Abstract
PURPOSE: To investigate the key factors influencing the plasma concentration of levetiracetam (LEV) in patients with epilepsy and to establish a predictive model for LEV steady-state trough concentration. PATIENTS AND METHODS: Clinical data from 130 epilepsy patients (175 steady-state trough concentration blood samples) were retrospectively collected at a single center. Univariate analysis and multiple linear regression modeling were employed to systematically quantify the independent effects of demographic characteristics, biochemical indicators, and concomitant medications on the LEV C/D. The robustness of the model was validated using the Bootstrap method (1000 resamples). RESULTS: Creatinine clearance rate (Ccr) emerged as the strongest independent predictor (unstandardized β = -0.168, p < 0.001), with every 1 mL/min increase in Ccr resulting in a 0.168 ng·mL(-1)/(g·d(-1)) decrease in C/D. Hemoglobin (HGB) exhibited a secondary negative association (unstandardized β = -0.070, p < 0.05), where every 1 g/L decrease led to a 0.070 ng·mL(-1)/(g·d(-1)) increase in C/D. Bootstrap validation confirmed the stability of these coefficients (95% CI: Ccr [-0.221, -0.123], HGB [-0.140, -0.008]). The final predictive equation was: C/D(pred) [ng·mL(-1)/(g·d(-1))] = 37.759 - 0.168 × Ccr (mL/min) - 0.070 × HGB (g/L) (adjusted R² = 0.440, p < 0.001). CONCLUSION: Routine clinical indicators Ccr and HGB are core influencing factors of LEV exposure. This model quantifies their independent effects on LEV steady-state trough concentration, providing novel insights into the interindividual variability of LEV pharmacokinetics and offering a potential tool for aiding individualized dosing strategies, especially in resource-limited settings where therapeutic drug monitoring is not readily available. However, its generalizability still needs to be demonstrated through subsequent external validation and prospective multicenter studies.