Tuberculous Uveitis

Mycobacterium tuberculosis (MTB) is an extraordinarily common human pathogen, having infected approximately one-third of the world’s population.1–3 According to the World Health Organization (WHO), 9.0 million people developed active tuberculosis (TB) in 2013 – an overall annual incidence of 126 per 100,000 persons – and among this group one in six, or 1.5 million, died from the disease. The highest incidence rates for active TB were reported from Africa (280 per 100,000 persons), Southeast Asia (183 per 100,000 persons) and India (168 per 100,000 persons). Persons infected by the human immunodeficiency virus (HIV) were at particularly high risk of developing active disease in the WHO report, with an annual incidence of 3142 per 100,000 persons, or 25 times the overall global rate.4 During this same period in the US, by comparison, the overall case rate of active TB was 3 per 100,000 persons. Foreign-born persons constituted approximately two-thirds of all cases of active TB in the US and as a group had an incidence rate 13 times higher than those born in the US (15.6 versus 1.2 per 100,000 persons).5

Inhalation is the most common route of exposure to MTB. In the vast majority of otherwise healthy exposed people, a robust adaptive immune response quickly contains the infection, such that those exposed to MTB are said to have only a 10% life-time risk of developing active pulmonary or extrapulmonary disease.1–3 Among those with active TB, the reported prevalence of intraocular tuberculosis (IOTB) varies widely, from approximately 1% in patients with pulmonary TB to over 20% in those with extrapulmonary infection.2^,3 The reported regional rates of tuberculous uveitis also vary, from less than 1% in tertiary referral clinics in North America, to 10% or more in similarly specialized clinics in highly endemic regions.2^,3

Although establishing prior exposure to MTB through the use of tuberculous skin testing (TST), an interferon-γ release assay (IGRA), or chest X-ray or CT is fairly straightforward and can usually be done with a positive predictive value in excess of 90%, exposure is by no means equivalent to active infection. In fact, given the global prevalence of latent infection, one would expect one-third or more of all uveitis patients in highly endemic areas to have some evidence of prior MTB exposure. Moreover, many techniques used to diagnose active pulmonary or extrapulmonary TB, such as culture and biopsy, have limited utility in uveitis.2^,3^,6^,7 In this issue of Ocular Immunology & Inflammation (OII), Dr Amod Gupta et al. address many of the issues related to diagnosis and propose a new classification scheme for IOTB based on their experience in a tertiary referral center in a highly endemic region of North India.8 Specifically, the authors propose classifying IOTB as “confirmed”, “probable” or “possible”. The designation of confirmed IOTB would be limited to those patients for whom other causes of uveitis had been excluded, with at least one clinical sign suggestive of IOTB and microbiological confirmation of MTB from ocular fluids or tissues. In the authors’ clinic in North India, suggestive clinical signs in eyes with active uveitis include broad posterior synechiae, retinal perivasculitis with or without discrete choroiditis/scars and often with capillary non-perfusion, multifocal serpiginoid choroiditis (MSC), choroidal or optic disc granuloma(s), and optic neuropathy. Retinal perivasculitis and MSC, in particular, have been reported to have positive predictive values of 90% or greater.9^,10 In non-TB-endemic areas, broad or extensive posterior synechiae occur much more frequently in eyes with HLA-B27- or sarcoid-associated uveitis than IOTB and so this particular sign may be less predictive in such settings. Microbiological confirmation of MTB might include a positive culture, histological identification of acid fast bacilli or polymerase chain reaction (PCR)-based amplification of MTB DNA. Given the limited size of most ocular samples and the fact that MTB infections tend to be paucibacterial, PCR for one or more MTB DNA-coding regions has become the most commonly employed technique for establishing infection. In fact, the sensitivity of the multiplex, multi-targeted PCR assay employed at Dr Gupta’s institution has been reported to approach 80 and 100%, respectively.11 Of note, the receiver operating characteristic (ROC) of single-target assays not standardized for testing ocular fluids can be much less favorable.12 The authors propose the diagnosis of “probable IOTB” be used in patients with active uveitis, but no microbiological confirmation when: (1) other causes of uveitis have been excluded; (2) at least one suggestive clinical sign of IOTB is present; (3) there is evidence of pulmonary or extrapulmonary TB; and (4) there is either a documented TB exposure history OR immunological evidence of TB infection – such as a positive TST or IGRA. “Possible IOTB” would require one or more similarly suggestive clinical findings in addition to EITHER direct evidence of pulmonary or extrapulmonary TB OR exposure history and TST/IGRA evidence of prior exposure, but not both. While a favorable response to anti-tubercular therapy (ATT) was not included in the classification scheme, the authors acknowledged that such evidence can, ultimately and in retrospect, lend support to the diagnosis. Of course, it is important to remember that IOTB may present with a wide spectrum of clinical findings, many of which are less suggestive than those mentioned above. These varied presentations are thoroughly summarized by Dr Vishali Gupta et al. in this issue of OII.13 Five additional original articles,14–18 one brief report19 and four letters20–23 in this issue of OII present important findings related to the diagnosis and treatment of tuberculous uveitis.

