The immune response in tubercular uveitis and its implications for treatment: From anti-tubercular treatment to host-directed therapies

( Mtb ) is present in the ocular tissues, elicits a heightened immune response without Mtb invasion in ocular tissues, or even induces an anti-retinal autoimmune response. Gaps in the immuno-pathological knowledge of TB-uveitis likely delay timely diagnosis and appropriate management. In the last decade, the immunopatho- physiology of TB-uveitis and its clinical management, including experts ’ consensus to treat or not to treat certain conditions with anti-tubercular treatment (ATT), have been extensively investigated. In the meantime, research on TB treatment, in general, is shifting more toward host-directed therapies (HDT). Given the complexities of the host- Mtb interaction, enhancement of the host immune response is expected to boost the effectiveness of ATT and help overcome the rising burden of drug-resistant Mtb strains in the population. This review will summarize the current knowledge on the immunopathophysiology of TB-uveitis and recent advances in treatment modalities and outcomes of TB-uveitis, capturing results gathered from high- and low-burden TB countries with ATT as the mainstay of treatment. Moreover, we outline the recent progress of HDT development in the pulmonary TB field and discuss the possibility of its applicability to TB-uveitis. The concept of HDT might help direct future development of efficacious therapy for TB-uveitis, although more in-depth research on the immunoregulation of this disease is still necessary.


Keywords:
Anti-tubercular treatment Autoimmunity Host-directed therapies Tubercular uveitis Tuberculosis A B S T R A C T Tubercular uveitis (TB-uveitis) remains a conundrum in the uveitis field, which is mainly related to the diverse clinical phenotypes of TB-uveitis. Moreover, it remains difficult to differentiate whether Mycobacterium tuberculosis (Mtb) is present in the ocular tissues, elicits a heightened immune response without Mtb invasion in ocular tissues, or even induces an anti-retinal autoimmune response. Gaps in the immuno-pathological knowledge of TB-uveitis likely delay timely diagnosis and appropriate management. In the last decade, the immunopathophysiology of TB-uveitis and its clinical management, including experts' consensus to treat or not to treat certain conditions with anti-tubercular treatment (ATT), have been extensively investigated. In the meantime, research on TB treatment, in general, is shifting more toward host-directed therapies (HDT). Given the complexities of the host-Mtb interaction, enhancement of the host immune response is expected to boost the effectiveness of ATT and help overcome the rising burden of drug-resistant Mtb strains in the population. This review will summarize the current knowledge on the immunopathophysiology of TB-uveitis and recent advances in treatment modalities and outcomes of TB-uveitis, capturing results gathered from high-and low-burden TB countries with ATT as the mainstay of treatment. Moreover, we outline the recent progress of HDT development in the pulmonary TB field and discuss the possibility of its applicability to TB-uveitis. The concept of HDT might help direct future development of efficacious therapy for TB-uveitis, although more in-depth research on the immunoregulation of this disease is still necessary.

Introduction
Tuberculosis (TB) is a major infectious uveitis cause, especially in countries with high TB endemicity. (Tsirouki et al., 2018) Tubercular uveitis (TB-uveitis),  the term used to describe intraocular inflammation due to proven and presumptive Mycobacterium tuberculosis (Mtb) infection, is responsible for about 22.9-48.0% of infectious uveitis in India and Indonesia. (La Distia Nora et al., 2018a;Putera et al., 2022;Tsirouki et al., 2018) The visual morbidity of TB-uveitis can be severe. It has been reported that the best visual acuity of nearly one-third of TB-uveitis patients was worse than 3/60.  Another report found that uveitic macular edema and secondary glaucoma might develop in around 30% of TB-uveitis patients. (Gunasekeran et al., 2018) Meanwhile, the prevalence of TB-uveitis, based on positive interferon-gamma release assay (IGRA) results and exclusion of other entities in non-endemic countries, is low. (Alli et al., 2022) However, the proportion of unknown causes of uveitis was higher in IGRA-positive than in IGRA-negative individuals (59% vs 39%). (Groen-Hakan et al., 2020) This stresses the difficulty of assigning a definitive diagnosis of TB-uveitis to a patient presenting with active uveitis and a positive IGRA but without other ancillary tests or findings supportive of TB infection.
There has been a significant advancement in TB-uveitis diagnosis during the past decade, as indicated by the emergence of new criteria and expert consensus. (Agrawal et al., 2021a;Jabs et al., 2021) However, the dilemmatic aspect of TB-uveitis diagnostics has still not come to an end, particularly when facing patients with the absence of active pulmonary or other extrapulmonary TB. (Testi et al., 2020) Caseating granuloma formation in ocular tissue has been described. (Basu et al., 2015;Wroblewski et al., 2011) Yet, the pathomechanisms underlying TB-uveitis are so far incompletely understood. It was proposed that ocular inflammation can be caused by: (1) direct Mtb infection of ocular tissues, (2) the induction of specific anti-retinal immune responses or an immunological reaction toward Mtb antigen (without live Mtb in the eye) or (3) a combination of those mechanisms. Although the different underlying mechanisms of TB-uveitis may benefit from different treatment approaches, it is so far not possible to clinically distinguish between the different pathomechanisms that may underlie TB-uveitis. (Agrawal et al., 2021c;Basu et al., 2015Basu et al., , 2020Ludi et al., 2023) This clearly emphasizes that improved pathophysiological knowledge of TB-uveitis is required to optimize diagnostic and treatment approaches for the future.
The conundrum of underlying pathogenesis and diagnostics in TBuveitis reflects its management approach. Anti-tubercular therapy (ATT) remains the mainstay treatment for TB-uveitis. (Agrawal et al., 2021b) In many clinical circumstances, prescribing ATT is a dilemma when the diagnosis is still in doubt. (Agrawal et al., 2021b) Interestingly, systemic corticosteroid treatment without ATT may also resolve TB-uveitis in a fraction of patients who present without active pulmonary/extrapulmonary TB. (Kee et al., 2016) Moreover, the combination of ATT and systemic corticosteroid treatment is often necessary to obtain optimal treatment results. (Betzler et al., 2021;Kee et al., 2016) Another rising concern is the duration of ATT, which has not been standardized in TB-uveitis. (Ang and Chee, 2017) Some reports described the benefit of prolonging the ATT duration beyond six months. (Agrawal et al., 2015;Ang et al., 2012b) In many clinical scenarios, the lack of proof of Mtb presence from ocular samples in suspected TB-uveitis patients raises the possibility of an autoimmune response. (Bigdon et al., 2022) Given the long duration and potential adverse reactions to ATT, patient-specific inflammatory processes need to be understood. Patients can be treated appropriately and in a balanced way between enhancing Mtb clearance in ocular tissue and dampening excessive immune responses, including autoimmunity. Despite continuous efforts to advent more effective anti-TB drugs, (Singh and Mizrahi, 2017) host-directed therapy (HDT) has emerged to overcome the unresolved quest for antibiotics in TB. HDT specifically targets the host immune response to either (1) enhance Mtb elimination through modulation of host immunity against Mtb, such as by increasing efficiency of intracellular Mtb killing mechanisms, improving T-cell responses, and strengthening granuloma to limit Mtb dissemination, or (2) limit immunopathology caused by an exacerbated Mtb-directed immune response. (Kolloli and Subbian, 2017;Young et al., 2020) Although there is still limited available (pre-)clinical evidence, (Kiran et al., 2016;Young et al., 2020) HDT seems applicable to TB-uveitis. In this review, we will give an overview of the immunopathophysiology of TB-uveitis and summarize achievements made within current TB-uveitis management, including HDT and its possible future application for TB-uveitis.

Overview of immunopathophysiology of TB-uveitis
The current understanding of TB-uveitis pathogenesis relies on the concept that ocular inflammation relates to infection of ocular tissue by Mtb and anti-retinal autoimmune responses that may even occur without apparent ocular Mtb infection. A detailed description of these concepts has been provided previously. (Agrawal et al., 2021c;Basu et al., 2020) Here, we summarize relevant basic knowledge on pathogenesis with specific emphasis on the role of retinal pigment epithelium (RPE) in regulating the immune response in TB-uveitis.

