万古霉素血清谷浓度与药时曲线下面积预测新生儿肾毒性的临床应用价值

Clinical utility of vancomycin serum trough concentrations and area under the curve in predicting nephrotoxicity in neonates

目的探讨万古霉素血清谷浓度、24 h曲线下面积(AUC_(24h))与新生儿急性肾损伤(AKI)相关性,确定最佳药物暴露阈值,为新生儿临床合理用药提供参考。方法回顾性收集2019年6月至2023年6月接受万古霉素静脉治疗的住院新生儿信息,包括血清谷浓度、人口学资料、临床资料及相关实验室指标等;AUC_(24h)采用贝叶斯方法计算;采用Spearman相关系数评价谷浓度与AUC_(24h)的关系;绘制受试者工作特征(ROC)曲线和校准曲线评价谷浓度与AUC_(24h)对AKI的预测效能,使用Youden指数选择最佳临界点;采用多因素Cox风险比例模型分析谷浓度临界值、AUC_(24h)临界值与AKI的相关性;绘制患者的生存曲线,使用Log-rank检验比较分组后肾损伤发生率的差异。结果共135例患儿,12例发生AKI,占8.9%。发生AKI患儿万古霉素谷浓度(16.10 mg/L对12.40 mg/L;P>0.05)及AUC_(24h)(647.53 mg·h/L对503.25 mg·h/L;P<0.01)均高于未发生AKI患儿;万古霉素谷浓度与AUC_(24h)呈弱相关(Rho=0.304);通过ROC曲线分析,AUC_(24h)高于谷浓度(0.763对0.608),Brier评分均<0.1;AKI的谷浓度暴露阈值为24.90 mg/L,AUC_(24)的暴露阈值为575.78 mg·h/L,Cox风险比例模型分析显示两者与AKI发生相关,Log-rank检验显示二分类后的万古霉素谷浓度、AUC_(24h)的AKI发生率差异均具有统计学意义(P<0.01)。结论万古霉素血清谷浓度及使用贝叶斯方法计算AUC_(24h)可以预测万古霉素对新生儿肾毒性,谷浓度>24.90 mg/L和AUC_(24h)>575.78 mg·h/L可作为风险预测因子,用于指导临床用药。

Objective To explore the correlation between vancomycin serum trough concentrations,24-hour area under the curve(AUC_(24h)),and neonatal acute kidney injury(AKI),and determine the optimal exposure thresholds for rational therapy in neonates.Methods A retrospective analysis was conducted the neonates who received intravenous vancomycin treatment from June 2019 to June 2023.The clinical data were collected,including serum trough concentrations,demographic data,clinical information,and relevant laboratory tests.AUC_(24h) was calculated using Bayesian methods.Spearman correlation coefficient was used to assess the relationship between trough vancomycin concentrations and AUC_(24h).Receiver Operating Characteristic(ROC)curves and calibration curve were used to evaluate the predictive performance of trough concentrations and AUC_(24h) for AKI.The optimal cutoff value was selected using the Youden index.Multivariable Cox regression analysis was used to examine the correlation between trough concentration cutoff value,AUC_(24h) cutoff value,and AKI.Survival curves of patients were plotted.Log-rank test was used to compare differences in the incidence of AKI between groups.Results A total of 135 cases were included.Overall,8.9%(12/135)of the patients experienced nephrotoxicity.Nephrotoxic patients showed higher trough vancomycin concentrations(16.10 mg/L vs.12.40 mg/L;P>0.05)and AUC_(24h)(647.53 mg·h/L vs.503.25 mg·h/L;P<0.01)compared to those without AKI.Trough concentrations showed weak correlation with AUC_(24h)(Rho=0.304).ROC curve analysis revealed a higher area under the curve of AUC_(24h) than trough concentrations(0.763 vs.0.608),with Brier scores<0.1.The cutoff value of trough concentration and AUC_(24h) was 24.90 mg/L and 575.78 mg·h/L respectively for AKI.Cox regression analysis demonstrated correlation between trough concentration and AUC_(24h) and AKI risk.Log-Rank tests indicated significant difference in AKI rates for both dichotomized trough concentrations and AUC_(24h)(P<0.01).Conclusions Vancomycin serum trough concentrations and AUC_(24h),calculated using Bayesian methods,can predict vancomycin-induced nephrotoxicity in neonates.Cutoff value of trough concentration>24.90 mg/L and AUC_(24h)>575.78 mg·h/L can be considered as risk predictors in clinical prescription.