Population Pharmacokinetic-Pharmacodynamic Target Attainment Analysis of Flomoxef in the Serum and Liver Tissue of Patients Undergoing Hepatic Resection

ABSTRACT

The purpose of this study was to investigate the population pharmacokinetics of prophylactic flomoxef based on serum and liver tissue concentrations and to demonstrate a pharmacodynamic target concentration in the serum and liver tissue exceeding the MIC in order to design an effective dosing regimen. Serum samples (n = 210) and liver tissue samples (n = 29) from 43 individuals were analyzed using a nonlinear mixed-effects model. The pharmacodynamics index target value was regarded as the probability of maintaining flomoxef serum trough and liver tissue concentrations exceeding the MIC₉₀ values, 0.5 mg/L and 1.0 mg/L, for Escherichia coli and methicillin-susceptible Staphylococcus aureus, respectively. The final population pharmacokinetic model was a two-compartment model with linear elimination. Creatinine clearance (CL_CR) was identified as a significant covariate influencing total clearance when CL_CR was less than 60 mL/min. The probability of achieving concentrations in the serum and liver tissue exceeding the MIC₉₀ for E. coli or methicillin-susceptible S. aureus for a 1 g bolus dose was above 90% at 2 h after the initial dose. Our findings suggest that population pharmacokinetic parameters are helpful for evaluating flomoxef pharmacokinetics and determining intraoperative flomoxef redosing intervals.

INTRODUCTION

Flomoxef (FMOX) is a cephamycin antibiotic with high antibacterial activity against Gram-positive, Gram-negative, and anaerobic bacteria (1, 2). Additionally, FMOX has demonstrable in vitro activity against extended-spectrum beta-lactamase-producing Escherichia coli and clinical therapeutic efficacy against extended-spectrum beta-lactamase-producing E. coli bacteremia, comparable to carbapenems (3, 4). It is widely used for antimicrobial prophylaxis in patients undergoing hepatic resection. To prevent surgical site infection, an appropriate dosage for antimicrobial prophylaxis is needed to achieve and maintain adequate tissue concentration of the antibiotic near the surgical site (5). Antibiotic administration should be repeated intraoperatively for prolonged surgery (if the procedure continues beyond two half-lives of the drug after the first dose) to ensure adequate antimicrobial concentration until the end of the surgery (6). In Japan, FMOX is administered 30 min before surgery, followed by 1 g every 3 h (7). However, there is insufficient evidence to confirm whether existing FMOX dosing methods maintain adequate tissue concentrations. Therefore, the objective of this study was to evaluate the population pharmacokinetics of FMOX in both serum and hepatic tissue and to calculate a pharmacodynamic target concentration exceeding the MIC in order to design an effective dosing regimen.

RESULTS

FMOX concentration was evaluated in 29 male and 14 female patients. Liver tissue was obtained from 29 patients. Only one patient was administered 2 g of FMOX. Patient characteristics are shown in Table 1. FMOX concentrations in the serum and liver tissue separately at different time points are shown in Fig. 1A and B, respectively. The values at each blood collection point for the serum samples (n = 210) and liver tissue samples (n = 29) from the 43 individuals are shown in Table 2. A two-compartment model adapts the data better than a three-compartment model based on our criteria. The final population parameters are presented in Table 3. The only covariate observed was the FMOX clearance, and the FMOX clearance was linearly correlated with the estimated creatinine clearance (CL_CR) when CL_CR was less than 60 mL/min.

TABLE 1.

Characteristics of the patients

Characteristic	No.	Median (IQR)a	Range
No. of patients (male/female)	29/14
Age (yrs)		69 (59–75)	35–87
Wt (kg)		63.3 (54.1–68.7)	42.9–86.1
Creatinine clearanceb (mL/min)		67 (54–85)	7–143
Albumin (g/dL)		4.2 (4.0–4.6)	3.5–4.9

^aIQR, interquartile range.

^bEstimated using the Cockcroft-Gault equation.

FIG 1.

Observed flomoxef concentrations. Individual flomoxef concentration in serum (A) and liver tissue (B) at various time points after dosing.

TABLE 2.

