Acute-on-chronic liver failure alters meropenem pharmacokinetics in critically ill patients with continuous hemodialysis: an observational study

The study was approved by the Ethics Committee of the Hamburg Chamber of Physicians, Germany (Reference: PV5415). Consent was obtained from the patients’ closest relatives or legal surrogates.

Patients eligible for this open-label observational prospective cohort study were receiving meropenem for clinical indication and required CVVHD. Patients < 18 years or with an extracorporeal circuit other than the CVVHD were excluded. Patients were grouped according to liver function as follows: patients with ACLF and patients without ACLF (“no liver failure”, NLF).

ACLF was defined according to the definition of the Chronic Liver Failure (CLIF) Consortium [3]. Presence of liver cirrhosis was diagnosed based on a combination of characteristic clinical (e.g., ascites, caput medusae, spider angiomata, etc.), laboratory and radiological findings (typical morphological changes of the liver, signs of portal hypertension, etc., in ultrasonography or computed tomography scanning), or via histology, if available [12].

All patients received meropenem 1 g quid 8 h (Dr. Friedrich Eberth Arzneimittel GmbH, Germany). Meropenem was diluted in 50 mL isotonic saline solution and given over 30 min by syringe pump via a central venous line (short-term infusion).

We obtained prefilter blood samples at the following time points: T0 as the baseline before the first monitored infusion, 1 h (T1), 2 h (T2), 4 h (T4), 6 h (T6) and 8 h after the start of infusion (T8). T8 was obtained before the next infusion of meropenem as a minimum concentration. Furthermore, we obtained values after 24 h (before and 30 min after end of infusion, T24 and T25) and after 48 h (T48 and T49). All samples were centrifuged immediately, and the supernatant stored at − 20 °C until assayed.

An aliquot of a human serum sample (250 µL) was mixed with 50 µL of the internal standard ertapenem (0.2 mg/mL). 500 µL acetonitrile was added to precipitate serum proteins. After centrifugation at 14500 rpm for 5 min at 20 °C, 100 µL of the supernatant was diluted to a total of 1000 µL with water in a glass microvial. The injection volume was 50 µL.

A validated high-performance liquid chromatography with diode array detection (HPLC–DAD) method was used for the analysis of meropenem serum samples. Chromatography was performed on a Thermo Scientific Accela Liquid chromatography system consisting of an autosampler, quaternary pump and a photo diode array detector with a thermostated column compartment (Thermo Fisher Scientific, Waltham, MA, USA). The chromatographic separation of meropenem and internal standard was carried out on a reversed phase Varian Polaris C18-A (250 × 4.0 mm) with particle size of 5 µm and a C18-pre-column (SecurityGuard^™ Phenomenex^®, Aschaffenburg, Germany) using a flow rate of 1.0 mL/min at 25 °C. The mobile phases consisted of water with 0.2% ortho-phosphoric acid added (A) and acetonitrile (B). The following gradient program was used (proportion of solution (B): 5% at 0.0 min, 20% at 6.0 min, 70% at 11.0 min and 5% at 14.0 min. Between injections, the sampling needle was flushed with 400 µL and washed with 200 µL methanol. The wavelength detection was set at 260 nm and 310 nm. Chromeleon™ 7 Chromatography Data Systems Software (Thermo Fisher Scientific, Waltham, MA, USA) was used for the control of the instruments and data acquisition. The assay was routinely calibrated using six standards of spiked blank human serum (1, 10, 20, 50, 100, 150 mg/L) and there were also two independently prepared quality control samples (16 and 64 mg/L) included in each analytical series. Validation parameters including accuracy, interferences, linearity of calibration, matrix effects and in-process stability complied with international standards. Regular external quality control is performed by periodical proficiency testing. The limit of quantification was 1 mg/L.

CVVHD was performed with Multifiltrate^® dialysis machines using an Ultraflux^® AV1000S hollow-fiber hemofilter (Fresenius Medical Care, Bad Homburg, Germany) with a membrane surface area of 1.8 m². Dialyzers and lines were steam sterilized. A regional citrate-calcium anticoagulation was used. No filter change occurred during the study period. The targeted dialysate dose was 30 ml/kg/h of actual body weight.

Additional data were obtained from the patients’ electronic records (Integrated Care Manager ICM, version 9.1, Drägerwerk, Lübeck, Germany, and Soarian Clinicals 4.01 SP08, Cerner Health Services, Idstein, Germany).

The Acute Physiology and Chronic Health Evaluation II (APACHE II) score [13] and the Sequential Organ Failure Assessment (SOFA) score [14] were recorded on the first day of examination as measures of disease severity. ACLF patients were further characterized by the Model of End-Stage Liver Disease (MELD) score, the Chronic Liver Failure Consortium (CLIF)-SOFA score, and the CLIF-lactate score [12].

Microsoft Excel 2016 (Microsoft Corp., Redmond, WA, USA) was used for data management. The SPSS statistical software package (version 25, IBM Inc., Armonk, NY, USA) was used for descriptive statistical analysis. The pharmacokinetic analysis and Monte Carlo analysis were done with the non-linear mixed-effects modeling software NONMEM, version 7.4 (Icon Development Solutions, Ellicott City, MD, USA). Data are given as median and quartiles.

