Vancomycin area under the curves estimated with pharmacokinetic equations using trough-only data

Machine learning and population pharmacokinetics: a hybrid approach for optimizing vancomycin therapy in sepsis patients

Predicting vancomycin exposure is essential for optimizing dosing regimens in sepsis patients. While population pharmacokinetic (PPK) models are commonly used, their performance is limited. Machine learning (ML) models offer advantages over PPK models, but it remains unclear which model-PPK, Bayesian, ML, or hybrid PPK-ML-is best for predicting vancomycin exposure across different clinical scenarios in sepsis patients. This study compares the performance of these models in predicting the 24 hour area under the blood concentration curve (AUC24) to support precision dosing in sepsis care. Data from sepsis patients treated with intravenous vancomycin were sourced from the MIMIC-IV database. The data set was split into training and testing sets, and four models-PPK, Bayesian, ML, and hybrid-were developed. In the testing set, AUC24 was predicted using all models, and performance was evaluated using mean absolute error, mean squared error, root mean squared error, mean absolute percentage error (MAPE), and R². A total of 4,059 patients were included. In the absence of vancomycin concentration data, the hybrid model outperformed both PPK and Bayesian models, with MAPE improvements of 58% and 17%, respectively. When vancomycin concentration data were available, the Bayesian model demonstrated the best performance (MAPE: 13.37% vs 68.17%, 34.17%, and 28.52% for PPK, Random Forest, and hybrid models). The hybrid model is recommended to predict AUC24 when concentration data were unavailable, while the Bayesian model should be used when concentrations were available, offering robust strategies for precise vancomycin dosing in sepsis patients.

IMPORTANCE

This study evaluates and compares the performance of four models-PPK, Bayesian, ML, and hybrid PPK-ML-in predicting vancomycin exposure (AUC24) in sepsis patients using real-world data from the MIMIC-IV database. These results underscore the importance of selecting appropriate models based on the availability of concentration data, providing valuable guidance for precision dosing strategies in sepsis care. This work contributes to advancing personalized vancomycin therapy, optimizing dosing regimens, and improving clinical outcomes in sepsis patients.

Automated calculation and reporting of vancomycin area under the concentration-time curve: a simplified single-trough concentration-based equation approach

Vancomycin, a crucial antibiotic for Gram-positive bacterial infections, requires therapeutic drug monitoring (TDM). Contemporary guidelines advocate for AUC-based monitoring; however, using Bayesian programs for AUC estimation poses challenges. We aimed to develop and evaluate a simplified AUC estimation equation using a steady-state trough concentration (Ctrough) value. Utilizing 1,034 TDM records from 580 general hospitalized patients at a university-affiliated hospital in Ulsan, we created an equation named SSTA that calculates the AUC by applying Ctrough, body weight, and single dose as input variables. External validation included 326 records from 163 patients at a university-affiliated hospital in Seoul (EWUSH) and literature data from 20 patients at a university-affiliated hospital in Bangkok (MUSI). It was compared with other AUC estimation models based on the Ctrough, including a linear regression model (LR), a sophisticated model based on the first-order equation (VancoPK), and a Bayesian model (BSCt). Evaluation metrics, such as median absolute percentage error (MdAPE) and the percentage of observations within ±20% error (P20), were calculated. External validation using the EWUSH data set showed that SSTA, LR, VancoPK, and BSCt had MdAPE values of 6.4, 10.1, 6.6, and 7.5% and P20 values of 87.1, 82.5, 87.7, and 83.4%, respectively. External validation using the MUSI data set showed that SSTA, LR, and VancoPK had MdAPEs of 5.2, 9.4, and 7.2%, and P20 of 95, 90, and 95%, respectively. Owing to its decent AUC prediction performance, simplicity, and convenience for automated calculation and reporting, SSTA could be used as an adjunctive tool for the AUC-based TDM.