The RPE in TB-uveitis
RPE is a monolayer of specialized epithelial cells that is located between the neurosensory retina and choroid (Fig. 1). It originates from the neuroectoderm and serves many physiological functions, including the transport of nutrients and metabolites between the adjacent layers and recycling retinol for the proper function of photoreceptors, for which the RPE cells also exhibit phagocytic activity with the routine task of clearing photoreceptor outer segment fragments. (Yang et al., 2021) Moreover, RPE constitutes the outer blood-retinal barrier (BRB). The latter function is critical for ocular immune privilege as the tight junctions between the RPE cells maintain the physical barrier toward externally eliciting stimuli, such as infectious pathogens (Fig. 1). (Yang et al., 2021) Besides their regular physiological functions, RPE cells serve a pivotal immunological function by orchestrating innate and adaptive immune responses within the eye. Human RPE cells express toll-like receptors (TLRs), a set of conserved sensing receptors that play a crucial role in innate immunity by recognizing pathogen-associated molecular patterns derived from various microbes. Recognition of pathogens by TLRs initiates inflammatory cascades leading to the generation of pro-inflammatory cytokines and chemokines, including interleukin (IL)-6, IL-8 (chemokine C-X-C motif ligand (CXCL)-8), monocyte chemoattractant protein (MCP)-1, ICAM-1, CXCL9, CXCL10 and vascular endothelial growth factors (VEGF). (Detrick and Hooks, 2020) In addition, RPE cells express the C3a and C5a complement receptors. Activation of these receptors leads to the production of IL-1β, IL-6, IL-8, (MCP-1/CCL2), and granulocyte-macrophage colony-stimulating factor (GM-CSF). (Detrick and Hooks, 2020) RPE cells are also involved in regulating adaptive immune responses as they express human leukocyte antigens (HLA) classes I and II that can present non-self and self-antigens to T-lymphocytes. (Detrick and Hooks, 2020) Besides this, several lines of evidence point at an immunoregulatory function for RPE. For example, RPE can secrete interferon (IFN)-β, a type 1 IFN family member, that can suppress immune reactivity and exert a retinal protective effect to control the potentially detrimental effects of excessive inflammation. (Detrick and Hooks, 2020) Also, IFN-β suppresses CXCL-9 and intercellular adhesion molecule (ICAM)-1 expression in RPE. These IFN-β driven effects reduce BRB permeability during inflammation and limit the migration of immune cells, including T-lymphocytes and natural killer (NK) cells, into the eye. (Detrick and Hooks, 2020) Moreover, RPE cells can secrete α-Melanocyte-stimulating hormone and neuropeptide Y that enhance the apoptosis of activated effector T-lymphocytes. (Taylor et al., 2021).

Direct infection of Mtb in TB-uveitis
Mtb can reach ocular tissue through hematogenous dissemination from distant infection foci. (Agrawal et al., 2021c;Basu et al., 2020) Further investigation using a zebrafish model revealed that Mtb could seep into the eye despite an intact BRB. This was followed by monocyte infiltration and subsequent granuloma formation in retinal tissue. (Takaki et al., 2018) Another study showed that the capability of Mtb to enter retinal tissue by extracellular route is dependent on the mycobacterial ESX-1 secretion system. (Damera et al., 2021) The localized Mtb would result in granulomatous inflammation. It has been demonstrated that such inflammation could occur in cases where only a single Mtb bacillus was detected in the ocular specimen, particularly in the RPE layer. (Wroblewski et al., 2011) Another histopathological study also confirmed the presence of Mtb in necrotic RPE of the eye clinically seen as tuberculoma (choroidal granuloma) and serpiginous-like choroiditis. (Rao et al., 2006) RPE cells phagocytose the invading Mtb in the posterior segment of the eye (Fig. 2). TLR-2 and TLR-4 appear critical for Mtb infection, as blockage of these TLRs suppresses Mtb growth and proliferation in RPE. (Nazari et al., 2014) Mtb-infected RPE secrete various cytokines and chemokines (including IL-1β, IL-6, IL-8, tumor necrosis factor (TNF)-α, IFNs, CXCL9, CXCL10, and CXCL11 and attract immune cells, including neutrophils, NK cells, and T lymphocytes. (Agrawal et al., 2021c;Detrick and Hooks, 2020;La Distia Nora et al., 2018d) Mtb-infected RPE appeared to be more immunoregulatory than Mtb-infected macrophages. (La Distia Nora et al., 2018d) A previous study demonstrated that Mtb-infected RPE cells had a dominant type 1 IFN (IFNα/β) signaling response and displayed enhanced expression of genes involved in the regulation of cell-death and survival that are important in limiting retinal damage due to inflammation. (La Distia Nora et al., 2018d) Both IFN-α and β bind to IFNAR1 and IFNAR2 subunit receptors, which signal through STAT1 and STAT2.(Moreira-Teixeira et al., 2018) As discussed elsewhere, type 1 IFN signalling can be either protective or detrimental to TB disease. Type 1 IFN primes the secretion of protective cytokines IL-12 and TNF-α. On the other hand, it promotes the production of IL-10 which represses macrophage/monocyte responsiveness to the antibacterial effects of IFN-γ and therefore is associated with disease progression. (Moreira-Teixeira et al., 2018) It was postulated that the highly activated type 1 IFN signalling in RPE benefits the early stage of Mtb infection by reducing potential local tissue damage through its immunoregulatory effect and may inhibit Mtb outgrowth by eliminating Mtb through activated cell-death signalling pathways. (La Distia Nora et al., 2018d) However, chronic infection or Mtb latency in the RPE might be expected to happen in the later stage, possibly causing prolonged ocular inflammation. Furthermore, IFN-γ-activated RPE can activate retinal antigen-specific T-lymphocytes by enhanced expression of HLA class II molecules. (Sun et al., 2003) Several reports have shown that active TB is associated with specific activation of type-1 IFN-stimulated genes (ISGs) in peripheral blood cells. The more severe the TB disease, the higher the expression levels of ISGs. (Berry et al., 2010;La Distia Nora et al., 2018b) In line with these observations and the significant activation of IFN signalling upon in vitro infection of RPE, (La Distia Nora et al., 2018d) our group also showed that peripheral blood cell expression of ISGs can be of use to stratify QFT-positive patients with uveitis of unknown cause into a group with a high likelihood of having active TB-uveitis (high ISG expression) or a low likelihood of having active TB-uveitis (low ISG expression). (La Distia Nora et al., 2018b) Consequently, it was proposed that this approach may prevent patients with low ISG expression from unnecessary ATT treatment, thus preventing them from possible severe side effects associated with this treatment. (La Distia Nora et al., 2018b) Interestingly, discrete granuloma formation within the nerve fiber  layer adjacent to the retinal vasculature without RPE involvement can also occur. This intraretinal granuloma lacks multinucleated giant cells but harbours elongated cells that share morphological similarity to activated microglia. (Basu et al., 2012) It seems that the interaction between pericytes and retinal endothelial cells, independent of RPE, controls this intraretinal granuloma formation. Currently, studies regarding the role of various cells other than RPE in the formation of TB-uveitis granulomas are lacking. Moreover, the composition and interactions between BRB constituent cells, namely retinal endothelial cells, microglia, Müller cells, pericytes, and RPE, in TB infection have not been extensively explored. (O'Leary and Campbell, 2023) It is necessary to have a complete in vitro model of inner BRB-Mtb interaction to gain an in-depth look at the mechanisms that underlie TB-uveitis granuloma formation.
It is worth mentioning that the current understanding of TB-uveitis is limited to posterior segment inflammation. From the clinical perspective, the involvement of anterior segment inflammation in TB-uveitis, reflected by the presence of iris nodules, large keratic precipitates (KPs), and profound anterior chamber cells without any posterior segment involvement, could not be explained simply by breached BRB. BRB disruption would be followed by ocular inflammation, but the reverse is not necessarily true. (Forrester et al., 2018)

