Article Text

Download PDFPDF

Original article
The use of positron emission tomography/CT in the diagnosis of tuberculosis-associated uveitis
  1. Deshka Doycheva1,
  2. Christoph Deuter1,
  3. Juergen Hetzel2,
  4. Julia-Stefanie Frick3,
  5. Phillip Aschoff4,
  6. Eva Schuelen2,
  7. Manfred Zierhut1,
  8. Christina Pfannenberg4
  1. 1Centre for Ophthalmology, University of Tuebingen, Tuebingen, Germany
  2. 2Department of Internal Medicine II, University of Tuebingen, Tuebingen, Germany
  3. 3Institute of Medical Microbiology and Hygiene, University of Tuebingen, Tuebingen, Germany
  4. 4Department of Radiology, University of Tuebingen, Tuebingen, Germany
  1. Correspondence to Dr Deshka Doycheva, Centre for Ophthalmology, University of Tuebingen, Schleichstr. 12, 72076 Tuebingen, Germany; deshka.doycheva{at}


Background Detection of tuberculosis as the underlying disease in uveitis can be problematic because of the limited sensitivity of conventional diagnostic methods, especially in the case of latent infection. The aim of our study was to evaluate the clinical usefulness of 18F-fluorodeoxyglucose positron emission tomography/CT (18F-FDG-PET/CT) in patients with uveitis and positive interferon-γ release assay.

Methods We screened 95 patients with different uveitis forms by QuantiFERON TB-Gold test. Positive results were found in 24 cases. 18F-FDG-PET/CT was performed in 20 QuantiFERON-positive patients. PET/CT images were evaluated for the presence, size and metabolic activity of hilar and mediastinal lymph nodes and pulmonary lesions.

Results In nine patients (45%) PET/CT detected increased FDG uptake in mediastinal or hilar lymph nodes. In two patients Mycobacterium tuberculosis was detected in culture after PET/CT-guided lymph node biopsy. In seven patients (35%) with serpiginous choroiditis partly calcified lymph nodes without FDG-uptake were found. Remission of uveitis was achieved in nine of 11 (82%) anti-tuberculosis-treated patients with progressive course of uveitis.

Conclusion In QuantiFERON-positive patients with severe uveitis forms, such as serpiginous choroiditis and occlusive retinal vasculitis, 18F-FDG-PET/CT is useful to identify lesions appropriate for biopsy and helps to establish the diagnosis and appropriate therapy for presumed tuberculosis-induced intraocular inflammation.

  • Uveitis
  • tuberculosis
  • 18F-FDG-PET/CT
  • diagnosis
  • immunology
  • infection
  • imaging
  • diagnostic tests/investigation

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Tuberculosis (TB) is one of the most widely distributed infectious diseases and a leading cause of mortality worldwide.1 2 Intraocular inflammation is a common extrapulmonary manifestation of the disease and can present as granulomatous anterior uveitis, choroiditis, panuveitis or retinal vasculitis. However, because of a lack of specific tests, the diagnosis of TB as an underlying disease in patients with uveitis is difficult. Until recently, the diagnosis of ocular TB was based on the presence of a positive tuberculin skin test and specific chest x-ray findings. The limitations of these methods are well known. As the tuberculin skin test uses a mix of antigens from mycobacteria, false-positive reactions can occur.3 The chest x-ray and CT can show that the pulmonary and lymph nodes are affected, but sensitivity and specificity are limited in cases of latent TB. The interferon-γ release assay (IGRA) is a new test for TB diagnosis that uses more specific antigens than the tuberculin test and can minimise the number of false-positive results, especially in BCG-vaccinated patients.4 Both methods seem to have similar sensitivity, but IGRA has a better sensitivity (more than 92%) than the tuberculin test.5 However, because of the lack of a gold standard in TB diagnosis, the reliability and sensitivity of IGRA is difficult to establish.

