Introduction

Tuberculosis (TB) remains one of the major cause of global morbidity and mortality (1). One in three persons worldwide representing over 2 billion individuals  are known to be infected with Mycobacterium tuberculosis (2). In 2011 alone, 8.7 million new TB cases were estimated with 1.4 million recorded deaths. Approximately one million deaths were among HIV-negative persons and 430,000 among HIV-positive individuals (3). Moreover, TB has been estimated to cause close to 35 million deaths in the next two decades (2). Although, TB is present in every country, majority of cases occur in low income and middle income countries especially Sub-Saharan Africa and Asia (1). This high incidence of TB in these regions presents a huge threat to development as approximately 98% of TB related deaths occur among adults in the productive age groups (2). Nevertheless, according to the 2012 Global TB Report (3), the global target of halting and reversing TB as part of the millennium development goals (MDGs) has been achieved. This statement rests in the consistent decline in new cases over the years with decline rate of 2.2 between 2010 and 2011 (3). Moreover, as part of the MDGs, the global mortality rate reduction of TB aimed at 50% by 2015 is recorded to be on track as a result of the attainment of 41% reduction rate between 1990 and 2011 (3). In the midst of these positive developments, the increasing emergence of MDR-TB and XDR-TB in most parts of the world poses a threat to the success attained by TB care and control (4). According to the World Health Organisation (WHO), about 3.7% TB of all new cases of TB in the world accounts for MDR-TB with a higher incidence of 20%  in previously treated patients (4). Worldwide, an estimated 440,000 MDR-TB cases emerged in 2008 contributing to an estimated 150,000 deaths (5). China and India were estimated to have contributed about 50% of these MDR-TB cases (5). China alone has recently reported 100,000 MDR-TB cases emerging annually. Also, most countries in Eastern Europe and central Asia have reported MDR-TB cases to account for 50% of previously treated TB patients (5). The cost of managing multi-drug resistant TB remains enormous. According to the Stop TB Partnership’s Global Plan to Stop TB, 2011-2015 (6), an estimation of over 2 million cases of MDR-TB will emerge between 2010 and 2015 if there is no significant increase in TB funding and political commitment. In the same Global Plan, 2011-2015, approximately US $1.7 billion is required towards MDR-TB with an annual increase of US $0.9 billion in 2011 to US $1.9 billion in 2015 (6). There have been many calls for greater efforts  at reducing the emergence of multi-drug resistance TB if the battle to reduce the global burden of TB is to be achieved (6, 7). The treatment options for drug resistant TB still remains limited as treatment is based on case series and expert opinions instead of randomised controlled trials (8). According to Alffenaar et al. (9), no new anti-tubercular drugs have been registered since the early 1980. Nonetheless, promising new drugs for MDR-TB and XDR-TB are in the pipeline of drug development. The Working Group on New Drugs (10), supported by the Stop TB partnership has been accelerating the drug development process of potentially new drugs for drug resistant TB since its establishment in 2001. Since 2006, Linezolid has been recommended for off-label treatment of drug resistant TB based on unclear efficacy resulting from case series and expert opinions (8). In vitro studies conducted by Alcala et al. (11), Guna et al. (12) and Yang et al. (13) have also concluded Linezolid’s effectiveness in both susceptible and drug resistant strains of Mycobacterium tuberculosis clinical isolates. Linezolid has been given at a dose of 600mg twice daily or lower in some case series reports which have hypothesised the effectiveness in MDR-TB and XDR-TB (14-16). In some instances, doses higher than 600mg daily have been given creating conflicting reports as to the most effective dose of the drug. Moreover, serious toxic side effects such as haematological adverse reactions and neuropathies have been reported with the use of Linezolid (17). To overcome the uncertainty surrounding Linezolid, we conducted a systematic review to summarize the available information towards answering three main questions (a) Are there substantial evidence on the efficacy and toxicity on Linezolid to inform its clinical use in MDR-TB and XDR-TB? (b) Is Linezolid potent enough to cause positive sputum culture conversion? (c) Are there severe or life-threatening toxicity associated with prolonged use of Linezolid in MDR-TB and XDR-TB patients?

Method

We conducted the review in accordance with the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) Statement (18) .

