Tuberculosis (TB) is an airborne infectious disease that kills almost two million individuals every year. Multidrug-resistant (MDR) TB is caused by strains of Mycobacterium tuberculosis (M. tb) resistant to isoniazid and rifampin, the backbone of first-line antitubercular treatment. MDR TB affects an estimated 500 000 new patients annually. Genetic analysis of drug-resistant MDR-TB showed that airborne transmission of undetected and untreated strains played a major role in disease outbreaks. The need for new TB vaccines and faster diagnostics, as well as the development of new drugs, has recently been highlighted. The major problem in terms of current TB research and clinical demands is the increasing number of cases of extensively drug-resistant and ‘treatment-refractory’ TB. An emerging scenario of adjunct host-directed therapies is intended to target pulmonary TB where inflammatory processes can be deleterious and lead to immune exhaustion. ‘Target-organ-saving’ strategies may be warranted to prevent damage to infected tissues and achieve focused, clinically relevant and long-lasting anti-M. tb cellular immune responses. Candidates for such interventions may be biological agents or already approved drugs that can be ‘re-purposed’ to interfere with biologically relevant cellular checkpoints. Here, we review current concepts of inflammation in TB disease and discuss candidate pathways for host-directed therapies to achieve better clinical outcomes.
In a recent literature and online source search, the economic data regarding the treatment costs of TB cases in 2011 for 27 EU member states were summarized . The costs for the EU-15 countries, plus Cyprus, Malta and Slovenia, on average per case were €10 282 for drug-susceptible TB, €57213 for multidrug-resistant (MDR) TB and €170 744 for extensively drug-resistant (XDR) TB. The ‘call for action’ to improve prevention, diagnosis and treatment of TB often demands an increase in efforts within broadly accepted concepts. It seems, however, that the testing of new TB drugs and vaccines requires improved characterization of patients with TB to carefully re-evaluate inclusion and exclusion criteria for clinical trials, as well as to understand why some individuals respond better than others to certain (drug or vaccine) protocols within clinical trials. The underlying pathophysiology of the ‘10/3/1 formula’ (i.e. 10 people exposed, 3 at risk to end with latent TB and 1 at risk to acquire disease) cannot be addressed; current challenges include the long incubation period of TB and the lack of robust markers (either related to the pathogen or host biomarkers) that would indicate protection or progression from latent to active TB. Markers that indicate increased risks of Mycobacterium tuberculosis (M. tb) infection, development of active TB, destructive lung disease or risk of M. tb re-infection are of high priority in the clinical management of TB.
In this review, we address the need to define the clinical and immunological response status of patients with TB and discuss the concept that tailored treatment of inflammation in TB would help to limit organ damage and immune exhaustion. Adjunct immunological therapies may also be helpful to restore clinically-relevant immune responses in cases with severe TB. Novel insights into targeted immune modulation could lead to TB clinical trial developments and new treatment strategies. Here, we consider several pathways in TB-related inflammation that could serve as examples of models in which host-directed therapies (HDTs) could augment standard anti-TB treatments.
