Abstract
Background
The optimal dosing strategy of linezolid for treating multidrug-resistant and rifampicin-resistant tuberculosis remains unclear. We conducted an individual patient data meta-analysis to determine the optimal linezolid dosing strategy.
Methods
We searched for randomised controlled trials and prospective cohort studies on short-course all‑oral regimens containing linezolid for treating multidrug-resistant and rifampicin-resistant tuberculosis in PubMed, Embase and Scopus up to 31 August 2023. Patients were grouped according to linezolid dosing patterns. Time to treatment success and adverse events of grade 3 and higher were analysed using the Fine–Gray sub-distribution hazard model.
Results
Of 12 eligible studies, eight (four randomised controlled trials, four prospective studies) were included. Overall, 945 patients were grouped as follows: group 1 (600 mg·day−1 linezolid for 8 weeks), group 2 (600 mg·day−1 for 16 weeks, then 300 mg·day−1 for 8 weeks), group 3 (600 mg·day−1 for 39 weeks) and group 4 (1200 mg·day−1 for 25 weeks). Proportions of patients achieving treatment success were 59.1%, 90.4%, 91.3% and 96.0%, respectively. Compared with group 2, group 1 (adjusted sub-distribution hazard ratio (SHR) 0.24, 95% CI 0.08–0.71) and group 3 (adjusted SHR 0.36, 95% CI 0.16–0.81) had lower success rates. While group 4 showed no significant difference in treatment success versus group 2 (adjusted SHR 0.57, 95% CI 0.23–1.43), it had a higher rate of adverse events of grade 3 and higher (adjusted SHR 2.29, 95% CI 1.37–3.83).
Conclusion
A dosing strategy of 600 mg·day−1 linezolid for 16 weeks then 300 mg·day−1 for 8 weeks could be optimal for treating multidrug-resistant and rifampicin-resistant tuberculosis when considering effectiveness and safety.
Shareable abstract
In all-oral regimens for multidrug-resistant TB, a dose-tapering strategy for linezolid (600 mg daily for 16 weeks, then 300 mg daily for 8 weeks) achieved a favourable balance between efficacy and adverse events compared to other dosing strategies. https://bit.ly/3EoDYUR
Introduction
Multidrug-resistant or rifampicin-resistant tuberculosis (MDR/RR-TB) is a global health threat. In 2023, there were an estimated 400 000 cases of MDR/RR-TB and 150 000 deaths worldwide [1]. MDR/RR-TB had conventionally been treated with regimens including injectable drugs for at least 18–24 months because of the limited availability of potent agents [2]. However, MDR/RR-TB treatment has been revolutionised over the last decade by new drugs, such as bedaquiline, delamanid and pretomanid, as well as repurposed drugs, including linezolid and fluoroquinolones, allowing short-course oral regimens without the use of injectable drugs [3–6].
Linezolid, an oxazolidinone derivative that inhibits protein synthesis, has significantly improved treatment outcomes and reduced mortality in MDR/RR-TB patients [7]. Although it has become a cornerstone of shorter, fully oral regimens, its use is frequently associated with clinically important adverse events, including optic or peripheral neuropathy and myelosuppression. The reported proportions of patients experiencing peripheral neuropathy and myelosuppression during MDR/RR-TB treatment with linezolid-containing regimens range from 36.1% to 47.1% and from 28.5% to 38.1%, respectively [8, 9]. These adverse events, correlated with higher trough linezolid concentrations, lead to frequent interruption, discontinuation and dose reduction of linezolid [10]. Adverse events may necessitate a dose reduction from 600 mg to 300 mg in up to 70% of patients and drug discontinuation in approximately 34% [11].
Given these findings, it is imperative to determine the optimal dosage and duration of linezolid administration to improve treatment outcomes while minimising adverse events. In this study, we assessed the impact of linezolid dose and duration on outcomes and adverse events during treatment for MDR/RR-TB using individual patient data (IPD) from randomised controlled trials (RCTs) and prospective cohort studies. Based on these analyses, we aimed to propose an optimal linezolid dosing strategy that maximises the probability of treatment success while minimising adverse events.
Methods
Study design and methods
This IPD meta-analysis is reported in accordance with the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) of IPD statement [12]. It was conducted after exemption from the need for ethical approval by the Institutional Review Board of Seoul National University Hospital (number 2308-038-1457). The study protocol was registered in the PROSPERO database (identifier: CRD42023451843).
Search strategy and selection criteria
We performed a review of the literature about trials on the use of short-course all-oral regimens containing linezolid to treat MDR/RR-TB across the PubMed, Embase and Scopus databases. The search query was “[(tuberculosis) AND (linezolid)] AND [(trial) OR (prospective)]” in all fields, with no language restrictions. Two independent investigators (NK and JYK) screened titles and abstracts for eligible studies, defined as RCTs and prospective cohort studies reporting the outcomes of short-course all-oral regimens containing linezolid for the treatment of MDR/RR-TB or extensively drug-resistant TB [13], published up to 31 August 2023. Papers reporting on drug-susceptible TB, regimens involving injectable drugs, regimens >12 months [14] and regimens without linezolid, as well as those describing preclinical and retrospective studies, were excluded. Conference papers, book chapters, reviews, editorials, guidelines, case reports and studies focused exclusively on paediatric or pregnant populations were also excluded. At the individual participant level, those with missing outcomes, later addition of linezolid after starting initial treatment, unknown dosage or duration of linezolid and extrapulmonary TB were excluded.
