Meropenem pharmacokinetic/pharmacodynamic target attainment and clinical response in ICU patients: A prospective observational study

Several studies report lack of meropenem pharmacokinetic/pharmacodynamic (PK/PD) target attainment (TA) and risk of therapeutic failure with intermittent bolus infusions in intensive care unit (ICU) patients. The aim of this study was to describe meropenem TA in an ICU population and the clinical response in the first 72 h after therapy initiation.


Editorial Comment
This study provides interesting data on the pharmacokinetic and pharmacodynamic properties of meropenem in ICU patients.

| INTRODUCTION
Meropenem is a broad spectrum β-lactam antibiotic, frequently used in patients with sepsis and septic shock.The meropenem antibacterial effect is dependent on the time the concentration of meropenem remains above the minimum inhibitory concentration (% T > MIC). 1 Meropenem is a hydrophilic drug with low protein affinity.Its serum concentrations are affected by changes in the volume of distribution and renal clearance. 2,3[7] These studies conclude that renal function is a good predictor of meropenem PK, and should be considered when treating this group of patients.However, renal function is frequently evaluated using the Cockcroft-Gault equation, which is an unreliable method for estimating renal function in ICU patients. 8Furthermore, patients on continuous renal replacement therapy (CRRT) are frequently not included.
Meropenem PK/PD targets for the critically ill are still disputable. 9It has been recommended longer and higher exposure, for example, 100% T > 2-5Â MIC compared with 40%-70% T > MIC. 9 MIC breakpoints defined by the European Committee of Antimicrobial Susceptibility Testing (EUCAST) are recommended to use rather than MIC measured.In clinical practice, trough concentrations >MIC is often used as a surrogate for 100% T > MIC. 4,10Independent of method, lack of target attainment (TA) has been equated to risk of therapeutic failure, while attaining levels above the highest target may increase the risk of toxicity. 9is study is a prospective observational single-center study on meropenem PK in an ICU population treated with intermittent bolus infusions of meropenem.CRRT and non-CRRT patients are included.The CRRT group includes both patients on continuous venovenous hemodialysis (CVVHD) and continuous venovenous hemofiltration (CVVHF).The non-CRRT group includes patients with normal renal clearance (NRC) and augmented renal clearance (ARC).
The aim of this study was firstly, to examine meropenem PK/PD TA with intermittent bolus infusions in a population with variable renal function.TA was defined as 100% T > 4Â MIC or trough concentrations above 4Â MIC.Furthermore, a PK model was developed for calculation of TA and dosing simulation, if applicable.
Secondly, to examine changes in clinical endpoints of infection control within each group, within the first 72 h after therapy initiation.

| Study design and participants
This prospective observational study was carried out as a part of the PharmacoCRRT2012 study (NCT01582360 at clinicaltrials.gov).The study protocol was approved by the Regional Committee for Medical and Research Ethics and the institutional Data Inspectorate Health Authority in November 2012.The study was performed at four ICUs at Oslo University Hospital (OUS), a 1200-bed tertiary hospital, including the major trauma hospital, and the national organ transplant center in Norway.
Patients were included from a neurosurgical ICU, a cardiovascular ICU, and two general ICUs.The patients were enrolled from February 1, 2014 until October 31, 2016, with a final follow-up on January 31, 2017.The inclusion criteria were initiation of meropenem therapy within the last 24 h, age ≥ 18 years, and expected treatment time with meropenem >72 h.Written informed consent from the patient or next of kin was obtained prior to inclusion.For the non-CRRT patients, 24 h urine sampling and calculation of creatinine clearance (CL cr ) were performed for three consecutive days after inclusion.ARC was defined as CL cr > 130 mL/min.For patients in need of CRRT, the ICUs consistently used either CVVHF mode (Prismaflex ® Gambro, Illinois, USA) or CVVHD mode (Multifiltrate ® Fresenius, Frankfurt, Germany).For CVVHF, the Prismaflex M-150 ® filter with an AN69 membrane was used.For CVVHD, the Ultraflux ® AV1000S filter with the Fresenius Polysulfone ® membrane was used.