Autoimmune response in TB-uveitis
Since detection of Mtb or its genome in the eye is challenging and often remains undetectable, uveitis is often assumed to be associated with systemic TB infection based on IGRA or tuberculin skin test (TST) positivity, even in cases without systemic abnormalities. The Collaborative Ocular Tuberculosis Study (COTS) group endorsed using the general term "tubercular uveitis" without the necessity of Mtb presence in ocular tissue/fluid.  However, there is a grey area in diagnosing latent TB-uveitis in patients without systemic TB disease. Patients with immunoreactivity against TB, based on IGRA/TST, have a higher risk of developing active TB disease later. Some experts challenge the term "latent" as Mtb may keep replicating and is not "dormant" before progressing to active TB. Moreover, with the possibility of an undetected paucibacillary state, an increased immune response may also contribute to ocular inflammation. (Basu, 2022) Given that some pieces of evidence suggest that uveitis might result from direct infection of the eye, recent investigations also suggest that there is plausible autoimmunity involved in the development of uveitis among individuals with latent TB (Fig. 3). The autoimmune process may involve hypersensitivity to ocular antigens due to immune cell priming that cross-reacts with Mtb antigens outside the eye. One report described a case of anterior uveitis in a patient following intravesical Bacille-Calmette-Guérin treatment for bladder carcinoma. This patient presented with mutton-fat KPs and the Koeppe nodule, which resembled granulomatous anterior uveitis typically found in TB-uveitis. The authors evaluated the potential of antigenic mimicry as the underlying cause of uveitis. Peripheral T-lymphocytes had high IL-2, IFN-γ, and TNF-α production in response to purified protein derivative (PPD) stimulation. Similarly, an elevated cytokine response towards three retinal autoantigens (retinal soluble antigen, interphotoreceptor retinoid-binding protein (IRBP), and cellular retinal-binding protein) was found. (Garip et al., 2009) Moreover, a considerably lower frequency and aberrant function of peripheral T regulatory cells (Tregs) might contribute to TB-uveitis.  (Sharma et al., 2018a) Intraocular cytokine analysis from TB-uveitis patients showed a higher level of autoimmune-associated cytokines, such as IL-2, IL-12, CXCL9, and TNF-α, than in idiopathic/unknown uveitis. (Ang et al., 2012a) Together, these data support the dysfunction of Treg immunity and strong T-helper (Th)-1 polarisation as contributing factors to TB-uveitis. (Malek, 2003) Intraocular T-lymphocyte responses upon stimulation with the Mtb protein Early Secreted Antigenic Target-6 (ESAT-6) or the retinal autoantigen IRBP was compared in patients stratified for being TB-uveitis (using new Standardization of Uveitis Nomenclature (SUN) classification criteria (see Table 1) or positive Mtb polymerase chain reaction (PCR) from ocular fluid), latent TB uveitis (not fulfilled SUN criteria despite being TST+), and non-TB-uveitis patients. T-lymphocytes of latent TB uveitis patients elicited the highest TNF-α, IL-17A, and IFN-γ polyfunctional responses (cells expressing ≥2 cytokines) upon ESAT-6 stimulation. (Alam et al., 2022) Whereas phenotyping of peripheral T lymphocyte showed no difference between TB-uveitis patients presenting with granulomatous versus non-granulomatous uveitis, presence versus absence of chest radiological changes suggestive of pulmonary TB, and between different anatomical sites of uveitis. (Hutchinson et al., 2021) Interestingly, intraocular IRBP-specific T-lymphocyte cytokine responses were equally observed among QFT positive patients with uveitis of unknown cause and patients with uveitis associated with pulmonary TB. (Alam et al., 2022) Tagirasa et al. identified intraocular autoreactive T-lymphocytes that responded to Mtb-derived ESAT-6 and crude retinal extract antigens. (Tagirasa et al., 2017) However, autoreactive T-lymphocytes towards crude retinal extract could not be detected in peripheral blood from patients with TB-uveitis, which was suggested to be related to a low frequency of retina-specific T-lymphocytes in peripheral blood. (Schrijver et al., 2021;Tagirasa et al., 2017) Ocular autoimmunity can be associated with the presence of antiretinal autoantibodies (ARAs). (Adamus, 2018) In further support of an autoimmune component in TB-uveitis is the higher prevalence of serum ARA in active and latent TB-uveitis in comparison to healthy controls. (La Distia Nora et al., 2018c;Ten Berge et al., 2016) Although the pathogenic relevance of ARA in TB-uveitis is unclear, it has been proposed that they are formed secondary to the response of inflammation-induced retinal damage and further support the occurrence of a T-lymphocyte-dependent anti-retinal immune response in a substantial number of TB-uveitis cases. (Schrijver et al., 2021) Furthermore, a recent animal study showed that panuveitis could occur following the injection of heat-killed Mtb (hkMtb) in the vitreous humor. Compared to mice without previous systemic exposure to hkMtb (unprimed mice), a significantly higher number of ocular T lymphocytes, NK cells, and macrophages were observed in primed mice. Also, higher levels of IL-17, VEGF, CXCL9, CXCL10, IL-12p40, and macrophage inflammatory protein-1α (MIP-1α/CCL3) were observed. Only the primed mice developed chronic uveitis. (Pepple et al., 2022) Whether these observations are related to an excessive immune response to TB-antigens, an induction of an anti-retinal immune response, or both, is unclear.

Clinical perspective of tubercular uveitis: From diagnosis to treatment initiation
TB-uveitis can present with unilateral or bilateral intraocular inflammation at various ocular locations: anterior, intermediate, posterior, as well as panuveitis. The presence of large-mutton fat KPs, iris nodule in the pupillary margin (Koeppe nodule) or anterior mid-iris (Busacca nodule), and broad-based synechia indicate anterior ocular granulomatous inflammation. The formation of cataract is common in chronic inflammation. Significant snowball or snowbanking with peripheral vasculitis is frequently present in tubercular intermediate uveitis. Tubercular posterior uveitis typically presents with retinal and choroidal inflammation. Chorioretinal lesions include choroidal tubercle/tuberculoma, multifocal choroiditis, serpiginous-like choroiditis, or lesions mimicking other entities. Tubercular retinal vasculitis can also manifest as an occlusive periphlebitis. Even though it is rarely reported, optic nerve involvement (papilitis, optic neuritis, or optic nerve tubercle) may also be present. (Betzler et al., 2021; In practice, diagnostic delay was variably reported across the globe as the diagnosis of TB-uveitis often relies on the presence of clinically granulomatous inflammation, the exclusion of other possible entities, and the corroborative evaluation of systemic TB infection (based on chest radiography, IGRA, or tuberculin skin test (TST)). (Betzler et al., 2021) As a granulomatous disease, TB-uveitis shares similarities with other diseases, such as sarcoidosis. (Mochizuki et al., 2019;Mortaz et al., 2016) The current diagnostic approach for TB-uveitis implies the challenge of directly detecting Mtb or its genome in ocular tissue/fluid. PCR has gained popularity as this method provides an easy and quick result for Mtb detection. However, based on our systematic review, the overall PCR positivity for detecting Mtb was only 55% with poor specificity when we set the response to ATT as the reference. (La Distia Nora et al., 2021) In India, where TB is highly prevalent, the reported positivity could be as high as 70%, particularly when the MPB64 primer was used, which is characterised by a lower detection limit.  Due to the challenge of Mtb detection in ocular tissue/fluid in TB-uveitis, direct Mtb detection in ocular samples is not consistently implemented,  although it remains essential and irreplaceable. Moreover, a recent statement by the British Thoracic Society (BTS) mentioned that ocular fluid analysis could be an option to help diagnose TB-uveitis in patients presenting with atypical ocular inflammation and a positive IGRA, particularly in those with high risk of TB, (Kon et al., 2022) even though a negative result will still be a clinical dilemma.
In 2015, Gupta et al. proposed a classification of intraocular TB based on direct and indirect evidence of TB infection (Table 1).(A.  However, this classification was not based on the pathogenesis of TB-uveitis but rather on the amount of clinical evidence. Generally applied, many patients frequently fulfil the probable and possible intraocular TB classifications. In these cases, a clear response towards ATT makes the diagnosis very likely. Considering the problem of making the decision to start ATT, the COTS group published a consensus on when to initiate ATT in patients without proven active pulmonary or extrapulmonary TB. The consensus is based on the country's endemicity of TB, location, type of ocular inflammation, and ancillary test results (chest radiography, IGRA, and TST). (Agrawal et al., 2021b) In brief, the consensus is mainly achieved when at least chest radiography shows evidence of TB infection with one or both IGRA and TST positive results in countries with a high TB-burden. Based on the presenting phenotype, only serpiginous-like choroiditis and choroidal tuberculoma phenotypes need to be either TST or IGRA positive without evidence of a chest radiological TB lesion to start ATT. (Agrawal et al., 2021b) Recently, the SUN working group has developed a set of criteria for TB-uveitis (Table 1). (Jabs et al., 2021) These criteria have strong discriminative properties to differentiate TB-uveitis from the other entities. The criteria emphasise several phenotypes that are strongly compatible with TB: anterior uveitis with iris nodules, serpiginous-like tubercular choroiditis, choroidal nodule (i.e., tuberculoma), in individuals with active systemic tuberculosis showing multifocal choroiditis, and occlusive retinal vasculitis. However, the authors stated that the applicability of the classification criteria in the clinical setting is limited as it was mainly developed for research purposes. The classification system uses relevant clinical appearances in the best agreement between experts on TB-uveitis diagnosis. The authors acknowledged that the criteria are more restrictive than the COTS guidelines described previously, however, the latter might be more suitable for clinical care. For instance, even though intermediate uveitis is not considered a phenotype strongly suggestive of TB-uveitis, a study by Sreenivasan et al. in an endemic TB country reported and highlighted the benefit of adding ATT in addition to systemic immunosuppressive medication for the treatment of intermediate uveitis patients with positive Mtb PCR from ocular fluid. (Sreenivasan et al., 2022) In light of systemic biomarkers, the combination of a peripheral blood type-1 IFN gene signature and serum C1q measurements may provide additional diagnostic value in classifying QFT + uveitis of unknown cause without apparent systemic TB disease. (Schrijver et al., 2020) In this situation, uveitis patients showing low serum C1q and low type-1 IFN signature score may not need ATT as they have a low likelihood of having active TB disease (Fig. 4). (Schrijver et al., 2020) However, further prospective validation is required before clinical implementation.

