Pharmacokinetics and Pharmacodynamics of Linezolid following Intragastric and Intravenous Administrations in ICU Patients

Background. Though intravenous infusion linezolid therapy is common for patients in the intensive care unit (ICU), intragastric linezolid therapy is also provided for those whose gastrointestinal function are feasible. If intragastric linezolid acquired similar pharmacokinetics (PK) and pharmacodynamics (PD) parameters, this might be preferred based on cost and ease of liquid volume management. Methods. Patients in the ICU treated with intragastric and intravenous linezolid were included. Serial blood samples were collected and linezolid concentrations were measured. PK data were analyzed using Pmetrics. Monte Carlo simulations were used to evaluate PD target achievement. Results. T max was 1.06 ± 0.82 h of the study period in 10 patients receiving intragastric linezolid and 0.65 ± 0.24 h in 10 patients receiving intravenous linezolid (p<0.001). C max was 9.07 ± 4.99 mg/mL of patients with intragastric linezolid and 12.30 ± 4.06 mg/mL of patients with intravenous linezolid (p=0.904). Clearance was 11.99 ± 11.24 L/h in patients with intragastric linezolid and 14.48 ± 3.56 L/h in patients with intravenous linezolid (p=0.342). For infections with a microorganism with a minimum inhibitory concentration (MIC) of 2 mg/L, simulations demonstrated that with 600 mg every 12 hours, 58.22% would have a linezolid concentration greater than the MIC during 100% of the dosing interval (%T > MIC = 100%) in intragastric group, whereas this was 71.36% in intravenous group. Higher SOFA score and body weight were associated with lower probability of target attainment (PTA) of linezolid with standard regimen. Conclusions. Patients in ICU may be at high risk for underexposure to linezolid by intragastric administration, especially when their SOFA score and body weight is high and when infected with pathogens with an MIC ≥ 2 mg/L.


Introduction
The preference for intravenous antimicrobial therapy is common in the intensive care unit (ICU), as timely and adequate antimicrobial treatment is essential for improving clinical outcome [1]. ICU patients often need multiple medications, careful uid management and more medical expenses [2]. Most infected patients in ICU should be treated with 7 days of antibiotics at least [3]. Intravenous administration of linezolid requires additional expense [4], equipment, and liquid volume as compared with intragastric therapy. If administration of intragastric linezolid could provide anti-infective effects comparable with intravenous administration, it might be alternative because of lower cost and greater ease of liquid volume management.
As the rst synthetic oxazolidinone antibiotic, linezolid is effective against gram-positive bacteria in critically ill patients [5]. The bioavailability of linezolid is about 100% in health population [6] and obese subjects before and after Roux-en-Y gastric bypass surgery [7]. For a patient who has feasible gastrointestinal function, it is expected to use intragastric linezolid therapy instead of intravenous therapy. However, whether intragastric administration of linezolid once the patient is in gastrointestinal feeding condition would exhibit e cacy similar to those with intravenous therapy is unknown.
Individualised dosing based on population pharmacokinetics (PK) / pharmacodynamics (PD) models has been shown to increase the probability of achieving therapeutic drug exposures and the likelihood of clinical success [8]. The e cacy of linezolid treatment is reached with %T > MIC of 100% and AUC/MIC values ranging from 80 to 120 [9]. Inter-and intraindividual PK variability is extreme in ICU patients [10], whether intragastric linezolid can achieve effective PK/PD target requires further exploration.
To determine whether intragastric linezolid achieves the similar anti-infective effect to intravenous linezolid in the critically ill population, we conducted a randomized controlled trial of intragastric linezolid vs intravenous linezolid in ICU patients.

Study Design and Patients
Patients with evidence of gram-positive bacteria cultured from sites of infection were enrolled in the study during March 2017 to March 2017, and written informed consent was obtained from all subjects.
Exclusion criteria included age ≤ 18 years, recent use of linezolid (within 2 weeks), gastrointestinal bleeding and undergoing hemodialysis or continuous renal replacement therapy.

