Pharmacokinetic/Pharmacodynamic Modeling of Linezolid in Relation to Lactate as a Predictive Biomarker

Background: Linezolid, a synthetic oxazolidinone antibiotic, is effective against multidrug-resistant Gram-positive bacteria. However, prolonged use is associated with mitochondrial toxicity and lactic acidosis. Predictive modeling can enhance risk assessment and support therapeutic drug monitoring. Objective: This study evaluates various PK/PD models to describe lactate elevation associated with linezolid therapy. Both simulated and clinical data were used to fit four mathematical models: Emax, piecewise linear, log-linear, and AUC-based. The goal is to assist clinicians in optimizing dosing strategies and anticipating potential toxicity. Methods: Simulated linezolid concentration data (0–15 mg/L) were used to construct models in Python, employing NumPy for computations and Matplotlib for visualization. Each model was fitted to time-series lactate data extracted from a published clinical case report. Key parameters—such as EC₅₀, Hill coefficient, and exposure thresholds—were estimated to evaluate model performance. Results: The Emax model captured threshold-like lactate dynamics, closely reflecting the onset of clinical lactic acidosis. The log-linear model showed a smooth, progressive increase in lactate and remained robust across all time points. The piecewise linear model offered interpretability but lacked physiological realism. The AUC-based model reflected cumulative exposure effects and was suitable for long-term monitoring, though less sensitive in early phases. Conclusion: The Emax model provided the best fit for abrupt lactate elevation, while the log-linear model offered a balance between predictive accuracy and clinical utility. These models support individualized dosing strategies for at-risk patients undergoing prolonged linezolid therapy. Further validation in clinical cohorts is recommended.