Abstract
Objective. To provide real-world evidence about the reasons why Australian rheumatologists cease biologic (b) and targeted synthetic (ts) disease-modifying antirheumatic drugs (DMARD) when treating patients with rheumatoid arthritis (RA), and to assess (1) the primary failure rate for first-line treatment, and (2) the persistence on second-line treatments in patients who stopped first-line tumor necrosis factor inhibitors (TNFi).
Methods. This is a multicenter retrospective, noninterventional study of patients with RA enrolled in the Australian Optimising Patient outcome in Australian RheumatoLogy (OPAL) dataset with a start date of b/tsDMARD between August 1, 2010, and June 30, 2017. Primary failure was defined as stopping treatment within 6 months of treatment initiation.
Results. Data from 7740 patients were analyzed; 6914 patients received first-line b/tsDMARD. First-line treatment was stopped in 3383 (49%) patients; 1263 (37%) were classified as primary failures. The most common reason was “lack of efficacy” (947/2656, 36%). Of the patients who stopped first-line TNFi, 43% (1111/2560) received second-line TNFi, which resulted in the shortest median time to stopping second-line treatment (11 months, 95% CI 9–12) compared with non-TNFi. The longest second-line median treatment duration after first-line TNFi was for patients receiving rituximab (39 months, 95% CI 27–74).
Conclusion. A large proportion of patients who stopped first-line TNFi therapy received another TNFi despite evidence for longer treatment persistence on second-line b/tsDMARD with a different mode of action. Lack of efficacy was recorded as the most common reason for making a switch in first-line treatment of patients with RA.
Rheumatoid arthritis (RA) is a chronic progressive systemic autoimmune disease that results in significant pain, progressive joint damage, functional disability, and impaired quality of life1,2. There are 8 biological disease-modifying anti-rheumatic drugs (bDMARD), more than 12 conventional DMARD, and 2 targeted synthetic DMARD (tsDMARD) approved for the treatment of RA in Australia. Comparisons between available drugs in head-to-head randomized controlled trials are rare3,4 and current guidelines recommend that patients can be started on any 1 bDMARD as first- and second-line treatment5,6. As a result, treatment choices can be difficult.
A number of factors may influence treatment choices in the real-world practice setting. Although there have been various studies investigating treatment choices and the effectiveness of these treatments using data derived from several patient registries7,8,9,10, the reasons why individual b/tsDMARD are discontinued have not been well documented.
Some patients are refractory to b/tsDMARD and experience primary failure to treatment, while others show initial clinical response but eventually lose responsiveness (secondary failure). The primary failure rate for b/tsDMARD is currently unknown in Australia.
Persistence on treatment has been suggested as a surrogate for treatment effectiveness and the persistence of bDMARD in Australian patients with RA has been investigated7,11. However, to our knowledge, no study has included all of the currently available b/tsDMARD modes of action when investigating second-line persistence after first-line tumor necrosis factor inhibitor (TNFi) therapy.
This study aimed to use the Optimizing Patient outcome in Australian RheumatoLogy (OPAL) dataset to provide real-world evidence about the reasons behind stopping b/tsDMARD treatment in the Australian RA population. The study also aimed to provide real-world evidence on the effectiveness of b/tsDMARD in the first- and second-line setting through assessment of the first-line primary failure rate and persistence of second-line treatment after first-line TNFi failure. The study focused on persistence after first-line TNFi failure because TNFi are the most commonly prescribed first-line bDMARD for the treatment of RA in Australia.
MATERIALS AND METHODS
Study design and objectives
This is a multicenter retrospective noninterventional study of patients with RA treated in Australia, in routine clinical practice. The objectives of the study were to identify the reasons why Australian rheumatologists are ceasing b/tsDMARD for the treatment of RA, to assess the primary failure rate for first-line b/tsDMARD treatment, and to identify treatment choices after cessation of first-line TNFi and the persistence of the second-line treatments.
First-line treatment refers to patients who received b/tsDMARD treatment for the first time. Second-line treatment refers to patients who received their second b/tsDMARD treatment. Primary failure was defined as stopping treatment within 6 months. In the absence of disease activity data, the 6-month cutoff was chosen based on the local reimbursement requirements for demonstration of response to treatment within 12–16 weeks of starting a b/tsDMARD. Australian clinicians are required to provided documentation of adequate response to treatment for the patient to continue to receive a reimbursed supply of treatment12. Under the Australian Pharmaceutical Benefit Scheme (PBS), when a patient is commenced on any of the available b/tsDMARD, the treating clinician must document in a written application a 20% reduction in erythrocyte sedimentation rate (ESR) or C-reactive protein (CRP) and a 50% reduction in active joint count after 12 weeks for approval to continue treatment. Following the approval of the first repeat prescription, the reduction in active joint count and ESR or CRP must be maintained and documented every 6 months in subsequent applications for approval of repeat prescriptions. Therefore, secondary failure was defined as stopping treatment any time after 6 months of treatment initiation.
