• Anti-TNF
• Ankylosing Spondylitis
• Rheumatoid Arthritis

Multiple sclerosis (MS) is a chronic inflammatory neurological disease with multifactorial aetiology.1 Evidence of tumour necrosis factor (TNF)-α as an important factor in the pathogenesis of MS has emerged.2 However, attempts of treating MS with TNF-inhibitors (TNFi) have increased disease activity.3 Several case reports have indicated that demyelinating diseases could be a serious adverse event following TNFi treatment while data from the Spanish Registry of biological therapies in rheumatic diseases did not bring clarification.4 ,5 Some, but not all, studies have suggested a negative association between rheumatoid arthritis (RA) and MS.6 ,7

We aimed to investigate whether TNFi treatment in patients with arthritis is associated with an increased risk of developing MS and whether RA is associated with a decreased risk compared with the general population.

The nationwide Danish DANBIO Registry started in 2000 and covers >90% of adults with rheumatological disease treated with TNFi in routine care.8 Since 2005, patients not treated with TNFi have also been registered. A total of 30 033 patients with arthritis registered in DANBIO were linked to the Danish Multiple Sclerosis Registry (DMSR), established in 1956.9 All MS cases have been thoroughly evaluated by one of the four neurologists with expertise in MS affiliated with the register. Every MS case is registered and information includes year of first recognised symptom and of final MS diagnosis. A total of 2123 patients had valid information on entry date and 30 patients with an MS diagnosis prior to first registration in DANBIO were subsequently excluded, leaving 27 880 patients with arthritis eligible for analysis. During follow-up 2000–2012, 10 294 patients started TNFi therapy (baseline characteristics, see table 1).

Standardized incidence ratios (SIRs) of MS were calculated. During 113 527 person-years, 12 incident MS cases occurred in the cohort, overall SIR=1.11 (95% CI 0.63–1.96) (table 2). An increased risk was observed in males treated with TNFi (SIR 3.48; 95% CI 1.45–8.37) and in patients with ankylosing spondylitis (AS) (SIR 3.91; 95% CI 1.47–10.42). The SIR for all patients with RA was 0.65 (95% CI 0.24–1.72).

The mean age at MS diagnosis was 47.6 years. Five cases of MS occurred within 1–2 years of TNFi treatment initiation, (SIR 2.46; 95% CI 1.03–5.92) when strictly analysing this period.

Several case reports of demyelinating disease occurring after TNFi treatment have been published.4 ,10 The strengths of this study are the high external validity for routine care, owing to the inclusion of a nationwide, large population of patients with various types of arthritides. There was no loss to follow-up. The completeness of the DMSR has previously been estimated to 90%, and the validity of the definite MS diagnosis based on autopsy cases is 94%.11 Limitations to our study include a possible underestimation of the risk of MS following TNFi treatment as there is a risk that patients who develop symptoms of demyelinisation during TNFi use are not reported to DMSR as they are considered to have an adverse event rather than ‘true’ MS and that TNFi could have been denied to patients with neurological symptoms. In this register-based cohort study, we did not find that RA protects from the development of MS per se. We found no overall increased rate of MS in patients with TNFi-exposed arthritis but an increased risk of MS in male patients with RA and AS treated with TNFi. Furthermore, cases of MS were mainly diagnosed 1–2 years after initiation of TNFi therapy, suggesting an adverse event. However, low statistical power, diagnostic delay of MS and potential uncontrolled confounding should be taken into consideration when interpreting these results.