Determinants of gentamicin concentrations in critically ill patients: a population pharmacokinetic analysis

When treating critically ill patients with gentamicin for severe infection, peak concentrations (C_max) determine clinical efficacy and trough concentrations (C_min) determine toxicity. Despite administration of body weight-standardised starting doses, a wide range of C_max is generally observed. Furthermore, in therapeutic drug monitoring, several measures of renal function are used to predict appropriate C_min and gentamicin dosing intervals, but the most accurate predictor is not known. This study aimed to quantify the impact of several patient parameters on gentamicin C_max values and to determine which measure of renal function best predicts gentamicin clearance (CL). Clinical data and serum gentamicin levels were retrospectively collected from all critically ill patients treated with gentamicin at our intensive care unit between 1 January and 30 June 2011. Data were analysed using non-linear mixed-effects modelling (NONMEM v.7.1.2). A two-compartmental model was developed based on 303 gentamicin concentration-time data from 44 critically ill patients. Serum albumin levels explained 25% of interindividual variability in the volume of distribution (V_d). Creatinine clearance calculated from the creatinine concentration in a 6-h urine portion (CalcCL_Cr) resulted in acceptable estimation of gentamicin CL, whilst serum creatinine (SCr) and creatinine clearance estimated by the Cockcroft-Gault formula (CGCL_Cr) overestimated gentamicin CL and therefore underestimated C_min. In conclusion, low albumin concentrations resulted in a larger V_d and lower C_max of gentamicin. These results suggest that use of a higher gentamicin starting dose in critically ill patients with hypoalbuminaemia may prevent underdosing. Urinary CalcCL_Cr is a better predictor of C_min than SCr or CGCL_Cr.