Positron emission tomography (PET) combined with CT is a hybrid imaging technique that detects and localises the accumulation of tracers such as the glucose analogue 18F-fluorodeoxyglucose (FDG) in lesions with increased glucose metabolism.6 FDG uptake parallels glucose uptake, but unlike glucose, FDG is trapped in the cells and is not metabolised further. Inflammatory as well as neoplastic cells have increased glucose metabolism, causing FDG accumulation in tumours and inflammatory tissues.7 The fusion of PET and CT images gives detailed functional and morphological information simultaneously.8 Recently, cases of TB detected by 18F-FDG-PET/CT were reported in patients with malignant diseases and immunocompromised subjects, where clinical manifestations were atypical and morphological abnormalities were not detectable with conventional imaging.9 10 PET/CT has been also reported to be a useful method in the diagnosis of ocular sarcoidosis.11–13

The aim of our study was to evaluate the diagnostic role of 18F-FDG-PET/CT in patients with intraocular inflammation who were positive by interferon-γ release assay.

Material and methods

Between January 2007 and June 2009 we screened 95 patients with different forms of uveitis by QuantiFERON-TB Gold test (Cellestis, Germany). The screening was done in patients with clinical suspicion of TB-associated intraocular inflammation, including patients with granulomatous uveitis, panuveitis, retinal vasculitis, chorioretinitis, multifocal choroiditis, serpiginous choroiditis and acute posterior multifocal placoid pigment epitheliopathy. The QuantiFERON-TB Gold test was performed according to the guidelines of the Center for Disease Control and Prevention.14 Positive results were found in 24 cases. The underlying diagnoses of the QuantiFERON-positive patients were retinal vasculitis (six patients), serpiginous choroiditis (12 patients), multifocal choroiditis (two patients), granulomatous anterior uveitis (one patient), posterior uveitis (one patient) and panuveitis (two patients). Nine patients were female and 15 patients male. The mean age of the patients was 51±17 (range 17–76) years.

All QuantiFERON-positive patients underwent extensive diagnostic workup, including chest x-ray (16 patients), chest CT (six patients) and both x-ray and CT (two patients), to search for findings of active or inactive TB such as mediastinal or hilar lymphadenopathy with or without parenchymal lesions. All conventional examinations except in three patients (table 1, patients no. 4, 17 and 19) were performed in various outpatient centres and evaluated by board-certified radiologists. In three patients only enlarged and/or calcified mediastinal or hilar lymph nodes and small pulmonary nodules were found by CT. Since conventional imaging did not reveal any abnormalities in 21 cases, 18F-FDG-PET/CT was scheduled in all QuantiFERON-positive patients. Three patients without pathological changes on chest x-ray and with inactive intraocular inflammation refused the examination. In one patient with enlarged mediastinal lymph nodes on CT, PET/CT was cancelled because of lack of activity of uveitis. Finally, 18F-FDG-PET/CT was performed in 20 QuantiFERON-positive patients.

Table 1

Patient characteristics, results of 18F-fluorodeoxyglucose positron emission tomography/CT (18F-FDG-PET/CT) and microbiological investigations and therapy in 20 QuantiFERON-positive uveitis patients

The protocol for the PET/CT examination was a standard one with 60 min uptake time after administration of 350 MBq 18F-FDG. The examinations were performed using the Hi-Rez Biograph 16 (Siemens Medical Solutions, Knoxville, Tennessee, USA), consisting of a high-resolution three-dimensional lutetium oxy-orthosilicate (LSO) PET and a 16-row multi-detector CT (MDCT). All FDG-PET/CT scans included a contrast-enhanced CT to obtain diagnostic CT data.

PET/CT images of the thorax were carefully evaluated by an experienced nuclear medicine specialist and an experienced CT radiologist for the presence, size and metabolic activity of hilar and mediastinal lymph nodes and pulmonary lesions. The uptake of FDG was evaluated visually as well as semi-quantitatively (maximum standard uptake value (SUVmax)). On PET images, any focal, even faint FDG uptake in a normal-sized or enlarged lymph node exceeding normal regional tracer accumulation was regarded ‘PET-positive’ (= metabolically active lymph node) and indicative for inflammatory changes (including TB), whereas lymph nodes without FDG uptake were classified as ‘PET-negative’ (= metabolically inactive lymph nodes). For CT the presence of pulmonary lesions and mediastinal and/or hilar lymph nodes with or without enlargement and calcifications, even small (<10 mm) in diameter was regarded as ‘CT-positive’ for inflammatory changes (including TB). For semi-quantitative analysis three-dimensional regions of interest (ROIs) were placed over the lymph nodes semi-automatically using a dedicated software program, with a threshold of 50% of the maximum tracer uptake. The SUVmax was calculated from each ROI according to the formula: SUV = measured activity concentration (Bq/ml)/injected activity (Bq) per body volume (cm3). If technically possible, PET/CT-guided transbronchial biopsy of suspicious lymph nodes was performed. PET/CT results were validated by the results of lymph node biopsy and indirectly based on the clinical response of intraocular inflammation to anti-tubercular therapy.


