Association of elevated linezolid trough concentrations with hyperlactatemia in elderly patients: a multicenter observational study in China

Abstract

Background

The association between hyperlactatemia (HLA) and linezolid concentrations in elderly patients remains poorly understood. This study aimed to explore the associations among linezolid trough concentrations, HLA, and clinical outcomes in elderly patients.

Methods

Elderly patients receiving linezolid treatment at four tertiary hospitals in Beijing between May 2021 and June 2023 were enrolled. Linezolid concentrations and arterial lactate levels were dynamically monitored, and risk factors for HLA and mortality were analyzed to develop a clinical prediction model.

Results

A total of 231 patients were studied, among whom 116 (50.2%) developed HLA. The median average trough concentration (C_ave) of linezolid was higher in the HLA group than in the non-HLA group (31.9 vs. 26.8 mg/L, P = 0.01). C_ave of linezolid increased with age and was positively correlated with arterial lactate levels (r = 0.766, P < 0.001) after adjusting for confounding factors. C_ave of linezolid (OR = 1.225 [per 10 mg/L], P = 0.017) and respiratory failure (OR = 1.934, P = 0.022) were independently correlated with HLA, with an AUC of 0.709 (sensitivity 65.8%, specificity 72.5%). The optimal cutoff value of C_ave of linezolid for predicting increased HLA risk was 22.4 mg/L. Elevated linezolid concentration was independently associated with 30-day mortality in exploratory analyses (OR = 1.029, P = 0.026).

Conclusion

Elevated linezolid concentration was independently associated with higher arterial lactate levels, and was exploratorily associated with 30-day mortality in elderly patients. Given the observational design and the limited number of events, these findings should be considered hypothesis-generating. Routine combined monitoring of linezolid concentration and arterial lactate is recommended to guide timely dosage adjustment and reduce adverse outcomes.

Clinical trial registration

Chinese Clinical Trial Registry (trial registration number: ChiCTR2100045707; April 23, 2021).

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s40360-026-01196-y.

Introduction

Linezolid is a synthetic oxazolidinone antibiotic commonly used in the treatment of infections caused by aerobic gram-positive bacteria, including vancomycin-resistant enterococci, methicillin-resistant Staphylococcus aureus, and vancomycin-resistant Staphylococcus aureus, among other multidrug-resistant pathogens [1, 2]. Bone marrow suppression, particularly thrombocytopenia, is the most common adverse effect of linezolid [3–5], while other less frequent adverse events include irreversible peripheral neuropathy, serotonin syndrome, and hyperlactatemia [6–10].

Linezolid-related lactate elevation was initially regarded as a rare complication and received insufficient clinical attention. However, increasing evidence indicates that its actual incidence rate is far higher than previously understood. A series of studies has reported that the incidence of lactate elevation (including lactic acidosis) caused by linezolid is approximately 10%, and the 30-day mortality rate is close to 30% [11–14]. Our previous research on patients aged over 85 years further demonstrated that the incidence of linezolid-related lactic acidosis reached 35.7%, and the 30-day mortality rate was as high as 48.6% [15]. It is essential to clarify that hyperlactatemia (HLA) and lactic acidosis are not the same concept: the former merely indicates an increase in blood lactate levels, while the latter occurs on the basis of HLA combined with metabolic acidosis. Nevertheless, both are closely associated with poor prognosis, particularly in elderly patients, in whom even slight lactate elevations can negatively impact outcomes [10]. Indeed, international guidelines such as the “Surviving Sepsis Campaign” have listed blood lactate levels > 2 mmol/L as a predictor of poor prognosis [16]. Therefore, clinicians should pay attention to linezolid-related HLA. However, current studies mostly focus on lactic acidosis, while few have addressed isolated HLA.

Our previous prospective study confirmed that linezolid-induced thrombocytopenia was significantly positively correlated with its plasma concentration [17]. Based on these findings, we hypothesize that linezolid-associated lactate elevation may also correlate with concentration. At present, relevant studies in this field are limited, and data specifically targeting the elderly population are particularly scarce. Therefore, this study aims to investigate the correlation between linezolid trough concentration and arterial blood lactate level in elderly patients, and analyze its prognostic characteristics.