The concept of adjunct therapy and TB infection
In a number of studies, the immune signatures in M. tb infection have been addressed. It has been suggested that alterations in TB-associated gene and protein expression profiles predict standard drug treatment responses , but without tangible clinical impact on guiding anti-M. tb adjunct therapies. An interferon-gamma (IFN-γ)-related neutrophil signature differentiated TB patients from control subjects, and this was correlated with radiographic extent of disease along with M. tb treatment . Although it is clear that cellular immune reactivity is crucial in mediating effective anti-M. tb responses, more recently the possibility of a role of B-cell response as a diagnostic tool in M. tb infection has been examined ([4-6]. This possibility was confirmed by the observation that alterations in gene expression in blood from patients during TB treatment reflected the modulation of the humoral immune response ). Humoral markers to identify M. tb molecules as vaccine targets or as measures of anti-M. tb-directed immune responses associated with the TB clinical presentation (latent TB, active TB or exposure without disease) have been described and are under further investigation [6, 8]. The involvement of B cells in anti-M. tb immune responses was further consolidated by gene expression profiling in immune cells from patients with active and latent TB, showing increased transcripts as well as protein production of BAFF (B-cell-activating factor belonging to the TNF family) and APRIL (a proliferation-inducing ligand, also known as TNFSF13) in active but not in latent disease . In addition to potentially being of use for developing improved diagnostics for the differentiation of active versus latent TB , this may also point to a role of B cells in immune pathology. B cells not only act as antibody-producing cells, but also as antigen-presenting cells. This has been observed in patients with multiple sclerosis and other inflammatory diseases by the removal of B cells using anti-CD20 monoclonal antibodies (e.g. rituximab). Ablation of B cells not only leads to a decline in B cells, but also in T cells and the chemokines CXCL13 and CCL19 in patients with autoimmune diseases such as multiple sclerosis . Increased BAFF levels in patients with active TB  may therefore reflect an inflammatory syndrome that involves B cells as both antibody-producing and antigen-presenting cells. Of note, BAFF has been shown to be increased in a number of diseases, including systemic lupus erythematosus , leading to chronic inflammation and immune pathology. Interleukin (IL)-17 has also been shown to interact with human B cells. IL-17 is able to promote B-cell survival and, in synergy with BAFF, prevents B-cell antigen receptor-mediated apoptosis . Thus, the findings of increased BAFF and IL-17 suggest the involvement of B cells in anti-M. tb responses and/or M. tb-associated pathobiology.
IL-17, which is increased in patients with active TB compared to individuals with latent infection , appears to be a ‘double-edged sword’ in TB infection. Whilst IL-17 may be beneficial for containment of M. tb in early infection, it may also contribute to immune pathology and chronic unproductive inflammation. A shift from Th1 to Th17 cells indicates increased neutrophil recruitment and pulmonary damage . Excessive IL-17 production not only reflects the host response, for example the consequence of repeated exposure to the Bacille Calmette–Guérin vaccine or M. tb [15, 16], it also reflects the nature and composition of the pathogen. Increased IL-17 production has been associated with MDR strains of M. tb that lead to high antigen load and subsequent cytokine production ; more detailed studies are needed to determine which M. tb-associated structures lead to increased IL-17 production and subsequent pro-inflammatory signals to T and B cells.
The ‘double-edged sword’ of inflammation in infectious diseases
Although a Th1 response is believed to enable containment of M. tb, strong Th1 responses have been identified in patients in whom the pathogen is not contained and who present with clinically severe forms of TB . Thus, adjunct therapies have been developed with the aim of reducing nonproductive pro-inflammatory responses in TB  using glucocorticosteroids , in part via reduction in tumour necrosis factor-alpha (TNF-α) . IL-18 has been implicated in the transition of latent to active TB: latent TB infection reactivation is orchestrated not only by the pathogen but also by the local development of pro-inflammatory signals along with reduced expression of the apoptosis inhibitor B-cell lymphoma 2 (Bcl-2). One of the key cytokines promoting Bcl-2 expression is IL-7, which is strongly expressed in granuloma lesions from patients with latent TB . We showed ex vivo, using material from humans  as well as from a nonhuman primate model , that local expression of IL-7 is associated with improved clinical outcomes and reduced local (pulmonary) production of transforming growth factor-beta (TGF-β). Increased TGF-β production contributes to immune suppression and re-organization of damaged pulmonary tissue. IL-7 (or IL-15)-dependent cellular immune responses are able to drive potent anti-M. tb cellular reactivity in an M. tb treatment setting using passive transfer of M. tb-primed immune cells (to animals infected with virulent M. tb) , emphasizing the critical role of cytokines in coordinating the clinical presentation and outcome in TB.
