Management of difficult multidrug-resistant tuberculosis and extensively drug-resistant tuberculosis: Update 2012


  • The Authors: Dr. KC Chang, MB, MSc, is a Senior Medical Officer at the Tuberculosis and Chest Service of the Department of Health of Hong Kong SAR. Dr. WW Yew, MB, is one of the Editors-in-Chief of the International Journal of Tuberculosis and Lung Disease and affiliated with the Chinese University of Hong Kong as an Honorary Professor in the Department of Microbiology. Their research interests include various clinical and laboratory aspects of tuberculosis.


Kwok-Chiu Chang, Tuberculosis and Chest Service, Department of Health, Hong Kong, Wanchai Chest Clinic 1st Floor, Wanchai Polyclinic 99, Kennedy Road, Wanchai Hong Kong, China. Email:


Multidrug-resistant (MDR) tuberculosis (TB) denotes bacillary resistance to at least isoniazid and rifampicin. Extensively drug-resistant (XDR) TB is MDR-TB with additional bacillary resistance to any fluoroquinolone and at least one second-line injectable drugs. Rooted in inadequate TB treatment and compounded by a vicious circle of diagnostic delay and improper treatment, MDR-TB/XDR-TB has become a global epidemic that is fuelled by poverty, human immunodeficiency virus (HIV) and neglect of airborne infection control. The majority of MDR-TB cases in some settings with high prevalence of MDR-TB are due to transmission of drug-resistant bacillary strains to previously untreated patients. Global efforts in controlling MDR-TB/XDR-TB can no longer focus solely on high-risk patients. It is difficult and costly to treat MDR-TB/XDR-TB. Without timely implementation of preventive and management strategies, difficult MDR-TB/XDR-TB can cripple global TB control efforts. Preventive strategies include prompt diagnosis with adequate TB treatment using the directly observed therapy, short-course (DOTS) strategy and drug-resistance programmes, airborne infection control, preventive treatment of TB/HIV, and optimal use of antiretroviral therapy. Management strategies for established cases of difficult MDR-TB/XDR-TB rely on harnessing existing drugs (notably newer generation fluoroquinolones, high-dose isoniazid, linezolid and pyrazinamide with in vitro activity) in the best combinations and dosing schedules, together with adjunctive surgery in carefully selected cases. Immunotherapy may also have a role in the future. New diagnostics, drugs and vaccines are required to meet the challenge, but science alone is insufficient. Difficult MDR-TB/XDR-TB cannot be tackled without achieving high cure rates with quality DOTS and beyond, and concurrently addressing poverty and HIV.


antiretroviral therapy


directly observed therapy, short-course


drug susceptibility testing


human immunodeficiency virus




public-private mix


second-line injectable drug




World Health Organization


extensively drug-resistant


Multidrug-resistant (MDR) tuberculosis (TB) denotes bacillary resistance to at least isoniazid and rifampicin. Extensively drug-resistant (XDR) TB is MDR-TB with additional bacillary resistance to any fluoroquinolone and at least one of the three second-line injectable drugs (SLID), namely, kanamycin, amikacin and capreomycin. The definition of XDR-TB reflects the essential roles of fluoroquinolone1–5 and SLID.5–7 Difficult MDR-TB may be arbitrarily defined as pre-XDR-TB with fluoroquinolone resistance, as its outcome is barely better than that of XDR-TB.5

Drug-resistant TB has existed since the beginning of the antibiotic era.8 Genetic resistance to an anti-TB drug occurs naturally, as a result of chromosomal mutations that accompanies mycobacterial replication. However, difficult MDR-TB/XDR-TB is a man-made phenomenon that have emerged because of improper TB treatment. The more active is an anti-TB drug, the more likely its improper use can cause drug-resistant TB by selecting out drug-resistant bacillary mutants.9 MDR-TB has emerged since isoniazid and rifampicin were used in the 1970s. Increasing use of fluoroquinolone in TB treatment in the 1990s promoted the development of pre-XDR-TB/XDR-TB.

In comparison with drug-susceptible TB, MDR-TB and XDR-TB are more difficult to treat, with treatment success rates of 60–80%10 and 44–60%,11–16 respectively, as reported in low human immunodeficiency virus (HIV)-prevalence settings with effective TB programmes. Mortality from MDR-TB and XDR-TB in high HIV-prevalence region is alarming, with 1-year rates reaching 71% and 83%, respectively.17


MDR-TB/XDR-TB has become a global epidemic. In 2010, an estimated 650 000 (5.4%) of the world's 12 million prevalent cases of TB had MDR-TB.18 About 9% of MDR-TB cases worldwide had XDR-TB. By the end of 2011, 77 countries have reported at least one case of XDR-TB. Of the estimated 290 000 globally notified cases of pulmonary MDR-TB in 2010, China and India accounted for 44%. Estimated proportions with MDR-TB were 3.4% among new cases, and 21% among re-treatment cases.18 A recent report from China demonstrated that the majority of MDR-TB cases in some settings with high prevalence of MDR-TB are due to transmission of drug-resistant bacillary strains to previously untreated patients rather than mismanagement in a previous treatment episode.19

