Prediction of subtherapeutic tigecycline plasma levels by model-based Bayesian individualization
Maria Sfouni, Irene-Ariadne Kechagia, Sophia Markantonis, Aris Dokoumetzidis
School of Pharmacy, University of Athens, Greece
Objectives: To investigate the optimal sampling times for estimating posthoc clearance of tigecycline and assess their performance, in terms of predicting response, as early as possible, in the treatment of complicated skin and skin-structure infections (cSSSIs) and complicated intra-abdominal infections (cIAIs).
Methods: A large set of combinations of 4 sampling times was evaluated in a dataset of 1000 subjects, generated in MATLAB. For each combination, concentration-time data were simulated in NONMEM 7.3 from a literature two-compartment population PK model [3] and the EBEs of the PK parameters were subsequently estimated. Greater emphasis was placed on systemic clearance (CL), since tigecycline follows linear kinetics and the AUC24/MIC ratio is the PK-PD index predictive of efficacy with breakpoints of 17.9 (cSSSIs) [1] and 6.96 (cIAIs) [2].The optimality criteria were the absolute relative mean prediction error (|MPE%|) and the 90th percentile of the |MPE%| of the AUC24 as measures of bias, the relative root mean square prediction error (RMSE%) of the AUC24 as a measure of precision and the ETA shrinkage of CL as a measure of informativeness. In addition to these, the D-optimality criterion of the Fisher information matrix (FIM) of the typical individual was maximized. At a second level, the assessment of predictive performance of the optimal time combinations was based on metrics derived from the confusion matrices corresponding to each MIC of interest [4] for both cSSSIs and cIAIs.
Results: According to the optimality criteria considered, four sampling schemes were chosen to have good performance for estimating the posthoc clearance, hence the AUC24/MIC ratio, taking also into account practical considerations. Up to 72 hours, the scheme 61, 64, 71, and 72 hours is the optimal, however schemes with performance close to that but with earlier sampling times were also identified, that may be more useful. The predictive performance of all sampling schemes varies across the MIC range, still the selected combinations exhibit a comparatively high performance.
Conclusions: Detecting subtherapeutic antibacterial plasma levels as early as possible in treatment, can contribute to the reduction of the development rate of resistance. The results of an optimality study for tigecycline, a mainly bacteriostatic antibacterial, characterized by a long elimination half-life with high inter-individual variability are presented, but the methodology can be expanded to other similar agents.
References:
[1] Meagher, A. K., Passarell, J. A., Cirincione, B. B., Van Wart, S. A., Liolios, K., Babinchak, T., Ellis-Grosse, E. J. and Ambrose, P. G. (2007). Exposure-response analyses of tigecycline efficacy in patients with complicated skin and skin-structure infections. Antimicrobial agents and chemotherapy, 51(6), 1939-1945.
[2] Passarell, J. A., Meagher, A. K., Liolios, K., Cirincione, B. B., Van Wart, S. A., Babinchak, T., Ellis-Grosse, E. J. and Ambrose, P. G. (2008). Exposure-response analyses of tigecycline efficacy in patients with complicated intra-abdominal infections. Antimicrobial agents and chemotherapy, 52(1), 204-210.
[3] Van Wart, S. A., Owen, J. S., Ludwig, E. A., Meagher, A. K., Korth-Bradley, J. M. and Cirincione, B. B. (2006). Population pharmacokinetics of tigecycline in patients with complicated intra-abdominal or skin and skin structure infections. Antimicrobial agents and chemotherapy, 50(11), 3701-3707.
[4] Xie, J., Wang, T., Sun, J., Chen, S., Cai, J., Zhang, W., Dong, H., Hu, S., Zhang, D., Wang, X., Dong, Y. (2014). Optimal tigecycline dosage regimen is urgently needed: results from a pharmacokinetic/pharmacodynamic analysis of tigecycline by Monte Carlo simulation. International Journal of Infectious Diseases, 18, 62-67.
