Introduction
Tumor necrosis factor inhibitors (TNFi) are effective therapies for controlling the signs and symptoms and reducing progression of erosive disease in many patients with moderate-to-severe rheumatoid arthritis (RA) [
1,
2]. The most commonly used TNFi agents approved by the US Food and Drug Administration for the treatment of moderate-to-severe RA are adalimumab, etanercept, and infliximab [
3‐
5]. Adalimumab and etanercept are self-administered injectable agents and infliximab is administered intravenously (IV) [
6‐
8]. Clinicians frequently switch treatment to a second TNFi agent when patients do not achieve an adequate response [
9‐
11], lose their initial response [
12], or experience adverse events with their first course of a TNFi agent [
13,
14].
The use of TNFi agents for the treatment of RA is associated with significant costs, estimated to be between $14,000 and $22,000 annually [
4,
5], especially for patients who require dose escalation to achieve or maintain a clinical response, which entails 2–44% higher costs compared with non-dose escalating patients [
15‐
17]. Additionally, RA is a chronic disease for which treatment-free remission is rare and thus requires long-term therapy, adding to the lifetime cost of treatment. Evaluation of treatment patterns and associated costs is important for clinicians to make informed treatment decisions and for payers to manage costs; this evaluation is also important specifically among US veterans with RA because such an assessment has never been conducted in this population. Therefore, we evaluated clinical outcomes and associated costs in US veterans with RA enrolled in the Veterans Affairs Rheumatoid Arthritis (VARA) registry who initiated TNFi therapy after VA enrollment. This analysis compared disease activity and drug costs in patients treated with TNFi agents as a class, rather than comparing specific agents. The comparison determined differences between patients who were persistent on a single TNFi agent and patients who interrupted therapy with a single TNFi agent or switched to a different TNFi agent. Patients were specifically evaluated on duration of treatment, clinical response, rate of switching, and drug costs associated with the initial TNFi treatment episode as well as subsequent treatment episodes within the VA health care system.
Discussion
One goal of this study was to assess clinical outcomes, duration of treatment, and costs of treatment associated with TNFi agents as a class rather than specific agents. Such an analysis enabled comparison between patients who were persistent on a single TNFi agent and patients who interrupted therapy with a single TNFi agent or switched to a different TNFi agent. Our findings show that all patients had a similar degree of improvement in disease activity as measured by DAS28 scores whether they were persistent on a single TNFi agent with and without an interruption of therapy or switched TNFi agents. DAS28 data indicate that the change in post-treatment disease activity did not differ significantly among all groups analyzed—single therapy, interrupted therapy, and switched therapy. Patients did differ with regard to DAS28 scores before starting TNFi therapy, which were higher in patients who eventually switched therapy than in patients with single therapy or interrupted therapy. The absolute disease severity level was higher in patients who switched TNFi agents. These data suggest that patients with a higher initial disease activity are more likely to switch to a second TNFi agent than patients with a lower initial disease activity.
Prior reports describing changes in TNFi therapy have focused on switching between TNFi agents and have not considered interrupted therapy. Bonafede et al. reported rates of 12–25% for interrupted therapy with a single TNFi agent during the first year of TNFi therapy based on a 45-day gap in treatment [
9], which was similar to the 25% rate of interruption in our patients based on a 90-day gap in treatment; however, their switching rate was 13%, which was lower than our observed switching rate of 28%. Scrivo et al. reported a 10% switching rate [
29] over a span of 3–47 months, Hyrich et al. and Zhang et al. both observed a 13% switching rate [
11,
30] over a mean of 15 months, and Virkki et al. found a 37% switching rate [
12] over a mean of 28 months. Schabert et al. reported a low switching rate of 9–11% in the first year of treatment [
5].
