The introduction of biological therapies in the management of rheumatoid arthritis (RA) drives the need to re-evaluate existing standard therapy regimens. Although tumour necrosis factor (TNF)α inhibitors (TNFi) have been approved to treat RA successfully, little information is available on the potential glucocorticoid (GC)-sparing effect of TNFi.1 2 3 4 For example, there are no firm data on detailed time courses of the GC dose changes in clinical practice. We therefore performed this study to evaluate the impact of TNFi on GC use in RA.

We performed a database search of patient files in our department between January 1999 and January 2007. Inclusion criteria were RA (American College of Rheumatology (ACR) criteria), first time starting on TNFi (etanercept, 25 mg twice a week or infliximab, 3 mg/kg bodyweight or adalimumab, 40 mg every 2 weeks), on TNFi for at least 3 months and on concomitant GC at TNFi initiation, respectively. The 110 patients thus identified (table 1) were retrospectively investigated over a period of 5 years to quantify the course of disease activity and to assess 28-joint Disease Activity Score (DAS28)5 as well as GC demand.6 TNFi initiation was defined as baseline (T₀).

Further characteristics at baseline were: 87 patients were women, 23 men; mean (SD) age = 55.4 (12.4) years, ranging from 24 to 77 years; mean (SD) RA disease duration = 12.5 (9.7) years, ranging from 1 to 48 years; 89 patients were RF positive, 21 patients were RF negative; 79 patients were cyclic citrullinated peptide (CCP) positive, 31 patients were CCP negative.

Disease activity decreased significantly after TNFi initiation (DAS28: 5.4 (1.2) at baseline vs 3.2 (0.9) after 60 months, p<0.001). Clinical remission (DAS28<2.6) was achieved in 27 patients (24.5%). As a consequence, GC doses could be significantly reduced from 7.5 (5; 12.5) mg/day to 2.5 (0; 5) mg/day (p<0.001; fig 1A). GC doses were reduced in 81 patients and even stopped in 28 patients (fig 1B). In 24 patients, GC intake remained stable (fig 1C); in 5 patients GC dose was increased in response to a continuously high DAS28 score despite TNFi treatment (fig 1D).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Median glucocorticoid doses and mean 28-joint Disease Activity Score (DAS28) results under tumour necrosis factor α (TNFα) inhibitor treatment in 110 patients (A), in 28 patients undergoing tapering of glucocorticoids (GCs) (B), in 24 patients with stable glucocorticoid doses (C) and in 5 patients undergoing an increase of glucocorticoid doses. GC values are given as median (interquartile range), DAS28 values are given as mean (SD) of n patients. Asterisks indicate values significantly different to baseline (*p⩽0.05, **p⩽0.01, ***p⩽0.001; the Wilcoxon signed rank test was applied for not normally distributed variables, Student t test for normally distributed variables; distribution was assessed by the Kolmogorov–Smirnov test; SPSS V. 17.0 (SPSS, Chicago, Illinois, USA)).

Correlation values between DAS28 and GC doses increased at subsequent visits from no correlation at baseline (r = 0.03; p = 0.76) to a low but significant correlation at month 24 (r = 0.38; p<0.01) and at month 60 (r = 0.31; p = 0.04), respectively.

In all, 37 patients discontinued TNFi treatment due to lack/loss of effectiveness (n = 18), toxicity/adverse events (n = 6), pregnancy (n = 1), incompliance (n = 1) and unknown reasons (n = 1). In this group of so-called inadequate TNFi responders, GC was reduced or even stopped in 17 patients before TNFi underwent discontinuation. In 15 patients, GC doses remained stable and 5 patients underwent an increase. A further 22 switched to a second TNFi. After switching, DAS28 significantly decreased (5.2 (1.3) at baseline vs 3.6 (1.4) after 24 months; p = 0.002). Correspondingly, GC doses were significantly reduced (7.5 (5;10) mg/day vs 5.0 (3.5;5) mg/day; p = 0.001; dose reduction: 14 patients, stable doses: 9 patients, no dose increase).

In summary, this is the report of a detailed time course of GC intake under TNFi treatment with a long duration of follow-up. The results of this study highlight the effectiveness of TNFi in allowing a reduction of GC to low doses in the majority of patients based on significant reduction of disease activity. This effect was shown as early as within the first 3 months and persisted up to the end of the observation after 5 years. Patients switching to a second TNFi showed significant response to treatment and, interestingly, a significant GC-sparing effect.7 8 Limitations of our study are that (i) the reduction of GC dosages was based on criteria of daily clinical practice rather than prespecified criteria and that (ii) there was no control group of patients undergoing conventional disease-modifying antirheumatic drug (DMARD) therapy.