Improvement or Worsening of Disease Activity After Switch to Sarilumab in Patients With Rheumatoid Arthritis With a Partial Response to Adalimumab

Abstract

Objective 

The aim of this study was to assess the effect of switching from adalimumab to sarilumab monotherapy in partial responders with rheumatoid arthritis from the MONARCH randomized trial and its open-label extension (OLE).

Methods 

Partial response was defined as improvement in Clinical Disease Activity Index (CDAI) of 12 or 6 units (baseline score: >22 or >10 and ≤22, respectively). Proportions of adalimumab partial responders with meaningful worsening or improvement at OLE weeks 12 and 24 were evaluated using 2 CDAI thresholds (≥6 and ≥12 points), 28-joint Disease Activity Score using erythrocyte sedimentation rate (≥0.6 and ≥1.2 points), Health Assessment Questionnaire Disability Index (≥0.22 and ≥0.30 points), Simple Disease Activity Index (≥7 and ≥13 points), physician and patient global assessments (≥10 and ≥20), and 28-joint swollen and tender joint counts (≥1 and ≥2 joints). Outcomes were analyzed using mixed-effect models with repeated measures for observed cases. The p values were produced using Wilcoxon tests.

Results 

Of 369 enrolled patients, 320 (87%) entered the OLE and 155 switched from adalimumab to sarilumab; 59% (91/155) were partial responders. At week 24, 4%–17% and 2%–12% of partial responders experienced a worsening using the lower and higher thresholds, respectively, whereas 47%–78% and 27%–66% experienced improvement.

Conclusions 

Partial responders to adalimumab who switched to sarilumab had a low likelihood of experiencing meaningful worsening, with most patients showing meaningful improvement or no change in disease activity. This may help alleviate patients' fears of worsening when considering switching to a treatment with a different mechanism of action.

Despite advances in the treatment of rheumatoid arthritis (RA), only approximately 50% of methotrexate (MTX)-naive patients achieve remission after 24 weeks with biologics.^1,2 Guidelines recommend adjusting therapy every 3–6 months if low disease activity (LDA) or remission is not achieved.³ However, despite these recommendations, a significant number of patients with active disease do not change treatment.⁴ This is often due to patient preference to remain on their current treatment, and many patients cite that they are satisfied with their level of disease control, while also reporting active disease.^5–8 Concerns about the risk of clinical worsening as well as uncertainty about toxicity of a new regimen are cited as key barriers to switching treatments.^6,8 Indeed, in a patient preference survey of 6135 RA patients, 68% reported that concerns about loss of control over their RA would dissuade them from changing treatment.⁶ In general, patients' risk aversion tends to encourage maintenance of the status quo and, in this case, to not change treatments.⁹

A reluctance to change treatment to avoid risking clinical worsening may be particularly relevant to patients who had experienced significant improvement, yet did not achieve LDA—for example, those who achieved a 50% improvement according to the American College of Rheumatology criteria, but not LDA. Unfortunately, there is a dearth of data evaluating the likelihood of clinical worsening in patients who changed therapy after a partial response to previous treatment.

The open-label extension (OLE) of the MONARCH study offered an opportunity to evaluate whether this reluctance to change treatment to avoid clinical worsening is supported by evidence. MONARCH was a 24-week randomized controlled trial (RCT) that demonstrated superiority of monotherapy with sarilumab, a human antibody against the interleukin 6 receptor, to monotherapy with adalimumab, a human antibody against tumor necrosis factor α.¹⁰ After 24 weeks, patients were allowed to enter an OLE, where everyone received open-label sarilumab. The results of the OLE have been reported previously.¹¹

The aim of this post hoc analysis of the MONARCH OLE was to assess meaningful worsening or improvement after switching from adalimumab to sarilumab in a subset of patients who had only partially responded to adalimumab treatment during the double-blind phase.

PATIENTS AND METHODS

Study Design

The design of MONARCH (NCT02332590) and its OLE have been described in detail previously.^10,11 Briefly, MONARCH was an international, multicenter, phase 3 RCT that evaluated the efficacy and safety of sarilumab monotherapy (200 mg subcutaneously every 2 weeks [q2w]) compared with adalimumab monotherapy (40 mg subcutaneously q2w) over 24 weeks in patients with RA who were unsuitable candidates for continued MTX treatment because of intolerance or inadequate response.¹⁰ The last visit of the double-blind phase of the RCT was the first visit of the OLE (OLE baseline). In the OLE, 2 groups of patients were switched to open-label sarilumab 200 mg q2w: those who had been randomized to double-blind adalimumab in MONARCH (switch group) and those randomized to double-blind sarilumab 200 mg in MONARCH (continuation group).¹¹ Here, we focus on the data reported for the adalimumab to sarilumab switch group. Data from the continuation group, which was included as a control to evaluate for potential bias related to receiving open-label treatment, are shown in the Supplemental Material (see Supplemental Digital Content 1, https://links.lww.com/RHU/A536).

