Abstract
Reactivation of latent Mycobacterium tuberculosis (Mtb) infection is a major complication of anti-tumour necrosis factor (TNF)-alpha treatment, but its mechanism is not fully understood. We evaluated the effect of the TNF antagonists infliximab (Ifx), adalimumab (Ada) and etanercept (Eta) on anti-mycobacterial immune responses in two conditions: with ex vivo studies from patients treated with TNF antagonists and with the in vitro addition of TNF antagonists to cells stimulated with mycobacterial antigens. In both cases, we analysed the response of CD4+ T lymphocytes to purified protein derivative (PPD) and to culture filtrate protein (CFP)-10, an antigen restricted to Mtb. The tests performed were lymphoproliferation and immediate production of interferon (IFN)-gamma. In the 68 patients with inflammatory diseases (rheumatoid arthritis, spondylarthropathy or Crohn's disease), including 31 patients with a previous or latent tuberculosis (TB), 14 weeks of anti-TNF-alpha treatment had no effect on the proliferation of CD4+ T lymphocytes. In contrast, the number of IFN-gamma-releasing CD4+ T lymphocytes decreased for PPD (p < 0.005) and CFP-10 (p < 0.01) in patients with previous TB and for PPD (p < 0.05) in other patients (all vaccinated with Bacille Calmette-Guérin). Treatments with Ifx and with Eta affected IFN-gamma release to a similar extent. In vitro addition of TNF antagonists to CD4+ T lymphocytes stimulated with mycobacterial antigens inhibited their proliferation and their expression of membrane-bound TNF (mTNF). These effects occurred late in cultures, suggesting a direct effect of TNF antagonists on activated mTNF+ CD4+ T lymphocytes, and Ifx and Ada were more efficient than Eta. Therefore, TNF antagonists have a dual action on anti-mycobacterial CD4+ T lymphocytes. Administered in vivo, they decrease the frequency of the subpopulation of memory CD4+ T lymphocytes rapidly releasing IFN-gamma upon challenge with mycobacterial antigens. Added in vitro, they inhibit the activation of CD4+ T lymphocytes by mycobacterial antigens. Such a dual effect may explain the increased incidence of TB in patients treated with TNF antagonists as well as possible differences between TNF antagonists for the incidence and the clinical presentation of TB reactivation.