Data collection
De-identified data were extracted from the Australian OPAL dataset from 37 rheumatology practices. The OPAL dataset collects information from individual physicians’ servers entered during routine clinical consultations, using purpose-built worksheets in Audit4 software (Software4Specialists). This software serves as the patient’s medical record. Physicians can choose reasons for treatment cessation from a prespecified list of options (Supplementary Table 1, available with the online version of this article). The reason for cessation of a b/tsDMARD is a mandatory field when a treatment stop date is recorded, and only 1 reason can be chosen for a given treatment.
The activities of OPAL Rheumatology Ltd. have received overarching ethics approval from the University of New South Wales Human Research Ethics Committee based on a patient opt-out arrangement. This research protocol was approved by the committee (HC17232).
Patient population and eligibility criteria
Patients were included if they were registered in the OPAL dataset, were at least 18 years of age, and started a b/tsDMARD for the treatment of their RA between August 1, 2010, and June 30, 2017. The date of August 1, 2010, was chosen because it was the date from which all bDMARD under consideration in our study were reimbursed through the Australian PBS12.
Patients were excluded if they or their physicians opted out of data collection.
Statistical and analytical assessment
Data were analyzed using SAS (version 9.4; SAS Institute Inc.). There were 546 (8%) instances where the stop dates for a b/tsDMARD were missing. For those cases, the stop dates were imputed as follows: if the stop date of a b/tsDMARD was missing (and there was a subsequent b/tsDMARD listed), the stop date was imputed as the day before the next b/tsDMARD start date. In addition, there were 425 (6%) instances where the stop date for the b/tsDMARD was after the start date of the next b/tsDMARD. Therefore, the stop date for the first b/tsDMARD was set to the day before the start date of the following b/tsDMARD. Treatment persistence was defined as the duration of time between initiation and cessation of treatment. If there was no medication end date and no other medication initiated, it was assumed that the medication was ongoing (censored) at the time of data extraction (June 30, 2017). Comparisons between treatment persistence by treatment type were done using Kaplan-Meier methodology. The persistence results for individual TNFi therapies were grouped together because there is evidence that the persistence is similar between the different TNFi therapies in Australia7,11.
Descriptive statistics (mean, median, range, and 95% CI) are provided for continuous variables, and frequency counts and 95% CI for categorical variables. The frequency and type of reasons given to justify cessation of b/tsDMARD medication were summarized and reported.
At the time of data extraction, 9 b/tsDMARD were approved in Australia, including the TNFi adalimumab [ADA; subcutaneous (SC)], etanercept (ETN; SC), golimumab (GOL; SC), certolizumab pegol (CZP; SC), and infliximab [IFX; intravenous (IV)]. Other b/tsDMARD investigated were tocilizumab (TCZ; SC and IV), abatacept (ABA; SC and IV), rituximab (RTX; IV), and tofacitinib (oral, tsDMARD).
RESULTS
Patients and treatments
There were 25,237 patients with RA in the OPAL dataset (Figure 1). Of the 7740 who started a b/tsDMARD for the treatment of their RA between August 1, 2010, and June 30, 2017, there were 6914 patients who received first-line b/tsDMARD treatment. The majority were female (n = 5186, 75%) with a median age of 61 years (18–96 yrs) and a median disease duration (onset to last visit) of 10 years (0–73 yrs; Table 1).
Flow diagram of patients with RA in the OPAL dataset included in the study. RA: rheumatoid arthritis; OPAL: Australian Optimising Patient outcome in Australian RheumatoLogy; TNFi: tumor necrosis factor inhibitor.
Demographics and baseline characteristics for patients receiving first-line b/tsDMARD treatment.
ETN was the most commonly prescribed first-line agent (n = 1868, 27%), followed by ADA (n = 1788, 26%), GOL (n = 832, 12%), ABA (n = 609, 9%), TCZ (n = 555, 8%), tofacitinib (n = 518, 7.5%), CZP (n = 457, 7%), RTX (n = 230, 3%), and IFX (n = 57, 1%; Table 1).
First-line treatment was stopped in 3383 patients (49%), of whom 1263 (37%) experienced a primary failure as assessed by our definition (Table 2). In patients who stopped treatment, the highest percentage of primary failure was for patients receiving tofacitinib (124/185, 67%) and the lowest percentage of primary failure was for patients receiving TCZ (57/288, 20%).
No. patients receiving and ceasing first-line b/tsDMARD treatment.
