Dosing and Pharmacokinetics of Polymyxin B in Patients with Renal Insufficiency

ABSTRACT

Polymyxin B remains the last-line treatment option for multidrug-resistant Gram-negative bacterial infections. Current U.S. Food and Drug Administration-approved prescribing information recommends that polymyxin B dosing should be adjusted according to the patient's renal function, despite studies that have shown poor correlation between creatinine and polymyxin B clearance. The objective of the present study was to determine whether steady-state polymyxin B exposures in patients with normal renal function were different from those in patients with renal insufficiency. Nineteen adult patients who received intravenous polymyxin B (1.5 to 2.5 mg/kg [actual body weight] daily) were included. To measure polymyxin B concentrations, serial blood samples were obtained from each patient after receiving polymyxin B for at least 48 h. The primary outcome was polymyxin B exposure at steady state, as reflected by the area under the concentration-time curve (AUC) over 24 h. Five patients had normal renal function (estimated creatinine clearance [CL_CR] ≥ 80 ml/min) at baseline, whereas 14 had renal insufficiency (CL_CR < 80 ml/min). The mean AUC of polymyxin B ± the standard deviation in the normal renal function cohort was 63.5 ± 16.6 mg·h/liter compared to 56.0 ± 17.5 mg·h/liter in the renal insufficiency cohort (P = 0.42). Adjusting the AUC for the daily dose (in mg/kg of actual body weight) did not result in a significant difference (28.6 ± 7.0 mg·h/liter versus 29.7 ± 11.2 mg·h/liter, P = 0.80). Polymyxin B exposures in patients with normal and impaired renal function after receiving standard dosing of polymyxin B were comparable. Polymyxin B dosing adjustment in patients with renal insufficiency should be reexamined.

INTRODUCTION

Parenteral polymyxins (polymyxin B and polymyxin E [colistin]) have become one of the most important antibiotics for therapy of extensively drug-resistant Gram-negative bacterial infections over the past decade, including infections caused by carbapenem-resistant nonfermenters and carbapenem-resistant Enterobacteriaceae. The similarities and differences between polymyxin B and colistin have been reviewed elsewhere (1). Polymyxin B has been increasingly used due to the active drug form used, more straight-forward dosing, more favorable pharmacokinetics, and potentially lower incidence of nephrotoxicity than colistin (2, 3). Current U.S. Food and Drug Administration-approved prescribing information recommends polymyxin B dosing adjustment in patients with renal insufficiency (package inserts from Bedford Laboratories, X-Gen Pharmaceuticals, Inc., APP Pharmaceuticals, LLC). Many patients who received polymyxin B treatment had impaired baseline renal function, and the dose of polymyxin B was usually reduced according to the aforementioned prescribing information (4). It was reported that receiving polymyxin B dosages of <1.3 mg/kg/day (frequently seen in the patients with impaired renal function) was significantly associated with 30-day mortality (4), and polymyxin B at a dosage of >200 mg/day was found to be associated with lower in-hospital mortality (5).

Pharmacokinetic studies of polymyxin B revealed that polymyxin B was eliminated mainly by nonrenal pathways and that polymyxin B total body clearance was poorly correlated with creatinine clearance (6, 7). Therefore, polymyxin B dose modification may not be necessary in patients with impaired renal function. We sought to determine here whether steady-state polymyxin B exposures in patients with normal renal function were different from those in patients with renal insufficiency at the initiation of polymyxin B therapy.

RESULTS

Baseline characteristics of patients.

Nineteen patients who received standard polymyxin B dosing were evaluated. The estimated creatinine clearance (CL_CR) ranged from 15 to 110 ml/min, and 5 patients had normal renal function at baseline. Polymyxin B was administered by intermittent infusions (60 to 180 min) every 12 h (in 14 patients) or every 24 h (in 5 patients). Key demographic and clinical characteristics of the patients are summarized in Table 1. The most common site of positive culture was the respiratory tract (42.1%), followed by the bloodstream (36.8%). The most common bacteria isolated were Acinetobacter baumannii (63.2%) and Pseudomonas aeruginosa (26.3%). The median duration of polymyxin B therapy was 14 days.