Lou et al.14 report the results of two international surveys of the approach to diagnosis and treatment of IOTB – one completed by 87 members of the American Uveitis Society (AUS), most of whom were from either the US or Europe, and a second by an expanded group of 143 participants, nearly half of whom were from the developing world.15 Both surveys highlight the lack of consensus regarding the approach to diagnosis and treatment – a common finding in surveys of this sort.24 Moreover, both studies point to the common practice of routine testing for prior MTB exposure, be it by TST, IGRA, or chest X-ray or CT, despite anatomical location and regardless of the presence or absence of high risk features. Overall, screening tests were performed more often in developing countries, many of which are in TB-endemic regions, and screening in these areas tended to be more extensive with inclusion of a chest CT and both a TST and IGRA. This practice was particularly common in India.15 While such broad screening may provide valuable information in patients in or from TB endemic regions,2^,3^,8^,25 and is recommended for patients with suggestive clinical features as described above9 and prior to initiating long-term immunosuppression – particularly with tumor necrosis factor (TNF)-α inhibitors,26 its routine use in low-risk populations use has been called into question.27 Moreover, in both developed and developing countries, the vast majority of respondents indicated that the presence of either radiographic or immunologic evidence of prior MTB exposure would prompt them to initiate ATT, even in the absence of suggestive clinical findings. Of note, such patients would not have qualified for the designation of “possible IOTB” as proposed by Gupta et al.,8 suggesting that a large proportion of respondents may use somewhat liberal treatment criteria. Regarding the choice of agents, most respondents agreed that ATT should include multiple drugs, with rifampin, isoniazid, pyrazinamide and ethambutol constituting the most common first-line choices. Overall, 86% of respondents indicated that ATT should be continued for six months or more, with roughly half treating for nine to 12 months. The concurrent use of oral corticosteroids was recommended by approximately 50 and 25% of respondents from the developing and developed world, respectively. Respondents from the developed world were much more likely to defer specific treatment decisions to an infectious disease or TB specialist (76 versus 34%; nominal p < 0.0001).15

Agrawal et al.16 evaluated the response of 175 patients with presumed ocular tuberculosis who were administered ATT. The treated cohort constituted 46.7% of all patients diagnosed with presumed ocular tuberculosis seen by the authors in a tertiary uveitis referral clinic in London. Of note, patient-specific criteria for the diagnosis of presumed ocular tuberculosis varied widely within the cohort and the choice of whether or not to administer ATT with or without corticosteroids or non-corticosteroid immunosuppressive agents, and for how long, was left to the discretion of the treating non-ophthalmologist physician with input from the referring ophthalmology team. Hence, as with all such retrospective, clinic-based studies, the potential for both confounding and bias – including referral, selection/testing and treatment bias – was high. Only eight patients (4.6%) had systemic TB, including four with active pulmonary infection. Just under two-thirds of patients were male (61.1%) and Asian (64.8%), and most (66.3%) had bilateral uveitis. While most patients (62.9%) had clinical findings suggestive of ocular tuberculosis – such as retinal vasculitis or choroiditis, the reader is left to assume that the remainder did not. The vast majority of patients (95.4%) were IGRA positive, whereas 14.5% had supportive chest X-ray findings. Duration of ATT varied from three to 12 months (mean 10.5 ± 2.5 months) and tended to be longer in patients with more than one risk factor for tuberculosis. Treatment failure was defined as persistent or recurrent inflammation within six months of completing ATT, or the inability to taper oral corticosteroids below 10 mg/day, to taper topical corticosteroids to less than twice daily, or to stop oral non-corticosteroid immunosuppressive agents. Overall, and importantly, 134 or the 175 patients (76.6%) administered ATT with or without adjunctive anti-inflammatory therapy achieved control – a figure in line with results of other clinic-based studies.28 Bivariate analyses were used to explore factors associated with treatment failure. Systemic non-corticosteroid immunosuppressive agents were required in 20 patients, and their use was associated most strongly with treatment failure (75%; nominal p value < 0.001). While immigration per se appeared not to have been evaluated as a risk factor for response to ATT, bivariate analysis suggested that the 25 patients from Africa (14.3%) were more likely to fail treatment (nominal p value 0.03). Other factors associated with a higher ATT failure rate on uncorrected bivariate analysis included intermediate location (nominal p = 0.04), ATT duration equal to or greater than 9 months (nominal p = 0.06), and failure to use oral corticosteroids (nominal p = 0.08). In multivariate analysis using logistic regression, however, only the use of systemic non-corticosteroid immunosuppressive agents achieved statistical significance (p < 0.001). It remains unclear, however, whether patients with inflammation, that is sufficiently severe to require the addition of a non-corticosteroid immunosuppressive agent, would have had even worse inflammation if treated with ATT and corticosteroids alone. The authors cite earlier studies that have reported improved success in the treatment of presumed intraocular MTB with longer duration ATT.28