Comparison of our results with those of other groups who have reported significant clinical improvement in disease severity when switching TNFi agents is challenging because of the differences in methodology and clinical settings for these studies. Many of these comparative studies do not have baseline disease activity assessment during the first TNFi course. Instead, these analyses compare the response during the second course of TNFi treatment to a clinical course during the first course of TNFi treatment [
13,
31,
32]. Others have reported similar responses with a first and second TNFi agent with increases in disease activity in between courses [
33]. In some cases, high persistence rates (71–74%) on second TNFi agents are reported, but without disease severity measures [
30]. A separate report of US veterans switching biologics demonstrated that veterans who switched had a higher DAS28 score before TNFi therapy than veterans who did not switch [
14]. Virkki et al. and Scrivo et al. emphasized the benefit of switching TNFi agents when a secondary loss of efficacy was seen [
12,
29]. In our analysis, the change in clinical response with the second TNFi agent from initial baseline was similar to the response with the initial TNFi agent. The disease activity during the first and second courses was similar. This observation may have represented clinical benefit until the secondary loss of efficacy occurred. There is speculation that the development of anti-TNFi antibodies may play a potential role in these cases with secondary loss of efficacy [
9].
Another goal of our analysis was to compare costs of TNFi therapy within a drug class between patients who used a single TNFi agent and patients who switched TNFi agents. This work demonstrates that, while the annualized drug costs of the first course of treatment for patients who switched TNFi therapy were slightly higher than the costs of the first course of treatment in patients who remained on a single TNFi agent, these cost differences were not statistically significant. The annualized costs for patients with interrupted therapy with a single TNFi agent were similar for the first and second courses of treatment. In contrast, the costs for the second course of TNFi treatment in patients who switched to a second TNFi agent were significantly higher than those for the second course of treatment in patients who had interrupted therapy with a single TNFi agent. Differences in cost may have been the result, in part, of differences in disease activity, which required dose escalation to address this higher disease activity. The higher costs for second courses of medication may have been related to dose escalation during the second course of treatment, which has been previously reported [
15].
While the use of biologic agents is associated with significant costs, most analyses report that these agents fall within a currently accepted threshold of cost effectiveness [
34‐
37]. Prior studies have specifically focused on costs per agent rather than costs of the TNFi therapy class. These studies have generally demonstrated that costs for infliximab are higher than those for injectable agents [
4,
5,
38‐
40]. We have previously published that dose escalation is more common with infliximab in subsequent courses, which is also associated with higher costs [
25].
Strengths of our study included rheumatologist-confirmed diagnosis of RA in contrast to the use of administrative data in many other studies, the wide geographic distribution of patients across the US, the collection of baseline and post-treatment disease severity information using the DAS28, and standardized medical records and administrative databases within the VA system across all participating sites. Veterans enrolled in VA care have access to TNFi therapy as needed.
Limitations of the study included a predominance of men with RA of long-term duration reflecting the US veteran population, which may limit the generalizability of these findings to other populations. This study population of US Veterans is principally male and elderly in comparison with most RA populations that are predominantly female and of younger age. The demographic characteristics of the patients in this study may result in terminating and/or switching therapy more frequently because of comorbidities and concurrent medications. This bias may explain the lower rate of switching reported in other studies, which included populations that were younger and mostly female [
5,
9,
11,
30]. The potential exists that patients may have received TNFi therapy outside the VA, which would not be captured by this analysis; however, in our experience, US veterans rarely seek TNFi therapy from other sources while receiving their care through the VA. The federally negotiated cost for TNFi therapy in the VA system may be less than costs in the community, which may limit comparisons with other systems. Our study did not identify the reasons for TNFi switching or interrupting TNFi therapy. Such factors could confound the observed results. Another limitation is our lack of evaluation of background drug-modifying antirheumatic drug therapy, which may have an impact on decisions regarding continuation, interruption, or switching of TNFi therapy. Because of potential confounding factors inherent to observational studies, more research is needed to understand reasons for switching TNFi therapy and the effects of switching on overall outcomes in RA.
Acknowledgments
This research, article processing charges, and the open access charge for this study were funded by Immunex Corp., a fully owned subsidiary of Amgen Inc., and by VA HSR&D Grant SHP 08-172. The sponsor contributed to study design and analysis and interpretation of data, provided critical input into the manuscript, and made the decision to submit the article for publication. All named authors meet the International Committee of Medical Journal Editors (ICMJE) criteria for authorship for this manuscript, take responsibility for the integrity of the work as a whole, and have given final approval to the version to be published. Medical writing support in the preparation of this manuscript was provided by Edward Mancini, DPM, Jessica Ma, Ph.D., (Amgen Inc.), and Julia R. Gage, Ph.D. (on behalf of Amgen Inc.).