Systematic Therapeutic Drug Monitoring for Linezolid: Variability and Clinical Impact

ABSTRACT

Linezolid serum trough (C_min) and peak (C_max) levels were determined prospectively in 90 patients. Adequate exposure was defined as a C_min of 2 to 8 mg/liter. Therapy was empirical (73.3%) or targeted (26.7%). Wide interindividual variability in linezolid C_min levels was recorded (0.1 to 25.2 μg/ml). Overall, 65.5% of the patients had out-of-range, 41.1% had subtherapeutic, and 24.4% had supratherapeutic trough levels. We did not find a correlation between abnormal levels and adverse events, in-hospital mortality, or overall poor outcome.

TEXT

Linezolid has become increasingly important for the treatment of multidrug-resistant infections caused by Gram-positive microorganisms (1, 2). Pivotal studies reported that 600 mg/12 h was the standard dose of linezolid for patients aged >12 years (3), and current guidelines do not recommend therapeutic drug monitoring (TDM). However, different studies reported significant variations in the serum levels of linezolid in specific situations and in patients taking concomitant medications, such as phenobarbital, dexamethasone, rifampin, proton pump inhibitors, calcium channel antagonists, and amiodarone (4–12). We questioned the need for linezolid TDM, as suggested by several authors (1, 8, 13–16).

We performed a prospective study of inpatients receiving linezolid for empirical or targeted treatment at a tertiary hospital. The patients agreed to participate and gave their written informed consent. The dosage of linezolid was standard and in accordance with guidelines. The study was approved by the Ethics Committee of Hospital General Universitario Gregorio Marañón, Madrid, Spain (study number MICRO.HGUGM.2016-014). Two blood samples were drawn from each patient at least 3 days after initiation of treatment. Trough levels (C_min) were obtained within 30 min of drug administration, and peak levels (C_max) were obtained 1 h after intravenous (IV) infusion or 2 h after oral administration. Trough levels of 2 to 8 μg/ml and peak levels of 10 to 20 μg/ml were considered normal (3).

Linezolid serum levels were detected by a validated high-performance liquid chromatograph method (17). The results were reported to the physician responsible for the patient without further recommendations. Clinical outcome was classified, prospectively, as favorable when there was a clinical improvement or cure and no evidence of adverse events (thrombocytopenia, anemia) and as poor when there was no clinical response, infection recurrence, related mortality, or adverse events. Episodes of thrombocytopenia and anemia during treatment were defined as a reduction of >30% in the platelet count or hemoglobin level from baseline, respectively (5).

Ninety patients were included in the study (68.9% male). Patient characteristics and underlying diseases are shown in Table 1. Linezolid was prescribed as empirical treatment in 73.3% and as targeted therapy in 26.7% of the patients (Table 2). The median dose of linezolid was 8.8 mg/kg (interquartile range [IQR], 7.5 to 10.0 mg/kg), and the median duration of linezolid treatment was 8 days (IQR, 5.0 to 13.2 days). The main reasons for linezolid therapy, accounting for 74.5% of the total, were pneumonia, complicated skin and soft tissue infections, and undocumented febrile episodes. Seventy-seven percent of the patients were treated with proton pump inhibitors (Table 2). Infection, caused mainly by Staphylococcus aureus or coagulase-negative Staphylococcus, was confirmed microbiologically in 30.0% of the patients. All isolates were susceptible to linezolid. Attending physicians adjusted the linezolid dose in only 3 patients based on serum levels.

TABLE 1.

Demographic data, department of admission, comorbidities, and other clinical characteristics

Characteristica	Value (n = 90)
Age (yr) (median [IQR])	69.3 (57.5–79.9)
Sex, male (no. [%])	62 (68.9)
Weight (kg) (median [IQR])	68.8 (60.0–82.0)
Body mass index (mg/kg2) (median [IQR])	25.1 (22.2–29.2)
Race (no. [%])
White	85 (94.4)
Black	2 (2.2)
Other	3 (3.3)
Department of admission (no. [%])
Medical	43 (47.8)
Surgical	13 (14.4)
ICU	33 (36.7)
Pediatric	1 (1.1)
Days in ICU (median [IQR])	15 (8.0–26.0)
Underlying disease (no. [%])
Cardiac disease	29 (32.2)
Neurologic disease	25 (27.8)
Diabetes mellitus	22 (24.4)
Liver disease	20 (22.2)
Solid tumor	18 (20.0)
Chronic renal failure	11 (12.2)
Psychiatric disease	10 (11.1)
Chronic obstructive pulmonary disease	9 (10.0)
HIV infection	3 (3.3)
Hematologic neoplasia	2 (2.2)
Solid organ transplantation	1 (1.1)
Other	3 (3.3)
Charlson comorbidity index (median [IQR])	3 (2–5)
McCabe index (no. [%])
Nonfatal	56 (62.2)
Ultimately fatal	27 (30.0)
Rapidly fatal	7 (7.8)
Glomerular filtration rate (MDRD) (no. [%])
Normal (≥60 ml/min per 1.73 m2)	61 (67.8)
Low (<60 ml/min per 1.73 m2)	27 (30.0)
Extracorporeal membrane oxygenation (no. [%])	1 (1.1)
Hemodialysis (no. [%])	3 (3.3)

^aICU, intensive care unit; IQR, interquartile range; MDRD, modification of diet in renal disease.