Study Protocol
The patients were randomly assigned to intragastric or intravenous linezolid group. All patients received linezolid (Zyvox, P zer, New York, USA) at 600 mg, scheduled every 12 h. The intravenous group received intravenous linezolid for 60 min. The intragastric group received oral or nasogastric administration for 2 min. Two trained research nurses crushed and dissolved linezolid tablets in 10 mL sterile water for the patients who received nasal feeding. Baseline characteristics, APACHE II, serum albumin, aspartate aminotransferase (AST), and creatinine concentrations were captured within the rst 24 hours of administration. Glomerular ltration rate (GFR) was estimated using the Cockroft and Gault formula [11].

Blood Sampling and Drug Assays
A blood sample was collected at the same time as the linezolid adiministration (pre-dose sample), and 0.5, 1, 2, 4, 6, 8, and 12 h (just before the subsequent dose) after the start of infusion of the rst and the fth dose in both intragastric and intravenous groups. Blood samples were centrifuged at 3000 rpm for 10 min at 4 °C. Plasma was then collected and stored at − 80 °C until the assay.
Linezolid concentrations in plasma were measured using a sensitive and selective high-performance liquid chromatography (HPLC; column: Diamonsil ODS, 4.6 × 250 mm, 5 µm; mobile-phase: acetonitrile − 0.01% phosphoric acid water solution (27:73, v/v)) based on previously described researches [12,13]. The precision and accuracy of the method were evaluated by performing replicate analyses of quality control samples against calibration standards. Analytical methods were linear (r 2 > 0.9999) over the calibration range of 0.25-25 mg/L. Intra-and inter-day precision (relative standard deviation, %) were 2.1% and 5.0%, respectively. The accuracy was within the range of 99.45-104.42%.

Antimicrobial Therapy Responses and Safety Assessment
The antimicrobial effect of linezolid was assessed using the body temperature normalized (< 37.3 °C) within 3 days and 28-day mortality. The safety was determined by the assessment of adverse events reported by the participant or medical provider and laboratory tests monitor. The relationship between adverse reactions and linezolid was graded as de nitely related, probably related, possibly related, probably not related or not related.

Pharmacokinetic Modeling
The PK modeling for linezolid was conducted using the Nonparametric Adaptive Grid (NPAG) algorithm within the freely available software Pmetrics package for R (Los Angeles, CA, USA) [14]. Onecompartment and two-compartment models were tted to the linezolid concentration data. PK parameters were estimated based on linezolid concentrations and clearance (CL), central volume of distribution (V) and rate constant for drug distribution from the central to peripheral compartment (Kcp), rate constant for drug distribution from the peripheral to central compartment (Kpc). Patient demographics and pathophysiological factors, which included age, sex, bodyweight, height, BMI, APACHE II, SOFA, serum albumin, AST and GFR, were tested for their association with the identi ed PK parameters.
Model performance was evaluated based on visual t plots and small Akaike information criterion (AIC).

Monte Carlo Simulation
Monte Carlo simulations (n = 10000) were conducted using Pmetrics to determine the probability of target

Statistical Analysis
Page 6/23 A statistical analysis was performed using SPSS version 13.0 (SPSS Inc., Chicago, IL). All data are presented as mean ± SD. The continuous variables were analyzed using the Mann-Whitney U test, and the categorical variables were analyzed using the Fisher's exact test. A P value of < 0.05 was considered statistically signi cant.

Patient Characteristics
A total of 21 patients with gram-positive bacteria infection were enrolled. Of these, 1 patient was excluded from analysis because of gastrointestinal ulcer bleeding. Finally, 320 blood samples were included in the drug concentration analysis. The characteristics of two study groups are shown in Table 1. No signi cant baseline differences were detected in age, body weight, APACHE II, SOFA score, ALB, AST and GFR between intragastric and intravenous group. The culture results of gram-positive microorganisms are shown in Table 1, the specimens included blood, sputum, bronchoalveolar lavage uid, ascites or urine.