Systemic treatment
ATT, the mainstay treatment of TB-uveitis, has been studied across different populations ( ) during the initial 2-month intensive phase, followed by a combination of two or three drugs (HR ± Z) for a minimum of 4 months. The World Health Organization currently has also introduced a 4-month ATT regimen consisting of isoniazid, rifapentine [P], moxifloxacin [M], and pyrazinamide (2HPMZ/2HPM)(WHO consolidated guidelines on tuberculosis Module 4: Treatment -drugsusceptible tuberculosis treatment, 2020) that may be further applied and tested for TB-uveitis. The benefit of prolonging the total ATT duration beyond six months has been reported. (Agrawal et al., 2015;Ang et al., 2012b) It is recommended to commence a more extended treatment (9-12 months) for those with initial severe inflammation or poor early response to ATT. (Kon et al., 2022) Meanwhile, the role of ATT in treating TB-uveitis patients remains controversial. Resolution of inflammation is also achieved among patients treated without ATT, both in countries with high and low TB burdens (Table 2). On the other hand, a considerable risk of latent TB reactivation or progression with the use of systemic corticosteroid alone has been described. (Betzler et al., 2021;Jick et al., 2006) Overall, the proportions of patients who achieved resolution of uveitis treated with ATT in high and low TB-burden countries were 80-100% and 23.9-100%, respectively. Of note, treatment responses vary considerably among studies, taking into consideration that TB-uveitis lacks a gold standard diagnostic test and can present with diverse ocular inflammatory phenotypes. Therefore, high caution is needed to interpret the treatment response based on a heterogeneous group of patients (see Supplementary Table). Therefore, the development of reliable laboratory tests to objectively monitor treatment response, TB-reactivation, or progression would be of great benefit.
With regard to the uveitis phenotypes, COTS reported that the outcomes varied, and ATT was not superior to other treatments (without ATT), which include topical or systemic corticosteroid alone. Clinical resolution of patients treated with and without ATT was comparable in tubercular anterior uveitis (TAU), ) tubercular intermediate uveitis (TIU), ) tubercular panuveitis (TPU), ) TB-uveitis with choroidal involvement (including serpiginous-like choroiditis, tuberculoma, multifocal choroiditis, and ampiginous choroiditis),  and tubercular retinal vasculitis (TRV). (Gunasekeran et al., 2019) However, it is worth noting that there was a tendency to give combination treatments in cases of severe ocular inflammation or significantly reduced visual acuity at presentation. Thus, interpreting treatment success rates in the studies above should be done with caution. A randomised controlled trial recruiting uveitis with positive IGRA who are initially treated with ATT versus corticosteroid is in progress and it is worth waiting for the results (ClinicalTrials.gov registration number: NCT05005637).
The COTS group provides an expert opinion-trained online calculator (https://www.oculartb.net/cots-calc) to help ophthalmologists decide if ATT is necessary for a given patient. The TB-uveitis phenotypes taken into account by this calculator are TB choroiditis, TAU, TIU, TPU, and TRV. Besides ocular phenotype and country-specific endemicity to TB, individual patient data, including TST, IGRA, and chest X-ray results, are also required by the calculator. The calculator will then display the median score of the survey result, ranging from 1 (very low recommendation to initiate ATT) to 5 (very high recommendation to initiate ATT) and the consensus on this: IQR 0 (absolute consensus achieved) to IQR 3 (poor consensus).  Even though there is still no well-conducted RCT, the recent BTS statement supports the regular use of a concomitant systemic corticosteroid (1 mg/kg prednisolone or equivalent) with ATT. (Kon et al., 2022) Systemic corticosteroids are expected to further control inflammation, especially in tubercular posterior uveitis. (Betzler et al., 2021;Kon et al., 2022) Increasing the dose of systemic corticosteroid is warranted in cases of paradoxical worsening (clinical or radiological worsening of pre-existing TB lesions or the presence of new lesions in patients who showed initial improvement with treatment) during ATT course, (Basu et al., 2013;Ganesh et al., 2019) which occurs in around a quarter of patients. (Basu et al., 2013) If satisfactory inflammation control is not achieved following a high dose of systemic corticosteroid, another immunosuppressive agent should be considered.  Several reports describe successful control of inflammation and relapse with adjunct immunosuppressive agents (methotrexate, cyclosporine, azathioprine, IFNα2, or a combination of these). (La Distia Nora et al., 2014;Oray et al., 2017;Urzua et al., 2017) The use of immunosuppressive agents and the outcomes are summarised in Table 3.

