Pharmacokinetics of Intravenous and Oral Linezolid in Adults with Cystic Fibrosis

Abstract

Linezolid is a treatment option for methicillin-resistant Staphylococcus aureus (MRSA) infections in cystic fibrosis (CF) patients. Little is known, however, about its pharmacokinetics in this population. Eight adults with CF were randomized to receive intravenous (i.v.) and oral linezolid at 600 mg twice daily for 9 doses in a crossover design with a 9-day washout. Plasma samples were collected after the first and ninth doses of each phase. Population pharmacokinetic analyses were performed by nonlinear mixed-effects modeling using a previously described 2-compartment model with time-dependent clearance inhibition. Monte Carlo simulation was performed to assess the activities of the linezolid dosing regimens against 42 contemporary MRSA isolates recovered from CF patients. The following pharmacokinetic parameter estimates were observed for the population: absorption rate constant, 1.91 h⁻¹; clearance, 9.54 liters/h; volume of central compartment, 26.8 liters; volume of peripheral compartment, 17.3 liters; and intercompartmental clearance, 104 liters/h. Linezolid demonstrated nonlinear clearance after 9 doses, which was reduced by a mean of 38.9% (range, 28.8 to 59.9%). Mean bioavailability was 85% (range, 47 to 131%). At steady state, 600 mg given twice daily produced 93.0% and 87.2% probabilities of obtaining the target pharmacodynamic exposure against the MRSA isolates for the i.v. and oral formulations, respectively. Thrice-daily dosing increased the probabilities to 97.0% and 95.6%, respectively. Linezolid pharmacokinetics in these adults with CF were well described by a 2-compartment model with time-dependent clearance inhibition. Standard i.v. and oral dosing regimens should be sufficient to reliably attain pharmacodynamic targets against most MRSA isolates; however, more frequent dosing may be required for isolates with MICs of ≥2 μg/ml.

INTRODUCTION

The incidence of methicillin-resistant Staphylococcus aureus (MRSA) in cystic fibrosis (CF) patients has been steadily increasing over the past 15 years. According to the Cystic Fibrosis Foundation patient registry annual report, 0.1% of positive respiratory specimen cultures were due to MRSA in 1995, and as of the last published report in 2009, this had increased to 23.7% (8). Until recently, it was unknown what influence this increase in incidence had on clinical outcomes. Pediatric patients with respiratory specimen cultures persistently positive for MRSA have significant reductions in lung function (10). Furthermore, the chronic presence of MRSA-positive cultures for CF patients of any age is associated with decreased survival (9). Due to these negative clinical outcomes, clinicians have become more conscious of the presence of respiratory MRSA and the need for treatment in these patients.

Although linezolid, an oxazolidinone antibiotic with in vitro activity against MRSA, is a treatment option for CF patients with acute exacerbations, little is understood about its pharmacokinetics in this population. Previous studies have proposed that CF patients with inadequate clinical responses may require more frequent dosing due to variability in their pharmacokinetic parameters (3, 22). Neither of these studies, however, considered the potential for nonlinear elimination of linezolid, which has been demonstrated in other populations after multiple doses (2, 19, 20). Additionally, no study has assessed the bioavailability of linezolid in CF patients, a population that often exhibits malabsorption. The present study was undertaken to describe the pharmacokinetics of intravenous (i.v.) and oral linezolid in adults with CF and determine the pharmacodynamic profile of multiple linezolid dosing regimens against contemporary MRSA isolates.

MATERIALS AND METHODS

Study design.

This study was a prospective, single-center, crossover pharmacokinetic analysis of i.v. and oral linezolid in outpatient adults with CF, conducted at the Clinical Research Center at Hartford Hospital (Hartford, CT). The protocol was approved by the Hartford Hospital Institutional Review Board (KUTI002983HE), and all participants provided written informed consent prior to study screening.

Participants.

Inclusion criteria required that each participant be greater than or equal to 18 years of age with known CF on the basis of a prior elevated sweat chloride level or an abnormal CF transmembrane regulator genotype. A prestudy screening consisting of a detailed medical history, physical examination, and diagnostic testing (including vital signs, blood/urine laboratory testing, and sputum culture) was performed within 28 days prior to study drug administration. Participants were excluded if they had clinically significant increases or decreases in baseline lab results, if they had a known linezolid-resistant MRSA infection or a known Burkholderia cepacia complex infection, or the presence of an ongoing acute exacerbation of their lung infection. The participants were to abstain from alcohol-, nicotine-, and caffeine-containing products during both study phases.

