Article Text

Concise report
Switching between TNF inhibitors in psoriatic arthritis: data from the NOR-DMARD study
  1. Karen Minde Fagerli1,
  2. Elisabeth Lie1,
  3. Désirée van der Heijde1,2,
  4. Marte Schrumpf Heiberg1,
  5. Synøve Kalstad3,
  6. Erik Rødevand4,
  7. Knut Mikkelsen5,
  8. Åse Stavland Lexberg6,
  9. Tore K Kvien1
  1. 1Deprtment of Rheumatology, Diakonhjemmet Hospital, Oslo, Norway
  2. 2Department of Rheumatology, Leiden University Medical Center, Leiden, The Netherlands
  3. 3Department of Rheumatology, University Hospital of Northern Norway, Norway
  4. 4Depatment of Rheumatology, St Olavs Hospital, Norway
  5. 5Department of Rheumatology, Lillehammer Hospital for Rheumatic Diseases, Lillehammer, Norway
  6. 6Department of Rheumatology, Vestre Viken/Drammen Hospital, Drammen, Norway
  1. Correspondence to Dr Karen Minde Fagerli, Deprtment of Rheumatology, Diakonhjemmet Hospital, P.b. 23, Vinderen, Oslo 0319, Norway; karen.fagerli{at}


Background Tumour necrosis factor inhibitors (TNFi) are efficacious in patients with psoriatic arthritis (PsA), but some patients do not respond or do not tolerate their first TNFi, and are switched to a different TNFi. Evidence supporting this practice is limited, and we wanted to investigate the effectiveness of switching to a second TNFi.

Material and methods From a longitudinal observational study (LOS) we selected patients with PsA who were starting their first TNFi, and identified patients who had switched to a second TNFi (‘switchers’). Three-month responses and 3-year drug-survival were compared between switchers and non-switchers, and within switchers.

Results Switchers (n=95) receiving their second TNFi had significantly poorer responses compared with non-switchers (n=344) (ACR50 response: 22.5% vs 40.0%, DAS28 remission: 28.2% vs 54.1%). There was a trend towards poorer responses to the second TNFi compared with the first TNFi within switchers. Estimated 3-year drug-survival was 36% for the second TNFi compared with 57% for the first TNFi overall.

Conclusions 20–40% of patients had a response on a second TNFi after having failed one TNFi in this LOS. This observation highlights the need for treatments with other mechanisms of action than TNF inhibition in patients with PsA.

  • Anti-TNF
  • Psoriatic Arthritis
  • Treatment

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Tumour necrosis factor inhibitors (TNFi) are recommended as treatment for psoriatic arthritis (PsA),1 with comprehensive evidence that they decrease disease activity and slow radiological progression.2–5 Some patients do not respond to, or do not tolerate, their first TNFi. Many of these patients receive a second TNFi, but there is limited evidence available to support this practice. Recently presented phase 3 data showing efficacy of the interleukin (IL)-12/IL-23 inhibitor Ustekinumab,6 and phase 2 data published on Abatacept,7 in PsA renews the interest in this issue, and further disease modifying therapies may be available in the future.

No randomised controlled trials have addressed the efficacy of switching to a second TNFi in PsA, and although some observational studies have shown good responses, the number of patients in these studies have been small.8–11 One study also examined efficacy of switching previous TNFi-failures from open-label trials to adalimumab.12 In 2009, the British Society of Rheumatology Biologics Register (BSRBR) published data on 178 PsA patients receiving their second TNFi, and found the 1-year drug survival to be only slightly lower than that of those receiving their first TNFi.13

The objective of this study was to assess the efficacy and drug survival of a second TNFi in patients with PsA in an observational study integrated in regular clinical practice.

Material and methods

The data for these analyses are from NOR-DMARD, a longitudinal observational study including patients with inflammatory arthropathies from five Norwegian centres starting a new regimen with disease-modifying antirheumatic drugs (DMARDs). Demographics, disease characteristics, disease activity and functional status measures are collected at baseline. Follow-up is at 3, 6, 12 months and annually thereafter, for as long as they continued treatment. Patients are not followed in the study while they are off DMARD treatment, but reincluded whenever a new treatment is initiated.

