Risk Factors for a Low Linezolid Trough Plasma Concentration in Acute Infections

Abstract

Linezolid is an antibiotic with time-dependent activity, and both the percentage of time that plasma concentrations exceed the MIC and the area under the concentration-time curve over 24 h in the steady state divided by the MIC (AUC₂₄/MIC ratio) are associated with clinical response. The aim of this study was to analyze the linezolid trough plasma concentration (C_min) and to determine factors associated with a C_min < 2 mg/liter and other clinically relevant thresholds. Characteristics of 78 patients receiving 600 mg/12 h of linezolid with a C_min determination at the steady state and within the first 10 days of treatment were retrospectively reviewed. Concentrations were measured using high-pressure liquid chromatography. Univariate and multivariate analysis were performed to identify risk factors of low C_min. A total of 29.5% of patients had a C_min < 2 mg/liter. The percentage was significantly higher in patients with an estimated glomerular filtration (eGF) > 80 ml/min, in intensive care unit (ICU) patients, and in patients with an infection due to Staphylococcus aureus. The independent predictors of C_min < 2 mg/liter were an eGF > 80 ml/min (odds ratio [OR], 10; 95% confidence interval [CI], 2.732 to 37.037; P = 0.001) and infection due to S. aureus (OR, 5.906; 95% CI, 1.651 to 21.126; P = 0.006). A linezolid C_min of <2 mg/liter was found in 29.5% of cases, and the risk was significantly higher among those with an eGF > 80 ml/min and in infections due to S. aureus. In patients with severe sepsis, a loading dose or continuous infusion and drug monitoring could improve the pharmacodynamic parameters associated with linezolid efficacy.

INTRODUCTION

Linezolid belongs to a family of antimicrobials (oxazolidinones) that inhibit bacterial protein synthesis by preventing the fusion of 30S and 50S ribosomal subunits (1). During the last 10 years, linezolid has been widely used with good clinical results for the treatment of skin and soft tissue infections or nosocomial and health care-related pneumonia (2, 3). Linezolid has time-dependent activity, and the best predictors of its efficacy are the percentage of time that plasma concentrations exceed the MIC and the area under the concentration-time curve over 24 h in the steady state divided by the MIC (AUC₂₄/MIC ratio) (4). In a retrospective study of linezolid in a compassionate use program, a higher success rate was observed when concentrations remained above the MIC for the entire dosing interval and when AUC₂₄/MIC values were 80 to 120 (5). Some recent data suggest that the linezolid trough plasma concentration (C_min), in some patients, is below the MIC₉₀ of linezolid for staphylococci (6).

The aim of our study was to analyze the linezolid trough plasma concentration in a consecutive series of patients with an acute infection and to determine how often trough concentrations were <2 mg/liter (MIC₉₀ of linezolid for Staphylococcus aureus) and other clinically relevant thresholds (0.5, 1, and 4 mg/liter).

MATERIALS AND METHODS

The linezolid trough plasma concentration (C_min) has been monitored at the steady state and within the first 10 days after linezolid prescription in our hospital since 2010. Only one sample per patient was obtained within 1 h previous to the next dose. Linezolid was administered orally or intravenously at 600 mg/12 h. The main clinical characteristics of all patients were retrospectively reviewed. Variables gathered were as follows: age, sex, comorbidity, concomitant administration of proton-pump inhibitors (PPI), rifampin, amlodipine, and amiodarone, intensive care unit (ICU) stay at the moment of obtaining linezolid plasma concentration, source of infection, isolated microorganisms, leukocyte count, serum creatinine level, and estimated glomerular filtration (eGF) using the modification of diet in renal disease (MDRD) formula at the moment of obtaining the linezolid plasma concentration.

Linezolid plasma concentration monitoring was carried out by a high-pressure liquid chromatography (HPLC) method previously validated and described by our group (7). Briefly, sample treatment consisted of precipitation of plasma proteins with trichloroacetic acid, and afterward, the supernatant was collected for the analysis. HPLC was performed with a Waters 2695 Alliance system (Waters, Milford, MA). The mobile phase consisted of a mixture of acetonitrile and water with a flow rate of 1 ml/min. The HPLC analytical column was a C₁₈ 3.5-μm-pore-size reverse-phase column (4.6 by 75 mm). A Waters 2487 UV/visible light detector was used at 254 nm to detect the peaks.

Statistical methods.