| Patient data
The following information was gathered from the electronic medical records (DIPS ® , Oslo, Norway and MetaVision ® IMD-soft, Telaviv, Israel): SAPS II score at admission to ICU; SOFA score at 24, 48, and 72 h after the start of therapy; 28-and 90-days mortality; indication for meropenem therapy; and clinical, biochemical, and microbiological laboratory data results.The data were entered into a Medinsight ® database (Oslo, Norway).

| Meropenem dosage and administration
Meropenem was administered as intermittent 30 min bolus infusions.
Blood samples for meropenem trough concentrations were drawn immediately prior to the meropenem doses at 24, 48, and 72 h.Meropenem peak concentrations were sampled 15 min after meropenem infusion, and the mid sample was taken 3 h before the next dose.

| Meropenem concentration measurements
Paired heparinized blood samples of 2 mL were obtained, conserved on ice before transport and immediately centrifuged, and stored at À80 C for later analysis.A deuterated internal standard was added to the plasma sample followed by protein precipitation with acetonitrile.
The supernatants were analyzed with liquid chromatography-mass spectrometry (LC-MS).The separation of the samples was performed on a C-18 BEH UPLC column.
The proposed method has been validated for the concentration range 0.1 and 260 μmol/L, with an average R 2 of 0.99 (3 days).The intermediate precision and bias were <10%.The method was accurate and rapid with 3 min run time, and suitable for quantitative determination of meropenem in human plasma.

| Microbiology
Bacterial isolates were identified at the species level using matrixassisted laser desorption/ionization-time-of-flight mass spectrometry (Brüker Daltonics, Bremen, Germany).All isolates were subjected to susceptibility test for meropenem using the current methods at the time of the study: MIC gradient strips, Etests ® (bioMeriéux, Marcy L' Etoile, France) or VITEK ® 2 automated system (bioMeriéux).

| PK/PD calculations
Calculations of meropenem PK and the PK/PD relationship T > 4Â MIC were performed for each patient using a PK/PD calculator developed for the study.Traditional pharmacokinetic equations modified after the Sawchuk-Zaske method were used as shown in Equation (A.1) in Supporting information. 11The EUCAST MIC of 2 mg/L for Pseudomonas aeruginosa was used for the calculation of PK/PD (www.eucast.org).

| Population pharmacokinetic analysis
Total serum meropenem concentrations were modeled with the nonparametric adaptive grid algorithm within Pmetrics ® package for R ® (Los Angeles, CA, USA).One-, two-, and three-compartment models and both lambda (additive) and gamma (multiplicative) error models were tested.Biologically plausible covariates tested were weight, age, serum creatinine level, creatinine clearance, urine output, SAPS score, and presence of CVVHD/CVVHF.These covariates were tested on the volume of distribution and clearance in a forward stepwise manner.If the inclusion of a variable resulted in an increase in the coefficient of determination of the linear regression (R 2 ) and in a reduction of the bias of the goodness-of-fit plot, as well as a statistically significant reduction in the log-likelihood, the covariate was supported for inclusion.
The R ® and the bias of the observed versus predicted plots as well as the log-likelihood of each run were considered for the Time was used as a categorical covariate, but the analyses were repeated with continuous time and gave the same conclusions.To check the sensitivity of the results on the assumption of normality, we conducted a pairwise Wilcoxon test on the median, which also resulted in the same conclusions.
Linear mixed models were used for all secondary endpoints, with the same covariates and structure as for primary endpoints.
Data were analyzed using IBM SPSS statistics version 28 and R studio version 3.6.1.
Visual inspection of the PK model outputs will be used to assess the statistical strength of the model and inform the utility of dosing simulations.

| Baseline patient characteristics
Eighty-seven patients were included in the study.The baseline characteristics of the patient population and the group distributions according to renal function and CRRT mode are presented in Table 1.
Pathogens relevant to meropenem therapy were identified in positive isolates from 42 out of 87 patients, as shown in Table 2.
T A B L E 2 Identified pathogens in positive isolates from 42 patients started with meropenem therapy.

| Meropenem dosing
Meropenem was mainly given as 30 min intermittent bolus infusions of median 1 g (IQR,1-1) at a median frequency of every sixth hour (IQR, 6-8), except for every eighth hour in the CVVHD group.
Fourteen meropenem doses were given as 5 min bolus injections, and 19 doses were 2 g.