Decrease in nonproductive, pro-inflammatory host immune responses in infections has been suggested to be of central importance considering increased resistance to antibiotics in general . New insights into the nature of infections and pathogen-induced damage to the host need to be conceptualized into clinical medicine. Chronic inflammation has been shown to lead to T-cell signalling defects and aberrant immune effector functions [25, 26], including decreased T-cell receptor (TCR) zeta chain expression  which could be restored with IL-2 . Reduced immune suppression has not only been shown in TB (e.g. reflected in detection of anti-M. tb T cells using major histocompatibility complex (MHC) class I peptide molecules in the absence of functional T-cell responses) [28, 29], but also in blood from individuals with a sudden high burden of infectious pathogens. Examples of this include patients with post-transplant lymphoproliferative disease, an Epstein–Barr virus (EBV)-driven B-cell expansion that may lead to immune paralysis and death due to uncontrolled expansion of EBV in an immune-compromised host , or those with cytomegalovirus infection, reflected by a lack of phosphorylation of signal transducer and activator of transcription (STAT) 5 .
Mycobacterium tuberculosis re-infection
Two alternative but not mutually exclusive models of M. tb re-infection have been discussed: M. tb primary infection may increase the risk of M. tb re-infection and/or infection ‘recruits’ high-risk individuals to a group of those at risk of M. tb (re-infection) . Following previous exposure to M. tb (or other pathogens), individuals become more prone to M. tb infection . Gender , human immunodeficiency virus (HIV) infection  and host response pattern markers [36, 37] all contribute to the increased risk of infection. Increased risk of re-infection with M. tb has been identified in western countries, including the USA  and Denmark . Severe clinical presentations of TB are also associated with strong pro-inflammatory responses. Patients with advanced TB disease show higher serum levels of TNF-α and TGF-β. This ‘hyperinflammatory’ reaction appears to be a sign of TB progression in humans; the destructive activity of pro-inflammatory cytokines has been reviewed . Immune pathology targeting M. tb-derived antigens is a typical clinical feature that can manifest as immune reconstitution inflammatory syndrome (IRIS) after antiretroviral treatment of HIV–M. tb co-infected patients . Corticosteriod treatment immune modulation in these individuals has recently been shown to act via reduction in pro-inflammatory cytokines, but not via reduction in M. tb antigen-specific adaptive T-cell responses .
Adjunct TB treatment and unmet clinical needs
Recent data show that latent TB represents a broad spectrum of M. tb infection. There appears to be a truly latent TB, which may remain so during the lifetime of the individual, with M. tb staying dormant; in other cases, individuals present with latent TB which is at risk of becoming clinically active (‘re-activation’) . Current technologies and diagnostic tools including platforms are not able to distinguish between these different states. Therefore, there is unmet need to define markers that may guide decision-making in the following clinical situations: (i) initiation of pre-emptive antibiotic therapy; (ii) prevention of disease spreading; (iii) identification of individuals at risk of developing clinical disease (i.e. individuals undergoing immune suppression using anti-TNF-α-directed drugs); (iv) co-infection with other noncommunicable diseases [44, 45]; and (v) use of immune/biomarker analysis to identify individuals who would benefit from adjunct therapy, in addition to standard drug treatment regimens [46-48]. The concept of limiting tissue destruction, to accelerate M. tb clearance and lead to better clinical outcomes, was investigated more than a decade ago. A number of clinical trials of corticosteroid treatment combined with standard antibiotic regimens to achieve faster TB culture conversion were conducted . Adjunct treatment of TB patients with etanercept; i.e. anti-TNF blockade via an anti-TNF receptor antibody, suggested a reduced time frame of sputum culture conversion in HIV–M. tb co-infected individuals . Similar promising data were obtained with thalidomide, which acts in part via reduction in TNF-α levels [51, 52].
Two strategies may facilitate M. tb clearance: enhancement of anti-M. tb immune responses with the aim of eradicating the pathogen and a targeted decrease in inflammation (organ-saving strategy, see above). The timing of the intervention is also of critical importance. For instance, IFN-γ treatment of patients with sepsis can be deleterious; it may worsen the ‘cytokine storm’ that contributes to clinical instability. However, these patients benefit clinically with better survival if IFN-γ treatment is initiated within a specific ‘window’ of immune paralysis during the later clinical course of sepsis ; at this time-point, the early acute pro-inflammatory reaction leads to counter-regulatory events, in part reflected by downregulation of the HLA-DR molecules on circulating CD14+ monocytes. Similarly, immunosuppressive (cyclophosphamide) treatment of patients with lethal influenza has also been reported to save lives in cases of severe disease . A similar situation may be found in patients with aberrant high Th1-type immune responses associated with target (pulmonary) tissue destruction and subsequent ‘immune paralysis’. T cells from patients with active TB disease show downregulation of the TCR zeta chain , associated with loss of or decreased immune effector functions by T cells. Furthermore, T cells from patients with active TB have been shown to produce only TNF-α (and no other pro-inflammatory cytokines including IFN-γ) in response to molecularly defined M. tb targets: reduced immune competence is an integral part of the clinical spectrum of active TB .
