Article Text
Abstract
Objectives According to guidelines, a nationwide non-medical switch from originator (INX, Remicade) to biosimilar infliximab (Remsima, CT-P13) was conducted in Danish patients with rheumatoid arthritis (RA), psoriatic arthritis (PsA) and axial spondyloarthritis (AxSpA). We investigated disease activity before/after switching and retention rates in the DANBIO registry.
Methods Disease activities 3 months before and after switch and changes over time were calculated. Flare was defined as change in 28 Joint Disease Activity Score (∆DAS28) ≥1.2 (RA/PsA) or Ankylosing Spondylitis Disease Activity Score (∆ASDAS) ≥1.3 (AxSpA). Crude and adjusted retention rates were compared with a historic cohort of INX-treated patients.
Results Eight hundred and two patients switched (403 RA/120 PsA/279 AxSpA; 51% women, age (median (IQR): 55 (44-66)) years). Follow-up was 413 (339–442) days. Prior INX treatment duration was 6.8 (4.3–9.5) years. Disease activities were similar 3 months before/after switch. Crude 1-year CT-P13 retention rate (84.1 (95% CI 81.3 to 86.5)) was similar to the historic IFX cohort (86.2 (95% CI 84.0 to 88.0), p=0.22). The adjusted absolute retention rates were 83.4 (95% CI 80.8 to 86.2) and 86.8% (95% CI 84.8 to 88.8), respectively (p=0.03). In total 132 patients withdrew (lack of effect: 71/132=54%, adverse events: 37/132=28%). Patients with previous INX treatment duration >5 years had longer CT-P13 retention.
Conclusion In 802 arthritis patients treated with INX for median >6 years, a nationwide non-medical switch to CT-P13 had no negative impact on disease activity. Adjusted 1-year CT-P13 retention rate was slightly lower than for INX in a historic cohort.
- Anti-TNF
- Outcomes research
- Ankylosing Spondylitis
- Psoriatic arthritis
- Rheumatoid arthritis
Statistics from Altmetric.com
Background
As patents on the originator biological disease modifying agents (bDMARDs) expire, less expensive biosimilars are marketed – the first being CT-P13, Remsima. Due to their complex biochemical structure, a biosimilar drug is never an exact copy of the originator.1 Before marketing, the equivalence of CT-P13 compared with originator infliximab (Remicade, INX) was demonstrated in pharmacokinetic studies and randomised controlled trials (RCT).1–3 However, it is debated whether the biosimilars perform equally to the originator when INX-treated patients are switched to biosimilar in routine care, as small differences in immunogenicity potentially may influence tolerability and outcomes. Postmarketing observational studies contribute important knowledge regarding biosimilar effectiveness in clinical practice.1
In Denmark, public hospital owners provide bDMARDs via a tax-based system. A national guideline by May 2015 dictated a non-medical switch, that is, all patients treated with INX should switch to CT-P13 for economic reasons.4 The patients (and physicians) had no say in the matter. The potential implications of the switch were economic savings without loss of beneficial treatment outcomes.5 Thus, on marketing in Denmark, the costs of CT-P13 was 36% of that of INX. However, the experience with non-medical switching is limited and stem from open-label studies,6 7 small cohorts8–13 and the NOR-SWITCH trial.14
In the nationwide quality registry, DANBIO, treatment outcomes of Danish adult patients with inflammatory arthritis are monitored prospectively.15 The aims of the present observational study were to investigate the impact of the nationwide switch from INX to CT-P13 on (1) 3 months’ disease activity and flare rates and (2) 1-year retention rates.
Methods
DANBIO covers >95% of adults with rheumatic diseases treated in routine care with bDMARDs. According to national treatment guidelines, disease activity and outcomes are monitored at least biannually and when medication is changed.15 According to Danish legislation, registration and publication of data from clinical registries do not require patient consent or approval by ethics committees.
Patients with inflammatory arthritis in DANBIO who had been followed since start of first bDMARD and who switched from INX to CT-P13 before 1 January 2016 were included (table 1). In addition, 28 patients from two hospitals, where the switch guideline was adapted later, were included. A time gap between planned INX and start of CT-P13 of 0–120 days was allowed to comply with differences in registration practice. All departments of rheumatology were invited to validate data regarding switch date, disease activities and reasons for CT-P13 withdrawal. Data were censored 9 September 2016.