Flomoxef concentration in serum and liver tissue

Time (type of sample)	Time from flomoxef administration (h)a	Flomoxef concn (mg/L)a
Initial incision (serum)	0.57 (0.26–0.75)	74.9 (48.3–107.9)
First redosing (serum)	3.2 (3.1–3.4)	10.8 (8.4–10.6)
Second redosing (serum)	6.5 (6.4–6.6)	11.3 (10.0–16.3)
Third redosing (serum)	9.9 (9.8–10.0)	12.3 (8.7–21.0)
Resection (serum)	6.8 (4.3–8.5)	26.7 (14.0–54.6)
Resection (liver)	6.8 (4.6–8.8)	9.7 (6.6–14.7)
Skin closure (serum)	7.2 (4.9–9.5)	26.3 (14.3–41.0)

^aData are expressed as median (IQR).

TABLE 3.

Parameter estimates of the population pharmacokinetic model

Structural model parametera	Parameter	Final model		Bootstrap estimates (n = 1,000)
		Estimate	RSEb (%)	Avg	Lower 2.5%	Upper 97.5%	RSE (%)
CL (L/h) = θ1(CLCR ≥ 60)	θ1	6.25	8.9	6.02	4.93	7.11	9.2
CL (L/h) = θ1 × CLCR (CLCR < 60)	θ2	0.0732	16.5	0.0684	0.045	0.092	17.7
Vc (L) = θ3	θ3	10.2	6.2	10.3	9.06	11.5	6.1
Q (L/h) = θ4	θ4	2.28	22.4	2.58	1.58	3.58	19.8
VL (L) = θ5	θ5	52.3	25.4	62.1	36.0	88.1	21.4
Between-subject variability
Clearance	CL (%CV)	29.4	15.2	32.1	23.3	40.8	13.9
Vol of distribution of the central compartment	Vc (%CV)	33.4	13.7	33.6	23.3	43.9	15.6
Central-mesenteric adipose tissue compartment clearance	Q (%CV)	36.1	42.6	45.2	14.9	75.3	34.1
Vol of distribution in the liver tissue compartment	VL (%CV)	74.9	21.6	69.2	37.6	100.8	23.3
Residual unidentified variability	RUVPROP CMZ	25.1	6.7	26.3	22.8	29.8	6.8

^aCL, clearance; Vc, volume of distribution of the central compartment; Q, central-liver tissue compartment clearance; V_L, volume of distribution in the liver tissue compartment.

^bRelative standard error.

Assessment of the predictive performance of the final model is presented in scatterplots of the observed concentration versus population-predicted (Fig. 2A and C) and individual-predicted concentrations (Fig. 2B and D) of FMOX. Plots of the conditional weighted residual concentration versus the population-predicted concentration in the serum and liver tissue are presented in Fig. 2E and F, respectively. These plots were symmetrically distributed around the line of identity, indicating that the model adequately described FMOX concentration in the serum and liver tissue. The prediction-corrected visual predictive check (VPC) is shown in Fig. 3A and B. Prediction distribution simulated from the final model agreed with the observations.

FIG 2.

Diagnostic plots of the flomoxef final covariate model. Observed versus population-predicted concentrations in serum (A) and liver tissue (C). Observed versus individual-predicted concentrations in serum (B) and liver tissue (D). Conditional weighted residuals (CWRES) versus population-predicted concentration in serum (E) and liver tissue (F).

FIG 3.

A prediction-corrected visual predictive check based on the final population pharmacokinetics of serum (A) and liver tissue (B). The solid and dotted lines are median profiles and the 97.5th percentile and 2.5th percentile of predicted intervals (PIs). The shaded blue areas represent the 95% prediction interval of the 2.5th and 97.5th percentiles of the observed concentrations; the red-blue areas represent the 95% prediction interval of the 50th percentile of observed concentrations.

The median and percentile intervals of the predicted values were relatively close to the observed values. In the bootstrap analysis of the final model, 875 of the 1,000 bootstraps showed positive results (Table 3).

A simulation was performed using the final model to determine the optimal dosing regimen based on CL_CR. FMOX trough concentrations in both serum and liver tissue were simulated for every 1,000 patients, with 24 levels of CL_CR ranging from 5 mL/min to 120 mL/min at 5 mL/min intervals. The target was the probability of maintaining FMOX trough concentration in the serum above 0.5 or 1.0 mg/L (Fig. 4A and B). For the liver tissue, the probability of target attainment was above 90% for all criteria.

FIG 4.

