BTS clinical statement for the diagnosis and management of ocular tuberculosis

Clinical practice point

• All patients suspected of OTB should be managed jointly by ophthalmic specialists and TB centres.

Summary of clinical practice points

Special groups

Patient considerations

• Non-randomised data have shown that most patients with likely TB uveitis will benefit from ATT with a reduction in uveitis activity and relapse compared with patients who are not treated.

• Consider withholding TB treatment when the evidence for TB exposure is separate to the cause of the uveitis, for example, old TB CXR or CT changes of TB or a positive IGRA with prior history of TB and a clear alternative cause for uveitis. This should also take into account the risks of TB treatment in the individual.

• Patients should be included in the decision to treat. Where OTB is the only manifestation of TB this usually also requires an MDT discussion involving the uveitis and TB specialists, including the risks of alternative treatments and potential visual loss.

Inflammation involving the uvea and retina (uveitis)

Uveitis (intraocular inflammation) is the the most common manifestation of TB-associated eye disease.

Anterior uveitis is typically granulomatous, with ‘mutton fat’ deposits on the inner cornea and nodules may be seen on the iris due to granulomatous inflammation. Chronic non-granulomatous anterior uveitis may also occur and TB should be excluded in all cases of atypical anterior uveitis not responding to conventional therapy, in patients with socioepidemiological risk factors for TB.

Intermediate uveitis where the predominant site of inflammation is in the vitreous gel (vitritis), peripheral retina (vascular sheathing) and pars plana (pars planitis), is a non-specific finding in many patients with ocular inflammation. Clumps of cells in the vitreous and ‘snowball’ opacities suggest chronic granulomatous inflammation. Recurrent macular oedema is a significant cause of visual morbidity in some patients with TB-associated intermediate uveitis.

Posterior uveitis is an umbrella term for inflammation of the back of the eye.

Retinal vasculitis is a description of inflammatory changes associated with retinal blood vessels, predominantly veins and with perivascular inflammatory infiltrates. There may also be significant vascular occlusion in OTB. Eales’ disease refers to a retinal vasculitis characterised by occlusive disease and neovascularisation, leading to recurrent vitreous haemorrhage and tractional retinal detachment, although this term is usually reserved for patients in whom TB is not thought to be a contributing factor. This boundary is difficult to draw and many patients with peripheral occlusive disease secondary to TB may have this label attached to them.

• Serpiginous choroidopathy, ampiginous and serpiginous-like choroidopathy are all diseases of the retinal pigment epithelium. Classic serpiginous disease originates around the optic disc and spreads in a contiguous, centrifugal manner. Ampiginous and serpiginous-like choroidopathy refers to placoid lesions present in the posterior pole and periphery, initially non-contiguous and evolving in a serpiginoid pattern with varying degrees of vitritis.

• Choroidal granulomas, which may be small and multiple (‘tubercles’), or present as a single, elevated choroidal mass (‘tuberculoma’). Multiple tubercles or multifocal choroiditis) signify disseminated disease and are more commonly observed in immunocompromised patients.

Other posterior segment findings of TB include:

• Tuberculous optic neuropathy, including tubercles on the optic nerve.

• Panuveitis—severe inflammation involving all layers and compartments of the eye, more common in immunocompromised patients.

Inflammation involving the ocular surface

• Episcleritis or scleritis is usually anterior (visible when the eye is wide open). Any persistent redness or irregularity of visible sclera should be referred for an ophthalmic opinion, particularly if associated with discomfort.

• Phlyctenulosis describes an inflamed nodule on the conjunctiva often abutting the limbal cornea, associated with a Type IV hypersensitivity reaction to Mtb. Lesions may advance over the cornea causing scarring.

• Corneal inflammation may occur deeper, in the stroma (interstitial keratitis) or adjacent sclera (sclerokeratitis).

Inflammation of the orbit

Orbital and periorbital disease is usually acquired by haematogenous or lymphatic spread and more rarely by direct spread from skin inoculation or the paranasal sinuses. Presentations include:

• Eyelid or periorbital cutaneous TB (lupus vulgaris).

• Lacrimal gland and sac involvement.

• Bone involvement.

• Inflammatory masses within the orbit. Patients may present with pain, proptosis, diplopia or reduced vision secondary to compressive optic neuropathy. Orbital biopsy may be required to exclude malignancy.

An image library providing examples of a number of the above conditions can be found here: https://www.aao.org/topic-detail/ocular-tuberculosis-tb--asia-pacific-2

Immunosuppression and OTB

The literature is sparse on particular presentations of OTB associated with immunosuppression, varying from mild iatrogenic immunosuppression to advanced AIDS. All patients with suspected TB should undergo HIV testing. The table below details particular clinical considerations and their treatment implications for immunosuppressed patients, such as those with advanced HIV infection (see table 2).