Search Strategy

A systematic search was conducted in PubMed, EMBASE, Web of Science and Clinicaltrials.gov. The keywords were deduced by the PICO (Patient population, Intervention, Comparison and Outcome respectively) approach (19). The research question was thus fragmented to identify the correlation with the PICO categories. Since the research question does not involve a comparison, the C component of PICO was exempted from the keyword search. The PICO approach was adopted for only PubMed, EMBASE and Web of Science. The fourteen keyword terms generated for the patient population search for  MDR-TB were multidrug resistant tuberculosis, multidrug resistant TB, multi-drug resistant tuberculosis, multi-drug resistant TB, multi-drug-resistant TB, multiple drug resistant tuberculosis, multiple drug resistant TB, multiple-drug resistant tuberculosis, multiple-drug resistant TB, drug resistant tuberculosis, drug resistant TB, MDRTB, MDR TB and MDR-TB. For XDR-TB, the following seven keywords were generated for the patient population search; extensively drug resistant tuberculosis, extensively drug resistant TB, extensively-drug resistant tuberculosis, extensively-drug resistant TB, extensively-drug-resistant TB, XDRTB, XDR-TB and XDR TB. For the intervention category, six keyword terms were employed; Linezolid, oxazolidinone antibacterial, oxazolidinone antibiotic, Zyvox, Zyvoxid and Zyvoxam. Zyvox is the trade name under which Linezolid is marketed in the United States of America and other countries. In Europe, it is marketed under the name Zyvoxid whiles in Canada and Mexico, Zyvoxam is used (20). Finally, the outcome category was subdivided under ‘efficacy’ and ‘toxicity’ which are the two main outcome measures in this study. Seven keyword terms were employed for ‘efficacy’; efficacy, effective, potency, effectiveness, effect, efficacious, potent and potency. Under ‘toxicity’, eight keyword terms were also generated as follows; toxicity, toxic effects, tolerability, safety, side effects, adverse effects, adverse events and adverse reactions. The Boolean operators ‘OR’ and ‘AND’ were used within and between each category respectively to combine the keywords. The results from the three databases; PubMed, EMBASE and Web of Science were not refined by any parameter with the exception of publication year between January, 2006 to December, 2013. Search through Clinicaltrials.gov was performed using one keyword each for drug intervention and disease. The keywords ‘Linezolid’ and ‘Tuberculosis’ were searched under intervention and disease respectively. PubMed, EMBASE and Web of Science were last searched on 11th May, 2013. Clinicaltrials.gov was last accessed on the 3rd of June, 2013 and 30th June, 2013 for the intervention (i.e. Linezolid) and disease (i.e. tuberculosis) respectively.

Study Selection

For a study to be included, it must involve an adult population ≥ 5 patients, sputum culture confirmed MDR TB/ XDR TB patients, DST report, data on outcome measures for efficacy or toxicity, original study, both pulmonary and extra pulmonary TB and exclusively English publications. The publication years was limited from 2006 to 2013 because this study aims to capture studies that have been conducted from the time WHO approved the use of Linezolid in MDR-TB and XDR-TB which was in 2006 (8). Studies were excluded for absence of sputum culture confirmed TB status and DST, animal and in vitro studies, case reports or series less than five subjects, unpublished literature, review articles and any language other than English. Case reports and case series were limited to five subjects to eliminate the influence of selection bias which is inherently associated with small sample size mostly employed by such study designs (21). Also the two recent systematic reviews performed with high quality which this study seeks to update employed inclusion sample size of ≥ 3 subjects (17) and ≥ 5 subjects (22).  Hence the higher sample size i.e. ≥ 5 subjects was applied to this study.

Outcome Measures

The efficacy outcomes were defined according to the definitions for TB treatment outcomes adapted by WHO (23). The primary outcome measure for efficacy was themed as treatment success from the definitions. Secondary efficacy outcome measures were defined according to the sputum culture conversion and completion of treatment themes. For the outcomes of toxicity, primary outcome measures included severe adverse events reported to be associated with Linezolid which are myelosuppression, optic and peripheral neuropathy (24). The secondary outcomes of toxicity were discontinuation of treatment and other reported adverse events.

