Real-world treatment persistence in patients with rheumatoid arthritis initiating DMARDs in Germany—a health insurance claims data analysis

The German Society for Rheumatology guidelines for the treatment of rheumatoid arthritis (RA) advocate a treat-to-target approach to achieve sustained remission or, in established disease, at least low disease activity [1]. The guidelines recommend initial treatment with a conventional synthetic (cs) disease-modifying antirheumatic drug (DMARD), preferably methotrexate (MTX) [1]. If a patient has an inadequate response to initial csDMARD therapy, the recommended treatment is a combination of csDMARDs, or if unfavorable prognostic factors are present, a combination of a csDMARD with a biologic (b) or targeted synthetic (ts) DMARD [1]. In addition to efficacy, factors such as safety, costs, and patient perceptions are important considerations in RA management [2, 3].

Many studies of RA treatment patterns in Germany report data collected from patients treated before the approval of Janus kinase (JAK) inhibitors in 2017 [4‐6]. Our aim was to investigate baseline characteristics and therapeutic pathways in patients with RA treated with DMARDs in Germany following the introduction of JAK inhibitors, with a specific focus on those who had received prior csDMARD or bDMARD therapy.

JAK inhibitors have enriched the therapeutic options for patients with RA who inadequately respond to csDMARDs. The extent to which JAK inhibitors have affected treatment patterns since they became available in Germany has not been studied in detail. To our knowledge, this is the first detailed analysis of baseline characteristics and treatment patterns in patients with RA in a real-world setting in Germany that includes JAK inhibitors and focuses on patients who previously received csDMARD or bDMARD therapy.

In the prior csDMARD cohort, the most commonly initiated treatment was a TNF‑α inhibitor; similar proportions of patients initiated treatment with a JAK inhibitor or other csDMARD. In the prior bDMARD cohort, JAK inhibitors were by far the most frequently initiated treatment. In both cohorts, anti-IL‑6 and anti-CD80/86 agents were rarely prescribed, and more patients received monotherapy than combination therapy. This finding is unexpected as the German guidelines advise that bDMARDs are prescribed in combination with MTX where possible. The proportion of patients on monotherapy in both cohorts of our study (61% and 78%) was higher than that reported in the RA Observation of Biologic Therapy (RABBIT), a German registry of patients with RA in which 45.6% of bDMARD-treated patients were receiving no concomitant csDMARD [10]. The higher proportion of patients prescribed monotherapy than combination therapy in our study may be due to poor tolerability experienced with prior MTX/other csDMARD treatment. A history of intolerance would be more relevant in the prior bDMARD cohort (which had the greater proportion of patients on monotherapy), having been escalated from prior csDMARD strategies. Alternatively, due to the selection process, it is possible that combination therapy was underreported in our study.

Glucocorticoid use was common throughout the assessment period. The majority of patients in each cohort (over 85%) received glucocorticoids in the pre-index period, as did over 65% of patients in each cohort in the post-index period while on index treatment. The proportion of patients receiving glucocorticoids during the pre-index phase was slightly higher than that reported in RABBIT (which was included in the JAK-pot study), in which 65% of patients initiating a new DMARD treatment were receiving glucocorticoids at baseline [10]. Data collected from the German national rheumatology database between 1996 and 2016 indicate that the proportion of patients with RA treated with glucocorticoids peaked at approximately 60% in 2001, decreasing to 45% in 2016 [11]. Of note, the German (and the European Alliance of Associations for Rheumatology) guidelines advise that glucocorticoids should be prescribed with initial csDMARD therapy, but reduction to a low dose is recommended within 8 weeks, and the duration of glucocorticoid therapy should be limited to 3–6 months [1, 12].

Of patients in the prior csDMARD cohort, 23% were switched to a different csDMARD or a second csDMARD was added to their treatment regimen as the index event. Furthermore, csDMARD-pretreated patients initiating treatment with JAK inhibitors or TNF‑α inhibitors continued treatment for longer and, as indicated by the log-rank test and Cox regression analysis, were less likely to discontinue their treatment than those remaining on csDMARDs.

This may indicate that another csDMARD therapy in csDMARD-pretreated patients does not result in a lasting response or that it is less well tolerated than other treatments. Consequently, these patients should be monitored even more closely to achieve the treatment goal in accordance with the treat-to-target principle.

However, it is notable that in the prior csDMARD cohort, the majority of patients who received another csDMARD strategy were treated with monotherapy at index (n = 175; Table 2). Only a minority of patients (n = 57) received combination csDMARD therapy, despite either MTX, SSZ, and HCQ or MTX and LEF being recommended combination treatment options [1]. This finding may partially explain the shorter continuation of a second csDMARD in the prior csDMARD cohort.

Our results indicate that the mean and median treatment duration was longer with monotherapy than with combination therapy within the prior csDMARD cohort, while no significant difference was observed in the prior bDMARD cohort. In contrast, previous German real-world studies have reported longer treatment duration or persistence with combination therapy versus monotherapy. For example, in a recent retrospective study based on claims data, Grellmann et al. found that treatment persistence in patients with RA was longest in those receiving anti-TNF therapy in combination with a csDMARD, with approximately 50% of patients remaining on treatment for 270 days, compared with approximately 30% of those on anti-TNF therapy alone [13]. In addition, based on data from RABBIT, Zink et al. found that drug continuation rates were higher for etanercept or infliximab when combined with MTX or another DMARD than for either agent alone [14]. Our contrasting finding may reflect a switch from combination therapy to monotherapy during the follow-up period once patients have achieved remission or low disease activity, as persistence was defined conservatively. Moreover, the group of patients on combination therapy might reflect a population with higher disease activity and treatment resistance, leading to an earlier switch to other drugs.