Data collection and quality assessment
We contacted the authors of reports on eligible studies or those with ownership of the data and asked them to participate in this study. If no response was obtained, two additional attempts were made to contact them. If agreement was obtained, anonymised data with the following variables were collected: age, sex, nationality, body mass index (BMI), HIV status, acid-fast bacilli (AFB) smear and culture results, presence of cavities, in vitro drug susceptibility, site of TB infection, comorbidities including diabetes mellitus, laboratory results, initial linezolid dose, companion drugs used, timing of linezolid adjustment(s) and subsequent dose, treatment duration, treatment outcomes and adverse events. For adverse events, we focused on grade 3 or higher events commonly associated with linezolid, including optic or peripheral neuropathy and myelosuppression, as assessed by clinicians in each study [15–17]. In general, grade 3 refers to severe events that result in hospitalisation or significant impairment of social and functional activities, while grade 4 indicates potentially life-threatening events [15–17]. In cases in which data were insufficient or further clarification was required, we communicated with the lead authors or those with ownership of the data. The data from all of the included studies were merged and transformed into a common dataset. The risk of bias in individual studies was assessed using the Cochrane risk-of-bias tool (RoB 2) for RCTs [18] and ROBINS-I for prospective cohort studies [19].
Treatment outcomes, grouping and statistical analysis
We adhered to the revised WHO definitions for treatment outcomes [20]. Treatment success was determined by summing the cases classified as cured and those that completed treatment. We employed a competing risk analysis method that accounts for patients no longer at risk of experiencing the primary event [21]. Treatment failure and death during treatment were considered competing events that precluded the occurrence of the primary event, namely, treatment success. Patients who were lost to follow-up or who were otherwise unsuccessful were censored. For analysis of the risk of adverse events, death before the occurrence of such events was considered a competing event.
Patients were grouped by clustering using the partitioning around medoids algorithm [22] based on their linezolid usage patterns, incorporating days of use and cumulative dose. The optimal number of clusters was determined using the elbow method [23]. Demographic and clinical characteristics, companion drugs, treatment duration, outcomes and adverse events were summarised as proportions for categorical variables and as medians with interquartile ranges (IQRs) for continuous variables. Differences in the proportion and distribution of variables across cluster groups were assessed using ANOVA, the Kruskal–Wallis test, Pearson's chi-square test or Fisher's exact test, as appropriate.
The sub-distribution hazard ratio (SHR) of the event of interest for each group, which reflects the relative instantaneous rate (probability per unit time) of the event of interest given that the event has not occurred before [24], was estimated using the Fine–Gray sub-distribution hazard model, accounting for the presence of competing risks. The reference group was set by the authors’ consensus as the one that best reflected the actual linezolid usage pattern in clinical practice [11, 25, 26]. We also accounted for the clustering effect per study in the pooled analysis of random effects and the robust variance [27]. In the multivariable model, we adjusted for age, sex, BMI, presence of cavities and AFB smear positivity. Covariates for which >10% of the values were missing were not included in the analysis because of the possibility of bias [28]. A two-tailed significance of p<0.05 was considered statistically significant. All analyses were conducted using R software (version 4.4.0; R Foundation for Statistical Computing, Vienna, Austria).
Results
We identified 5353 records using the predefined keywords. Of these, 1631 articles were assessed for eligibility and 12 met the inclusion criteria. The PRISMA IPD flow diagram is depicted in figure 1. We contacted the corresponding authors of the papers on these studies or those with ownership of the data. Of the 12 included studies, anonymised data were provided from eight studies, four RCTs [3, 5, 6, 29, 30] and four prospective cohort studies [4, 31–33]. The risk of bias in these studies is provided in supplementary figure S1, and all studies had a low risk of bias. In these eight studies, 965 patients were treated with a short-course all-oral regimen that included linezolid. After excluding 20 patients (11 because of missing outcomes, six because linezolid was added to the regimen at a later stage, two with unknown linezolid dosage and duration, and one with extrapulmonary TB), 945 patients were included in the final analysis. The clinical characteristics of these patients and the design of each study are summarised in supplementary table S1.
FIGURE 1.
Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) individual patient data (IPD) flow diagram. TB: tuberculosis.
Based on the clinical contexts and the within-cluster sum of squares (supplementary figure S2), the linezolid usage patterns were classified into four groups. The partitioning around medoids clustering results are provided in supplementary figure S3 as a scatter plot. Group 1 comprised 215 patients who received 600 mg·day−1 of linezolid for 8 weeks. Group 2 comprised 447 patients who received 600 mg·day−1 of linezolid for 16 weeks, followed by 300 mg·day−1 of linezolid for 8 weeks. Group 3 comprised 183 patients who received 600 mg·day−1 of linezolid for a median of 39 weeks. Group 4 comprised 100 patients who received 1200 mg·day−1 of linezolid for 25 weeks (table 1). The allocation of patients into groups for each study is presented in supplementary table S2.
TABLE 1.
