Modelling of plasma unbound Ceftriaxone: can we improve efficacy in clinical practice ?
Johnny Michel (1), Tania Petersen (2), Francesco Monti (3), Sandrine Dahyot (2), Tony Pereira (4), Martine Pestel-Caron (2), Julien Grosjean (3), Fabien Lamoureux (4,5), Thomas Duflot (4,5)
(1) Emergency Department, CHU Rouen, F-76000 Rouen, France, (2) Department of Bacteriology, CHU Rouen, F-76000 Rouen, France, (3) Department of Digital Health, CHU Rouen, F-76000 Rouen, France, (4) Department of Pharmacology, CHU Rouen, F-76000 Rouen, France, (5) INSERM EnVI UMR 1096, Health Campus, University of Rouen Normandie
Introduction: Ceftriaxone, a third generation semisynthetic cephalosporin with a long half-life, is a cornerstone antibiotic for critically ill patients with severe infections. Despite its widespread use, further investigations are needed to predict plasma unbound (active) Ceftriaxone concentrations since it is highly bound to albumin in a saturable manner.
Objectives:
- predict the unbound fraction of Ceftriaxone in plasma according to different models
- evaluate which model best fits human plasma unbound Ceftriaxone quantified in our laboratory
- determine the best thresholds of plasma total Ceftriaxone concentrations according to minimum inhibitory concentration (MIC) using fT> MIC 100% and fT > 4 × MIC 100%
- analyze retrospectively total Ceftriaxone trough concentrations measured for therapeutic drug monitoring in our institution
Methods: A literature review was conducted to search for published models of unbound Ceftriaxone. Eleven total and unbound human plasma concentrations of ceftriaxone were measured in the Department of Pharmacology of the Rouen University Hospital in addition to albuminemia and compared with the predicted values of the different models. A target MIC of 1mg/L, considered as target attainment by the European Committee on Antimicrobial Susceptibility Testing in the case of community-acquired pneumonia (CAP), caused by Enterobacterales and Moraxella cattarhalis infections, was used to determine total ceftriaxone thresholds according to albuminemia. The healthcare data warehouse of our local institution was used to extract data from patients treated by Ceftriaxone in the past 5 years. Three hundred seventy six samples of plasma total Ceftriaxone and albuminemia were collected, allowing unbound Ceftriaxone prediction.
Results: Six models for unbound Ceftriaxone modelling has been investigated: the linear binding (89.5% of protein binding) [1], the empirical in vitro saturation model [2] and 4 in vivo models of saturable protein binding (Bos, Gijsen, Leegwater and Hartman) [1, 3-5]. In vivo models relied on the total density (concentration) of receptors (Bmax) and the dissociation constant (Kd). Albuminemia was used as a predictor of Bmax for all in vivo models and hypoalbuminemia was associated with an increased unbound fraction of Ceftriaxone. A high between model variability which increased with total Ceftriaxone concentration was observed for unbound fraction. Comparison of quantified total and unbound Ceftriaxone from our institution and model predictions revealed that Hartman model best fit our data (R²=0.89) followed by Leegwater model (R²=0.82), Bos model (R²=0.80), Gijsen model (R²=0.79), linear model (R²=0.63) and empirical in vitro model (R²=0.50). Using the Hartman model, total Ceftriaxone trough concentration of 10.3 mg/L and 37.8 mg/L were needed to reach fT > MIC 100% and fT > 4 × MIC 100% respectively for patients with an albumin concentration of 35 g/L. In the case of hypoalbuminemia (20g/L), these concentrations decreased to 6.3 mg/L and 23.3 mg/L respectively. Retrospective analysis of total Ceftriaxone trough concentrations from the healthcare data warehouse showed a median predicted unbound Ceftriaxone of 11.2 mg/L [interquartile range = 6.1 mg/L - 19.5 mg/L]. Importantly, 10/376 (2.7%) and 43/376 (11.5%) patients did not meet the required fT> 1 × MIC 100% and fT > 4 × MIC criteria respectively.
Conclusions: In vivo models using saturable binding of Ceftriaxone to albumin exhibited better prediction than linear and in vitro estimation. The four studied in vivo models showed pretty close goodness-of-fit based on R² with excellent predictions for the Hartman model. Overall, depending on the target attainment criteria (fT> 1 × MIC 100% and fT > 4 × MIC), dosage adjustment may be suggested to improve efficacy. Further prospective investigations are needed to confirm that Hartman model may be used as an effective formula to predict unbound Ceftriaxone concentration.
References:
[1] Gijsen M, Dreesen E, Van Daele R, et al. Pharmacokinetic/Pharmacodynamic Target Attainment Based on Measured versus Predicted Unbound Ceftriaxone Concentrations in Critically Ill Patients with Pneumonia: An Observational Cohort Study. Antibiotics (Basel). 2021;10(5):557. Published 2021 May 11. doi:10.3390/antibiotics10050557
[2] McNamara PJ, Trueb V, Stoeckel K. Protein binding of ceftriaxone in extravascular fluids. J Pharm Sci. 1988;77(5):401-404. doi:10.1002/jps.2600770509
[3] Bos JC, Prins JM, Mistício MC, et al. Pharmacokinetics and pharmacodynamic target attainment of ceftriaxone in adult severely ill sub-Saharan African patients: a population pharmacokinetic modelling study. J Antimicrob Chemother. 2018;73(6):1620-1629. doi:10.1093/jac/dky071
[4] Leegwater E, Kraaijenbrink BVC, Moes DJAR, Purmer IM, Wilms EB. Population pharmacokinetics of ceftriaxone administered as continuous or intermittent infusion in critically ill patients. J Antimicrob Chemother. 2020;75(6):1554-1558. doi:10.1093/jac/dkaa067
[5] Hartman SJF, Upadhyay PJ, Hagedoorn NN, et al. Current Ceftriaxone Dose Recommendations are Adequate for Most Critically Ill Children: Results of a Population Pharmacokinetic Modeling and Simulation Study. Clin Pharmacokinet. 2021;60(10):1361-1372. doi:10.1007/s40262-021-01035-9