Outcome
The temperature change between rst and third day is shown in Fig. 1. The temperature decreased after linezolid treatment in both groups. Intragastric group had longer ICU stay (12.8 days vs. 9.00 days, P = 0.978) and fewer patients of 3-day temperature normalized (5 vs. 7, P = 0.361) compared to intravenous group as shown in Table 1. None ICU death occurred in both groups. One patient dead at 28 days postdischarge ICU and was attributed to septic shock. Multiple logistic regression analysis showed no evidence of a signi cant difference between intragastric and intravenous group when adjusted for age, BMI, infection diagnosis and bacteria: the odds ratio (OR) was 2.333 (95% CI: 0.373-14.613; P = 0.365).
No adverse effects with linezolid therapy were noted in either group.

Linezolid Concentrations and Model Development
The observed linezolid plasma concentrations for both intragastric and intravenous group are shown in Fig. 2. At the rst dose, the patients in the intragastric group had lower linezolid concentrations than those in the intravenous group. Model building was performed using observed 160 plasma linezolid concentrations in each group. A two-compartment model was best tted the data whether intragastric or intravenous linezolid. The nal models had the goodness-of-t evaluations shown in Fig. 3, as determined by an R-squared of 0.813 in intragastric group and 0.854 in intravenous group. The small AIC score of 655 in intragastric group and of 730 in intravenous group also could be estimated for nal models.
Between-patient variability could be estimated for linezolid clearance, and coe cient of variation (CV%) was 22.72% in intragastric group and 17.82% in intravenous group. Covariate analysis shown that CL of linezolid was signi cantly associated with body weight and SOFA score, explaining the between-patient variability.

Pharmacokinetics Parameters
Plasma pharmacokinetic parameters for linezolid from the nal covariate pharmacokinetic models analysis are summarized in Table 2. Following intragastric administration, a mean T max occurred at

Dosing Simulations
For microorganisms with a range MIC of 0.5 to 4 mg/mL, the PTAs of linezolid for four dosing regimens in both intragastric and intravenous groups are shown in Fig. 4. The plots show that the PTA decreased in patients receiving linezolid by intragastric administration (Fig. 4A) compared to those by intravenous administration (Fig. 4B). Neither intragastric group nor intravenous group had the PTA of > 75% for the standard linezolid regimen of 600 mg q12h at an MIC of 2 mg/L. Decreasing the linezolid dosing interval or augmenting the linezolid dose improved the PTA of the %T > MIC = 100% PD target. At a MIC of 4 mg/L, less than 70% of the patients achieved a %T > MIC = 100% with all dosing regimens in the intragastric group. However, more than 70% of patients achieved it with a dosage regimen of 600 mg q 8 h or 900 mg q 8 h in the intravenous group. The PTAs of linezolid for the standard linezolid dose with different MICs based on the SOFA score and body weight are shown in Fig. 5A and 5B. It shows that patients with a higher SOFA score or over weight had lower PTA of linezolid for the dosing regimen of 600 mg q 12 h.

Fractional Target Attainment
For 7 different gram-positive pathogens observed in 20 patients, the FTAs achieved a %T > MIC = 100% with four linezolid dosing regimens by intragastric and intravenous administration against the MIC distribution are shown in Table 3. When treating MRSA, only patients with intravenous linezolid of 600 mg or 900 mg q 8 h can achieve FTA > 85% for %T > MIC = 100%. The FTA of PD target of %T > MIC = 100% for the E. faecalis and E. faecium with dose 600 mg q 12 h were less than 85% in both intragastric and intravenous patients. In the context of empiric therapy for V. Streptococci, S. hominis, and S. pyogenes, FTA > 85% can be achieved for %T > MIC = 100% in intravenous patients with four dose regimens, but in intragastric patients with more than 600 mg q 12 h.

