Objectives In 2018, a nationwide mandatory switch from originator to biosimilar adalimumab was conducted in Denmark. The available biosimilar was GP2017 (Hyrimoz) in Eastern regions and SB5 (Imraldi) in Western regions. We aimed to assess the comparative effectiveness of GP2017 versus SB5 in patients with rheumatoid arthritis (RA)/psoriatic arthritis (PsA)/axial spondyloarthritis (AxSpA).
Methods Observational cohort study based on the DANBIO registry with geographical cluster pseudo-randomisation, analysed by emulating a randomised clinical trial. Main outcome was adjusted 1-year treatment retention (Cox regression). Furthermore, 6 months’ remission rates (logistic regression), reasons for withdrawal and back-switching to originator were investigated (overall and stratified by indication).
Results Overall, of 1570 eligible patients, 1318 switched and were included (467 RA/321 PsA/530 AxSpA); 623 (47%) switched to GP2017, 695 (53%) to SB5. Baseline characteristics of the two clusters were largely similar, but some differences in registration practice were observed. The combined 1-year retention rate for the two biosimilars was 89.5%. Compared with SB5, estimated risk of withdrawal for GP2017 was lower (HR 0.60; 95% CI 0.42 to 0.86) and 6 months’ remission rate was higher (OR 1.72; 95% CI 1.25 to 2.37). Stratified analyses gave similar results (statistically significant for RA). During 1 year, 8.5% and 12.9% withdrew GP2017 and SB5, respectively (primarily lack of effect and adverse events), of whom 48 patients (3.6%) back-switched.
Conclusion This head-to-head comparison of GP2017 versus SB5 following a mandatory switch from the originator indicated differences in effectiveness in routine care. This may reflect a true difference, but other explanations, for example, differences in excipients, differences between clusters and residual confounding cannot be ruled out.
- biosimilar pharmaceuticals
Data availability statement
All data relevant to the study are included in the article or uploaded as supplementary information.
Statistics from Altmetric.com
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.
What is already known about this subject?
Biosimilars are increasingly used in routine care of patients with inflammatory rheumatic disease. Currently, no head-to-head comparison of the different adalimumab biosimilars has been conducted.
What does this study add?
Based on geographical cluster pseudo-randomisation, we explored outcomes in two biosimilars following a nationwide mandatory non-medical switch from originator adalimumab.
In 1318 included switch patients, the estimated risk of withdrawal was lower and 6 months remission rate was higher for GP2017 compared with SB5. This may reflect a true difference, however other factors (differences in excipients, differences between clusters, residual confounding) cannot be ruled out.
The 1-year treatment retention and disease remission rates for both biosimilars were high.
How might this impact on clinical practice or future developments?
Switch outcomes following a non-medical mandatory switch in routine care for both adalimumab biosimilars were considered effective and safe.
Over the past decades, biological disease-modifying antirheumatic drugs (bDMARDs) have revolutionised the treatment of inflammatory arthritis, including rheumatoid arthritis (RA), psoriatic arthritis (PsA) and axial spondyloarthritis (AxSpA).1 However, due to their high costs, bDMARDs pose a significant financial burden on healthcare systems around the world. The development of less expensive biosimilar drugs potentially improves accessibility of bDMARDs for patients.1–3
For adalimumab, several biosimilars have been approved and marketed in Europe based on results from randomised clinical trials (RCTs) comparing the effectiveness and safety of the biosimilar versus originator adalimumab.4–10 Effectiveness of adalimumab biosimilars might potentially vary due to minor differences in molecular structures, excipients or injection devices.11–13 RCTs are considered the gold standard for assessing efficacy and safety of medications,14 but head-to-head comparison of biosimilars has not been conducted. Well-designed observational studies based on real-world patient data monitored in quality registries constitute a valuable alternative, especially when analysed in a way that emulates an RCT.15 16
In November 2018, Danish National guidelines dictated a mandatory switch from originator to biosimilar adalimumab for economic reasons (a so-called non-medical switch).17 In Denmark, bDMARD treatment is funded by taxpayers and provided free of charge to patients. For drugs considered to be equally effective, a tendering process takes place, after which it is mandatory to prescribe the less expensive drug.18 The expected cost reduction was 34%–49% compared with the originator (personal communication); corresponding to approximately €50 million annually in Denmark.19 The available biosimilar was SB5 (Imraldi) in the Western regions of Denmark (Region North, Middle and South) and GP2017 (Hyrimoz) in the Eastern regions (the Capital Region and Region Zealand). Thus, the guideline provided a surrogate cluster pseudo-randomisation tool where geography rather than patient-related factors determined the choice of biosimilar.20 21 The ‘ideal hypothetical trial’ that we attempted to emulate was pragmatic, in which eligible participants were patients with RA, PsA or AxSpA treated with originator adalimumab, who switched to either GP2017 or SB5 according to the national guideline.