Clustering of linezolid use patterns by days of linezolid use and cumulative dose
| Total | Group 1 | Group 2 | Group 3 | Group 4 | p-value# | |
|---|---|---|---|---|---|---|
| Subjects (n) | 945 | 215 | 447 | 183 | 100 | |
| Total treatment duration (days) | 181 (167–236) | 181 (108–270) | 167 (167–175) | 272 (184–281) | 181 (181–186) | <0.001 |
| Duration of linezolid use (days) | ||||||
| Total | 166 (126–181) | 63 (57–63) | 166 (166–167) | 272 (182–276) | 181 (148–181) | <0.001 |
| 1200 mg·day−1 | 0 | 0 (0–18) | 0 | 0 | 178 (116–181) | <0.001 |
| 600 mg·day−1 | 111 (55–112) | 57 (7–63) | 111 (109–111) | 272 (181–276) | 0 (0–9) | <0.001 |
| 300 mg·day−1 | 0 (0–55) | 0 | 55 (55–56) | 0 | 0 | <0.001 |
| Cumulative dose of linezolid (mg) | 83 100 (75 600–127 200) | 37 800 (36 000–63 000) | 83 100 (83 100–83 700) | 163 200 (111 000–166 200) | 181 000 (161 700–200 100) | <0.001 |
Data are presented as median (interquartile range), unless otherwise indicated. #: indicates the overall difference among the four groups.
The median age of the 945 patients was 35 years (IQR 29–45 years), with a predominance of male patients (592 patients, 62.7%). The prevalence rates of HIV infection, the presence of cavities and resistance to fluoroquinolones were higher in groups 1 and 4 than in the other two groups (table 2).
TABLE 2.
Demographic and clinical characteristics of the patients
| Total | Group 1 | Group 2 | Group 3 | Group 4 | p-value# | |
|---|---|---|---|---|---|---|
| Subjects (n) | 945 | 215 | 447 | 183 | 100 | |
| Age (years) | 35 (29–45) | 35 (29–43) | 36 (28–46) | 38 (30–50) | 35 (29–44) | 0.108 |
| Male sex | 592 (62.7) | 141 (65.6) | 268 (60.0) | 124 (67.8) | 59 (59.0) | 0.188 |
| Weight (kg) | 57 (50–66) | 57 (51–66) | 57 (50–66) | 55 (50–64) | 61 (51–67) | 0.465 |
| BMI (kg·m−2) | 19.7 (17.9–22.4) | 20.0 (17.9–22.8) | 19.8 (18.0–22.3) | 19.1 (17.7–21.6) | 20.3 (17.9–23.2) | 0.085 |
| HIV-positive | 218/741 (29.4) | 54/122 (44.3) | 114/435 (26.2) | 19/121 (15.7) | 31/63 (49.2) | <0.001 |
| Diabetes | 45/769 (5.9) | 8/154 (5.2) | 29/445 (6.5) | 1/70 (1.4) | 7/100 (7.0) | 0.349 |
| Acid-fast bacilli smear positivity | 524/900 (58.2) | 114/211 (54.0) | 266/437 (60.9) | 96/179 (53.6) | 48/73 (65.8) | 0.111 |
| Presence of cavity on chest radiography | 579/929 (62.3) | 135/209 (64.6) | 269/444 (60.6) | 99/176 (56.3) | 76/100 (76.0) | 0.007 |
| Resistance to fluoroquinolone | 256/788 (32.5) | 99/169 (58.6) | 82/379 (21.6) | 37/169 (21.9) | 38/71 (53.5) | <0.001 |
Data are presented as median (interquartile range), n (%) or n/N (%), unless otherwise indicated. BMI: body mass index. #: indicates the overall difference among the four groups.
Overall, 873 patients (92.4%) received bedaquiline, most frequently in groups 3 and 4 (100%). Delamanid was used by 92 patients (9.7%), mainly in group 2 (13.4%). Pretomanid was used by 650 patients (68.8%), with the highest usage in groups 2 (84.7%) and 4 (100%). Fluoroquinolones were used by 412 patients (43.6%) and clofazimine by 290 patients (30.7%), most commonly in group 3 (table 3).
TABLE 3.
Antituberculous drugs used among the patients
| Total | Group 1 | Group 2 | Group 3 | Group 4 | p-value# | |
|---|---|---|---|---|---|---|
| Subjects (n) | 945 | 215 | 447 | 183 | 100 | |
| Bedaquiline | 873 (92.4) | 203 (94.4) | 387 (86.6) | 183 (100.0) | 100 (100.0) | <0.001 |
| Delamanid | 92 (9.7) | 18 (8.4) | 60 (13.4) | 14 (7.7) | 0 | <0.001 |
| Pretomanid | 650 (68.8) | 130 (60.5) | 374 (84.7) | 46 (25.1) | 100 (100.0) | <0.001 |
| Later-generation fluoroquinolone | 412 (43.6) | 89 (41.4) | 200 (44.7) | 123 (67.2) | 0 | <0.001 |
| Levofloxacin | 275 (29.1) | 79 (36.7) | 73 (16.3) | 123 (67.2) | 0 | <0.001 |
| Moxifloxacin | 137 (14.5) | 10 (4.7) | 127 (28.4) | 0 | 0 | <0.001 |
| Clofazimine | 290 (30.7) | 70 (32.6) | 107 (23.9) | 113 (61.8) | 0 | <0.001 |
| High-dose isoniazid | 84 (8.9) | 60 (27.9) | 6 (1.3) | 18 (9.8) | 0 | <0.001 |
Data are presented as n (%), unless otherwise indicated. #: indicates the overall difference among the four groups.