Intraocular treatment
Adjunctive intraocular anti-inflammatory treatment in TB-uveitis has a minor role and is supported only by limited studies. Although it is difficult to draw conclusions regarding the success rate, case reports with a limited sample size showed promising results. There was a report on the beneficial effect of a dexamethasone implant for treating secondary cystoid macular edema (success in 10/10 eyes), vitritis (success   in 10/13 eyes), and paradoxical worsening (success in 2/2 eyes) during the ATT course. The timing of a dexamethasone implant varied between 0 and 4 weeks after ATT initiation.  Another report on nine eyes with a specific subset of multifocal serpiginous choroiditis also supported (inflammation control in 8/9 eyes) the use of a dexamethasone implant. (Jain et al., 2018) Besides, dexamethasone implants successfully controlled inflammation among those experiencing a recurrence of inflammation when oral corticosteroids were tapered below 10 mg/day. (Fonollosa et al., 2016) In addition, although only described in a few cases, intravitreal methotrexate (Julian et al., 2013;Sahin and Ziaei, 2014;Tsui et al., 2021) and bevacizumab (Invernizzi et al., 2015) have also been reported to control inflammation satisfactorily. VEGF is known to be elevated in the serum of active pulmonary TB patients and reflects changes in endothelial function. (Polena et al., 2016;Saghazadeh and Rezaei, 2022) VEGF is involved in the modulation of angiogenesis and granuloma formation, which serves as the niche for Mtb. (Saghazadeh and Rezaei, 2022) It is postulated that secreted VEGF is also responsible for monocyte/macrophage accumulation and inflammation. The newly formed blood vessels tend to be fragile and hyperpermeable, potentially involved in Mtb dissemination, recruitment of inflammatory cells incapable of Mtb elimination or even support its growth, and participation in cell death signalling. (Maison, 2022) In vitro studies showed that Mtb-infected RPE cells secreted higher amounts of VEGF than uninfected RPE cells. (La Distia Nora et al., 2018d;Singh et al., 2021) In further support of enhanced local ocular VEGF production is the observation that vitreous VEGF level was higher in TB-uveitis than in other uveitis entities. However, available data on this latter was inconsistent. (Singh et al., 2021), (Schrijver et al., 2022) Studies with a significant number of patients are scarce, however, a recently published case report shows the utility of anti-VEGF injection in the resolution of choroidal granuloma.(M. Agarwal et al., 2020) The authors described a TB-uveitis case who received weekly intravitreal bevacizumab injections in combination with moxifloxacin (three injections) along with ATT. With each injection, the vitreous level of VEGF was measured, and a significant decline in its level was observed. Visual acuity improved (from hand movement to 6/9) within one month after the first injection, with noticeable regression of the choroidal granuloma.(M. Agarwal et al., 2020) However, a conflicting case has been reported by Babu et al. where a patient with bilateral choroiditis received bilateral intravitreal ranibizumab injection alongside ATT and systemic corticosteroids. While the injection improved the left eye, the right eye worsened. Subsequent intravitreal injection of moxifloxacin combined with dexamethasone (without anti-VEGF) helped resolve the lesion. (Babu et al., 2022) Besides anti-VEGF, octreotide (a somatostatin analog) offers an alternative to treat uveitic macular edema. Somatostatin receptors are present in human RPE. (van Hagen et al., 2000) Additionally, macrophages express somatostatin, which is involved in granulomatous inflammation. Somatostatin receptors are upregulated during the differentiation of macrophages from monocytes and the maturation of thymocytes. (Dalm et al., 2003;Ferone et al., 2004) Even though the Not analysed separately *MTX = methotrexate, CsA = Cyclosporine, AZA = azathioprine, ADA = adalimumab, BCVA = best-corrected visual acuity, H = Isoniazid, R = rifampicin, Z = pyrazinamide, E = Ethambutol, CS = corticosteroid, SC = subcutaneous, IGRA = interferon-gamma release assay, ATT = anti-tubercular therapy. a Remission: Clinically inactive ocular inflammation (further detail was not described). b Progression: clinical worsening of the ocular inflammation, including assessment by multimodal ocular imaging modalities.
precise mechanism of reducing uveitic macular is still unclear, two studies depicted the efficacy of octreotide. A 100 μg subcutaneous injection (three times daily) or a long-acting depot formulation of octreotide acetate (monthly 20-mg intramuscular injection), resolved chronic uveitic macular edema in 7 out of 9 eyes that were unresponsive to prior corticosteroids (local or systemic), oral nonsteroidal anti-inflammatory agents, and acetazolamide. (Kafkala et al., 2006) Another case series further confirmed the potential efficacy of long-acting octreotide as this treatment had a macular edema-reducing effect in 70% of treated episodes. (Missotten et al., 2007) Prospective studies with a larger sample size are warranted to elucidate the potential benefit of additional octreotide treatment for TB-uveitis. Interestingly, octreotide is being developed for local intraocular application by using nanoparticle technology, which brings hope for widespread use in treating macular edema in the upcoming years. (Amato et al., 2020)

Paradoxical worsening
Clinical worsening or progression of the active lesion(s) while on ATT (with or without concomitant corticosteroids) frequently occurs between 3 and 8 weeks after ATT initiation, especially among those showing serpiginous-like choroiditis (SLC)(A. Agarwal et al., 2020b;Esen et al., 2016;Kalra et al., 2022;Mishra et al., 2020) and multifocal choroiditis (MFC).  The mechanism underlying this worsening in TB-uveitis has not been clearly elucidated. The release of mycobacterial antigens following ATT administration and the occurrence of a delayed-type hypersensitivity reaction have been proposed as the underlying mechanisms. This is partly supported by the higher baseline level of IL-10 as well as increasing levels of IFN-γ, transforming growth factor-β (TGF-β), and TNF-α during treatment among those with paradoxical reactions.  Several clinical factors can help to predict paradoxical worsening, including baseline best corrected visual acuity (BCVA) > 0.7 logMAR, advanced grade of initial lesion opacity, and disc involvement (optic neuritis, disc elevation/edema, or active choroiditis extending to within 1 disc diameter of the optic nerve head).(A. Agarwal et al., 2020b) An algorithm using automated colour fundus image processing showed promising utility in developing an ocular image-based prediction model for paradoxical worsening (Kalra et al., 2022), but this requires further validation in other studies.
Because of the current well-described paradoxical reaction, the attending ophthalmologist should stick with the ATT treatment and concomitant use of corticosteroids. However, an alternative differential diagnosis, such as other autoimmune-related uveitis or another infection mimicking TB-uveitis, as a cause for the paradoxical worsening, should be kept in mind. The diagnosis can be revised if the clinical phenotype changes or remains unresponsive to treatment. Further diagnostics, including drug resistance tests (i.e., GeneXpert), are recommended if the continuation of ATT with escalated corticosteroid therapy fails to show improvement beyond the initial 2-3 months of treatment,  although this is practically difficult to test in an ocular fluid sample.

Current consensus on treatment outcome definition
The COTS nomenclature working group defined remission and cure as the outcomes of TB-uveitis treatment, as stated below. Of note, patients unable to taper systemic corticosteroids to <10 mg/day or topical steroid drops to <2 drops/day and the continuation of steroid-sparing immunosuppressants do not fit into these criteria.(A. Agarwal et al., 2020a;Agrawal et al., 2019) • Remission: inactive disease (grade 0 cells/no inflammation) for at least 3 months following a complete ATT course.
• Cure: inactive disease (grade 0 cells/no inflammation) 24 months following a complete ATT course.

Host-directed therapies for tuberculosis: possible implications on tubercular uveitis treatment
Modulation of host immune cell function against Mtb can be achieved via several approaches (Kilinç et al., 2021) (Fig. 5).
1) Enhancing phagolysosome activity and autophagy to improve intracellular killing of Mtb, 2) Reducing granulomatous lesions to limit Mtb spread, 3) Reducing inflammation-associated tissue damage, 4) Enhancing cellular responses against the infection and associated inflammation, and 5) Skewing T-lymphocytes responses towards Th1 to stimulate cellular immunity.
Knowledge of the underlying immune responses at each stage of TB pathogenesis underpins studies in the development of HDT, mostly from in vitro or animal studies on pulmonary TB. This section will focus on selected HDTs that have been applied to human subjects. Of note, none of these mentioned modalities has been applied to TB-uveitis so far. However, as we gain more immune-pathogenic knowledge in TB-uveitis, application of specific HDT for TB-uveitis is to be expected in the future.

Metformin
Metformin has been repurposed for TB treatment as it may enhance autophagy. It may limit Mtb growth by stimulating the formation of reactive oxygen species and phagolysosome fusion. (Guler et al., 2021) In a Mtb guinea pig model for granuloma formation, metformin dampened inflammation and significantly reduced lung pathology. (Frenkel et al., 2020) Further study observed the downregulation of genes involved in oxidative phosphorylation, mammalian target of rapamycin (mTOR) signalling, and type I IFN response pathways. (Lachmandas et al., 2019) Meanwhile, genes involved in phagocytosis and the production of reactive oxygen species were upregulated. (Lachmandas et al., 2019) Adjunct metformin, combined with ATT, increased the probability of sputum conversion at two months of treatment. (Yu et al., 2019) It also lowered the mortality rate among people with diabetes, based on a recent systematic review. (Yu et al., 2019) However, another study reported that additional metformin treatment significantly accelerated lung healing but did not accelerate sputum culture conversion. (Padmapriydarsini et al., 2022) Autophagy is a conserved process that delivers proteins, organelles, and invading pathogens to lysosomes for degradation. (Liang et al., 2017) However, Mtb can escape and block the maturation of autophagolysosomes, preventing it from an effective pathogen-killing mechanism. (Liang et al., 2017) Although initial findings showed the benefit of adjunct metformin in TB patients with diabetes mellitus, (Yu et al., 2019) the potential use of metformin for TB-uveitis through its autophagy enhancement needs further investigation. As Mtb can infect RPE, (La Distia Nora et al., 2018d) we assume that theoretically, the discovery of an effective drug that enhances the killing of Mtb residing in RPE can accelerate clinical improvement and prevent chronic ocular inflammation.