TABLE 2.

Treatment characteristics; indications; microbiological isolates; dose, duration, and C_min, and C_max of linezolid; concomitant medications; and clinical outcome

Characteristic	Value (%) (n = 90)
Type of treatment (no. [%])
Empirical	66 (73.3)
Targeted	24 (26.7)
Main indication for linezolid (no. [%])
Pneumonia	50 (55.6)
Complicated skin and soft tissue infection	10 (11.1)
Undocumented febrile episode	7 (7.8)
Osteoarticular infection	4 (4.4)
Mediastinitis	4 (4.4)
Intra-abdominal infection	4 (4.4)
CNSa infection	3 (3.3)
Bacteremia	2 (2.2)
Other	6 (6.7)
Microbiological isolate (no. [%])
Coagulase-negative Staphylococcus	13 (14.4)
Staphylococcus aureus	9 (10)
Enterococcus faecalis	2 (2.2)
Enterococcus faecium	1 (1.1)
Enterococcus sp.	1 (1.1)
Corynebacterium sp.	1 (1.1)
Linezolid treatment
Duration of treatment (days) (median [IQR])	8.0 (5.0–13.2)
Dose (mg/kg) (median [IQR])	8.8 (7.5–10.0)
Route of administration (no. IV/no. oral)	61/29
Cmin (mg/liter) (median [IQR])	2.9 (0.7–7.7)
Cmax (mg/liter) (median [IQR])	11.9 (7.8–17.9)
Concomitant medication (no. [%])
Phenobarbital	0 (0)
Dexamethasone	4 (4.4)
Rifampin	1 (1.1)
Proton pump inhibitor	70 (77.8)
Calcium channel antagonist	11 (12.2)
Amiodarone	5 (5.6)
Overall outcome (no. [%])
Favorable	66 (73.3)
Poor	24 (26.7)

^aCNS, central nervous system.

The median C_min in our study was 2.9 mg/liter, with a wide range of distribution (0.1 to 25.2 mg/liter) (Table 2). C_min was below the therapeutic range (<2 mg/liter) in 41.1% and >8 mg/liter in 24.4% of patients. The median C_max was 11.9 mg/liter and also had a wide range of distribution (1.7 to 36.8 mg/liter) (Table 2). C_max was below the optimal range (<10 mg/liter) in 34.1% and >20 mg/liter in 20.4% of patients.

We analyzed the variables that might influence linezolid serum levels. Multivariate analysis confirmed that patients with higher body weight had lower C_min values and those with a higher age-adjusted Charlson comorbidity index value or lower glomerular filtration rate had higher C_min levels (Table 3). The adjusted R² of 0.404 proved that 40% of the variability in C_min was related to these variables. Multivariate analysis confirmed that higher weight correlated with lower C_max and that lower glomerular filtration rate correlated with increased C_max (Table 3). The adjusted R² of 0.234 proved that one-quarter of the variability in C_max of linezolid among our population was related to these variables. Conversely, we did not identify a significant correlation between phenobarbital, dexamethasone, rifampin, proton pump inhibitors, calcium channel antagonists, and amiodarone and the C_min and/or C_max (Table 3) of linezolid in the univariate analysis.

TABLE 3.

Univariate and multivariate analyses of variables associated with C_min and C_max of linezolid