TABLE 1.

Characteristics of all patients^a

Variable	CLCR ≥ 80	CLCR < 80	P
No. of patients	5	14
No. (%) of patients enrolled from a U.S. study	2 (40.0)	4 (28.6)	1.00
Mean age (yr) ± SD	61.0 ± 9.4	62.8 ± 14.3	0.76
No. (%) of male patients	3 (60.0)	7 (50.0)	1.00
Mean baseline CLCR (ml/min) ± SD	90.0 ± 12.5	40.8 ± 21.8	<0.001
Mean ABW (kg) ± SD	73.6 ± 25.7	54.8 ± 10.4	0.18
Mean polymyxin B daily dose (mg/kg [ABW]) ± SD	2.2 ± 0.2	1.9 ± 0.3	0.08
Mean duration of therapy (days) ± SD	12.4 ± 2.2	12.6 ± 6.2	0.90

^aCL_CR, creatinine clearance (in ml/min); ABW, actual body weight.

Pharmacokinetics.

Overall, the pharmacokinetic profiles were reasonably characterized; the r² values between observed and best-fit concentrations for all the profiles evaluated were 0.94 (standard two-stage approach) and 0.93 (maximum-a posteriori [MAP]-Bayesian approach), respectively (Fig. 1). Despite using different methods, the parameter estimates were highly correlated (r² = 0.97 for total body clearance), and thus only the MAP-Bayesian estimates are shown. Based on the best-fit parameters, the mean ± standard deviation (SD) for the area under the concentration-time curve (AUC) observed in the normal renal function cohort was 63.5 ± 16.6 mg·h/liter compared to 56.0 ± 17.5 mg·h/liter in the renal insufficiency cohort (P = 0.42) (Fig. 2). Adjusting the AUC for the daily dose (in mg/kg of actual body weight) did not result in a significant difference (28.6 ± 7.0 mg·h/liter versus 29.7 ± 11.2 mg·h/liter, P = 0.80). Similarly, the clearances observed were 2.5 ± 0.4 and 2.0 ± 0.6 liters/h, respectively (P = 0.06). Moreover, sensitivity analysis revealed that our primary finding would not change if the definition of normal renal function was set at any thresholds between creatinine clearances of 40 and 80 ml/min (Table 2).

FIG 1.

Correlation between observed and best-fit concentrations. (A) Standard two-stage approach. (B) MAP-Bayesian approach. The solid line depicts the line of best fit; the dashed line is the line of identity where y = x.

FIG 2.

Comparison of overall drug exposures stratified by renal function. Data means ± the standard deviations. Note that the normalized AUC was adjusted to 1 mg/kg of polymyxin B daily.

TABLE 2.

AUC comparisons based on various CL_CR thresholds^a

Threshold for normal renal function	No. of subjects with normal renal function	Mean CLCR (ml/min) ± SD in patients with:		P
		Normal renal function	Renal insufficiency
CLCR ≥ 80	5	63.5 ± 16.6	56.0 ± 17.5	0.42
CLCR ≥ 60	8	58.7 ± 17.7	57.4 ± 17.6	0.88
CLCR ≥ 40	10	55.6 ± 17.6	60.6 ± 17.3	0.54

^aCL_CR, creatinine clearance.

DISCUSSION

In view of the escalating trends of multidrug resistance among Gram-negative bacteria, the increased use of polymyxin B is also anticipated to meet this unmet medical need. In vitro data suggested that an AUC/MIC ratio of polymyxin B was the pharmacokinetic/pharmacodynamic index most closely linked to bactericidal activity (8). In order to prevent underdosing, dosing strategies in patients with renal insufficiency should focus on normalizing/achieving an AUC value similar to those seen in patients with normal renal function.

