FormalPara Key Summary Points

Why carry out this study?

In patients with rheumatoid arthritis (RA), clinical practice guidelines recommend switching from a biologic (b) or targeted synthetic (ts) disease-modifying anti-rheumatic drug (DMARD) to another b/tsDMARD with an alternative mechanism of action (without preference to any class) in case the treatment target is not achieved.

Although data exist for the effect on clinical outcomes while switching from a tumor necrosis factor inhibitor (TNFi) to another b/tsDMARD, there is limited research on switching between an interleukin-6 receptor inhibitor (IL-6Ri) and a Janus kinase inhibitor (JAKi), given both classes act on the IL-6 pathway.

What was learned from this study?

This observational study showed improvement in clinical outcomes when switching patients with RA from an IL-6Ri to a JAKi (and vice versa), and the responses were generally comparable between the two groups, indicating that the healthcare providers may consider such switch, without concerns that commonalities in their mode of action may hamper clinical effectiveness.

This is the first study to measure clinical outcomes when switching from an IL-6Ri to a JAKi and from a JAKi to an IL-6Ri in the same patient population and suggests that the distinct mechanisms of action of the two classes account for the observed clinical responses.

Introduction

Rheumatoid arthritis (RA) is a chronic, systemic inflammatory disease, characterized by synovial inflammation in multiple joints, leading to progressive and irreversible joint damage. If appropriate treatment is not provided, RA can cause significant pain and swelling in the hands and feet, loss of physical function, and a deterioration in the overall quality of life [1, 2].

Conventional synthetic (cs) disease-modifying anti-rheumatic drugs (DMARDs) have been in use for decades in RA treatment [1]. However, in the last 20 years, significant advances have been made in the treatment landscape with the advent of other DMARDs. Various biologic (b) and targeted synthetic (ts) DMARDs (b/tsDMARDs) are now available as treatment options for RA, including tumor necrosis factor inhibitors (TNFi), interleukin-6 receptor inhibitors (IL-6Ri), cytotoxic T-lymphocyte-associated protein 4-immunoglobulin (CTLA4-Ig), anti-CD20, and Janus kinase inhibitors (JAKi) [2,3,4,5,6,7,8].

Despite these advances, many patients fail to achieve or sustain the treatment target (i.e., low disease activity [LDA] or remission) and may require multiple therapies and switching between drugs. For patients with RA who fail treatment with a csDMARD, clinical practice guidelines (the American College of Rheumatology [ACR] 2021 and the European Alliance of Associations for Rheumatology [EULAR] 2022) recommend that a bDMARD or a tsDMARD should be added to achieve the treatment target [3, 4]. Further, if a bDMARD or tsDMARD fails, treatment with another bDMARD or tsDMARD (with an alternative mechanism of action) should be considered [3, 4].

Although data exist for the effect on clinical outcomes while switching from a TNFi to other b/tsDMARDs [9], there is limited research on switching from a non-TNFi bDMARD to a tsDMARD [10], and from a tsDMARD to a bDMARD [11, 12]. In these situations, healthcare providers may particularly hesitate to switch between IL-6Ri and JAKi, since they both impact IL-6 signaling [8, 13, 14].

This US-based observational study aimed to explore the effect of switching between IL-6Ri (following discontinuation of JAKi) and JAKi (following discontinuation of IL-6Ri) on disease activity and patient-reported outcomes (PROs) in patients with RA.

Methods

Data Source and Patients

The CorEvitas (formerly known as Corrona) RA Registry is a prospective, multicenter, observational disease-based registry launched in US in 2001 [15, 16]. Adult patients with RA enrolled in the registry, who initiated either IL-6Ri (i.e., sarilumab or tocilizumab) or JAKi (i.e., baricitinib, tofacitinib, or upadacitinib) during or after December 2012 (baseline visit) and had a follow-up visit at 6 (± 3) months after therapy initiation, were included in the present study. Eligible patients were divided into two groups: (1) “IL-6Ri initiators” who were initiated on an IL-6Ri following discontinuation of a JAKi and (2) “JAKi initiators” who were initiated on a JAKi following discontinuation of an IL-6Ri.