A total of 5002 patients received first-line TNFi; 2560 (51%) patients stopped first-line treatment, of whom 964 (38%) and 1596 (62%) were classified as primary failures and secondary failures, respectively.
Physician-reported reasons for treatment cessation
There were 2656 reasons for cessation recorded by the treating physician for the 3383 patients who stopped first-line b/tsDMARD, and 928 reasons for the 1263 patients who were classified as primary failures of first-line b/tsDMARD (Table 3A and Table 3B).
Reason for treatment cessation in patients who stopped all first-line b/tsDMARD.
Reason for treatment cessation in patients classified as primary failures of first-line b/tsDMARD.
For all patients who stopped first-line b/tsDMARD, the most common reason for cessation was recorded as “lack of efficacy” (947/2656, 36%). CZP had the highest (114/221, 52%) and IFX had the lowest (4/25, 17%) proportion of treatment cessation recorded as lack of efficacy. Tofacitinib had the highest proportion (22/105, 21%) and GOL had the lowest proportion (22/327, 7%) of discontinuations due to “adverse reactions.”
In the subset of patients classified as primary failures, the most common reason for treatment cessation was recorded as lack of efficacy (388/928, 25%). CZP (60/92, 65%) had the highest proportion and RTX had the lowest (1/11, 9%) proportion of discontinuation due to lack of efficacy. IFX (2/6, 33%) had the highest proportion and GOL (12/139, 7%) the lowest proportion of discontinuation due to adverse events (Table 3B).
Persistence on b/tsDMARD post first-line TNFi
Persistence on second-line b/tsDMARD was assessed for patients who stopped first-line TNFi only, considering the low number of patients on other b/tsDMARD.
Of the patients who stopped first-line TNFi, 43% (1111/2560 patients) received second-line TNFi and this resulted in the shortest median time to stopping second-line treatment (11 months, 95% CI 9–12; Figure 2). The longest second-line median treatment duration after first-line TNFi was for patients receiving RTX (39 months, 95% CI 27–74).
Kaplan-Meier plot of second-line persistence post discontinuation of first-line TNF inhibitors regardless of reason for cessation. TNF: tumor necrosis factor.
Of the 964 patients classified as having a primary failure to first-line TNFi, 322 patients had a physician-recorded lack of efficacy as the reason for treatment cessation, of whom 309 (96%) received second-line treatment. The majority received other TNFi (n = 130, 42%; Table 4). The persistence rate on second-line b/tsDMARD treatment (Table 4) was highest for TCZ (78%) at 6 months and RTX (75%) at 12 months, and lowest for TNFi (60% at 6 months and 40% at 12 months). The median time to stopping second-line treatment was the longest for RTX (49 months, 95% CI 17–74) and shortest for TNFi (9 months, 95% CI 7–12; Table 4).
Kaplan-Meier estimates of persistence on second-line b/tsDMARD after discontinuation of first-line TNFi because of “lack of efficacy.”
Of the 1596 patients classified as secondary failures to the first-line TNFi, 429 patients had lack of efficacy recorded as the reason for cessation by the treating physician. The majority of these patients received other TNFi (n = 200, 47%) as second-line treatment (Table 4). The second-line persistence rates in this group were highest for RTX at both 6 and 12 months (83% and 69%, respectively) and lowest for tofacitinib (58% at 6 months and 29% at 12 months); it is important to keep in mind that the data are not mature enough for persistence on tofacitinib, considering that it became available on the PBS from October 2015. The median time to stopping second-line treatment was the longest for RTX (39 months, 95% CI 9–not determined) and shortest for TNFi (11 months, 95% CI 9–15; Table 4).
DISCUSSION
Data from controlled clinical trials suggest that about one-third of patients with RA experience primary response failure to a TNFi13,14,15 and a significant proportion of patients will also experience a secondary failure after an initial response16. One might expect that patients with a primary failure to a TNFi would be less likely to respond to a second TNFi than those who have sustained a secondary failure. This question has not been previously addressed in a large cohort of patients and particularly not in a real-world evaluation, to our knowledge. Our study used the Australian OPAL dataset, which is one of the largest clinical practice datasets in the world, to determine the reasons for cessation of first-line treatment with b/tsDMARD in patients with RA, and also whether the effectiveness of second-line TNFi was dependent on whether the failure to first-line TNFi was primary or secondary.
The most common reason for b/tsDMARD treatment cessation was recorded by the treating physician as lack of efficacy, which is consistent with findings from a number of registries as well as other retrospective observational studies10,17,18,19,20.
The rate of first-line treatment discontinuation in our study is slightly higher (49%) than that reported in a Canadian (38%) and UK (45%) study21,22. The overall primary failure rate for all b/tsDMARD as defined by cessation of treatment within 6 months was 37%, with a large variation observed across the individual b/tsDMARD. Of interest is the high primary failure rate observed for patients treated with tofacitinib (67%). To our knowledge, this is the first study that has investigated the primary failure rate of tofacitinib in a real-world setting.