Phosphodiesterase and mTOR inhibition
Phosphodiesterase inhibitors suppress TNF-α while increasing intracellular signalling pathways involving cyclic adenosine monophosphate (cAMP) and cyclic guanosine monophosphate (cGMP). (Guler et al., 2021) Autophagy is regulated by the serine/threonine kinase mTOR. Inhibition of mTOR stimulates autophagy in macrophages and has been suggested to enhance intracellular killing of Mtb. (Singh and Subbian, 2018) mTORC1 inhibition by nitazoxanide could enhance autophagy and inhibit Mtb proliferation in vitro. (Lam et al., 2012) A recent trial showed that adjunctive everolimus (mTOR inhibitor) and CC-11050 (type 4 phosphodiesterase inhibitor) had beneficial effects. These compounds were safe and may facilitate lung function recovery but not early sputum culture conversion in pulmonary TB. (Wallis et al., 2021) Similar to metformin, more research is needed to evaluate its potential use in TB-uveitis.

Matrix metalloproteinases (MMPs) inhibition
TB granuloma is currently thought to be a protective immune response to TB. However, the granuloma can also serve as a potential niche for further Mtb replication and dissemination. MMPs are a family of endopeptidases that have properties to degrade extracellular matrix (ECM) components, and their activity is controlled by tissue inhibitors of metalloproteinases (TIMPs). (Sabir et al., 2019)Although the primary function of MMPs is related to ECM degradation and remodelling, (Sabir et al., 2019) these endopeptidases are also involved in angiogenesis, inflammation, and immune responses. (Ohno-Matsui et al., 2003;Xu et al., 2018) During inflammation, the expression and activity of most MMPs increases in immune (including monocytes/macrophages) and non-immune cells, which is controlled by different cytokines and transcription factors, including IFN-γ, IL-12, IL-17, TNF-α, NF-κB, and STAT3. (Sabir et al., 2019) Increased MMP expression, including MMP-1, -3, -9, and -12 and MMP-10, has been demonstrated in TB lung pathology. (Parasa et al., 2017) Furthermore, elevated levels of MMP-1, -2, -7, -8, and -9 have also been detected in circulation. (Kumar et al., 2020) MMP-9, induced by ESAT-6, is involved in monocyte/macrophage recruitment during the initial phase of granuloma formation at the site of infection. (Sabir et al., 2019) Meanwhile, MMP-3, MMP-8, and MMP-13 are more involved in tissue remodelling at the site of granuloma. (Sabir et al., 2019) To date, doxycycline is the only FDA-approved MMP inhibitor and has been explored as an HDT for TB. Adjunct doxycycline treatment for two weeks at 100 mg twice a day during ATT resulted in reduced pulmonary cavity volumes that, however, did not reach statistical significance compared to the placebo-treated patients. (Miow et al., 2021) Reduced plasma MMP-1, -8, -9, -12, and − 13 levels, suppressed degradation of type I collagen and elastin, and reduced blood transcriptional levels of several immune response genes (IRF1, APOL1, FCGR1A, FCGR1B, GBP5, and GBP6) were observed upon doxycycline treatment. (Miow et al., 2021) Although the precise mechanism of granuloma formation and remodelling in TB-uveitis is not fully understood, the additional use of doxycycline (or another MMPs inhibitor) to limit retinal damage in TB-uveitis with posterior segment inflammation warrants further investigation.

N-acetylcysteine
The biological effect of N-acetylcysteine is attributed to its antioxidant properties. It may also lower inflammatory cytokine production (e. g TNF-α) by replenishing intracellular glutathione, which is decreased in the peripheral blood mononuclear cells of TB patients. (Venketaraman et al., 2008) Furthermore, it may limit Mtb growth by its immunomodulatory function, resulting in adequate production of Th1 cytokines (e.g. IL-2, IFN-γ), which are crucial to controlling Mtb effectively. (Guerra et al., 2011;Guler et al., 2021) The result of a study on adjunctive N-acetylcysteine seems promising, as it significantly increased glutathione peroxidase levels, accelerated sputum conversion, and enhanced radiological improvement among pulmonary TB patients. (Mahakalkar et al., 2017) Further investigation is warranted, especially in TB-uveitis presenting with clinical pictures strongly suggestive of Mtb presence in the eye, such as choroidal tuberculoma and serpiginous-like choroiditis.

Cyclooxygenase (COX) inhibitor
Non-steroidal anti-inflammatory drugs (NSAIDs), including ibuprofen, indomethacin, diclofenac, and aspirin, have been explored as candidates for HDT against Mtb. It was observed that the inhibition of COX reduces the prostaglandin E2 (PGE2) level, which may inhibit phagocytosis and Mtb killing at the late stage of infection. (Kroesen et al., 2017) An in-vitro study revealed that indomethacin reduced Treg activity and enhanced Th1 immunity, as shown by the increased production of IL-2, TNF-α and IFN-γ by CD4 + T-lymphocytes. (Tonby et al., 2016) Thereby, it can be expected that COX inhibitor may increase Mtb clearance. Moreover, a hypercoagulable state that may be present in the advanced stage of TB can also be managed with low-dose aspirin. (Kroesen et al., 2017) Although it does not reduce the mortality rate, aspirin significantly reduces the onset of stroke among patients with TB meningitis. (Rohilla et al., 2021) In addition, adjunctive aspirin for treating diabetic patients with pulmonary TB increased the sputum conversion rate (86.7% vs 53.8%; p = 0.031). (Wang et al., 2020) The observed beneficial effect of NSAIDs raises the possibility of repurposing these drugs to be used to combat TB. (Kroesen et al., 2017;Maitra et al., 2016) COX inhibitors might be considered an adjunct treatment in TB-uveitis, particularly for those with a clinical picture of retinal vasculitis or patients with an increased risk of cardiovascular complications.

