Introduction

Ocular tuberculosis in non-endemic areas is usually more discreet and diagnosis is more difficult than in endemic areas. Moreover, unlike in endemic areas, polymerase chain reaction (PCR) identification of Mycobacterium tuberculosis DNA in ocular fluids is often negative. The choroid is frequently involved in ocular tuberculosis [1] but choroiditis can sometimes be very faint. The most sensitive method to explore and monitor the inflammatory involvement of the choroid is indocyanine green angiography (ICGA) often detecting subclinical choroidal inflammation [2]. For example, subclinical presumed tuberculous choroiditis was detected using ICGA in a patient presenting with isolated papillitis and responded to anti-tuberculous therapy alone [3]. ICGA was also the most useful method to follow interferon-gamma release assay (IGRA)-positive multifocal-serpiginous choroiditis [4].

Significant clinical evidence indicates that IGRAs are more specific than the purified protein derivative (PPD) skin tests in diagnosing tuberculosis as it is not confounded by prior bacillus Calmette-Guérin (BCG) vaccination and one of the two tests also being significantly more sensitive than the PPD skin test [5]. QuantiFERON®-TB Gold (Cellestis; Carnegie, Australia) is one of the IGRA tests using a whole-blood enzyme-linked immunosorbent assay, which measures the release of interferon gamma (IFN-γ) in response to three allegedly M. tuberculosis complex-specific proteins (M. tuberculosis, M. africanum, M. bovis). The detection and quantification of IFN-γ released by T cells in response to ESAT-6, CFP-10 and TB7.7 (p4) proteins is the basis of this test. These proteins are absent from BCG strain and from most non-tuberculous mycobacteria (NTM) with the exception of M. kansasii, M. szulgai, M. marinum, M. gordonae, M. flavescens, and M. gastri [6, 7]. Ingen et al. [8], however, were unable to demonstrate that genes encoding ESAT-6 and CFP-10 are present in M. flavescens, but revealed it in M. riyadhense. Consequently, a positive result of the IGRA test is a support for diagnosing tuberculous uveitis, but can also indicate diseases caused by a few other mycobacteria (Table 1). Therefore disease caused by some NTMs cannot be excluded by this test. Although NTMs are rare causative agents of ocular infections, NTMs produced sporadic cases of keratitis, scleritis, endophthalmitis, orbital granuloma and chorioretinitis [9–12]. M. riyadhense and M. gastri, however, have not yet been shown to cause ocular infections.

Table 1 Cross-reactivity in IGRA tests
Full size table

It is commonly accepted that the combination of a positive IGRA test and a compatible clinical picture indicates ocular tuberculosis until proven otherwise and specific therapy should be applied [3, 4, 13, 14].

We present a case with resistant post-surgical inflammation presenting cystoid macular oedema (CMO) in one eye as well as bilateral occult, ICGA-detected choroiditis associated with a positive IGRA test where culture was finally positive for M. kansasii and which responded to standard anti-tuberculous therapy.

Case report

A 66 year-old male patient had undergone surgery for a combined cataract and epiretinal membrane (ERM) in his left eye one year prior to being seen in our centre. Vision recovered to 1.0 after the operation; however, 4 months later vision decreased again to 0.6 OS and a CMO was seen on fluorescein angiography (FA) and optical coherence tomography (OCT; Heidelberg Engineering, Inc., Heidelberg, Germany). An intraocular injection of 4 mg of triamcinolone was performed with an improvement of the OCT scans but a stagnation of visual acuity around 0.5 and an increase of intraocular pressure to 32 mmHg.

He was referred to the Centre for Ophthalmic Specialised care (COS), Lausanne, Switzerland, for a resistant post-operative CMO at the end of February 2011. At entry, his treatment for the left eye included Azopt® BID, prednisolone acetate 1 % drops BID, indomethacin 0.1 % drops 6× daily and Diamox® 375 mg/day. The best-corrected visual acuity was 1.25 for far and 1.0 for near OD and 0.5 for far and near OS. Intraocular pressures were 22 and 28 mmHg, respectively, for the left and right eyes. Intraocular inflammation measured by laser flare photometry (LFP) was elevated on the left with a value of 54.4 ph/ms (normal values, 3–6 ph/ms). Surprisingly, his right eye also had an elevated flare with a value of 50.9 ph/ms. A CMO was seen in the left eye and confirmed by OCT (Fig. 1). Visual field testing using Octopus® (Haag-Streit, Bern, Switzerland) showed no significant visual field impairment on either side and microperimetry displayed a mild diminished retinal sensitivity in the left eye with a test score of 364/560 versus 460/560 OD.