Materials and methods

Study design, setting, and population

This multicenter, open-label observational study was conducted from May 2021 to June 2023 at the First, Second, Fourth, and Eighth Medical Centers affiliated with Chinese PLA General Hospital in Beijing, China. Inpatients aged ≥ 65 years who were treated with intravenous or oral linezolid (Zyvox; Pfizer, USA) at a standard dose of 1200 mg/day (600 mg twice daily) were eligible for inclusion.

Exclusion criteria were: (1) incomplete clinical data; (2) absence of baseline lactate measurement or dynamic lactate monitoring; (3) receipt of renal replacement therapy at study entry; (4) linezolid treatment duration < 3 days; and (5) estimated survival time < 7 days.

Based on these preliminary data [15], our study estimated that the incidence of linezolid-associated HLA would exceed 40%. Assuming a statistical power of 80%, a two-sided significance level of α = 0.05, and an anticipated dropout rate of 15%, the minimum required sample size was estimated to be approximately 216 patients. A total of 231 patients were ultimately enrolled, which met the prespecified sample size requirement.

Measurement of linezolid concentrations and arterial lactate

Peripheral blood samples (5 mL) were collected from patients’ antecubital fossa into Vacutainers^® containing ethylenediaminetetraacetic acid (Becton Dickinson, Milan, Italy) immediately before the morning dose after 3 days of linezolid treatment. Serum samples were separated by centrifugation at 2500 × g for 10 min, and 2 mL of supernatant were stored at − 20℃ until analysis. Linezolid trough concentrations were determined using a validated liquid chromatography-tandem mass spectrometry method as previously described [18, 19]. The calibration curve was linear over the concentration range of 0.8–100 mg/L, with a lower limit of quantification of 0.8 mg/L (coefficient of variation < 5.0%). Evaluation was performed using the average trough concentration (C_ave) of linezolid, which was calculated from independent measurements. If only one measurement was available, then the average was replaced by the single available measurement.

During linezolid treatment, arterial lactate monitoring was recommended, with actual testing frequency determined at the discretion of treating clinicians. Arterial blood samples (1 mL) were collected in BD Preset syringes (BD, UK) and analyzed within 15 min using blood gas analyzer (Cobas B221, Roche, Switzerland). The baseline lactate level was defined as the arterial lactate level measured within 3 days prior to linezolid initiation. The maximum lactate level was defined as the highest value recorded during linezolid treatment up to 48 h after drug discontinuation.

Data collection

Routinely collected clinical data were extracted from medical records and anonymized. The data-extracting investigator was not involved in patient care. Basic information on the subjects included age, sex, body mass index, underlying diseases (respiratory failure, obstructive pulmonary disease, hypertension, coronary heart disease, acute myocardial ischemia, shock, diabetes, chronic kidney disease, chronic liver disease, malignancy, neurological disease), concomitant medications, Sequential Organ Failure Assessment (SOFA) scores, laboratory findings, sites of infection, duration of linezolid use, and 30-day mortality.

To distinguish linezolid-associated lactate elevation from infection-driven (sepsis-related) hyperlactatemia, the temporal relationship between the onset of lactate elevation and linezolid exposure was evaluated for each case. Alternative causes of hyperlactatemia, including septic shock and tissue hypoperfusion, hepatic failure, and metformin use, were systematically reviewed by two independent investigators. In addition, sensitivity analyses were performed after excluding patients with shock or metformin use (Table S2) and after excluding patients with elevated baseline lactate levels (Table S3).

Definition

HLA was defined as an arterial blood lactate level of ≥ 2 mmol/L in patients with a normal baseline level (< 1.8 mmol/L) or an arterial lactate level more than 1-fold higher than the baseline level in patients with an elevated baseline level (≥ 1.8 mmol/L). Elevated arterial lactate levels were defined using the difference between the maximum and baseline lactate levels. The patients were categorized into the HLA and non-HLA groups according to the presence or absence of HLA.

Additional clinical definitions included: (1) anemia: hemoglobin level < 90 g/L [20]; (2) respiratory failure: PaO₂ <60 mmHg [21]; (3) heart failure: NT-pro-BNP > 450 pg/mL for patients < 50 years old, > 900 pg/mL for patients 50–75 years old, or > 1800 pg/mL for patients > 75 years old [22]; (4) renal failure: estimated glomerular filtration rate < 60 mL/min/1.73 m² calculated using the CKD-EPI formula [23].