The global epidemic of MDR-TB/XDR-TB is rooted in inadequate TB treatment, compounded by a vicious circle of diagnostic delay and improper treatment that amplifies drug resistance, and fuelled by conditions that promote transmission, infection and development of disease.9 Previous TB treatment is a strong risk factor for MDR-TB.19–28 As shown by the situation in South Africa,29 HIV is one of the most important factors that has worsened the global TB epidemic. People living with HIV who are also infected with Mycobacterium tuberculosis are about 21–34 times more likely to develop TB disease compared with those who are HIV-negative.18 A systematic review has alluded to a significant association between HIV and primary MDR-TB.30 Although the number of HIV-infected people screened for TB has almost quadrupled from 600 000 in 2007 to 2.3 million in 2010, this represents less than 7% of the 34 million people estimated to be living with HIV.

MDR-TB is currently under-diagnosed and inadequately treated, Globally, less than 2% of new cases and 6% of previously treated cases were tested for MDR-TB,18 and only 16% of MDR-TB cases notified in 2010 were enrolled on treatment.18 The treatment of MDR-TB/XDR-TB is so costly that it can drain valuable resources needed for global TB control. Without timely implementation of preventive and management strategies, difficult MDR-TB/XDR-TB can cripple global TB control efforts.

Global efforts in controlling MDR-TB can no longer focus solely on high-risk patients, as this will miss a substantial proportion of MDR-TB cases.31Figure 1 shows a multifaceted approach that is required for the management of difficult MDR-TB/XDR-TB. Preventive strategies include prompt diagnosis with adequate TB treatment using the directly observed therapy, short-course (DOTS) strategy and drug-resistance programmes, airborne infection control, preventive treatment of TB/HIV, and optimal use of antiretroviral therapy (ART). Management strategies for established cases mainly rely on specific alternative treatment regimens complemented with surgery in carefully selected cases.

Figure 1.

Strategies for the management of difficult multidrug-resistant tuberculosis (MDR-TB)/extensively drug-resistant tuberculosis. Arrows with dotted lines allude to a downregulating influence on the emergence of MDR-TB. Lines without arrows connote components related to the central object. CRI, colorimetric redox indicator methods; DOTS, directly observed therapy, short-course; FDC, fixed-dose drug combination; LPA, line probe assay; MODS, microscopic observation of drug susceptibility; NRA, nitrate reductase assay; TDM, therapeutic drug monitoring.


The first principle in the management of difficult MDR-TB/XDR-TB is preventing its emergence. Treatment non-adherence was soon recognized as a common behavioural problem that caused treatment failure and acquired resistance since a milestone study in Madras triggered a paradigm shift from sanatorium care to community-based TB treatment.32

DOTS and beyond

Ambulatory treatment supervision, the origin of directly observed therapy, was first explored in the 1960s to enhance treatment adherence.33 It soon became the cornerstone of TB treatment. In response to the resurgence of TB, World Health Organization (WHO) declared TB as the first global emergency in 1993 and coined DOTS to emphasize directly observed therapy with short-course combination chemotherapy using first-line drugs.34 DOTS became well-known as a public health strategy that embraced political commitment, diagnosis by sputum-smear microscopy, delivery of chemotherapy under directly observed settings, reliable supply of quality drugs, and a sound recording and reporting system. Given proper implementation, DOTS can achieve cure rates of at least 90% and prevent MDR-TB.34 Additionally, the Orizaba study has demonstrated that DOTS with only first-line drugs can reduce transmission of MDR-TB.35 However, DOTS with first-line drugs is inadequate for treating MDR-TB. A descriptive study in the Philippines showed that MDR-TB occurred most frequently among patients who failed treatment with the Category II regimen.20 The Category II regimen is inadequate for treatment failures due to Category I regimens,36 with cure rate <50% among smear-positive MDR-TB patients and a high risk of amplifying drug resistance.37,38 Among a prison cohort with a high prevalence of drug-resistant TB, a study demonstrated drug-resistance amplification in 3.4% of the cohort after standardized chemotherapy with the Category II regimen.21 The subsequent addition of fluoroquinolones and/or SLID can generate XDR-TB.38

Recently called drug-resistance programmes, DOTS-plus was introduced in 1998 to tackle MDR-TB with second-line drugs using the DOTS strategy plus drug susceptibility testing (DST).39 Ideally, specific alternative treatment regimens that are individually tailored to the drug-resistance pattern may give the best chance of cure. In reality, it is prohibitively expensive to individualize treatment for all patients with drug-resistant TB.34 Cohort studies in Peru and Bangladesh demonstrated the feasibility of using standardized treatment with second-line drugs to improve MDR-TB treatment in resource-limited settings.40,41

