Are interferon-gamma release assays reliable to detect tuberculosis infection in patients with rheumatoid arthritis treated with Janus kinase inhibitors?

BACKGROUND: Screening for latent tuberculosis infection is recommended in patients with rheumatoid arthritis (RA) starting Janus kinase inhibitors (Jaki). Interferon (IFN)-gamma release assays (IGRAs) are increasingly used for this purpose. Jaki tend to decrease the levels of IFNs, questioning the reliability of IGRAs during treatment with these drugs. OBJECTIVES: To compare the performance of QuantiFERON-TB Gold Plus (QFT-P) and QFT Gold In-tube (QFT-GIT) in RA patients treated with Jaki. METHODS: RA patients underwent QFT-P and QFT-GIT at baseline (T0), and after 3 (T3) and 12 months (T12) of treatment with Jaki. The agreement between the two tests was calculated. The agreement between IGRAs and tuberculin skin test (TST) or chest radiography at baseline was also determined. The variability of QTF-P results was longitudinally assessed. RESULTS: Twenty-nine RA patients (F/M 23/6; median age/IQR 63/15.5 years; median disease duration/IQR 174/216 months) were enrolled. A perfect agreement was found between QFT-P and QFT-GIT at all times (κ = 1). At T0, no agreement was recorded between IGRAs and TST (κ = -0.08) and between TST and chest radiography (κ = -0.07), a low agreement was found between QFT-P and chest radiography (κ = 0.17). A variation of 33.3% in the results of QFT-P was recorded at T3 vs T0, of 29.4% at T12 vs T0, and of 11.8% at T12 vs T3. The median levels of IFN-γ produced by lymphocytes in response to the mitogen of QFT-P decreased after 3 months followed by an increase after 12 months (not significant). No change in the median number of circulating lymphocytes was documented. Glucocorticoids intake was associated with a higher probability of negative or indeterminate IGRA results at T0 (p<0.0001). CONCLUSION: A response to IGRAs is detectable during treatment with Jaki. However, fluctuations in the results of IGRAs have been observed in the absence of correlation with clinical outcomes, thus challenging their interpretation.