Among the 945 patients, 794 (84.0%) achieved treatment success. The proportion of treatment success was lowest in group 1 (59.1%) but exceeded 90% in groups 2–4. Group 1 also had the highest proportions of treatment failure, death and loss to follow-up. Of the 215 patients in group 1, 34 (15.8%) experienced treatment failure, while only 13 patients across groups 2–4 did so (table 4).
TABLE 4.
Treatment outcomes according to group
| Total | Group 1 | Group 2 | Group 3 | Group 4 | p-value# | |
|---|---|---|---|---|---|---|
| Subjects (n) | 945 | 215 | 447 | 183 | 100 | |
| Cured or treatment completed (n) | 794 | 127 | 404 | 167 | 96 | <0.001 |
| % (95% CI) | 84.0 (81.7–86.4) | 59.1 (52.5–65.6) | 90.4 (87.6–93.1) | 91.3 (87.2–95.3) | 96.0 (92.2–99.8) | |
| Treatment failed (n) | 47 | 34 | 5 | 6 | 2 | |
| % (95% CI) | 5.0 (3.6–6.4) | 15.8 (10.9–20.7) | 1.1 (0.1–2.1) | 3.3 (0.7–5.9) | 2.0 (0–4.7) | |
| Died (n) | 21 | 17 | 1 | 3 | 0 | |
| % (95% CI) | 2.2 (1.3–3.2) | 7.9 (4.3–11.5) | 0.2 (0–0.7) | 1.6 (0–3.5) | ||
| Lost to follow-up (n) | 26 | 15 | 6 | 5 | 0 | |
| % (95% CI) | 2.8 (1.7–3.8) | 7.0 (3.6–10.4) | 1.3 (0.3–2.4) | 2.7 (0.4–5.1) | ||
| Unevaluated or unassigned outcome (n) | 57 | 22 | 31 | 2 | 2 | |
| % (95% CI) | 6.0 (4.5–7.5) | 10.2 (6.2–14.3) | 7.0 (4.6–9.3) | 1.1 (0–2.6) | 2.0 (0–4.7) |
CI: confidence interval. #: indicates the overall difference among the four groups.
The impact of each variable on treatment success is presented in table 5. In the adjusted analysis, age (adjusted SHR 0.98, 95% CI 0.97–0.99) was negatively associated with treatment success, whereas BMI (adjusted SHR 1.03, 95% CI 1.01–1.06) was positively associated with it. However, neither the presence of cavities nor a positive AFB smear at enrolment was significantly associated with treatment success.
TABLE 5.
SHRs for treatment success
| Crude SHR (95% CI) | Adjusted SHR (95% CI) | |
|---|---|---|
| Age (years) | 0.99 (0.98–0.10) | 0.98 (0.97–0.99) |
| Male sex | 0.86 (0.82–0.90) | 0.89 (0.79–1.01) |
| BMI (kg·m−2) | 1.03 (1.00–1.06) | 1.03 (1.01–1.06) |
| Cavity | 1.03 (0.86–1.24) | 1.08 (0.98–1.19) |
| Acid-fast bacilli smear positivity | 1.06 (0.90–1.25) | 1.14 (0.92–1.42) |
| Linezolid use pattern | ||
| Group 1 | 0.28 (0.08–0.96) | 0.24 (0.08–0.71) |
| Group 2 | Reference | Reference |
| Group 3 | 0.36 (0.13–0.96) | 0.36 (0.16–0.81) |
| Group 4 | 0.70 (0.22–2.20) | 0.57 (0.23–1.43) |
BMI: body mass index; CI: confidence interval; SHR: sub-distribution hazard ratio.
The relative treatment success rate varied according to linezolid dosing. Compared with group 2 (the reference group), group 1 (adjusted SHR 0.24, 95% CI 0.08–0.71) and group 3 (adjusted SHR 0.36, 95% CI 0.16–0.81) exhibited significantly lower rates of treatment success. In contrast, the rate of treatment success in group 4 (adjusted SHR 0.57, 95% CI 0.23–1.43) did not significantly differ from that in group 2. The cumulative incidence function plot for treatment success is provided in figure 2.
FIGURE 2.
Cumulative incidence function plot for treatment success.
During the treatment period, 235 patients (24.9%) experienced adverse events of grade 3 or higher. This proportion was highest in group 4 (43.0%) and lowest in group 3 (13.7%). Peripheral neuropathy occurred most frequently in group 4 (15.0%), and myelosuppression developed most frequently in group 1 (9.8%), followed by group 2 (4.7%). A total of 11 of the 21 patients who experienced myelosuppression in group 1 discontinued linezolid earlier than scheduled. Optic neuropathy was reported in only two patients, both of whom were in group 4, in which the highest daily dose of linezolid (1200 mg) was used (table 6). Compared with group 2, group 1 had a higher rate of adverse events (adjusted SHR 1.84, 95% CI 0.75–4.50), while group 3 showed a lower rate (adjusted SHR 0.55, 95% CI 0.18–1.67), although these differences did not reach statistical significance. Group 4, however, had a significantly higher rate of adverse events than group 2 (adjusted SHR 2.29, 95% CI 1.37–3.83) (table 7).