Kataria et al.17 from Chandigarh, India, studied the sensitivity and specificity of PCR-based testing for the devR and MPB64 genes of MTB in 25 patients with presumed IOTB based on the presence of suggestive clinical findings and corroborative evidence – such as a positive TST, IGRA or chest X-ray. The two control groups, each with 25 patients, included a heterogeneous group with non-tuberculous uveitis and a group without uveitis. The sensitivities of devR and MPB64 gene amplification in eyes with presumed IOTB infection were 64 and 72%, respectively. Specificity for the two PCR-based tests was 100% in that no patient in either control group had a positive test. Among the 20 patients with presumed intraocular MTB who were positive for one or both tests, 14 had concordant and 6 had discordant results. The sensitivity of either test being positive was 80%, a figure similar to that reported by Sharma et al.11 using multi-targeted PCR. Although the numbers were small in the study by Kataria et al.,17 trends suggested that PCR-based testing might be more likely to be positive in eyes with highly suggestive clinical findings, including retinal vasculitis with vitreous hemorrhage and choroiditis.

Rifkin et al.18 performed multimodal imaging on a 60-year-old female immigrant from Poland with bilateral MSC. Evidence of MTB infection included a positive IGRA, radiographic evidence of inactive granulomatous lung disease, and history of exposure to her father who worked in a TB sanatorium. Whereas spectral domain-optical coherence tomography (SD-OCT) imaging of inactive areas in each eye showed atrophy of both the outer retina and underlying choroid, imaging of an active leading edge in the right eye revealed irregular disruption of the outer retinal hyper-reflective bands associated with both the photoreceptors and retinal pigment epithelium (RPE), along with pronounced thickening of the underlying choroid on enhanced depth imaging (EDI)-OCT. Fundus autofluorescence showed hypoautofluorescence in inactive, atrophic areas, but stippled hyperautofluorescence associated with the active leading edge on the right. Although the patient initially improved on systemic corticosteroids and four-drug, first-line ATT, new lesions appeared bilaterally and the previously active lesion in the right eye showed re-activation with an increased thickness of the underlying choroid on EDI-OCT when the systemic corticosteroids were tapered and she went from four- to two-drug ATT. Increasing both the corticosteroids and number of ATT agents lead to dramatic resolution of the lesions. The authors suggested that localized thickening of the choroid on EDI-OCT imaging may be a sign of choroidal granuloma formation, supporting the diagnosis of tuberculous MSC.

Agrawal et al.19 describe a 22-year-old Asian Indian man with multi-drug resistant (MDR) spinal tuberculosis. Four weeks following initiation of second-line ATT, which included pyrazinamide, moxifloxacin, cycloserine and linezolid, the patient presented with 6/24 vision on the right and 6/9 vision on the left. Ophthalmoscopy revealed disc hyperemia with peripapillary hemorrhages in each eye. The presence of a right afferent pupillary defect (APD), bilaterally decreased color vision and cecocentral scotoma with low amplitude waveforms on pattern visual evoked potentials (VEP) in each eye confirmed the diagnosis of bilateral optic neuropathy. Given previous reports of linezolid-induced optic neuropathy, the linezolid was replaced by prothiomanide and within one month the patient’s vision returned to 6/6 bilaterally, the disc hyperemia decreased, and both the peripapillary hemorrhages, visual field defects and the right APD resolved. The authors cite earlier reports suggesting that 10–15% of patients receiving linezolid for MDR-MTB will develop evidence of optic neuropathy, particularly at doses at or above 600 mg daily.

