Factors associated with successful discontinuation of certolizumab pegol in early rheumatoid arthritis

Abstract Aim The Certolizumab–Optimal Prevention of joint damage for Early Rheumatoid Arthritis (C‐OPERA) study demonstrated that in methotrexate (MTX)‐naïve early RA patients with poor prognostic factors, 1‐year certolizumab pegol (CZP) therapy added to the first year of 2‐year optimized MTX therapy brings radiographic and clinical benefits through 2 years even after stopping CZP. This exploratory analysis aimed to identify factors at baseline and at CZP discontinuation associated with successful CZP discontinuation. Methods MTX‐naïve early RA patients with poor prognostic factors entered C‐OPERA (NCT01451203), a multicenter, randomized controlled trial. Patients were randomized to CZP + MTX (n = 159) or PBO + MTX (n = 157); those who completed the 1‐year, double‐blind period received MTX alone in Year 2 (CZP + MTX→MTX, n = 108; PBO + MTX→MTX, n = 71). Association between factors at baseline or at discontinuation of CZP and clinical/radiographic outcomes were evaluated by multiple logistic regression analysis. Predictive value cut‐offs were calculated using receiver operating characteristic analysis. Results Sex (male) and low baseline Disease Activity Score of 28 joints – erythrocyte sedimentation rate (DAS28‐ESR) were associated with simple disease activity index (SDAI) remission (≤3.3), whereas high baseline DAS28‐ESR and modified total Sharp score (mTSS) were associated with clinically relevant radiographic progression (yearly progression mTSS > 3) at Week 104 (across both treatment arms). Low DAS28‐ESR (<2.1) and rheumatoid factor (RF; <74 IU/mL) at discontinuation of CZP were associated with SDAI remission at Week 104. At Week 104, SDAI remission was achieved by 75.0% (42/56) of patients with low DAS28‐ESR and RF at discontinuation, compared to 15.4% (2/13) of patients with high DAS28‐ESR and RF. Conclusion Patients with low RF and low disease activity after treatment with CZP + MTX may be able to discontinue CZP without risk of loss of response.


| INTRODUC TI ON
Rheumatoid arthritis (RA) is a chronic inflammatory disease characterized by progressive inflammatory synovitis, joint destruction and chronic disability. Anti-tumor necrosis factor (anti-TNF) agents provide an effective treatment option for patients with RA. 1,2 Certolizumab pegol (CZP) is a humanized anti-TNF antibody fragment conjugated to polyethylene glycol, approved for the treatment of inflammatory diseases, including RA. 3 Early treatment of patients with CZP has been associated with a reduction in inflammation, inhibition of structural damage and improved long-term outcomes. 4,5 Recent studies have investigated whether the clinical benefits of early CZP treatment are maintained following CZP withdrawal.
The Certolizumab-Optimal Prevention of joint damage for Early RA (C-OPERA) study was a phase 3, 2-year, multicenter study. 5,6 C-OPERA demonstrated the clinical benefit of initial 1-year CZP treatment on 2-year optimized methotrexate (MTX) therapy, in MTXnaïve early RA patients with poor prognostic factors. C-OPERA was double-blind (DB) for the first year, with patients randomized to receive either CZP or placebo (PBO) alongside MTX. 6 For the second year (post-treatment [PT] period), CZP and PBO were discontinued and all patients were maintained on MTX alone. 5 During the PT period, although most patients who had received CZP + MTX maintained good clinical status after withdrawal of CZP, approximately 25% of patients (28/108) went on to flare. 5 Identifying factors at baseline or at CZP discontinuation that are associated with disease flare may help clinicians decide whether to cease or continue CZP treatment after attaining the desired response with initial CZP + MTX treatment.
In this present study, we performed post-hoc analyses on the C-OPERA data to identify factors at baseline and at CZP discontinuation associated with successful CZP discontinuation.

| ME THODS
In the C-OPERA study (NCT01451203), the efficacy and safety of CZP + MTX were compared with PBO + MTX in MTX-naïve patients with early RA in Japan. Details of the C-OPERA study have been published elsewhere. 5,6 In brief, eligible patients were within 1 year of disease onset and had poor prognostic factors including: positive for anti-cyclic citrullinated peptide antibody (anti-CCP, >3 times the upper limit of the normal range [ULN]), positive for rheumatoid factor (RF) and/or bone erosions on radiographs of the hands or feet.
Prior to the DB period, patients were randomized to receive either PBO + MTX or CZP + MTX. Optimized oral MTX was started at 8 mg/ wk and escalated up to 16 mg/wk by Week 8, if tolerated. Patients who completed the 52-week DB period were eligible to enter the 52-week PT period. In the PT period, administration of PBO or CZP was discontinued but patients remained on optimized MTX therapy. Disease activity and van der Heijde modified total Sharp score (mTSS) were evaluated at baseline, Week 52, Week 104, and at withdrawal. Last observation carried forward analysis was employed for clinical outcomes, and linear extrapolation was used for estimation of mTSS in patients who withdrew from the study.
Multivariate logistic regression analysis was applied for factors with a P value <0.1 to identify independent predictive factors. Factors with a P value <.05 were considered to be associated.

