Empirical Vancomycin Dosing in Pediatric Patients with Congenital Heart Disease and the Impact of Cardiopulmonary Bypass on Trough Concentrations

Optimizing vancomycin dosing in pediatrics: a machine learning approach to predict trough concentrations in children under four years of age

BACKGROUND

Vancomycin trough concentration is closely associated with clinical efficacy and toxicity. Predicting vancomycin trough concentrations in pediatric patients is challenging due to significant inter-individual variability and rapid physiological changes during maturation.

AIM

This study aimed to develop a machine learning model to predict vancomycin trough concentrations and determine optimal dosing regimens for pediatric patients < 4 years of age using ML algorithms.

METHOD

A single-center retrospective observational study was conducted from January 2017 to March 2020. Pediatric patients who received intravenous vancomycin and underwent therapeutic drug monitoring were enrolled. Seven ML models [linear regression, gradient boosted decision trees, support vector machine, decision tree, random forest, Bagging, and extreme gradient boosting (XGBoost)] were developed using 31 variables. Performance metrics including R-squared (R2), mean square error (MSE), root mean square error (RMSE), and mean absolute error (MAE) were compared, and important features were ranked.

RESULTS

The study included 120 eligible trough concentration measurements from 112 patients. Of these, 84 measurements were used for training and 36 for testing. Among the seven algorithms tested, XGBoost showed the best performance, with a low prediction error and high goodness of fit (MAE = 2.55, RMSE = 4.13, MSE = 17.12, and R2 = 0.59). Blood urea nitrogen, serum creatinine, and creatinine clearance rate were identified as the most important predictors of vancomycin trough concentration.

CONCLUSION

An XGBoost ML model was developed to predict vancomycin trough concentrations and aid in drug treatment predictions as a decision-support technology.

Vancomycin Dosing and Its Association With Acute Kidney Injury in Pediatric Cardiac Intensive Care Patients Under 3 Months of Age

BACKGROUND

The standard vancomycin regimen for term neonates is 45 mg/kg/day. However, the optimal starting vancomycin dosing for achieving therapeutic levels in young infants in cardiac intensive care units remains unknown. Moreover, data on the association of supratherapeutic vancomycin levels with acute kidney injury (AKI) are limited.

METHODS

Retrospective study of infants ≤3 months old, receiving vancomycin following congenital heart surgery at postoperative intensive care unit admission. Assessed were vancomycin dosing, achievement of therapeutic trough concentration of 10-20 mg/L and development of AKI, based on the modified Kidney Disease Improving Global Outcomes criteria.

RESULTS

Inclusion criteria were met by 109 patients with a median age of 8 days (IQR: 6-16). The mean (SD) vancomycin dose required for achieving therapeutic concentration was 28.9 (9.1) mg/kg at the first postoperative day. Multivariate logistic regression identified higher preoperative creatinine levels and shorter cardiopulmonary bypass time as predictors of supratherapeutic vancomycin concentrations (c-index 0.788). During the treatment course, 62 (56.9%) developed AKI. Length of stay and mortality were higher in those who developed AKI as compared with those who did not. Multivariate logistic regression identified higher vancomycin concentration as a predictor for postoperative AKI, OR, 3.391 (95% CI: 1.257-9.151), P = 0.016 (c-index 0.896).

CONCLUSION

Our results support a lower starting vancomycin dose of ~30 mg/kg/day followed by an early personalized therapeutic approach, to achieve therapeutic trough concentrations of 10-20 mg/L in cardiac postoperative term infants. Supratherapeutic concentrations are associated with an increased risk for AKI, which is prevalent in this population and associated with adverse outcomes.

Use of Machine Learning for Dosage Individualization of Vancomycin in Neonates

BACKGROUND AND OBJECTIVE

High variability in vancomycin exposure in neonates requires advanced individualized dosing regimens. Achieving steady-state trough concentration (C0) and steady-state area-under-curve (AUC0-24) targets is important to optimize treatment. The objective was to evaluate whether machine learning (ML) can be used to predict these treatment targets to calculate optimal individual dosing regimens under intermittent administration conditions.

METHODS

C0 were retrieved from a large neonatal vancomycin dataset. Individual estimates of AUC0-24 were obtained from Bayesian post hoc estimation. Various ML algorithms were used for model building to C0 and AUC0-24. An external dataset was used for predictive performance evaluation.

