Application of population pharmacokinetic-pharmacodynamic modelling evaluate and optimise aminoglycoside therapy in patients with cystic fibrosis

Rights statement
Awarding institution
  • University of Strathclyde
Date of award
  • 2013
Thesis identifier
  • T13574
Qualification Level
Qualification Name
Department, School or Faculty
  • Cystic fibrosis (CF) is an inherited autosomal recessive disorder that is characterised by frequent lung infections commonly caused by P.aeruginosa. The standard treatment for this infection is an aminoglycoside combined with a β-lactam and patients often receive multiple courses of these antibiotics over many years. Aminoglycosides are narrow therapeutic index drugs where the margin between safety and toxicity is small. Therefore, it is important to monitor patients who are on aminoglycosides to ensure safety and efficacy of therapy and advise on current and future dosage regimens. The focus of this thesis was to use population pharmacokinetic methodologies to examine how aminoglycoside pharmacokinetic parameters change over time in this group of patients and to develop and evaluate dosing regimens and data interpretation methods. A population pharmacokinetic analysis was first conducted using the package NONMEM with the FOCE (parametric) algorithm. Aminoglycoside concentration-time profiles were available from 166 patients treated within the Glasgow Cystic Fibrosis Unit and comprised 1075 courses of therapy and 2238 concentration measurements collected over 15 years. The final, two compartment, population model identified an influence of height and creatinine clearance on clearance and height on volume of distribution of the central compartment. Inclusion of these descriptors reduced between subject variability from 23% to 18% for clearance and 14% to 12% for volume of distribution of the central compartment. Within-subject variability was low at 11%, and there were no changes in aminoglycoside clearance over time. Internal valuation of the population model using bootstrap, prediction corrected visual predictive check and normalised prediction distribution errors indicated that this model was stable and with good predictive ability. In addition, an external model evaluation was conducted using data from The Hague that comprised tobramycin concentration measurements from 165 patients who received 415 courses of therapy. The results of this analysis indicated good performance of the model in predicting pharmacokinetic parameters and concentrations in another group of patients with cystic fibrosis. The combined Glasgow and The Hague datasets were subsequently analysed using a non-parametric approach with the software Pmetrics (Neely MN et al., 2012). In total, data from 331 patients with 1490 courses of therapy and 3690 aminoglycoside concentration measurements were analysed. Despite the different assumptions of the two methods, the final models were the same and the final parameter estimates were very similar. The standard dose of aminoglycoside used in patients with cystic fibrosis is 10 mg/kg administered once daily. The typical daily area under the concentration-time curve (AUC) arising from this dose was determined using pharmacokinetic parameter estimates reported in the TOPIC study (Smyth A et al., 2005) and by examining the raw data from patients within The Hague dataset who received this dosage regimen. The results of this analysis led to a target daily AUC of 106 mg.h/L (range 80-120 mg.h/L). A simulated dataset of 5000 patients was created with clinical characteristics based on patients with cystic fibrosis from Glasgow and The Hague. The final population model was then used to estimate pharmacokinetic parameters and to predict concentrations at defined time points according to the standard dose of 10 mg/kg/day and three alternative regimens (13 mg/kg/day lean body weight, 3 mg/cm/day and 326 mg/m2/day). It was found that the dose based on height (3mg/cm/day) had the highest probability of achieving the combined targets of daily AUC range, peak concentrations of 20-30 mg/L and trough concentrations < 1 mg/L. For standard "once daily" aminoglycoside therapy, dosage adjustment nomograms are available that help clinicians to interpret aminoglycoside concentrations and advise on dose adjustments (Nicolau DP et al., 1995). For adult patients with cystic fibrosis there is no dosage adjustment nomogram available for the 10 mg/kg aminoglycoside dose. Therefore, one of the aims of the thesis was to develop an aminoglycoside (tobramycin) dosage adjustment nomogram that could be used with doses of 10 mg/kg/day and 3 mg/kg/day. The nomogram was derived from the concentration-time profiles that were generated from the simulation approach described above and consisted of three areas representing below, within and above the target ranges. Preliminary validation work indicated that the nomogram could identify patients with low, within and above target daily area under the concentration-time curve range from one sample point. Importantly, the nomogram was able to identify patients with poor renal function. The goal for any antibacterial therapy is to ensure efficacy against treated organism and to achieve high probability of treatment success. In this thesis, the likelihood of treatment success for the 10 mg/kg and 3 mg/cm dosage regimens were determined against P.aeruginosa. This was achieved by determining the susceptibility breakpoint and cumulative fraction of response for these dosage regimens and comparing them with the breakpoints obtained from the European Committee on Antimicrobial Susceptibility Testing (EUCAST) and the British Society for Antimicrobial Chemotherapy (BSAC). Breakpoints and cumulative fraction of response were also examined for doses of 12 mg/kg/day and 4 mg/cm/day. The results showed that these regimens had similar MIC breakpoints of ≤ 2 mg/L to achieve a Peak/MIC ratio ≥ 10, and MIC breakpoint of ≤ 0.5 mg/L for a daily AUC/ MIC ratio ≥ 100 mg.h/L against P.aeruginosa. However, they were lower than the EUCAST and BSAC susceptibility breakpoints against gram-negative pathogens (≤ 4 mg/L). Analysis of the cumulative fraction of response identified an overall treatment success of more than 90 % with all regimens for a Peak/MIC ratio ≥ 10 against P.aeruginosa using the EUCAST MIC distribution. At a daily AUC/MIC ratio greater than 100, the cumulative fraction of response for tobramycin indicated a success rate between 70 - 80 % for all dosage regimens, with the higher values being observed with the doses of 12 mg/kg/day and 4 mg/cm/day. However, the 12 mg/kg/day dosage regimen was associated with high peak and daily exposure and might result in more toxicity compared with the high 4 mg/cm/day dosage regimen. The population model developed in this thesis was able to describe and predict the handling of aminoglycoside in patients with cystic fibrosis. The model was used to evaluate the current 10 mg/kg/day dosage regimen, develop a new dosage regimen and develop a dosage adjustment nomogram for clinical application. Furthermore, the model was used to predict the efficacy of the standard and new dosage regimens through determining the susceptibility breakpoints and cumulative fraction of response against gram-negative organisms. In the future, the models could be used to help estimate individual pharmacokinetic parameters and design individualised dosage regimens using both parametric and non-parametric clinical pharmacokinetic software.
Resource Type
Date Created
  • 2013
Former identifier
  • 1001650