The aim of this study was to assess the comparative effectiveness of GP2017 versus SB5 following a mandatory switch regarding 1-year treatment retention rates, 6-month disease remission rates and reasons for withdrawal. In addition, to investigate changes in disease activity 6 months prior to and after the switch, and the frequency and reasons for back-switching to originator adalimumab.
This was a population-based, observational cohort study with surrogate cluster (ie, geographical) pseudo-randomisation. Eligible patients were identified in the Danish DANBIO quality registry. DANBIO, established in year 2000, has prospective follow-up of >95% of adults with rheumatic disease treated with bDMARDs in routine care22 and validity is high.23 By use of the unique Danish civil registration number that each Dane receives at birth, DANBIO data were enriched with patient-level information regarding comorbidities from The Danish National Patient Registry (DNPR), which has nearly complete data on inpatient and outpatient contacts,24 and with information regarding vital status from The Danish Civil Registry.25 We report our findings in accordance with the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) statement.26
Patients with a clinical diagnosis of RA, PsA or AxSpA, who were treated with originator adalimumab by 1 November 2018 were identified in DANBIO (figure 1).
The following patient groups were identified: (1) Switchers: patients, who switched from originator adalimumab to biosimilar (date of switch=baseline) between 1 November 2018 and 30 April 2019 (inclusion period). According to geographical cluster pseudo-randomisation, patients were allocated to treatment with GP2017 or SB5. (2) Non-switchers: patients treated with originator adalimumab by 1 November 2018 (=baseline), who did not switch to biosimilar during the inclusion period. (3) Back-switchers: switchers, who during 1 year’s follow-up discontinued the biosimilar and resumed treatment with originator adalimumab.
Comparative effectiveness was assessed in patients, who switched to GP2017 versus those who switched to SB5: The adjusted comparative 1-year treatment withdrawal was the primary outcome. Key secondary outcomes were (a) the adjusted comparative 1-year treatment withdrawal stratified by indication (RA, PsA and AxSpA) and (b) disease remission at 6 months (adjusted), overall and stratified by indication. Other secondary outcomes were reasons for withdrawal across treatments.
In addition, the following outcomes were assessed: (a) changes in disease activity 6 months pre-switch versus 6 months post-switch and (b) frequency of back-switch to originator adalimumab, including patient characteristics, reasons, and time to back-switch.
Clinical characteristics, disease activity and DMARD treatment history were retrieved from DANBIO (data extracted by 23 October 2020) and included the following covariates: age (years), gender (female/male), disease duration (years), body mass index, current smoker (yes/no), number of previous bDMARDs, concomitant methotrexate (MTX) (yes/no), treatment duration for originator adalimumab (years), C-reactive protein (CRP; mg/L), physician global assessment (Visual Analogue Scale (VAS), 0–100), patient reported outcomes: pain (VAS 0–100), fatigue (VAS 0–100), global assessment (VAS 0–100), functional status (Health Assessment Questionnaire (HAQ); 0–3) and Patient Acceptable Symptom State (PASS; (yes/no)).
Disease activity was assessed by, for RA and PsA: Disease Activity Score-28 (DAS28)-CRP, Clinical Disease Activity Index (CDAI); for AxSpA: Bath Ankylosing Spondylitis Indices of Disease Activity (BASDAI; 0–10) and BAS functional status (BASFI; 0–10), Ankylosing Spondylitis Disease Activity Score (ASDAS). Disease remission (yes/no) was defined as DAS28 <2.6 (RA and PsA) and ASDAS <1.3 (AxSpA).
Previous comorbidities at baseline were identified in DNPR (International Classification of Diseases, tenth revision codes, online supplemental table S1).
Switchers were followed up for 1 year. Treatment duration was calculated as the number of days that each patient maintained treatment with GP2017 or SB5, or until withdrawal (=first missed dose irrespective of reason), death or lost to follow-up, whichever came first. Treatment pauses less than 90 days (eg, due to infections or surgery) were disregarded. For treatments with no stop date, data were censored 1 year after the last visit in DANBIO. Furthermore, reasons for withdrawal (according to predefined categories in DANBIO22) were retrieved.
For back-switchers, reasons for back-switch and time to back-switch were retrieved from DANBIO. Further, disease activity at the time of biosimilar start and at the time of back-switching was compared, and changes (=delta values) were calculated in each patient.