| Meropenem PK
For pharmacokinetic calculations, 699 plasma samples were analyzed.
A meropenem concentration in the middle of the dosing interval was obtained in 58% of patients.For the individual PK calculations, the peak and trough values from 470 plasma samples were used.
These results are presented in Table 3.

| Meropenem trough concentrations
There were statistically significant differences between trough concentrations in the CRRT and the non-CRRT groups, with median values across all time-points 15.5 mg/L (IQR, 8.8-27.2) and 5.9 mg/L (IQR, 2.0-12.7)respectively ( p < 0.001).Meropenem trough concentrations according to renal clearance and CRRT mode are shown in Figure 1.The trough concentrations in the ARC group were significantly lower, and in the CVVHF group, they were significantly higher than in all other groups ( p < 0.001).

| Meropenem T > 4Â MIC, individually calculated
There were statistically significant differences between the CRRT and the non-CRRT groups, with median values 100% ( p < 0.001).Meropenem T > 4Â MIC according to renal clearance and CRRT mode are shown in Figure 2.

| Pharmacokinetic model building
The population pharmacokinetic model that best described the data was a two-compartment model, incorporating a multiplicative error.
Weight was supported for inclusion as a covariate in the model on For all patients: For patients not on CRRT: For patients on CRRT: The population pharmacokinetic model data are summarized in Table 4.
The individual predicted concentration diagnostic plots for the final covariate model for total meropenem concentrations are given in Figure 3.

| Observed changes in clinical endpoints
There were significant differences in SOFA score at 24 h between the non-CRRT and the CRRT group, with median 8 (IQR, 5-10) and The main proportion of deaths, 24 out of 28, occurred in the ICU within 28 days.

| DISCUSSION
This study shows that intermittent bolus infusions of meropenem provide satisfactory median 100% TA for all groups except for patients with ARC.The increased dosing frequency of q6h, compared to standard q8h, may explain the higher TA in the present study. 413 Despite the lack of TA in the ARC group, a significant reduction in CRP and SOFA score was observed across the first 3 days of therapy.
Other studies have shown reduced PK/PD TA of β-lactams and a lack of coherence between TA and patient outcome data in ARC patients. 14In the present ICU population, the observed MIC for identified pathogens was lower than the EUCAST MIC used for PK/PD calculations.This adds up to the higher target of PD to 4Â MIC.The aggressive threshold may lead to a perception of underestimation of exposure and explain the lack of association of TA with clinical response. 15,16tisfactory TA for CRRT patients and patients with NRC with intermittent bolus infusions have been previously reported. 17,18rthermore, studies on β-lactam antibiotic administration comparing continuous or extended infusions with bolus infusions report no or negligible differences in TA or outcome results, thus questioning the necessity of prolonged or continuous infusions for all critically ill. 7,19 a recent study on critically ill patients with sepsis, compared with intermittent administration, continuous meropenem infusions did not improve the outcome of mortality. 20However, there were no reports on meropenem PK or TA in this study.
The highest TA in the present study was observed in the CRRT groups, being a consequence of reduced meropenem clearance.Furthermore, acute kidney injury is independently associated with higher morbidity and mortality.This exemplifies the challenges in analyzing the relationship between TA and clinical endpoints in an ICU population. 21e threshold of 100% T > 4Â MIC has been recommended in the presence of resistant strains, and has been used in clinical studies with an assumed worst case scenario in the absence of identified pathogens. 22It may cause unnecessary high dosing recommendations, leading to increased risk of β-lactam toxicity. 9In our study, the CVVHF group with a dosing regimen of 1 g q6h had median trough levels above 20 mg/L.Although no toxicity thresholds for meropenem are established to date, trough levels above 50 mg/L is associated with increased likelihood of neurotoxicity and nephrotoxicity. 9In a recent study, the optimal β-lactam target for infection resolution and improved outcome was 100% T > MIC, and no benefit observed for patients reaching the higher targets of 100% T > 4Â MIC. 22e surviving sepsis campaign 2021 suggests using prolonged infusion of β-lactams over conventional intermittent infusion. 23olonged or continuous infusion should be administered to selected groups of patients with verified or suspected increased renal clearance, and suspected multidrug resistant bacteria at initiation of therapy, until renal clearance, the MIC value and meropenem TDM provides a better basis for dosing and de-escalation. 24udies with population PK modeling reports variable results and lower meropenem PK/PD TAs for ICU patients than in the present study. 5,6,25,266][7] The high pharmacokinetic variability observed in the present population did not support the progress to dosing simulations from PK modeling.High variability in PK frequently observed in ICU populations requires more frequent sampling early in the dosing interval. 27Further challenges with limited sampling strategies for use in population pharmacokinetic models have been thoroughly discussed by Hovd et al. 28

