Predictors of Inadequate Linezolid Concentrations after Standard Dosing in Critically Ill Patients

Abstract

Adequate linezolid blood concentrations have been shown to be associated with an improved clinical outcome. Our goal was to assess new predictors of inadequate linezolid concentrations often observed in critically ill patients. Fifty-two critically ill patients with severe infections receiving standard dosing of linezolid participated in this prospective observational study. Serum samples (median, 32 per patient) were taken on four consecutive days, and total linezolid concentrations were quantified. Covariates influencing linezolid pharmacokinetics were identified by multivariate analysis and a population pharmacokinetic model. Target attainment (area under the concentration-time curve over 12 h [AUC12]/MIC ratio of >50; MIC = 2 mg/liter) was calculated for both the study patients and a simulated independent patient group (n = 67,000). Target attainment was observed for only 36% of the population on both days 1 and 4. Independent covariates related to significant decreases of linezolid concentrations included higher weight, creatinine clearance rates, and fibrinogen and antithrombin concentrations, lower concentrations of lactate, and the presence of acute respiratory distress syndrome (ARDS). Linezolid clearance was increased in ARDS patients (by 82%) and in patients with elevated fibrinogen or decreased lactate concentrations. In simulated patients, most covariates, including fibrinogen and lactate concentrations and weight, showed quantitatively minor effects on target attainment (difference of ≤9% between the first and fourth quartiles of the respective parameters). In contrast, the presence of ARDS had the strongest influence, with only ≤6% of simulated patients reaching this target. In conclusion, the presence of ARDS was identified as a new and strong predictor of insufficient linezolid concentrations, which might cause treatment failure. Insufficient concentrations might also be a major problem in patients with combined alterations of other covariate parameters. (This study has been registered at ClinicalTrials.gov under registration number NCT01793012.).

FIG 1

Strategy for identifying covariates that influence linezolid concentrations. AUC, area under the linezolid concentration curve; Vd, volume of distribution for linezolid; Cl, linezolid clearance; PK, pharmacokinetic. ¹, study patients (for characteristics, see Table 1); ², predicted AUC values from the basic population pharmacokinetic model; ³, independent patient group of 134 patients (for characteristics, see Table 1); ⁴, simulated patient group (67,000 patients) with 500-fold simulation of all patients from group 2.

FIG 2

Covariates in study patients (patient group 1). ¹, medians of AUC₁₂ are presented in red. *, P < 0.05; **, P < 0.01. 1st indicates the lowest quartile of concentrations of parameters shown in panels A to G, 4th indicates the highest quartile of concentrations of parameters shown in panels A to G. The dashed line indicates the lower threshold for the target range.

FIG 3

Percentage of simulated linezolid AUC₁₂ values expected for the simulated patient group, ranked from lowest to highest. Shown are effects of covariates in simulated patients. Cumulative frequency distributions of AUC₁₂ values are presented for the specified values of respective covariates.

FIG 4

Proposed model for cofactors influencing linezolid pharmacokinetics. Analyses in this study are indicated in red. Cr Cl, creatinine clearance; Fib, fibrinogen; AT, antithrombin. ¹, also described in references and ; ², described in reference ; ³, described in references , , and ; ⁴, described in references , , and .