Vancomycin dosing in high-intensity continuous renal replacement therapy: A retrospective cohort study

INTRODUCTION

An inverse relationship exists between vancomycin serum concentrations and the intensity of continuous renal replacement therapy (CRRT), reflected through the dialysate flow rate (DFR). There remains a lack of evidence to guide initial vancomycin dosing in the setting of high-intensity CRRT (i.e., DFR >30 mL/kg/h). Additionally, recommendations for pharmacokinetic monitoring of vancomycin have transitioned from a trough-based to area under the curve (AUC)-based dosing strategy to optimize efficacy and safety. Therefore, an improved understanding of the impact of CRRT intensity on AUC/MIC (minimum inhibitory concentration) has the potential to enhance vancomycin dosing in this patient population.

OBJECTIVES

The goal of this study is to evaluate current vancomycin dosing strategies and achievement of pharmacokinetic targets in patients on high-intensity CRRT.

METHODS

This was a single-center, retrospective cohort study of adult critically ill patients admitted to Houston Methodist Hospital between May 2019 and October 2021 and received vancomycin therapy while on high-intensity CRRT. High-intensity CRRT was defined by a DFR that was both ≥3 L/h and >30 mL/kg/h. Depending on the initial vancomycin dosing strategy, patients were stratified into either the traditional (15 mg/kg/day) or enhanced (≥15 mg/kg/day) dosing group. The primary outcome was the percent of patients who attained steady-state AUC24 /MIC ≥400 mg*h/L at the first obtained vancomycin level in the enhanced group compared with the traditional group.

RESULTS

A total of 125 patients were included in the final analysis, 56 in the traditional and 69 in the enhanced dosing group. The primary end point occurred in 74% and 54% of patients in the enhanced and traditional dosing groups, respectively (p = 0.029). Therapeutic vancomycin trough levels (10-20 μg/mL) were more commonly achieved in the enhanced dosing group compared with the traditional dosing group (66.7% vs. 45%, p = 0.013). As DFR rose, increasingly higher doses of vancomycin, up to 27 mg/kg/day, were required to achieve the therapeutic targets.

CONCLUSION

This is the first study to evaluate the influence of variable CRRT intensities on vancomycin AUC/MIC. Our findings suggest that vancomycin doses of ≥15 mg/kg/day are needed to achieve early therapeutic targets in patients on high-intensity CRRT.

Comparison of Open-Access, Trough-Only Online Calculators Versus Trapezoidal Method for Calculation of Vancomycin Area Under the Curve (AUC)

BACKGROUND

Vancomycin area-under-the-curve (AUC) monitoring is associated with reduced nephrotoxicity but may increase cost and workload for personnel compared to trough monitoring.

OBJECTIVE

The purpose of this study was to compare the accuracy of vancomycin AUC calculated by open-access, online, trough-only calculators to AUCs calculated by the trapezoidal method (TM) using peak and trough concentrations.

METHODS

This retrospective, multi-center study included adults ≥18 years old with stable renal function who received vancomycin with steady-state peak and trough concentrations. Areas under the curve calculated by TM were compared to AUCs calculated by 3 online calculators using trough-only options for calculation: ClinCalc, VancoVanco, and VancoPK. The primary outcome was actual difference in AUC between TM and the online calculators. Secondary outcomes were percent difference in AUC and clinical alignment in dose adjustments between methods.

RESULTS

Seventy patients were included for analysis. There was a statistically significant difference in AUC between TM and ClinCalc (median actual difference: -52, P < 0.001) and VancoVanco (median actual difference: 95, P < 0.001), whereas there was no significant difference between TM and VancoPK (median actual difference: -0.8, P = 0.827). Discordant dose adjustments were indicated when comparing ClinCalc, VancoVanco, and VancoPK to TM in 28%, 36%, and 12% of cases, respectively.

CONCLUSION

The AUC calculator most closely aligned with TM was VancoPK, whereas other included calculators were statistically different. Owing to the cost and complexity of obtaining multiple levels, our findings support using a single steady-state trough using VancoPK as an alternative to TM for calculation of vancomycin AUC.