Statistical analysis

Statistical analysis was conducted using IBM SPSS Statistics 23.0 (IBM, Armonk, NY, USA). The Kolmogorov–Smirnov test was employed to assess the normality of continuous variables. Normally distributed quantitative data were presented as the mean and standard deviation and analyzed using a t-test. Non-normally distributed quantitative data were expressed as the median and interquartile range and analyzed using the Mann–Whitney U test. Numerical data were compared using the chi-squared test or Fisher’s exact probability test. Factors exhibiting significant differences (P < 0.1) in univariate analysis or clinically relevant factors were included in multivariate regression analysis to identify independent influencing factors. Spearman’s correlation analysis was performed to determine the linear relationship between these factors. A receiver operating characteristic (ROC) curve was developed using the R Base Package and used to evaluate the predictive value of multifactor results. The Kaplan–Meier plot was used to illustrate 30-day mortality rates. Among the 231 enrolled patients, the overall missing rate of laboratory parameters, concomitant medication history and other variables was 3.03%, with 7 patients having only one missing index. Missing data were imputed using multiple imputation prior to analysis. P < 0.05 was considered significant.

Results

Patients’ characteristics

A total of 231 patients were enrolled in this study, with the participant recruitment flow outlined in Fig. 1. Baseline demographic and clinical characteristics of the study cohort are summarized in Table 1. The mean age of the cohort was 92.6 ± 6.4 years, and 93.1% (215/231) of patients presented with pulmonary infection at enrollment. The median duration of linezolid treatment was 11[8,14] days, and the median C_ave of linezolid across the entire cohort was 28.8 [20.6, 40.9] mg/L. Linezolid trough concentration was observed to increase continuously with advancing age and rising arterial lactate levels (Fig. 2a and b).

Fig. 1.

Flow chart of patient enrollment and study design. TDM, Therapeutic Drug Monitoring; SOFA, Sequential organ failure assessment; HLA, hyperlactatemia

Table 1.