Although it is commonly recommended in package inserts that polymyxin B dosing should be adjusted in renal insufficiency, data from several recent studies have provided contradicting insights refuting this practice. In experimental animals with transient renal impairment, similar drug exposures were observed after receiving the same dose (in mg/kg) despite drastic differences in renal function (9, 10). These data were consistent with human data, suggesting minimal renal clearance of polymyxin B (7, 11). Furthermore, there are also anecdotal clinical reports showing reduced drug exposure with adjusted doses in renal insufficiency (12). Collectively speaking, these studies provided reasonably convincing circumstantial evidence that dosing adjustment might not be appropriate in renal insufficiency. To the best of our knowledge, this is the first study to date directly comparing drug exposures achieved in patients with different renal functions given standard dosing of polymyxin B.

In patients given standard polymyxin B dosing, drug exposures were found to be comparable in patients with normal or impaired renal function. We recognize that the definition of renal insufficiency could be somewhat subjective and may vary among different studies. Consequently, we performed a secondary sensitivity analysis to demonstrate the consistency of our primary finding using different thresholds of creatinine clearance. Although the polymyxin B daily dose was within the normal range for all patients, we noted there was a trend that patients with renal insufficiency were given a lower dose than patients with normal renal function (1.9 mg/kg versus 2.2 mg/kg). However, the magnitude of difference (approximately 11% in the mean AUC) would still have been within intersubject variability expected in critically ill patients (35 to 50%), and thus a dosing adjustment would not have been indicated (i.e., typically >2-fold change in AUC is needed for dosing adjustment). Adjusting the AUC for the daily dose (in mg/kg of actual body weight) also did not result in a significant difference in patients with renal insufficiency (Fig. 2). We hope that insights from our study would provide the impetus in the next step of advancing the level of patient care.

We have previously compared the pharmacokinetics of different polymyxin B components. We found that the overall drug exposure derived by adding the AUC of individual component was reasonably close to that estimated directly by total polymyxin B concentrations (13). The binding of polymyxin B to human serum has also been examined previously (14). Instead of using a population pharmacokinetic analysis with a sparse sampling strategy, multiple samples were obtained from each patient in this study. This approach would allow us to determine the individual drug exposure in each patient directly and facilitate comparing drug exposures between two different cohorts (subpopulations) in a straightforward manner. Studies linking various demographic variables to the overall drug exposure observed (as model covariates) in a single population are anticipated in the future.

There are several limitations in the study. First, a relatively small number of patients were examined; a greater sample size would enhance the robustness of our findings. Second, the renal functions of the patients were estimated and not measured using urine collected over a time period. Third, the clinical outcomes of the patients with normal renal function and those with impaired renal function were not compared; the sample sizes in both patient groups would be too small to reach any definite conclusions. Finally, patients examined in this study had stable renal function at baseline. If a patient has deteriorating renal function while on polymyxin B therapy, the necessity to adjust dose should be further explored in a future study.

In conclusion, we found comparable drug exposures in patients with different renal function given standard dosing of polymyxin B. If our preliminary findings are verified in a large prospective study, the dosing guidelines of polymyxin B should be revised for optimal patient care.

MATERIALS AND METHODS

Clinical study design and sites.

This prospective, multicenter, open-labeled, observational study was conducted from October 2014 to June 2016 in three clinical sites: Siriraj Hospital (a 2,300-bed academic tertiary care hospital in Bangkok, Thailand), New York University Langone Medical Center (an 800-bed academic medical center in New York, NY), and Baylor St. Luke's Medical Center (an 850-bed teaching hospital in Houston, TX). Institutional Review Board approval at each study site and at the University of Houston was obtained prior to the initiation of this study. Written informed consent was obtained from each patient prior to study enrollment.

Patient selection criteria and study variable definitions.

