Population Pharmacokinetics and Monte Carlo Dosing Simulations of Meropenem during the Early Phase of Severe Sepsis and Septic Shock in Critically Ill Patients in Intensive Care Units

Abstract

Pathophysiological changes during the early phase of severe sepsis and septic shock in critically ill patients, resulting in altered pharmacokinetic (PK) patterns for antibiotics, are important factors influencing therapeutic success. The aims of this study were (i) to reveal the population PK parameters and (ii) to assess the probability of target attainment (PTA) for meropenem. The PK studies were carried out following administration of 1 g of meropenem every 8 h during the first 24 h of severe sepsis and septic shock in nine patients, and a Monte Carlo simulation was performed to determine the PTA of achieving 40% exposure time during which the free plasma drug concentration remains above the MIC (fT_>MIC) and 80% fT_>MIC. The volume of distribution (V) and total clearance (CL) of meropenem in these patients were 23.7 liters and 7.82 liters/h, respectively. For pathogens with MICs of 4 μg/ml, the PTAs of 40% fT_>MIC following administration of meropenem as a 1-h infusion of 1 g every 8 h and a 4-h infusion of 0.5 g every 8 h were 92.52% and 90.29%, respectively. For pathogens with MICs of 2 μg/ml in immunocompromised hosts, the PTAs of 80% fT_>MIC following administration of 1-h and 4-h infusions of 2 g of meropenem every 8 h were 84.32% and 94.72%, respectively. These findings indicated that the V of meropenem was greater and the CL of meropenem was lower than the values obtained in a previous study with healthy subjects. The maximum recommended dose, i.e., 2 g of meropenem every 8 h, may be required for treatment of life-threatening infections in this patient population.

INTRODUCTION

Severe sepsis and septic shock represent one of the most important reasons for admitting critically ill patients to an intensive care unit (ICU) and remain a major cause of high rates of morbidity and death (1). Early appropriate antimicrobial therapy has been shown to be the crucial factor for therapeutic success, leading to reductions in the mortality rates for these patients (2–4). During the early phase of severe sepsis and septic shock in critically ill patients, however, pathophysiological changes resulting in altered pharmacokinetic (PK) patterns, including the volume of distribution (V) and total clearance (CL), have been found with several antimicrobial agents, which may affect therapeutic plasma drug concentrations and the achieving of pharmacodynamic (PD) targets for antimicrobial therapy (5, 6).

Meropenem, a broad-spectrum carbapenem that is effective against Gram-negative bacilli, Gram-positive cocci, and anaerobic bacteria, is commonly prescribed for empirical treatment of highly resistant pathogens in patients with life-threatening severe sepsis or septic shock in ICUs (7). In common with other β-lactams, this agent exhibits primarily time-dependent antimicrobial activity, and the PK/PD index that best predicts the in vivo antimicrobial activity is the exposure time during which the plasma concentration remains above the MIC (T_>MIC) of the pathogen (8, 9). However, standard dosage recommendations for antibiotics for treatment of infections are obtained from PK data from patients with less severe sepsis (10, 11), and standardized dosage regimens for critically ill patients with severe sepsis or septic shock, particularly during the early phase, have not been determined to date. Therefore, the aims of this study were (i) to determine the population PK of meropenem and (ii) to assess the probability of target attainment (PTA) for meropenem during the early phase of severe sepsis or septic shock in critically ill patients, in order to optimize dosing recommendations.

MATERIALS AND METHODS

Subjects.

The study was conducted during the initial 24 h of severe sepsis or septic shock in nine patients admitted into the ICU of Songklanagarind Hospital, the largest tertiary care center in southern Thailand, between January and December 2013. Patients were eligible for the study if they met the following criteria: (i) >18 years of age and (ii) with a diagnosis of severe sepsis or septic shock, either at admission or during the ICU stay. Sepsis is the systemic response to an infection, defined by two or more of the following conditions: body temperature of >38°C or <36°C; heart rate of >90 beats/min; respiratory rate of >20 breaths/min or arterial partial pressure of carbon dioxide (PaCO₂) of <32 mm Hg; and leukocyte count of >12,000 cells/mm³, <4,000 cells/mm³, or 10% immature (band) forms. Severe sepsis is defined as sepsis associated with organ dysfunction, hypoperfusion, or hypotension (systolic blood pressure of <90 mm Hg, mean arterial pressure of <70 mm Hg, or a reduction of ≥40 mm Hg from baseline). Septic shock is defined as severe sepsis associated with hypotension despite adequate fluid resuscitation (12). Patients were excluded from the study if they were pregnant, had documented hypersensitivity to carbapenems, or had a history of chronic kidney disease. The severity of illness for each patient was assessed at the time of enrollment into the study, using acute physiology and chronic health evaluation II (APACHE II) and sepsis-related organ failure assessment (SOFA) scores. The protocol for the study was approved by the ethics committee of Songklanagarind Hospital. Written informed consent was obtained from each subject's legally acceptable representative before enrollment.