The study was approved by the Danish Data Protection Agency (RH-2015–209, 04145). In Denmark, registry studies require neither patient consent nor ethical approval.
All statistical analyses were done using R V.3.6.1. P values <0.05 were considered statistically significant. All analyses were conducted using an intention-to-treat approach according to a predefined statistical analysis plan. Clinical characteristics and disease activity are presented as medians (IQR) and frequencies (percentages) for continuous and categorical variables, respectively. No formal power analysis was done as all available patients in DANBIO were included in the study.
Primary and key secondary outcomes
Comparative treatment withdrawal in GP2017 versus SB5 switchers was explored with univariable and multivariable Cox proportional hazards regression analyses (overall and stratified by indication). Analyses were performed as crude, age-adjusted and gender-adjusted and fully adjusted. Adjustment was according to the following a priori defined baseline variables: age, gender, current smoker, indication, disease remission, PASS, number of previous bDMARDs, treatment duration for originator adalimumab and number of comorbidities. In stratified analysis including patients with RA, concomitant MTX was included.
Disease remission at 6 months in GP2017 versus SB5 switchers was estimated with univariable and multivariable logistic regression analyses (OR with 95% CI). Analyses were adjusted as mentioned above.
For multivariable analyses, missing baseline variables were imputed by multiple imputation using chained equations (30 imputations).27 Proportional hazards assumption for Cox regression models was assessed in stacked imputed data sets. A rule of five events per variable was applied to avoid overfitting.
Primary and key secondary outcomes, sensitivity analyses
To mimic common inclusion and exclusion criteria in RCTs, we performed the following sensitivity analyses: S1: restricted to patients with no more than one prior bDMARD treatment in addition to originator adalimumab, and excluding patients with age >75 years, prior malignancy, >2 comorbidities and patients with RA not receiving concomitant MTX; S2: restricted to patients, who scored PASS=yes at baseline. Furthermore, Cox regression analyses of 1-year treatment withdrawal were performed, where only discontinuation due to lack of effect (LOE) or adverse events (AE) was considered an event (S3). Finally, a sensitivity analysis (S4) was performed replacing DAS28 remission with CDAI remission (CDAI <2.9) for patients with RA and PsA.
Of 1570 patients treated with originator adalimumab by 1 November 2018, 1318 (84%) switched to GP2017 (623, 47%) or SB5 (695, 53%) before 30 April 2019 and were included (table 1, figure 1). All patients were assigned the biosimilar treatment in compliance with the guideline, that is, according to their geographical region. In general, the patients had established disease with median 8 years of originator adalimumab treatment. At baseline, the majority (61%) of the patients were in remission. A total of 173 patients (11%) did not switch and were excluded (baseline characteristics in online supplemental table S3).
Baseline characteristics of switchers stratified by indication
Compared with patients who switched to SB5, those who switched to GP2017 had similar age, gender (mainly RA and AxSpA), disease duration, concomitant use of MTX and comorbid burden (table 1), but appeared to have higher disease activity and patient reported outcomes (mainly RA and PsA), more prior bDMARD treatments and more were current smokers. The number of follow-up visits during the first year was similar for both groups (table 1, online supplemental table S2).
Comparative 1-year treatment withdrawal
For all switchers, the combined 1-year crude treatment retention rate was 89.5% (Kaplan-Meier estimation). For GP2017, the overall estimated risk of withdrawal was lower than for SB5 (HR 0.60; 95% CI 0.42 to 0.86) (multivariable Cox regression analysis, fully adjusted, table 2, figure 2). Stratified by indication, similar results were found for patients with RA (HR 0.50; 95% CI 0.28 to 0.90), whereas for PsA and AxSpA the results did not reach statistical significance: HR 0.97 (95% CI 0.46 to 2.02) and HR 0.75 (95% CI 0.42 to 1.33), respectively (age and gender adjusted, table 2).
Remission rates at 6 months
At 6 months, 56% of switchers were in DAS28/ASDAS remission. The odds of being in remission were higher for GP2017 compared with SB5 (OR 1.72; 95% CI 1.25 to 2.37). Similar results were found when stratified by indication (logistic regression analysis, fully adjusted, table 3), but only statistically significant for patients with RA (OR 1.81; 95% CI 1.07 to 3.06).
Reasons for withdrawal
During 1-year follow-up, 53 patients (8.5%) withdrew from treatment with GP2017 and 90 (12.9%) from SB5, respectively, mainly due to LOE and AEs (table 4, online supplemental table S4). Among SB5 switchers, more patients withdrew due to remission, loss to follow-up and other reasons, including change of hospital.