Data Extraction

The data extraction was performed by the author (AAA). Variables to aid in descriptive characterisation of the selected studies included name of first author, year of publication, country of study, study design, presence or absence of control group, number of patients enrolled and study duration (Table 4). The following variables were extracted for efficacy and toxicity measures; number of subjects exposed to Linezolid, number of MDR-TB and XDR-TB patients in each study, HIV co-infection status, type of anti-TB regimen administered, dose of Linezolid received, treatment outcomes and reported adverse effects.

Assessment of study quality

The methodological quality was assessed by the author (AAA) based on protocol involved in the conduct of selected studies (i.e. Institutional Review Board (IRB) approval and informed consent), Linezolid dose indicated, type of treatment regimen indicated based on DST, Linezolid initiated under hospital environment, DOT applied to treatment monitoring and treatment outcome measures in conformity to WHO adapted definitions. In addition, the individual studies were critically appraised using the McMaster critical review for quantitative studies (25) which was modified and used by Deenadayalan et al. (26) [Appendix I]. The critical appraisal focused on individual study aims, study design, sampling method, outcome measurement, description of intervention, the quality of results analysis and reporting, the main findings, strengths and weaknesses.

Statistical Analysis

The results were analysed by meta-analysis proportions which were performed with MedCalC meta-analysis software version 12.7.0 (27). The analysis of efficacy measures was based on the number of patients who completed treatment. Hence culture conversion and cure outcomes were evaluated for only patients who completed treatment to evaluate efficacy on the common ground of treatment completion so that results will not be biased by patients who had culture conversion but later relapsed on follow up. For the analysis of toxicity, the proportions were determined based on the number of patients exposed to Linezolid but not limited to those who completed treatment. A further analysis was performed to compare proportions on efficacy and toxicity between doses ≤ 600mg and > 600mg. Statistical significance was set at p<0.05.

Results

A total of 1,007 search results were retrieved. Upon screening for duplicates and irrelevant papers based on titles and abstracts, 37 articles were found relevant for full-text analysis and reference list screening. Out of 37 articles, 20 articles were excluded with reasons. Hence, a total of 17 studies were included in this review (Figure 4). The seventeen included studies were all primary studies including 10 retrospective case series, 3 prospective case series, 1 phase 1 clinical trial, 2 phase 2a clinical trial and 1 retrospective cohort study. Majority (9 out of 17) of the studies was conducted in Asia, followed by the Americas (5 out of 17) and finally 3 out of 17 in Europe. No study was included which was performed in Africa. Only one multinational study was identified which was conducted across four countries in Europe (28). A total of 868 MDR-TB patients were enrolled in all the 17 studies including 201 reported XDR-TB cases. Of the 868 patients enrolled, 413 were exposed to Linezolid at once or twice daily dosing of at doses ≤ 600mg or ≥ 600mg.  The treatment duration for all the studies was in a range of 1month to 29months.  362 HIV-negative and 10 HIV-positive cases were reported from all the studies. Detailed epidemiological description of included studies is given in Table 1. Critical appraisal and methodological quality assessment scores are provided in Appendices 3 and 4 respectively.

Figure 1

PRISMA Flow diagram of review process

Table 1

Descriptive characteristics of studies included in the review

Analysis of Results

The effect of interest was proportion and individual study proportions were assessed at 95% confidence interval (CI) as well as the pooled effect. Test for heterogeneity was performed for all the proportions based on Cohran’s Q (Q) and degree of inconsistency (I2) (42). In all the summary or pooled analysis, random effect model was preferred to fixed effect model due to the presence of heterogeneity resulting from variations of effects from individual studies confirmed by I2 being greater than 0% and thus the random effect model takes into account the weighted average of all variations (42). General proportions concerning XDR-TB cases and completion of treatment were determined. Also, different indicators for efficacy and toxicity were analysed. With respect to efficacy, sputum culture conversion and treatment success were the main indicator measures. Toxicity was also evaluated based on four parameters namely; the occurrence of myelosuppression, neuropathy, discontinuation of Linezolid treatment and other adverse reactions. Furthermore, comparison of proportions between doses ≤ 600mg and > 600mg doses were tested for statistical difference in terms of the aforementioned efficacy and toxicity parameters.