The population PK analysis comprised the development of a structural and statistical base model followed by a categorical evaluation of PK parameter differences between the ACLF and NLF patient population. One and two-compartment PK models with serum data attributed to the central compartment were evaluated. Elimination from the central compartment and intercompartmental distribution were assessed using linear processes via ordinary differential equations. Interindividual variability as described by the coefficient of variation of the exponential model was implemented on clearance, intercompartmental clearance, and peripheral volume of distribution (V₂). Interindividual variability in PK parameters, and different residual unexplained variability models to quantify the difference between the model predictions and observations were investigated. The informativeness of interindividual variability and residual unexplained variability parameters was assessed based on shrinkage [15]. The model quality was evaluated by visual inspection of the observed and predicted meropenem concentrations, residuals and individual predicted PK parameter scatter plots as well as visual predictive checks (n = 1000 simulations) [16]. Statistical comparisons between nested models with additional covariates were made using the likelihood-ratio test [17]. Because of limited data for the first declining phase of the meropenem concentration–time profile and large parameter imprecision and parameter correlation between intercompartment clearance and central volume of distribution (V₁), V₁ was fixed at 8.31 L as plausible value previously described for critically ill patients [18].

Based on the PK parameters of the final population PK model, Monte Carlo simulations (n = 1000) were performed to determine the PTA for different meropenem dosing regimens in ACLF and NLF patients. PTA was defined as time above 4× the European Committee on Antimicrobial Susceptibility Testing (EUCAST) epidemiological cut-off values (ECOFF) as typical minimum inhibitory concentration (MIC) values [19, 20] of the free fraction of meropenem (fT > _4×ECOFF). PTA for achieving a PK/pharmacodynamic target of 95% of fT > _4×ECOFF was calculated on the first day of the therapy and of 100% fT > _4×ECOFF at steady-state on day 7. A dosing regimen was considered adequate if the PTA was ≥ 90%.

PTA was calculated for the following ECOFFs: 0.25 mg/L for Enterobacterales and 2.0 mg/L for Pseudomonas aeruginosa or Acinetobacter baumannii and the following dosing regimens: meropenem 1 g short-term infusion (over 30 min) quid 8 h, 1 g quid 8 h as prolonged infusion over 4 h after a short-term infusion loading dose of 2 g and without loading dose, 2 g short-term infusion quid 8 h, and continuous infusion of meropenem 3 g/day after a short-term infusion loading dose of 2 g.

In total, 19 meropenem measurements were available per sampling time point for the first study dosing interval on day 1 and day 2 with values missing for T4 to T8 for 2 patients and 17 measurements per sampling time point available on day 3 (n_total = 180). The meropenem clearance of the CVVHD was 4.99 (4.32–5.75) L/h.

Meropenem PK in patients with and without ACLF was described by a two-compartment PK model with linear clearance from the central compartment. Interindividual variability for clearance was coefficient of variation = 28% and on V₂ = 66%. Separate estimation of V₂ yielded a V₂ of 18.5 L for NLF patients and 35.5 L for ACLF patients (p = 0.05, likelihood-ratio testing), which largely explained interindividual variability in V₂ (61.9% relative reduction). The remaining interindividual variability in V₂ was estimated imprecisely and suffered from a large shrinkage (50.8%), indicating non-informativeness of this parameter by the data. Due to the model stability, it was excluded in the final model. No difference in clearance between the two patient populations was evident (ACLF: 5.20 L/h, NLF: 5.11 L/h; p = 0.90). The half-lives of ACLF compared to NLF patients were t_½α: 0.43 vs. 0.40 h (p = 0.60) and t_½β: 9.0 vs. 5.0 h (p = 0.002). The final model parameter values are depicted in Table 2.

PTA for the lower ECOFF of 0.25 mg/L (Enterobacterales) was 100% for all dosing regimens on day 1 (Fig. 1) and on day 7 (Fig. 2) in both groups. For the higher ECOFF of 2.0 mg/L (P. aeruginosa, A. baumannii), meropenem 1 g short-term infusion quid 8 h yielded a PTA of 18% in the ACLF group and 48% in the NLF group on day 1 and 80% for ACLF and 65% for NLF patients on day 7. With prolonged infusion of 1 g quid 8 h, PTA was increased to 30% and 67% on day 1 and 88% and 81% on day 7 for ACLF and NLF patients, respectively. For a prolonged infusion after a short-term infusion loading dose of 2 g, PTA was higher with 94% and 91% on day 1 and 88% and 83% on day 7. The dosing regimen of 2 g short-term infusion quid 8 h yielded a PTA of 85% and 91% on day 1 and 98% and 93% on day 7. The continuous infusion of 3 g quid 8 h after a short-term infusion loading dose of 2 g reached a PTA of 100% for both groups on days 1 and 7.

Therefore, adequate PTA was achieved for all dosing regimens up to an ECOFF of 0.25 mg/L. However, for the higher ECOFF of 2 mg/L, only a 2 g short-term infusion followed by a continuous infusion of 3 g/day reached adequate PTA on day 1 and 7 in both groups. A dosing regimen of 2 g quid 8 h was sufficient on day 1 only in the NLF group and in both groups at steady-state.