Monte Carlo simulations. Probability of achieving serum trough concentration of flomoxef above the MIC₉₀ at a dose of 1 g for CL_CR ranging from 5 mL/min to 120 mL/min at 2, 3, 6, and 12 h postadministration. (A) MIC₉₀, 0.5 mg/mL for E. coli and K. pneumoniae; (B) MIC₉₀, 1.0 mg/mL for MSSA.

DISCUSSION

To the best of our knowledge, this is the first study to investigate the pharmacokinetics and pharmacodynamics of FMOX in the serum and liver tissue of patients undergoing surgical resection of hepatic lesion, and it is also the first to calculate the FMOX dosing regimen based on renal function to achieve serum and liver tissue concentrations above the MIC of 0.5 mg/L for E. coli and Klebsiella pneumoniae and 1.0 mg/mL for methicillin-susceptible Staphylococcus aureus (MSSA).

Our data revealed that the concentration of FMOX in the liver tissue was approximately 33% of the corresponding serum concentration. Consistent with this observed relationship between serum and liver tissue FMOX concentrations, Nakamura et al. reported that the ratio of prostate tissue-to-plasma area under the drug concentration-time curve of FMOX at a dose of 1 g was 0.42 (8). Igawa et al. (9) reported the ratio of maxillary bone and mandibular bone concentrations to the serum concentration of FMOX at 6 h after intravenous administration of 2 g FMOX over 3 min. Based on these results, the FMOX tissue-to-serum concentration ratio appears to be approximately 30%, although it may vary depending on the tissue type.

Population pharmacokinetics data of this study revealed CL_CR as a significant covariant for the central compartment clearance of FMOX. FMOX has been reported to undergo renal elimination (10). According to Fig. 2, the conditional weighted residuals were acceptable within three standard deviations. In addition, the parameters from the bootstrap procedure followed a normal distribution and contained all parameter estimates from the final population model. Therefore, the final model had good predictive capability.

The probability of target attainment in the serum and liver tissue for the FMOX redosing interval of 2 h was above 90% (Fig. 4A and B) regardless of the renal function results. In a phase 1 study in Japan (11), the half-life of FMOX in the serum was approximately 50 to 60 min. Considering this, additional FMOX should be considered for redosing every 2 h for patients with normal renal function. Conversely, for those with failing renal function (20 mL/min < CL_CR ≤ 40 mL/min), our data suggested an FMOX redosing interval of 6 h because the probability of target attainment in the serum and liver tissue was above 90%. Additionally, for patients with severe failing renal function (CL_CR ≤ 20 mL/min), our data suggested an FMOX redosing interval of 12 h. To the best of our knowledge, there are no reports on the relationship between the FMOX concentration and occurrence of side effects. For cefepime and ceftriaxone, there is a known correlation between concentration and neurotoxicity (12, 13). There is a possibility that unexpected side effects may occur with accumulation of FMOX concentrations. Therefore, it is important to design the redosing interval based on renal function.

The current study had some limitations. First, the liver tissue concentration was the only aspect examined. As there is a large interindividual variation in the volume of liver distribution, observed versus population-predicted concentrations in the liver tissue (Fig. 2C) were poor. Therefore, this limited information may be insufficient for accurate prediction of liver tissue concentration. Second, Fig. 3A and B show that the predictability of high serum concentrations is poor, so caution is needed when predicting serum concentrations immediately after administration. Thirdly, we did not measure the free serum concentrations and instead used the values in the reference to estimate the free serum concentrations (14), so it is unclear whether they accurately reflect the probability of achieving the target. However, as shown in Table 1, most of the patients had normal albumin levels, and the protein binding of FMOX was 35% lower; therefore, the effect of FMOX on the free form concentration would be minimal.

Notwithstanding the limitations of our study, we reported the population pharmacokinetics of FMOX in both serum and liver tissue concentrations. Furthermore, the optimal dosing regimen reported in this study may be helpful in determining intraoperative FMOX redosing intervals.

MATERIALS AND METHODS

Patients.

Patients who underwent elective surgical resection of hepatic lesions from April 2019 to December 2020 at Kitasato University Hospital, Kanagawa, Japan were enrolled in this study. All patients received 1 or 2 g of intravenous FMOX over a 5-min period. Informed consent was obtained from all patients before surgery. The study was performed in accordance with the Declaration of Helsinki and was approved by the ethical review board of our hospital (approval number B 16-277).

Sample collection.