Discussion
This study compared the differences of the PK/PD parameters between intragastric and intravenous linezolid in ICU patients. Our results provide evidence that ICU patients with intragastric administration of linezolid are at substantial risk for underexposure to linezolid than those with intravenous administration. This can be explained by low bioavailability in patients with intragastric administration. The risk for underexposure is also associated with high SOFA score, overweight, and when pathogens with MIC ≥ 2 mg/mL. Our population PK model of linezolid was best described by a two compartment model, similar to previous studies [16,17]. The high variability of linezolid PK was observed in present study. It has been shown that the PK of linezolid can be extremely altered in critically ill patients, which is associated with body weight, organ function and renal replacement therapy [18]. We found that patinets with SOFA > 8 and body weight > 68 kg had an increase of linezolid clearance, then resulted in a lower PTA.
The results of our study do not allow us to conclude de nitively that intragastric linezolid is noninferior to intravenous in critical patients. The bioavailability of intragastric linezolid observed in this study is inconsistent with that reported in previous studies, which was 100% in health populations and unaltered in the presence of enteral feedings [19,20]. The lower bioavailability in our study is caused by the reduction of linezolid absorption through the gastrointestinal tract and the loss during crushing and transferring. Inhibition of gastrointestinal tract motility and delayed gastric emptying is a serious problem in critically ill patients [21]. We speculated that linezolid absorption was affected by gastrointestinal dysfunction, especially long T max and low C max in intragastric group. In addition, four patients' linezolid tablets were crushed and dissolved with 10 ml water and injected though nasogastric tube, this procedure may cause the loss of linezolid.
We found that volume of distribution was similar between intragastric and intravenous linezolid administration. However, Kcp was different between intragastric and intravenous therapies. It is generally believed that the rate of drug distribution depends on tissue blood ow and membrane permeability. Low serum album is observed in intravenous group in our study, which lead to accelerating the rate of drug distribution between the peripheral and central compartment, as hypoalbuminaemia is likely to increase the unbound fraction of linezolid which is the only fraction available for distribution [22].
Our results suggested that intravenous linezolid could achieve higher PTA in critically ill patients than intragastric administration. For treatment of gram-positive cocci infections, target %T > MIC = 100% or AUC/MIC ≥ 80 have been proposed for linezolid [23]. The PTA of %T > MIC = 100% target was applied because of the short sampling periods determined by the linezolid dosage intervals and the estimating of ICU stay. C min ≤ 8.2 mg/L was a signi cant predictor for minimising linezolid-induced thrombocytopenia during treatment [24]. The C min of linezolid was all below 8.2 mg/L in 20 patients of our study population.
This simulation showed that the pathogens with MIC ≥ 2 mg/mL did reduce the PTA. This range of MIC is meaningful for clinical treatment, as we found 3 of S. aureus, 1 of E. faecium and 1 of S. hominis had an MIC of 2 mg/L. Another two studies also revealed the risks of linezolid underdosing in empirical antibiotic therapy for pathogens (MIC ≥ 2 mg/L) in critically ill patients and obese patients [25,26]. Increasing the linezolid dose or reducing the dose interval was associated with adequate probability of target attainment for MIC ≥ 2 mg/L.
There are some limitations in present study. The size of our study population is small. Our results may not be directly translatable into all ICU patients due to the complexity of their critical illness conditions. In addition, nasal feeding is common in ICU patients. Tablets have to be dissolved for administration in this population, which may cause different absorption effects compared with those who take the whole tablet.
In the intragastric linezolid group, four patients were administrated linezolid through nasogastric tube after dissolving the pills because they are unable to take oral tablets.

Conclusions
Intravenous linezolid therapy is superior to intragastric linezolid in critical patients, although our endpoints were surrogate PK/PD end-points and not clinical end-points. Using 600 mg every 12 h for linezolid at the EUCAST breakpoint at an MIC ≥ 2 mg/L did not achieve optimal results, especially in the patients with high SOFA score and overweight. Higher dose linezolid therapy combined therapeutic drug monitoring should be considered in these patients.  Figure 1 Daily maximum temperature between the rst and third day of linezolid administration.

Figure 1
Daily maximum temperature between the rst and third day of linezolid administration.

Figure 2
Observed total linezolid concentration-time data after the initial dose (A) and multiple doses (B) in ICU patients with intragastric and intravenous linezolid.

Figure 2
Observed total linezolid concentration-time data after the initial dose (A) and multiple doses (B) in ICU patients with intragastric and intravenous linezolid.