 (a) Proportion of XDR-TB

Fourteen studies reported XDR-TB cases. The studies which did not report XDR-TB cases were conducted in USA (n=2) and Norway (n=1).  Of the studies which reported XDR-TB cases, Lee et al. (35) reported the highest number of 41 XDR-TB patients in South Korea. Another cohort in South Korea reported the least XDR-TB patients with a total of 4 patients (14). Five studies were exclusively XDR-TB cases conducted in Asia (n=3) and the Americas (n=2). Seven studies showed more than 50% proportion while three studies showed less than 50% proportion of XDR-TB patients. Wide confidence intervals were observed in all studies except Lee et al. (35) with a confidence interval of 91.4 to 100. High variability was observed which corresponds to I2 of 94.05% (95% CI of 91.58% to 95.80%).  A proportion of 72.5% was obtained for the pooled effect (95% CI 51.6%-89.2).

Figure 2

Meta-analysis proportion for XDR-TB

(b) Proportion of treatment Completion

All the studies reported some cases having completed treatment. 100% treatment completion was achieved in five studies whiles more than 50% but less than 100% and less than 50% were observed in twelve and two studies respectively.  In all five studies (14, 16, 31, 33, 39) with 100% treatment completion, a maximum dose 600mg Linezolid was given daily with the exception of Tang et al. (39) where 600mg twice daily was administered for the first two months followed by 600mg daily for the remaining treatment duration.  For the two studies (35, 36) achieving less than 50% completion of treatment, a maximum dose of 600mg daily was investigated. On the contrary, varying doses between 300mg daily to 600mg twice daily were administered in studies attaining more than 50% but less than 100% (15, 28-30, 32, 34, 37, 38, 40, 41) treatment completion. Wide confidence intervals were common to all the studies except four; Carroll et al. (31), Koh et al. (33), Koh et al. (34) and Udwadia et al. (43). For the pooled random effect, 79.421% treatment success was observed at a 95% confidence interval of 66.837% to 89.621%. The result of heterogeneity was also 88.2% for the degree of inconsistency.

Figure 3

Meta-analysis proportion for treatment completion

(c) Proportion of Sputum culture conversion

Sputum conversion was observed in all seventeen studies with respect to patients who completed treatment. 100% culture conversion was achieved in eight studies (15, 29, 30, 35, 36, 38-40). On the other hand, nine studies (14, 16, 28, 31-34, 37, 41) achieved more than 50% but less than 100% sputum culture conversion.  None of the studies achieved sputum culture conversion of less than 50%.The test for heterogeneity showed that minimum variability in individual response depicted by I2 value of 33.81% (95% CI= 0.00%-63.08%). The summary proportion was 91.090% (95% CI 86.7%-94.7%). The individual proportion had CI with the exception of Condos et al. (32) (80%, 95% CI=28.4% to 99.495%) and Park et al. (36) (100%, CI 2.5% -100.0%) with extremely wide CI.

Figure 4

Proportion of Sputum culture conversion

(d) Proportion of Treatment success

A proportion of patients showed treatment success in all the seventeen studies. Three studies (15, 30, 36) showed 100% treatment success. Koh et al.(33), Lee at al. (34), Schecter et al. (37) and  Singla et al. (38) reported treatment success of less than 50%. A total of ten studies (14, 16, 28, 29, 32, 34, 39-41) recorded more than 50% but less than 100% treatment success. Five studies were observed to have very wide CI; Koh et al. (33) (8.33%, 95% CI 1.026 to 26.997), Condos et al. (32) (60%, 95% CI 14.663%-94.726%), Lee et al.(35) (23.077%, 95% CI 5.038% to 53.813%), Park et al. (36) (100%, 95% CI 2.5% to 100%) and Schecter et al. (37) (4.545%, 95% CI 0.115%-22.844). The summary proportion also observed a wide 95% CI of 46.814% to 77.275% at a point estimate of 62.690%. High heterogeneity was observed at I2 of 87.19% (95% CI 81.01% to 91.35%) and Cohran’s Q of 122.6 at a p-value of < 0.0001.