Innovations for promoting treatment adherence

Dosing intermittency has been introduced to facilitate DOTS because clinical trials of intermittent regimens suggested high cure with low relapse rates.42 Cumulative evidence in recent years suggest that daily treatment, especially in the initial phase in the presence of cavitation, reduces the risk of treatment failure, recurrence and acquired drug resistance including acquired rifamycin monoresistance.43

Fixed-dose drug combination formulations comprising two, three and even four drugs have been used to enhance ease of prescription, reduce inadvertent medication errors, simplify drug supply and improve patient acceptability.44–46 With modest rates of incomplete interruption, fixed-dose drug combination formulations can reduce the risk of acquiring MDR-TB.47 But fixed-dose drug combination tablets cannot substitute for directly observed therapy, as drug resistance can still arise as a result of irregular treatment.48

Proper use of incentives can improve treatment adherence. A retrospective cohort analysis of the TB directly observed therapy programme in New York City showed that more incentives improved treatment adherence.49 Among homeless patients in Russia, food incentives along with help from a social worker improved treatment adherence from 31% to 59–95%.50 However, food incentives may have nonsignificant impact on treatment adherence that exceeds 90%.51

Patient-centred care may also improve treatment adherence. Among patients at high risk of treatment default in Tomsk, a patient-centred TB treatment delivery programme called Sputnik improved treatment adherence from 52% to 81% for TB treatment recipients.52 Mean adherence and cure rate were, respectively, 79% and 71.1% for MDR-TB patients, and 89% and 60% for all others. A study in Pakistan demonstrated that TB treatment adherence increased from 23.3% to 56.1% as scores for hospital service quality and patient satisfaction improved.53 A Thai study suggested that a triad model, which emphasized interactions between the health-care provider, the TB patient and the treatment supporter, significantly improved the TB cure rate from 79% to 95%.54 A cluster randomized, controlled trial in a resource-constrained country showed that patient counselling and communication, decentralization of treatment, and reinforcing supervision activities with patient's choice of directly observed therapy supporter significantly reduced the defaulter rate from 16.8% to 5.5% and increased the treatment success rate from 76% to 88%.55

First published in 2001,56,57 the idea of engaging private practitioners to improve TB control rapidly evolved into a strategy known as public-private mix DOTS (PPM DOTS)58 that soon gained support by the International Standards for TB Care.59 PPM DOTS, now known as PPM for TB care and control, has become a core component of the WHO Stop TB Strategy. A review of the Global Fund's official data showed that PPM contributed to detecting more than 25% TB cases with high treatment success rates in China, India, Nigeria and the Philippines.60 A systematic assessment suggests that PPM has improved case detection and treatment outcomes among patients seeking care from private practitioners, but there is a need for guidelines on engaging practitioners that primarily contribute to health care of the poor.61 A retrospective cohort study demonstrated the feasibility of using PPM DOTS to improve treatment outcomes among both HIV-infected and HIV-negative TB patients.62 A recent exploratory study in the Philippines also showed promising results for integrating MDR-TB management into PPM DOTS.63 Nonetheless, not all PPM projects are successful. A cohort study of PPM in Vietnam showed worse treatment outcomes in PPM in comparison with the national TB programme and tentatively attributed this to lack of regulatory enforcement and subsidization of drug costs.64 A study in Indonesia found that a substantial proportion of TB patients cared for at PPM DOTS hospitals are not managed under the DOTS strategy.65 It is perhaps necessary to monitor the quality of care provided by the partnership66 with attention to underlying motivations.67

Innovations for ascertaining prompt diagnosis and DST

Conventional diagnostic and DST methods are inadequate for tackling the global epidemic of MDR-TB/XDR-TB. Sputum acid-fast bacilli smear examination misses the diagnosis of TB in 50–70% and provides no information about drug susceptibility. WHO recommendation of replacing conventional fluorescence and Ziehl–Neelsen microscopy with light-emitting diodes microscopy does not address the intrinsic limitations of microscopy.68 Conventional diagnostic and DST methods that depend on culture are slow, with a turnaround time of 9–12 weeks for first-line DST and 12–16 weeks for second-line DST.69

Rapid culture and DST in liquid medium and molecular assays are available to reduce diagnostic delay of MDR-TB. This may help protect fluoroquinolones and pyrazinamide, which are both recommended in MDR-TB treatment. A decision analysis suggested that the most cost-effective strategy in settings with moderate to high burdens of drug-resistant TB would be performing rapid DST of isoniazid and rifampicin for all patients before treatment initiation.70 A mathematical model suggests that better use of rapid culture in liquid medium or molecular tests could reduce TB prevalence and mortality in a high HIV-prevalence setting by at least 20%.71

