Biological agents interfering with proinflammatory tumour necrosis factor α (TNFα) signalling have become an established tool in the treatment of rheumatoid arthritis and other chronic inflammatory conditions. Their use is associated with an increased risk for some infections including the reactivation of latent tuberculosis infection (LTBI).1 Consequently, screening for LTBI prior to onset of anti-TNFα therapy is recommended. Interferon γ release assays (IGRAs) are novel diagnostic tools for the diagnosis of LTBI and show a high specificity compared with the tuberculin skin test (TST).2 However, optimal screening procedures are still a matter of debate.1 Furthermore, questions remain with regard to the performance of IGRAs in patients receiving immunosuppressive therapy and the comparative performance of the two commercially available IGRAs.2 To address these questions, 90 successive patients from the outpatient department of rheumatology (representing an LTBI low prevalence patient group) at the University Hospital of Würzburg, Germany were prospectively screened for LTBI using TST, TSPOT.TB (TSPOT; Oxford Immunotec, Abingdon, Oxfordshire, UK) and Quantiferon Gold (QFT; Cellestis Europe, Darmstadt, Germany) before initiation of anti-TNF treatment). Underlying disease entities included ankylosing spondylitis (n=18 patients), rheumatoid arthritis (n=51), psoriatic arthritis (n=10) and other autoimmune diseases (n=11). Nineteen patients were receiving a steroid regimen of ≥7.5 mg prednisone equivalent and 28 were being treated with at least two different disease-modifying antirheumatic drugs (DMARDs). To determine risk factors for LTBI, chest x-rays were obtained from all patients and evaluated by two experienced radiologists unaware of the clinical information for the presence of: (1) calcified hilar or mediastinal lymph nodes; (2) small calcified upper lobe pulmonary nodules; (3) apical pleural thickening; and (4) upper lobe architectural distortion. Information regarding previous contact with cases of active tuberculosis (eight patients) and own history of active tuberculosis (one patient) was obtained by direct questioning. The association of positive results with age, sex and established risk factors for LTBI (contact, positive chest x-ray, history of tuberculosis) was evaluated using multivariate logistic regression with program R (Foundation for Statistical Computing, Vienna, Austria).

The risk factor ‘contact’ was associated with a positive result in both IGRA tests (TSPOT: adjusted OR 13.68 (95% CI 1.72 to 108.74); QFT: adjusted OR 92.77 (95% CI 4.21 to 2044.67)). Stepwise model simplification by removing the non-significant variables age and sex did not alter this association. The association of positive QFT with ‘positive chest x-ray’ was marginally significant, but deletion of non-significant variables (age, sex) rendered it non-significant (data not shown). The association with a history of tuberculosis could not be evaluated (only one positive patient). A combination of all risk factors led to a significant odds ratio for the QFT only (table 1). However, matched pairs logistic regression was not able to confirm the superiority of QFT (OR 0.333 (95% CI 0.0347 to 3.20)) compared with the TSPOT. In contrast to the IGRAs, multivariate regression revealed no association of a positive TST with any of the mentioned variables. However, there was a relatively high agreement of the three tests of 83.1% (Fleiss κ=0.437, p=1.88×10⁻¹⁰). Sixteen out of 90 QFT assays (17.8%) were indeterminate compared with only three of 90 TSPOT assays (3.3%; p<0.01, Fisher exact test; figure 1). These indeterminate QFT assays were associated with a steroid use of ≥7.5 mg prednisone equivalent (OR 3.21, 95% CI 1.11 to 9.23). There was no significant association between the indeterminate QFT assays and the use of DMARDs.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Distribution of latent tuberculosis infection test results in relation to results of the other test systems. Numbers display positive (grey), negative (white) or indeterminate (grey and white) test results. QFT, Quantiferon Gold; TST, tuberculin skin test; TSPOT, TSPOT.TB.

In conclusion, both IGRAs were associated with the risk factor ‘contact’ in our study in accordance with previously published results. As this association was not found for TST, these data are consistent with the proposed higher specificity of IGRAs. None of the IGRAs was found to be superior over the other, but this could be due to the low number of positive results. In contrast, other studies have suggested a higher sensitivity of the TSPOT assay.2 In our patient group we had significant difficulty in interpreting the QFT, resulting in 17.8% indeterminate test results (compared with 3.3% for TSPOT). Previous studies using QFT have reported indeterminate results in different patient groups of 1.9–36%.3,–,9 The indeterminate QFT results in our study were significantly associated with a steroid dose of ≥7.5 mg prednisone equivalent. We assume that this problem could be due to lymphopenia induced by corticosteroids10 or functional impairment of T cells or antigen-presenting cells. Interestingly, previously reported rates of indeterminate results were especially high in patients with HIV (24%) and in those receiving chemotherapy for malignancy (36% vs 21% for TSPOT.TB).3 4 However, a correlation with high-dose steroid treatment has so far not been shown. Finally, our data support the clinical practice of adding IGRAs to the screening procedure of LTBI, yet suggest further evaluation in patients receiving immunosuppressive therapy, especially those on high-dose steroids. It has to be kept in mind that IGRAs need to be used in combination with other screening procedures, especially the patient's detailed history.