The variables collected in DANBIO have been described previously.15 Predefined time windows were applied for visits on switch (=baseline) and 3 months preswitch/postswitch (see notes to table 2).
Through linkage by social security numbers, comorbidities were identified in national registries (hospitalisations and outpatient care 10 years back) and numbers (0–7) calculated.16
Statistics
Statistical analyses were performed by SPSS 22 and SAS 9.4. Descriptive data are presented by medians (IQR). Non-parametric statistics were used for comparisons. p Values <0.05 were considered statistically significant.
Disease activity 3 months before switch (preswitch), at the time of switch, after 3 months (postswitch) and changes over time (∆preswitch and ∆postswitch) were calculated in each patient. Missing data at the 3 months’ visit were imputed with the 6 months’ visit. For patients who withdrew ≤3 months postswitch (n=18), data from the latest registered visit after baseline were carried forward.
Disease flare was defined as changes in 28 Joint Disease Activity Score (DAS28) ≥0.6 or ∆DAS28 ≥1.2 (rheumatoid arthritis (RA), psoriatic arthritis (PsA)), and ∆Ankylosing Spondylitis Disease Activity Score (ASDAS) ≥1.3 (axial spondyloarthritis (AxSpA)). Remission was defined as DAS28 <2.6 and ASDAS <1.3, respectively.
Treatment retention among switchers was explored with Kaplan-Meier plots and log-rank tests. Univariable and multivariable Cox regression analyses and HR stratified by diagnosis (RA/PsA/AxSpA) were used to identify baseline predictors of CT-P13 retention (gender/age/methotrexate(yes–no)/CT-P13 dose/CT-P13 interval/comorbidities (number) and baseline disease activity (patient’s global score/Bath Ankylosing Spondylitis Disease Activity Index score/ASDAS/DAS28)). Age and gender were forced into the model. Variables with p<0.1 on univariate analysis were included in the multivariate model (stepwise backwards selection).
The 1-year treatment retention was compared with that of a cohort of all patients in DANBIO receiving treatment with INX by 1 January 2014. Cox proportional hazards regression analysis was used to compare the crude 1-year retention rate in the two groups with a robust variance calculation implemented to account for repeated subjects. Multivariable Cox regression analysis with left truncation (1 January 2014) and years since start of INX as timescale was performed to calculate HR for withdrawal adjusted for the following baseline variables — age/gender/diagnosis/methotrexate(yes–no)/comorbidities(number)/patient’s global score — and to calculate adjusted 1-year retention rates for INX and CT-P13.
Results
We included 802 switch patients (table 1). The median (IQR) time gap between planned INX and first CT-P13 was 1 (1–1) day. In 76%, INX was the first biological drug. Follow-up time was 413 (339–442) days during which 16% stopped CT-P13, mostly due to lack of effect (71/132=54%) or adverse events (37/132=28%) (table 1 and online supplementary table 1).
Supplementary Material
Disease activity 3 months preswitch/postswitch was largely unchanged in the majority of patients (table 2) with no clinically meaningful differences observed. Flare rates preswitch/postswitch were similar (table 2).
The characteristics of the comparison and switch cohorts were similar (see online supplementary table 2). One-year crude retention rates (INX: 86.2% (95% CI 84.0 to 88.0) and CT-P13: 84.1% (95% CI 81.3 to 86.5), p=0.22) are shown in figure 1A. The adjusted absolute rates were 86.8% (95% CI 84.8 to 88.8) versus 83.4% (95% CI 80.8 to 86.2) (p=0.03), corresponding to an absolute difference of 3.4%. Correspondingly, CT-P13-treated patients had significantly higher relative risk of withdrawal than the INX cohort (HR 1.31 (1.02–1.68), p=0.03).
CT-P13 retention rate tended to be poorer for RA than for PSA and AxSpA (figure 1B). Duration of INX treatment (<5 years) was associated with poorer retention (figure 1C) as was not being in DAS28 remission at baseline (RA, figure 1D). Higher patient global score at baseline (RA (borderline significant), AxSpA), higher CT-P13 doses (AxSpA) and monotherapy (RA) were associated with poorer retention (table 3).