Liquid culture can shorten the expected time for MDR-TB detection to 3–5 weeks.69 Direct use of line probe assay in most smear-positive, and some smear-negative, clinical specimens allows the diagnosis of MDR-TB within 48 h.69 WHO has recommended the use of commercial line probe assay in direct testing of smear-positive sputum specimens and on isolates of M. tuberculosis complex for early detection of MDR-TB.72 WHO has also strongly recommended the use of Xpert MTB/RIF as the initial diagnostic test in patients suspected of having MDR-TB or HIV-associated TB, and conditionally recommended its use as a follow-on test to microscopy in other settings.73 Xpert MTB/RIF is a TB-specific, automated, cartridge-based nucleic amplification assay that uses a real-time polymerase chain reaction method involving short segments of single-stranded DNA called molecular beacons. In comparison with microscopy, the use of Xpert MTB/RIF is expected to increase the diagnosis of drug-resistant TB by three times and double the yield of TB/HIV diagnoses.74 For resource-limited settings, WHO has recommended interim use of microscopic observation of drug susceptibility and nitrate reductase assay for direct testing of sputum specimens, and colorimetric redox indicator methods, microscopic observation of drug susceptibility and nitrate reductase assay for indirect DST of M. tuberculosis isolates.75

Rapid culture and DST in liquid medium can detect XDR-TB in 4–9 weeks.69 Genotype MTBDRsl is a line probe assay that can be used for rapid diagnosis of fluoroquinolone-resistant MDR-TB and XDR-TB,76,77 but its role in XDR-TB is limited by the modest sensitivity for mutations associated with bacillary resistance to SLID.77 As rapid molecular assays other than polymerase chain reaction DNA sequencing can reliably screen for ofloxacin resistance in MDR-TB culture isolates,78 liquid culture may be combined with molecular assays to detect fluoroquinolone-resistant MDR-TB in 3–5 weeks for smear-negative specimens.

Despite the availability of rapid culture and molecular assays, we still need conventional microscopy, culture and DST in solid medium for confirming diagnosis, monitoring treatment and epidemiological studies.79

Infection control

A well-designed experiment that exposed guinea pig to infectious TB patients has convincingly demonstrated that TB is an airborne infectious disease.80 Although resistance-associated mutations may render M. tuberculosis strains less fit for transmission, MDR-TB strains can be up to 10 times more transmissible than pan-susceptible strains possibly because genetic evolution can mitigate fitness defects.81 A retrospective nested case–control study in Latvia showed that MDR cases were more likely to be found in clusters than drug-susceptible cases (74.0% vs 33.6%).82 A recent study also suggests that XDR-TB significantly increases transmission among household contacts of MDR-TB patients.83 The spread of MDR-TB and XDR-TB in South Africa highlighted the importance of infection control in health care and other congregate settings.84,85

A mathematical model of inpatient and community-based data in a resource-limited setting showed that a synergistic combination of infection control measures with rapid DST and ART could avert 48% of XDR-TB cases, with nearly one third prevented by use of face masks and a shift to outpatient therapy.86 The model also showed an unexpected association between rise in XDR-TB incidence and involuntary detention, which could be explained by restricted isolation capacity. A recent study demonstrated that the risk of nosocomial MDR-TB transmission could be halved by requiring infectious MDR-TB patients to use face masks.87 Findings allude to the importance of incorporating community-based strategies and effective airborne infection control measures into the TB programme in resource-limited settings.88,89 It cannot be overemphasized that the foundation of TB infection control is early and rapid diagnosis of TB, plus its proper management.90

Preventive treatment

A retrospective analysis has demonstrated a significant association between reduced TB incidence and use of both isoniazid preventive therapy and ART in HIV-infected patients.91 Thus, WHO has recommended the three ‘I's for TB/HIV: intensified TB case finding, isoniazid preventive therapy and infection control for TB.90,92 A randomized, controlled trial of preventive treatment among HIV-infected subjects found that rifapentine and isoniazid once weekly for 12 weeks, or rifampicin plus isoniazid twice weekly for 12 weeks, had comparable effectiveness as isoniazid once daily for 6 months.93

The role of preventive treatment for MDR-TB contacts is uncertain. A cross-sectional study of 302 household contacts of MDR-TB patients showed that nearly 90% of MDR-TB contacts had drug-susceptible TB, and the MDR-TB rate was only 0.66%.94 A systematic review showed insufficient evidence to support or reject preventive treatment of MDR-TB.95 Although preventive treatment for high-risk MDR-TB contacts may be necessary, no consensus guidance exists on how best to manage these cases.96 Using efficacy estimates extrapolated from mouse models and a wide range of assumptions, a computerized Markov model of total treatment cost suggested that preventive treatment with moxifloxacin/ethambutol would be preferable to other combinations of pyrazinamide, ethambutol, ethionamide and PA-824.97

HIV treatment as prevention

ART has considerable benefit both as treatment of HIV and prevention of TB.98 A recent WHO-led meta-analysis of observational studies from low- and middle-income countries found that ART decreased the risk of TB by up to 65% irrespective of CD4+ cell count.99 Findings suggest that earlier initiation of ART may be strategically important for controlling the HIV-associated TB syndemic.