The study was conducted in accordance with the Declaration of Helsinki, and all participating investigators were required to obtain full ethics or institutional review board (IRB) approval for conducting research in patients. The Sponsor approval and continuing review were obtained through central IRB, New England Independent Review Board (NEIRB No. 120160610). For academic sites that did not receive a waiver to use the central IRB, approval was obtained from the respective governing IRBs, and documentation of approval was submitted to the Sponsor prior to initiating any study procedures. All registry patients were required to provide written informed consent prior to participation.

Study Assessments and Outcomes

The main objectives of the study were to assess the changes in disease activity and PROs in the patients with RA switching from an IL-6Ri to a JAKi (and vice versa), and to compare these clinical outcomes between the two groups.

Demographics, clinical characteristics, and medication parameters (concurrent medications, prednisone use, prior therapy, and line of therapy) were assessed at baseline. Patient therapy patterns were evaluated between the initiation and the 6-month follow-up visit; the proportion of patients continuing the index therapy at the 6-month follow-up visit and the proportion of patients discontinuing the index therapy at or prior to the 6-month follow-up visit were assessed.

The primary outcome of the study was the mean change in the Clinical Disease Activity Index (CDAI) from baseline until 6 (± 3) months post-initiation. Continuous secondary outcomes included the mean change from baseline in the Health Assessment Questionnaire (HAQ), patient-reported pain (0–100 mm visual analog scale [VAS]), patient-reported fatigue (0–100 mm VAS), tender joint count (TJC, 0–28), swollen joint count (SJC, 0–28), Physician Global Assessment (MDGA, 0–100 mm VAS), Patient Global Assessment (PtGA, 0–100 mm VAS), and morning stiffness duration. Categorical secondary outcomes included the achievement of LDA (CDAI ≤ 10) among those with moderate or high disease activity at baseline (baseline CDAI > 10), and remission (CDAI ≤ 2.8) among those with low, moderate, or high disease activity at baseline (baseline CDAI > 2.8) [17]. Further, minimal clinically important differences (MCIDs) were assessed for the following outcomes: HAQ using ≥ 0.3 units improvement from baseline [18], and patient-reported pain, patient-reported fatigue, MDGA, and PtGA using ≥ 10 units improvement from baseline [17, 19,20,21,22,23].

Statistical Analyses

Outcomes after the index b/tsDMARD (IL-6Ri or JAKi) were evaluated at the 6-month follow-up visit post initiation. Patients who switched to any other b/tsDMARD prior to the follow-up visit were excluded; however, those who discontinued without starting any other b/tsDMARD prior to the follow-up were included in this analysis and their responses were evaluated at the 6-month follow-up visit.

Further, to assess the potential impact of the intercurrent event of switching to any other b/tsDMARD prior to the follow-up visit, a sensitivity analysis was performed. For the sensitivity analysis, all outcomes were reanalyzed by imputing non-response for patients who discontinued their index medication prior to the 6-month follow-up. Specifically, “non-response” was imputed for binary outcomes and last observation carried forward (LOCF) was used to impute continuous outcomes.

Disease activity measures and PROs were evaluated at baseline and the 6-month follow-up visit. Continuous outcomes were summarized as unadjusted mean changes from baseline (with 95% confidence intervals [CIs]), while categorical outcomes were summarized as response rates (with 95% CIs). Differences in the outcome measures between therapy groups were evaluated using both unadjusted and covariate-adjusted linear and logistic regression models. We reported β (95% CI) for continuous outcomes and odds ratio (OR [95% CI]) for binary outcomes, with JAKi initiators as the reference group.

In the adjusted models, covariates were selected based on an imbalance (defined as |standardized difference (SDi)| > 0.1) between the groups at baseline and effects on the outcome and the primary independent variable (therapy class). |SDi| provides a measure of clinically important difference even when no statistically significant difference is present; |SDi| < 0.1 is commonly taken to indicate a negligible difference between the treatment groups.

