Background/Purpose

For rheumatoid arthritis it is well established that co-medication (co-med) with methotrexate (MTX) increases the efficacy and drug survival of TNF inhibitors (TNFi), while there has been little evidence of such a beneficial effect of co-med for ankylosing spondylitis (AS). The purpose of this study was to assess the effect of co-med with conventional synthetic disease modifying anti-rheumatic drugs (csDMARDs) on retention to TNFi therapy in AS.

Methods

Data on patients (pts) with AS starting adalimumab (ADA), etanercept (ETN) or infliximab (IFX) as their first TNFi from Jan 2003 through Dec 2010 were extracted from the Swedish Biologics Register ARTIS, including use of csDMARD co-med at baseline. Follow-up data through Dec 2011 were available at the time of analysis. Additional covariate data were obtained by linkage to the National Patient Register and Statistics Sweden. We analysed 5-year drug survival by multivariable Cox regression, including covariates that were selected a priori (Table). Additional potential confounders (ESR, CRP, patient global, 28-swollen joint count (SJC), and history of uveitis, psoriasis and inflammatory bowel disease) were also tested in the model.

Results

1365 pts with AS were included (605 IFX, 406 ADA, 354 ETN). At baseline 557 pts (40.8%) used csDMARD co-med (429 used MTX), and co-med was more often given with IFX (55.4%) than with ADA (28.1%) or ETN (30.5%). Patients on co-med had higher levels of ESR and CRP and more often ≥1 swollen joint at baseline, and remained on therapy longer vs. pts on TNFi monotherapy. Estimated median survival time (based on Kaplan-Meier of overall drug survival) was 6.0 vs. 4.2 years (log rank test p<0.001). By separate analyses per drug (Kaplan-Meier with log rank test), statistically significantly better survival rates in pts on co-med were observed for IFX (p=0.015), ADA (p=0.015) and ETN (p=0.020). The multivariable Cox regression model is shown in the Table. The hazard ratio (HR) for TNFi discontinuation for pts on csDMARD co-med was 0.71 vs. those not on co-med. We also found statistically significant associations between drug survival and ESR, CRP, patient global, 28-SJC and uveitis, respectively, while the association with csDMARD co-med remained highly statistically significant (p<0.001) with a HR of 0.58-0.70 in the models with additional adjustments (data not shown). We also tested csDMARD co-med as MTX vs. other csDMARDs vs. no csDMARD in the model shown in the Table: MTX vs. no csDMARD HR (95% CI) 0.74 (0.61-0.91) (p=0.004), and other csDMARDs vs. no csDMARD 0.59 (0.41-0.83) (p=0.002).

Conclusion

AS patients who received csDMARD co-med with their first TNFi remained on therapy significantly longer than those who were not on co-med. The association remained statistically significant when adjusting for potential confounders.