Difficult MDR-TB/XDR-TB is predominantly treated with combination anti-TB drug therapy complemented with surgery in highly selected cases. Adjunctive immunotherapy may have a role in the future.

Anti-TB drug therapy

Classification of anti-TB drugs

For MDR-TB treatment, WHO has grouped anti-TB drugs according to efficacy, experience of use and drug class.100Table 1 shows a slightly different group order that may better reflect the clearly essential roles of fluoroquinolones1–5 and SLID.5–7 Drugs from groups 1–3 should be used if there is good evidence from DST or clinical history that they are effective.100 Drug resistance is a critical determinant of treatment success, and prior TB treatment confers an increased risk.111,112 DST is fairly reliable for drugs from groups 1–3 and less so for the others.100

Table 1. Antituberculosis drugs: classification and dosages used in multidrug-resistant tuberculosis treatment
Group: descriptionComponentsDaily dosage (mg/kg)Usual daily dosage (mg)
  • This table is adapted from two reviews.106,107 Unless otherwise stated, drugs are generally given on a once-daily basis except for injectable agents (which may be given three to five times per week), amoxicillin-clavulanate (given twice daily) and para-aminosalicylic acid (PAS) (given twice to thrice daily). Dosing of second-line injectable drugs, ethambutol, pyrazinamide and possibly other drugs, should be based on ideal rather than actual body weight among obese patients.101,108,109

  • † 

    Some (but not all) patients with body weight >70 kg can receive higher dosages if they are tolerable.

  • ‡ 

    Drugs or food containing divalent or trivalent cations (calcium, magnesium, iron and aluminium) should be administered at least 2 h before or after fluoroquinolone administration because they impair absorption of fluoroquinolones.110

  • § 

    For patients aged ≥60 years or with mild renal insufficiency (creatinine clearance 30–60 mL/min), the dose size is reduced to 10–12 mg/kg with a daily maximum of 750 mg.

  • ¶ 

    For ethionamide, prothionamide, cycloserine and terizidone, some patients may require administration in two split doses per day. For PAS, it is necessary to give it in split dosages two to three times per day.

  • †† 

    Optimal dosage has not been fully delineated for linezolid, amoxicillin-clavulanate and clofazimine. Long-term safety has not been fully confirmed for linezolid, amoxicillin-clavulanate and meropenem-clavulanate. Thiacetazone should be used only in the absence of alternative group 5 drugs in patients documented to be human immunodeficiency virus-negative. Meropenem is given intravenously, while clavulanate is given orally in the form of amoxicillin plus clavulanate.104

  • NA, not applicable; —, the optimal dosages for these drugs have not been fully delineated. Please refer to the text for the relevant discussion.

1: First-line oral drugsIsoniazid16–18800–1200
2: FluoroquinolonesLevofloxacin101750–1000
3: Injectable agents§Capreomycin12–15750–1000
4: Oral bacteriostatic second-line agentsEthionamide15500–750
Para-aminosalicylic acid2008000–12000
5: Agents with efficacy that is not totally clear (not recommended for routine use)††Linezolid600
Amoxicillin-clavulanate875-125 b.i.d.
Clarithromycin500 b.i.d.
Meropenem-clavulanate104Initially 2000-125 t.i.d, later 2000-125 b.i.d.
Thioridazine105Initially 25 mg daily.
Increase dosage weekly until 200 mg daily.

First-line oral drugs (group 1) are the most potent and best tolerated.100 A randomized, controlled trial among relatively young MDR-TB patients suggests that in comparison with non-recipients, high-dose isoniazid (16–18 mg/kg) recipients became sputum-negative 2.38 times more rapidly and had a 2.37 times higher likelihood of being sputum-negative at 6 months113 Considering the pharmacokinetics of isoniazid,114 high-dose isoniazid (16–18 mg/kg) is likely effective when isoniazid minimal inhibitory concentration ≤1 mg/L,115 and potentially beneficial when isoniazid minimal inhibitory concentration ≤5 mg/L. Pyrazinamide may improve treatment of pyrazinamide-susceptible MDR-TB.116 Before rifampicin and fluoroquinolones were available for the treatment of drug-resistant TB, high cure and culture conversion rates were achieved by combining pyrazinamide, ethionamide and cycloserine.117–119 A small retrospective analysis suggested significantly better outcomes among MDR-TB patients given ethambutol and pyrazinamide.120 One analysis included in a WHO review suggested a slightly added adjusted benefit due to pyrazinamide.121 A retrospective cohort analysis suggests that pyrazinamide may substantially improve early sputum culture conversion and 2-year treatment success among MDR-TB patients given fluoroquinolone-based regimens.122 The potentially important role of pyrazinamide and the high prevalence of pyrazinamide resistance in MDR-TB123 underscores the need of protecting pyrazinamide in settings with substantial MDR-TB burden.