Blood samples were collected at initial incision within 60 min after antibiotic administration, during liver resection, at redosing (every 3 h), and before skin closure. Blood samples were centrifuged immediately after collection, and the obtained serum samples were stored at −30°C until further analysis. Liver tissue samples were collected immediately after liver resection. The samples were rinsed with phosphate buffer solution and stored at −30°C until further analysis.

Measurement of FMOX concentration in the serum and liver tissue.

FMOX concentration in the serum and liver tissue was measured using high-performance liquid chromatography according to a previously described procedure (15) with minor modifications. Briefly, the liver tissue was pulverized in liquid nitrogen using Cryo-Press (Microtec Nition, Chiba, Japan) to obtain powdered samples that were subsequently homogenized in 200 μL of phosphate buffer solution. The homogenate was centrifuged, and the supernatant was collected. The tissue supernatant (100 μL) and serum sample (100 μL) were mixed separately with 90 μL of deproteinizing agent (1 M HClO₄) and centrifuged at 10,000 × g for 10 min. The sample solution (50 μL) was then injected into a C₁₈ column at 10°C. The mobile phase consisted of 80% 50 mM phosphate buffer (pH 7.0) and 20% methanol. The samples were eluted in the mobile phase at a flow rate of 1.0 mL/min and absorbance of the eluate was measured at 254 nm using a UV absorption detector. The specific gravity of the liver tissue was defined as 1 (kg = L) with reference to existing reports (8). The quantification limits for FMOX were 0.5 mg/L, and calibration curves were linear up to 250 mg/L with interday and intraday coefficients of variation of <10%.

Population pharmacokinetics analysis.

Population pharmacokinetic modeling was performed using NONMEM software (version 7.3.0; ICON Development Solution, Ellicott City, MD, USA). All serum and liver tissue concentrations were simultaneously adapted to a two-compartment or three-compartment model, which were discriminated using the Akaike information criterion (16). All serum concentrations and liver tissue concentrations were simultaneously fitted to a two-compartment model (PREDPP subroutines ADVAN3 and TRANS4). Structural model parameters, including clearance, volume of distribution in the central compartment, central-liver tissue compartment clearance, and volume of distribution in the liver tissue compartment, were assessed. The interindividual variability for all pharmacokinetic parameters was calculated and assessed using an exponential error model. Residual (intraindividual) variability was determined using a proportional error model. Body weight and CL_CR (estimated using the Cockcroft-Gault formula) were chosen as pharmacokinetic covariate candidates. Covariance, showing a correlation with pharmacokinetic parameters, was introduced into the model. The significance of the influence of covariates was evaluated by a change of −2 log-likelihood (the minimum value of the objective function value [OFV]). An OFV decrease of more than 3.84 from the basic structural model (P < 0.05) was considered statistically significant during the forward inclusion process. The full model was structured by incorporating significant covariates, and the final model was developed using a backward elimination method. When one covariate factor was excluded from the full model, an OFV that increased by more than 6.63 from the full model (P < 0.01) was considered statistically significant.

The adequacy of fit was assessed by plotting the predicted versus the observed concentrations of FMOX, individual-predicted concentration after each Bayesian step versus the observed concentration, and weighted-residual concentration versus the predicted concentration. A prediction-corrected VPC was performed to evaluate the final model and parameter estimates. For prediction-corrected VPC, 1,000 simulation replicates of the original data set were performed using the final model. Nonparametric bootstrap analysis was performed using Perl-speaks-NONMEM (PsN) software to assess the reliability and stability of the estimated parameters (17). The final model was fitted repeatedly to 1,000 additional bootstrap data sets. The average, 95% confidence interval, and relative standard errors were calculated from the empirical bootstrap distribution and compared with estimates from the original data set.

Pharmacodynamics analysis using Monte Carlo simulation.

Monte Carlo simulations were conducted to simulate FMOX trough concentrations in the serum and liver tissue at 2, 3, 6, and 12 h after FMOX bolus dose (1 g). The free FMOX serum concentration was corrected using a protein binding rate of 35% (14). Virtual patients were randomly generated by uniform random numbers based on the population pharmacokinetic model, yielding mean estimates (θ) and interindividual variances (ω). The pharmacodynamics index target value was the probability of maintaining FMOX concentration exceeding the MIC₉₀ for the target-contaminating bacteria in the field of hepatic surgery. These include MSSA, E. coli, and K. pneumoniae. The MIC₉₀ values of the clinical isolates were 0.5 mg/L for E. coli and K. pneumoniae and 1.0 mg/mL for MSSA (18).