Drugs and chemicals.

Meropenem (Meronem) was donated by AstraZeneca (Bangkok, Thailand). Meropenem standard powder was donated by AstraZeneca (Macclesfield, United Kingdom), and cefepime standard powder (internal standard) was donated by Bristol-Myers Squibb (Sermoneta, Italy), as pure powder. All solvents were of high-performance liquid chromatography (HPLC) grade.

Study design.

Following the manufacturer's instructions, the recommended dose of meropenem was administered in a 1-h infusion of 1 g diluted in 100 ml of normal saline solution, delivered via infusion pump at a constant flow rate, every 8 h for 14 days. A Monte Carlo simulation (MCS) was performed to assess the efficacy of standard dosages of meropenem. Each patient received meropenem at room temperature (32 to 37°C).

Blood sampling.

Meropenem PK studies were carried out during administration of the first and second doses of meropenem (0 to 16 h after the start of meropenem administration), during the initial 24 h of severe sepsis and septic shock. Blood samples (∼3 ml) were obtained by direct venipuncture at the following times: shortly before (time zero) and 0.25, 0.5, 1, 1.25, 1.5, 2, 2.5, 3, 4, 5, 8, 8.5, 9, 9.5, 10, 12, 14, and 16 h after the start of meropenem administration. All blood samples were added to a heparinized tube, immediately stored on ice, and centrifuged at 2,000 × g for 10 min at 4°C within 5 min. All plasma samples were stored at −80°C until analysis within 1 week.

Meropenem assay.

Blood concentrations of meropenem were determined by reverse-phase HPLC. The samples were prepared by the modified method described by Ozkan et al. (13). Briefly, 500 μl of plasma was subjected to ultrafiltration using a Nanosep 10K device (Pall Corp., Northborough, MA); the device was centrifuged at 13,000 × g for 30 min at 4°C. A 50-μl aliquot of the sample was injected onto a μBondapak C₁₈ column (3.9 by 300 mm; Waters Associates) using an automated injection system (Waters 717 Plus autosampler; Waters Associates, Milford, MA). The mobile phase was 15 mM KH₂PO₄–acetonitrile–methanol (84:12:4 [vol/vol/vol] [pH 2.8]; flow rate, 1 ml/min). The column effluent was monitored at 308 nm with a photodiode array detector (Waters 2996; Waters Associates, Milford, MA). Peaks were recorded and integrated with a Waters 746 data module (Waters Associates). The limit of detection for meropenem was 0.05 μg/ml, and the limit of quantitation was 0.08 μg/ml. The intra-assay reproducibility values, characterized by coefficients of variation (CVs), were 2.58%, 1.77%, and 3.45% for samples containing 2, 32, and 128 μg/ml, respectively. The interassay reproducibility precision values, calculated as CVs, were 3.21%, 2.98%, and 3.74% for samples containing 2, 32, and 128 μg/ml, respectively. The accuracy values were 102.91%, 105.49%, and 108.08% and the recovery values were 117.85%, 103.37%, and 109.15% for samples containing 2, 32, and 128 μg/ml, respectively.

Pharmacokinetic analysis. (i) Model building.

The concentration-time data for meropenem in plasma were analyzed by a nonlinear mixed-effects modeling approach using NONMEM version 7.2 (Icon Development Solutions, Ellicott City, MD). The NONMEM runs were executed with PDx-Pop version 5.1 (Icon Development Solutions, Ellicott City, MD). Data were analyzed using the first-order conditional estimation with interaction (FOCE-I) method. The plasma meropenem concentrations were fitted to one-, two-, and three-compartment models using subroutines from the NONMEM library, to obtain the appropriate base model.