Among group 2 drugs, preference is given to newer generation fluoroquinolones, which include levofloxacin, moxifloxacin and gatifloxacin.100 Ciprofloxacin is no longer recommended. A systematic review with meta-analysis suggests that newer generation fluoroquinolones significantly improves cure or treatment completion among XDR-TB patients.124 Among predominantly second-line treatment-naïve MDR-TB patients, high relapse-free cure rates with good tolerance have been achieved by the inclusion of gatifloxacin 400–800 mg once daily for at least 9 months,103 notwithstanding initial concerns about gatifloxacin-induced dysglycaemia.125 A murine model suggested that moxifloxacin 400 mg once daily could be effective when moxifloxacin minimal inhibitory concentration ≤2 mg/L.126 An in vitro pharmacodynamic infection model suggests that moxifloxacin 800 mg once daily likely achieves excellent M. tuberculosis microbial kill with suppression of drug resistance.127 Tolerance and safety of high-dose moxifloxacin will be evaluated in a randomized, controlled trial, using standardized regimens in MDR-TB treatment.102 In view of likely cost-effectiveness,106,128 levofloxacin 1000 mg once daily may be prescribed,101 but long-term tolerance data in the literature are very limited.107

Among injectable agents (group 3), kanamycin or capreomycin is the drug of choice for second-line treatment-naïve MDR-TB because many M. tuberculosis strains resistant to either drug are still susceptible to amikacin but probably not vice versa.129 The choice between kanamycin and capreomycin is determined by cost and availability. Streptomycin is not recommended owing to relatively high rates of bacillary resistance and ototoxicity.100

Oral bacteriostatic agents (group 4) include thioamide (ethionamide or prothionamide), cycloserine, terizidone and para-aminosalicylic acid. In view of the molecular mechanism of bacillary resistance,9 a thioamide may still have activity in vitro in the presence of high-level phenotypic resistance to isoniazid. Use of a thioamide with para-aminosalicylic acid may increase the risk of gastrointestinal side-effects and hypothyroidism.100,130–132 Combined use of a thioamide with pyrazinamide or high-dose isoniazid may increase the risk of hepatotoxicity, although a retrospective study suggests that hepatotoxicity during MDR-TB treatment may not adversely affect outcome.133,134

Group 5 drugs, generally representing repurposed agents, is perhaps indispensable in the treatment of difficult MDR-TB/XDR-TB.100,121 Abundantly supported by in vitro135–138 and clinical data,139,140 linezolid is likely beneficial in the treatment of difficult MDR-TB/XDR-TB.139,140 Although linezolid 300 mg once daily may reduce the risk of neurotoxicity,141,142 uncertainty still prevails regarding the optimal dosage.143 The carbapenem class of beta-lactams are very poorly hydrolysed by BlaC gene products that are encoded by M. tuberculosis.144 When combined with the beta-latamase inhibitor clavulanate, meropenem showed potent anti-TB activity in vitro, including anaerobically grown cultures that mimic mycobacterial persisters.144 Thioridazine may improve the cure of XDR-TB by several mechanisms. It affects the transport of K+ and Ca2+ from the phagolysosome, thereby rendering better acidification and activation of hydrolases and enhancing the killing of intracellular M. tuberculosis.145 It also inhibits the genetic expression and the activity of existing efflux pumps that contribute to the MDR phenotype.146 Clinical evidence that may support use of meropenem-clavulanate104,147 or thioridazine105 in XDR-TB treatment is limited and confounded by the concomitant use of linezolid in many assessed patients. Although clofazimine may be beneficial among second-line treatment-naïve patients,103 its role among previously treated patients, who were more frequently treated with a fluoroquinolone and SLID, appeared less impressive.148 This may cast doubt on the role of clofazimine in the treatment of difficult MDR-TB/XDR-TB. Use of amoxicillin-clavulanate is based on anecdotal evidence and inconsistent in vitro findings.149–155

New drugs with potentials

New drugs for TB treatment have been recently reviewed.156 Three new drugs with potential for improving MDR-TB treatment in the near future may deserve some attention, namely bedaquiline, PA-824 and delamanid.