We included patients with a clinical diagnosis of PsA starting their first TNFi (February 2001–October 2011). From these patients, we identified those who had been reincluded in NOR-DMARD when starting a second TNFi (‘switchers’). The remaining group is referred to as ‘non-switchers’. A similar approach was used previously when switching of TNFi in AS-patients was assessed in the NOR-DMARD study.14

Baseline characteristics were compared for non-switchers, first and second TNFi of switchers. Disease activity measures described at baseline and 3 months were patient and physician global assessment, Modified Health Assessment Questionnaire (MHAQ) score,15 Short Form 6-dimensions (SF-6D, based on SF-36).16 Number of swollen joints by a 32 joint-count (standard 28 joint-count+ankles and combined MTP-joints), DAS28,17 American College of Rheumatology (ACR) responses (20/50/70),18 EULAR response and DAS28 remission rates (DAS28 <2.6) were applied only to the subgroup of patients with swollen joints at baseline to improve the performance of these outcome measures.

Baseline and 3-month responses were described, and responses of switchers were compared with non-switchers using independent samples t test, Mann–Whitney U test and χ2 test. Comparison of responses to first and second TNFi within switchers was performed using paired samples t test, Wilcoxon signed rank test and McNemars test as appropriate. Data from the last recorded visit prior to switch were also reported. Separate response analyses were performed for patients receiving their second TNFi comparing patients who previously discontinued due to adverse events (AE) to those who had inadequate response (IR). All response analyses were completer analyses. Drug survival was assessed by Kaplain–Meier analysis, and switchers and non-switchers were compared by log rank test.


We included 439 patients (median (IQR) follow-up time: 547 (165–1141)), of whom 95 patients were identified as switchers, and the remaining 344 were ‘non-switchers’. Table 1 shows their baseline characteristics. The only significant differences between groups were lower C-reactive protein (CRP) and swollen joint count in switchers at the time of receiving their 2nd TNFi. Etanercept was the most commonly prescribed first TNFi and adalimumab more commonly prescribed as the second.

Table 1

Baseline characteristics of the different patient groups

Results after 3 months are shown in table 2. Differences in states, changes and response rates were pronounced between non-switchers and the second TNFi of switchers after 3 months, with statistically significant differences in all outcome measures except changes in MHAQ and CRP. When comparing the first and second course of TNFi of switchers, we observed a trend towards poorer responses to the second TNFi for all outcome measures, reaching statistical significance for EULAR response and change in DAS28 (table 2). Switchers also had some improvement with their second TNFi compared with the last observation on their first TNFi. Eight of the non-switchers discontinued treatment prior to 3-month assessment, 10 and 6 patients, respectively, discontinued prior to 3-month assessment of the first/second course among the switchers. No difference was shown in 3-month responses to the second TNFi between patients who previously discontinued due to AE (n=35) and those who discontinued due to IR (n=49) (see online supplementary table S1). Remaining switchers (n=11) had discontinued due to other (n=8) or unknown (n=3) causes.

Table 2

Three-month responses of different patient groups

Three-year drug survival for patients receiving their first TNFi (switchers+non-switchers) and patients receiving their second TNFi is shown in figure 1a, and the drug survival of non-switchers, first and second TNFi of switchers in figure 1b. Drug survival of the second TNFi was significantly poorer than observed in non-switchers (p<0.001). Similar patterns were observed for 1-year drug survival (first TNFi (all): 0.74; non-switchers: 0.83; second TNFi (switchers):0.56). There was no difference in drug survival of the second TNFi of switchers when comparing those who had previously discontinued due to AE/IR (p=0.51, data not shown).

Figure 1

Three-year drug survival. (a) 1st vs 2nd TNFi. (b) Switchers vs non-switchers.


This study investigates the value of switching to a second TNFi in patients with PsA. Switchers had significantly poorer responses to their second TNFi than non-switchers had to their first. We also observed a trend towards poorer responses to the second TNFi compared with the first TNFi in switchers. ACR50 responses at 3 months to the second TNFi in switchers was just over 20%, and less than 30% of these patients were in DAS28 remission at 3 months. Drug survival was somewhat better for the second TNFi than the first in switchers, but still considerably poorer than for the first TNFi overall.

The patients in this study reflect normal clinical practice which gives generalisable results. However, guidelines to start a TNFi are more lenient in Norway than in many other countries. Diagnoses of PsA in NOR-DMARD are based on clinical judgement, which is potentially inaccurate, but the proportion of wrongly diagnosed patients should be small in a population treated with TNFi. As with any observational study, there are several confounding factors which may affect the results. By dividing patients into non-switchers and switchers, we selected patients with better responses in one group, but comparing the response of all patients receiving their first TNFi with that of those receiving their second TNFi would violate the requirement of independent observations for statistical testing. We therefore compared switchers to non-switchers and performed paired analyses in the switchers. The switcher analyses may be influenced by some patients still having residual effects of their first TNFi when starting the second, and we therefore present both changes in disease activity and states, and the pattern of poorer response to the second TNFi was consistent for both analytic approaches. The completer analysis approach is likely to overestimate effectiveness due to discontinuations prior to assessment in poor responders. In this analysis, the group with the poorest response (switchers, second TNFi), indeed had less patients discontinuing prior to the 3-month assessment compared with the first course of TNFi, but differences were small.