Figure 5

Proportion of treatment success

(e) Proportion of Myelosuppression

Myelosuppression occurred in all the studies; however, only fourteen studies indicated the number of patients who suffered from it. Abbate et al. (29), Carroll et al. (31) and Villar et al. (40) are the three studies which reported the incidence of myelosuppression but did not indicate the number of patients. Myelosuppression occurred at the highest rate of 81.250% (95% CI 54.4-95.9) in the American cohort with a sample size of 16 patients (30). Anger et al. (30) investigated variable doses of linezolid i.e. 600mg twice daily, 400mg daily and 600mg daily for a mean duration of fifteen months. The least myelosuppression was observed in the South Korean cohort of 24 patients having a proportion of 4.167% (95% CI 0.1%-21.1%).  In this study, linezolid was administered at a dose of 300mg daily (n=17) and 600mg daily (n=7) for an interquartile range of 268 days to 443 days. Heterogeneity testing was high with I2 of 81.48% (95% CI 69.9% -88.6%). A pooled proportion of 32.9% (95% CI 21.8%-45.0%) was observed.

Figure 6

Proportion of Myelosuppression

(f) Proportion of Neuropathy

All studies reported the incidence of neuropathy with the exception of  Villar et al. (40) who reported occurrence but did not indicate the number of patients. All the 11 patients enrolled in the study conducted by Nam et al. (14) experienced neuropathy (100%, 95% CI 71.5%-100%) with administered dose of 300mg twice daily. The recorded range of treatment duration was 3.5 months to 24 months. Migliori et al. (28) with a large sample size of 85 patients recorded the least occurrence of neuropathy with a proportion of 3.529% (95% CI 0.7%-9.9%) for a mean treatment duration of 222 days. A dose of 600mg daily was administered to 28 patients whiles 600mg twice daily was administered to 57 patients. For the summary proportion, 40.938% point estimate with 95% CI of 24.4%-58.5% was obtained with a corresponding inconsistency of 92.10%.

Figure 7

Proportion of Neuropathy

(g) Proportion of linezolid treatment discontinuation

A total of twelve studies reported incidence of treatment discontinuation relating to linezolid. Carroll et al. (31) reported linezolid discontinuation but did not indicate number of patients corresponding to the occurrence. Five studies (16, 29, 32, 39, 40) had no patient being discontinued from linezolid treatment. Nam et al. (14) had the highest incident of linezolid discontinuation with a proportion of 72.7% (95% CI 39.0%-93.9%). Heterogeneity was present at an inconsistency value of 79.7% (95% CI 67% -87.3%). The pooled proportion also observed a proportion of 14.9% (95% CI: 7.796% -23.8%).

Figure 8

Proportion of linezolid treatment discontinuation

(h) Proportion of other adverse events

The absence or presence of other adverse events was reported in all the studies. However two studies (29, 31) did not indicate the number of patients associated with these other adverse events which were predominantly gastrointestinal disturbances, nausea and headache. Seven studies (14-16, 32-34, 36, 38) had no incidence of other adverse events apart from myelosuppression and neuropathy. Xu et al.(41) recorded the highest occurrence of gastrointestinal disturbances with a proportion of 83.3% (95% CI 58.58% to 96.42%). Variability was present at an inconsistency value of 89.06% (95% CI 83.66%-92.68%). The summary proportion yielded a proportion of 10.865% (95% CI 3.1%-22.3%).