Bedaquiline (a.k.a. TMC207 and J compound) is a novel diarylquinoline that inhibits mycobacterial adenosine triphosphate synthase.157 Metabolized by cytochrome P450 3A4, plasma levels of bedaquiline may be affected through interaction with rifampicin and some antiretroviral drugs (protease inhibitors/non-nucleoside reverse transcriptase inhibitors). Its use in mice suggested synergism with pyrazinamide, and potential for shortening treatment158 and enabling once-weekly dosing.159 A double-blind, randomized placebo-controlled phase II clinical trial showed that adding bedaquiline to a standard five-drug MDR-TB regimen significantly hastened and increased the proportion with sputum culture conversion (48% vs 9%),160 and helped prevent acquired resistance to companion drugs.161

As derivatives of metronidazole, PA-824 (a nitroimidazo-oxazine) and delamanid (a.k.a. OPC-67683, a nitro-dihydro-imidazooxazole) are nitroimidazopyrans that share cross-resistance. They probably act by inhibiting cell wall biosynthesis, among other possible actions,162 with OPC-67683 having 20 times higher potency. High protein binding (94%) may render PA-824 less accessible in cavities of pulmonary TB.163 Use of PA-824 in mice suggested synergism with moxifloxacin and pyrazinamide and potential for shortening treatment.164,165 Likewise, use of OPC-67683 with rifampicin and pyrazinamide in mice demonstrated TB treatment-shortening potential and considerable intracellular post-antibiotic effects.162 A randomized trial that compared delamanid with placebo used alongside a background MDR-TB treatment regimen showed that delamanid significantly improved 2-month sputum culture conversion from 29.6% to 41.9%/45.4%,166 with a substantially increased risk of asymptomatic QT prolongation as a side-effect.166

General principles for designing a desirable treatment regimen

Without specific evidence-based guidelines on treatment of difficult MDR-TB/XDR-TB, we may perhaps adapt the latest WHO guidelines on programmatic management of MDR-TB in designing treatment regimens for these formidable diseases. On the basis of low-quality clinical evidence for non-XDR-TB, WHO has recommended the following principles.121 First, the intensive-phase treatment should include pyrazinamide in addition to at least four second-line anti-TB drugs likely to be effective. Although pyrazinamide may improve fluoroquinolone-based treatment of MDR-TB,122 its role in the treatment of difficult MDR-TB/XDR-TB may be substantially reduced by the high prevalence of pyrazinamide resistance in fluoroquinolone-resistant MDR-TB167 and XDR-TB.168 To balance treatment efficacy and toxicity, it appears prudent to identify pyrazinamide susceptibility with molecular assays whenever possible.116 Second, the four second-line drugs should include a fluoroquinolone, a SLID, a thioamide and cycloserine, which may be replaced by para-aminosalicylic acid if necessary. A rule of thumb is to include two core drugs with potent bactericidal activity plus two accompanying drugs. Third, the number of second-line drugs may be further increased in case of uncertainty about effectiveness but not for extensive disease per se. Thus, for the reasons given earlier, linezolid and high-dose isoniazid may be included after carefully weighing tolerability and safety.

Dosing frequency

Daily rather than intermittent scheduling is generally recommended in MDR-TB treatment.101 For patients at risk of otovestibular or renal toxicity, especially those aged ≥60 years or with mild renal insufficiency (creatinine clearance 30–60 mL/min), dosing frequency of SLID may be reduced to five times per week, which is generally considered to be effective,101 in the initial 2–3 months and then thrice weekly.101 When renal dysfunction is significant (creatinine clearance <30 mL/min), the dosing frequency of SLID, ethambutol, pyrazinamide, cycloserine and levofloxacin is preferably reduced to thrice weekly with little change in the dose size.100,101,110,169 Although use of daily high-dose isoniazid among relatively young MDR-TB patients was not associated with hepatotoxicity,113 thrice-weekly high-dose isoniazid may warrant evaluation in view of its track record of safety. Intermittent dosing of linezolid in the continuation phase has been used with promising results, albeit preliminary, to strike a balance between efficacy and toxicity of the oxazolindinone.170

Treatment duration

Based on low-quality evidence, WHO has conditionally recommended a minimum of 8 months for the intensive phase, and a minimum of 20 months for the entire course in the treatment of newly diagnosed MDR-TB.121 Tolerability and safety must be carefully weighed. Evidence for the effectiveness of a 9-month treatment regimen is limited and confined to the treatment of predominantly treatment-naïve MDR-TB.103 Further studies/trials are warranted to ascertain the optimal duration of the entire course as well as use of SLID in the treatment of difficult MDR-TB/XDR-TB, especially when the regimen contains high-dose gatifloxacin/moxifloxacin in the presence or absence of pyrazinamide with in vitro activity.

