Objective: To compare the safety and retention rate of TNF antagonists used in approved indications (AIs) and non-AIs.
Methods: Analysis of the Spanish registry BIOBADASER 2.0 (February 2000 to October 2009). Patients were classified into AIs and off-label uses (OUs), according to the European Medicines Agency approval. Retention rates, incidence rates (IRs) and IR ratios (IRRs) of adverse events (AEs) with 95% CI were compared between uses, by log-rank test, cause-specific Cox regression models and generalized linear models with Poisson's distribution.
Results: First treatment with TNF antagonist was available in 5150 patients, of whom 4594 (89%) were AIs (2854 RA, 882 AS and 858 PsA) and 556 (11%) were OUs [437 chronic arthropathies in the spectrum of SpAs (CA) and 119 chronic immune-mediated diseases (CIDs)]. The IR of AE was largest in CID (649 events per 1000 patient-years) and lowest in PsA (250 events per 1000 patient-years). The occurrence of AEs was significantly associated with OU [IRR of CA vs RA 1.33 (95% CI 1.19, 1.49); IRR of CID vs RA 1.94 (95% CI 1.62, 2.31). The largest hazard ratio for discontinuation was for CID vs RA (1.33; 95% CI 1.02, 1.71) and especially vs AS (2.18; 95% CI 1.63, 2.90).
Conclusions: OUs of TNF antagonists need a very close ascertainment of risk/benefit. The safety and retention pattern for CID is similar to that for RA and the pattern for CA resembles that of AS. This study shows an additional value of a national registry.