Cellular therapies
Mesenchymal stromal cells (MSCs) exhibit multipotent differentiation capacity and display tissue repair and immunosuppressive capacities by limiting the proliferation of T lymphocytes (CD4 + and CD8 + ), B lymphocytes, and NK cells. (Joshi et al., 2015) MSCs also modulate dendritic cells (DC) and induce Tregs. (Joshi et al., 2015) The immunosuppressant properties of MSCs rely on cell-to-cell interactions as well as the secretion of soluble factors, such as TGF-β1, hepatocyte growth factor (HGF), and PGE2. (Joshi et al., 2015) MSCs can phagocytose Mtb and restrict Mtb growth through autophagy. (Khan et al., 2017) A phase I trial with a single infusion of autologous bone marrow-derived MSCs as an adjunct treatment was conducted in multidrug-resistant (MDR) and extensively drug-resistant (XDR) TB and was found to be safe. (Skrahin et al., 2014) In the follow-up study, the cure rate for patients who received MSCs was three times higher than that for those treated with anti-TB treatment only. (Skrahin et al., 2016) Circulating Vγ9Vδ2 T lymphocytes represent a major innate-like peripheral T-lymphocyte subset activated by non-protein phosphoantigens derived from microbial pathogens, including mycobacteria. (Poupot and Fournié, 2004) Expansion of the Vγ9Vδ2 T lymphocyte pool occurs following infection with a variety of microbial pathogens, including Mtb. (Gay et al., 2022;Poupot and Fournié, 2004) These cells can directly kill infected cells, mainly through the release of lytic molecules, cytokines, and apoptosis but also modulate functions of other innate and adaptive immune cells. (Gay et al., 2022) However, a loss of cytotoxic activity of Vγ9Vδ2 T lymphocytes has also been described in active TB. (Gay et al., 2022) Vγ9Vδ2 T lymphocyte-based immunotherapy could be of interest in treating TB. A preliminary report on the utility of allogeneic Vγ9Vδ2 T lymphocytes treatment in tuberculosis has been recently published (ClinicalTrials.gov registration number: NCT03575299). (Liang et al., 2021) In that study, Vγ9Vδ2 T lymphocytes were obtained from healthy donors, expanded ex vivo, and subsequently infused into patients. Based on the published report, this procedure was safe, and lung lesions were reduced in all patients. (Liang et al., 2021) Moreover, CD56 high-expressing NK cells were reduced, while CD56 low-expressing NK cells were increased. (Liang et al., 2021) CD56 low CD16 + high expressing NK cells have a lesser capability in cytokine production than CD56 high CD16 low-expressing NK cells, the former exhibits significantly higher cytotoxic activity. (Poli et al., 2009) This suggests that the anti-TB effect of Vγ9Vδ2 T lymphocytes treatment is, potentially, mediated through a shift in the distribution of NK cell sub-populations and related functions. (Liang et al., 2021) Besides the effect on NK cells, Vγ9Vδ2 T lymphocyte treatment altered the distribution of other immune cell populations in peripheral blood, including increased effector memory T-lymphocytes (both CD4 + effector memory and CD8 + effector memory). (Liang et al., 2021) To be adopted in TB-uveitis treatment, the ability of cellular therapies to modulate other potential cells contributing to the granulomatous inflammation needs to be investigated. As the inner and outer BRB are important upon Mtb infection, cellular therapies that also strengthen the immunological response of microglia, Müller cells, and astrocytes against Mtb would be of interest. The integrity of the inner BRB depends upon pericyte-retinal endothelial cell interactions that involve autocrine and paracrine signaling pathways, including the VEGF and TGF-β signaling pathways. (Huang, 2020) Besides, retinal microglia, a resident immune cell involved in immune surveillance and the regulation of neurogenesis, would likely be important in the immune response in uveitis, (Okunuki et al., 2019) but so far, studies on this are limited. Moreover, recently it was demonstrated that IL-1β stimulates the proliferation and reactivity of microglia, which was associated with a reduction in retinal damage. (Todd et al., 2019) Importantly, microglia are crucial to facilitate passage of primed inflammatory cells across the BRB, and the absence of microglia prevents uveitis, as shown in the experimental autoimmune uveitis. (Okunuki et al., 2019) Furthermore, the retinal microenvironment contains cell and molecules involved in the establishment of ocular immune privilege, for instance, through CD95 ligand (CD95L/Fas ligand) that induces apoptosis of infiltrated inflammatory cells. (Keino et al., 2018) RPE cells also have the ability to convert activated T lymphocytes into Tregs via TGF-β signaling and express PD-L1 that suppresses IRBP-specific T lymphocytes exposed to IL-17 and IFN-γ. (Keino et al., 2018) T lymphocyte profiling from the peripheral blood of TB-uveitis patients revealed a low frequency of Tregs and a high CD4+/CD8+ T lymphocyte ratio. (Sharma et al., 2018a(Sharma et al., , 2018b However, in vitreous from TB-uveitis patients, comparable CD4 + and CD8 + lymphocyte numbers were observed. (Sharma et al., 2018b) Interestingly, even though T lymphocytes in the vitreous of TB-uveitis were significantly more abundant than in the non-uveitis controls, the percentage of viable T-lymphocytes was significantly lower. (Sharma et al., 2018b) These findings highlight the difference in the composition of the infiltrating T lymphocyte population in the eye compared to the systemic compartment, with a higher proportion of CD8 + T lymphocytes and the immune-privilege associated regulatory mechanisms to which the ocular immune response is exposed in TB-uveitis. Thus, the response of these cells toward cellular therapies warrants further exploration and should be taken into account when deciding on cellular therapies in TB-uveitis that are multipotent for their immunosuppression and retina-specific repair capacities.

Enhancement of cellular and humoral immune responses
IL-2 is a crucial cytokine within the immune system as it is involved in the activation of T-lymphocytes. (Ross and Cantrell, 2018) Cell-mediated immunity involving CD4 + and CD8 + T lymphocytes is crucial in TB pathogenesis. Treatment with recombinant human IL-2 has been tried to enhance Th1 activity against Mtb. Accelerated sputum smear conversion, sputum culture conversion, and pulmonary radiological resolution rates were achieved with IL-2 treatment. (Shen et al., 2015) However, further studies with a large number of patients yielded contradictory results. (Johnson et al., 2003;Nie et al., 2022) IFN-γ could enhance Mtb killing within macrophages via induction of autophagy by counteracting IL-6. (Dutta et al., 2012) Importantly, it was shown that Mtb could dampen IFN-γ signalling and inhibit HLA class II-antigen processing, allowing Mtb to escape cellular immune surveillance. (Pai et al., 2003) Thus, IFN-γ treatment is expected to improve Mtb-antigen presentation. The direct administration of IFN-γ to enhance Th1-mediated immunity has also been explored. Subcutaneous IFN-γ was used in the treatment of MDR pulmonary TB but did not improve the clinical outcome. (Park et al., 2007) Also, aerosolised IFN-γ treatment was not beneficial in the treatment of MDR pulmonary TB, although no control group was included in this study. (Koh et al., 2004) Furthermore, adjunct treatment with IL-12 (recombinant IL-12, 300 ng/kg body weight subcutaneously twice weekly for three months), an enhancer of IFN-γ, alongside ATT improved the treatment outcome of a patient with multiple site TB abscesses. (Greinert et al., 2001) Unfortunately, the effect of the treatment on Th1 skewing was not explored in those aforementioned studies.
H56:IC31subunit vaccine, containing the Mtb antigens (Ag85B, ESAT-6, and Rv2660c (H56)) was deployed to enhance humoral and cellular immune responses (antigen-specific T lymphocytes). (Aagaard et al., 2011) Jenum et al. conducted a four-arm phase I/II randomised controlled trial using the H56:IC31 vaccine and etoricoxib (a cyclooxygenase-2-inhibitor) in addition to standard ATT (ClinicalTrials. gov registration number: NCT02503839). (Jenum et al., 2021) In the eterocoxib group, there was no significant difference in the level of IFN-γ and IL-2 responses from isolated T-lymphocytes from TB patients in response to H56 fusion protein, Ag85B, or ESAT-6 stimulation. However, patients receiving two doses of H56:IC31 showed increased IFN-γ and IL-2 responses. It was concluded that H56:IC31 vaccination in TB patients resulted in an improved immune response compared to ATT alone, without major safety concerns. (Jenum et al., 2021) Since TB-uveitis patients with a higher QFT level have a greater chance for an ATT response, (Danjou et al., 2021) the induction of the host's IFN-γ and IL-2 responses seems promising to expedite clinical resolution in TB-uveitis. Also, it is important to study the effect of the aforementioned induction of the host IFN-γ and IL-2 responses on retinal glial cells. IFN-γ, for instance, stimulates TNF-α and IL-6 as well as ICAM-1 expression by Müller cells. (Drescher and Whittum-Hudson, 1996) Stimulation of ICAM-1 expression influences the migration of leucocytes in the inflamed tissue. (Drescher and Whittum-Hudson, 1996). Moreover, retinal glial cells not only exert innate and adaptive immune functions but also possess capacity for autophagy and apoptosis, which may also play a role in Mtb control and surveillance within the retinal-specific microenvironments. (Lorenz et al., 2021;Qin et al., 2022) Therefore, exploring the contribution of the different retinal cell types to Mtb surveillance and TB-uveitis and the role of cytokines such as IFN-γ herein requires further study.

The burden of drug-resistant TB
The use of second-line ATT, including the use of alternative ATT combinations to treat potential drug-resistant Mtb, has not been widely incorporated in clinics. The extent of drug-resistant Mtb strains detected in the ocular fluid/tissue has not been widely explored, as we still face challenges in finding Mtb footprints. Mtb resistance can be expected in patients from high-burden countries. Rifampicin-resistant Mtb can be expected in 14% of the cases. (Zignol et al., 2018) Bansal et al. reported that rifampicin-and isoniazid-resistant Mtb was present in their TB-uveitis cohort, with a positive response towards a combination of MDR TB treatment, systemic corticosteroids, and azathioprine.  The extent of MDR and XDR problems should be considered among those residing in countries with a high MDR TB burden. It is hoped that TB-uveitis treatment with HDT in the future will also be able to overcome this MDR and XDR problem, which may be currently overlooked.