RESULTS

Before starting treatment, C0 can be predicted a priori using the Catboost-based C0-ML model combined with dosing regimen and nine covariates. External validation results showed a 42.5% improvement in prediction accuracy by using the ML model compared with the population pharmacokinetic model. The virtual trial showed that using the ML optimized dose; 80.3% of the virtual neonates achieved the pharmacodynamic target (C0 in the range of 10-20 mg/L), much higher than the international standard dose (37.7-61.5%). Once therapeutic drug monitoring (TDM) measurements (C0) in patients have been obtained, AUC0-24 can be further predicted using the Catboost-based AUC-ML model combined with C0 and nine covariates. External validation results showed that the AUC-ML model can achieve an prediction accuracy of 80.3%.

CONCLUSION

C0-based and AUC0-24-based ML models were developed accurately and precisely. These can be used for individual dose recommendations of vancomycin in neonates before treatment and dose revision after the first TDM result is obtained, respectively.

Vancomycin dosing required to achieve a therapeutic level in children post-surgical correction of congenital heart disease

The vancomycin dosing range for safe and effective treatment remains uncertain for children who had corrective surgery for a congenital heart disease (CHD). We aimed to determine the vancomycin dosing requirements for this subgroup of patients. This prospective cohort study included children younger than 14 years old with CHD who received intravenous vancomycin for at least 3 days at the Pediatric Cardiology section of King Abdulaziz Medical City, Riyadh. In total, 140 pediatric patients with CHD were included with a median age of 0.57 years (interquartile range 0.21-2.2). The mean vancomycin total daily dose (TDD), 37.71 ± 6.8 mg/kg/day, was required to achieve a therapeutic trough concentration of 7-20 mg/L. The patient's age group and the care setting were significant predictors of the vancomycin dosing needs. Neonates required significantly lower doses of 34 ± 6.03 mg/kg/day (P = .002), and young children higher doses of 43.97 ± 9.4 mg/kg/day (P = .003). The dosage requirements were independent of the type of cardiac lesion, cardiopulmonary surgery exposure, sex, and BMI percentile. However, the patients in the pediatric cardiac ward required higher doses of vancomycin 41.08 ± 7.06 mg/kg/day (P = .039). After the treatment, 11 (8.5%) patients had an elevated Scr, and 3 (2.3%) patients developed AKI; however, none of the patients' sociodemographic factors or clinical variables, or vancomycin therapy characteristics was significantly associated with the renal dysfunction. Overall, the vancomycin TDD requirements are lower in pediatric post-cardiac surgery compared to non-cardiac patients and are modulated by several factors.

Pharmacokinetics of Antibiotics in Pediatric Intensive Care: Fostering Variability to Attain Precision Medicine

Children show important developmental and maturational changes, which may contribute greatly to pharmacokinetic (PK) variability observed in pediatric patients. These PK alterations are further enhanced by disease-related, non-maturational factors. Specific to the intensive care setting, such factors include critical illness, inflammatory status, augmented renal clearance (ARC), as well as therapeutic interventions (e.g., extracorporeal organ support systems or whole-body hypothermia [WBH]). This narrative review illustrates the relevance of both maturational and non-maturational changes in absorption, distribution, metabolism, and excretion (ADME) applied to antibiotics. It hereby provides a focused assessment of the available literature on the impact of critical illness—in general, and in specific subpopulations (ARC, extracorporeal organ support systems, WBH)—on PK and potential underexposure in children and neonates. Overall, literature discussing antibiotic PK alterations in pediatric intensive care is scarce. Most studies describe antibiotics commonly monitored in clinical practice such as vancomycin and aminoglycosides. Because of the large PK variability, therapeutic drug monitoring, further extended to other antibiotics, and integration of model-informed precision dosing in clinical practice are suggested to optimise antibiotic dose and exposure in each newborn, infant, or child during intensive care.

Pharmacokinetics and Target Attainment of Antibiotics in Critically Ill Children: A Systematic Review of Current Literature

BACKGROUND

Pharmacokinetics (PK) are severely altered in critically ill patients due to changes in volume of distribution (Vd) and/or drug clearance (Cl). This affects the target attainment of antibiotics in critically ill children. We aimed to identify gaps in current knowledge and to compare published PK parameters and target attainment of antibiotics in critically ill children to healthy children and critically ill adults.

METHODS

Systematic literature search in PubMed, EMBASE and Web of Science. Articles were labelled as relevant when they included information on PK of antibiotics in critically ill, non-neonatal, pediatric patients. Extracted PK-parameters included Vd, Cl, (trough) concentrations, AUC, probability of target attainment, and elimination half-life.