Figure 9

Proportion of other adverse events

Discussion

Efficacy

In all seventeen studies, DST was performed to guide individualised treatment regimen. Linezolid was also not given as monotherapy but in combination with other anti-tubercular drugs. The indicators for efficacy as mentioned earlier were treatment success and sputum culture conversion with treatment success as the main indicator. The pooled proportion of patients who successfully completed respective treatment regimen was 79% (95% CI 67%-90%) in a total sample of 413 patients. Although, the range (1month to 29months) of treatment duration was mostly above the recommended duration of 28 days, this high proportion of treatment completion makes linezolid promising in TB management where by the treatment last for a recommended extended duration of 24 months (23).  A high culture conversion of 91% (CI 87% -95%) was obtained in the pooled result as against a pooled treatment success of 62% (CI 45% to 77%). The result of the culture conversion is comparable to the review conducted by Sotgui et al.(44); 93% (CI 87% to 97%). On the contrary, a higher  treatment success was obtain by Sotgui et al. (44) (82%, 95%CI 74%-88%) and Cox and Ford (17) (68%, 95%CI 58% to 78%). Nevertheless, similar observation of culture conversion proportion being higher than treatment success was made. This infers that, not all patients who attain sputum culture conversion at the completion of regimen attained the cured status upon follow up. This leads to another point of reasoning which may suggest that the duration of treatment correlates with the chances of attaining treatment success. For example, at the same dose of 600mg daily, Lee et al. (35) attained 23% treatment success for a duration of 18 months whiles Tang et al. (39) attained  79% treatment success for 24 months treatment duration. Thus there is the need to optimise treatment duration because in as much as short treatment duration could lead to treatment failure and resistance; too long treatment durations could possibly initiate resistance (45).  In this light, Horsburg et al. (45) has proposed a new method for antibiotic therapy which employs logistic regression model in randomised  clinical trials which links varying treatment duration to the corresponding cure proportions to determine the shortest effective treatment duration (45). According to the researchers, this model is most suited for anti-tubercular regimen targeted at decreasing resistance, minimising toxicity, reducing cost and decreasing antibiotic pill burden. Furthermore, the four cases of treatment failure which were observed by Lee et al. (35) and also found to be associated with linezolid draws the spotlight on further investigation into likely emergence of linezolid resistance. Comparison of dose effect on culture conversion and treatment success showed no significant difference for culture conversion while significant difference was observed for treatment success. For the treatment success, statistically significant difference of 33% was attained in favour of the doses greater than 600mg. This observation is contrary to Sotgui et al. (44) where no statistical significance was observed between lower and higher doses with respect to achieving curative outcome. This observation may be as a result of more studies identified by this study (n=17) compared with Sotgui et al. (44) (n=12); however, much larger clinical trials is warranted to ascertain the true effect to expand generalisation of results. Also, a recent pharmacokinetic study on linezolid 300mg twice daily in MDR-TB patients conducted by Bolhuis et al. (46) showed a synergistic effect with Clarithromycin 500mg resulting in an increased linezolid serum concentration  (median=44%, 95%CI p=0.043). Another pharmacokinetic study performed by Alffenaar et al. (9), investigated  linezolid 300mg twice daily for three days and obtained an AUC24/MIC of at least 100  and always higher than the minimum inhibitory concentration (MIC) in MDR-TB and XDR-TB patients concentration in 99% of MDR-TB and XDR-TB patients enrolled. Another interesting finding from a pharmacokinetic study (47) investigating the early and extended bactericidal action (EBA) of linezolid observed linezolid at 600mg twice daily (EBA= 0.17) shows a higher early bactericidal action than 600mg once daily (EBA= 0.13). Nonetheless, both doses showed reduced bactericidal action on the extended use; 600mg once daily (EBA=0.09) and 600mg twice daily (EBA=0.04). This may suggest a likelihood of attaining favourable concentration when lower dose of linezolid is augmented with 500mg daily Clarithromycin as inferred by Bolhuis et al. (46). Also, since linezolid was given as a combination therapy, the observations made in these studies cannot be exclusively associated with linezolid and thus Clarithromycin may be further investigated through randomised controlled trials for possible effective combination.