TABLE 6.
Adverse events of grade 3 or higher by group
| Total | Group 1 | Group 2 | Group 3 | Group 4 | p-value# | |
|---|---|---|---|---|---|---|
| Subjects (n) | 945 | 215 | 447 | 183 | 100 | |
| Total (n) | 235 | 66 | 101 | 25 | 43 | <0.001 |
| % (95% CI) | 24.9 (22.1–27.6) | 30.7 (24.5–36.9) | 22.6 (18.7–26.5) | 13.7 (8.7–18.6) | 43.0 (33.3–52.7) | |
| Peripheral neuropathy (n) | 33 | 4 | 8 | 6 | 15 | <0.001 |
| % (95% CI) | 3.5 (2.3–4.7) | 1.9 (0.1–3.7) | 1.8 (0.5–3.0) | 3.3 (0.7–5.9) | 15.0 (8.0–22.0) | |
| Myelosuppression (n) | 47 | 21 | 21 | 2 | 3 | <0.001 |
| % (95% CI) | 5.0 (3.6–6.4) | 9.8 (5.8–13.7) | 4.7 (2.7–6.7) | 1.1 (0–2.6) | 3.0 (0–6.3) | |
| Optic neuropathy (n) | 2 | 0 | 0 | 0 | 2 | 0.011 |
| % (95% CI) | 2.1 (0–0.5) | 2.0 (0–4.7) |
CI: confidence interval. #: indicates the overall difference among the four groups.
TABLE 7.
SHRs for adverse events of grade 3 or higher
| Crude SHR (95% CI) | Adjusted SHR (95% CI) | |
|---|---|---|
| Age (years) | 1.01 (0.99–1.03) | 1.02 (1.00–1.04) |
| Male sex | 0.87 (0.73–1.04) | 0.84 (0.73–0.97) |
| BMI (kg·m−2) | 1.00 (0.97–1.03) | 0.96 (0.94–0.99) |
| Cavity | 0.91 (0.51–1.63) | 0.69 (0.49–0.98) |
| Acid-fast bacilli smear positivity | 0.82 (0.53–1.28) | 0.91 (0.57–1.45) |
| Linezolid use pattern | ||
| Group 1 | 1.60 (0.61–4.21) | 1.84 (0.75–4.50) |
| Group 2 | Reference | Reference |
| Group 3 | 0.51 (0.16–1.65) | 0.55 (0.18–1.67) |
| Group 4 | 2.48 (1.28–4.82) | 2.29 (1.37–3.83) |
BMI: body mass index; CI: confidence interval; SHR: sub-distribution hazard ratio.
Discussion
We conducted an IPD meta-analysis using data from patients who underwent short-course linezolid-containing oral treatment for MDR/RR-TB to determine the optimal dose and duration of linezolid treatment. Our analyses revealed that a dose-tapering strategy, starting with linezolid at 600 mg·day−1 (for 16 weeks) followed by a reduction to 300 mg·day−1 (for 8 weeks), was potentially the most optimal in terms of effectiveness and safety.
Although linezolid is one of the pivotal drugs for short-course all-oral regimens, its dosage and duration of use vary considerably, even in the clinical trials [3, 4, 6, 29, 30]. In our study, a considerable proportion of patients experienced dose reductions, temporary interruptions or discontinuation of linezolid earlier than scheduled. To account for this variability, we categorised linezolid usage patterns based on the actual dose and duration administered at the individual patient level rather than adhering to the original intent of the protocol. For example, among 109 patients in the Nix-TB study [4] who were scheduled to receive linezolid at 1200 mg·day−1 for 26 weeks, 30 were allocated to group 2 and 11 to group 1. This allocation was influenced not only by the original intent of the study but also by adverse events, death and patient response to treatment. For instance, group 1 included patients who were originally intended to receive linezolid for only 8 weeks, as well as those who discontinued linezolid prematurely due to serious adverse events or death. Censoring patients whose linezolid schedules were altered due to treatment failure or death could introduce bias and overestimate the effectiveness of short-term linezolid treatment on treatment success [24]. To address this issue, we employed the Fine–Gray sub-distribution hazard model, which accounts for competing events, such as treatment failure or death, that preclude the achievement of treatment success [24].
The proportion of treatment success was comparable between groups 2 and 3. However, when evaluating the probability of treatment success over time using the SHR, group 2 demonstrated more favourable outcomes than group 3 after adjusting for age, sex, BMI, presence of cavities and AFB smear positivity. The higher rate of treatment success in group 2 can be explained by two factors. First, the shorter treatment duration in group 2 may have contributed to its higher success rate compared to group 3. Shortening the treatment duration has been shown to improve treatment outcomes in MDR/RR-TB patients by reducing loss to follow-up [34]. Second, the relatively higher number of competing events in group 3 likely mitigated the overestimation of its impact on treatment success, as illustrated in figure 2. This adjustment revealed a clearer distinction between the outcomes of group 2 and group 3.