We lacked an extensive joint count, which may have caused an underestimation of the number of patients with peripheral arthritis and number of joints swollen, and prevented us from utilising PsA-specific measures. The included composite disease activity score and response criteria were all developed for rheumatoid arthritis, but have been widely used in PsA-studies. The validation of DAS28 in PsA was, however, carried out in a patient population who all had swollen joints at baseline.17 Measures of entheseal, axial and skin disease could potentially have provided additional interesting results, but were not available in NOR-DMARD. We believe that a considerable proportion of patients were treated with TNFi mainly on the basis of axial disease, but did not have specific information about indication for treatment. Another limitation in the data collection was the lack of specification of primary and secondary non-responders, preventing us from further exploring the importance of previous response in predicting response to a second TNFi.

Direct comparison of our results with those of other studies addressing this issue is difficult due to heterogeneity in presented outcome measures. However, overall, our data show poorer drug survival of a second TNFi than previously reported. One-year drug survival rates have been presented from the BSRBR and the Spanish BIOBADASER register. The overall 1-year survival of first TNFi in our population was similar to that presented from the BSRBR (0.74 vs 0.76),13 while BIOBADASER presented better drug survival (0.87).19 The one-year drug survival of the second TNFi was, however, much poorer in our population than the two others (0.56 vs 0.74 (BSRBR) and 0.81 (BIOBADASER)).13 ,19 The BSRBR did, however, only include patients from 2002 to 2006, while during our observation time there were up to four TNFi available for use in PsA, which may have increased the likelihood of switching to a third TNFi. There is also higher baseline disease activity in joint-specific measures and inflammatory markers in the BSRBR patients compared with our patients, which may influence response and drug survival. This discrepancy in disease activity may reflect the differences in British and Norwegian guidelines to start a TNFi, but also possibly a decreasing threshold for starting a TNFi over the decade.20 We did not observe a time trend of decreasing drug survival of the second TNFi, but the statistical power was inadequate to address this issue further in our data.

Even though responses to the second TNFi were surprisingly low compared with the first TNFi, a minority of the patients (20–40%) did show a response at 3 months. Our results, therefore, do not imply that a second TNFi should not be attempted in those who fail their first TNFi, but rather highlight the need for treatments with other mechanisms of action for patients with PsA.


The authors thank the patients for participating in this study and the local rheumatology staff for data collection.


Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Files in this Data Supplement:


  • Handling editor Gerd R Burmester

  • Contributors Study design: TKK, KMF, DvdH, EL. Data acquisition: SK, KM, ER ÅL, MSH. Data analysis: KMF, EL, DvdH. Manuscript preparation: KF, EL, DvdH, MSH, TKKvien. All authors have approved the final manuscript.

  • Funding This work was supported by the Norwegian South-Eastern Health Board through a grant for Karen M Fagerli's salary as a PhD-student. The NOR-DMARD study have received funding in form of unrestricted grants from Abbott, Amgen, Aventis, Bristol Myers Squibb, MSD, Roche, Schering-Plough/Centocor and Wyeth.

  • Competing interests Karen M Fagerli has received speaker's honoraria from Abbott and Pfizer. Elisabeth Lie has received speaker's/consulting honoraria from Roche, Pfizer, BMS and Abbott. Désirée van der Heijde has received consulting fees and/or research grants from Abbott, Amgen, AstraZeneca, BMS, Centocor, Chugai, Eli-Lilly, GSK, Merck, Novartis, Otsuka, Pfizer, Roche, Sanofi-Aventis, Schering-Plough, UCB, and Wyeth. ÅS Lexberg has received speaker's honoraria from Pfizer, MSD, Roche and Abbott. Tore K Kvien has received honoraria as speaker and/or consultant from Abbott, BMS, MSD/Schering-Plough, Pfizer/Wyeth, Roche and UCB.

  • Ethics approval Regional Committee for Medical and Health Research Ethics—South-Eastern Norway.

  • Provenance and peer review Not commissioned; externally peer reviewed.