Analysis of risk factors of linezolid-associated hyperlactatemia

	All patients(n = 231)	Groups		Univariate	Multivariate
		non-HLA (n = 115)	HLA (n = 116)	P	OR(95%CI)	P
Characteristics
Age, years, x±s	92.6 ± 6.4	92.3 ± 6.4	92.9 ± 6.3	0.471§
Gender, male, N (%)	213(92.2)	106(92.2)	107(92.2)	0.589*
BMI, median (IQR)	23.1[20.2,25.6]	23.2[20.1,25.6]	22.6[20.2,25.5]	0.694#
Linezolid exposure, median (IQR)
Duration, days	11[8,14]	12[9,14]	11[7,14]	0.122#
Time to HLA onset, days			9[6,13]
Cave, mg/L	28.8[20.6,40.9]	26.8[19.9,38.7]	31.9[23.5,43.6]	0.010#	1.225(1.038–1.446)Model 1	0.017
					1.223(1.039–1.439)Model 2	0.015
Underlying diseases, N (%)
Respiratory failure	77(33.3)	30(26.1)	47(40.5)	0.014*	1.934(1.098–3.405)Model 1	0.022
COPD	50(21.6)	26(22.6)	24(20.7)	0.423*
Mechanical ventilation	40(17.3)	16(13.9)	24(20.7)	0.117*
Hypertension	169(73.2)	84(73.0)	85(72.4)	0.543*
Coronary heart disease	154(66.7)	75(65.2)	79(68.1)	0.212*
AMI	20(8.7)	11(9.6)	9(7.8)	0.401*
Arrhythmia	77(33.3)	39(33.9)	38(32.8)	0.481*
Diabetes mellitus	100(43.3)	50(43.5)	50(43.1)	0.530*
Chronic kidney disease	70(30.3)	36(31.3)	34(29.3)	0.426*
Chronic liver disease	9(3.9)	5(4.4)	4(3.4)	0.484*
Malignancy	60(26.0)	29(25.2)	31(26.7)	0.456*
Neurological disease	96(41.6)	45(39.1)	51(44.0)	0.270*
Septic shock	54(23.4)	23(20.0)	31(26.7)	0.146*
Infection sites, N (%)				0.315*
Pulmonary infection	215(93.1)	105(91.3)	110(94.8)
Non-pulmonary infection	16(6.9)	9(7.8)	7(6.0)
Use of other drugs, N (%)
Metformin	11(4.7)	6(5.2)	5(4.3)	0.494*
Carbapenems	163(70.6)	83(72.2)	80(69.0)	0.348*
Cephalosporin	68(29.4)	32(27.8)	36(31.0)	0.348*
Antifungal drugs	81(35.1)	36(31.3)	45(38.8)	0.146*
P-glycoprotein inhibitors	212(91.8)	105(91.3)	107(92.2)	0.684*
Blood transfusion	76(32.9)	34(29.6)	42(36.2)	0.175*
Diuretic	20(8.7)	12(10.4)	8(6.9)	0.235*
Laboratory findings, median (IQR)
Lactate0, mmol/L	1.2[0.9,1.5]	1.1[0.8,1.4]	1.2[0.9,1.5]	0.169#
Lactatemax, mmol/L	2.1[1.6,3]	1.6[1.4,1.8]	2.9[2.2,4.2]	<0.001#
Hemoglobin, g/L	111[99,124]	108[97,125]	112[101,124]	0.310#
Platelet count, 109/L	199[160,244]	200[160,234]	198[157,257]	0.430#
Erythrocyte, 1012/L	3.6[3.2,4.1]	3.5[3.1,4.1]	3.7[3.2,4.1]	0.291#
Albumin, g/L	34.7[30.6,37.3]	35.3[31.0,37.4]	34[30.3,37.3]	0.224#
Creatinine, µmol/L	90[66,139]	88[66,133]	92[67,144]	0.140#
eGFR, ml/min/1.73m2	72[44,97]	73.0[46,101]	69.0[40,97]	0.495#
Bilirubin, µmol/L	11.5[7.9,16.8]	11.1[7.9,15.3]	11.8[7.8,19.5]	0.245#
ALT, U/L	14.7[9.6,24.6]	15.6[9.6,23.8]	14.4[9.7,26.0]	0.886#
Arterial blood glucose, mmol/L,	7.1[5.8,9.3]	7.3[5.7,9.4]	7.1[5.9,9.0]	0.905#
NT-proBNP, pg/ml	1210[539,2712]	1300[569,2483]	1189[504,3208]	0.782#
SOFA, median (IQR)	6[4,9]	5[3,8]	6[4,10]	0.020#	1.078(1.009–1.151)model 2	0.025
30-day mortality, N (%)	16(6.9)	4(3.5)	12(10.3)	0.037*

HLA, hyperlactatemia; BMI, Body mass index; C_ave, Average trough concentration of linezolid; COPD, Chronic obstructive pulmonary disease; P-gp inhibitors, P-glycoprotein inhibitors: Refer to a class of pharmacological agents capable of suppressing the transport activity of the P-g efflux transporter. Representative members include, but are not limited to, calcium channel blockers (e.g., amlodipine), proton pump inhibitors (e.g., omeprazole), amiodarone, cyclosporine, and itraconazole; Lactate₀, Baseline of Arterial lactate; Lactate_max, Maximum lactate level; eGFR, Estimated glomerular filtration rate; ALT, Alanine aminotransferase; NT-proBNP: N-Terminal Pro-Brain Natriuretic Peptide; SOFA, Sequential organ failure assessment

^*Chi-square test

^§t test

^#Mann-Whitney U test

^{Model 1}Including respiratory failure and C_ave of linezolid, per 10 mg/L

^{Model 2}Including SOFA and C_ave of linezolid, per 10 mg/L

Fig. 2.

(a) C_ave of linezolid at different ages. The 40 concentrations of linezolid in non-elderly patients were derived from the previous research results of our group [15]. (b) C_ave of linezolid at different elevated arterial lactate levels. (c) The baseline and maximum arterial lactate levels between the non-HLA and HLA groups. (d) HLA cases at different C_ave of linezolid. (e) C_ave of linezolid in HLA and non-HLA groups. ns, P > 0.05, *P < 0.05, **P < 0.01. HLA, hyperlactatemia; C_ave, Average trough concentration

The incidence of HLA was 50.2% (116/231) in the total population (Table 1). The median time to the first diagnosis of HLA following linezolid treatment was 9 [6,13] days. A total of 102 patients in the HLA group were classified into the category of “normal baseline lactate level with lactate ≥ 2 mmol/L”, and 14 patients were assigned to the category of “elevated baseline lactate level with lactate doubling from baseline”. Baseline arterial lactate levels were comparable between the two groups, while the maximum lactate level was significantly higher in the HLA group compared to the non-HLA group (P < 0.001; Table 1; Fig. 2c).