Adult patients (18 years or older) who were given at least 48 h of intravenous polymyxin B daily for suspected or documented infections were included in this study. Patients on any form of renal replacement therapy or with fluctuating renal function (increase or decrease in serum creatinine of more than 50% from the first day of therapy) were excluded. Patients deemed to be in a hypermetabolic state with augmented drug clearance (estimated creatinine clearance > 140 ml/min) were also excluded. Data collected included demographics (e.g., age, ethnicity, and gender) and pertinent laboratory findings (e.g., serum creatinine, site of infection, and isolated microorganisms). Polymyxin B regimens (daily dose, dosing interval, duration of intravenous administration, and duration of therapy) were at the discretion of the responsible attending medical teams. The primary outcome of the study was daily polymyxin B exposure at steady state, as expressed by the AUC over 24 h (AUC_0–24). For the purpose of this study, normal renal function was defined as an estimated creatinine clearance of ≥80 ml/min (based on the serum creatinine level according to the Cockcroft-Gault equation) on the first day of polymyxin B therapy. Standard polymyxin B dosing was defined as 1.5 to 2.5 mg/kg of actual body weight daily.

Sampling schedule and preparation.

After at least 48 h of intravenous polymyxin B therapy, four serial blood samples were obtained over a dosing interval from each patient (e.g., immediately prior to a scheduled dose, 1 to 2 h after the end of drug infusion). Each sample was specifically timed in relation to the dose given. Plasma samples were obtained by centrifugation within 30 min of blood collection and were stored at −80°C until analysis. After thawing at room temperature, plasma samples were spiked with 20 μl of the internal standard, and 5% trichloroacetic acid (180 μl) was added for protein precipitation. After centrifugation, the supernatants were evaporated to dryness and further processed as described previously (15).

Drug assay.

Major polymyxin B components were quantified using a validated ultraperformance liquid chromatography tandem mass-spectrometry method with the following modifications: Acquity UPLC HSS C₁₈ column (50 mm by 2.1 mm internal diameter; 1.7 μm) from Waters (Milford, MA); mobile phase A, 0.1% formic acid in water; and mobile phase B, 0.1% formic acid in acetonitrile (15). Standard stock solutions of polymyxin B1, B2, B3, and isoleucine-B1 were prepared by dissolving a known amount of the reference standards in liquid chromatography-mass spectrometry-grade water. The working solutions were obtained by serial dilution of the prepared mixture stock solution in 0.1% formic acid to achieve a final concentration range of 0.0625 to 32 μl/ml. To establish a calibration curve, blank serum (100 μl) was spiked with 40 μl of the above working solutions of the standard mixture and 20 μl of the internal standard (carbutamide) solution. To precipitate proteins, 5% trichloroacetic acid (140 μl) was added to the mixture to yield a final concentration range of 0.025 to 12.8 μl/ml.

Pharmacokinetic modeling and statistical analysis.

The pharmacokinetics of polymyxin B was derived using two different methods. In both methods, serum concentrations for each polymyxin component (polymyxin B1, B2, B3 and isoleucine B1) were quantified individually and collectively reported as the total polymyxin B concentration. With the standard two-stage approach, a one-compartment linear model with a zero order input was used to fit the total polymyxin B concentration-time profiles. Alternatively, a MAP-Bayesian approach was used to fit a two-compartment model to the data. The point estimates and the covariance matrix of unscaled population estimates previously published were used as the Bayes priors (6). Log-normal parameter distributions were assumed. All modeling were performed using ADAPT 5 (University of Southern California, Los Angeles, CA). Using the best-fit model parameter estimates of each patient, the overall drug exposure was calculated using daily dose/clearance.

Descriptive statistics were used to characterize the study cohorts. Patients with normal renal function and given standard daily dosing were considered the reference group. The drug exposures observed were compared to those in patients with reduced renal function and given standard daily dosing. Continuous variables were compared using the Student t test. Categorical variables were compared using a Fisher exact test. P values of ≤0.05 were considered significant. All statistical analyses were performed by using SYSTAT version 12 (SYSTAT Software, Inc., Chicago, IL).