Toxicity

As mentioned earlier, the major adverse effects associated with linezolid are neuropathy and myelosuppression (24). In this review, anaemia, thrombocytopenia and leukopenia were recorded as myelosuppression. The pooled proportion of myelosuppression in this review was 33% (95% CI 22% -45%). This observation was consistent with Cox and Ford (17) (28%, 95% CI 15%-42%) and Sotgui et al. (44) (38% CI 28%-49%). When the incidence of myelosuppression is compared between doses ≤ 600mg and > 600mg, a statistically significant (p=0.0001) difference of 42% is observed in favour of doses > 600mg. Similar conclusion was also drawn by Sotgui et al. (44).  Nam et al. (14) specifically investigated the effect of dose reduction (300mg twice daily) on incidence of the major adverse events and concluded a significant reduction in myelosuppression but not neurotoxicity. Condos et al. (32) after investigating a dose of 600mg twice daily, concluded that major adverse events can be effectively managed under appropriate conditions. In their study, serious symptomatic anaemia occurred in three patients who were successfully managed via blood transfusion. Von der Lippe (15) also employed blood transfusion successfully to manage severe anaemia in three patients. However, two other patients normalised upon withdrawal of linezolid without having to be transfused blood.  For neuropathy, a summary proportion observed was 41% (95% CI 24% -59%) which was also in conformity with Cox and Ford (17) (36%, 95% CI 19%-53%) and Sotgui et al. (44) (47%, 95% CI 36%-58%). In comparison with the incidence of myelosuppression, the results obtained for neuropathy depicts a higher chance of occurrence than myelosuppression. This is in line with another observation made by Nam et al. (14) on low dose of linezolid (300mg twice daily) which they conclude that reduced dose of linezolid reduces incidence of bone marrow suppression but not neurotoxicity. This hypothesis may be true as no statistically significant difference on the incidence of neurotoxicity was observed between ≤ 600mg and > 600mg linezolid in this review. The same result was reported by Sotgui et al. (44) in their review when they compared the incidence of neuropathy associated the aforementioned doses of linezolid. Also, in one of the studies (15) reviewed, reported incidence of peripheral neuropathy in one patient which was not resolved after 14 months of treatment discontinuation and caution on the likelihood of irreversible peripheral neuropathy associated with linezolid was raised. A similar observation was made by Udwadia et al. (16) reporting of persistent crippling peripheral neuropathy in 17 MDR-TB patients in the Indian cohort following several years of treatment discontinuation. In comparison with the European study conducted by Migliori et al.(28), Udwadia et al.(16) relates these crippling effects to the malnourished condition of the Indian cohort indicating that 16 out of the 18 patients had body mass index less than 18kg.m-2 . These observations may indicate that the incidence of neuropathy may not necessary be high dose dependent but rather the long term exposure resulting from prolonged use. Hence for linezolid to be included in the long treatment regimen of drug resistant tuberculosis, dose optimisation is essential. The occurrence of other adverse events which are usually headache, nausea and gastrointestinal disturbances was observed with a pooled proportion of 11% (95% CI 3% -22%). This observation was very low compared with  Cox and Ford (17) (61%, 95% CI 40% to 83%) and Sotgui et al. (44) (59%, 95% CI 49%-68%). This may be as a result of close to half (n=8) of the included studies in this review did not report the occurrence other adverse events and two studies did not indicate the number of occurrence which led to their exclusion from the meta-analysis. In addition, Sotgui et al. (44) reported a statistical significance of 0.004 of other reported adverse events occurring with higher doses of linezolid. It is therefore necessary to watch out for minor adverse events when initiating linezolid treatment. A case of a 64 year old diabetic man reported by Bodnar et al. (48) in whom linezolid was used to treat Methicillin resistant Staphylococcus Aureus  infected cellulitis severed linezolid associated hypoglycaemia. This incident was attributed linezolid mimicking monoamine oxidase inhibitory properties which causes hypoglycaemia. Linezolid has also been reported to be associated with lactic acidosis (49) with the authors concluding on the mechanism as being unknown while raising an assumption of being associated with mitochondrial toxicity similar to those exhibited by nucleoside reverse-transcriptase inhibitors. This assumption is in line with the finding from Carroll et al. (31) who investigated the mechanism of linezolid toxicity and concluded on the toxicity being associated with linezolid induced decrease in mitochondrion function. One case of rhabdomylosis was reported in this review by Lee et al. (35) which need to be well documented to inform future classification as a rare adverse event or not.