Discussion
This study of a nationwide non-medical switch in routine care included 802 patients with arthritis previously treated with INX for >6 years. Three-months’ disease activity and flare rates were largely unaffected by the switch. One-year crude retention rate of CT-P13 was not statistically different from that of INX in a comparison cohort, and just reached statistical significance in adjusted analysis.
To our knowledge, this is the first study of large-scale, non-medical switching in routine care with prospective data collection. Similar results of largely unchanged disease activity after the switch have been reported in smaller observational studies of <40 patients with inflammatory arthritis8 9 and inflammatory bowel disease,10 12 17 whereas few have reported negative outcomes.13 In the open-label extension studies of two RCTs (PLANETRA and PLANETAS), INX-treated patients switched to CT-P13 after 1 year (144 and 86 patients, respectively).6 7 These trials indicated similar safety, immunogenicity and efficacy between the two drugs. Preliminary data in a subgroup of patients in our study indicated that infliximab drug levels and presence of antidrug antibodies were unaffected by the switch.18
The NOR-SWITCH RCT, which examined non-medical switch of CT-P13 across indications, included 481 patients (198 with arthritis).14 Across diagnoses, the proportion of patients with flare in NOR-SWITCH was 26.2% for INX versus 29.6% for CT-P13. The higher flare rates might be explained by the different time periods (1 year in NOR-SWITCH and 3 months in the present), inclusion of patients with inflammatory bowel disease and the randomised controlled design of NOR-SWITCH.
Retention rates were slightly lower in the CT-P13 cohort versus the historic INX cohort, with an adjusted absolute risk difference of 3.4%. This difference is not necessarily attributable to CT-P13, but could also represent a ‘nocebo-effect’, that is, negative expectations towards the drug19 or residual confounding.
In the present study ≈84% of patients were still on drug after 1 year, which was lower than in NOR-SWITCH (96%).14 This may reflect differences between ‘real-life’ patients and patients included in an RCT. Differences in study design (patients with arthritis in the current study versus patients from rheumatology/dermatology/gastroenterology in NOR-SWITCH) may also have contributed. The 14% 1-year withdrawal rate in our historic INX cohort illustrates that cessation of treatment also occurs after many years of treatment.20
CT-P13 retention rates across diagnoses were comparable. This is reassuring, as treatment of PsA with CT-P13 had not been investigated before marketing. Patients with RA who were on monotherapy were at increased risk of withdrawal, and there was a tendency towards poorer retention among women and patients with RA with more comorbidities. No new safety signals were detected for CT-P13.
The long average INX treatment duration of >6 years at the time of switching reflects that INX was not first-line bDMARD in Denmark prior to the switch.4 As longer treatment with INX was associated with better CT-P13 retention, extrapolation of our results to other cohorts of shorter treatment duration should be done with caution.
This study of a large cohort of real-life patients contributes important knowledge of postmarketing effectiveness of non-medical switching.1 The availability of historic DANBIO data enabled us to use the patients as their own controls regarding fluctuations in disease activity before and after switch, and to identify a historic INX cohort for comparison of retention rates. Limitations include incomplete data due to the observational study design, for example skin status and 66/68 joint counts in PsA. We applied a time interval of 13 weeks around the baseline visit to reduce missing data at baseline, but in the majority of patients data were available within few days before/after the switch date.
In conclusion, a nationwide non-medical switch from INX to CT-P13 in 802 patients with inflammatory arthritis, who had previously been treated with INX for >6 years, had no apparent negative impact on disease activity. The adjusted retention rate during ≈1 year of follow-up was slightly reduced (3.4%) compared with a historic cohort.
Acknowledgments
Thanks to all the Danish departments of rheumatology for reporting to the DANBIO registry.
References
Footnotes
Contributors BG and MLH contributed to the study design.
BG, MLH, NSK and LP contributed to data analyses and interpretation.
All authors contributed to data collection and contributed to and approved the final manuscript.
Funding The study was funded in part by a research grant from AbbVie.
Competing interests BG: AbbVie; IMJH: Roche; AGL, MLH: AbbVie, BMS, MSD, Pfizer, Roche and UCB; the remaining authors: none declared.
Ethics approval According to Danish legislation, registration and publication of data from clinical registries do not require patient consent or approval by ethics committees.
Provenance and peer review Not commissioned; externally peer reviewed.