The high effectiveness of the tapering regimen in group 2 can be explained through a two-stage approach. Initially, a front-loaded 600 mg·day−1 dose of linezolid enhances bactericidal activity during the early stages of treatment [35, 36]. Subsequently, reducing the dose to 300 mg·day−1 maintains the sterilising effect while minimising the risk of adverse events [37]. In the hollow fibre model, a 300 mg·day−1 dose of linezolid alone was sufficient to achieve a sterilising effect in approximately 90% of the population [37]. Our study highlights that a two-stage approach, which balances bactericidal activity, sterilising efficacy and adverse events, is an effective strategy in clinical settings. This finding is further supported by a recent clinical trial conducted in India, which demonstrated higher treatment success with a dose-tapering strategy of linezolid beginning at 600 mg·day−1 and tapering to 300 mg·day−1 [25].
In this study, adjustments for the effects of companion drugs were not performed because the timing and duration of companion drug use varied widely among patients. Because we focused on the dosing and duration of linezolid administration over time, adjusting for the timing and duration of companion drug administration would have complicated the determination of optimal linezolid dosing and duration. To mitigate these limitations, at the trial level, we adjusted for the effects of individual studies to average the influence of diverse regimens. At the individual patient level, we also adjusted for age, sex, BMI and disease severity.
Grade 3 or higher adverse events were more frequent in the group receiving 1200 mg·day−1 of linezolid. This finding aligns with the characteristics of peripheral neuropathy, a common adverse effect in groups with higher cumulative doses [38]. In contrast, myelosuppression was more prevalent in group 1. Unlike peripheral neuropathy, which typically occurs between 3 and 6 months after treatment initiation [39], myelosuppression is known to occur earlier in the treatment course [39]. This suggests that some patients in group 1 may have discontinued linezolid earlier than scheduled due to the rapid onset of myelosuppression. Indeed, of the 21 patients in group 1 who experienced myelosuppression, 11 discontinued linezolid immediately after its onset, although it remains unclear whether myelosuppression directly caused the discontinuation.
This study has several limitations. Owing to the lack of availability of such data from most studies, we could not incorporate the minimal inhibitory concentration of linezolid against clinical isolates or patients’ pharmacokinetic data, despite these variables potentially impacting treatment outcomes. Second, we were unable to confirm whether tapering linezolid at another timepoint, rather than at 16 weeks, would still yield comparable effects to maintaining the 600 mg dose [25]. Third, the unavailability of four studies and the uneven distribution of patients from studies across our groups may have introduced unintended bias. Fourth, no adjustments were made for the use of companion drugs, and factors such as drug resistance pattern, alcohol abuse, diabetes or HIV infection were not adjusted for due to the high proportion of missing data.
Despite these limitations, this study obtained robust findings by collecting individual patient-level data from RCTs and prospective studies with clearly defined treatment protocols. It expands on previous IPD meta-analyses [38] by including a larger cohort of patients with diverse background regimens, thereby not only examining adverse events but also proposing an optimal dose and duration for optimal treatment outcomes. The inclusion of nearly 1000 patients across diverse geographic settings support the generalisability of our findings.
In conclusion, given the high treatment success rate and acceptable risk of adverse events, we propose a dose-tapering strategy for linezolid involving 600 mg·day−1 for the first 16 weeks followed by 300 mg·day−1 for the next 8 weeks as the potentially optimal approach for the use of linezolid in a short-course all-oral regimen.
Shareable PDF
Footnotes
Ethics approval: This IPD meta-analysis is reported in adherence to the PRISMA of IPD statement. It was conducted after exemption from ethical approval by the Institutional Review Board of Seoul National University Hospital (approval number 2308-038-1457).
This article has an editorial commentary: https://doi.org/10.1183/13993003.00927-2025
Conflict of interest: M. Beumont and S. Foraida are employees of TB Alliance. R.A. Murphy reports payment for expert testimony and support for attending meetings from the University of Nevada. J-J. Yim has participated or is currently participating as the overall or institutional principal investigator in multiple clinical trials sponsored by Insmed Incorporated, AN2 Therapeutics and LigaChem Biosciences Inc. The remaining authors have no potential conflicts of interest to disclose.
Supplementary material
Please note: supplementary material is not edited by the Editorial Office, and is uploaded as it has been supplied by the author.