Strengths and Limitations of study

The major strength of this review compared to previously conducted review is the large sample size of 413 patients exposed to linezolid compared with 121 (44) and 218 (17). This sample size was fairly distributed in Europe and Asia covering WHO regions where strategic plans have been put in place to stop TB as well as combat drug resistant strains (50, 51). This study was also limited between 2006 and 2013 which gives a concise picture of how prescribing pattern have been influenced since WHO’s (8) recommendation on linezolid use as part of group five medications in 2006. It also serves as an update to previously conducted review in this subject area. Subgroup analysis comparing doses ≤ 600mg and > 600mg of linezolid minimises the dose size as a confounder to the efficacy and toxicity outcome measures.  Another limitation is the inconsistency in reporting outcome measures of included studies. Although studies included reported outcomes in a manner similar to WHO (52) adapted recommendations, there still variations which may introduce bias in the overall outcome measures. For instance, the treatment duration varied among studies (1month to 29 months) coupled with varying follow-up times. Hence a longer follow up may be good enough to ascertained cured patients, whiles a shorter follow-up time, may classify a patient who may relapse at a much longer period as cured due to the short time follow-up assessment.

Conclusions

(a) Identification of evidence-based primary research

It is documented that linezolid use in  drug resistant TB is based on series of case reports and expert opinions (8). This trend was observed in this review as more than half of the studies (n=13) retrieved were case series and case reports. Three early phase clinical trials were also identified. One cohort study, which happens to the largest retrospective cohort study (28) so far identified in the investigation of linezolid use in drug resistant TB. The only randomised placebo controlled trial (53) identified was reported unsuccessful with results not published to date. Hence, as it concludes in this review, the only high evidence-based study conducted on linezolid use in MDR-TB and XDR-TB is the multi-national TBNET study conducted by Migliori et al. (28).

(b) Efficacy  

It can generally be concluded that linezolid has a promising efficacy in MDR-TB and XRD-TB. Good in vitro studies (11-13) have been identified which have also been supported by positive pharmacokinetic studies (9, 46, 47) depicting the MIC of linezolid being above that reported in in-vitro studies (9). The only identified retrospective cohort study (28) concluded no statistical difference in treatment success between patients receiving linezolid and those not receiving linezolid. For dose comparisons, the significant test showed no statistical significance between culture conversion (p=0.3247) whiles statistical significance was observed for treatment success (0.0001) considering doses ≤ 600mg and >600mg in favour of the higher dose. The resistant strains of Mycobacterium tuberculosis to linezolid observed in four patients in the South Korean (35) cohort raises concern about the promising efficacy of linezolid and as such the need for further investigation into this observed resistance.

(c) Toxicity

Associated adverse events can be both life threatening and manageable as well. The reported major adverse events were myelosuppression and neuropathy. Other side effects included gastrointestinal disorders, nausea and headache. Adverse events resulting in myelosuppression can be salvage by treatment suspension or administration of blood transfusion in severe cases (15, 32). Myelosuppression may be associated with higher dose as reported by significant p-value of 0.0001 in favour of dose > 600mg when compared with ≤ 600mg. On the other hand neuropathy may not be directly related to higher doses as statistical significance (p= 0.1186) was not achieved in the dose comparison study. Neuropathy may also be also not be reversed and may persist for longer years (16, 35). All these reported adverse events may also complicate treatment success by compelling discontinuation. 

Recommendations

The results of this study add to the long standing need to carry out randomised placebo controlled trial to inform evidence based clinical use of linezolid in drug resistant TB. Despite efforts being made by the WHO’s Working Group on New TB drugs partnership (10) with the pharmaceutical industries, only one randomised placebo controlled trial have been completed with unsuccessful results. Though the main objectives of this pilot trial were not achieved, Padayatchi et al. (53) shares an important lesson learnt in the conduct of MDR-TB trial that, the availability of adequate number of trained staff and the strict compliance to administrative controls is paramount and may contribute to successfully accomplishing MDT-TB trial. In the same trial, success was also hampered by non-adherence to treatment by patients possibly due to adverse events. There is therefore the need for dose optimisation to ensure that favourable response is achieved at lower doses. The model for anti-tubercular dose optimisation proposed by Horsburg et al. (45) may be recommended for dose optimisation in future randomised controlled trials. Also further studies into dose optimisation may be investigated for combination therapy with Clarithromycin 500mg daily (46). Finally, drug sensitivity testing will strongly be recommended for initiation of linezolid treatment in MDR-TB and XDR-TB so that possible resistant strains to linezolid may be identified early and necessary preventive measures taken to protect this promising ray of hope from emerging into another strain of drug resistant TB.  

Appendix 1

Modified McMaster critical appraisal tool for quantitative studies (25, 26)