Risk factors for HLA

Univariate analysis revealed no significant difference in median duration of linezolid treatment between the HLA group and the non-HLA group (11 [7, 14] days vs. 12 [9, 14] days; P = 0.122; Table 1). Median C_ave of linezolid was significantly higher in the HLA group compared to the non-HLA group (31.9 [23.5, 43.6] mg/L vs. 26.8 [19.9, 38.7] mg/L, P = 0.01; Table 1; Fig. 2e). Additionally, the HLA group had a significantly higher SOFA score (6 [4, 10] vs. 5 [3, 8]; P = 0.02; Table 1) and a higher prevalence of respiratory failure (40.5% vs. 26.1%; P = 0.014; Table 1) compared with the non-HLA group.

Two multivariate analysis models were constructed to identify independent correlates of HLA. In Model 1, C_ave of linezolid (OR = 1.225 [per 10 mg/L], 95% CI 1.038–1.446; P = 0.017) and respiratory failure (OR = 1.934, 95% CI 1.098–3.405; P = 0.022) emerged as independent risk factors for HLA. In Model 2, C_ave of linezolid (OR = 1.225 [per 10 mg/L], 95% CI 1.039–1.439; P = 0.015) and SOFA score (OR = 1.078, 95% CI 1.009–1.151; P = 0.025) remained independently associated with HLA (Table 1).

After excluding the effects of shock and metformin, C_ave of linezolid (OR = 1.230 [per 10 mg/L], 95% CI 1.016–1.364; P = 0.001) and respiratory failure (OR = 2.399, 95% CI 1.173–4.905; P = 0.001) emerged as independent risk factors for HLA (Table S2). After excluding patients with abnormal baseline lactate levels (≥ 1.8 mmol/L), two multivariate analysis models were constructed to identify independent correlates of HLA. In Model 1, C_ave of linezolid (OR = 1.231 [per 10 mg/L], 95% CI 1.042–1.455; P = 0.013) and respiratory failure (OR = 2.073, 95% CI 1.110–3.873; P = 0.022) emerged as independent risk factors for HLA. In Model 2, C_ave of linezolid (OR = 1.227 [per 10 mg/L], 95% CI 1.052–1.457; P = 0.015) and SOFA score (OR = 1.092, 95% CI 1.012–1.178; P = 0.023) remained independently associated with HLA (Table S3).

The diagnostic performance of combined biomarkers for the early detection of HLA was assessed using ROC curve analysis. The model incorporating C_ave of linezolid and respiratory failure yielded an AUC of 0.709 (95% CI 0.634–0.774; P < 0.001). When C_ave of linezolid was combined with SOFA score, the predictive model achieved a comparable AUC of 0.702 (95% CI 0.534–0.734; P = 0.032). The optimal cutoff value of C_ave of linezolid for predicting increased HLA risk was 22.4 mg/L, corresponding to a sensitivity of 83.8% and specificity of 39.0% (Table 2; Fig. 3d).

Table 2.

Predictive performance of total points and other factors for predicting hyperlactatemia and 30-day mortality in the elderly population

	Variables	AUC 95% CI	Sensitivity, %	Specificity,%	P
HLAModel 1	Total points	0.709(0.634–0.774)	65.8	72.5	< 0.001
	Cave of linezolid, per 10 mg/L	0.665(0.593–0.738)	83.8	39.0	< 0.001
	Respiratory failure	0.585(0.508–0.661)	41.4	75.5	0.033
HLAModel 2	Total points	0.702(0.534–0.734)	61.3	70.5	0.032
	Cave of linezolid, per 10 mg/L	0.598(0.525–0.671)	83.8	39.0	0.035
	SOFA, median (IQR)	0.588(0.515–0.662)	30.2	84.3	0.037
30-day mortality	Total points	0.846(0.772–0.920)	93.8	68.8	< 0.001
	Cave of linezolid, mg/L	0.655(0.520–0.790)	50.0	81.9	0.039
	Lactatemax, mmol/L	0.704(0.580–0.829)	65.3	77.2	0.006
	SOFA	0.835(0.756–0.913)	75.0	80.9	< 0.001
	Septic shock	0.710(0.567–0.853)	62.5	79.5	0.005

C_ave, Average trough concentration; Lactate_max, Maximum lactate level; SOFA, Sequential organ failure assessment

Fig. 3.