Supplementary material
ERJ-00315-2025.SUPPLEMENT
References
- 1.World Health Organization . Global Tuberculosis Report 2024. Geneva, World Health Organization, 2024. [Google Scholar]
- 2.Reuter A, Tisile P, Von Delft D, et al. The devil we know: is the use of injectable agents for the treatment of MDR-TB justified? Int J Tuberc Lung Dis 2017; 21: 1114–1126. doi: 10.5588/ijtld.17.0468 [DOI] [PubMed] [Google Scholar]
- 3.Conradie F, Bagdasaryan TR, Borisov S, et al. Bedaquiline–pretomanid–linezolid regimens for drug-resistant tuberculosis. N Engl J Med 2022; 387: 810–823. doi: 10.1056/NEJMoa2119430 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Conradie F, Diacon AH, Ngubane N, et al. Treatment of highly drug-resistant pulmonary tuberculosis. N Engl J Med 2020; 382: 893–902. doi: 10.1056/NEJMoa1901814 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Nyang'wa B-T, Berry C, Kazounis E, et al. Short oral regimens for pulmonary rifampicin-resistant tuberculosis (TB-PRACTECAL): an open-label, randomised, controlled, phase 2B-3, multi-arm, multicentre, non-inferiority trial. Lancet Respir Med 2024; 12: 117–128. doi: 10.1016/S2213-2600(23)00389-2 [DOI] [PubMed] [Google Scholar]
- 6.Mok J, Lee M, Kim DK, et al. 9 months of delamanid, linezolid, levofloxacin, and pyrazinamide versus conventional therapy for treatment of fluoroquinolone-sensitive multidrug-resistant tuberculosis (MDR-END): a multicentre, randomised, open-label phase 2/3 non-inferiority trial in South Korea. Lancet 2022; 400: 1522–1530. doi: 10.1016/S0140-6736(22)01883-9 [DOI] [PubMed] [Google Scholar]
- 7.Ahmad N, Ahuja SD, Akkerman OW, et al. Treatment correlates of successful outcomes in pulmonary multidrug-resistant tuberculosis: an individual patient data meta-analysis. Lancet 2018; 392: 821–834. doi: 10.1016/S0140-6736(18)31644-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Cox H, Ford N. Linezolid for the treatment of complicated drug-resistant tuberculosis: a systematic review and meta-analysis. Int J Tuberc Lung Dis 2012; 16: 447–454. doi: 10.5588/ijtld.11.0451 [DOI] [PubMed] [Google Scholar]
- 9.Sotgiu G, Centis R, D'Ambrosio L, et al. Efficacy, safety and tolerability of linezolid containing regimens in treating MDR-TB and XDR-TB: systematic review and meta-analysis. Eur Respir J 2012; 40: 1430–1442. doi: 10.1183/09031936.00022912 [DOI] [PubMed] [Google Scholar]
- 10.Song T, Lee M, Jeon HS, et al. Linezolid trough concentrations correlate with mitochondrial toxicity-related adverse events in the treatment of chronic extensively drug-resistant tuberculosis. EBioMedicine 2015; 2: 1627–1633. doi: 10.1016/j.ebiom.2015.09.051 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Kwon YS, Jang JG, Lee J, et al. Risk factors for peripheral neuropathy in patients on linezolid-containing regimens for drug-resistant TB. Int J Tuberc Lung Dis 2023; 27: 232–234. doi: 10.5588/ijtld.22.0423 [DOI] [PubMed] [Google Scholar]
- 12.Stewart LA, Clarke M, Rovers M, et al. preferred reporting items for systematic review and meta-analyses of individual participant data: the PRISMA-IPD statement. JAMA 2015; 313: 1657–1665. doi: 10.1001/jama.2015.3656 [DOI] [PubMed] [Google Scholar]
- 13.World Health Organization . WHO Consolidated Guidelines on Tuberculosis. Module 4: Treatment-Drug-Resistant Tuberculosis Treatment, 2022 Update. Geneva, World Health Organization, 2022. [PubMed] [Google Scholar]
- 14.World Health Organization . WHO Consolidated Guidelines on Tuberculosis. Module 4: Treatment-Drug-Resistant Tuberculosis Treatment. Geneva, World Health Organization, 2020. [PubMed] [Google Scholar]
- 15.Department of Health and Human Services . Division of AIDS (DAIDS) Table for Grading the Severity of Adult and Pediatric Adverse Events: Corrected Version 2.1. Bethesda, MD, National Institutes of Health, 2017. [Google Scholar]
- 16.Médecins Sans Frontières . PV-TB-D12 TB Severity Grading Scale. Available from: https://endtb.org/toolkit/endtb-trials-pharmacovigilance/
- 17.National Institutes of Health . Common Terminology Criteria for Adverse Events (CTCAE). 2010. Available from: https://ctep.cancer.gov/protocoldevelopment/electronic_applications/ctc.htm#ctc_40 Date last accessed: 21 August 2022.