Linezolid dosing and blood sampling.

Commercially available prepackaged preparations of 600 mg i.v. (lot 09D07Z10; expiration date, April 2012) and oral (lot C091093; expiration date, December 2011) linezolid (Zyvox; Pfizer Inc., New York, NY) were supplied by Pfizer Inc. and stored according to manufacturer recommendations. Participants were randomized to initially receive either i.v. or oral linezolid at 600 mg twice daily for 9 doses. The i.v. formulation of linezolid was completely infused over 30 min through a peripheral i.v. catheter, and the oral formulation was swallowed with a glass of water. Food was prohibited for 1 h prior to and 2 h after each dose. Blood samples were collected from a peripheral catheter contralateral to the one used for dosing at time zero (prior to dosing), 0.5, 0.75, 1, 1.25, 1.5, 2, 4, 6, 8, and 12 h after the first and ninth doses. After at least a 9-day washout period, participants returned to the study center and were crossed over to receive the alternate formulation, and the same dosing and blood sampling schemes were repeated. Blood samples were centrifuged (1,000 × g at 10°C for 10 min) immediately after collection. Separated plasma was stored at −80°C and protected from light until further analysis.

Linezolid concentration determination.

Linezolid concentrations in plasma were assessed using a validated high-performance liquid chromatography (HPLC) method modified from a previously published assay (5). The standard curve was extended from 20 μg/ml to 30 μg/ml, and the extraction was modified from deproteinization with 200 ml of acetonitrile to deproteinization with 150 ml of 7.5% trichloroacetic acid with no drying step. The lower limit of detection of the assay was 0.2 μg/ml. The mean interday coefficients of variation for high-concentration (20 μg/ml) and low-concentration (0.5 μg/ml) check samples were 4.3% and 2.0%, respectively. The mean intraday coefficients of variation were 3.2% and 2.4%, respectively.

Pharmacokinetic analyses.

Population modeling of linezolid plasma concentrations was performed by nonlinear mixed-effects modeling (NONMEM, version VI; ICON Development Solutions, Manchester, United Kingdom) using a 2-compartment model with time-dependent inhibition (Fig. 1), which has been well described previously (20). First-order conditional estimation with interaction was used as the estimation method. The following parameters were estimated: absorption rate constant (k_a), volume of distribution of the central compartment (V₂), intercompartmental clearance (Q), volume of distribution of the peripheral compartment (V₃), rate constant for transfer into the inhibition compartment (k_ic), total body clearance, and the maximum fraction of clearance that cannot be inhibited after infinite doses (RCLF). Model fit was assessed by changes in the objective function value (OFV); goodness-of-fit plots; visual predictive checks; and plausibility and precision of estimated parameters. Absolute bioavailability was calculated for each participant by comparing model-derived profiles between i.v. and oral regimens. Covariate relations of height, weight, and lean body weight on clearance, V_2, V₃, k_a, bioavailability, Q, and RCLF were implemented as a proportional change of the population value, centered on the median of the covariate. The covariate analysis was performed using the stepwise forward inclusion (statistical significance level of delta OFV, P < 0.05) and backward elimination procedure (P < 0.001).

Fig. 1.

Population pharmacokinetic model for linezolid plasma concentrations in cystic fibrosis patients. Clearance is inhibited on the basis of the concentration in a theoretical inhibition compartment. KA, absorption rate constant; ALAG, absorption lag time; CL, clearance; KIC, rate constant into inhibition compartment (inhib. comp.); IC₅₀, inhibition compartment concentration yielding 50% of maximum clearance inhibition; periph., peripheral.

Susceptibility testing.

Forty-two MRSA isolates collected from adolescent patients in 6 states between October 2008 and December 2009 during a separate CF surveillance study were supplied for linezolid MIC determination. MICs were determined in triplicate by Etest (lot BJ0773; expiration date, April 2014; AB bioMérieux, Solna, Sweden) and read at 90% inhibition according to manufacturer instructions.

Monte Carlo simulation.