Current concepts and adjunct treatment modalities
Immune activation-based therapies, such as IL-2  or application of IFN-γ via aerosol, for improving TB treatment outcomes or shortening the duration of therapy have been described with limited results . A thorough review of the clinical use of corticosteroids for the prevention of mortality in individuals with TB showed no clinically relevant adverse effects of the systemic use of these agents for TB treatment; however, TB mortality and both pulmonary as well as extrapulmonary manifestations were reduced . The use of adjunct therapies in TB and the important effect of inflammation on clinical outcome were substantiated in a preclinical (zebra fish) model showing that susceptibility to M. tb infection is associated with the quality and degree of inflammation. Mutations in the leukotriene (LT) A4 hydrolase locus (LTA4H, which controls production of eicosanoids) affected the interactions between excess damaging LTB4 and lipoxins, which mediate TNF-α production. A reduction in inflammation led to better outcomes in a preclinical animal model, and subsequent analysis of a single-nucleotide polymorphism in the human LTA4H promoter supported the notion that inflammatory responses (associated with LTA4H variants) are correlated with increased rate of survival and responsiveness to biological therapy in TB patients . The LTA4H-associated differences in immune responses affect both the innate (i.e. containment of M. tb by macrophages) and adaptive cellular immune responses. LT-driven differentiation of immune effector T cells may in part be responsible for this observation. The usefulness of the concept of reducing damaging inflammation is further supported by the observation that aspirin is beneficial as an adjunct therapy in patients with meningeal TB . Levels of prostaglandin (PG) E2 are increased in the lungs from patients with TB, and it has been shown to inhibit phagocytosis, M. tb clearance and the production of pro-inflammatory cytokines. Further clinical trials are needed to examine the effect of nonsteroidal anti-inflammatory drugs (NSAIDs) in TB treatment. NSAIDs interfere with cyclooxygenase (COX)-1 and COX-2 which convert arachidonic acid to PGH2, prostacyclin and thromboxane A2 (the latter mediates vasoconstriction and thrombocyte aggregation). Ibuprofen (IBP) is able to inhibit both COX-1 and COX-2; selective COX-1 inhibition is associated with bleeding and ulceration, whereas selective COX-2 inhibitors may be associated with hypertension and cardiac complications.
Another marker of nonproductive immune activation (and potentially counter-regulatory mechanisms) is the decreased expression of the suppressor of cytokine signalling (SOCS) molecules. It was found that SOCS1 mRNA is lower in individuals with extrapulmonary TB. This has been confirmed in independent studies in which SOCS3 expression was shown to be critically important in cells of the lymphoid and myeloid lineage as a marker of decreased immune competence to control M. tb [60, 61]. Further, functional studies showed that a general lack of STAT5 phosphorylation determines reduced ‘immune competence’ and suggests a reduced response to pathogens in general [26, 62]. Immune-competent T cells respond with strong STAT5 phosphorylation in response to IL-7 and IL-2; a reduced response to cytokines is associated with (or is due to) the inability to produce cytokines in response to pathogens. Thus, clinical TB disease is complex and shaped by the host–pathogen relationship. Several markers have been identified, using different technologies, that would enable better interventional strategies to be designed for TB treatment. For instance, levels of cytokines in serum (IFN-γ, TNF-α, IL-18, TGF-β or IL-10) have been associated with latent or active TB [11, 63-65]. Measurement of infection parameters would indicate the extent of TB disease; for example, response to anti-TB treatment could be measured as the level of C-reactive protein [66, 67]. Candidate humoral parameters have been shown to be associated with the M. tb infection status and response to treatment [4, 25]. Immune cell gene expression signatures cell signatures have been used to identify SOCS3 as an attractive candidate for determining the capacity of the immune system to control M. tb, and IFN-γ signatures have been associated with M. tb infection status and the extent of disease [60, 68]. Genetic markers may contain mutations in the LTA4H locus controlling TNF-α production and possibly the response to anti-inflammatory treatment . Functional response analysis may include determination of SOCS3 protein expression in immune cells  and possibly HLA-DR expression on CD14 monocytes, which has been shown to be a clinically relevant marker of immune exhaustion and subsequent immune intervention in patients with sepsis .