Clinical practice points

• Consider OTB as a cause of ocular disease particularly when individuals have risk factors for TB.

• In immunosuppressed individuals, TB may be more disseminated and include the eye.

• OTB may be present despite a lack of clinical or radiological evidence of pulmonary TB, particularly in immunosuppressed individuals.

• Radiological appearances and immunological tests for TB may be atypical or falsely negative in immunosuppressed individuals.

Immunological tests and radiology

Uveitis patients with suspected TB should have an IGRA and a CXR. It is important that other causes of a similar uveitis phenotype are excluded, and tests are also performed to prepare for TB treatment and/or immunosuppression. Tests to investigate the cause of the uveitis should be driven by the presenting ocular phenotype, systematic enquiry of all systems and clinical findings (see figure 1).

Clinical practice points

• Tests on all patients with uveitis and positive IGRA: U&E, eGFR, LFT, FBC, CXR. Syphilis serology, HIV, Hepatitis B and C screen, sputum samples for AFB and culture if coughing.

• Tests depending on ocular phenotype, systematic enquiry and clinical findings elsewhere: HLA-B27 & A29, serum ACE, CRP, ESR, ANCA, ANA. Toxoplasmosis serology, Lyme (Borrelia) serology.

Ocular imaging

The fundal appearance should be documented with ocular imaging especially if electronic records preclude fundus drawings. A widefield colour image is ideal, and an autofluorescence image can also indicate areas of active chorioretinitis.

WFFA is required to assess retinal vasculitis. Vascular leakage generally indicates active disease. WFFA may also help to define areas of active chorioretinitis, neovascularisation and areas of non-perfusion.19

OCT should be used to evaluate and monitor macular oedema. Enhanced depth Imaging-OCT can be used to measure choroidal thickness and is useful for characterising and monitoring choroidal granulomas.20 It can also detect characteristic changes in retinal vessels where it includes sections of vessels affected by vasculitis. Such imaging modalities are routinely employed in the uveitis clinic .

Clinical practice points

• Colour fundus images are recommended for documentation and monitoring, preferably wide-field with autofluoresence images where available.

• Macular OCT is standard practice for evaluating macular lesions and macular oedema.

• WFFA is recommended where retinal vasculitis is suspected, or to further characterise chorioretinitis.

Visual field testing

Patients with optic nerve involvement or peripheral non-perfusion may benefit from automated visual field testing, both for assessment purposes and to monitor response to treatment. The most useful visual field test on a Humphrey Visual Field Analyser (Zeiss) for uveitis is 30–2, measuring 30⁰ of visual field from the centre.21 A 10–2 test (10° from centre) may be more appropriate for paracentral scotomas. An Estermann visual field (performed with both eyes open) can be performed to evaluate whether or not a patient meets the DVLA driving standard. If central or peripheral vision is permanently affected in both eyes, the ophthalmologist can advise if the patient should inform the DVLA.

Aqueous or vitreous sampling for PCR

DNA amplification tests by PCR can identify TB DNA in ocular fluids. If present this may demonstrate active intraocular TB rather than a hypersensitivity reaction. Aqueous can be sampled in an outpatient setting whereas vitreous sampling is usually performed in an operating theatre. Any penetration of a needle into the eye comes with a risk of introducing infection (endophthalmitis). The risk is low (approximately 1 in 3000 for therapeutic intravitreal injection), but the consequences are dire (painful vision loss). Aqueous sampling has additional risks of iris damage or traumatic cataract. Therefore, the result of PCR testing must have the potential to change the patient’s management.

The yield of PCR testing is affected by the method of DNA extraction, amplification targets, the presence of inhibitors and the volume of fluid.22 23 It is crucial to check local requirements for the ideal volume required as tests vary significantly. In general, the specificity of PCR tests are high, but the sensitivity in TB uveitis is variable24: standard PCR 33%–67%, quantitative real-time PCR 57% (Eales’ disease), multiplex PCR 78%, normalised quantitative PCR 85%.25 A study from an endemic area found the rate of aqueous PCR positivity was affected by the ocular disease phenotype: 40% in serpiginous-like choroiditis, ampiginous choroiditis and multifocal choroiditis, vs 80% in choroidal abscess, miliary TB and choroidal tubercle.26 This perhaps reflects the likelihood of direct infection as compared with an immune phenomenon in response to remote infection.

A number of other studies have reported that a positive PCR is more likely in patients with ocular phenotypes suggestive of OTB, when compared with control groups, though again sensitivity is variable.1–5 26–29 A large clinicopathological study on 42 eyes showed very few bacilli identifiable in ocular tissues, thus highlighting the challenges in demonstrating intraocular infection by traditional means of culture.6 30

Overall, the diagnostic value of PCR techniques have improved, however a negative test does not exclude TB uveitis.

Clinical practice points

• PCR to detect Mtb of ocular fluids has a high specificity but has a low sensitivity.