Patients

Individuals aged ≥18 years with a 28-joint Disease Activity Score using erythrocyte sedimentation rate (DAS28-ESR) >5.1, C-reactive protein ≥8 mg/L or ESR ≥28 mm/h, RA duration ≥3 months, ≥6 swollen joints on the 66-joint swollen joint count (SJC), and ≥8 tender joints on the 68-joint tender joint count (TJC) were eligible for inclusion in the RCT.¹⁰ Only patients with a partial response to treatment in the double-blind phase were included in this analysis. Partial response was defined as an improvement of at least the minimal clinically important difference (MCID) in Clinical Disease Activity Index (CDAI) during the double-blind phase without achieving LDA or remission. The MCID was defined based on patient disease activity at baseline. For those with high disease activity (CDAI >22) at the RCT baseline, the MCID improvement threshold was 12 units, whereas for those with moderate disease activity (CDAI >10 to ≤22), the MCID improvement threshold was 6 units.¹² Written informed consent was obtained from patients before study participation. The study was conducted in compliance with institutional review board regulations and in accordance with ethics principles from the International Conference on Harmonization Guidelines for Good Clinical Practice and the Declaration of Helsinki.

Assessments

Meaningful worsening or improvement in efficacy and patient-reported outcomes in partial responders were assessed at OLE weeks 12 and 24, relative to OLE baseline, using a higher (more specific) and a lower (more sensitive) threshold for each parameter (Table 1). The parameters assessed included the CDAI, DAS28-ESR, Health Assessment Questionnaire Disability Index (HAQ-DI), Simple Disease Activity Index (SDAI), physician global assessment (MDGA), patient global assessment (PtGA), 28-joint swollen joint count (SJC28), and 28-joint tender joint count (TJC28).

TABLE 1 - Thresholds for Meaningful Worsening or Improvement

Outcome, Point(s)	Meaningful Change: Lower Threshold (More Sensitive)	Meaningful Change: Higher Threshold (More Specific)
	Worsening (Score Increase) or Improvement (Score Decrease)	Worsening (Score Increase) or Improvement (Score Decrease)
CDAI 12	≥6	≥12
DAS28-ESR	≥0.6	≥1.2
HAQ-DI	≥0.22	≥0.30
SDAI	≥7	≥13
MDGA	≥10	≥20
PtGA	≥10	≥20
SJC28	≥1	≥2
TJC28	≥1	≥2

For each parameter, any value between the thresholds of improvement or worsening was considered to represent no meaningful change. For example, a CDAI score decrease of 7 points would be considered a meaningful change from the perspective of the lower threshold (≥6 points), but not a meaningful change from the perspective of the higher threshold (≥12 points). Safety assessments reported in the OLE included treatment-emergent adverse events (AEs) and serious AEs, recorded by the investigator. Adverse events of special interest were identified using prespecified search criteria.

Statistical Analyses

Outcomes listed in Table 1 were analyzed using mixed-effect models with repeated measures in observed cases only. The p values associated with change from OLE baseline were produced using Wilcoxon tests.

RESULTS

Patient Characteristics

Of 369 patients enrolled in the MONARCH double-blind phase, 320 (87%) entered the OLE: 155 switched from adalimumab to sarilumab open-label (switch group) and 165 continued sarilumab, with administration changing to open-label (continuation group) (Fig. 1). Overall, 165 (52%) of patients were classified as partial responders at OLE baseline, of whom 91/165 (55%) were from the switch group and 74/165 (45%) were from the continuation group.

FIGURE 1: Patient disposition. ^aOpen-label. DBP, double-blind phase.

At RCT baseline, demographic characteristics of partial responders were similar between those initially randomized to receive adalimumab (switch group) and sarilumab (continuation group) (see Table 1, Supplemental Digital Content 1, https://links.lww.com/RHU/A536, which depicts the characteristics of partial responders who continued sarilumab in the OLE). Overall, 84% of partial responders were women, and the mean age was 52 years. Demographic and clinical characteristics at RCT baseline for the switch group are summarized in Table 2.

