Abstract
Objective To study changes in retention of first biologic disease-modifying antirheumatic drug (DMARD) therapy over a period of 15 years in an inception cohort of patients with rheumatoid arthritis (RA), psoriatic arthritis (PsA), and ankylosing spondylitis (AS).
Methods We assessed patient and disease characteristics and drug survival of patients starting a biologic (tumor necrosis factor inhibitor [TNFi]) therapy between 2004 and 2019 in routine care at the Amsterdam Rheumatology and Immunology Center, Reade, the Netherlands. Starts were classified as early (2004-2008), intermediate (2009-2013), and recent (2014-2018). Kaplan-Meier plots and log-rank tests assessed the overall difference in drug survival between the 3 observation groups and between diagnoses, followed by Cox regression to estimate hazard ratios (HRs).
Results We included 1938 consecutive patients starting TNFi therapy, 63% with RA, 19% with PsA, and 19% with AS; 65% were female. Drug survival decreased significantly over time (overall P < 0.001), mostly caused by decreases in the most recent 4-year period. The HR for drug continuation was 2.04 (95% CI 1.71-2.43, P < 0.001) for the early vs the recent group and 1.92 (95% CI 1.58-2.35, P < 0.001) for the intermediate vs the recent group. Drug survival time was significantly different between diseases (overall P < 0.001), mostly caused by shorter survival in RA. The HR for drug continuation was 0.58 (95% CI 0.47-0.73, P < 0.001) for RA vs PsA and 0.63 (95% CI 0.51-0.78, P < 0.001) for RA vs AS.
Conclusion Patients with RA, PsA, and AS currently initiating biologic (TNFi) therapy discontinue the drug much sooner than those starting shortly after the drugs were introduced. This is most likely because of the availability of alternative novel biologic and targeted synthetic DMARD treatments and treat-to-target protocols enabling and necessitating earlier switching.
- ankylosing spondylitis
- biological therapy
- psoriatic arthritis
- rheumatoid arthritis
- tumor necrosis factor inhibitors
The introduction of tumor necrosis factor inhibitors (TNFi) for the treatment of patients with rheumatoid arthritis (RA), psoriatic arthritis (PsA), and ankylosing spondylitis (AS) has improved clinical outcomes over the last 15 to 20 years. However, a substantial proportion of patients discontinue TNFi treatment,1 making treatment retention an important aspect to study. Drug survival is a marker for the effect of treatment in routine care.2 Since the introduction of TNFi treatment, patient characteristics, treatment options, targets, and strategies have changed in major ways.3 Patients who started TNFi in recent years had less active disease on initiation, were treated much earlier and more aggressively, and had more treatment options than patients treated 15 years ago.4,5 In addition, targeting states of minimal disease activity and remission has become more common, as well as tapering strategies. Importantly, the above changes may not apply equally to the diseases most frequently treated with these agents (ie, RA, PsA, and AS).
In our open cohort of patients initiating biologic therapy, we examined changes in drug survival in patients with RA, PsA, and AS initiating TNFi as their first biologic over a period of 15 years.
METHODS
Consecutive patients with RA, PsA, or AS starting their first TNFi treatment were followed in a large prospective observational cohort of patients spanning up to 15 years of biologic use at the Amsterdam Rheumatology and Immunology Center (location Reade) in Amsterdam, the Netherlands (NTR68686).6 The patients started the TNFi between February 6, 2004, and December 28, 2018. Only data for the first TNFi were used for this study. In January of 2019, all patients on etanercept (ETN) and infliximab (IFX) were switched to a biosimilar and 98% of adalimumab (ADA) users were switched to a biosimilar. For golimumab (GOL) and certolizumab pegol (CZP), no biosimilars were used.
To ensure a minimum potential exposure time of 3 years for all patients, those starting later than December 31, 2018, were not included in this study. All patients with RA, PsA, or AS (as diagnosed by a rheumatologist), aged ≥ 18 years, and in whom TNFi therapy was initiated as their first biologic in routine daily practice were invited to participate; 98% agreed to enter the cohort. All patients were treated according to registered dosage and frequency.
Drug survival was calculated from the dates of the first and last dose as documented in the case record. Brief stops (eg, because of surgery) were treated as continuous drug use, longer stops (eg, complication from surgery) were treated as drop-out, and the data after eventual restart were not used in the study. Patients remaining on TNFi therapy were censored at the last recorded visit to the clinic. Additional baseline factors comprised concomitant medication, disease duration, sex, and age. In addition, the reason for drug discontinuation was examined for the entire cohort and separately for the diseases.
Statistics. The metric for the primary outcome was the change in survival time on TNFi over the time of the cohort (2004-2018). To facilitate comparisons, patients were categorized as early (starting the TNFi in 2004-2008), intermediate (starting in 2009-2013), or recent (starting in 2014-2018).
To investigate drug survival, Kaplan-Meier plots and log-rank tests were used to test the overall difference in drug survival between the 3 observation groups (early, intermediate, and recent); these were followed by Cox regression analyses to estimate hazard ratios (HRs). Likewise, the effect of diagnosis was also examined by log-rank test followed by Cox regression analyses, and subsequently the overall difference in drug survival between the 3 observation groups for each diagnosis was examined. In addition, corrected analyses were performed adjusting for diagnosis and the interaction between observation group and diagnosis. This was done by including dummy variables for the diagnosis and corresponding interaction terms. All analyses were done in IBM SPSS Statistics version 27, and P values < 0.05 were considered statistically significant. This study was performed in compliance with the declaration of Helsinki and approved by the local Medical Ethics Committee of Slotervaart Hospital and Reade. All patients gave written informed consent.