- 18.Sterne JAC, Savović J, Page MJ, et al. RoB 2: a revised tool for assessing risk of bias in randomised trials. BMJ 2019; 366: l4898. doi: 10.1136/bmj.l4898 [DOI] [PubMed] [Google Scholar]
- 19.Sterne JA, Hernán MA, Reeves BC, et al. ROBINS-I: a tool for assessing risk of bias in non-randomised studies of interventions. BMJ 2016; 355: i4919. doi: 10.1136/bmj.i4919 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.World Health Organization . Meeting Report of the WHO Expert Consultation on Drug-Resistant Tuberculosis Treatment Outcome Definitions. Geneva, World Health Organization, 2021. Available from: https://www.who.int/publications/i/item/9789240022195 [Google Scholar]
- 21.Fine JP, Gray RJ. A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc 1999; 94: 496–509. doi: 10.1080/01621459.1999.10474144 [DOI] [Google Scholar]
- 22.Kaufman L, Rousseeuw PJ. Finding Groups in Data: An Introduction to Cluster Analysis. Hoboken, NJ, John Wiley & Sons, 1990. [Google Scholar]
- 23.Nainggolan R, Perangin-Angin R, Simarmata E, et al. Improved the performance of the k-means cluster using the sum of squared error (SSE) optimized by using the elbow method. J Phys Conf Ser 2019; 1361: 012015. doi: 10.1088/1742-6596/1361/1/012015 [DOI] [Google Scholar]
- 24.Coemans M, Verbeke G, Döhler B, et al. Bias by censoring for competing events in survival analysis. BMJ 2022; 378: e071349. [DOI] [PubMed] [Google Scholar]
- 25.Padmapriyadarsini C, Oswal VS, Jain CD, et al. Effectiveness and safety of varying doses of linezolid with bedaquiline and pretomanid in treatment of drug-resistant pulmonary tuberculosis: open-label, randomized clinical trial. Clin Infect Dis 2024; 79: 1375–1385. doi: 10.1093/cid/ciae388 [DOI] [PubMed] [Google Scholar]
- 26.Wasserman S, Brust JC, Abdelwahab MT, et al. Linezolid toxicity in patients with drug-resistant tuberculosis: a prospective cohort study. J Antimicrob Chemother 2022; 77: 1146–1154. doi: 10.1093/jac/dkac019 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Zhou B, Fine J, Latouche A, et al. Competing risks regression for clustered data. Biostatistics 2012; 13: 371–383. doi: 10.1093/biostatistics/kxr032 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Bennett DA. How can I deal with missing data in my study? Aust NZ J Public Health 2001; 25: 464–469. doi: 10.1111/j.1467-842X.2001.tb00294.x [DOI] [PubMed] [Google Scholar]
- 29.Nyang'wa B-T, Berry C, Kazounis E, et al. A 24-week, all-oral regimen for rifampin-resistant tuberculosis. N Engl J Med 2022; 387: 2331–2343. doi: 10.1056/NEJMoa2117166 [DOI] [PubMed] [Google Scholar]
- 30.Esmail A, Oelofse S, Lombard C, et al. An all-oral 6-month regimen for multidrug-resistant tuberculosis: a multicenter, randomized controlled clinical trial (the NExT study). Am J Respir Crit Care Med 2022; 205: 1214–1227. doi: 10.1164/rccm.202107-1779OC [DOI] [PubMed] [Google Scholar]
- 31.Govender T, Jham MA, Zhang JC, et al. Decentralized, integrated treatment of RR/MDR-TB and HIV using a bedaquiline-based, short-course regimen is effective and associated with improved HIV disease control. J Acquir Immune Defic Syndr 2023; 92: 385–392. doi: 10.1097/QAI.0000000000003150 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Fadeyi MO, Decroo T, Ortuño-Gutiérrez N, et al. A four-drug standardized short regimen for highly resistant TB in South-West Nigeria. Int Health 2024; 16: 123–125. doi: 10.1093/inthealth/ihad023 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Nguyen TMP, Le THM, Merle CSC, et al. Effectiveness and safety of bedaquiline-based, modified all-oral 9–11-month treatment regimen for rifampicin-resistant tuberculosis in Vietnam. Int J Infect Dis 2023; 126: 148–154. doi: 10.1016/j.ijid.2022.11.007 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Abidi S, Achar J, Assao Neino MM, et al. Standardised shorter regimens versus individualised longer regimens for rifampin- or multidrug-resistant tuberculosis. Eur Respir J 2020; 55: 1901467. doi: 10.1183/13993003.01467-2019 [DOI] [PubMed] [Google Scholar]
- 35.Tsuji BT, Bulitta JB, Brown T, et al. Pharmacodynamics of early, high-dose linezolid against vancomycin-resistant enterococci with elevated MICs and pre-existing genetic mutations. J Antimicrob Chemother 2012; 67: 2182–2190. doi: 10.1093/jac/dks201 [DOI] [PubMed] [Google Scholar]
- 36.Tsuji BT, Brown T, Parasrampuria R, et al. Front-loaded linezolid regimens result in increased killing and suppression of the accessory gene regulator system of Staphylococcus aureus. Antimicrob Agents Chemother 2012; 56: 3712–3719. doi: 10.1128/AAC.05453-11 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Srivastava S, Magombedze G, Koeuth T, et al. Linezolid dose that maximizes sterilizing effect while minimizing toxicity and resistance emergence for tuberculosis. Antimicrob Agents Chemother 2017; 61: e00751-17. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Hasan T, Medcalf E, Nyang'wa BT, et al. The safety and tolerability of linezolid in novel short-course regimens containing bedaquiline, pretomanid, and linezolid to treat rifampicin-resistant tuberculosis: an individual patient data meta-analysis. Clin Infect Dis 2024; 78: 730–741. doi: 10.1093/cid/ciad653 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Imperial MZ, Nedelman JR, Conradie F, et al. Proposed linezolid dosing strategies to minimize adverse events for treatment of extensively drug-resistant tuberculosis. Clin Infect Dis 2022; 74: 1736–1747. doi: 10.1093/cid/ciab699 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
This PDF extract can be shared freely online.
Shareable PDF
ERJ-00315-2025.Shareable
Please note: supplementary material is not edited by the Editorial Office, and is uploaded as it has been supplied by the author.
Supplementary material
ERJ-00315-2025.SUPPLEMENT