Sharma et al.20 describe a 28-year-old Asian Indian woman from Chandigarh, India, with a large choroidal granuloma in the setting of a markedly positive TST who failed to respond to prolonged treatment with combined systemic corticosteroids and first-line four drug ATT, including isoniazid, rifampicin, ethambutol and pyrazinamide. Polymerase chain reaction-based testing of intraocular fluid, followed by sequencing of the rpoB gene using the Cepheid GeneXpert MTB/RIF assay confirmed the presence of MTB DNA and a specific mutation at codon 516 resulting in rifampicin resistance. The patient’s treatment was then changed to a non-rifampicin containing second-line ATT, resulting in regression of the choroidal granuloma and resolution of inflammation. The authors state that while systemic corticosteroids and first-line ATT results in recurrence-free disease resolution in 85% of patients in their setting,16^,28 drug resistance needs to be considered when response to therapy is poorer than expected.

Yilmaz et al.21 describe a 20-year-old man from Turkey who presented with mildly decreased vision in his left eye for one month. Ophthalmoscopic examination reviewed two separate choroidal lesions with overlying serous retinal detachment. The right eye was unremarkable. A TST was positive and a chest CT showed diffuse, small radio-opacities in the lungs consistent with miliary tuberculosis. The patient was treated with ATT, but, paradoxically, following one month of therapy the lesion had enlarged and vision worsened. Treatment was continued and, by two months both the pulmonary and ocular findings improved. Vision at last visit and following 12 months of ATT stabilized at 20/100. The authors cited earlier reports of paradoxical worsening, or the localized Jarisch–Herxheimer-like reaction, in eyes with tuberculous uveitis, and note that while not used in their patient, concomitant corticosteroids can be used effectively in this setting.

Ostheimer et al.22 describe a 62-year-old female immigrant from Philippines who reported decreased vision in her right eye for an year. On clinical examination, the affected eye had light perception vision and was noted to have a large anterior chamber mass. The fellow eye had moderate, granulomatous, anterior uveitis and a vision of 20/35. Both a TST and an IGRA were positive, whereas a chest X-ray was normal. Treatment with ATT and topical corticosteroids for nine months resulted in resolution of the inflammation in the left eye, but no improvement in vision in the eye with the presumed tuberculous granuloma, which became painful and was removed. Histological examination of the enucleated eye revealed necrotizing granulomatous inflammation, but neither culture nor PCR testing of tissue sections from the gross specimen provided evidence of MTB, which the authors attributed to testing having been done four months after initiation of ATT. The authors cite previous reports of tuberculous granulomas, none of which occupied the entire anterior chamber.

Chong et al.23 describe a 12-year-old Caucasian girl with persistent, painless, non-pruritic skin lesions found on biopsy to be granulomatous and caused by disseminated Mycobacterium avium-intracellular complex (MAC). Genetic testing provided evidence of a defect in the patient’s interleukin-12 (IL-12) receptor, confirming the suspicion of a primary immunodeficiency. The lesions improved following a 12-month course of systemic ethambutol and clarithromycin, but after 10 months into this therapy the patient developed a scotoma in her right eye and was found to have 6/120 vision and an inactive, fibrotic scar in the central macula. Multiple choroiditis scars where also present. The authors remind us that the occurrence of MAC infection in patient with no obvious cause for immunosuppression, such as human immunodeficiency virus (HIV) infection, use of systemic immunosuppressive agents, or an underlying malignancy, should prompt consideration of an underlying defect in IL-12 or interferon-γ (IFN-γ) signaling, conditions collectively known as Mendelian Susceptibility to Mycobacterial Diseases (MSMD).29 In healthy individuals, mycobacterial infection leads to IL-12 induced production of IFN-γ, which in turn directs antigen-specific T-cells and macrophages toward granuloma formation and tumor necrosis factor-α (TNF-α) release required for killing of intracellular pathogens. In a similar fashion, systemic TNF-α inhibition puts patients at an increased risk of mycobacterial infection.26

Together, these studies highlight a number of important issues related to the diagnosis and treatment of IOTB. While the classification scheme proposed by Dr Amod Gupta et al.8 is an important advance, both widespread availability of standardized multi-target PCR testing of ocular fluids for MTB DNA of the sort used by Dr Gupta and his colleagues,11^,17 as well as consensus testing and treatment guidelines are still lacking. These, it would seem, should be near-term priorities for both the field and our patients.

Acknowledgements

We thank Dr Amod Gupta for thoughtfully commenting on an earlier version of this editorial.

Declaration of interest

The authors have no relevant financial conflicts.

Supported in part by The Pacific Vision Foundation (ETC) and The San Francisco Retina Foundation (ETC).