Irina Bondareva (1), Sergey Zyryanov (1,2), Mikhail Chenkurov (1,2)
(1) RUDN University, Moscow, Russia, (2) State Budgetary Institution of Healthcare
Objectives: Meropenem is a broad-spectrum carbapenem antibiotic widely used in the ICU to treat severe infections. In critically ill patients, the main alterations of meropenem pharmacokinetics are augmented renal clearance, impaired renal clearance and increased volume of distribution [1]. In ICU patients, significant PK/PD variability leads to high risk of underdosing or overdosing with standard dosing, and individualizing dosage regimens are increasingly advocated [1- 5]. Meropenem is a time-dependent antibiotic, whose antibacterial activity is associated with the fraction of time (fT) for which the free concentration is maintained above the minimum inhibitory concentration (MIC).
The study objectives are:
- Develop a population model of meropenem pharmacokinetics from TDM data of adult ICU patients;
- Investigate relationship between creatinine clearance (CLCr), weight, age, gender and meropenem PK parameters in the ICU patients;
- Calculate PD index (fT>MIC) using PK parameter estimates for different dosage regimens and MIC levels by category of renal function.
Methods: TDM data were routinely collected in the ICU. Meropenem daily doses (1-6 gram) were administered by 0.5-3hrs IV infusion 1-3 times a day. For the PK analysis, demographic and clinical characteristics as well as TDM data were retrieved from patients’ records retrospectively. Blood samples were collected at predose (Ctrough) and at the end of IV infusion (peak levels). Meropenem concentrations were measured by high performance liquid chromatography. PK analysis was performed using the Pmetrics software based on the linear one-compartment model and TDM data of 131 patients. Totally, 223 concentrations of 116 patients were included in the meropenem population model. For 90(77.6%) patients, TDM was performed within first four days of their antibiotic therapy. PK data of 15 patients on dialysis were described separately.
Results:
The mean age of the included subjects was 67.7±13.6 (24 – 93) years, about 60% of patients were male. At blood probe, weight was 79.5±19.7 kg; CLCr (estimated by the Cockcrofte – Gault formula) was 4.24±2.27 L/h. The geometric means of predose and peak concentrations were 5.9 and 36.0 μg/ml. In 15 (11.5%) of patients, their Ctrough levels on the empirical dosage regimens were below 2 μg/ml (the S/I EUCAST breakpoint), Ctrough of ≥ 8 μg/ml (the I/R EUCAST breakpoint) was reached in 50(38.2%) patients, 12(9.2%) patients had Ctrough above 16 μg/ml, Ctrough in 3(2.3%) patients exceeded toxic breakpoint of 44.5 μg/ml [1]. The mean ± SD (median) values for constant rate of elimination (Kel) and volume of distribution (Vd) of meropenem were estimated as 0.351±0.160 (0.321) 1/h and 27.2±15.91 (25.0) L with interindividual variability (CV) of 45.4 and 58.6%, respectively. The median values for total clearance (CL) and elimination half-life (T1/2) were calculated as 8.1 L/h and 2.14 h with CV= 55.0 and 74.5%. Results of the predictive performance demonstrated that the population model by itself gives poor prediction (all concentrations were predicted based on the population mean parameter values), especially for higher serum levels, while the individualized Bayesian posterior models give much improved prediction (bias: 10.4 versus 0.41; R2: 39.4 versus 99.2%). For external validation, a data set (TDM of 25 patients), which has not been included in the model parameter estimation was used.
Different regression lines between meropenem CL and CLCr were obtained in patients with CLCr ≤ 7L/h versus > 7L/h: statistically significant regression CL=0.95*CLCr+4.3 (n=96, p<0.001) versus no correlation (n=20, р=0.62), respectively. The regression analysis revealed that age negatively influenced the meropenem CL (p=0.009). The mean CL was significantly lower in the elderly compared to that estimated in younger patients (10.3 versus 8.0L/h, p=0.032) due to reduced renal function. On average, the lower %T>MIC values were associated with higher clearance.
Conclusions: The proposed population model can be used as a Bayesian prior information to estimate individual PK parameter values in ICU patients and to obtain predictions of desired PK/PD target once the patient’s TDM samples become available (even for the first IV dose). The intraindividual PK alterations may be influenced by the course of the disease, especially during first days of therapy (estimated median decrease in Vd=26.7%), and repeated TDM can be used to adjust the dosage accordingly.
References:
[1] Abdul-Aziz Mohd H et al. Intensive Care Med, 46:1127–1153, 2020.
[2] Gonçalves-Pereira J, Póvoa P. Critical Care, 15:R206, 2011, http://ccforum.com/content/15/5/R206.
[3] Roberts J, Kumar A, Lipman J. Crit Care Med, 45: 331–6, 2017.
[4] Scharf Ch et al. Antibiotics, 9, 131, 2020,
[5] Dhaese S et al. Expert Review of Anti-infective Therapy, 2020, https://doi.org/10.1080/14787210.2020.1788387
Reference: PAGE 29 (2021) Abstr 9827 [www.page-meeting.org/?abstract=9827]
Poster: Drug/Disease Modelling - Infection