• Aqueous fluid is easier to obtain and has similar yield to vitreous.

• Ocular fluid sampling may be considered in IGRA positive patients at high risk of TB but with a uveitis phenotype less typical of TB, or patients with another potential cause of uveitis.

Risk stratification for referral timeline

Introduction

A key principle of investigating OTB is to look for potential sites of TB outside the eye. In patients with suspected OTB further investigations are valuable to help support the diagnosis, rule out differential diagnoses and to guide sampling. Initial tests can also help rule out infectious TB, which could pose a risk to personal contacts, staff and other patients. Biopsy of the eye carries considerable risks and the sensitivity of TB PCR and culture is limited, so a key principle is to look for other potential sites of TB (see prior section on potential role for TB PCR in ocular fluid) which can allow:

• Microbiological confirmation of diagnosis.

• Exclusion of mimics such as sarcoidosis and non-tuberculous mycobacterial infection.

• Resistance testing to ensure the optimal treatment regimen is given.

• Epidemiological analysis by whole-genome sequencing.

However, there is no consensus on the optimal investigation pathway for patients with suspected OTB, in part reflecting the sparse evidence in this field. Several web-based surveys have been undertaken among uveitis specialists to assess international practice in areas of high and low TB endemicity,31–33 suggesting significant variation, and with a preference for CXR in areas of low TB endemicity (low TB countries 84% compared with 71% in high TB countries),33 but a preference for CT in areas of high TB endemicity (high TB countries 51% compared with 19% in low TB countries). In practice, a retrospective multinational cohort study of 945 patients diagnosed and treated for TB uveitis also highlighted significant regional variation in investigation practices with few centres conducting CT scans of the thorax.10

The London OTB Pathway is a pragmatic algorithm for investigation, and was created by consensus among practitioners in London.12 This formed the basis for discussions by this CSG (see figure 1).

Respiratory/TB clinic investigations

Clinical practice points

• Patients suspected of having OTB, should have an urgent CXR requested by ophthalmology.

• Consider CT thorax with contrast to guide sampling (for example induced sputum, bronchoscopy or EBUS) in patients with suspected OTB.

• The role of PET-CT to guide sampling is unclear, but it may detect additional areas of TB in extrapulmonary sites or within normal sized intrathoracic nodes. PET-CT may therefore be considered where it is available.

The tuberculin skin test

When tuberculin is injected intradermally, it generates a delayed hypersensitivity reaction within the skin of individuals who have previously been exposed to TB. Current clinical practice is to use purified protein derivative extract (PPD—a standardised mixture of over 200 mycobacterial proteins) for the test; this is known as the Mantoux test, where the borders of the reaction to intradermal PPD at 48–72 hours are defined by visual inspection and palpation and then measured.

Definitions of a positive reaction are determined by the dose and type of PPD used. In the UK, a dose is of 2 international units (IU) of RT 23 is used and a positive reaction (‘cut-off’) is seen when the skin induration is greater than or equal to 5 mm irrespective of BCG status.48 A recent multicentre prospective UK-based study (PREDICT)49 has demonstrated that all cutoffs for the TST evaluated still had a poor prognostic performance for progression to active disease from LTBI although a BCG stratified 5 or 15 mm cut-off was superior to the 5 mm cut-off .

Other issues with skin testing include the need for two clinic visits to complete the test, severe localised skin reactions including blistering and oedema can also occur and cause significant discomfort for the individual. Lastly, the test is operator dependent in terms of instillation and reading, and therefore, subject to a degree of potential error.

Interferon-gamma release-assays

IGRAs are immune-based blood tests for detecting Mtb infection by measuring T-cell responses (currently Interferon-gamma) to highly specific Mtb antigens (ESAT-6 and CFP-10) being the most well-established in commercial assays). They are not confounded by prior BCG vaccination and provide higher diagnostic specificity than the TST. Two assay platforms exist in current clinical practice: the T-Spot.TB, which utilises an ELISpot assay as its readout after stimulation of peripheral blood mononuclear cells by Mtb antigens; and the Quantiferon (QFN-GIT), which has an ELISA-based readout after a whole-blood stimulation. Both benefit from a single blood draw at a single clinic visit, in contrast to the TST. Now established as a standard-of-care for diagnosing LTBI, they offer a modest advantage over the TST for the prediction of progression from LTBI to active TB.50

A large-scale prospective head-to-head comparison of diagnostic performance of IGRAs in routine practice has been performed in the UK, and neither existing IGRA, that is, the T-SPOT.TB nor QFN-GIT, have sufficient sensitivity nor negative predictive value to rule out a diagnosis of active TB (IDEA study 2019).51 IGRAs are therefore not recommended in the diagnosis of active TB.