All analyses were performed using Stata 15 and/or SAS 9.4.

Results

Demographic and Clinical Characteristics

Of 55,069 patients enrolled in the registry, 340 patients initiated an IL-6Ri (following discontinuation of a JAKi; IL-6Ri initiators) and 608 patients initiated a JAKi (following discontinuation of an IL-6Ri; JAKi initiators) during or after December 2012; the enrollment period ended in March 2020. However, 218 IL-6Ri initiators and 464 JAKi initiators did not have a 6-month follow-up visit (Fig. 1). There were three primary reasons for this reduction in the sample of initiators with 6-month follow-up visit: (1) the follow-up visits for the patients occurred outside the 6-month window, (2) the patients did not accumulate 6 months of follow-up at the time of analysis, and (3) the patients were lost to follow-up. Remaining 122 IL-6Ri initiators and 144 JAKi initiators were initially considered for this study. However, 22 IL-6Ri initiators (18%) and 15 JAKi initiators (10%) were further excluded from the main analysis due to switching to a new b/tsDMARD before the 6-month follow-up. Thus, the main analysis included 100 IL-6Ri initiators (82%) and 129 JAKi initiators (90%), and the sensitivity analysis included all IL-6Ri (N = 122) and JAKi (N = 144) initiators.

Fig. 1
figure 1

Study population. b/tsDMARD biologic/targeted synthetic disease-modifying anti-rheumatic drug, IL-6Ri interleukin-6 receptor inhibitor, JAKi Janus kinase inhibitor, N number of patients

Patient characteristics were balanced across various measures at baseline; however, some differences were noted in the main analysis. Those switching to IL-6Ri (vs. JAKi) were younger (mean [SD] age: 57.2 [11.3] vs. 59.2 [12.7] years; |SDi|: 0.167), had a higher baseline CDAI (mean [SD] CDAI: 23.9 [13.1] vs. 19.7 [12.9]; |SDi|: 0.324), had used more csDMARDs (prior use of 0, 1, and 2 + csDMARDs: < 5%, 22%, and 74% vs. 4.7%, 30%, and 65%, respectively; |SDi|: 0.197), and had a different concomitant therapy pattern at the time of switch (monotherapy, combination with methotrexate, and combination with other csDMARD: 42%, 24%, and 34% vs. 50%, 30%, and 20%, respectively; |SDi|: 0.317). The detailed demographic and clinical characteristics of IL-6Ri and JAKi initiators are summarized in Table 1 and Table 2, respectively. The findings from the sensitivity analysis were similar and are summarized in Supplementary Materials Table S1 and Table S2.

Table 1 Demographic characteristics of IL-6Ri vs. JAKi initiators
Table 2 Clinical characteristics of IL-6Ri vs. JAKi initiators

Among the patients included in the main analysis, 29% of IL-6Ri initiators (n = 29/100) and 25% of JAKi initiators (n = 32/129) discontinued the therapy by the 6-month follow-up visit (Table 3). In the sensitivity analysis population, 42% of IL-6Ri initiators (n = 51/122) and 33% of JAKi initiators (n = 47/144) discontinued the therapy by the 6-month follow-up visit (Supplementary Materials Table S3).

Table 3 Therapy patterns for IL-6Ri and JAKi initiators between initiation and the 6-month follow-up visit

Change in Continuous Clinical Outcomes from Baseline to Follow-Up

Within each group, both IL-6Ri and JAKi initiators saw numerical improvement from baseline for all continuous outcomes, including CDAI, HAQ, patient-reported pain, patient-reported fatigue, TJC, SJC, MDGA, PtGA, and morning stiffness duration. For IL-6Ri initiators, the change from baseline to the 6-month follow-up visit was statistically significant for all outcomes, except the HAQ; for JAKi initiators, a significant improvement was observed only for the patient-reported pain (Table 4). In sensitivity analysis, the change from baseline was significant for most of the outcomes among IL-6Ri initiators, except HAQ, patient-reported fatigue and PtGA, while among JAKi initiators, a significant improvement from baseline was noted only for CDAI, patient-reported pain, and TJC (Supplementary Materials Table S4).