RESULTS

50 relevant articles were identified. Studies focusing on vancomycin were most prevalent (17/50). Other studies included data on penicillins, cephalosporins, carbapenems and aminoglycosides, but data on ceftriaxone, ceftazidime, penicillin and metronidazole could not be found. Critically ill children generally show a higher Cl and larger Vd than healthy children and critically ill adults. Reduced target-attainment was described in critically ill children for multiple antibiotics, including amoxicillin, piperacillin, cefotaxime, vancomycin, gentamicin, teicoplanin, amikacin and daptomycin. 38/50 articles included information on both Vd and Cl, but a dosing advice was given in only 22 articles.

CONCLUSION

The majority of studies focus on agents where TDM is applied, while other antibiotics lack data altogether. The larger Vd and higher Cl in critically ill children might warrant a higher dose or extended infusions of antibiotics in this patient population to increase target-attainment. Studies frequently fail to provide a dosing advice for this patient population, even if the necessary information is available. Our study shows gaps in current knowledge and encourages future researchers to provide dosing advice for special populations whenever possible.

Vancomycin Dosage and Its Association with Clinical Outcomes in Pediatric Patients with Gram-Positive Bacterial Infections

Aim

The aim of this study was to evaluate whether vancomycin trough concentrations at initial steady state are associated with clinical and microbiological outcomes along with vancomycin-related nephrotoxicity in pediatric patients with Gram-positive bacterial (GPB) infections.

Methods

A retrospective cohort study of pediatric patients who received vancomycin for ≥72 hours during 2008–2016 was conducted. Study patients were divided into three cohorts in accordance with their first trough levels at steady state: <5 mg/L (lower-trough), 5–10 mg/L (low-trough), and >10 mg/L (high-trough; reference) cohorts.

Results

Of the 201 patients eligible for study inclusion, 60 patients in the lower- and low-trough cohorts, respectively, were idect 3ntified via propensity score matching and analyzed against 30 high-trough patients in each comparison pair (neonates were excluded due to small sample size). Lower-trough patients were at a greater risk for prolonged therapy, retreatment, and dose adjustment than high-trough patients. Final steady-state troughs remained substantially lower in both the lower- and low-trough cohorts (p<0.001 and p=0.005, respectively), despite greater dose up-titration in the lower-trough cohort and percent change in daily dose in both the lower- and low-trough cohorts than in the high-trough cohort (p<0.001 for all). Clinical cure and death risk, along with the risks of isolation of resistant strains and renal events, were not significantly different between cohorts in both comparison pairs.

Conclusion

Vancomycin troughs of <5 mg/L at initial steady state were associated with significantly compromised clinical outcomes in terms of risk of therapy prolongation, retreatment, and aggressive dose up-titration, compared to >10 mg/L troughs in pediatric patients with GPB infections.

External Validation of Model-Based Dosing Guidelines for Vancomycin, Gentamicin, and Tobramycin in Critically Ill Neonates and Children: A Pragmatic Two-Center Study

BACKGROUND

The Dutch Pediatric Formulary (DPF) increasingly bases its guidelines on model-based dosing simulations from pharmacokinetic studies. This resulted in nationwide dose changes for vancomycin, gentamicin, and tobramycin in 2015.

OBJECTIVE

We aimed to evaluate target attainment of these altered, model-based doses in critically ill neonates and children.

METHODS

This was a retrospective cohort study in neonatal intensive care unit (NICU) and pediatric ICU (PICU) patients receiving vancomycin, gentamicin, or tobramycin between January 2015 and March 2017 in two university hospitals. The first therapeutic drug monitoring concentration for each patient was collected, as was clinical and dosing information. Vancomycin and tobramycin target trough concentrations were 10-15 and ≤ 1 mg/L, respectively. Target gentamicin trough and peak concentrations were < 1 and 8-12 mg/L, respectively.

RESULTS

In total, 482 patients were included (vancomycin [PICU] n = 62, [NICU] n = 102; gentamicin [NICU] n = 97; tobramycin [NICU] n = 221). Overall, median trough concentrations were within the target range for all cohorts but showed large interindividual variability, causing nontarget attainment. Trough concentrations were outside the target range in 66.1%, 60.8%, 14.7%, and 23.1% of patients in these four cohorts, respectively. Gentamicin peak concentrations were outside the range in 69% of NICU patients (term neonates 87.1%, preterm infants 57.1%). Higher creatinine concentrations were associated with higher vancomycin and tobramycin trough concentrations.

CONCLUSION

This study illustrates the need to validate model-based dosing advice in the real-world setting as both sub- and supratherapeutic concentrations of vancomycin, gentamicin, and tobramycin were very prevalent. Our data underline the necessity for further individualization by addressing the high interindividual variability to improve target attainment.