(a) The correlation between elevated arterial lactate levels and C_ave of linezolid (r = 0.412, P < 0.001). (b) The correlation between elevated arterial lactate levels and C_ave of linezolid after excluding patients with respiratory failure (r = 0.605, P < 0.001). (c) The correlation between elevated arterial lactate levels and C_ave of linezolid after excluding patients with respiratory failure, shock, anemia, renal failure, and heart failure(r = 0.766, P < 0.001). (d) The area under the ROC curve (AUC) for C_ave of linezolid combined with respiratory failure for the early detection of HLA was 0.709 (95% CI = 0.634–0.774, P < 0.001). The optimal cutoff value of C_ave of linezolid for predicting increased HLA risk was 22.4 mg/L. (e) Kaplan-Meier plot showing that 30-day survival in HLA group was lower than that in non-HLA group (mortality: 10.3% vs. 3.5%, P = 0.037). (f) ROC curve to evaluate the predictive value of C_ave of linezolid and other factors for predicting 30-day mortality in the elderly patients. AUC was 0.846(95%CI = 0.772–0.920, P < 0.001). HLA,hyperlactatemia; Arterial lactate 0, Arterial lactate at baseline; Arterial lactate max, Maximum lactate level; C_ave, Average trough concentration; SOFA, Sequential organ failure assessment

When stratified by C_ave of linezolid, the incidence of HLA was 40% (2/5) in patients with trough concentration < 8 mg/L, 25% (7/28) in those with 8–16 mg/L, 50% (45/90) in those with 16–30 mg/L, and 57.4% (62/108) in those with > 30 mg/L, respectively (Fig. 2d). The incidence of HLA was higher in the 16–30 and > 30 mg/L groups than in the 8–16 mg/L group (P < 0.05), and the incidence was significantly higher in the > 30 mg/L group than in the < 8 mg/L group (P < 0.001, Fig. 2d).

Correlation between C_ave of linezolid and HLA

In the total population, elevated arterial lactate levels displayed a mild linear correlation with C_ave of linezolid (r = 0.412, P < 0.001, Fig. 3a). Considering the effect of respiratory failure on lactate levels, a moderate linear correlation was found between elevated lactate levels and C_ave of linezolid (r = 0.605, P < 0.001, Fig. 3b) after excluding patients who developed respiratory failure. Furthermore, after excluding confounding factors that affect arterial lactate levels such as respiratory failure, anemia, shock, renal failure, and heart failure, there was a strong linear correlation between arterial lactate elevation and C_ave of linezolid (r = 0.766, P < 0.001, Fig. 3c).

Risk factors for 30-day mortality

The 30-day mortality rate was significantly higher in the HLA group compared to the non-HLA group (10.3% [12/116] vs. 3.5% [4/115]; P = 0.037; Table 1; Fig. 3e). Multivariate analysis revealed that C_ave of linezolid (OR = 1.029, 95% CI = 1.003–1.056, P = 0.026), maximum arterial lactate levels (OR = 1.268, 95% CI = 1.029–1.562, P = 0.026), SOFA score (OR = 1.271, 95% CI = 1.115–1.448, P < 0.001), and septic shock (OR = 6.087, 95% CI = 2.068–17.917, P = 0.01) were independent risk factors for 30-day mortality (Table 3). The AUC for 30-day mortality for combined risk factors using ROC curves was 0.846 (95% CI = 0.772–0.920, P < 0.001) with a sensitivity of 93.8% and specificity of 68.8% (Table 2; Fig. 3f).

Table 3.