A large number of patients (43%) received a second-line TNFi after stopping first-line TNFi, which resulted in the lowest median treatment persistence compared to second-line b/tsDMARD with a different mode of action. This is consistent with results from other studies demonstrating that patients who switched to another TNFi after first-line TNFi failure had an inferior persistence compared to those who switched to non-TNFi treatment20,23,24,25.
Our results have also shown that the persistence on a second-line TNFi was lower than on bDMARD with a different mode of action, regardless of whether the patients experienced primary or secondary failure to the first-line TNFi due to lack of efficacy. Further, primary or secondary failure of the first-line TNFi does not appear to affect the physician’s decision to prescribe a second TNFi, with the proportion of patients receiving a second TNFi being similar in both groups.
Patients receiving second-line RTX and TCZ after failure of TNFi had higher persistence rates and longer median treatment duration compared to other b/tsDMARD. This is in agreement with current literature24,25,26,27. Other observational studies reported that patients who had failed 1 TNFi demonstrated better disease activity when switched to RTX compared to switching to an alternative TNFi28,29,30.
It is important to acknowledge the limitations of our study; being a retrospective study with some missing data points has resulted in assumptions being made in some cases regarding treatment start and stop dates. Our study is also limited by the lack of specific clinical disease activity measures and could not document the magnitude of response for each of the drugs under study. However, built into the Australian PBS approval for continuing treatment is a documentation of clinical response, whereby persistence on a drug requires the treating clinician to document a reduction in inflammatory markers and a reduction in active joint counts, which are indicators of persistent clinical response. Also, given the type of data available in this dataset, we did not attempt to match underlying disease conditions across treatment groups, and therefore treatment choices may reflect other clinical characteristics that were not accounted for in this analysis. However, the large number of patients studied reduces the likelihood that differences in the baseline clinical characteristics in the treatment groups altered the outcomes. Results for subgroups with small numbers of patients, particularly for IFX and RTX, should be interpreted with caution.
This large real-world study found that lack of efficacy as classified by the treating physician was the most common reason for cessation of the first-line b/tsDMARD. Switching to a second TNFi after discontinuation of first-line TNFi therapy resulted in the lowest treatment persistence when compared to switching to b/tsDMARD with other modes of action. Nevertheless, a large proportion of the patients who stopped first-line TNFi therapy switched to another treatment with the same mode of action.
Acknowledgment
The authors acknowledge Software4Specialists (S4S) for assistance with data extraction. Medical writing assistance was provided by Dr. Theresa Wade from WriteSource Medical Pty Ltd. and funded by Roche Products Pty Ltd. Dr. Joseline Ojaimi from Roche Products Pty Ltd. provided editorial assistance, contributed to the development of the manuscript outline, and prepared the figures. Members of the OPAL Consortium who contributed data to the project include Professor Geoff Littlejohn, Chris Fong Rheumatology, Coast Joint Care, Barwon Rheumatology Service, Canberra Rheumatology, Coburg Rheumatology, Combined Rheumatology Service, Concord Repatriation General Hospital, Dr. Mark Collins Newcastle, Dr. Anna Finnis, Dr. Kate Franklyn, Dr. Alex Klestov, Dr. Laila Girgis, Drs. Godfrey and Wong, Dr. Jennifer Harmer, Dr. Daniel Lewis, Dr. Mona Marabani, Dr. John Moi, Dr. Jane Oliver, Frank Laska Rheumatology, Footscray Specialists Rooms, Georgetown Arthritis, Gold Coast Rheumatology, Gold Coast Specialist Centre, Hills Rheumatology, Hobart Specialists Group, Melbourne Arthritis Associates, Northern Rheumatology and Specialists Group, Peninsula Rheumatology, Redcliffe and Northside Rheumatology, Rheumatology ACT, Rheumatology Tasmania, Rheumatology United, Southern Rheumatology, Subiaco Rheumatology, Susan Street Specialists Centre, and Townsville Hospital-Rheumatology.
Footnotes
This study was supported by Roche Products Pty Ltd. (Australia). B.M. is an employee of Roche Products Pty Ltd. and reports stock ownership at Roche. P.B. was an employee of Roche Products Pty Ltd. until December 2017 and then worked as a consultant statistician during the conduct of the study. M.T. was an employee of Roche Products Pty Ltd. until December 2017 and then worked as a consultant statistician during the conduct of the study; and he reports stock ownership at Roche.
- Accepted for publication November 20, 2019.
REFERENCES
ONLINE SUPPLEMENT
Supplementary material accompanies the online version of this article.