PET/CT detected increased FDG uptake (mean SUVmax 3.3±1.6, range 1.6–9.6) in normal-sized or slightly enlarged mediastinal or hilar lymph nodes in nine QuantiFERON-positive patients (45%). PET/CT images and fundus photographs of one patient with retinal vasculitis (no. 2) and another patient with serpiginous choroiditis (no. 6) are presented in figure 1. Lymph nodes without FDG uptake, partly calcified, were found in seven patients (35%), all suffering from serpiginous choroiditis. Four patients showed additional calcified granulomas in the lung. The PET/CT did not reveal any abnormalities in four QuantiFERON-positive patients (20%). The size of all evaluated lymph nodes, including calcified nodes, varied between 8 and 31 mm (median 14.2 mm). The median size of metabolically active lymph nodes was 17.3 mm in comparison to 10.1 mm in metabolically inactive lymph nodes (p<0.01, r=0.7).

Figure 1

18F-FDG-PET/CT and fundus photographs of patients with retinal vasculitis (no. 2) and serpiginous choroiditis (no. 6). (A, B) PET/CT of a patient with retinal vasculitis, showing a clearly increased FDG uptake (maximum standard uptake value (SUVmax) 9.6) of enlarged right hilar and mediastinal lymph nodes (arrows). (C, D) Fundus photographs of patient no. 2, showing retinal haemorrhages, vessel occlusions and neovascularisation. The patient was treated with anti-tubercular therapy, systemic steroids, intravitreal bevacizumab and laser photocoagulation because of retinal ischaemia and neovascularisation. Remission of uveitis was achieved after the treatment. (E, F) PET/CT of a patient with serpiginous choroiditis, demonstrating a slightly increased FDG-uptake (SUVmax 2.6) in a marginally enlarged subcarinal mediastinal lymph node (arrow). (G, H) Fundus photographs of patient no. 6, showing peripapillar and central chorioretinal atrophic lesions. The patient was treated with anti-tubercular therapy and systemic steroids, and no recurrences of choroiditis were observed.

Bronchoscopy was carried out in ten patients: eight patients with PET-positive lymph nodes and two patients with PET-negative lymph nodes. Bronchoscopy included blind transbronchial needle aspiration biopsy (TBNA) in five patients; in five patients TBNA was technically impossible because of small size and difficult localisation of lymph nodes. Microbiological examination of bronchial lavage for Mycobacterium tuberculosis was negative in all patients. Polymerase chain reaction (PCR) was negative in two patients with retinal vasculitis (no. 2 and no. 3) in whom bronchoscopy with TBNA was performed, but the culture disclosed M. tuberculosis. The microbiological examination was negative in other three patients with TBNA (no. 4, 14 and 16).

An anti-TB treatment was instituted in 11 patients with a progressive and sight-threatening course of intraocular inflammation—three patients with vasculitis (including both patients with culture-proven TB), one patient with multifocal choroiditis and seven patients with serpiginous choroiditis. Six of these patients were PET- and CT-positive, four patients were CT-positive but PET-negative, and one patient was PET- and CT-negative. Two PET-positive patients refused anti-tubercular treatment because of sufficient control of intraocular inflammation, and one patient because of potential side effects. In 10 of 11 anti-TB-treated patients the therapy was carried out with a combination of three anti-tubercular agents (isoniazid, rifampicin and pyrazinamid) and systemic steroids for at least 6 months. One patient (no. 4) was treated with isoniazid and steroids only.