Data from different groups of patients were compared by chi-square test or Fisher's exact test (if expected cell counts were less than 5) for categorical variables, and the Student t test or Mann-Whitney U test for continuous variables. A multivariable regression model was used to identify independent risk factors associated with low linezolid C_min. For the statistical analysis, continuous variables were categorized according to median or percentile values. Variables with a P value of less than 0.15 in the univariate analysis were subjected to further selection by using a forward nonconditional logistic procedure, and the criterion for entering or not entering a variable was a value of 0.05. A two-tailed P value of less than 0.05 was considered significant. The goodness of fit of the model was assessed using the Hosmer-Lemeshow test. Statistical analyses were performed using Statistical Package for the Social Sciences (SPSS) version 20.0 (SPSS, Chicago, IL).

RESULTS

A total of 78 patients treated with linezolid for an acute infection had a C_min. The sources of infections and the isolated microorganisms are listed in Table 1. Orthopedic implant infections, skin and soft tissue infections, and pneumonia were the most common sources. The most frequent microorganisms were S. aureus and coagulase-negative staphylococci. Since the linezolid MIC₉₀ for staphylococci is 2 mg/liter, patients were divided in 2 groups (those with a C_min < 2 mg/liter and those with a C_min ≥ 2 mg/liter), and their characteristics are compared in Table 2. A total of 23 of 78 patients (29.5%) had a linezolid C_min < 2 mg/liter. Data regarding sex, comorbidity, and concomitant administration of a PPI were not different between the two groups. Other concomitantly administered drugs such as rifampin (n = 6), amlodipine (n = 9), and amiodarone (n = 2) were studied, but the number of patients receiving these drugs was small, and no differences between groups were found. Patients with a low C_min were younger (60.8 versus 66.8 years of age), more frequently received linezolid through an intravenous route (56% versus 36%), and had higher leukocyte counts than those with C_min ≥ 2 mg/liter; however, the differences were not statistically significant. A significantly higher percentage of patients with a C_min < 2 mg/liter had an infection due to S. aureus or were in the ICU. The median (interquartile range [IQR]) eGF was significantly higher in patients with a low linezolid concentration, while the median (IQR) SCr was significantly lower. The median (IQR) linezolid C_min values for 3 strata of eGF (≤40, 40 to 80, and >80) are shown in Table 3. A significant inverse relationship between eGF and C_min was observed (P = 0.002; Kruskal-Wallis test). Different percentiles of eGF were studied in order to better differentiate the two groups. Among patients with a C_min < 2 mg/liter, 78% had an eGF > 80 ml/min (50th percentile of eGF) and 91% an eGF > 60 ml/min (25th percentile of eGF). The cutoff point with the highest sensitivity and specificity was an eGF > 80 ml/min, and for this reason, this value was included in the univariate and multivariate analyses. Although nonsignificant, there was a progressive increase in the C_min throughout the number of days on linezolid treatment (Fig. 1) and the percentage of patients with a low linezolid C_min was higher within the first 4 days (Table 2).

Table 1.

Infection sources and isolated microorganisms

Source or species	No. (%) of patients
Source (n = 78)
Bone and joint infections	36 (46.1)
Skin and soft tissue infections	11 (14.1)
Pneumonia	8 (10.2)
Central nervous system infections	7 (9)
Bacteremia	5 (6.4)
Other	10 (12.8)
Isolated microorganism species (n = 78)
S. aureus	24 (30.8)
Coagulase-negative staphylococci	23 (29.5)
Enterococcus spp.	4 (5.1)
Streptococcus spp.	4 (5.1)
Other gram-positive bacteria	4 (5.1)
Unknown	19 (24.3)

Table 2.

Characteristics of patients according to the trough plasma concentration of linezolid^a

Patient characteristic	Values for indicated linezolid Cmin		P
	<2 mg/liter (n = 23)	≥2 mg/liter (n = 55)
Mean age in yr (SD)	60.8 (17.4)	66.8 (16.6)	0.156
No. (%) of males	14 (60.9)	26 (47.3)	0.273
No. (%) with indicated comorbidity
Diabetes mellitus	5 (21.7)	9 (16.4)	0.747
COPD	3 (13)	4 (7.3)	0.414
Liver cirrhosis	0	2 (3.6)	1
Chronic renal failure	2 (8.7)	10 (18.2)	0.492
No. (%) receiving proton pump inhibitors	19 (82.6)	46 (83.6)	1
No. (%) with indicated source of infection
Bone and joint	9 (39.1)	27 (49.1)	0.464
Skin and soft tissue	1 (4.3)	10 (18.2)	0.160
Pneumonia	3 (13)	5 (9.1)	0.687
Central nervous system	2 (8.7)	5 (9.1)	1
Bacteremia	3 (13)	2 (3.6)	0.149
Other	5 (21.7)	6 (10.9)	0.285
No. (%) receiving intravenous drug administration	13 (56.5)	20 (36.4)	0.100
No. (%) receiving oral drug administration	10 (43.5)	35 (63.6)
No. (%) infected with S. aureus vs other species	12 (52.2)	12 (21.8)	0.008
No. (%) in intensive care unit	10 (43.5)	11 (20)	0.033
No. with indicated total days of linezolid treatment (%)b
3–4	12 (52.2)	21 (38.2)	0.254
5–7	7 (30.4)	23 (41.8)
8–10	4 (17.4)	11 (20)
Median (IQR) LC (cells/mm3)c	8,800 (6,800–12,500)	7,640 (5,875–9,757)	0.051
Median (IQR) SCr (mg/dl)	0.70 (0.57–0.90)	0.90 (0.76–1.39)	0.005
Median (IQR) eGF (ml/min)	101.2 (84.8–125.2)	71.9 (48.2–96.5)	0.001
No. with eGF > 80 ml/min	18 (78.3)	18 (32.7)	0.0001