(ii) Covariate exploration.

After the appropriate base model was established, 15 covariates, including age, body weight, body mass index, systolic blood pressure, diastolic blood pressure, fluid intake (per day), fluid output (per day), arterial blood pH, blood urea nitrogen level, SOFA score, APACHE II score, serum creatinine level, serum albumin level, and creatinine clearance (CL_Cr) estimated with the Cockcroft-Gault equation and with the Modification of Diet in Renal Disease (MDRD) formula, were analyzed. Individual parameters were plotted against covariate values to assess relationships. If a trend between a covariate and a PK parameter was found, then it was considered for inclusion in the base model. Covariates were kept in the model if there was significant improvement in the fit over the base model. Based on the χ² test, a decrease in the minimum objective function value (MOFV) of 3.84 units was considered significant (P < 0.05) for an addition step and a more stringent criterion (P < 0.01) was used in backward deletion, to avoid any possible false-positive results.

(iii) Covariate model diagnostics.

Statistical comparisons of models were based on differences in the MOFVs. Goodness-of-fit of models were evaluated by visual inspection of diagnostic scatter plots, including observed and predicted concentrations versus time, weighted residual errors versus time, and weighted residual errors versus predicted concentrations.

(iv) Model validation.

One thousand bootstrap runs were performed to assess the robustness of all PK parameter estimates in the final model. Model stability was indicated by a condition number of less than 1,000. In addition, a visual predictive check was performed by simulating 100 subjects to assess the predictive performance of the final model. The visual checks and representative percentiles (5th, 10th, 50th [median], 90th, and 95th percentiles) were visually assessed.

Pharmacodynamic assessment using Monte Carlo simulation.

MCS of concentration-time profiles was performed using the Box-Muller transform to simulate the log-normal PK parameters of the population (14). The covariate was included in the MCS; therefore, the behavior of these simulated parameters of the population retained the characteristics of the parameters obtained from the actual patients. Additionally, for more reliable MCS, 10,000 iterations were simulated to calculate the fT_>MIC at 40% and 80% target attainment. The MCS code was written in the Basic language, using Runge-Kutta order 4 as the algorithm for solving differential equations, and was compiled by a Quick Basic compiler.

RESULTS

Nine patients were enrolled in the study (eight male patients and one female patient). Their mean age was 57.22 ± 16.10 years (range, 33 to 83 years), and their mean weight was 62.88 ± 11.64 kg (range, 49 to 80.5 kg). The characteristics of all patients are shown in Table 1.

TABLE 1.

Characteristics of 9 critically ill patients with severe sepsis and septic shock in ICU^a

Patient no.	Age (yr)	Sex	BW (kg)	MDRD CLCr (ml/min)	Comorbidity	Source of infection	Pathogen	APACHE II score	SOFA score	Concomitant medications
1	46	F	51	142.46	Meningioma, high-dose steroid therapy	VAP	GNB	25	5	Cefotaxime, dexamethasone, omeprazole, metoclopramide
2	58	M	58	31.89	DM	CAP with septic shock	GNB	21	14	Actrapid, dopamine, doxycycline, levofloxacin, morphine, omeprazole
3	59	M	49	79.78	DM	Bacteremia with septic shock	Escherichia coli and Streptococcus bovis	25	8	Ceftazidime, levofloxacin, norepinephrine, oseltamivir
4	83	M	56	59.43	PV	UTI with septic shock	E. coli, Klebsiella pneumoniae, and Pseudomonas aeruginosa	33	9	Dopamine, norepinephrine, omeprazole
5	71	M	80	48.10	COPD, MI	VAP with septic shock	K. pneumoniae	17	6	Dicloxacillin, vancomycin, isosorbide dinitrate, norepinephrine, metoprolol, nifedipine, omeprazole, prednisolone
6	65	M	60	214.55	DM, AA	Bacteremia	Salmonella spp.	18	12	Cardipine, ceftriaxone, metoprolol, nifedipine, omeprazole, simvastatin
7	33	M	70.8	52.09	Thalassemia	Severe leptospirosis with septic shock	NA	29	8	Dopamine, doxycycline, norepinephrine, morphine, omeprazole, hydrocortisone
8	63	M	80.5	12.37	NF	Bacteremia	Group A β-hemolytic streptococci	18	14	Clindamycin, omeprazole
9	37	M	60.6	65.08	ASD, PH	VAP	GNB	16	9	Cloxacillin, morphine, omeprazole, lorazepam