In nine of 11 patients (82%) receiving anti-TB treatment, stabilisation and remission of uveitis was achieved. Recurrences of intraocular inflammation were observed in two anti-TB treated patients only—one patient with retinal vaculitis (no. 4), who received one anti-mycobacterial agent, and one patient with serpiginous choroiditis (no. 5). Worsening inflammatory activity with relapses of uveitis occurred in five of nine patients (56%) who did not use anti-tubercular therapy and were treated with systemic steroids and/or immunosuppressive drugs (no. 8 and 18) only. Comparing the incidence of recurrences in both groups, anti-TB-treated and untreated, reduction in the recurrence rate with achieving remission was observed in 82% of the anti-TB-treated patients and in 47% of the untreated patients.

Tables 1 and 2 summarise the results of radiological and microbiological investigations in all patients.

Table 2

Summary of results of 18F-fluorodeoxyglucose positron emission tomography/CT (18F-FDG-PET/CT) and bronchoscopy of 20 QuantiFERON-positive uveitis patients


Although patients with uveitis undergo extensive diagnostic workup, in many cases the aetiology of intraocular inflammation remains unclear. Detection of TB as an aetiological cause of uveitis can be difficult when there is no direct evidence for M. tuberculosis infection and because of the limited sensitivity of conventional imaging methods, especially in the case of latent and asymptomatic TB infection.15 Therefore, it is not surprising that in the majority of our patients without active TB the previous chest x-ray was normal. Recently, new methods for TB diagnosis have been used in clinical practice. IGRA is a sensitive laboratory method that shows high specificity in the diagnosis of TB.5 However, no studies were performed on the correlation between IGRA and imaging in patients with intraocular inflammation suspicious for TB. To show the benefit of combined anatomical–functional imaging in this setting, we performed 18F-FDG-PET/CT in 20 QuantiFERON-positive patients with different forms of intraocular inflammation, aiming to biopsy suspicious lymph nodes. Interestingly, in 80% of the patients PET/CT detected mediastinal and hilar lymph nodes, metabolically active in 45% of cases and metabolically inactive, partly calcified, in 35% of the cases. The higher incidence of positive findings, detected by PET/CT in comparison to conventional imaging methods is mainly due to the tightened classification criteria for affected lymph nodes in comparison to routine clinical practice. In our cohort, PET/CT findings associated with culture-proven TB included a clearly increased FDG uptake by several moderately enlarged hilar and mediastinal lymph nodes. We found a positive correlation between the lymph node size and FDG uptake with a significantly larger size of metabolically active lymph nodes. However, taking into account all patients with positive PET or CT results, we have to emphasise that the findings we considered as suspicious of inflammatory disease (including TB) were non-specific and often subtle in terms of calcified lymph nodes and normal-sized or slightly enlarged lymph nodes with only faint FDG uptake.

After PET/CT-guided lymph node biopsy M. tuberculosis was isolated in culture in two of our vasculitis patients. The two cases illustrate that the ability of FDG-PET/CT to show the metabolic activity of pathological lesions can increase the rate of successful lymph node biopsies. The microbiological detection of M. tuberculosis in these two patients provides strong evidence for the role of TB as an aetiological cause of uveitis. However, the number of negative biopsies in our group underlines that the aspiration lymph node biopsy for M. tuberculosis is still a ‘blind’ biopsy, in which a limited number of cells are retrieved, with the possibility of missing M. tuberculosis. Additional problems include the lymph nodes being localised near big vessels (aorta or pulmonary artery), and which are thus difficult to biopsy.

We analysed patients who were suggestive for TB-induced uveitis or at least possibly induced by TB, such as serpiginous choroiditis and occlusive retinal vasculitis. The association between serpiginous choroiditis and TB was first described in Indian patients. Gupta et al reported on seven patients with serpiginous-like choroiditis who had a strong positive tuberculin test and positive chest radiography findings.16 In our series PET/CT findings such as metabolically active or enlarged, partly calcified, inactive lymph nodes were found in more than 80% of the QuantiFERON-positive patients with serpiginous choroiditis, suggesting a role for TB in serpiginous choroiditis also in the Western world. In India, patients with serpiginous-like choroiditis improved after anti-tubercular treatment, but in our patients no extraocular clinical activity of TB was observed and the intraocular inflammation was treated with a combination of anti-tubercular drugs and corticosteroids.