Table 4 Continuous clinical outcomes for IL-6Ri and JAKi initiators: change from baseline to 6 months post-initiation

Although unadjusted estimates of mean change were numerically greater for IL-6Ri (vs. JAKi) initiators, both unadjusted and adjusted comparisons of IL-6Ri initiators and JAKi initiators showed no significant differences in the clinical outcomes across treatment groups (Table 4). Similarly, no significant differences were noted between treatment groups in the sensitivity analysis, except for a reduced duration of morning stiffness noted in IL-6Ri vs. JAKi initiators (Supplementary Materials Table S4).

Change in Categorical Clinical Outcomes from Baseline to Follow-Up

Within each group (for both IL-6Ri and JAKi initiators), a significant proportion of patients achieved CDAI LDA, CDAI remission, and MCID for other measures (HAQ, patient-reported pain, patient-reported fatigue, MDGA, and PtGA) (Table 5 and Supplementary Materials Table S5).

Table 5 Categorical clinical outcomes for IL-6Ri and JAKi initiators: change from baseline to 6 months post-initiation

Although IL-6Ri (vs. JAKi) initiators had numerically higher unadjusted response rates for most of the clinical outcomes, there were no significant differences noted in the unadjusted comparisons between the two groups (Table 5). The odds of achieving the clinical outcomes were similar between the groups based on the adjusted analysis, except for CDAI LDA. The IL-6Ri initiators (with moderate-to-severe disease) had higher odds of achieving CDAI LDA (adjusted odds ratio, aOR [95% CI]: 3.30 [1.01, 10.78]) than JAKi initiators (Table 5 and Fig. 2).

Fig. 2
figure 2

Adjusted odds ratio for categorical clinical outcomes: change from baseline to 6 months post-initiation (IL-6Ri vs. JAKi initiators). aOdds ratio are shown for IL-6Ri initiators with JAKi initiators as the reference group (presented on a scale of log 2). CDAI Clinical Disease Activity Index, HAQ Health Assessment Questionnaire, IL-6Ri interleukin-6 receptor inhibitor, JAKi Janus kinase inhibitor, LDA low disease activity, MCID minimal clinically important difference

In the sensitivity analysis, there were no significant differences noted between the treatments, except for the improvement of 10 + units in MDGA (Supplementary Materials Table S5). While the absolute difference in achievement of 10 + units’ improvement in MDGA was similar between IL-6Ri and JAKi initiators (26.8% vs. 31.9%), IL-6Ri initiators reported a higher baseline disease activity based on MDGA (40.3 vs. 31.4; |SDi|= 0.41). Therefore, after adjusting for differences in baseline MDGA, the odds of achieving 10 + units’ improvement in MDGA was lower for IL-6Ri vs. JAKi initiators (OR = 0.49; 0.26–0.94).

Discussion

To our knowledge, this is the first study to evaluate the effect of switching from JAKi to IL-6Ri vs. switching from IL-6Ri to JAKi in the same patient population. Our study found that a proportion of patients experienced improvement when switching from one class to the other; the responses were comparable for IL-6Ri and JAKi initiators, except for the achievement of CDAI LDA, which favored IL-6Ri (vs. JAKi) initiators.

Within each group, IL-6Ri and JAKi initiators showed improvement in continuous outcomes, although the changes were not significant for the HAQ among IL-6Ri initiators, and for all the outcomes (except patient-reported pain) among JAKi initiators. This could be due to the relatively low disease activity of the included patients at baseline, which may have made it difficult to measure any significant improvement with the available sample size [24, 25]. A significant proportion of patients still experienced improvement in categorical outcomes within each group, consistent with the well-established efficacy of IL-6Ri and JAKi [26,27,28].