In the setting of immunosuppression as a result of underlying disease or iatrogenic causes, both the TST and IGRAs are attenuated and hence are less sensitive in those settings.52

With these general principles in mind, there is some limited literature on the utility of the TST and IGRAs in the diagnosis of OTB. In general, these studies do not consistently show a definite role for either approach to aid the diagnosis of OTB.53–55 Notably, a UK-based retrospective evaluation showed that of a group with clinically defined OTB, the sensitivity of the QFN was only 70.1%56; in contrast, the increased rate of QFN positivity in unexplained uveitis in a low incidence setting indicates that the test may identify some cases of OTB but with uncertain sensitivity and specificity.57

Clinical practice points

• The TST and IGRAs are useful immunological tools in identifying individuals with prior TB infection and are an additional and useful tool in a composite approach to diagnosing presumed OTB alongside ocular phenotype, epidemiology and imaging.

• IGRAs may be associated with false negative results in active TB and should not be used as the sole tool to rule out OTB. IGRAs have the advantage of the single clinic visit and increased specificity but where available, a combination of both IGRAs and skin tests may offer the best sensitivity.

Treatment with ATT

The treatment of OTB is controversial as the underlying pathogenesis of disease is unclear. The following issues remain unknown:

1. Whether the disease is primarily through direct infection of the eye in aetiology, or an inflammatory reaction to remote infection elsewhere.

2. Whether all ocular phenotypes should be treated in the same manner.

3. Whether there is any need to use adjunctive treatments such as laser or intravitreal steroids.

4. Whether ATT be given for 12 months in line with CNS disease or for 6 months, in line with conventional therapy.

In this section the literature to date and is presented and clinical practice points are based on the evidence available.

Although some studies have shown that administering ATT to patients with OTB is beneficial, not all studies have shown significant benefit. In a large meta-analysis, Kee et al did not find any significant difference between those treated with ATT and those who were not.58 However, this result likely reflects the nature of the studies analysed. The lack of high-quality studies reflects the difficulty in confirming the diagnosis (especially in areas of high incidence of TB, patients may have latent TB and uveitis of unrelated cause). In addition drug related toxicity, particularly drug-induced liver injury from ATT may be an issue for some patients.

Unfortunately, there are no published trials determining ideal treatment regimens. Most papers published are retrospective case series (ie, without control groups), often small in size, treatment was not standardised, and overall there is a high risk of bias.58

NICE guideline NG033 recommends a total of 6 months of treatment for TB in any part of the body except the CNS (brain, spinal cord or meningeal) for which 12 months is recommended.48 Clinical practice in the UK and globally for the treatment of OTB is variable which reflects the lack of consensus in the literature and lack of guideline recommendations.10

Antimicrobials and duration of ATT

A retrospective cohort study from a non-endemic area studied 175 patients with presumed OTB, all of whom were treated with ATT.8 Forty-three patients were given moxifloxacin instead of ethambutol. Treatment failure was defined as recurrence of inflammation within 6 months of completion of ATT or the inability to reduce prednisolone to <10 mg/day or topical corticosteroids to less than twice daily. Overall there was no significant difference in outcomes between those given regimens containing ethambutol or moxifloxacin. In those given longer courses of ATT (≥9 months vs ≤9 months), there was a trend to better outcomes on a bivariate analysis (p=0.06 with OR 0.5 (95% CI 0.24 to 1.05)). Sixty patients from this study were analysed in a separate paper.59 This paper showed increased treatment failure associated with ATT durations of less than 12 months (p=0.034).

In a retrospective cohort study from Singapore—a country of medium TB incidence—182 patients with a diagnosis of OTB were analysed.60 Of these, 64 were given ATT. The authors found that those given over 9 months of ATT had a lower rate of recurrence of eye disease than those who received no ATT (OR 0.09, 95% CI 0.01 to 0.76, p=0.027). In those given <6 or 6–9 months, there was less recurrence than those who received no ATT but this did not reach statistical significance.

Clinical practice points

• Standard regime ATT should be given for OTB (unless alternative drug sensitivities are available), in conjunction with a specialist team with experience treating OTB. Services should consider setting up regional MDTs which includes specialists in uveitis and TB.

• ATT should be given for at least 6 months. Teams can consider giving treatment for longer (9–12 months), especially if there is slow improvement in eye disease or disease is severe initially.

• It is reasonable to replace ethambutol with a fluoroquinolone (moxifloxacin or levofloxacin). This decision should be made in conjunction with a specialist in OTB (or MDT). Consideration should be made of potential adverse effects of fluoroquinolones (QTc prolongation, tendon rupture and aortic aneurysm rupture).

Systemic and local steroids in the treatment of OTB

Regardless of whether OTB is primarily due to direct infection of the eye or inflammatory disease of the uvea, rapid recognition and early intervention with high-dose corticosteroids is required to prevent loss of vision. It is important to note that systemic steroid doses should be doubled in the context of ATT using rifampicin. There are several intraocular steroid implants: Ozurdex (Allergan, Irvine, California, USA) and Iluvien, (Alimera, Alpharetta, Georgia, USA) both of which have been licensed to treat different aspects of uveitis. The next section covers therapy for the various ocular phenotypes associated with OTB.

