popPK models must be improved to better estimate peak samples.
Background: Both parametric and nonparametric methods have been proposed to support model-informed precision dosing (MIPD). However, which approach leads to better models remains uncertain. Using open-source software, these 2 statistical approaches for model development were compared using the pharmacokinetics of vancomycin in a challenging subpopulation of class 3 obesity. Methods: Patients on vancomycin at the University of Vermont Medical Center from November 1, 2021, to February 14, 2023, were entered into the MIPD software. The inclusion criteria were body mass index (BMI) of at least 40 kg/m 2 and 1 or more vancomycin levels. A parametric model was created using nlmixr2/NONMEM, and a nonparametric model was created using Pmetrics. Then, a priori and a posteriori predictions were evaluated using the normalized root mean squared error (nRMSE) for precision and the mean percentage error (MPE) for bias. The parametric model was evaluated in a simulated MIPD context using an external validation dataset. Results: In total, 83 patients were included in the model development, with a median age of 56.6 years (range: 24–89 years), and a median BMI of 46.3 kg/m 2 (range: 40–70.3 kg/m 2 ). Both parametric and nonparametric models were 2-compartmental, with creatinine clearance and fat-free mass as covariates to clearance and volume parameters, respectively. The a priori MPE and nRMSE for the parametric versus nonparametric models were −6.3% versus 2.69% and 27.2% versus 30.7%, respectively. The a posteriori MPE and RMSE were 0.16% and 0.84%, and 13.8% and 13.1%. The parametric model matched or outperformed previously published models on an external validation dataset (n = 576 patients). Conclusions: Minimal differences were found in the model structure and predictive error between the parametric and nonparametric approaches for modeling vancomycin class 3 obesity. However, the parametric model outperformed several other models, suggesting that institution-specific models may improve pharmacokinetics management.
Model-informed precision dosing (MIPD) approaches typically apply maximum a posteriori (MAP) Bayesian estimation to determine individual pharmacokinetic (PK) parameters with the goal of optimizing future dosing regimens. This process combines knowledge about the individual, in the form of drug levels or pharmacodynamic biomarkers, with prior knowledge of the drug PK in the general population. Use of "flattened priors" (FPs), in which the weight of the model priors is reduced relative to observations about the patient, has been previously proposed to estimate individual PK parameters in instances where the patient is poorly described by the PK model. However, little is known about the predictive performance of FPs and when to apply FPs in MIPD. Here, FP is evaluated in a data set of 4679 adult patients treated with vancomycin. Depending on the PK model, prediction error could be reduced by applying FPs in 42-55% of PK parameter estimations. Machine learning (ML) models could identify instances where FPs would outperform MAPs with a specificity of 81-86%, reducing overall root mean squared error (RMSE) of PK model predictions by 12-22% (0.5-1.2 mg/L) relative to MAP alone. The factors most indicative of the use of FPs were past prediction residuals and bias in past PK predictions. A more clinically practical minimal model was developed using only these two features, reducing RMSE by 5-18% (0.20-0.93 mg/L) relative to MAP. This hybrid ML/PK approach advances the precision dosing toolkit by leveraging the power of ML while maintaining the mechanistic insight and interpretability of PK models.
Precision dosing aims to tailor doses to individual patients with the goal of improving treatment efficacy and avoiding toxicity. Clinical decision support software (CDSS) plays a crucial role in mediating this process, translating knowledge derived from clinical trials and real-world data (RWD) into actionable insights for clinicians to use at the point of care. However, not all patient populations are proportionally represented in clinical trials and other data sources that inform CDSS tools, limiting the applicability of these tools for underrepresented populations. Here, we review some of the limitations of existing CDSS tools and discuss methods for overcoming these gaps. We discuss considerations for study design and modeling to create more inclusive CDSS, particularly with an eye toward better incorporation of biological indicators in place of race, ethnicity, or sex. We also review inclusive practices for collection of these demographic data, during both study design and in software user interface design. Because of the role CDSS plays in both recording routine clinical care data and disseminating knowledge derived from data, CDSS presents a promising opportunity to continuously improve precision dosing algorithms using RWD to better reflect the diversity of patient populations.
Abstract Background Model‐informed precision dosing (MIPD) optimizes drug doses based on pharmacokinetic (PK) model predictions, necessitating careful selection of models tailored to patient characteristics. This study evaluates the predictive performance of various vancomycin PK models across diverse age and BMI categories, drawing insights from a large multi‐site database. Methods Adults receiving vancomycin intravenous therapy at United States health systems between January 1, 2022, and December 31, 2023, were included. Patient demographics, vancomycin administration records, and therapeutic drug monitoring levels (TDMs) were collected from the InsightRX database. Age and body mass index (BMI)‐based subgroups were formed to assess model performance, with predictions made iteratively. The optimal model for each age‐BMI subgroup was chosen based on predefined criteria: models were filtered for mean percentage error (MPE) ≤ 20% and normalized root mean squared error (RMSE) < 8 mg/L, and then the most accurate among them was selected. Results A total of 384,876 treatment courses across 155 US health systems were analyzed, contributing 841,604 TDMs. Eleven models were compared, showing varying accuracy across age‐BMI categories (41%–73%), with higher accuracy observed once TDMs were available for Bayesian estimates of individual PK parameters. Models performed more poorly in younger adults compared to older adults, and the optimal model differed depending on age‐BMI categories and prediction methods. Notably, in the a priori period, the Colin model performed best in adults aged 18–64 years across most BMI categories; the Goti/Tong model performed best in the older, non‐obese adults; and the Hughes model performed best in many of the obese categories. Conclusion Our study identifies specific vancomycin PK models that demonstrate superior predictions across age‐BMI categories in MIPD applications. Our findings underscore the importance of tailored model selection for vancomycin management, especially highlighting the need for improved models in younger adult patients. Further research into the clinical implications of model performance is warranted to enhance patient care outcomes.
Dose personalization improves patient outcomes for many drugs with a narrow therapeutic index and high inter-individuality variability, including busulfan. Non-compartmental analysis (NCA) and model-based methods like maximum a posteriori Bayesian (MAP) approaches are two methods routinely used for dose optimization. These approaches vary in how they estimate patient-specific pharmacokinetic parameters to inform a dose and the impact of these differences is not well-understood. Using busulfan as an example application and area under the concentration-time curve (AUC) as a target exposure metric, these estimation methods were compared using retrospective patient data (N = 246) and simulated precision dosing treatment courses. NCA was performed with or without peak extension, and MAP Bayesian estimation was performed using either the one-compartment Shukla model or the two-compartment McCune model. All methods showed good agreement on real-world data (correlation coefficients of 0.945-0.998) as assessed by Bland-Altman plots, although agreement between NCA and MAP methods was higher during the first dosing interval (0.982-0.994) compared to subsequent dosing intervals (0.918-0.938). In dose adjustment simulations, both NCA and MAP estimated high target attainment (> 98%) although true simulated target attainment was lower for NCA (63-66%) versus MAP (91-93%). The largest differences in AUC estimation were due to different assumptions for the shape of the concentration curve during the infusion phase, followed by how the methods considered time-dependent clearance and concentration-time points collected in earlier intervals. In conclusion, although AUC estimates between the two methods showed good correlation, in a simulated study, MAP lead to higher target attainment. When changing from one method to another, or changing infusion duration and other factors, optimum estimated exposure targets may require adjusting to maintain a consistent exposure.