HDTs in TB: exploiting controlled and host-beneficial inflammation to improve treatment outcome
Intracellular M. tb bacilli are known to reside within phagosomal compartments in a broad range of human cell types, especially those of the myeloid lineage such as macrophages and dendritic cells [69-73]. HDTs are designed to harness biologically relevant checkpoints in the host to complement currently available interventional methods. These treatments would therefore primarily target patients with drug-resistant TB or those with clinically challenging pulmonary conditions associated with M. tb infection. HDTs utilize intrinsic cellular mechanisms such as autophagy, apoptosis and possibly pyroptosis to kill bacilli persisting in host cells (Fig. 1). These processes have been associated with the host control of TB [74-77], thus providing the basis for HDTs to achieve potentially optimal benefits without harming the host (summarised in Table 1).
Table 1. Clinically-approved drugs for host-directed indications in tuberculosis (TB). Licensed drugs for treating malignancies other than TB provide a highly promising platform to explore combinatorial therapy. Note that this is not a complete list of the available drugs and their biological effects
Although excessive inflammation is detrimental to the host, aspects of inflammatory responses may be exploited to allow early and efficient control of M. tb infection. Pharmacological agents that promote pro-inflammatory activity within M. tb-infected cells could be of great use in eliminating ‘silent bacterial reservoirs’ in the host, and acting at the molecular level with significant implications for cell function and physiology.
Histone deacetylase inhibitors
The biochemistry of histones is important in the epigenetic control of gene expression, governed by chemical modifications. Amongst these modifications are lysine acetylation, serine phosphorylation and methylation of arginine and lysine . Of particular clinical relevance is the acetylation of lysine residues on amino-terminal tails of histones. This dynamic reaction is jointly catalysed by two groups of enzymes: histone deacetylases (HDACs) and histone acetyltransferases . To date, up to 18 different HDACs have been identified and categorized into four classes [79, 80]. As a consequence of HDAC acetylation of lysine residues, histones lose their ability to bind in a compact manner to chromosomal DNA, resulting in DNA unwinding and thereby allowing access to the RNA polymerase complex. HDACs can be pharmacologically interrupted using specific chemical compounds termed HDAC inhibitors (HDIs). Clinically, the use of HDIs has been most common and successful in experimental cancer therapy. By inhibiting histone acetylation in cancer cells, gene expression may be enhanced allowing better immunological control of tumours via apoptosis and autophagy [81, 82]. In this regard, HDIs such as valproic acid (VPA) and suberoylanilide hydroxamic acid (SAHA) are being tested to treat various haematological pathologies, such as T-cell lymphomas, lymphoproliferative disorders and multiple myelomas, with much promise as a novel type of anticancer drug [83-86]. From an infectious disease viewpoint, HDIs have been investigated in the context of HIV infection of T cells. VPA was the first class I HDI shown to induce HIV-1 reactivation in CD4 + T cells from latently infected individuals [87, 88]. Similar results were obtained with SAHA, using the different in vitro approach of a genetically modified primary human CD4 + T-cell line harbouring green fluorescent protein-positive latent HIV-1 provirus . After treatment of these cells with SAHA, there was a marked increase in viral replication and, subsequently, improved immune attack by virus-specific autologous CD8 + T cells due to high antigen turnover [89, 90]. VPA and SAHA are now being clinically tested in HIV-infected individuals in Denmark [79, 91]. As M. tb can replicate within host cells, or remain latent therein, VPA could be considered a suitable candidate for investigation for treatment of drug-resistant TB as a means of enhancing intracellular immunological control of bacterial proliferation. For example, exposing infected antigen-presenting cells to HDIs such as VPA may promote expression of genes involved in initiation of xenophagy, the autophagic mechanism that triggers phagolysosome formation and bacterial death . This process facilitates antigen processing and presentation and activates in turn cytokine production which engages T and B cells to amplify antimycobacterial activity (Fig. 1).
