Optimizing Vancomycin Exposure for Early Treatment Response in Very-Low-Birth-Weight Neonates With Late-Onset Sepsis.

BACKGROUND: Vancomycin is widely used to treat neonatal late-onset sepsis (LOS). However, its optimal pharmacokinetic/pharmacodynamic targets in neonates, particularly those with very low birth weight (VLBW), remain inadequately defined. Therefore, the objective of this study is to identify vancomycin exposure thresholds associated with early clinical and microbiologic effectiveness in VLBW neonates with LOS. METHODS: A multicenter observational study included VLBW neonates who received vancomycin for LOS and had at least 1 steady-state trough concentration measured. Based on clinical and microbiological outcomes within the first 72 hours of vancomycin therapy, patients were categorized into 2 groups: early clinical and microbiological cure (ECMC) and late clinical and microbiological cure. Vancomycin exposure threshold predictive of ECMC was identified using classification and regression tree analysis and the receiver operating characteristic curve. RESULTS: Thirty-six VLBW neonates with confirmed Gram-positive LOS were included, of whom 27 (75%) achieved ECMC. A significant correlation was observed between area under the curve (AUC)24 and trough concentrations (R2 = 0.568; P < 0.001). Both AUC24 and trough concentrations were significantly higher in the ECMC than in the late clinical and microbiological cure group [median (interquartile range), 257 (229-330) vs. 175 (151-223) mg·h/L; P = 0.005; 10.2 (6.0-13.3) vs. 5.1 (3.0-10.0) mg/L; P = 0.043, respectively]. An AUC24/MIC threshold of ≥ 253 mg.h/L was identified as the optimal threshold associated with ECMC with better discriminative performance than trough concentrations (AUC-receiver operating characteristic, 0.813 vs. 0.726; P = 0.005 and P = 0.045, respectively). CONCLUSIONS: While both exposure metrics demonstrated good predictive performance for ECMC in VLBW neonates with LOS, an AUC24/MIC ≥253 mg·h/L showed superior discriminative performance compared with trough concentrations. Prospective validation of this exposure threshold in larger cohorts is warranted.