Anterior uveitis

Clinicians are increasingly choosing to manage anterior uveitis with ATT. A recent paper from Elangovan et al explored the effect of ATT on OTB, the authors described 83% of their cases as unilateral, with non-granulomatous anterior uveitis representing 24.1% of their overall cases.61 Tomkins-Netzer et al explored the effect of ATT in a larger cohort from a non-endemic area and described a similar proportion of patients with non-granulomatous anterior uveitis.53 These findings are of interest as they reflect a paradigm shift in the management of OTB. Historically, only patients with posterior segment findings would be referred for consideration of ATT as these were considered ‘sight threatening’. Recognition of the risks associated with prolonged use of topical corticosteroids (cataract, glaucoma, keratopathy) along with a greater understanding from ophthalmologists around the world that patients with OTB may benefit from ATT has resulted in a higher proportion of patients with anterior uveitis being treated. A joint publication created by ophthalmologists, respiratory, infectious diseases and public health physicians from London was published in 2019. Based on consensus methodology, the group agreed that patients requiring treated for active OTB should be referred for 9 months of treatment.62

To date, there remains no strong evidence that ATT is of benefit in the management of anterior uveitis. This is because there is neither an effective diagnostic test for OTB nor a well-defined outcome measures in anterior uveitis. Studies have been retrospective and relied on patients presenting to ophthalmic services and reporting a flare of their uveitis or collecting data on the number of topical corticosteroids prescribed.

Clinical practice points

• Patients in whom AU could be a manifestation of OTB should be seen by an ophthalmologist experienced in the management of OTB to determine a management plan. This is because of the difficulty in determining the likelihood of the AU being caused by TB.

• Patients with chronic granulomatous anterior uveitis typical of OTB and a high clinical or epidemiological suspicion of OTB, should be managed with ATT and topical corticosteroids regardless of their IGRA or TST result.

• Consider ATT in patients with chronic anterior uveitis of unclear cause requiring more than two drops of corticosteroid per day and a positive IGRA or TST.

• ATT may also be considered for positive IGRA or TST patients with recurrent anterior uveitis, without other cause, who suffer more than two episodes per year.

Intermediate uveitis

Agarwal et al describes the use of Ozurdex (Allergan, Irvine, California, USA), a slow release dexamethasone implant in the management of OTB.63 Eleven of 17 cases described had either intermediate uveitis or retinal vasculitis and the majority demonstrated an improvement in vitreous haze, visual acuity and macular oedema over 3 months.

Interferon alpha 2a has also been reported as an effective treatment in two small studies in patients with intermediate uveitis secondary to OTB.64 65 Both papers described a good response at a dose of 3 million IU per day in patients who had been treated with ATT and had continued inflammation. Of note, no patient developed signs of systemic TB while taking interferon alpha 2a.

Clinical practice points

• The use of systemic corticosteroids, local corticosteroid and other immunosuppressive therapy should be guided by the extent of the disease, evidence of structural damage and response to ATT.

• It is reasonable to use intravitreal steroids in the management of OTB.

Posterior uveitis

Posterior uveitis is often sight-threatening and treatment should be adjusted accordingly. There are a number of papers reporting encouraging outcomes for patients with inflammatory lesions affecting the posterior pole, however, none of these were conducted prospectively in a controlled fashion. Ghauri et al describe a cohort of patients with OTB from Pakistan.66 Ninety percent of patients in this cohort had posterior uveitis and all received 12 months of ATT. The authors reported that 80% of patients achieved disease quiescence at 12 months with only three patients requiring oral corticosteroids.

The COTS study also reported posterior uveitis as the most common manifestation of OTB (36.3%). Detailed outcomes for these patients were not possible due to study design. In contrast to the study by Ghauri et al, 89% of patients required oral corticosteroids (9.3% went on to be treated with other immunosuppressive medications). The COTS data also suggested that patients with a positive IGRA who received systemic corticosteroids had a higher incidence of treatment failure. The authors commented that this finding was largely due to patients commencing oral corticosteroids prior to ATT, however, this finding remains a surprising outcome from this study. As ATT is not thought to directly affect inflammatory lesions in the uvea, local tissue destruction will continue unless the inflammatory aspect of the disease is addressed. Of interest, another retrospective series from Spain found that ATT alone was associated with improved visual outcomes in OTB, although this paper did not adjust for the location of the inflammatory lesions.67

