Population pharmacokinetics and humanized dosage regimens matching the peak, area, trough, and range of amikacin plasma concentrations in immune-competent murine bloodstream and lung infection models

Abstract

Amikacin is an FDA-approved aminoglycoside antibiotic that is commonly used. However, validated dosage regimens that achieve clinically relevant exposure profiles in mice are lacking. We aimed to design and validate humanized dosage regimens for amikacin in immune-competent murine bloodstream and lung infection models of Acinetobacter baumannii. Plasma and lung epithelial lining fluid (ELF) concentrations after single subcutaneous doses of 1.37, 13.7, and 137 mg/kg of body weight were simultaneously modeled via population pharmacokinetics. Then, humanized amikacin dosage regimens in mice were designed and prospectively validated to match the peak, area, trough, and range of plasma concentration profiles in critically ill patients (clinical dose: 25-30 mg/kg of body weight). The pharmacokinetics of amikacin were linear, with a clearance of 9.93 mL/h in both infection models after a single dose. However, the volume of distribution differed between models, resulting in an elimination half-life of 48 min for the bloodstream and 36 min for the lung model. The drug exposure in ELF was 72.7% compared to that in plasma. After multiple q6h dosing, clearance decreased by ~80% from the first (7.35 mL/h) to the last two dosing intervals (~1.50 mL/h) in the bloodstream model. Likewise, clearance decreased by 41% from 7.44 to 4.39 mL/h in the lung model. The humanized dosage regimens were 117 mg/kg of body weight/day in mice [administered in four fractions 6 h apart (q6h): 61.9%, 18.6%, 11.3%, and 8.21% of total dose] for the bloodstream and 96.7 mg/kg of body weight/day (given q6h as 65.1%, 16.9%, 10.5%, and 7.41%) for the lung model. These validated humanized dosage regimens and population pharmacokinetic models support translational studies with clinically relevant amikacin exposure profiles.

Keywords: S-ADAPT; amikacin; aminoglycoside; bloodstream infection; lung epithelial lining fluid; lung infection; model-based humanized dosage regimen design; mouse/mice; murine infection model; population pharmacokinetics.

Fig 1

Population PK model structure for amikacin describing the plasma and lung ELF concentrations in immune-competent murine models with bloodstream or lung infection. The k_abs is the first-order absorption rate constant from the subcutaneous injection site, V1 is the volume of distribution of the central compartment, and CL is the total body clearance. The X_plasma represents the amount of amikacin in the central compartment, X_ELF is the amount in the ELF compartment, and F_ELF is the AUC in ELF divided by the AUC in plasma. The volume of distribution of the ELF compartment (V_ELF) was fixed to a small value, which only minimally affected the plasma concentrations. The t_1/2,eq is the equilibration half-life between the ELF and the plasma compartment.

Fig 2

Visual predictive check for the population PK analysis of the bloodstream infection model for the PK dose ranging studies using single doses of 1.37, 13.7, and 137 mg/kg of body weight. Plasma concentrations were plotted on linear scale (row A) and log scale (row B). The observations (markers) were plotted along with the 50th percentile (i.e., the median; black line), the 80% prediction interval (10th–90th percentile), and the interquartile range (25th–75th percentile) of the model predictions. Ideally, the median should reflect the central tendency of the observations, and 10% of observations should fall outside the 80% prediction interval on either side.

Fig 3

Visual predictive check for the population PK analysis of the lung infection model for the PK dose ranging studies using single doses of 1.37, 13.7, and 137 mg/kg of body weight. Plasma concentrations were plotted on linear scale (row A) and log scale (row B). Likewise, ELF concentrations were plotted on linear scale (row C) and log scale (row D). The observations (markers) were plotted along with the 50th percentile (i.e., the median; black line), the 80% prediction interval (10th–90th percentile), and the interquartile range (25th–75th percentile) of the model predictions.

Fig 4

Individual fits for the single-dose PK dose range studies (1.37, 13.7, and 137 mg/kg of body weight). Plasma concentrations from the bloodstream and lung infection models are shown in rows A and C on linear scale, as well as in rows B and D on log scale, respectively. The ELF concentrations from the lung infection model are plotted in row E on linear scale and in row F on log scale. Markers are observations and curves are individual fits.

Fig 5

Prospective validation of humanized dosage regimens for amikacin after multiple dosing in murine models with bloodstream infection (row A) and lung infection (rows B and C) at linear scale (left column) and log scale (right column). The green line is the 10th percentile, and the purple line is the 90th percentile of the amikacin plasma concentrations in critically ill patients. The blue line is the typical simulated PK profile for amikacin in mice based on the established model after a single dose (Table 1). The dots are observed concentrations. The humanized dosage regimens used q6h dosing for the bloodstream infection model (61.9% at 0 h, 18.6% at 6 h, 11.3% at 12 h, and 8.21% at 18 h, 117 mg/kg of body weight/day in total) and the lung infection model (65.1% at 0 h, 16.9% at 6 h, 10.5% at 12 h, and 7.41% at 18 h, 96.72 mg/kg of body weight/day in total).

Fig 6

Individual fitting plots for the population PK analysis of the PK validation study with humanized dosing. Row A represents the plasma concentrations from the bloodstream infection model. Rows B and C show the plasma and ELF concentrations of the lung infection model.