A 1,000-patient Monte Carlo simulation was conducted to determine the probability of achieving requisite in vivo exposures for a simulated CF patient receiving 600 mg twice daily or thrice daily (for both i.v. and oral regimens) using the pharmacokinetic parameter estimates and variability derived from the model. The median lean body weight for the population was fixed in the covariate model to predict clearance and k_a. The total drug area under the concentration-time curve over 24 h (AUC) was calculated by NONMEM as the integral of the simulated concentration-time profile. The probability of target attainment (PTA), which is the likelihood that the regimen will meet or exceed the predefined pharmacodynamic target at a specific MIC dilution, was calculated. The pharmacodynamic exposure target was a ratio of the AUC to the MIC (AUC/MIC) of ≥83 (1, 18, 21). PTAs for each regimen were used to calculate the cumulative fraction of response (CFR) against the MRSA population. CFR is the probability that the dosing regimen will attain its pharmacodynamic index against the entire population of organisms.

RESULTS

Eight participants with mild to moderate lung disease were enrolled in the study; however, 1 male was withdrawn after completing only the i.v. formulation phase due to the development of a maculopapular rash. His i.v. plasma concentrations were included in the pharmacokinetic model. Participant characteristics are provided in Table 1. Overall, linezolid was well tolerated, with only mild and transient adverse events. Diarrhea (n = 3), headache (n = 2), and nausea (n = 2) were the most commonly reported adverse events, and no thrombocytopenia was observed in any of the participants.

Table 1.

Characteristics of the 8 cystic fibrosis participants at baseline^a

Subject no.	Sex	CF genotype	Age (yr)	ABW (kg)	LBWb (kg)	BMI (kg/m2)	Ht (m)	FVC (%c)	FEV1 (%c)
1	Male		35	63.4	54.0	19.4	1.81	81	66
2	Male	ΔF508/ΔF508	24	93.6	69.5	27.9	1.83	74	60
3	Male	ΔF508/—	47	69.5	54.0	25.2	1.66	36	32
4	Male	ΔF508/ΔF508	25	57.7	50.3	17.8	1.80	70	46
5	Male	ΔF508/G551D	29	79.1	59.3	27.3	1.70
6	Male	G524Z/R1303K	24	52.3	44.4	19.7	1.63	41	23
7	Female		24	55.5	39.7	24.0	1.52
8	Male	ΔF508/ΔF508	23	65.5	54.9	20.4	1.79	70	44
Mean			28	67.1	53.3	22.7	1.72	62.0	45.2
SD			8	13.7	9.1	3.9	0.11	18.7	16.3

^aABW, actual body weight; LBW, lean body weight; BMI, body mass index; FVC, forced vital capacity; FEV1, forced expiratory volume over 1 s.

^bCalculated by the James equation (16a).

^cPercent predicted.

The mean i.v. and oral linezolid plasma concentrations over 12 h after the first and last (i.e., steady-state) doses are presented in Fig. 2. Typical values for the population pharmacokinetic parameter estimates are listed in Table 2. The final model's goodness-of-fit plot for observed versus predicted concentration and weighted residuals versus time are presented in Fig. 3A and B. After multiple doses, linezolid displayed nonlinear clearance, where the mean individual initial clearance was 9.7 liters/h (range, 3.6 to 14.2 liters/h) and the mean clearance after the ninth dose was 6.1 liters/h (range, 1.8 to 9.8 liters/h). This was a reduction in clearance by 38.9% (range, 28.8% to 59.9%), with a maximum potential model estimated reduction of 67.9%. Increases and decreases in median lean body weight were associated with proportional changes in both k_a (4.74% per kilogram) and clearance (3.35% per kilogram). Mean bioavailability was 85.1% and ranged from 47% to 137%.

Fig. 2.

Mean and standard deviation intravenous (A) and oral (B) linezolid plasma concentrations over 12 h for cystic fibrosis participants after the first dose and at steady state.

Table 2.