^aF, female; M, male; BW, body weight; MDRD CL_Cr, creatinine clearance estimated with the Modification of Diet in Renal Disease formula; DM, diabetes mellitus; PV, polycythemia vera; COPD, chronic obstructive pulmonary disease; MI, myocardial infarction; AA, aortic aneurysm; NF, necrotizing fasciitis; ASD, atrial septal defect; PH, pulmonary hypertension; VAP, ventilator-associated pneumonia; CAP, community-acquired pneumonia; UTI, urinary tract infection; GNB, Gram-negative bacilli; APACHE, acute physiology and chronic health evaluation; SOFA, sepsis-related organic failure assessment; NA, not available.

PK modeling was performed using the data from the 171 plasma concentration samples. The PK data were best described by a one-compartment model with combined additive and proportional residual variability (minimum objective function value of 717.728). The two-compartment model provided an insignificant decrease in the objective function, compared to the one-compartment model, while the three-compartment model seemed to be overparameterized for our data, as the minimization did not terminate successfully or the model was not stable. The Akaike information criterion (AIC) values for the one-, two-, and three-compartment models were 729.728, 735.508, and 744.287, respectively. The values of the parameters for the one-compartment model used in this study are given in Table 2.

TABLE 2.

Population pharmacokinetic parameters of meropenem in 9 critically ill patients

Population PK parametera	Estimate (% RSE)	Interindividual variability (%)
V (liters)	23.7 (12.6)	35
CL (liters/h)	7.82 (22.1)	64
t1/2 (h)	2.54 (68.1)

^aV, volume of distribution; CL, total clearance; t_1/2, elimination half-life; %RSE, percentage of relative standard error.

Only the MDRD CL_Cr was a significant covariate describing the CL of meropenem (MOFV decreased by 5.063), and it reduced the estimated interindividual variability of clearance from 64% to 48%. There was no significant covariate that explained the V. The final model was represented by the following:

$$\mathrm{C}\mathrm{L}=\mathrm{T}\mathrm{V}\mathrm{C}\mathrm{L}\times e^{{\eta }_1}$$

$$\mathrm{T}\mathrm{V}\mathrm{C}\mathrm{L}=({\theta }_1+{\theta }_2\times \mathrm{M}\mathrm{D}\mathrm{R}\mathrm{D}\hspace{0.17em}\mathrm{C}{\mathrm{L}}_{\mathrm{C}\mathrm{r}})$$

$$V=\mathrm{T}\mathrm{V}\mathrm{V}\times e^{{\eta }_2}$$

where TVCL and TVV are the typical values of CL and V, respectively, η₁ and η₂ are the interindividual random effects of CL and V, respectively, and θ₁ and θ₂ are shape parameters.

Goodness-of-fit plots for the final model were evaluated and showed no apparent visual bias for the predictions, as shown in Fig. 1. The 95% confidence intervals for the parameters from the final model, from 1,000 bootstrap runs, are presented in Table 3. Table 3 shows that all parameter estimates were within the ranges of the 95% confidence intervals from 1,000 bootstrap runs, indicating the robustness of the final model. The condition number for the final model was 172.6. A visual predictive check confirmed the predictive performance of the model. There was a resemblance between observed and simulated data. The observations outside the percentile range were randomly scattered and not aggregated at a particular time point. These findings imply that the final model had adequate predictive ability to describe the measured meropenem concentrations. The PTAs for all meropenem regimens achieving 40% fT_>MIC and 80% fT_>MIC at specific MICs are shown in Table 4 and Fig. 2.

FIG 1.

Basic goodness-of-fit plots. (A) Plot of observations versus population predictions. (B) Plot of observations versus individual predictions. (C) Plot of conditional weighted residuals versus time. (D) Plot of conditional weighted residuals versus individual predictions.

TABLE 3.

