IV and oral fosfomycin pharmacokinetics in neonates with suspected clinical sepsis

Abstract

Background: Fosfomycin has the potential to be re-purposed as part of a combination therapy to treat neonatal sepsis where resistance to current standard of care (SOC) is common. Limited data exist on neonatal fosfomycin pharmacokinetics and estimates of bioavailability and CSF/plasma ratio in this vulnerable population are lacking.

Objectives: To generate data informing the appropriate dosing of IV and oral fosfomycin in neonates using a population pharmacokinetic analysis of plasma and CSF data.

Methods: The NeoFosfo study (NCT03453177) was a randomized trial that examined the safety and pharmacokinetics of fosfomycin comparing SOC versus SOC plus fosfomycin. Sixty-one neonates received fosfomycin (100 mg/kg IV q12h for 48 h) and then they converted to oral therapy at the same dose. Two plasma pharmacokinetic samples were taken following the first IV and oral doses, sample times were randomized to cover the whole pharmacokinetic profile and opportunistic CSF pharmacokinetic samples were collected. A population pharmacokinetic model was developed in NONMEM and simulations were performed.

Results: In total, 238 plasma and 15 CSF concentrations were collected. A two-compartment disposition model, with an additional CSF compartment and first-order absorption, best described the data. Bioavailability was estimated as 0.48 (95% CI = 0.347-0.775) and the CSF/plasma ratio as 0.32 (95% CI = 0.272-0.409). Allometric weight and postmenstrual age (PMA) scaling was applied; additional covariates included postnatal age (PNA) on clearance and CSF protein on CSF/plasma ratio.

Conclusions: Through this analysis a population pharmacokinetic model has been developed that can be used alongside currently available pharmacodynamic targets to select a neonatal fosfomycin dose based on an infant's PMA, PNA and weight.

Figure 1.

Observed CSF and plasma concentration versus time data for all subjects. The dashed lines represent mean concentrations, which are 37.6 mg/L in CSF and 70.1 and 201.7 mg/L in plasma following oral and IV dosing, respectively; the solid lines represent the loess smooth curves. PO, oral.

Figure 2.

VPCs showing the observed data (black circles) and the 2.5th, 50th and 97.5th percentiles of the observed data (black lines) compared with the 95% CIs of the corresponding simulations (prediction intervals) from the final model (shaded areas). The top panel shows plasma following IV dosing, the middle panel shows plasma following oral dosing and the bottom panel shows CSF. The 2.5th and 97.5th percentiles, and corresponding prediction intervals, are not presented in the bottom panel due to the size of the CSF dataset evaluated.

Figure 3.

Visualization of the PNA effect on clearance; individual predicted clearances have already been scaled for PMA and weight. Data points are grouped by subject ID (n = 60); the left-hand panel shows individual fractional clearance at time = 0 and the right-hand panel shows individual fractional clearance at all PK sampling timepoints. The solid black line in each panel represents the model estimated PNA maturation function.

Figure 4.

TA plots for various dose schemes using the full simulation population. The top row presents predicted AUC/MIC ratio in plasma following IV dosing, while the bottom row shows results following oral dosing. A comparison of 100, 150 and 200 mg/kg q12h is given for IV and 100, 200 and 300 mg/kg q12h for oral. The continuous black line is the predicted AUC/MIC ratio achieved by 95% of the population (5th percentile), while the typical patient (50th percentile) is shown by the dotted line. AUC/MIC target ratios for stasis (19.3), 1 log reduction (87.5) and resistance suppression (3136) are shown by the grey horizontal reference lines. The grey vertical reference lines highlight MIC values of 4 and 32 mg/L. BID, twice daily; PO, oral.