Another interesting patient group in our study are the patients with retinal vasculitis. In 18F-FDG-PET/CT we found clearly enlarged metabolically active lymph nodes in all four patients with retinal vasculitis. In two patients (no. 2 and 3) with severe occlusive vasculitis M. tuberculosis was detected in culture after lymph node aspiration biopsy. In a case series of 13 patients with ocular TB, Gupta et al described the most common clinical findings as being retinal vasculitis with vitritis, retinal haemorrhages, neovascularisation, focal choroiditis and neuroretinitis.17 These clinical findings resemble the clinical features of Eales' disease.18 In some studies M. tuberculosis has been detected by PCR in vitreous samples from patients with Eales' disease and it has been suggested that this disorder is induced by TB.19 The clinical findings in our patients with culture-proven TB (no. 2 and 3) are very similar to the clinical picture of Eales' disease.

M. tuberculosis is known to lead to active infection or to a latent stage without clinical signs. However, in cases of TB-associated uveitis, an immune-mediated mechanism has also been suggested. Since some uveitis forms are induced by direct invasion of M. tuberculosis into the ocular tissues, other forms may result from ‘a hypersensitivity reaction’ in the absence of any infectious agent.18 20 21 It is possible that M. tuberculosis in lymph nodes may induce specific anti-TB T-cells, which probably cross-react with ocular structures and cause intraocular inflammation. Therefore, the course of uveitis and the response to therapy seems to be different for infectious and immune-mediated disease. In cases of infectious TB the uveitis could worsen under immunosuppressive treatment, while in immune-mediated forms the intraocular inflammation takes a chronic course, but without exacerbation under immunosuppressive therapy. The difference in the mechanism of the two forms is also reflected in both different imaging appearance and different therapy regimens. While the infectious form needs only strong anti-TB treatment, the immune-mediated disease improves only after a combination therapy with anti-tubercular and anti-inflammatory drugs, for example corticosteroids or even immunosuppressive agents.18 22 We treated our patients with anti-mycobacterial drugs and systemic corticosteroids. During the follow-up a remission of uveitis was observed in 82% of anti-TB-treated patients.

Our current approach to the diagnosis of possible TB-associated uveitis is individualised for each patient. The approach, presented schematically in figure 2, is recommended for regions that are non-endemic for TB. In contrast to other authors who recomend starting anti-TB treatment in each uveitis patient with positive IGRA, our aim is to confirm the TB aetiology before introducing a specific treatment with potential severe side effects. For this purpose we perform CT or PET/CT (if CT-negative) in all QuantiFERON-positive patients with additional bronchoscopic and microbiological investigations. In patients with microbiologically proven TB, anti-TB therapy is initiated directly. Culture-negative uveitis patients with clinical suspicion for TB association receive anti-TB treatment if uveitis is unresponsive to conventional anti-inflammatory therapy.

Figure 2

Current diagnostic approach in patients with possible tuberculosis-associated uveitis. CS, corticosteroids; IGRA, interferon-γ release assay; IOI, intraocular inflammation; IS, immunosuppressive treatment; TB, tuberculosis; PET/CT, positron emission tomography/CT.

We are aware of several limitations of this study, arising from the small number of patients, the lack of standardised criteria for interpretation of PET/CT findings in these patients and the very small number of cases with biopsy-proven diagnoses. The relative utility of PET/CT versus other imaging modalities is only partly addressed in our work and remains unknown. However, this is the first study evaluating the clinical usefulness of 18F-FDG-PET/CT in uveitis patients with positive QuantiFERON test. We found that 18F-FDG-PET/CT is beneficial to identify metabolically active lymph nodes, appropriate for biopsy in patients with presumed TB-induced intraocular inflammation, and thereby helps to establish the diagnosis and appropriate therapy, especially in the case of latent infection in regions with low prevalence of TB.


The authors thank Regina Hoffer (Centre for Ophthalmology, University of Tuebingen) for expert technical assistance.



  • Competing interests None declared.

  • Ethics approval This study was conducted with the approval of the Ethics Committee of University of Tuebingen, Germany.

  • Provenance and peer review Not commissioned; externally peer reviewed.