Parameter estimates, standard errors, and bootstrap confidence intervals

Population PK parametera	Estimate (% RSE)	Interindividual variability (%)	Bootstrap confidence interval
V (liters)	23.7 (12.6)	35	18.80–30.90
CL (liters/h)
θ1	3.01 (47.5)	48	0.07–8.11
θ2	0.07 (32.9)		0.01–0.14

^aV, volume of distribution; CL, total clearance; θ₁ and θ₂, shape parameters; % RSE, percentage of relative standard error.

TABLE 4.

Probability of target attainment for meropenem regimens achieving 40% and 80% fT_>MIC in 9 critically ill patients in ICU following administration by 1-h and 4-h infusions

MIC (μg/ml)	PTA
	40% fT>MICa						80% fT>MIC
	1-h infusion			4-h infusion			1-h infusion			4-h infusion
	0.5 g	1 g	2 g	0.5 g	1 g	2 g	0.5 g	1 g	2 g	0.5 g	1 g	2 g
0.125	99.69	99.76	99.90	100.00	100.00	100.00	93.23	95.26	96.80	98.61	99.29	99.57
0.25	99.40	99.65	99.81	100.00	100.00	100.00	89.73	92.91	95.20	97.41	98.58	99.23
0.5	98.73	99.33	99.65	100.00	100.00	100.00	84.01	89.46	93.11	94.41	97.19	98.53
1	97.28	98.77	99.37	100.00	100.00	100.00	75.22	83.89	89.73	88.55	94.49	97.31
2	92.75	97.02	98.77	99.59	100.00	100.00	60.47	74.43	84.32	74.69	88.49	94.72
4	79.38	92.52	97.04	90.29	99.59	100.00	37.44	59.51	75.19	45.36	74.63	88.64
8	43.56	79.05	92.66	43.46	90.45	99.63	11.04	36.54	60.13	9.02	45.49	75.08
16	5.51	43.18	79.65	3.72	43.28	90.24	0.72	11.31	37.27	0.11	8.79	45.95
32	0.06	5.73	43.51	0.02	3.61	43.88	0.01	0.67	11.83	0.00	0.10	8.70
64	0.00	0.07	5.98	0.00	0.01	3.54	0.00	0.00	0.66	0.00	0.00	0.09

^afT_>MIC, exposure time during which the free plasma drug concentration remains above the MIC.

FIG 2.

Probability of target attainment (PTA) for meropenem regimens achieving 40% fT_>MIC (A) or 80% fT_>MIC (B) at specific MICs during the early phase of severe sepsis and septic shock in 9 patients, after administration of a 1-h infusion of 0.5 g every 8 h (▲), a 4-h infusion of 0.5 g every 8 h (△), a 1-h infusion of 1 g every 8 h (■), a 4-h infusion of 1 g every 8 h (□), a 1-h infusion of 2 g every 8 h (●), or a 4-h infusion of 2 g every 8 h (○). Dashed lines, 90% PTA. fT_>MIC, exposure time during which the free plasma drug concentration remains above the MIC.