^aAbbreviations: C_min, trough plasma concentration; COPD, chronic obstructive pulmonary disease; SD, standard deviation; LC, leukocyte count; SCr, serum creatinine; IQR, interquartile range; eGF, estimated glomerular filtration.

^bData represent comparisons between results determined at ≤4 days (n = 33) and >4 days (n = 45).

^cIn 1 patient, the leukocyte count was not available.

Table 3.

Linezolid trough concentrations according to different estimated glomerular filtration strata

eGF-MDRD (ml/min)a	No. (%) of patients	Median Cmin in mg/liter (IQR)b
0–40	11 (14.1)	10.40 (2.32–18.40)
41–80	31 (39.7)	7.40 (3.10–11.90)
>80	36 (46.2)	1.921 (0.85–5.85)

^aGF-MDRD, estimated glomerular filtration according to MDRD formula; IQR, interquartile range.

^bData represent the results of a Kruskal-Wallis test comparing the 3 median concentrations (P = 0.002).

Fig 1.

Distribution of linezolid C_min according to the number of days on linezolid treatment (Kruskal-Wallis test; P = 0.421). Boxes represent medians and interquartile ranges.

The independent predictors of C_min < 2 mg/liter were an eGF > 80 ml/min (odds ratio [OR], 10; 95% confidence interval [CI], 2.732 to 37.037; P = 0.001) and infection due to S. aureus (OR, 5.906; 95% CI, 1.651 to 21.126; P = 0.006). Infection due to S. aureus and an eGF > 80 ml/min were the variables associated with a C_min of <1 and 4 mg/liter, and there was a trend toward a higher eGF > 80 ml/min in patients with C_min < 0.5 mg/liter (see Tables S1 to S3 in the supplemental material).

The main variables associated with a C_min < 2 mg/liter in the global cohort were evaluated for the 3 subgroups of patients on treatment for ≤4, 5 to 7, or 8 to 10 days for all thresholds (0.5, 1, 2, and 4 mg/liter), and the results are provided in Tables S4 to S7 in the supplemental material. The same trends observed in the global cohort were observed in the different groups and for different thresholds; however, significance was not always achieved due to the low number of cases in each group.

Patients with a platelet count ≥ 130,000 cells/mm³ (normal reference value) before starting linezolid treatment and in whom the platelet count decreased to less than 130,000 cells/mm³ at the same day C_min was obtained were analyzed. There were 6 patients (7.7%) in this situation with a median (IQR) C_min of 12.9 (3.98 to 20.85) mg/liter compared to 4.20 (1.35 to 9.76) in the rest of the cohort (P = 0.10; U Mann-Whitney test).

