Individualized Vancomycin Regimens Based On Population Pharmacokinetics In Pediatric Patients

Objectives:To evaluate the published vancomycin population pharmacokinetic(PPK)model in pediatric patients and to verify its applicability to our population.To develop a PPK model of vancomycin based on serum Cystatin C(CysC)in pediatric patients and a PPK model of vancomycin in pediatric patients with hematological malignancies complicated with augmented renal clearance(ARC).To define the initial dosage regimens based on the two established vancomycin PPK model.Using a clinical decision supporting system for individualized dosing and evaluate the validity and practicality of the PPK model for guiding the individualized dosage in clinical application.To evaluate the association between vancomycin AUC_0-24/MIC and clinical outcome in pediatric patients with invasive gram-positive infection.Methods:(1)Literatures were collected with the Web of Science,PubMed,Scopus,CNKI and WANFANG databases up to 1 January 2020.Published models of 2-18 years pediatric patients were screened from the literatures.Prediction and simulation diagnostics and Bayesian forecasting were performed to evaluate the predictive performance of models.(2)A vancomycin PPK model in pediatric patients was built by nonlinear mixed effects modeling(NONMEN)method with CysC as the indicator of renal function.Internal evaluation of the final model is performed through goodness of fit plots(GOFs),non-parametric bootstrapping(Bootstrap)and visual predictive check(VPC)and normalized prediction distribution error(NPDE).The final model was examined by external validation and internal evaluations.(3)Pediatric patients with hematologic malignancy complicated by ARC were enrolled in this retrospective study.The vancomycin PPK model was established by NONMEN program.Goodness-of-fit(GOF)plots,nonparametric bootstrap,normalized prediction distribution error(NPDE),and visual predictive checks(VPC)were carried out for internal evaluation of the final model.(4)The optimal dosing regimens was obtained by Monte Carlo simulation approach,which was used to simulate different initial dosage regimens and calculate the probability of target acquisition(PTA)for AUC_0-24/MIC?400 based on the two established vancomycin PPK model.Using a clinical decision supporting systems for individualized dosing to clinical application.The accuracy and practicality of the PPK model was evaluated by the predicted error between individual prediction value and measured value of steady-state through Bayesian estimation in clinical real cases.(5)AUC_0-24 was computed using a Bayesian approach based on the population pharmacokinetics PPK model.Classification and regression tree analysis(CART)was used to identify the AUC_0-24/MIC threshold associated with clinical outcome.The vancomycin clinical failure of independent predictors were identified by multivariable logistic analysis.Results:(1)Ten published models were externally validated.In prediction-based diagnosis,the prediction error of all models was less than 40%in the ±30%range.In simulation diagnosis,the simulated values of visual predictive checks(VPC)and observations had a large bias in all models.Normalized prediction distribution error(NPDE)did not obey the normal distribution.In Bayesian prediction,Bayesian forecasting significantly improved the prediction performance of the model with prior 1-2 observations.However,none of the models had a proportion greater than 35%in the ±20%individual prediction error range.(2)A vancomycin PPK model in pediatric patients was built based on Cys C.The final model was CL(L/h)=2.33×0.747((CYSC)/(0.71))× 1.94((WT)/(28.71));V(L)=8.16×2.44((WT)/(28.71)).Internal and external evaluations demonstrated good stability and predictability.(3)Fifty-three patients with 106 samples were included.A onecompartment model with first-order elimination was developed,the final model was:CL(L/h)=6.32×((WT)/(70))0.75×e^0.0467;V(L)=39.6×((WT)/(70)).The internal validation of the model showed a good prediction performance.(4)The initial dosage regimen was recommended to achieve a target of AUC_0-24/MIC 400-700 in different age and renal function pediatric patients.The average value of probability of target attainment(PTA)was 66.9%.The initial dosage regimen was recommended to achieve a target of AUC_0-24/MIC?400 in pediatric patients with hematological malignancies complicated with ARC.The average value of PTA was 94.48%.A clinical decision supporting systems for individualized dosing SmartDose combined with Bayesian estimation was used to provide individualized dosing regimen guidance.It shows that the prediction error of the model was small,indicating high accuracy and good practicality in clinical application.(5)Sixty pediatric patients with invasive gram-positive infection were enrolled.The median age in week of the sample was 112 weeks(interquartile range(IQR),19-456 weeks).A CART-derived vancomycin AUC_0-24/MIC of<273 was associated with increased failure rate(P=0.026),and multivariable logistic analysis showed that AUC_0-24/MIC of<273 might be an independent predictor of vancomycin clinical failure(OR 5.567,95%CI 1.30-23.80,P=0.021).Conclusions:Published vancomycin models in children and adolescents had poor extrapolation for our dataset.Bayesian forecasting significantly improved the prediction performance of the model with prior 1-2 observations.The pharmacokinetic parameters for vancomycin in pediatric patients were estimated using a serum CysC model.A vancomycin PPK model was established in Chinese pediatric patients with hematologic malignancy and ARC.The Initial dosage regimens were recommended in different renal function pediatric patients and hematologic malignancy complicated with ARC patients to achieve AUC_0-24/MIC?400.A clinical decision supporting systems and PPK models combined Bayesian estimation can provide individualized dosing regimen guidance.We found that vancomycin PK/PD indicator AUC_0-24/MIC of<273 was an independent predictor of vancomycin clinical failure.Larger prospective researches are required to verify the target in pediatric patients.