Compared with the overall switch population, patients who withdrew biosimilar treatment were more often women, had received fewer prior bDMARD treatments and tended to have higher disease activity at baseline both regarding subjective and objective measures (online supplemental table S4).
In sensitivity analysis S1 applying strict inclusion/exclusion criteria, the estimated risk of withdrawal for GP2017 switchers (with SB5 as the reference) showed results similar to the main analysis (HR 0.59; 95% CI 0.37 to 0.92). The estimates remained similar across all sensitivity analyses with HR ranging from 0.59 to 0.64 (fully adjusted, table 2). Stratified by indication, age and gender adjusted estimates had wide CIs, which included 1.
Sensitivity analyses for remission at 6 months demonstrated estimates largely similar to the main analyses (table 3). When replacing DAS28 remission with CDAI remission (sensitivity analysis S4), the estimates were lower. Baseline characteristics for patients included in the sensitivity analyses are shown in online supplemental tables S5 and S6.
Disease activity 6 months pre-switch and post-switch
For both GP2017 and SB5, changes in disease activity 6 months pre-switch and post-switch in individual patients were close to zero for all measures (eg, DAS28, CDAI, VAS scores) (table 5).
Back-switchers to originator adalimumab
Among patients, who withdrew GP2017 or SB5 during follow-up, 48 switched back to originator adalimumab (=back-switchers), 52 commenced treatment with another bDMARD, 1 died, 5 were lost to follow-up and 37 did not restart bDMARDs. Treatment duration before back-switching to originator was median 174 days (IQR 101–250) (online supplemental table S7). Back-switch was mainly due to LOE (54%) and AE (27%). Back-switchers were mostly women, had RA and >50% had ≥3 prior bDMARD treatments (online supplemental table S7).
At the time of back-switch, fewer patients were in remission (58% vs 33%) or scored PASS=yes (82% vs 32%), respectively, compared with when they switched, whereas changes in CRP and HAQ were close to zero (online supplemental table S7).
In this nationwide study, we investigated 1-year comparative effectiveness of two adalimumab biosimilars (GP2017 vs SB5) in >1300 real-world patients with arthritis, who were mandated to switch from originator adalimumab. We emulated an RCT comparing these two active treatments using geographical residence (GP2017 in East and SB5 in West Denmark) as the surrogate randomisation tool based on national treatment guidelines.
Knowledge regarding non-medical switching mainly stems from phase III extension of RCTs where patients, who initially received the originator drug were randomised to either maintain treatment or switch to the biosimilar drug.4 5 7 8 10 These studies included patients who are highly selected and often bDMARD naïve, and did not study patients with PsA or AxSpA.4 5 7 8 10 Since no RCT with head-to-head comparison of biosimilars has been performed, post-marketing registry studies of real-life patients may help fill the knowledge gap and provide important evidence for clinical decision-making. The few minor observational studies currently available, mainly conducted in patients with inflammatory bowel disease, have shown stable disease activity following non-medical switch to biosimilar adalimumab.28 29
Strengths of the current study include the nationwide prospective follow-up, high data completeness and high compliance to the guideline with only 11% non-switchers. The cluster pseudo-randomisation was successful, since all patients were assigned biosimilar treatment according to their geographical region, and the number of patients in the two treatment groups was comparable.
We found lower risk of withdrawal and higher remission rates for GP2017 compared with SB5. The difference, which was most consistent in patients with RA, was observed also after adjustment for a large number of baseline variables, and in both the main analyses and in the sensitivity analyses. Due to the observational nature of the study, the difference can not necessarily be attributed to the drug per se, and potential confounders and biases need to be addressed.
First, differences in devices, injection site reactions or differences in buffers are examples of factors, which may have affected drug retention and patient satisfaction. It appeared that itching and burning at the injection site were more often reported in SB5-treated patients, which could be speculated to arise from differences in buffers (citrate or phosphate) (personal communication).30 Unfortunately, we do not have data to further explore this.
Second, geographical clusters (in this study the Western and Eastern regions of Denmark) are never exactly alike, which is a known challenge in the interpretation of cluster-randomised studies.31 This raises the question whether the differences in effectiveness that we found between GP2017 and SB5 relate to the individual patients or to the geographical region. More patients in the Western regions (ie, treated with SB5) were lost to follow-up or withdrew due to other reasons than LOE and AE, including change of hospital during follow-up. Although the number of follow-up visits were similar for Eastern and Western regions, follow-up bias or registration bias may have been present.