Comorbidities such as cardiovascular disease, infections, and metabolic disorders were common in the prior csDMARD and prior bDMARD cohorts, which is in line with previous reports [13, 15‐17]. When baseline cardiovascular diseases were compared across index classes, a comparatively high proportion of patients treated with anti-CD80/86 agents had hypertension or ischemic heart disease. These comorbidities were reported in a higher proportion of patients treated with JAK inhibitors than TNF‑α inhibitors, indicating a greater baseline cardiovascular risk in the former group. The finding of longer treatment duration with JAK inhibitors compared with TNF inhibitors in the current study, despite less favorable cardiovascular risk profile in the JAK inhibitor cohort, is of note in light of the recently published Oral Surveillance study which failed to show non-inferiority of tofacitinib compared to TNF inhibitors in a cardiovascular risk population [18]. In the Oral Surveillance study, the mean treatment duration was similar in the tofacitinib 5 mg and 10 mg groups and the TNF inhibitor group (41.1 and 38.5 months and 40.2 months, respectively); therefore, the cardiovascular risk appeared to have no obvious impact on treatment persistence [18].

There are a number of limitations to the current analyses, including the low patient numbers, which may explain the lack of significant differences for some of the regressions performed. Another limitation was the applied hierarchy used to determine the index treatment (JAK inhibitor > bDMARD > csDMARD), which resulted in a shift towards JAK inhibitors in terms of patient numbers. The hierarchy could have also influenced the switching and augmentation patterns within the prior csDMARD and prior bDMARD groups in the 12-month follow-up, which was limited to the index years 2017 and 2018. For example, csDMARD-treated patients who switched to JAK inhibitors in 2017 were included in the JAK inhibitor cohort and did not qualify as “csDMARD switchers.” This means that the numbers of switches and augmentations in particular could be underestimated, which, in turn, could have led to overestimation of the duration of treatment (“time on treatment”) in the csDMARD and bDMARD groups. More specifically, individuals who initiated treatment with a csDMARD and switched to a JAK inhibitor in 2017/2018 were not included in the csDMARD cohort by default, and so forth. This could have overestimated treatment duration in csDMARD/bDMARD-treated patients in comparison with those receiving JAK inhibitor treatment. Nevertheless, patients initiating treatment with JAK inhibitors or TNF‑α inhibitors remained on treatment for longer than did those on csDMARDs. A further limitation was that patients in the prior csDMARD and prior bDMARD cohorts had had at least one previous csDMARD or bDMARD prescribed, respectively, in the 24-month pre-index period. As this was based on prescriptions rather than clinical factors, patients experiencing inadequate response to treatment (failing to achieve treatment targets) could not be differentiated from those with an intolerance to treatment. As the pre-index period was restricted to 2 years, earlier data were not captured and, therefore, the number of patients reported may be underestimated. In addition, at the time the analysis was performed, tofacitinib and baricitinib were the only JAK inhibitors available in Germany. Since that time, upadacitinib and filgotinib have also been approved for treatment of RA in Europe. Finally, claims data are primarily collected for accounting purposes; thus, clinical parameters such as disease severity grades, duration of disease, prognostic factors, and laboratory results were not captured. In addition, the numbers of previous treatment changes were unavailable. Time on treatment was based on prescription data, including the date of medication collection and the recommended dose as per the package leaflet. Therefore, it is not known if a physician prescribed a different dosing pattern or non-standard dose, or whether the patient took their medication as prescribed (e.g., they may have stopped treatment before taking the whole pack).

This analysis provides detailed information regarding treatment patterns and baseline characteristics in patients with RA in a real-world setting in Germany since the introduction of JAK inhibitors. Such data are important to understand how patients are treated in clinical practice and to identify whether improvements can be made to optimize patient care and outcomes. We found that TNF‑α inhibitors and JAK inhibitors were the most commonly prescribed DMARD class in the prior csDMARD and prior bDMARD cohorts, respectively. Only a minority of patients received another csDMARD strategy in the prior csDMARD group, and this mainly as monotherapy. In both the prior csDMARD and prior bDMARD cohorts, more patients received monotherapy than combination therapy, which was unexpected, as the German guidelines advocate combination treatments when escalating treatment from initial csDMARD therapy. Time on treatment was significantly longer for patients initiating treatment with JAK inhibitors or TNF‑α inhibitors than for those starting on other csDMARDs in the prior csDMARD cohort. However, there was no statistically significant difference in drug persistence between JAK inhibitors and TNF‑α inhibitors in this cohort. In the prior bDMARD cohort, approximately 40% of patients initiating treatment with JAK inhibitors continued treatment during the 12-month follow-up, compared with approximately 29% of those initiating treatment with TNF‑α inhibitors.