Variable	Univariate analysis		Multivariate analysis
	Unstandardized β coefficient (95% CI)	P	Unstandardized β coefficient (95% CI)	P
Cmina (n = 90)
Age (yr)	0.098 (0.015 to 0.181)	<0.01
Sex	−0.224 (−0.515 to 0.068)	0.131
Weight (kg)	−0.012 (−0.018 to −0.007)	<0.01	−0.013 (−0.018 to −0.008)	<0.01
Body mass index (kg/m2)	−0.026 (−0.041 to −0.010)	<0.01
Glomerular filtration rate (MDRDb)	0.446 (0.161 to 0.731)	<0.01	0.364 (0.131 to 0.597)	<0.01
Concomitant treatment
Phenobarbital	NAc
Dexamethasone	−0.132 (−0.795 to 0.531)	0.694
Rifampin	0.659 (−0.638 to 1.956)	0.315
Proton pump inhibitor	−0.158 (−0.485 to 0.169)	0.340
Calcium channel antagonist	0.282 (−0.131 to 0.695)	0.178
Amiodarone	−0.393 (−0.984 to 0.198)	0.190
Charlson comorbidity index	0.121 (0.063 to 0.180)	<0.01	0.100 (0.049 to 0.150)	<0.01
McCabe index	0.151 (−0.062 to 0.364)	0.162
Cmax (n = 88)
Age (yr)	0.098 (0.015 to 0.181)	0.022
Sex	2.911 (−0.515 to 6.337)	0.095
Weight (kg)	−0.119 (−0.183 to −0.056)	<0.01	−0.139 (−0.203 to −0.075)	<0.01
Body mass index (kg/m2)	−0.266 (−0.449 to −0.084)	<0.01
Glomerular filtration rate (MDRD)	4.074 (0.625 to 7.522)	0.021	3.725 (0.585 to 6.865)	0.021
Concomitant treatment
Phenobarbital	NA
Dexamethasone	−2.368 (−10.061 to 5.324)	0.542
Rifampin	2.589 (−12.552 to 17.729)	0.735
Proton pump inhibitor	−0.269 (−4.101 to 3.562)	0.889
Calcium channel antagonist	2.583 (−2.447 to 7.612)	0.310
Amiodarone	−5.461 (−12.472 to 1.189)	0.104
Charlson comorbidity index	0.773 (0.051 to 1.495)	0.036
McCabe index	1.346 (−1.154 to 3.847)	0.287

^aTrough levels were log₁₀-transformed before being compared because of their non-normal distribution.

^bMDRD, modification of diet in renal disease.

^cNA, not available.

Overall, 66 patients (73.3%) achieved a favorable outcome. As for adverse events, only 2 patients (2.2%) developed anemia during treatment. However, 12 out of 90 patients (13.3%) experienced a >30% reduction in platelet count. The median C_min in the 14 patients was 4.6 mg/liter (range, 0.1 to 18.1 mg/liter). Discontinuation of linezolid was deemed necessary in only 1 of these patients.

No clear correlation was observed between clinical outcome with the dose administered or abnormal trough levels of linezolid (subtherapeutic and supratherapeutic independently or out of range together) (Fig. 1). To further evaluate the potential relationship of linezolid levels with clinical outcome, we classified levels as normal or out of range. No differences were seen between normal and out-of-range levels with respect to adverse events (15.6% versus 15.5%, respectively; P = 0.9), in-hospital mortality (18.8% versus 15.5%, respectively; P = 0.7), or overall poor outcome (25.0% versus 27.6%, respectively; P = 0.8). Similarly, we were unable to demonstrate differences between normal and out-of-range peak levels or empirical and targeted therapy when they were analyzed separately.

FIG 1.

Correlation of administered doses of linezolid, obtained trough levels, and clinical outcomes (adverse events and mortality).

In our study, a high proportion of patients had inadequate linezolid levels, but we did not observe a clinical impact in patients who were outside the therapeutic range.

Different retrospective studies reported that 29% to 50% of patients had linezolid trough levels of <2 mg/liter (5, 7, 18), and trough levels of <1 mg/liter were observed in 50% of critically ill patients with sepsis (19).

Evaluation of potential factors affecting the C_min of linezolid in the present study showed that part of the interpatient variability was due to weight, renal function, and the presence of comorbidities adjusted by age. The correlation between linezolid serum levels and weight and renal function has been reported by others (20–24). Age may also influence linezolid trough concentrations (25), since elderly patients with low body weight had a higher risk of accumulating linezolid (26). However, other studies did not find a correlation between age, weight, and renal function when evaluating the pharmacokinetics-pharmacodynamics of linezolid (5, 27, 28). Of note, relevant pharmacokinetic interactions of linezolid with several drugs have been reported in adult and pediatric patients (5, 7, 29, 30).

Our findings, with lack of correlation between levels and mortality or adverse events, do not support the need to assess linezolid serum trough levels. Note that none of the previous studies searched for a correlation between drug levels and clinical outcome, although patients with renal dysfunction showed a higher rate of adverse events due to accumulation of linezolid in some of the studies (8, 15, 16, 31, 32). Some authors suggest that the high frequency of low drug exposure with standard dosages of linezolid may explain the appearance of coagulase-negative staphylococci with resistance to linezolid (5).

Our study is based on real-life data in a tertiary hospital. However, the data were subject to limitations. The small sample size, heterogeneity of the cases, and fact that most of the patients were receiving linezolid as empirical therapy during short periods are the most relevant. In addition, dose adjustments were made in only 3 patients at the discretion of the attending physicians, and we did not intervene directly with that decision.

We conclude that a high proportion of linezolid levels were off target at initiation of antibiotic therapy. Despite the significant influence of weight, renal function, and comorbidities in trough and peak serum linezolid levels, we found no correlation between abnormal levels and clinical outcome.