TABLE 2 - Double-Blind Phase Baseline and OLE Baseline Characteristics of Partial Responders to Adalimumab Who Switched to OLE Sarilumab

Parameter	Adalimumab → Sarilumab Switch Group (n = 91)a
Double-blind phase baseline
Age, mean ± SD, y	52.5 ± 11.3
Women, n (%)	76 (84)
White, n (%)	81 (89)
BMI ≥30 kg/m2, n (%)	24 (26)
Time since RA diagnosis, mean ± SD, y	5.3 ± 5.7
No. prior nonbiologic DMARDs, n (%)
1	46 (51)
2	29 (32)
≥3	16 (18)
Rheumatoid factor positive, n (%)	59 (66)b
Anti-CCP antibody positive, n (%)	68 (77)c
OLE baseline
CDAI, mean ± SD	19.6 ± 8.2
DAS28-ESR, mean ± SD	4.9 ± 0.9
HAQ-DI (0–3), mean ± SD	1.4 ± 0.6
SDAI, mean ± SD	21.3 ± 8.8
MDGA (0–100 mm), mean ± SD	31.9 ± 16.8
PtGA (0–100 mm), mean ± SD	47.6 ± 20.2
SJC28, mean ± SD	4.1 ± 3.3
TJC28, mean ± SD	7.5 ± 5.4

^aIntention-to-treat population.

^bTwo patients had missing values and are not included.

^cThree patients had missing values and are not included.

BMI, body-mass index; CCP, cyclic citrullinated peptide; DMARD, disease-modifying antirheumatic drug; SD, standard deviation.

Meaningful Changes in Efficacy and Patient-Reported Outcomes

Depending on the outcome measure, week 12 OLE data were available for 82–88 (90%–97%) partial responders to adalimumab who switched to sarilumab. At week 24 OLE, data were available for 81–84 (89%–92%) patients in that group.

At both OLE visits, across the outcome measures and regardless of the threshold used, partial responders who switched from adalimumab to sarilumab were more likely to achieve meaningful improvement (week 24 OLE: more sensitive threshold, 47%–78%; more specific threshold, 27%–66%) than to experience meaningful worsening (week 24 OLE: more sensitive threshold, 4%–17%; more specific threshold, 2%–12%) (Fig. 2). Twenty-four weeks after switching to sarilumab, partial responders to adalimumab were between 4 and 24 times more likely to experience improvement than worsening, depending on the outcome. For example, for SJC (both thresholds), the ratio of improvement to worsening was approximately 4 (eg, low threshold: 53 patients improved/14 patients worsened = 3.79). A similar calculation for DAS28-ESR (high threshold) yields a ratio of 24.

FIGURE 2: Outcomes at weeks 12 and 24 of the OLE phase in partial responders to adalimumab who switched to sarilumab at the lower (left) and higher (right) thresholds for outcomes. At week 12, n = 87 for all measures, except CDAI (n = 86), DAS28-ESR (n = 88), SDAI (n = 82), and MDGA (n = 86). At week 24, n = 83 for all measures, except DAS28-ESR (n = 84), HAQ-DI (n = 82), and SDAI (n = 81).

Partial responders to sarilumab who continued to receive sarilumab into the open-label phase experienced additional improvement in all outcomes except for HAQ-DI (see Fig. 1, Supplemental Digital Content 1, https://links.lww.com/RHU/A536, which shows outcomes of the OLE in partial responders to sarilumab [continuation group]). However, their improvement from OLE baseline to OLE week 24 was, on average, smaller than the improvement experienced by the switch group (see Table 2, Supplemental Digital Content 1, https://links.lww.com/RHU/A536, which shows adjusted mean changes in efficacy and patient-reported outcomes in adalimumab to sarilumab switch and sarilumab continuation groups).

Safety

The nature and occurrence of the treatment-emergent AEs following the switch were in line with the established safety profile of sarilumab. Seven (8%) partial responders who switched to sarilumab experienced a serious AE (Table 3 and see Table 3, Supplemental Digital Content 1, https://links.lww.com/RHU/A536, which summarizes adverse events in the OLE in the partial responder population who continued sarilumab).

TABLE 3 - Summary of Adverse Events in the OLE in the Partial Responder Population Who Switched From Adalimumab to Sarilumab

n (%)	Adalimumab → Sarilumab Switch Group (n = 91)
Overview of treatment-emergent AEs
Any AE	69 (76)
Any serious AE	7 (8)
Any AE leading to sarilumab discontinuation	5 (5)
Any AE leading to death	2 (2)
AEs of special interest
Infections	39 (43)
Serious infections	1 (1)
Opportunistic infections	2 (2)
Tuberculosis	1 (1)
Leukopenia	8 (9)
Thrombocytopenia	1 (1)
Hepatic disorders	8 (9)
Diverticulitis/potential GI perforations	0
GI ulcerations	0
Elevation in lipids	1 (1)
Hypersensitivity	4 (4)
Anaphylaxis	0
Injection site reactions	8 (9)
Malignancy	2 (2)
Malignancy excluding NMSC	2 (2)
Lupus-like syndrome	0
Demyelinating disorders	0

AE, adverse event; GI, gastrointestinal; NMSC, nonmelanoma skin cancer.

