Modifications of antimicrobials’ pharmacokinetic parameters have been reported in critically ill patients, resulting in a risk of treatment failure. We characterized amikacin pharmacokinetic variability in critically ill patients with ventilator-associated pneumonia (VAP) and evaluated several dosing regimens.

We conducted a prospective multicenter study in critically ill patients with presumptive diagnosis of Gram-negative bacilli (GNB) VAP. Patients empirically received imipenem and a single-dose of amikacin, which was administered as a 30-min infusion (20 mg/kg). Concentrations were measured 0.5, 1, 8, 16, and 24 h after beginning of infusion. Pharmacokinetic parameters were estimated using a population approach. Main pharmacodynamic target was a ratio ≥10 between the concentration achieved 1 h after beginning of infusion (C _1h) and the minimal inhibitory concentration of the liable bacteria (MIC). We simulated individual C _1h for several dosing regimens by Monte Carlo method and computed C _1h/MIC ratios for MICs from 0.5 to 64 mg/L.

Sixty patients (47 males), median (range) age, and body weight, 61.5 years (28–84) and 78 kg (45–126), respectively, were included. Amikacin median C _1h was 45 mg/L (22–87). Mean value (between-patients variability) for CL, V1, Q, and V2 were 4.3 L/h (31 %), 15.9 L (22 %), 12.1 L/h (27 %), and 21.4 L (47 %), respectively. CL increased with CrCL (p < 0.001) and V1 with body weight (p < 0.001) and PaO₂/FIO₂ ratio (p < 0.001). With a 25 mg/kg regimen, the pharmacodynamic target was achieved in 20 and 96 % for a MICs of 8 and 4 mg/L, respectively.

Amikacin clearance was decreased and its volume of distribution was increased as previously reported. A ≥25 mg/kg single-dose is needed for empirical treatment of GNB-VAP.