DISCUSSION

Since the percentage of time that plasma concentrations exceed the MIC is a good predictor of linezolid efficacy, we have retrospectively analyzed the prevalence and risk factors associated with a C_min < 2 mg/liter, the MIC₉₀ of linezolid for S. aureus, and other clinically relevant thresholds. In our study, 29.5% of patients had a C_min < 2 mg/liter, and the percentage increased in patients with infections due to S. aureus (52% versus 21%), in patients in the ICU (43% versus 20%), and in those with an eGF > 80 ml/min (78% versus 32%). These results suggest that patients with severe sepsis are at risk of being underexposed to linezolid during the first days of treatment. Our results are in agreement with two recent studies about linezolid pharmacokinetics (PK). The first one analyzed the C_min of linezolid in 92 patients, and 29% had a C_min below the MIC₉₀ of linezolid (6). The second experiment was performed in 8 critically ill patients who received linezolid at 600 mg/12 h, and 50% of them had a linezolid C_min < 1 mg/liter (8). In contrast, Thallinger et al. (9) studied linezolid concentrations in plasma and interstitial fluid, using a microdialysis technique, in 24 patients (8 with severe sepsis and 16 with septic shock) and they did not find statistically significant differences in plasma AUC_0–24 compared with the concentrations in a control group of healthy volunteers; however, the lowest plasma AUC_0–24 was observed in patients with severe sepsis (100 mg · h/liter versus 146 mg · h/liter in septic shock patients or 159 mg · h/liter in volunteers), and those authors detected a high interindividual variability. For this reason, it is necessary to identify predictors of this variability to optimize the dose of linezolid in patients with severe sepsis, most especially when the MIC is >1 mg/liter. Linezolid has a volume of distribution ranging from 50 to 60 liters; this makes linezolid less susceptible to extensive changes of the extracellular fluid volume, contrasting with hydrophilic antibiotics such as beta-lactams (10); therefore, other parameters should explain the variability in septic patients. Although the level of renal clearance of linezolid is less than 30%, the early phase of sepsis is often a hypermetabolic condition leading to increased renal blood flow, glomerular filtration rate (GFR), renal creatinine clearance, and clearance of renally eliminated drugs (11) and it could explain the progressive decrease in C_min by increasing the values of eGF presented in Table 3. In addition, recent data have shown higher levels of linezolid (12, 13) and a higher toxicity (14, 15) in patients with renal failure, suggesting that renal function impacts linezolid clearance.

The risk of thrombocytopenia within the first 10 days of linezolid therapy is <10% (16). In our study, a decrease in platelet counts of <130,000 cells/mm³ (reference normal value) was observed in 6 patients (7.7%) and they had a median C_min of 12.9 mg/liter, whereas in those without a platelet count decrease, the C_min was 4.2 mg/liter, supporting the hypothesis of a relationship between trough levels and thrombocytopenia that has been previously described by others (13).

A nonsignificant increase in C_min values during treatment was observed (Fig. 1), suggesting a potential accumulation of linezolid. Indeed, Plock et al. (17), using the population pharmacokinetic analysis technique, examined the PK of unbound linezolid in plasma after single and multiple dosing. The examination of the individual concentration-time profiles revealed a change in the disposition of linezolid in comparisons of the concentration-time profiles obtained after single dosing to those obtained after multiple dosing, suggesting PK nonlinearity, which might presumably be attributed to clearance reduction. Those authors hypothesized that linezolid inhibits its own metabolism due to the inhibition of the mitochondrial respiratory chain enzyme activity previously described for linezolid (18). In order to elucidate this aspect, it is necessary to perform a longitudinal study determining consecutive trough concentrations in patients requiring >3 weeks of linezolid.

The major limitations here are the retrospective nature of our study, the lack of a standard protocol to obtain the samples on the same day, and the evaluation of renal function according to the estimated GF using only the MDRD formula, which is not accurate at GF ≥ 60 ml/min (19). Unfortunately, patient weight data were not available and we could not calculate GF using the Cockcroft-Gault formula (CG). Recently, Park et al. (20) found that the CG classification and the MDRD classification of renal function generally agreed but they found important differences in patients with advanced age, low weight, and modestly elevated serum creatinine levels. Their recommendation is to use CG and not MDRD for drug adjustments for subjects with a combination of age > 80 years, weight < 55 kg, and serum creatinine level of >0.7 and ≤1.5 mg/dl. In our study, there were 15 patients >80 years of age and, although the weight data for our patients were not available, it is unlikely that many patients were <55 kg. Last, the best pharmacodynamic predictor of the efficacy of time-dependent antibiotics is probably the AUC₂₄/MIC ratio; therefore, this should be considered in future research.

Our results are of particular interest because, in a recent resistance surveillance of linezolid, the MIC₅₀ of this antibiotic for S. aureus and Enterococcus spp. was 2 μg/ml (21). In a recent clinical trial, the success rate in the linezolid arm was 57.6% and the MIC of linezolid was 2 μg/ml in 67.2% and 4 μg/ml in 27% of the cases (3). Therefore, it is possible to hypothesize that increasing the linezolid dosage would increase the success rate of linezolid treatment. Some authors have suggested a continuous infusion of 1,200 mg/daily after a 300-mg loading dose, demonstrating an improvement of pharmacodynamic indices (8), good diffusion to epithelial lining fluid (22), and good clinical outcome in a small series of cases (23).

In conclusion, a linezolid C_min < 2 mg/liter was found in 29.5% of patients, and the risk was significantly higher among patients with an eGF > 80 ml/min and in patients with severe sepsis. In these patients, a loading dose or continuous infusion and drug monitoring could improve the efficacy of linezolid treatment. Further clinical studies are necessary to validate our results.