Clofazimine (CLO), an antibacterial drug used in the treatment of leprosy , has been shown to drive inflammation in human THP-1-derived macrophages via caspase 3-dependent apoptosis, mediated by the nuclear enzyme poly (ADP-ribose) polymerase [94, 95]. A recent study demonstrated substantial efficacy of CLO in mice infected with an isoniazid-resistant M. tb strain, achieving up to a 5-log unit reduction in pulmonary bacillary load when administered in combination with first- and second-line TB drugs . The efficacy of CLO has already been investigated in MDR TB patients with overall treatment success rates of up to 62% , whilst recently it was shown in a South Korean study that treatment outcome was improved when CLO was combined with linezolid and second-line drugs . It has been reported that virulent M. tb induces expression of Bcl-2 by infected murine macrophages as a means of inactivating apoptosis, thus allowing stable intracellular M. tb survival [99, 100]. Additionally, there is evidence that Bcl-2 protects murine macrophages from nitrosative stress by inhibiting apoptosis . Pharmacological inhibition of Bcl-2 using the experimental anticancer drug ABT-263 is capable of selectively killing metabolically inactive leukaemia stem cell (which give rise to acute and chronic myeloid leukaemia, ) populations via oxidative phosphorylation and induction of apoptosis . Myeloid cells constitute the main cellular repertoire generally targeted by M. tb for intracellular survival and persistence. As such, Bcl-2 inhibition presents yet another strategy by which intracellular M. tb bacilli may be destroyed (Table 1).
Bacterial toxins and nucleic acids are potent stimuli for intracellular assembly of the NACHT, LRR and PYD domains-containing protein 3 (NALP3) inflammasome complex which leads to caspase 1 activation [104, 105]. Caspase 1 cleaves cytosolic pro-IL-1β and pro-IL-18 to release mature IL-1β and IL-18, respectively, which are both highly potent pro-inflammatory mediators . Pyroptosis, a caspase 1-dependent cell death mechanism, is closely associated with control of intracellular pathogens including M. tb in macrophages and dendritic cells . M. tb can activate the NALP3 inflammasome complex and trigger IL-1β release via secretion of early secreted antigenic target-6 [107, 108]. Animals lacking either IL-1β or its receptor fail to efficiently control M. tb burden in the lung . By contrast, M. tb zinc metalloprotease 1 exerts inhibitory effects on NALP3 inflammasome activity, leading to caspase 1 downregulation and arrest of phagosome maturation in murine macrophages to establish intracellular infection and persistence . The role of IL-1β in protection against TB has been extensively explored, and it has been found that the NALP3 inflammasome appears to participate in early control of intracellular M. tb replication [107, 108, 111, 112]. It was recently shown that the principal trigger of NALP3 inflammasome activation, shared by bacterial toxins as well as pharmacological agents, is efflux of K+ ions from the (host) cytosol . K+ efflux, which is invariably coupled to Na+ influx, is regulated by the activity of ion pumps located in the cell membrane. Diazoxide, which has been used for the treatment of hypertension, stimulates the Na+/K+ ATPase and therefore K+ efflux . Diazoxide could be investigated in the context of TB as a possible novel NALP3 inflammasome agonist in this setting.