Pharmacokinetic parameter estimates for the final model describing linezolid plasma concentrations in cystic fibrosis patients^a

Model parameter	Units	Final model
		Estimate	% RSE
Clearance	liters/h	9.54	15.4
V2	liters	26.8	8.3
V3	liters	17.3	42.3
Q	liters/h	104	18.4
ka	liters/h	1.91	12.9
Bioavailability	%	85.1	10.0
Kic	liter/h	0.0005
RCLF	%	32.1	36.1
IC50	mg/liter	0.38	32.3
Covariate influence
LBW on ka	%/kg	4.74	5.76
LBW on clearance	%/kg	3.35	35.5
Between-patient variability
Clearance	% CV	36.3	46.7
V3	% CV	85.8	70.4
RCLF	% CV	58.3	100
Bioavailability	% CV	23.0	47.6
Residual variability, proportional error	% CV	22.4	18.6

^aRSE, relative standard error; IC₅₀, inhibition compartment concentration yielding 50% of maximum clearance inhibition; LBW, lean body weight; CV, coefficient of variation.

Fig. 3.

Goodness-of-fit plot of the final model. (A) Dashed line, observed versus model-predicted linezolid concentrations (r² = 0.66, y = 1.08x − 0.665); solid line, line of unity; (B) weighted residuals.

The linezolid MIC range for the 42 MRSA isolates was 0.5 to >256 μg/ml, with only a single isolate (2.5%) found to be resistant to linezolid. The MIC₅₀ and MIC₉₀ were 1.0 and 1.5 μg/ml, respectively. The probabilities of achieving target AUC/MIC exposures of ≥83 at steady state for simulated linezolid regimens are shown in Fig. 4. (PTA curves for the first 24 h of dosing are not shown.) Due to higher clearances after initial doses, the CFRs for the first day of i.v. or oral dosing were 73.5% and 59.6%, respectively, for the twice-daily regimens. Increasing the dose to three times a day provided CFRs of 90.9% and 82.2% for the i.v. and oral formulations, respectively, on the first day of therapy. After multiple doses, the CFRs for 600 mg i.v. and orally twice daily increased to 93.0% and 87.2%, respectively, while the CFRs for 600 mg i.v. and orally three times daily increased to 97.0% and 95.6%, respectively.

Fig. 4.

Probability of target attainment to achieve an AUC/MIC of ≥83 for two dosing regimens of intravenous and oral linezolid at steady state. BID, twice a day; TID, three times a day.

DISCUSSION

As the incidence of MRSA exacerbations in CF patients increases, the need for alternative antibiotic options to that of i.v. vancomycin has become apparent. Case reports have documented successful treatment of MRSA infections with linezolid in patients with CF (12, 25, 26). It has also been evident that prolonged treatment with low doses of linezolid results in the development of resistance (13, 15). Thus, dosing regimen selection is critical in achieving a successful response while minimizing the development of resistance and adverse events. Herein, we describe the population pharmacokinetics of linezolid in eight adult CF participants using a previously described 2-compartment model with a theoretical inhibition compartment (20). This model accurately described linezolid pharmacokinetics in these participants and generally resulted in mean parameter estimates similar to those for healthy volunteers and septic shock patients, albeit with lower absolute bioavailability (20). Importantly, nonlinear clearance was evident in all participants, thereby increasing linezolid exposure after multiple dosing and suggesting that standard linezolid dosing regimens (i.e., 600 mg i.v. or oral twice daily) should be sufficient to achieve critical pharmacodynamic targets for most MRSA isolates.

Few studies have evaluated linezolid pharmacokinetics in CF patients (3, 22, 23), and all of these studies determined the pharmacokinetics of linezolid by noncompartmental methods, making direct comparisons with our pharmacokinetic parameters inappropriate. The earliest study described linezolid disposition after a single i.v. dose of 600 mg in 12 adult CF patients admitted to the hospital for an acute exacerbation (3). By comparing dose-normalized AUCs, similar results between their study and ours (0.187 μg·h/ml mg and versus 0.212 μg·h/ml or mg, respectively) were observed from a single 600-mg i.v. dose. A second study in 10 CF adults receiving multiple oral doses of linezolid observed lower concentrations at 2 and 4 h after dosing compared with those in healthy volunteers; however, this study had a sparse sampling strategy, did not collect serum after the first dose, and did not report pharmacokinetic parameter estimates (23). Their mean 2-hour serum concentration after multiple dosing was in concordance with our 2-hour steady-state oral plasma concentration (13.5 μg/ml versus 12.7 μg/ml, respectively); however, their reported mean 4- and 12-hour concentrations (8.1 μg/ml and 2.3 μg/ml, respectively) were substantially lower than our concentrations (11.9 μg/ml versus 5.5 μg/ml, respectively) for these time points. This may be partially explained by differences in the number of doses prior to sampling; although steady state would typically be reached by either 6 or 9 doses, clearance inhibition is time dependent; thus, our participants may have had higher concentrations due to more apparent clearance inhibition. The final study (22) also had a sparse sampling strategy (peak and trough only), used noncompartmental methods, and enrolled 10 children (age range, 4 to 20 years), with only a single patient being over the age of 16 years, so it is difficult to make comparisons with our results.