Fig. 1
figure 1

OCTs of right and left maculas at presentation (top two pictures) and after four sub-Tenon’s triamcinolone injections OS (two bottom pictures). Extensive cystoid macular oedema seen OS at presentation (top right picture) and quasi identical cystoid macular oedema after four subTenon’s triamcinolone injections (bottom right)

Full size image

As the case was considered a resistant post-operative CMO no uveitis work-up was performed at initial presentation. Four subTenon’s injections of 40 mg of triamcinolone were performed on the left side at 3–4 week intervals. Four months later the situation had not changed substantially on the left with a BCVA of 0.6, intraocular pressure of 14 mmHg, inflammation remaining at 52.1 ph/ms and a CMO quasi identical on OCT to the situation at entry (Fig. 1).

Therefore a complete uveitis work-up was performed with the following findings. FA showed CMO on the left and disc hyperfluorescence OU (Fig. 2).

Fig. 2
figure 2

FA showing cystoid macular oedema in the left eye (bottom pictures)

Full size image

ICGA indicated hypofluorescent dots and fuzziness of choroidal vessels in both eyes (Fig. 3). Among performed laboratory examinations, the QuantiFERON®-TB Gold test was positive suggesting sensitisation to the M. tuberculosis complex. When this result was given to the patient he recalled having been treated for tuberculosis as a youngster.

Fig. 3
figure 3

ICGA showing hypofluorescent dark dots OU and fuzziness of choroidal vessels (OD left figure, OS right figure)

Full size image

Prior to the initiation of a specific treatment, a chest X-ray was performed since the patient had been treated a few months earlier by amoxicillin/clavulanic acid for a right upper lobe pneumonia with nocturnal sweats and a loss of 5 kg. An infiltrate in the right upper lobe was still present. Smears of bronchial aspirate obtained during a flexible bronchoscopy disclosed 1–10 acid-fast bacilli/field but PCR for M. tuberculosis was negative. The patient was then started on a triple anti-tuberculous therapy including isoniazid, rifampicin and ethambutol. Surprisingly, the bronchial aspirate culture came back positive for M. kansasii, which is susceptible to classical anti-tuberculous treatment. The therapy was therefore continued.

Five months later, BCVA was 1.25 OD and increased to 0.9 OS. Intraocular pressure was normal at 14 mmHg OU. Inflammation measured by LFP still amounted to 35.1 ph/ms OD and had improved to 42.8 ph/ms OS. The CMO improved markedly on OCT for the first time (Fig. 4).

Fig. 4
figure 4

OCTs of right and left maculas showing resolution of CMO OS after anti-tuberculous therapy; the left macula still showing an increased thickness (right picture)

Full size image

FA was within normal limits on both sides and there were neither hypofluorescent spots nor fuzziness of choroidal vessels seen on ICGA any more. At no time during nine months of follow-up did the CMO recur and BCVA improved in the left eye to 1.0 with a decrease of flare to 40.2 ph/ms.

Discussion

When intraocular inflammation is not responding to adequate inflammation suppressive therapy (IST), an infectious cause should be suspected. The patient presented here was first treated as a common post-surgical inflammation in his left eye after a combined cataract and ERM operation but was resistant to both one intraocular and repeated sub-Tenon’s injections of triamcinolone. Since a complete uveitis work-up showed occult choroiditis on ICGA as well as a positive IGRA test, the diagnosis of presumed ocular tuberculosis was made. A classical triple anti-tuberculous therapy was initiated following flexible bronchoscopy performed because of a persistent upper right lobe infiltrate.