DISCUSSION

In critically ill patients with severe sepsis and septic shock, PK changes, including changes in V and CL, for antimicrobial agents can occur as a result of the patients' altered pathophysiological condition (15). The presence of extensive fluid extravasation and tissue edema, associated with increased capillary leakage and the use of inotropes during septic shock, can yield a larger V than the values obtained from healthy subjects. Moreover, increased cardiac output during the initial hyperdynamic state of severe sepsis, leading to increased renal blood flow and increased free fraction of antibiotics, as observed with hypoalbuminemia, can result in increased renal clearance, particularly for highly protein-bound hydrophilic antimicrobial agents; on the other hand, the renal clearance may be decreased with end-organ dysfunctions that can occur with severe sepsis and septic shock (12, 15, 16). As a consequence of these alterations in V and CL, the half-life (t_1/2) of antimicrobial agents can be affected, leading to undesirable therapeutic outcomes. In the current study, alterations of the PK of meropenem were found in comparison with an earlier study with healthy volunteers (17). The V of meropenem was greater and the CL of meropenem was lower than the values obtained from the previous study with healthy volunteers (17), resulting in >2-fold prolongation of the t_1/2 of this agent. A possible explanation for the PK changes in this study is that the study was undertaken with seriously ill patients with life-threatening infections and more than one-half of the recruited patients were experiencing renal dysfunction, as defined by MDRD CL_Cr values of ≤60 ml/min. All enrolled patients were in very seriously ill condition, with four patients having severe sepsis and five septic shock. The majority of the patients had APACHE II scores of ≥18 and SOFA scores of ≥8. However, the current study was not conducted to compare meropenem PK changes between early and late phases during antimicrobial therapy, in order to characterize the PK of meropenem when the patients had recovered from severe sepsis and septic shock. A previous study with critically ill septic patients for assessment of PK changes during meropenem therapy found that the PK of this agent had large interpatient variability and the mean value of V decreased from 18.5 liters during the early phase to 17.3 liters during the late phase of severe sepsis, and the mean value of CL increased from 7.2 liters/h during the early phase to 8.10 liters/h during the late phase of severe sepsis, but the PK changes between the two phases were not statistically significantly different (18). In addition, we compared the PK of meropenem during the early phase of therapy between the current study and a previous study (18) and found that the V and CL of meropenem from our study were still greater than the values obtained from critically ill septic patients in the other study, while the t_1/2 values were similar. Therefore, the altered PK of meropenem found in the current study might have affected the achievement of the PD targets associated with maximal antimicrobial efficacy.

Carbapenems, including meropenem, are some of the most important and commonly prescribed drugs for coverage of highly resistant nosocomial infections in critically ill patients in an ICU. The PK characteristics indicate that meropenem, a hydrophilic antimicrobial agent, is primarily distributed to extracellular compartments, such as the pulmonary epithelial lining fluid (ELF) in ventilator-associated pneumonia (VAP). The ability of this agent to penetrate the infected site to achieve the exposure targets is altered by the primary infection site, and drug concentrations in extracellular compartments are difficult to determine; thus, correlations between the PK/PD index in the tissue and antimicrobial effects are less well understood (19). Therefore, plasma drug concentrations are most commonly used as surrogate measures for the PK/PD indices, and T_>MIC is the best parameter that correlates with the bactericidal activity of β-lactams. High peak concentrations do not enhance the bactericidal activity of these agents, and bacterial growth resumes rapidly when the levels of antibiotics decrease below the MIC (8, 9). Studies in animal infection models have shown that, for most β-lactams, concentrations do not need to exceed the MIC for 100% of the dosing interval to achieve significant antibacterial effects (8, 9), and bactericidal effects of carbapenems against Escherichia coli and Pseudomonas aeruginosa in a murine thigh infection model are observed when plasma drug concentrations are above the MIC for 40% of the dosing interval (20). Moreover, optimal killing properties have been observed in critically ill patients when concentrations are maintained at 4×MIC, with higher concentrations providing little added benefit (21, 22). However, there is no consensus regarding which strategy (T_>MIC versus T_{>4× MIC}) is better.