Regulation of inflammatory processes may also provide an ideal treatment for drug-resistant TB. LTA4 arising from arachidonic acid catabolism in humans has been shown to markedly influence inflammation in TB via regulation of TNF-α production . Mutations in lta4h, the genetic locus in humans that encodes LTA4, determine whether or not equilibrium is maintained between physiological levels of the LTA4 derivatives lipoxin A4 (LXA4, anti-inflammatory mediator) and leukotriene B4 (LTB4, pro-inflammatory mediator). A balance between LXA4 and LTB4 leads to optimal antitubercular immune responses and to adequate amounts of TNF-α for facilitating effective clearance of M. tb with minimal immunopathology . Aspirin possesses anti-inflammatory activities and is able to prevent excessive neutrophil migration, thus playing a role in inflammatory extravasation of cells into tissue . As such, aspirin could be tested for efficacy in preclinical TB models as an adjunct therapeutic agent in combination with currently administered anti-TB drugs. Pyrazinamide (PZA) is a highly potent antimycobacterial drug that has been part of the first-line anti-TB regimen since the 1970s . Inside the mycobacterial cell cytosol, pyrazinimidase converts PZA to pyrazinoic acid which then interrupts the cell membrane potential leading to bacterial death . However, it was only recently discovered that PZA also coordinates anti-inflammatory effects during M. tb infection in mice, represented by downregulation of pro-inflammatory cytokines such as TNF-α, IL-1β, macrophage chemotactic protein 1 (MCP-1) and IL-6 . Thalidomide is another antimycobacterial drug used to treat leprosy, which has been shown to possess anti-inflammatory properties via downregulation of TNF-α production, as well as anti-angiogenic effects possibly via inhibition of vascular endothelial growth factor (VEGF) activity [21, 22]. TNF-α is a potent mediator of VEGF release , and VEGF has already been established as having a role in perpetrating pulmonary TB [24-26]. Of note, TNF-α- and IL-12p40-driven angiogenesis has been associated with active TB disease . VEGF levels, measured in in vitro stimulation assays with blood from adults and children with latent TB infection, may provide a potential biomarker of progression to active TB disease . An experimental thalidomide analogue, IMiD3, was tested in a rabbit model of tuberculous meningitis and found to contribute to reduced amounts of TNF-α in the cerebrospinal fluid (CSF) . This observation was directly coupled with decreased TB disease severity and higher survival rates compared to isoniazid and rifampicin treatment although colony-forming unit burden in the CSF was similar between the treatment groups. The calcium channel blocker verapamil is used for treating a number of conditions including hypertension and cardiac arrythmias, and has been shown in a mouse model of TB to complement the activity of currently administered antimycobacterial drugs by interfering with mycobacterial efflux pumps [120, 121]. Also, the NSAID ibuprofen, a routinely used over-the-counter painkiller, appears to exert antimycobacterial effects against MDR M. tb, both in vitro  and in a murine experimental model of active TB . Imatinib mesylate, which is licensed for the treatment of cancer, inhibits the activity of the oncogenic BCR–ABL tyrosine kinase  and has been shown to promote reduction in inflammation and control of bacterial burden in lungs, liver and spleen of mice infected with rifampicin-resistant M. tb .
Taken together, current findings suggest that timely interventions coupled with novel strategies based on a better understanding of host-targeted factors and basic translational research efforts may contribute to improved management of drug-resistant TB in the near future.
The inflammatory response: models and new concepts
The increased risk of M. tb re-infection in individuals with HIV co-infection raises two important issues. First, animal models may not be adequate to mimic repeated exposures to M. tb or to mycobacteria other than tuberculosis; secondly, repeated host infections may affect immune-competence and antigen-specific (protective) immune responses. The usefulness of mice with targeted gene deletions (such as deficiencies in pathogen recognition receptor adaptor molecules, MyD88, TIR8, CARD9  or pro-inflammatory cytokines that help to coordinate granuloma formation) in elucidating M. tb–host relationships is clear (for review see ). However, the usefulness of murine studies for understanding human immunopathology is questionable, at least in complex treatment models, as some key immune effector molecules are different in mice and humans; response in mice poorly resembles human inflammatory diseases . For instance, recent studies addressing the effect of anti-PD1-directed immunotherapies in patients with cancer compared with preclinical animal models underlined the need for careful consideration of whether animal models may contribute to decisions regarding whether or not to conduct future clinical trials, whereas anti-PD1-directed therapies showed limited effects in preclinical models, promising responses were observed in clinical anti-PD1 trials suggesting the need for caution in using murine studies for the development of novel therapies .