Further, no difference was observed for most outcomes when comparing IL-6Ri with JAKi initiators (after adjusting for differences in the baseline characteristics). Previously, two network meta-analyses (NMAs) compared the efficacy of IL-6Ri with that of JAKi in TNF-inadequate responders (IR), and showed higher ACR20/50 responses with tocilizumab than tofacitinib [29] and higher ACR50 response with sarilumab than baricitinib [30]. However, patients included in the present study (compared with those included in the NMAs) were switching between IL-6Ri and JAKi (not switching from TNFi) and had less disease activity at baseline, history of multiple prior therapies, and longer disease duration, all of which could have made it difficult to detect any differences [24, 25, 31,32,33]. Recently, few other NMAs showed comparable efficacy (relative to placebo) of non-TNFi biologics and JAKi in patients with RA refractory to TNFi treatment, and of bDMARDs and JAKi in patients with inadequate response to bDMARDs [34, 35]. The findings of the present study are consistent with data from a cohort study (based on two observational registries), in which no significant differences were observed for CDAI remission, CDAI 50/70/85, and MCID-based CDAI improvement between tofacitinib and tocilizumab in patients with bDMARD failure [36].

In the present study, patients who switched between IL-6Ri and JAKi had failed treatment with multiple prior b/tsDMARDs and had active disease. For such patients, the ACR and EULAR guidelines recommend treatment with either a bDMARD or a tsDMARD, without preference to any class [3, 4]. This is supported by various studies, which have demonstrated better outcomes in patients with RA, when switching to an alternative mechanism of action therapy after failure of the first-line TNFi [9, 11, 37,38,39], although there is published research showing equivalent results between TNFi cycling and switching to non-TNFi [40]. Limited data exist regarding switching between non-TNFi and JAKi [10], and our study endeavors to address this gap.

While considering the alternative class to switch, there may be hesitancy to switch from an IL-6Ri to a JAKi (and vice versa) as they both work on the IL-6 pathway [8, 13, 14]. However, these therapies have unique mechanisms of action, which could account for the observed outcomes. IL-6 binds to the membrane-bound or the soluble forms of IL-6R, leading to the activation of various intracellular signaling pathways. IL-6Ri works extracellularly to block this IL-6 signaling (and thus, activation of JAK-STAT3, PI3K-PKB/Akt, and Ras/MAPK pathways) along with the regulation of CD4+ T-cells, VEGF, and NLRP3 inflammasome, which are involved in the pathophysiology of RA [14, 41, 42]. On the other hand, JAK enzymes are bound to the intracellular domains of multiple type 1 and type 2 cytokine receptors and are activated by the ligand binding, leading to auto-phosphorylation, activation of the STAT proteins, and ultimately transcription of the pro-inflammatory genes. JAKi primarily acts by interrupting the JAK–STAT signaling (and thus, activation of multiple interleukins [including IL-6], colony stimulating factors, and interferons) along with the suppression of RANKL, CD80/CD86, natural killer cells, TYK2, and NLRP3 inflammasome, which are known to be involved in RA [7, 8, 43, 44].

Although patients experienced improvements in this study (with either direction of switch), it is possible that switching to an alternative mechanism of action therapy that does not directly impact IL-6 signaling could have resulted in a better response. However, this was beyond the scope of our analyses. The main strength of this study was its dataset, based on the large US-based registry with vast geographic representation and robust evaluation [15]. However, as with every retrospective observational study, this study has a few limitations. The study outcomes were evaluated through 6 months only and less than one-third of the IL-6Ri/JAKi initiators could be included in the analyses. There could also be unidentified sources of bias and confounding in the real-world setting, and factors affecting treatment adherence, which would have impacted the study outcomes [45].

Conclusions

In this observational study, both IL-6Ri and JAKi initiators were associated with improvements in clinical outcomes, with comparable responses noted when switching from an IL-6Ri to a JAKi and vice versa. Both IL-6Ri and JAKi therapies are known to have effects on the IL-6 pathway; however, they have distinct mechanisms of action that could account for improvement in disease activity and other outcomes after switching. Overall, switching to an IL-6Ri or a JAKi can be a suitable choice in patients with RA who have an inadequate response to a previous JAKi or IL-6Ri, respectively.