Univariate and multivariate analysis of 30-day mortality

	Outcome		Univariate	Multivariate
	Survival(215)	Death(16)	P	OR(95%CI)	P
Duration, days, median (IQR)	11[8,14]	7.5[7,12]	0.023#
Cave of linezolid, mg/L	28.7[20.2,40.7]	40.6[25.6,52.3]	0.039#	1.029(1.003–1.056)	0.026
Lactate0, mmol/L, median (IQR)	1.1[0.9,1.4]	1.55[1.13,1.8]	0.011#
Lactatemax, mmol/L, median (IQR)	2.1[1.6,2.9]	3.1[2.1,6.4]	0.006#	1.268(1.029–1.562)	0.026
HLA, N (%)	104(48.4)	12(75)	0.035*
SOFA, median (IQR)	6[4,9]	11.5[9.3,13.5]	<0.001#	1.271(1.115–1.448)	<0.001
Respiratory failure, N (%)	67(31.2)	10(62.5)	0.013*
Blood transfusion, N (%)	65(30.2)	11(68.7)	0.012*
Septic shock, N (%)	44(20.5)	10(62.5)	0.001*	6.087(2.068–17.917)	0.01

C_ave, Average trough concentration; Lactate₀, Baseline of Arterial lactate; Lactate_max, Maximum lactate level; HLA, hyperlactatemia; SOFA, Sequential organ failure assessment

^*Chi-square test

^#Mann-Whitney U test

Discussion

This study investigated the association between C_ave of linezolid and lactate elevation and survival in elderly patients. Our findings demonstrate that over 50% of elderly patients developed HLA during linezolid therapy, with each 10 mg/L increment in C_ave of linezolid associated with a 22.5% higher odds of HLA occurrence. Furthermore, elevated C_ave of linezolid was independently associated with 30-day all-cause mortality in exploratory analyses; given the limited number of death events, this finding should be regarded as hypothesis-generating.

In this study, it was observed that the trough concentration of linezolid increased with age, which was consistent with our previous findings [17]. The median trough concentration of linezolid in the entire study cohort was 28.8 [20.6, 40.9] mg/L, which is a relatively high value. This might be attributed to the disproportionately high proportion of patients aged 90 or above and the significant variations in their pathophysiology. Furthermore, HLA occurred in more than half of the patients in our study, and C_ave of linezolid in the HLA group was significantly higher than in the non-HLA group, with a median exceeding 30 mg/L, suggesting that patients who developed HLA had substantial linezolid overexposure. In addition, the risk of HLA increased as C_ave of linezolid increased, especially in patients with concentration > 16 mg/L. Numerous previously published case reports describing linezolid-associated lactate elevation have similarly observed that trough concentrations of linezolid were consistently above 20 mg/L at the onset of this adverse reaction [24–30].

It should be noted that the trough concentrations observed in this cohort considerably exceeded the commonly referenced therapeutic window. In this study, all patients received the standard dose of 1200 mg/day (600 mg twice daily) throughout TDM (Therapeutic Drug Monitoring), in accordance with the approved drug label and national guidelines for the clinical use of linezolid, which do not recommend age- or renal-function-based dose adjustment at treatment initiation. This uniform standard dosing reflects real-world prescribing practice. Nevertheless, given that reduced drug clearance is highly predictable in very elderly patients (the mean age of this cohort was 92.6 years), our findings demonstrate that standard dosing in elderly patients frequently leads to supratherapeutic trough levels and an increased risk of HLA, thereby underscoring the critical necessity of routine TDM and TDM-guided dose individualization in this population.

Our study observed a significant independent association between higher C_ave of linezolid and increased risk of HLA in elderly patients. Each 10 mg/L increase in C_ave of linezolid was associated with a 22.5% higher odds of HLA in adjusted analyses. By contrast, Cheng et al. [31] recruited 50 non-elderly patients receiving linezolid treatment and found no significant correlation between elevated lactate and trough concentration of the drug. This discrepancy may be attributed to the fact that all participants in their study were non-elderly individuals, and the peak plasma concentration of linezolid was relatively low, with a mean value of 9.6 mg/L.

Notably, in the total population, there was a weak linear correlation between C_ave of linezolid and elevated lactate levels. This may reflect the disruptive effect of comorbidities on lactate metabolism. This correlation was strengthened after excluding factors affecting arterial lactate, such as respiratory failure, heart failure, anemia, renal failure, and shock. Collectively, these findings indicate that linezolid-associated lactate elevation is independently associated with increased linezolid exposure, potentially mediated by inhibition of mitochondrial ribosomal protein synthesis, which impairs oxidative phosphorylation and respiratory chain enzyme activity, ultimately enhancing anaerobic metabolism and arterial lactate production [32, 33].