Challenges of immune pathology
The pathogen–host relationship in TB is complex, involving host genetic determinants, pathogen diversity, mutations and re-infection and mediation of pro-inflammatory reactions not only via viable bacilli but also via their structural products. The M. tb envelop contains lipids that activate cellular host immune responses; this is often underestimated as the extent and nature of these lipids are difficult to assess. Moreoever, M. tb lipid structures are also modified during early pro-inflammatory reactions: lipids are altered via cis-cyclopropane modification of mycolic acids on trehalose dimycolate which is specifically mediated via proximal cyclopropane synthase of alpha mycolates. The loss of this process (trans-cyclopropanation) increases M. tb-mediated activation of macrophages . Cyclopropane synthesis appears to be crucial in chronic M. tb persistence and inflammation-associated immune pathology , and some M. tb enzymes may themselves represent targets for antibody and T-cell responses in M. tb-infected individuals. The impact of these M. tb-derived soluble factors on inflammatory responses is difficult to assess ex vivo and may require novel strategies for determining which effects are mediated by viable cell (M. tb)–cell (host) interactions, or by M. tb-associated molecules. The production of M. tb-associated factors is also modulated by the host environment which influences M. tb gene expression. This has been shown for the successful use of metronidazole (which acts on anaerobic and proliferating bacteria). Metronidazole (MTZ) has been shown to kill (in vitro) M. tb in a hypoxic but not in an aerobic environment . Although sputum conversion was not found to be affected by metronidazole (MTZ) treatment, patients receiving MTZ experienced an improved clinical disease course  (which was confirmed in a nonhuman primate model but not in a murine model of TB ). It has been suggested therefore that MTZ is active in preventing reactivation of latent M. tb infection .
Use of cellular therapy
We have recently completed a Phase I, open-label, safety study in 30 individuals with pulmonary MDR/XDR TB using autologous mesenchymal stromal cells with the aim of providing adjunct treatment modalities for these patients . The primary end-point of safety was met in all 30 individuals. Although this was a Phase I safety study, we also included 30 individuals who fulfilled the inclusion criteria for studying the natural course of MDR/XDR TB, yet did not receive (autologous) MSC treatment. All individuals in the study (plus or minus adjunct MSC treatment) received standardized drug treatment according to the World Health Organization (WHO). We found that the MSC-treated patients performed clinically better and showed a higher rate of M. tb culture conversion. A recent meta-analysis of data regarding autologous and allogeneic MSC therapy, from publications found by searching MEDLINE, EMBASE and the Cochrane Central Register of Controlled Trials (to June 2011), demonstrated no major side effects in patients other than an association between MSC infusion and transient fever . Strategies to reduce damaging inflammation and facilitate tissue repair could therefore be helpful in augmenting standard TB drug regimens.
Several open questions remain: (i) What is the nature of immature immune responses in children compared with adults?  and (ii) What are the roles of co-infection and co-morbidity in patients with TB? However, these are beyond the scope of the current review. Any adjunct therapy may also have an impact on co-infections and other concomitant diseases which may limit the use of some ‘repurposed’ drugs. In addition, drug–drug interactions should be considered carefully, for example as reported for concomitant aspirin and isoniazid treatment . Future adjunct therapies may also explore combinatorial treatments, given the successful clinical trials using immune modulatory agents in patients with cancer . Therefore, robust and clinically meaningful technical platforms are needed that would enrich and guide future studies of adjunct therapies in TB. The biological mechanisms underlying the clinical presentation of TB (latent vs. active TB, IRIS and risk of TB re-activation and re-infection) remain poorly understood. Better understanding of the immune pathology of TB will lead to more informed decisions regarding which type of HDTs may bring the greatest benefit for those patients who are not able to respond to current standard treatment protocols.
This work was funded by HLF, Vinnova, VR, SIDA, Sweden and EDCTP (TBNeat) (to MM), and the UK Medical Research Council, European Union 7th Framework project Rid-RTI, European Developing Countries Clinical Trials Partnership (EDCTP), UBS Optimus Foundation, Switzerland, and the NIHR Biomedical Research Centre, University College Hospitals, London, UK (to AZ).
Conflict of interest statement
All authors have no conflict of interests to declare. None of the funding bodies had any role in the preparation of the manuscript.