DISCUSSION

Our data are based on RA patients who attained a meaningful but only partial response to adalimumab as their first-line biologic therapy and remained in moderate or high disease activity at 24 weeks and subsequently switched to sarilumab. We showed that these patients face a relatively low likelihood (2%–12%) of experiencing meaningful clinical worsening, and a relatively high likelihood of experiencing meaningful improvement (up to 78%, depending on the measure used and its threshold) upon switching to sarilumab. Across all RA disease activity and functional measures, approximately 50% of patients experienced improvement using more stringent thresholds, and approximately 70% experienced improvement using less stringent (and more sensitive) but nevertheless well-accepted thresholds.

Taken together, our findings suggest that switching to treatment with a different mode of action at approximately 6 months is an effective approach in improving the outcomes in partial responders to biologic therapies, rather than only extending the duration of existing treatment. Indeed, the European League Against Rheumatism committee now recommend switching to a biologic disease-modifying antirheumatic drug with a different mode of action ahead of switching to a second TNF inhibitor for patients who are inadequately responding to their first tumor necrosis factor α inhibitor.³

Our data are consistent with those of the SELECT-COMPARE study, which evaluated switching (but in a blinded manner) from adalimumab to the Janus kinase inhibitor (JAKi) upadacitinib, and vice versa, on a background of MTX in patients who achieved an incomplete response on their original medication, defined as CDAI >10, after 26 weeks of treatment.^13,14 In this study, rates of worsening based on DAS28–C-reactive protein thresholds of 0.6 and 1.2 were only 20% and 7%, respectively (upadacitinib to adalimumab), and 4% and 1% (adalimumab to upadacitinib).¹³

As mentioned earlier, patients with RA who achieved only partial clinical improvement may be unwilling to switch therapies^6,8 out of concerns that may stem from loss aversion,¹⁵ despite the fact that continued partial response is associated with adverse long-term outcomes, compared with the achievement of LDA or remission.³ Moreover, RA patients who change therapy and attain lower disease activity states incur lower healthcare costs.¹⁶ To address patients' and their providers' concerns as broadly as possible, we studied a range of disease activity and PRO thresholds that reflect clinically important changes from both perspectives.¹⁷ In addition, the magnitude of meaningful worsening in our analysis, captured over a period of 24 weeks, was consistent with the occurrence of reporting of RA worsening as an AE in RA clinical trials, which generally ranges from 0% to 6%.^10,18

The safety profile of sarilumab in partial responders who either switched from adalimumab or continued existing sarilumab therapy was generally consistent with that reported in the double-blind phase of MONARCH and in other phase 3 trials and also with that anticipated for an interleukin 6 receptor inhibitor, without any requirement to wash out or otherwise delay sarilumab initiation.^19–21

Our study's limitations include the potential for introduction of expectation bias when all patients in a double-blinded study voluntarily switch to open-label active treatment. Although there was some evidence of this expectation bias in that patients who continued sarilumab had continued meaningful improvement at OLE week 24 than week 12, it is also possible that additional time on therapy may improve disease control for some patients. Notably, the rates of meaningful worsening were similar between the 2 visits. This was true for both patient-reported outcomes and for outcomes that include more objective measures of inflammation (eg, SJC). However, these changes were of lesser magnitude than those observed in patients switching from adalimumab, indicating that it is unlikely that this expectation bias had a material effect on the results. A further potential limitation is the lack of validated thresholds for meaningful worsening, in contrast to the thresholds for improvement, which have been more extensively used for most of the end points that we assessed.¹² It was for that reason that we studied several clinical outcomes commonly used in RA assessment. Finally, we can only speculate what would have happened with partial responders to adalimumab had they stayed on that regimen. Available RCTs do not suggest that there would be much or any improvement between 24 and 52 weeks with continued adalimumab treatment.^22,23 On the other hand, changes noted from baseline in several parameters evaluated (see Table 2, Supplemental Digital Content 1, https://links.lww.com/RHU/A536, which shows adjusted mean changes in efficacy and patient-reported outcomes) suggest that partial responders who continued on sarilumab into the open-label phase did experience some improvement, but generally of lesser magnitude than the switch group. Taken together, our data suggest that, in partial responders to adalimumab, the combination of switching mechanism of action and switching from double-blind adalimumab to open-label sarilumab is more beneficial than just switching from double-blind to open-label sarilumab. It would be of interest to assess whether these results translate to any partial responders who are considering switching to any other treatment with a different mechanism of action. It is unclear if switching between other drug combinations or switching to sarilumab following multiple biologic therapies would lead to similar results.

In summary, our findings may help alleviate fears of disease worsening in patients with RA who have shown only a partial response to adalimumab and are considering a switch to sarilumab. Our data suggest the risk of worsening is low (eg, 2%–12%), and such patients are 4- to 24-fold more likely to experience improvement than worsening, and thereby improve their long-term outcomes.