Irene-Ariadne Kechagia, Zoe Athanassa, Sophia Markantonis, Aris Dokoumetzidis
School of Pharmacy, University of Athens, Greece
Objectives: Colistin is an antimicrobial agent administered to treat multidrug-resistant gram-negative bacterial infections [1]. The aim of the present study was to develop a population PK model of colistin after inhalation of colistin methanesulfonate (CMS), an inactive prodrug of colistin. Nebulized CMS was administered via a vibrating-mesh nebulizer to critically ill patients suffering from ventilator-associated tracheobronchitis (VAT).
Methods: Pharmacokinetic data were obtained from a previous study [2] and included concentrations of colistin in serum and in lung epithelial lining fluid (ELF) after the first inhalation of CMS. Population pharmacokinetic analysis was performed using the FOCE method with interaction in NONMEM (version 7.2). Model parameters clearance (CL/F), volume (V/F), Weibull scale (a) and shape parameters (b) together with OMEGAs for CL/F, V/F and b were estimated. Diagnostic graphs were generated by the Xpose program (version 4). Demographical and biochemical data were tested as covariates for the pharmacokinetic parameters. Visual predictive check and a nonparametric bootstrap resampling method were performed to evaluate the model. Also, a monoexponential model was used to fit each patient's ELF data to obtain the elimination rate constant of colistin from ELF (ke,ELF) which was tested as a covariate on absorption parameters.
Results: Colistin pharmacokinetics was best described by a one compartment model with first order elimination and absorption being modeled by the weibull distribution function. Residual variability was modeled by a time dependent error model to account for higher residual error during the absorption phase. Furthermore, due to unusually high Cmax values attributed to suspected hydrolysis of CMS to colistin during sample analysis, a systematic bias was introduced to the residual error. Creatinine clearance was a significant covariate for CL/F and ke,ELF for the shape parameter of weibull distribution, b. The population mean estimates for apparent clearance (CL/F) and apparent volume of distribution (V/F) were 6.76 liters/h (CV 28%) and 37.2 liters (CV 19%), respectively.
Conclusions: A population PK model for colistin was developed. A residual error including systematic bias allows interpretation of unusually high Cmax values suspected to be erroneous.
References:
[1] Couet W, Grégoire N, Marchand S, Mimoz O. Colistin pharmacokinetics: the fog is lifting. Clin Microbiol Infect. 2012 Jan;18(1):30-9.
[2] Athanassa ZE, Markantonis SL, Fousteri MZ, Myrianthefs PM, Boutzouka EG, Tsakris A, Baltopoulos GJ. Pharmacokinetics of inhaled colistimethate sodium (CMS) in mechanically ventilated critically ill patients. Intensive Care Med. 2012 Nov;38(11):1779-86.
Reference: PAGE 22 (2013) Abstr 2868 [www.page-meeting.org/?abstract=2868]
Poster: Infection