It is well recognized that CF patients can portray pharmacokinetic profiles for antibiotics different from those of non-CF patients (17). Most notably, CF patients tend to have a more rapid clearance and a greater volume of distribution for antibiotics eliminated via glomerular filtration, tubular secretion, as well as nonrenal metabolism pathways (6, 7, 16, 27). In non-CF populations, linezolid concentration data are best described by nonlinear elimination models (2, 19, 28). The previously well-described population pharmacokinetic model applied in this study best predicted parameter estimates and minimized model misspecifications (20). We also undertook model comparisons with other nonlinear approaches and found this model to be the most accurate for describing linezolid concentrations in these CF patients (24). Our final model estimates for clearance and volume of distribution were very similar to those reported for septic shock patients and healthy volunteers (20). Notable differences included a more rapid Q in the CF patients (104 liters/h versus 75 liters/h) and a lower RCLF (0.321 versus 0.764). With respect to the maximal fraction of linezolid clearance that could not be inhibited, lower estimates suggest that linezolid clearance may be reduced to a greater degree in CF patients than other patient populations.

Additionally, oral bioavailability in our CF population was found to be approximately 85.1% and was quite variable (range, 47% to 137%). This observation is substantially different from those in studies in non-CF subjects, which have reported linezolid bioavailability of nearly 100% (2, 29). Linezolid is a moderately lipophilic molecule (4), and pancreatic enzyme deficiency, which is common in CF patients, can impact the absorption of lipophilic medications (14). It was also not surprising, then, to observe a positive relationship between lean body weight and k_a (Table 2), suggesting a link between nutritional status and the rate of antibiotic absorption (i.e., those who are malnourished might need higher oral dosages to obtain similar concentrations). All eight patients received exogenous oral supplementation of pancreatic enzymes, but at various doses.

Perceived pharmacokinetic differences in CF patients, combined with a propensity for infection caused by multidrug-resistant bacteria, often result in the need to administer higher antibiotic doses to the CF population. Indeed, the previous pharmacokinetic assessment in adults attempted to elucidate achievable exposure by dividing the mean AUC by the susceptibility breakpoint (i.e., 4 μg/ml), resulting in exceedingly low AUC/MIC values and a conclusion that linezolid dosing should be increased in this population (3). However, the MIC₅₀ and MIC₉₀ observed in our collection of 42 CF isolates were 1.0 and 1.5 μg/ml, respectively, which are similar to MIC values from the most recent linezolid national surveillance program (LEADER program, 2008) (11). Using the pharmacokinetic model derived from these 8 participants, we aimed at simulating AUC/MIC exposures for 1,000 CF patients after the first 24 h and at steady state. Although exposures over the first 24 h were low for twice-daily dosing, the CFRs at steady state were 93.0% and 87.2% for i.v. and oral doses, respectively. The likelihood of achieving an AUC/MIC ratio of ≥83 declines rapidly once MICs are 2 μg/ml or greater (Fig. 4). Empirically increasing linezolid dosing may put a patient at unnecessary risk for drug-induced toxicity; therefore, testing the linezolid MIC prior to selection of a dosing regimen would be of value in this population. Lastly, this pharmacodynamic target comes from animal infection models of MRSA and non-CF patient studies (1, 18, 21). The pharmacodynamic target required for a successful response in CF patients is unknown.

Linezolid pharmacokinetics in these 8 CF adults were well described by a 2-compartment model with time-dependent clearance inhibition. Although interpatient variability was apparent and absolute bioavailability was 85%, standard i.v. and oral dosing regimens should be sufficient to reliably attain pharmacodynamic targets against most MRSA isolates. Higher or more frequent dosing, however, may be required for isolates with MICs of ≥2 μg/ml. The pharmacokinetics and safety of higher or more frequent linezolid dosing require further study in adult patients with CF.