RESULTS
A total of 1938 patients were included; 63% had a diagnosis of RA, 19% PsA, and 19% AS; 65% were female. The mean age was 51 (SD 13) years, with a median disease duration of 5.4 (IQR 2.0-12.5) years (Table 1; Supplementary Table S1, available from the authors upon request). Most patients started on either ETN (51%) or ADA (44%), median observation time until withdrawal or censoring was 2.19 (IQR 0.61-4.74) years, and 36% discontinued treatment during the observation period. The main reasons for discontinuing the TNFi were failure (21%) or adverse events (11%; Table 2). The early group was treated with ADA and ETN only; the intermediate and the recent group were treated with all 5 TNFi (ADA, ETN, CZP, GOL, and IFX; Supplementary Table S2, available from the authors upon request).
Baseline characteristics (N = 1938).
Reasons for discontinuing TNFi.
Median survival time periods are not reported because in several instances 50% survival was not reached, leading to results that could not be compared across subgroups. Drug survival strongly decreased during the observation period in the recent group, as shown by the survival curves of the 3 observation groups (Figure 1; log-rank test for overall difference: P < 0.001). Most of the decrease in drug survival was apparent in the final observation period, as confirmed by the pairwise HRs: 2.04 (95% CI 1.71-2.43, P < 0.001) for the early vs recent group, 1.06 (95% CI 0.87-1.29, P = 0.55) for the early vs intermediate group, and 1.92 (95% CI 1.58-2.35, P < 0.001) for the intermediate vs recent group (Figure 1 and Table 3).
Survival by groups of inclusion year. Log-rank test for overall difference: P < 0.001.
Results of the Cox regression analyses.
Drug survival differed between the 3 disease groups (Figure 2; log-rank test for overall difference: P < 0.001). The greatest difference was between RA and the other 2 diseases, as confirmed by the pairwise HR: 0.58 (95% CI 0.47-0.73, P < 0.001) for RA vs PsA, and 0.63 (95% CI 0.51-0.78, P < 0.001) for RA vs AS. No significant difference was found between PsA and AS (HR 1.08, 95% CI 0.82-1.42, P = 0.57; Figure 2).
Survival by disease. Log-rank test for overall difference: P < 0.001. AS: ankylosing spondylitis; PsA: psoriatic arthritis; RA: rheumatoid arthritis.
Further analyses on the interaction between diagnosis and observation period showed a similar pattern for RA and AS (lower survival especially in the recent group), whereas the decline occurred more gradually in PsA (Figure 3 and Table 3).
Survival by groups of inclusion year for RA, PsA, and AS separately. AS: ankylosing spondylitis; PsA: psoriatic arthritis; RA: rheumatoid arthritis.
DISCUSSION
In our large routine care clinic, the initial survival time of biologic initiators starting TNFi declined substantially over the 15 years since their introduction, especially in the last 5 years. The effect was more prominent for RA than for PsA and AS. This is most likely a result of the number of available alternative biologic/targeted synthetic disease-modifying antirheumatic drugs (DMARDs) for RA and the year these new drugs became available. This in turn also dictates the possibilities for treat-to-target protocols, which are more developed in RA than in the 2 other diseases, as more treatment options lead to more elaborate protocols. Over time, since treat-to-target has become routine since 2009, striving to achieve remission or a low disease activity state in as many patients as possible has become a major treatment goal.7-9 This has led to patients being switched to a different biologic much sooner, as stricter guidelines have been developed and implemented widely.
A major strength of this study is that although drug survival has been studied for TNFi, to our knowledge, none of the studies have looked at the changes in survival over time with recent patients. Although studies have looked at survival over time for RA (eg, Neovius et al),10 those studies focused on RA only, and none investigated recent patients, with the exception of Christiansen et al11,12 who studied patients with PsA and AS up to 2018. Although they did not look at patients with RA, they found results similar to ours in patients with PsA and AS, with shorter drug survival for more recent patients over the first 2 years of treatment.
Other strengths of this study include its setting in routine care, the large sample size and long observation time of the cohort, and the use of multiple different TNFi. In addition, we purposefully studied only the first response, avoiding complex analysis issues in patients switching treatment, especially in later years as a wider variety of biologics and targeted synthetic DMARDs have become available.
A limitation of this study is that in 2014 and 2015, fewer patients were included in the cohort because of the transition from paper to electronic patient records. Also, we did not study radiographic damage, nor did we investigate the application of glucocorticoid bridging strategies over the years. An interesting topic for further study would be to see if more frequent/earlier switching to another biologic or targeted synthetic DMARD would result in better outcomes for the switchers vs those remaining on their original drug.
As sex and age did not change over the observation groups, no adjustments were made in the survival analysis. However, although the focus of this paper was diagnosis and the observation group, sex and age are also important factors to consider and warrant further study in another investigation.
In conclusion, patients with RA, PsA, and AS starting TNFi in recent years have a much lower drug survival than those starting shortly after the drugs were introduced. This is most likely because of the availability of alternative treatments and treat-to-target protocols. As patients often expect that they will be using these drugs for years, or even for the rest of their lives, it is important for rheumatologists to keep in mind the likelihood that a patient may switch treatments early on and manage patients’ expectations accordingly.
Footnotes
The authors declare no conflicts of interest relevant to this article.
- Accepted for publication September 8, 2023.
- Copyright © 2024 by the Journal of Rheumatology