IGRA tests are very helpful, especially in non-endemic areas, to orient the diagnosis of IST-resistant uveitis cases towards possible ocular tuberculosis in case of IGRA positivity with a compatible clinical presentation. In such a situation the patient should be considered to have ocular tuberculosis until proven otherwise and specific treatment is warranted [3, 4, 13, 14]. However, it is not well known that IGRA tests are in fact not only identifying M. tuberculosis complex but also at least five NTMs including M. kansasii, M. szulgai, M. marinum, M. gordonae and M. flavescens that can cause ocular infections [6, 7]. This is indicated in Table 1 showing significant cross-reactivity between M. tuberculosis complex and environmental strains such as M. kansasii, M. szulgai, M. marinum, M. gordonae, M. flavescens, M. gastri, M. flavescens and M. riyadhense (Table 1).

After having started anti-tuberculous therapy in our patient, a bronchial aspirate culture yielded the NTM M. kansasii. Non-tuberculosis mycobacteria are subdivided into three groups of slow-growing strains (SGMs) where M. kansasii, M. marinum, M. szulgai and M. gordonae are included and one group of rapidly growing strains [15]. Ocular infections are mostly caused by M. abcessus and M. chelonae, a rapidly growing mycobacteria (RGM), which is mostly involved in keratitis reports [16–18] and which does not produce a positive IGRA test [19]. Moreover, tuberculosis-like chorioretinitis has also been reported in association with a non-characterised RGM [9]. Grenzebach et al. [11] have reported a case of granulomatous tuberculous-like endophthalmitis caused by an atypical rapidly growing mycobacterium strain. In a case of scleritis, Tanemoto et al. [12] were able to identify an atypical mycobacterium by PCR in eye discharge and gastric juices while culture and microscopy were negative.

Although very rare, all five SGM NTMs cited above that produce a positive IGRA test have been reported to cause intraocular inflammation. Table 2 shows non-tuberculous mycobacteria which cause different ocular diseases together with their IGRA cross-reactivity characteristics and type of growth. Lai et al. [10] reported a case of systemic dissemination of M. kansasii including tuberculosis-like chorioretinitis and cytomegalovirus retinitis in a patient with acquired immune deficiency syndrome. Other cases of non-ocular infections by M. kansasii in immunosuppressed as well as in immunocompetent patients have been described [20, 21]. One case of keratitis, one case of sclerokeratitis and one case of preseptal cellulitis due to M. marinum [22–24], one case of keratouveitis and two cases of post-LASIK keratitis due to M. szulgai have been reported in the literature [16, 25]. M. gordonae was also shown to cause keratitis and was isolated by culture [26, 27]. Bullington et al. [28] published a case of M. flavescens keratitis and confirmed that cultures and sensitivity tests are mandatory in determining appropriate treatment.

Table 2 Eye infections caused by non-tuberculous mycobacteria
Full size table

Fortunately, there are no negative therapeutic consequences in case of M. kansasii infections misdiagnosed as M. tuberculosis following a positive IGRA test, as this strain is usually sensitive to standard anti-tuberculous therapy [21, 29]; however, the recommended duration of treatment is 18 months [30, 31]. In contrast, if infections due to M. szulgai, M. gordonae, M. flavescens or M. marinum are misdiagnosed as tuberculosis due to a positive IGRA test, this can have deleterious therapeutic consequences as these SGM NTMs are resistant to standard anti-tuberculous therapy [32, 33]. The combination of trimethoprime/sulfametoxazole and doxycycline has been suggested for infections due to M. szulgai and M. marinum [34]. M. gordonae was reported to be sensitive to azithromycin, rifampin and quinolones [35, 36]. In the present case, the clinical response after nine months of therapy has been spectacular with no recurrence so far.

In conclusion, if ocular clinical findings are compatible with a tuberculous uveitis in presence of a positive IGRA test, ocular tuberculosis should be strongly suspected and a therapeutic trial is warranted. As IGRA tests do not react with most of the NTMs, they do react with M. kansasii, M.szulgai, M. gordonae, M. flavescens and M. marinum and an IGRA test cannot exclude an infection by these agents. Although this is a very rare occurrence, it deserves to be known for potential cases resistant to standard anti-tuberculous therapy. In the reported case we strongly suspect that the same organism as the one found in the lung was causing the ocular disease. Fortunately, the consequences would not have been deleterious, unlike for M. szulgai, M. gordonae, M. flavescens or M. marinum, if the organism had not been isolated in the lung as both M. tuberculosis and M. kansasii are sensitive to the same drugs.