| Limitations
Firstly, this study was not powered to conclude on statistical significance of differences in clinical endpoints between groups.Secondly, we were not able to gather precise weight for all the patients.Where precise weights could not be obtained, estimated weight was used.This might influence the calculations of meropenem PK.Thirdly, the sampling approach used in this study was not adequate in describing the pharmacokinetic variability for use in the final PK modeling, and not suitable in dosing simulation.
the central volume of distribution (V c ) as it improved the loglikelihood and the AIC, as well as the bias and imprecision of the observed-predicted plot.Weight was normalized to the mean weight of all the patients (81 kg) in Equation (1).Serum creatinine was normalized to the mean serum creatinine concentration of the non-CRRT patients (110 μmol/L) in Equation (2).Patients on CRRT were designated a separate non-renal clearance (CLnr) in Equation (3).

F
I G U R E 3 Diagnostic plots for the final covariate model for meropenem.Observed versus population (left) and individual (right) predicted plots.The visual predictive check associated with the final population pharmacokinetic model of the total serum meropenem concentration is presented in Figure A.1 in Supporting information.The high PK variability observed did not support the progress to dosing simulations.
-of-fit evaluation.Predictive performance evaluation was based on the mean predicted error (bias) and the mean bias-adjusted prediction error (imprecision) of the population and individual prediction models.The visual predictive check plot and the normalized prediction distribution errors as well as the Akaike information criterion (AIC), Bayesian information criterion, and the log-likelihood ratio were considered to compare different models.The primary endpoints were analyzed at each time-point(24, 48,and 72 h) with pairwise t-tests adjusted for multiple comparisons, and with one-way ANOVA to estimate the overall group effect.To include all the three time-points, a linear mixed model was fitted, with group and time as fixed effects and patient ID as random effects to account for dependence across measurements taken on the same patient.
Creatinine clearance, mL/min non-CRRT patients (n = 53) Augmented renal clearance (n = 15) 177 (142-202) Note: Data are presented as median with interquartile range (IQR), unless stated otherwise.Abbreviations: CRRT, continuous renal replacement therapy; CVVHD, continuous venovenous hemodialysis; CVVHF, continuous venovenous hemofiltration; INR, international normalized ratio; SAPS II, simplified acute physiology score; SOFA, sequential organ failure assessment.a The number of diagnosis adds up to a greater total than 87 because many patients were admitted to the intensive care unit (ICU) with more than one diagnosis.goodnessSecondly to examine change in clinical endpoints of infection control within each group in the first 72 h after therapy initiation: SOFA score, CRP, leukocytes, and defervescence.2.9 | Statistical analysis 2.9.1 | Statistical methods Continuous data are presented as medians with interquartile range (IQR) and categorical data as numbers and percentages.
ET AL.Two patients in the CVVHF group ended CRRT after 48 h and two were transferred to another hospital after 24 h; in one patient, samples were not obtained at 48 h.In the NRC group, four patients were transferred to another hospital after 24 h and three after 48 h; in one patient, samples were not obtained at 24 h.In the CVVHD group, samples were not obtained from one patient at 24 h and two patients at 72 h.In the ARC group, samples are missing from one patient at 72 h.
Abbreviations: BAL, bronchoalveolar lavage; ESBL, extended spectrum betalactamase; MIC, minimum inhibitory concentration.aCatheter: central venous lines, dialysis catheters, arterial lines, external cerebral ventricular drainage lines.HELSET Pharmacokinetic parameter estimates for the final covariate model.K cp , rate of transfer from the central compartment to the peripheral compartment; K pc , rate of transfer from the peripheral compartment to the central compartment; PK, pharmacokinetic; SD, standard deviation; shrink%, model shrinkage; V c , central volume of distribution of meropenem.