Kaza et al describe the use of therapeutic pars plana vitrectomy in the management of OTB affecting the posterior pole. Only patients with bilateral disease were analysed and the better of the two eyes was used as a control. Twenty-five per cent of patients in this study had posterior uveitis with the rest having retinal vasculitis. The authors reported complete resolution of disease at 3 months in 96.7% of operated eyes compared with 83.3% of control eyes (p=0.1945) While this paper remains the only study to use vitrectomy as a therapy for OTB, it is notable for obtaining a 72.2% positive TB PCR from the vitreous, significantly higher than other studies.15 27

Jain et al reported outcomes using the Ozurdex implant to manage posterior uveitis secondary to OTB.68 They reported that 88.8% of eyes treated with ATT and Ozurdex showed no sign of recurrence. Similar findings were reported by Agrawal et al.63 Importantly, the use of intravitreal corticosteroids avoided the need for these patients to be exposed to high-dose systemic corticosteroids.

For the most part, the literature on OTB uses the terms ‘retinal vasculitis with or without occlusive features’. Occlusive retinal vasculitis is thought to be a typical manifestation of OTB. It presents in two phases: an initial inflammatory phase and a later ‘surgical’ phase. In the initial phase, ATT and steroids are commonly given, with the addition of anti-inflammatory agents (disease-modifying or immunomodulatory drugs and anti-TNF therapy) in those unable to wean prednisolone to below 10 mg/day. In the surgical phase medical therapy is likely to be ineffective and surgery (pan retinal photocoagulation) is needed.

The largest study to date examining the effect of ATT on retinal vasculitis is a subset of the COTS data.11 The authors described a series of 251 patients from 25 centres around the world. The authors did not find any effect of ATT on treatment failure (as defined by recurrence of inflammation within 6 months of completing ATT; inability to taper below 10 mg/day of oral prednisolone; or need to escalate to further immunosuppression).

No evidence has been published which supports the use of specific anti-inflammatory or anti-TNF therapy.

Clinical practice points

• Patients with a high clinical or epidemiological suspicion of OTB, and occlusive retinal vasculitis typical of OTB should be managed with ATT and high-dose corticosteroids (1 mg/kg prednisolone or equivalent) regardless of their IGRA or TST positivity. Laser retinal photocoagulation of ischaemic retina should be considered if there are signs of neovascularisation.

• Patients with a high clinical or epidemiological suspicion of OTB, and choroidal lesions typical of OTB should be managed with ATT and high-dose corticosteroids (1 mg/kg prednisolone or equivalent) regardless of their IGRA or TST positivity.

• All other patients with posterior uveitis and a positive IGRA or TST should have other causes of uveitis excluded before being referred for ATT. The use of systemic corticosteroids, local corticosteroid and other immunosuppressive therapy should be guided by the extent of the disease, evidence of structural damage and response to ATT.

Clinical practice points

• Visual acuity alone is insufficient as a measure of outcome from treatment.

• Improvement in fundus imaging can be used as an outcome measure when chorioretinitis and/or retinal vasculitis is affecting the peripheral retina.

• The SUN criteria are useful standardised methods of assessing uveitis activity.

• A reduction in steroid dose (typically below 7.5 mg per day) can be used as an outcome measure.

Incidental LTBI

In some patients, LTBI will be identified by a positive IGRA/TST as a coincidental finding in the absence of any TB symptoms, with a normal CXR or chest CT and without consistent eye disease. The risk for TB reactivation and development of active disease in these patients depends on multiple factors. Treatment is advised based on an assessment balancing the risk of infection, the likelihood of progression to active disease and the benefits of treatment weighed against the risk of potential significant drug-related adverse effects such as hepatotoxicity.3 48 75 76 It is therefore recommended that people identified coincidentally to have LTBI who are aged younger than 65 years and who have prior TB contact, or who are recent new entrants to the country (<5 years from a high incidence country, where TB incidence ≥100/100 000), or who have other risk factors for TB reactivation (eg, HIV infection, diabetes, silicosis, on immunosuppressive therapy) are referred to the local TB service for possible LTBI treatment, according to current guidance.48 77

Clinical practice points

• Patients identified to have LTBI in the absence of eye disease compatible with OTB, and who are defined as eligible, should have prophylactic treatment according to current NICE guidance.

Prebiological screening for inflammatory eye disease

The introduction of biological therapies has had a significant beneficial impact on many inflammatory diseases particularly those affecting joints, the GI tract and skin; recent evidence suggests that they may also have a role in managing inflammatory ocular disease. In particular, anti-TNF-α therapies seem most promising, with an expert international panel recommending Infliximab and Adalimumab as first-line or second-line immunomodulatory agents for the treatment of ocular manifestations of Behçet’s disease, juvenile arthritis and severe ocular inflammatory conditions.78 There is ongoing research into other biological targets including recombinant interferon (IFN)-α and monoclonal antibodies targeting different interleukins.79 80 Currently, adalimumab is the only anti-TNF agent licensed by the European Medicines Agency for the treatment of uveitis.