Future directions
The current understanding of intraocular TB infection is especially based on the observation of the RPE response in the presence of Mtb. Mtb may reach the ocular structure via intra-or extracellular routes, which is argued to explain the diverse phenotypes of TB-uveitis as both routes differ in BRB breaching. (Damera et al., 2021) TLRs on RPE recognise Mtb structures. (Nazari et al., 2014) Once phagocytosed, RPE would orchestrate further immune regulation against Mtb. (La Distia Nora et al., 2018d;Nazari et al., 2014) Meanwhile, the current development of HDT for TB is based on lung pathology, mainly involving macrophages, lung alveolar cells, and cytokines involved with the lung lesions. (Guler et al., 2021) Granulomas are important structures in the pathology of TB and are also formed in the ocular structures. Granuloma formation in TB can occur during the primary infection or reactivation phase in a previously latently infected person. As a dynamic structure, classic TB granuloma, as found in the lungs, might be constructed by complex interplay between various types of immune and non-immune cells. (Guirado and Schlesinger, 2013) Formation of TB granuloma involves interaction between macrophages, epithelioid cells, multinucleated giant cells, B and T lymphocytes, neutrophils, and fibroblasts, that results in a central necrotic area within the granuloma where Mtb is mainly sequestered. (Cronan, 2022;Guirado and Schlesinger, 2013) Secreted CCL2/MCP-1, CCL12, and CCL13 that bind to the CCR2 receptor lead to early macrophage recruitment in the infected tissue. (Guirado and Schlesinger, 2013) Infected macrophages secrete TNF-α which is essential for appropriate granuloma formation, as altered TNF-α signaling resulted in fatal TB progression with a lack of epithelioid cells, less-organized granuloma formation, and profound mycobacterial replication. (Cronan, 2022;Domingo-Gonzalez et al., 2016;Kindler et al., 1989) This granulomas not only combat Mtb but also become a source of inflammation, which warrants further exploration to limit self-tissue damage. (Guler et al., 2021;Kolloli and Subbian, 2017) NK cells also participate in cytokine signaling in response to Mtb antigens, induce cell-mediated cytotoxicity, and bridge innate and adaptive immune responses against Mtb. (Garand et al., 2018) The adaptive immune response induced by Mtb infection is mainly evoked by dendritic cells trafficking to the lymph nodes and subsequent priming of naïve T lymphocytes. This will result in a dominant Th1 response that is characterized by marked IFN-γ secretion by CD4 + T lymphocytes. (Domingo-Gonzalez et al., 2016) In addition, activated CD8 + T lymphocytes can also produce IFN-γ and exert cytotoxic activity. (Guirado and Schlesinger, 2013) Moreover, Th17 cells will secrete IL-17, which then primarily stimulates the production of the G-CSF and IL-8 that subsequently attract neutrophils. (Mourik et al., 2017;Zenaro et al., 2009) Neutrophils also exert Mtb killing capacity through nicotinamide adenine dinucleotide phosphatase (NADPH) mechanism. (Yang et al., 2012) However, accumulation of necrotic neutrophils is associated with chronic inflammation and tissue damage. (Hilda et al., 2020) B cells, aside from acting as antigen presenting cells, are also critical in enhancement of Th1 immunity and granuloma formation as they promote the proliferation and capacity of IFN-γ-producing CD4 + T lymphocytes. (Chen et al., 2023) The general view of an adaptive immune response to TB is thus depicted by a dominant Th1 response with robust IFN-γ secretion, although the pathogenesis clearly involves other cellular components, proinflammatory cytokines (such as TNF-α, IL-1β, IL-6, IL-8, and IFN-α/β), as well as anti-inflammatory cytokines (such as TGF-β and IL-10). (Cronan, 2022) The local production of various cytokines thus induces immune cells recruitment and Mtb killing. Eventually, a well-regulated balance between pro-and anti-inflammatory mechanisms within TB granulomas will determine whether Mtb can be controlled. Within the microenvironment of a granuloma, Mtb can persist in the granuloma for a long time, so-called latent TB, which is asymptomatic. Several eliciting stimuli that alter host immunological status, including malnutrition, diabetes, and even aging, can reactivate the contained Mtb. (Guirado and Schlesinger, 2013) Optimization of al HDT development in the setting of TB-uveitis requires further in-depth knowledge on the immunopathogenesis of granuloma formation within the ocular structures. Alternative regulation of ocular granuloma formation can be expected as RPE cells display a lower phagocytic capacity and a different transcriptomic response to Mtb infection when compared to Mtb -infected macrophages. (La Distia Nora et al., 2018d) This raises the possibility of a specific pathological process involved in granuloma formation within the eye that does not occur in pulmonary/systemic TB.
Even though significant research efforts to improve the understanding of TB-uveitis pathophysiology have been made, still large gaps in our current knowledge interfere with effective diagnosis and adequate treatment management. With multi-omics approaches, pathways of host-pathogen responses and immunoregulatory processes involved in TB-uveitis might be better understood. Identification of key pathways involved in the pathology of TB-uveitis might uncover novel diagnostic approaches as well as target molecules and/or pathways for disease treatment. Moreover, we consider accurate detection of Mtb in ocular tissue and identification of biomarkers that are capable of distinguishing the various subtypes of patients, particularly those who might pose induced immune responses, to be critical for optimal patientdirected treatment. (Ludi et al., 2023)

A lesson learned from immune checkpoint inhibitor treatment for host directed therapy in TB
Immune checkpoint inhibitors (ICI), such as anti-programmed cell death 1 (PD-1)/programmed cell death ligand 1 (PD-L1) and cytotoxic T-lymphocyte-associated antigen 4 (CTLA-4) signalling, have been used for cancer treatment. (Elkington et al., 2018;Langan et al., 2020) There are concerns about new onset TB disease following such treatment, even though the estimated incidence rate might be low (3/1,144 cancer patients treated with ICI). (Zaemes and Kim, 2020) Despite being used for reversing tolerance to tumor tissue, the development of TB following ICI that enhance effector T lymphocytes seems counterintuitive. (Elkington et al., 2018) However, it is hypothesised that the immune homeostasis within TB granulomas is altered and facilitates Mtb spread. As a result of PD-L1 inhibition, augmented cytokine responses could lead to excessive inflammatory cell activation and the destruction of the granuloma-associated extracellular matrix, thereby enhancing Mtb growth and spread. (Tezera et al., 2020) Altogether, this data suggests that finding an appropriate HDT is challenging and should take differences between systemic and local (within the granuloma) immune regulation in TB into account.

Harnessing HDT for tubercular uveitis: Are we there yet?
The applicability of the current HDTs to TB-uveitis patients may be of limited value unless: (1) substantial evidence supports their utility in both pulmonary and extrapulmonary TB, (2) a specific drug target involved in TB-uveitis that dampens various inflammatory pathways relevant to the ocular condition is discovered, and (3) treatmentdirected patient classification that focuses on underlying pathological processes despite various clinical spectrums is validated. Many other HDT candidates than those discussed above are being investigated. The drug candidates might serve as an adjunct treatment of ATT for TBuveitis.

Conclusion
TB-uveitis is one of the uveitis entities that leads to diagnostic and treatment problems among clinicians. The diversity of its clinical picture without a reliable single gold-standard diagnostic test makes it challenging to diagnose and treat TB-uveitis accurately. On one side, Mtb might be detected in a paucibacillary manner in the ocular tissue, but an autoimmune process without demonstrable presence of ocular Mtb might also play a pivotal role. Some conditions where the diagnosis of TB is not straightforward pose difficulties in deciding who needs and who does not need ATT, particularly if one is facing patients with a positive TST or IGRA but without an ocular phenotype strongly suggestive of TB-uveitis or other signs of TB-disease. Meanwhile, successful treatment with ATT was depicted in many studies, even though treatment results without ATT were not immensely inferior; the latter finding should not be overlooked. The long duration and potential adverse reactions to ATT underline the need for further studies to identify those patients in actual need of ATT. Moreover, the current development of HDTs that specifically modulate host responses to Mtb without ignoring the importance of antimycobacterial therapy might in the future also prove to be of value for the treatment of TB-uveitis. It is imperative to emphasise the importance of more studies on developing a new approach in TB-uveitis diagnostics and the further development of precise treatment strategies based on the patient-specific underlying inflammatory process. This requires us to greatly increase our understanding of the pathogenesis of TB-uveitis in the coming years.

Declaration of competing interest
None.

Data availability
No data was used for the research described in the article.