Our analyses also revealed that the 30-day mortality rate was significantly higher in the HLA group than in the non-HLA group. Further analysis confirmed that C_ave of linezolid was an independent risk factor for all-cause mortality. A predictive model constructed by combining this indicator with other clinical risk factors demonstrated excellent performance in predicting 30-day mortality (AUC = 0.846). These findings not only confirm the elevated mortality risk in patients with HLA but also establish a link between this risk and linezolid exposure levels, indicating that a high trough concentration itself is an independent driver of poor prognosis, rather than a mere confounding epiphenomenon.

Previous studies have indicated that prolonged linezolid therapy was associated with an elevated risk of lactic acidosis [34–38]. In the present cohort, however, no statistically significant association was identified between treatment duration and the development of HLA. A plausible explanation for this observation is that HLA represents an early stage in the spectrum of lactic acidosis; consequently, at the time of assessment, the potential effect of differential treatment duration may have remained undetectable.

This study has several limitations. First, arterial lactate measurement was performed at the discretion of the physician, and sicker patients are monitored more frequently, which could lead to selection bias. Second, analyses only focused on HLA, without exploring linezolid’s effect on acidosis requiring intervention, due to widespread use of HCO3-/pH-altering medications in the elderly cohort. Third, due to the small number of death events (n = 16) in this study, the multivariate mortality model had a low events-per-variable ratio (EPV = 4) and should be interpreted as hypothesis-generating only. The results require validation in larger cohorts. Fourth, linezolid-induced mitochondrial toxicity and sepsis exacerbation share overlapping clinical features. Although we evaluated the temporal relationship between lactate elevation and linezolid exposure and performed sensitivity analyses, SOFA scores were captured at baseline without serial reassessment; therefore, we could not fully adjust for the dynamic progression of sepsis. Residual confounding by infection-driven multiorgan failure cannot be excluded, and the observed associations—particularly with 30-day mortality—should be interpreted with caution. Fifth, latent vitamin B1 (thiamine) deficiency, which is common in frail elderly patients and impairs pyruvate dehydrogenase activity, represents a strong potential confounder for hyperlactatemia. Baseline nutritional status and vitamin B1 levels were not routinely assessed in this cohort and could not be adjusted for in our analyses.

Conclusions

In this observational study, elevated C_ave of linezolid was independently associated with HLA in elderly patients, with each 10 mg/L increase in C_ave corresponding to a 22.5% increased odds of HLA. These findings suggest that monitoring blood concentration and arterial lactate levels during linezolid therapy in elderly patients may facilitate the early identification of HLA. Furthermore, elevated C_ave of linezolid was identified as an independent risk factor for 30-day all-cause mortality. Accordingly, maintaining linezolid trough concentration within an appropriate therapeutic range through therapeutic drug monitoring may improve survival outcomes in elderly patients. Nevertheless, this hypothesis warrants further validation in prospective studies.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

LHX, YYP, LTT, XLX and FXQ contributed to the study design. WJH, NP, CXM and RWQ were responsible for the centrifugation and testing of blood samples. YYP, LTT, WJH, WC, WX, MGX, ZC, ZJS, LJX and DZM contributed to the collection of clinical data. YY, WJ, MGX, ZQK and WC contributed to collecting blood samples. YYP, LHX and FXQ contributed to the data analysis. YYP, LTT and WX drafted the manuscript. All authors read and approved the final manuscript.

Funding

The authors declare that no funds, grants, or other support were received during the preparation of this manuscript.

Data availability

To protect study participant privacy, our data cannot be shared openly. However, the data are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

The study follows the principles of the Declaration of Helsinki of the World Medical Association. The study protocol was approved by the Medical Ethics Committee of the Chinese People’s Liberation Army (PLA) General Hospital (Ethical approval number: S2020-206-01), and written informed consent was obtained from all participants or their legal agents. This study fully protected patient privacy and personal information, and participants or legal representatives also consented to their data being published in a journal article.

Competing interests

The authors declare no competing interests.