Anti-TNF-α treatment is associated with increased rates of TB reactivation81–83 as a result of these agents modifying the host immune response against TB. Screening and treatment for LTBI before beginning biological therapy significantly reduces the risk of reactivation.84 The reactivation risk varies with different biological agents,83 with anti-TNF-α drugs (especially Infliximab and adalimumab) more highly associated with reactivating LTBI and the non-anti TNF-biologicaltreatments being less risky.

Wakefield et al published recommendations based on expert committee opinion on the assessment of patients with inflammatory eye disease preimmunosuppressive and biological therapy,85 including screening for infections including TB. The CSG was unable to find any other published guideline. However, multiple safety guidelines exist, from different countries, for the use of biologics in other inflammatory conditions. There is universal agreement that all patients should undergo TB screening and if identified to have LTBI should begin treatment for this, before starting any biologics.86 However, there is variation regarding the method of diagnosis of LTBI (by TST and/or IGRA) in patients who are frequently immunocompromised.86

WHO recommend the use of either TST or IGRA,77 whereas the UK 2016 NICE guidelines48 recommend testing all immunocompromised patients with IGRA alone or with IGRA and TST; a positive result in either should prompt treatment.

Hence all patients as part of a screening baseline assessment preinitiation of biologics should be clinically assessed to exclude active TB disease based on a thorough clinical history including exploration of possible previous TB exposure or past history of TB treatment, alongside clinical examination and CXR. If there is any evidence of active TB the patient should be referred to a TB specialist team for further investigation and treatment. A patient who has had inadequate or unclear TB treatment for prior active disease or latent infection should be referred to a TB specialist (for exclusion of active TB, or consideration of LTBI treatment).

Patients with LTBI should be referred to a TB specialist for LTBI treatment. There is a lack of data suggesting when patients may begin biological therapy; ideally biological therapy should be started after completion of TB therapy, however, this may not always be possible due to the severity of disease requiring biological therapy. Evidence is lacking; expert opinion suggests patients should complete at least 1-month LTBI treatment to allow time to tolerate the treatment or identify side effects, before starting biological therapy.87 Patients treated for active TB should complete at least 3 months treatment and have clear evidence of improvement including culture negativity prior to initiating biological therapy.87

Clinical practice points

• Patients due to start disease modifying treatments, including immunosuppressive or biological therapy, should be discussed with their local TB centre to confirm if screening for LTBI is required for that agent.

• Ideally screening for LTBI prior to initiating disease-modifying treatments should be by IGRA alone or by IGRA and TST .

• Patients identified to have LTBI should have prophylactic treatment according to current NICE guidance.

• Patients should ideally complete treatment for LTBI for at least a month before starting disease-modifying treatments, as indicated by their eye disease.

Ethambutol toxicity

Ethambutol is commonly used in treatment regimes for the eradication of mycobacterial species. It is bacteriostatic and, with other antibiotics, effective. However, one of the most common and potentially serious side effects is an optic neuropathy that can lead to blindness. Ethambutol’s principal action is as a metal chelator and it has been hypothesised that the optic neuropathy may arise from disrupted oxidative phosphorylation due to lack of copper available to mitochondria or from the chelation of zinc leading to the inhibition of lysosymal activation.

The prevalence of the optic neuropathy depends on the dose and the duration of the treatment. Overall it is in the region of 1%–2%. It has been observed in the intermittent dosing of directly observed treatment strategies.88 Higher doses are used in non-tubercular infections such as M. avium complex and M. kansasii. Ethambutol toxicity in adults is uncommon, occurring in less than 2% of patients on the standard dosage of 15 mg/kg/day. The risk of optic neuropathy has been estimated to be 3% with a dose of 20 mg/kg/day, 5%–6% with 25 mg/kg and 18%–33% with 35 mg/kg. Other risk factors for the development of ethambutol optic neuropathy include age >65 years, hypertension, smoking and the presence of renal disease as the drug is excreted through the kidney.89 90 Given that isoniazid has been independently associated with an optic neuropathy, the combination with ethambutol may put the patient at higher risk. The toxicity usually manifests from 2 to 5 months after the start of treatment but has been reported within days. The prevalence appears to be lower in children than in adults. This may be because the serum concentration in children is smaller in children than adults for a given dose.

The clinical features of ethambutol toxicity are well known and comprise bilateral, painless, subacute and symmetrical loss of central vision. Examination reveals loss of central as well as colour vision with a central or centrocaecal scotoma. Bitemporal hemianopia and peripheral field constriction have been reported. In the later stages the optic disc may become pale and this is regarded as a poor prognostic sign. Ancilliary testing with visual evoked potentials, optical coherence tomography and multifocal electroretinography have all been used to try and detect the neuropathy at a preclinical stage. While these tests offer the potential to achieve this, their results are currently inconsistent and the tests are not readily available in resource poor settings.