A previous PD study of a first dose of 1 g of meropenem administered during the early phase of severe sepsis and septic shock in critically ill patients found that serum concentrations of this agent remained above 4 times the MIC of 2 μg/ml for 57% of the dosage interval, and the authors concluded that the standard dosing regimen for meropenem is sufficient to be empirically used for coverage of less susceptible pathogens in the early phase of severe sepsis and septic shock in this patient population (5). In the current study, we conducted a study to examine the population PK of meropenem during the first 24 h of severe sepsis and septic shock in patients in an ICU, and we performed a Monte Carlo dosing simulation to determine the probability of attaining specific PD targets using various regimens, including 0.5 g every 8 h, 1 g every 8 h, and 2 g every 8 h, for meropenem. A prolonged 4-h infusion regimen for achieving PTAs of 40% fT_>MIC and 80% fT_>MIC was a more effective strategy to achieve optimal PD exposure for pathogens with higher MICs than a 1-h infusion regimen. The high PTAs (≥90%) achieving 40% fT_>MIC for a MIC of 4 μg/ml were observed when meropenem was administered by either a 1-h infusion of 1 g every 8 h or a 4-h infusion of 0.5 g every 8 h. For pathogens with a MIC of 8 μg/ml, the high PTAs were achieved when the dosage of meropenem was increased to either a 1-h infusion of 2 g every 8 h or a 4-h infusion of 1 g every 8 h; moreover, a 4-h infusion of 2 g every 8 h achieved the high PTA of 40% fT_>MIC for a MIC of 16 μg/ml. These data indicate that prolonged infusion of 1 g of meropenem every 8 h can provide good coverage for pathogens with MICs of ≤8 μg/ml; for less susceptible pathogens with MICs of >8 μg/ml, the dosage regimen should be increased to a 4-h infusion of 2 g every 8 h to achieve optimal antimicrobial activity. Therefore, the standard manufacturer's dosage recommendation of 0.5 g to 1 g of meropenem every 8 h is sufficient for empirical coverage of the pathogens normally encountered in the early phase of severe sepsis and septic shock. A previous clinical study with immunocompromised patients with febrile neutropenia found that the optimal clinical response of meropenem for the treatment of bacteremia was achieved when the percentages of T_>MIC of meropenem were greater than 75% of the dosing interval (23). The current study examining the treatment of life-threatening infections during the early phase of severe sepsis and septic shock found that high PTAs for achieving 80% fT_>MIC for MICs of 1 μg/ml and 2 μg/ml were obtained when meropenem was administered through 4-h infusions of 1 g every 8 h and of 2 g every 8 h, respectively. Therefore, the results from this study indicate that, for treatment during the early phase of a life-threatening infection in an immunocompromised host, a high dosage regimen of a 4-h infusion of 2 g every 8 h may be required in order to achieve the optimal pharmacodynamic targets. In the current study, four critically ill patients had pneumonia, 3 with VAP and 1 with community-acquired pneumonia. Adequate penetration of antimicrobial agents to the site of infection is a crucial factor for achieving successful therapeutic outcomes for critically ill patients. The determination of drug concentrations in the ELF for extracellular respiratory tract pathogens provided the best estimate of antibiotic exposure in this patient population (24). A previous study of meropenem PK in plasma and ELF among seriously ill patients with VAP found the mean plasma CL of meropenem to be variable, due to differing physiological conditions across the patient population, leading to variability in the penetration of this agent into the ELF, with the 10th-to-90th-percentile range of penetration being 3.67% to 177.90%, and inadequate drug exposure at the primary infection site for some patients (25). A study in a murine pneumonia model with P. aeruginosa to determine the penetration of meropenem into the ELF of mice also demonstrated different levels of penetration (26). The exposure targets developed in the murine model to ascertain how well a dose of 2 g of meropenem administered as a 3-h infusion could achieve those targets were simulated. The target attainment of this dosage regimen in achieving a 2-log₁₀(CFU/g) cell kill was <80% at a MIC of 2 μg/ml, while values for a 3-log₁₀(CFU/g) cell kill and resistance suppression were <90% at a MIC of 0.25 μg/ml and <75% at a MIC of 1 μg/ml (25, 26). These findings indicate that PD targets may not be achieved even when the largest licensed dose of meropenem, administered as a prolonged infusion, is prescribed for critically ill patients being treated for VAP, because of the variability of penetration and the high exposure targets for meropenem in ELF in the lungs for this patient population.

This study had a few limitations that must be considered. First, the results of this study could be difficult to extrapolate to other situations, because the low body weights of the patients could have had effects on V and CL. Second, the small number of patients could be considered a potential limitation. In the absence of data from a larger sample size, however, a MCS based on a small number of patients such as in this study can be instructive in illuminating the effects of different dosing approaches.

In conclusion, our current study of population PK during the early phase of severe sepsis and septic shock in critically ill patients in an ICU found that the V of meropenem was greater and the CL of meropenem was lower than the values obtained from a previous study with healthy subjects. In addition, the V and CL of meropenem in the current study were greater than the values obtained from a previous study with critically ill patients in the early phase of sepsis. In our study, we found that 4-h infusion regimens were more effective than 1-h infusion regimens for achieving optimal PTAs of 40% fT_>MIC and 80% fT_>MIC. The maximum recommended dose, i.e., 2 g of meropenem every 8 h, administered as a 3-h infusion, may be required to maintain adequate PK for treatment of life-threatening infections in critically ill patients with severe sepsis and septic shock in an ICU. However, further large well-defined clinical trials in this patient population are required to confirm these findings.