Performance of Bayesian Area Under the Concentration-Time Curve-Based Pharmacokinetic Dosing Based on a One-Compartment Model and Trough-Only Monitoring for Vancomycin

A novel Bayesian method was developed to interpret serum vancomycin concentrations (SVCs) following the administration of one or more vancomycin doses with potential varying doses and intervals based on superposition principles. The method was evaluated using retrospective data from 442 subjects from three hospitals. The patients were required to receive vancomycin for more than 3 days, have stable renal function (fluctuation in serum creatinine of ≤0.3 mg/dL), and have at least 2 trough concentrations reported. Pharmacokinetic parameters were predicted using the first SVC, and the fitted parameters were then used to predict subsequent SVCs. Using only covariate-adjusted population prior estimates, the first two SVC prediction errors were 47.3 to 54.7% for the scaled mean absolute error (sMAE) and 62.1 to 67.8% for the scaled root mean squared error (sRMSE). "Scaled" refers to the division of the MAE or RMSE by the mean value. The Bayesian method had minimal errors for the first SVC (by design), and for the second SVC, the sMAE was 8.95%, and the sRMSE was 36.5%. The predictive performance of the Bayesian method did degrade with subsequent SVCs, which we attributed to time-dependent pharmacokinetics. The 24-h area under the concentration-time curve (AUC) was determined from simulated concentrations before and after the first SVC was reported. Prior to the first SVC, 170 (38.4%) patients had a 24-h AUC of <400 mg · h/L, 186 (42.1%) had a 24-h AUC within the target range, and 86 (19.5%) had a 24-h AUC of >600 mg · h/L. After the first SVC was reported, 322 (72.9%) had a 24-h AUC within the target range, 68 (15.4%) had low values, and 52 (11.8%) had high values based on the model simulation. Target attainments were 38% before the first SVC and 73% after the first SVC. The hospitals had no policies or procedures in place for targeting 24-h AUCs, although the trough target was typically 13 to 17 mg/L. Our data provide evidence of time-dependent pharmacokinetics, which will require regular therapeutic drug monitoring regardless of the method used to interpret SVCs.

Vancomycin AUC values estimated with trough-only data: Accuracy in an adult academic medical center population

PURPOSE

Vancomycin area under the concentration-time curve (AUC) can be calculated using steady-state serum peak and trough concentrations; however, compared to traditional trough-only monitoring, this approach requires an additional blood sample. Recently published data demonstrated vancomycin AUC estimations using trough-only data with a volume of distribution (Vd) model incorporating age and actual body weight were reasonably accurate and precise in a veteran population. This study sought to extend these methods to a Mayo Clinic adult population.

METHODS

A retrospective, observational cohort of adult patients with documented steady-state vancomycin peak and trough concentrations was evaluated. Vancomycin AUCs were estimated using trough-only data, and 4 Vd models were assessed for accuracy and precision. Estimated AUCs were compared to AUCs calculated using 1-compartment intermittent infusion equations and steady-state peak and trough ("peak-trough") data.

RESULTS

The study population (N = 95) was 46% female, with a median age of 59 years and a median weight of 97 kg. Using the VancoPK equation Vd = 0.29 (age in y) + 0.33 (actual weight in kg) + 11, the mean peak-trough and estimated trough-only AUC were 533 and 534, respectively, with a correlation of 0.936. The root mean square error was 47.7, meaning about 95% of AUCs were within 95 mg · h/L of peak-trough AUCs.

CONCLUSIONS

Accuracy and precision of Vancomycin AUC estimations using trough-only data and the described Vd model were demonstrated in a Mayo Clinic cohort. Targeting an estimated AUC of 500 mg · h/L using the VancoPK model would likely result in an actual AUC within 400 to 600 mg · h/L.