If ethambutol optic neuropathy is suspected the drug should be stopped immediately but withdrawal is not always effective in reversing its effects. It has been estimated that between 35% and 65% of patients will show some improvement of their visual disturbance over a period of some months.91

Regular screening of adults and children taking the drug is not considered necessary but a baseline visual assessment should be undertaken before starting the drug and patients should be instructed to discontinue the drug if they have fresh visual symptoms and seek immediate review. Since Ishihara colour testing is commonly used to detect early neuropathy, alternative strategies may be required in those who are congenitally colour blind.92–95

Ethambutol is often not included in treatment regimens for OTB. This may be due to concerns of causing an optic neuropathy in a patient with pre-existing eye pathology. However, there is no evidence to suggest that optic neuropathy is more likely to occur in a patient with OTB than a patient with healthy eyes or indeed any other eye pathology. The ophthalmologist should ensure that if the patient’s vision allows, colour vision is checked at baseline and during therapy. In the case of doubt over the cause of visual loss it is safer to switch to a regime without ethambutol.

Prescribers and TB teams should continue to enquire about changes in visual symptoms and refer any reported changes from baseline back to the ophthalmic team involved. If ethambutol toxicity is suspected the drug should be stopped immediately pending an ophthalmologist review.

Clinical practice points

• Establish whether there is any history of eye disease and a baseline visual assessment (visual acuity and Ishihara colour plates) should be obtained before prescribing the drug.

• Patients should be advised to stop taking ethambutol if they experience new visual symptoms and seek prompt medical/ophthalmic review. Patients should be asked about vision at every visit.

• If the patient reports changes in colour vision or reduced central vision, leading to suspicion of ethambutol toxicity, the drug should be stopped immediately pending urgent ophthalmic review.

• Screening for ethambutol toxicity should be conducted at eye checks for their inflammatory eye disease, but the TB team should ensure the eye team are aware that ethambutol is part of the regime.

Who to avoid treating

A careful discussion of risks and benefits of TB treatment is required with patients, particularly those at high risk of side effects, for example, those over 75 years, or with pre-existing liver disease. The discussion should also include the reduced risk of recurrence of uveitis with treatment; and risks of long-term immunosuppression, and also visual loss if the uveitis is severe without ATT. This should usually be an MDT discussion with the ophthalmologist, TB physician and the patient.

Most trials of ATT in OTB show a greater relapse rate, and higher rates of systemic treatment in uveitis when ATT is not given, although to the authors’ knowledge there are no randomised controlled trials (RCTs) of ATT vs no ATT in OTB, which is important in a relapsing remitting condition to identify different cohorts. In one trial of ATT in OTB,96 14/350 (4%) of patients did not complete the TB treatment. In counterpoint, medication can cause intolerance and adverse effects with rates of hepatotoxicity increasing with age.97 Older patients can have higher treatment interruptions (18%) due to toxicity, and lower completion rates (70%),98 though others have found no difference.99 Consider not treating for OTB where the evidence of TB is considered unrelated to ocular disease—for example if there is a clear alternative cause of uveitis and the evidence for TB is a positive IGRA, or changes of old TB on the CXR alone. If eligible, however, consider treatment for latent TB in such patients. The decision to immunosuppress for uveitis may also influence the decision to treat TB.

It is recommended not to retreat patients for TB who have had previous adequate treatment for TB uveitis with no overt evidence of active TB elsewhere, after a careful search for TB reactivation and exclusion of drug resistance and alternative diagnoses, as described in section 5, as ocular inflammation can persist after treatment.100

Clinical practice points

• Non-randomised data have shown that most patients with likely TB uveitis will benefit from ATT with a reduction in uveitis activity and relapse compared with patients who are not treated.

• Consider withholding TB treatment when the evidence for TB exposure is separate to the cause of the uveitis, for example old TB CXR or CT changes of TB or a positive IGRA with prior history of TB and a clear alternative cause for uveitis. This should also take into account the risks of TB treatment in the individual.

• Patients should be included in the decision to treat. Where OTB is the only manifestation of TB this usually also requires an MDT discussion involving the uveitis and TB specialists, including the risks of alternative treatments and potential visual loss.

Research recommendations

RCTs are needed, to determine:

• Whether ATT is of benefit in presumed OTB with no evidence of clinical TB in other sites.

• The sensitivity and specificity of IGRAs or other biomarkers in the diagnosis of OTB.

• The optimal modality or combination of modalities using imaging to detect OTB.

• The optimal composition of ATT, especially to clarify the role of ethambutol and fluoroquinolones.

• The optimal duration of ATT.

• The role of extending ATT if disease does not respond in the timescale or manner expected.

• The benefit of ATT and steroids in OTB in the different eye compartments.

• The role of immunomodulatory/disease-modifying treatment in posterior uveitis.

• The role of surgery in intermediate uveitis.