Background Treatment with TNF-α inhibitors (TNFi) is associated with an increased incidence rate (IR) of tuberculosis (TB).1 Proper screening can reduce TB IR.2,3 Recent reports suggest comparable TB IR for certolizumab (CZP) and monoclonal TNF-α inhibiting antibodies adalimumab ADA, golimumab (GOL) and infliximab (IFX) (0.47–0.69 vs 0.29–0.50 per 100 person years (PY), respectively).4 Etanercept (ETA) was associated with a TB IR of 0.009–0.08 per 100 PY.5–7 In February 2012 we showed that in our setting with a background annual TB IR of 8.4 per 105 persons and stringent screening for latent tuberculosis infection prior to the initiation of the first TNFi since 2002 a TB IR of 0.16 (95% CI 0.02–0.58) per 100 PY in RA patients treated with TNFi.8
Objectives To compare the TB IR in patients treated for RA with different TNFi.
Methods In February 2015 we cross-linked the data from the mandatory BioRx.si and the national TB registries to identify cases of TB in patients who were ever treated with TNFi to avoid underreporting. TB IR for RA patients were calculated and compared for different TNFi.
Results At the time of data extraction the BioRx.si contained data for 1693 RA, ankylosing spondylitis (AS) and psoriatic arthritis (PsA) patients treated with at least one TNFi for 4574 PY. There were no TB cases among the patients with AS and PsA. In 945 RA patients treated with TNFi for 2453 PY (41% ADA, 9% CZP, 36% ETA, 4% GOL, 10% IFX) 5 cases of TB were identified (Table 1) which translates into an overall IR of 0.20 (95%CI 0.07–0.47) in RA patients. There were no TB cases in RA patients on GOL or IFX, but 3 on CZP, 1 on ADA, and 1 on ETA. The TB IR for CZP was 1.39 (95% CI 0.29–4.02) per 100 PY which is significantly higher than TB IR for ADA at 0.10 (95% CI 0.002–0.55; p=0.002) per 100 PY or ETA at 0.11 (95% CI 0.003–0.63; p=0.005) per 100 PY or all TNFi other than CZP combined at 0.09 (95% CI 0.01–0.32; p=0.006) per 100 PY.
Conclusions In our TNFi treated observational RA cohort, admitting all the shortcomings of the registry data, we observed a very strong signal suggesting a significantly higher TB IR in patients treated with CZP than those treated with any other TNFi.
Gόmez-Reino JJ et al. Arthritis Rheum 2003;48:2085–91.
Carmona L et al. Arthritis Rheum 2005;52:1766–72.
Gόmez-Reino JJ et al. Arthritis Rheum 2007;57:756–61.
Ramiro S et al. Ann Rheum Dis 2010;69:522–8.
Askling J et al. Arthritis Rheum 2005;52:1986–92.
Tubach F et al. Arthritis Rheum 2009;60:1884–94.
Dixon WG et al. Ann Rheum Dis 2010;69:522–8. Tomsic M et al. Ann Rheum Dis 2012;71:1909–11.
Disclosure of Interest None declared