TB-HIV disease

ART significantly reduces the impact of HIV on TB by lowering HIV viral load and restoring the immune system.171 In HIV-infected patients with documented MDR-TB/XDR-TB, ART should be initiated within 2–4 weeks of confirmation of TB drug resistance and initiation of second-line TB therapy.171–176 Immune reconstitution inflammatory syndrome may occur after initiation of ART. In general, both ART and TB treatment should be continued while managing immune reconstitution inflammatory syndrome.172

Model of care

Based on low-quality evidence, WHO has conditionally recommended that MDR-TB be treated using mainly ambulatory (or community-based) rather than hospitalization care models.121 This can possibly improve the overall cost-effectiveness and reduce secondary transmission provided that proper infection control measures are also in place at home and in the clinic. These measures include proper use of surgical face masks by MDR-TB patients,87 prioritizing community-care approaches and avoiding overcrowding and unnecessary hospitalization/outpatient visits.90

The concept of community-based patient care in MDR-TB was conceived in Peru in the 1990s.177 Similar success stories have been witnessed in South Africa.178,179 A recent report from rural South Africa about integrated home-based treatment for MDR-TB and HIV has further strengthened the rationale of community-based management of these diseases.180

Adjunctive surgery

Adjunctive lung resection may be considered for patients with drug-resistant TB after satisfying three basic criteria: a high probability of failure or relapse with medical therapy alone, sufficiently localized disease for resection with adequate postoperative cardiopulmonary capacity, and sufficient drug activity for facilitating postoperative healing of bronchial stump.181 Poor postoperative outcomes are more likely in the presence of low body mass index <18.5 kg/m2, bacillary resistance to fluoroquinolones and unresectable cavities.182

Large cohort studies have demonstrated that the best outcomes in MDR-TB are achieved by the use of fluoroquinolones and adjunctive surgery.3,183 Lung resection can help achieve sustainable sputum culture conversion in selected XDR-TB patients.184 Large case series have also reported better outcomes among XDR-TB patients treated with adjunctive lung resection.3,185,186 A retrospective study suggests that early aggressive treatment comprising at least four effective drugs and lung resection may improve the outcome of MDR-TB/XDR-TB.187Table 2 summarizes outcomes of adjunctive surgery of MDR-TB. The median proportion (interquartile range) was 89% (78–95%) for treatment success, 1% (0–3%) for operative mortality and 16% (12–23%) for postoperative complications.

Table 2. Outcomes of adjunctive surgery of multidrug-resistant tuberculosis
First authorYear of publicationNo. of patientsTreatment success (%)Operative mortality (%)Postoperative complication (%)
van Leuven18919976275223
Jeon18620091656Not availableNot available
Gegia20520123778Not availableNot available

When cardiopulmonary reserve is insufficient, collapse therapy using thoracoplasty,206 plombage207 or artificial pneumothorax208 may be considered.

Adjunctive immunotherapy

Although preliminary data on supplemental cytokines209–213 and M. vaccae214 are encouraging, the limited number of enrolled patients and the frequent lack of comparable controls leaves considerable uncertainty regarding their definitive role in the treatment of difficult MDR-TB/XDR-TB.215,216 Notwithstanding a theoretical basis,217 clinical evidence has not substantiated vitamin D supplementation.218 We hope the next decade may witness rewarding results in the continual quest for effective immunomodulating agents.

Palliative care

Difficult MDR-TB/XDR-TB can be incurable. When the risk of treatment outweighs its benefits for both the patient and the community, palliative care is indicated to provide symptomatic treatment and ensure a dignified end of life.219–221


We need new diagnostics, drugs and vaccines to meet new challenges, notably HIV co-infection and rampant drug resistance in the global TB scene.222 Importantly, we must learn how to use newly developed tools optimally to enable sustainability. But, science alone is insufficient.223 The principles of managing drug-resistant TB have existed for over 50 years.224 Without a sound public health approach, new drugs may be rapidly lost, and it is unlikely that the discovery of new drugs/regimens can keep pace with the emergence of drug resistance. To make matters worse, the wide availability of inadequate treatment will reduce TB death but aggravate the global disease burden by increasing chronic releasers of drug-resistant M. tuberculosis. To meet the challenge of MDR-TB/XDR-TB, tireless efforts must be devoted to achieve high cure rates with quality DOTS and drug-resistance programmes.

TB is a disease of poverty225 fuelled globally by the HIV epidemic. The association has been illustrated in countries that witness a breakdown of health services and further socioeconomic inequality in access of health care, thus culminating in higher TB morbidity and mortality, alongside an upsurge of MDR-TB. Difficult MDR-TB/XDR-TB cannot be tackled without concurrently addressing socioeconomic issues associated with unfavourable outcomes.226 It is encouraging to see that many countries have joined efforts to turn back the clock and that international efforts are geared towards strategic use of ART to prevent TB/HIV.227,228