Comparison of vancomycin area under the concentration-time curve (AUC) using two-point pharmacokinetics versus two open-access online single-concentration vancomycin calculators

WHAT IS KNOWN AND OBJECTIVE

Current vancomycin monitoring guidelines recommend the use of area under the concentration-time curve (AUC₂₄ ) monitoring in patients with serious Methicillin-Resistant Staphylococcus aureus (MRSA) infections by utilizing either a Bayesian approach or first-order analytic equations. Several open-access websites exist that allow estimation of vancomycin AUC₂₄ with the use of a single steady-state concentration. It is uncertain how these open-access calculators perform against guideline-recommended methods. The objective was to compare AUC₂₄ estimates from two online, open-access, single-concentration vancomycin calculators compared with the two-point pharmacokinetic (2PK) method.

METHODS

AUC₂₄ estimates were made using the 2PK reference method and the single-concentration vancomycin calculators, ClinCalc and VancoPK. The AUC₂₄ estimates from the 2PK reference method were compared to the online calculators by assessing bias (median AUC₂₄ difference) and precision (AUC₂₄ difference ± 100 mg*h/L). Clinical precision was also assessed by characterizing the frequency that the 2PK reference method and the online calculators showed clinical disagreement based on the following AUC₂₄ categories: (1) AUC₂₄  < 400 mg*h/L; (2) AUC₂₄ 400-600 mg*h/L and (3) AUC₂₄  > 600 mg*h/L.

RESULTS AND DISCUSSION

A total of 253 patients were included in the study. The AUC₂₄ estimates from the ClinCalc and VancoPK single-concentration vancomycin calculators showed some bias and imprecision, though VancoPK appeared to have less. Clinical disagreement versus the 2PK reference method occurred in 31.2% and 19.4% of AUC₂₄ estimates from the ClinCalc and VancoPK single-concentration vancomycin calculators, suggesting clinical imprecision.

WHAT IS NEW AND CONCLUSION

The AUC₂₄ estimates from single-concentration, online vancomycin calculators showed some bias and imprecision in comparison to the 2PK method. Institutions should validate these online, trough-only calculators relative to a 2PK method in their patient populations prior to adoption as standard-of-care.

Assessment of an online calculator's vancomycin dosing and exposure appropriateness in persons who inject drugs with methicillin-resistant Staphylococcus aureus bloodstream infections

WHAT IS KNOWN AND OBJECTIVE

While the gold standard for calculating AUC involves two steady-state concentrations, online calculators can empirically estimate AUC and other pharmacokinetic (PK) parameters. In patients with potentially altered PK, such as persons who inject drugs (PWID), the reliability of these predictions is unclear. Our objectives were to characterize the PK of vancomycin in PWID with methicillin-resistant Staphylococcus aureus (MRSA) bloodstream infections (BSI) and to assess the impact of these PK parameters on dosing regimens when compared to regimens generated by an online calculator.

METHODS

This descriptive pilot study included a retrospective chart review of 48 inpatient PWID with MRSA BSI from 30 April 2018 through 31 August 2020. Demographic and clinical data along with vancomycin dosing and serum concentrations were collected. Patient-specific PK parameters were used to calculate the AUC of each empiric regimen compared with the originally predicted AUC.

RESULTS AND DISCUSSION

The study population had a median volume of distribution of 0.74 L/kg, clearance of 0.081 L/kg/h, elimination rate constant of 0.110/h and half-life of 6.3 h. The online calculator empirically predicted 6 subtherapeutic and 42 appropriate AUC values with its recommended empiric dosing regimens. Using the actual patient-specific PK parameters, the empiric vancomycin regimens actually resulted in 21 (43.75%) underexposures, 24 (50%) appropriate exposures and 3 (6.25%) overexposures.

WHAT IS NEW AND CONCLUSIONS

In PWID, empiric vancomycin dosing strategies suggested by an online calculator frequently resulted in lower-than-predicted vancomycin exposures. These findings suggest that PWID with MRSA BSI may require higher and/or more frequent vancomycin doses than those empirically recommended by the population-based methods of an online calculator.