Thirdly, confounding by indication was reduced, since the choice of biosimilar was based on geographical region, not on patient-related factors. Although baseline patient characteristics were well balanced in our study, some noteworthy differences were found. GP2017 switchers appeared to have higher disease activity, more frequently to be smokers and have failed more prior bDMARD treatments, which we adjusted for in the multivariable analyses. However, we cannot exclude residual confounding, for example, socioeconomic factors or other patient characteristics that were not available.32
Finally, although this study was a mandated nationwide switch, regional differences in the transition and communication strategy with patients cannot be excluded. The information was mainly given by nurses in the outpatient clinics, and no specific patient-material was developed. Patients included in the study had been treated with originator adalimumab for median 8 years. They were generally satisfied with the originator as illustrated by high remission rate and high patient satisfaction (PASS). Neither patients, nor physicians were blinded for the switch, and patient-related and/or physician-related factors may have affected outcomes. Thus, negative expectations towards the biosimilar (the so-called nocebo effect) and/or incorrect causal attributions, may potentially have affected patients’ experience and consequently outcomes.30 33 34
The combined 1-year retention rate for the two biosimilars was high with 9 in 10 patients maintaining treatment, and the remission rate at 6 months remained largely unchanged. In individual patients, changes in disease activity pre-switch and post-switch were negligible for both GP2017 and SB5. The main reasons for withdrawal were LOE and AE, which is in agreement with what has been reported for the originator drug.4 5 7 8 10
The use of biosimilars including switch procedures in routine care varies substantially between countries.35–37 In Denmark, bDMARDs are provided free-of-cost by public hospital owners via a tax-based system, and mandated switch procedures are implemented after marketing of less expensive biosimilars.38 39 Previous experiences with non-medical switching in routine care have been positive34 40 41 in rheumatological patients treated with infliximab and etanercept. Thus, we have previously reported that 78%–84% of patients maintained treatment 1 year after switching, similar to that of a historic cohort,38 39 which is of magnitude similar to that of the current study. Results from a recent Danish study have shown that Denmark has benefitted from substantial price reductions after patent expiration of the originator adalimumab.18 Only few patients (4%) back-switched to the originator adalimumab during follow-up. At the time of back-switch, patients had higher patient-reported outcomes, whereas the more objective markers remained unchanged (swollen joint count, CRP), suggesting that subjective health experiences influenced the perception of treatment outcomes and AEs, in line with results from our previous study on etanercept switchers.38
In conclusion, we emulated an RCT by using a surrogate randomisation tool (geographical region) in this observational study of >1300 patients, who conducted a mandatory switch from originator adalimumab to one of two biosimilars according to a national guideline. This allowed us to directly compare two adalimumab biosimilars, GP2017 and SB5. Our results indicated a difference between the two biosimilars. This may be a true difference between the active drugs, but other explanations, for example, differences in excipients, differences between clusters and residual confounding cannot be ruled out.
Data availability statement
All data relevant to the study are included in the article or uploaded as supplementary information.
Patient consent for publication
We are grateful to all Danish departments of rheumatology for reporting to the DANBIO registry.
Handling editor Josef S Smolen
MLH and BG shared last authorship and contributed equally to the paper.
Correction notice This article has been corrected since it published Online First. The author, Dorte Vendelbo Jensen, has been added.
Contributors Study conception and design: HN, SG, BG and MLH. Acquisition of data: HN, SG, BG, MLH, FM and NSK. Statistical analysis: HN, SG, BG and MLH. All authors contributed to the interpretation of the data. HN, SG, BG and MLH wrote the manuscript. All authors critically revised the manuscript. All authors revised and approved the final manuscript to be published.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests AGL: AbbVie, Eli Lilly Denmark A/S, Janssen-Cilag A/S, MSD, Novartis, Pfizer, UCB teaching or consultancy fees. OH: AbbVie, Pfizer, Novartis. MLH: AbbVie, Biogen, BMS, Celtrion, Eli Lilly Denmark A/S, Janssen Biologics BV, Lundbeck Fonden, MSD, Pfizer, Roche, Samsung Bioepis, Sandoz. Furthermore, chair of the steering committee of the Danish Rheumatology Quality Registry (DANBIO), which receives public funding from the hospital owners and funding from pharmaceutical companies. Co-chair EuroSpA, which generates real-world evidence of treatment of psoriatic arthritis and axial spondyloarthritis based on secondary data and is partly funded by Novartis. BG: BMS, Pfizer, Sandoz (research grants).
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.