| Factors at CZP discontinuation associated with maintenance of clinical response
All analyses in this section were performed for patients who were originally allocated to the CZP + MTX group and entered the PT period. Association between factors at Week 52 (discontinuation of CZP) and Week 104 outcomes were analyzed by logistic regression analysis. Factors included in the analysis were: DAS28-ESR, HAQ-DI, mTSS, RF, anti-CCP antibody, and MMP-3. Log value + 1 was used for mTSS, RF, anti-CCP antibody, and MMP-3 to normalize values. Univariate logistic regression analysis was used to identify factors associated with the clinical outcomes. Factors with P value <.1 were included in the multiple logistic regression analysis to identify independent predictive factor(s); the Youden index on the receiver operating characteristic curves was used to estimate appropriate cut-off values for these factors. The proportion of patients achieving SDAI remission was calculated for patients with values that were lower than, and equal to or higher than, the cut-off value of the relevant predictive factor(s).

| RE SULTS
Patients in the C-OPERA study were randomized to CZP + MTX

| Baseline factors associated with a clinical response at Week 52 in the PBO + MTX treatment group
In the MTX + PBO arm, low baseline DAS28-ESR and low HAQ-DI were associated with SDAI remission at Week 52 by univariate analysis (

| Baseline factors associated with maintenance of clinical response
In univariate analyses, CZP treatment was associated with a higher SDAI remission rate and a lower cRRP rate at Week 104. Out of the remaining factors included in the analysis, only gender, BMI, baseline DAS28-ESR and HAQ-DI were also associated with SDAI remission at Week 104 (Table 2). After adjustment for these factors, Week 104 SDAI remission was associated with gender (P = .026), baseline DAS28-ESR (P = .012), and CZP treatment (P = .047; Table 2).

| Factors at CZP discontinuation associated with maintenance of clinical response
Univariate and multiple logistic regression analyses showed that DAS28-ESR and RF at CZP discontinuation (Week 52) were  Table 3). There was a rapid reduction in disease activity (indicated by both DAS28-ESR and SDAI measures) with CZP treatment, and only a slight increase was observed after discontinuation of CZP ( Figure 1). RF at discontinuation of CZP treatment (Week 52) was lower compared to baseline (Figure 2 Figure 3). The distribution of disease activity at baseline, Week 52, and Week 104 stratified by Week 52 DAS28-ESR and RF is presented in Figure 4. In patients with high disease activity at baseline, 16 out of 24 patients (66.7%) with low DAS28-ESR and low RF at Week 52 were in SDAI remission at Week 104; whereas only 1 patient out of 9 (11.1%) with high DAS28-ESR and high RF at baseline was in remission at Week 52.

| D ISCUSS I ON
In Some studies have suggested the possibility of discontinuing or interrupting treatment with biologics after attainment of clinical response. [7][8][9] However, identifying the patients who can discontinue treatment without going on to flare, can prove challenging.
Therefore, it is of therapeutic value to identify factors that can be used to predict maintenance of a "good disease state" following discontinuation of treatment with biologics. Furthermore, if biologic treatment could be stopped successfully in some patients, it would dramatically reduce the economic burden of RA therapy and any risk of adverse events related to prolonged biologic use.
The post-hoc analyses presented here aimed to provide clinicians with information to aid their decision-making both at the start of treatment, to help determine whether to opt for biologic co-therapy with MTX, and following disease control, to guide their decision regarding the discontinuation of biologic treatment.

| Baseline factors associated with maintenance of clinical response
In C-OPERA, a substantial number of patients withdrew from the study, particularly from the placebo group. In the post-hoc analyses reported here, patient characteristics and baseline disease status were associated with clinical and radiographic outcomes; therefore, patient withdrawal due to disease activity has the potential to bias findings should be interpreted with caution due to the high number of patient dropouts, these results are in line with findings from previous studies. [10][11][12] Other factors, including auto-antibodies, were not associated with Week 104 outcomes (P > .05). In the C-OPERA study, only patients with high levels of anti-CCP antibody at baseline were included, therefore, it is not possible to draw conclusions regarding the importance of baseline auto-antibodies, especially anti-CCP antibodies, in the prediction of long-term outcomes.

| Factors at CZP discontinuation associated with maintenance of clinical response
It has been suggested that discontinuation of biologic treatment without subsequent disease flare may be an achievable target, for example in early RA patients. 13 In a previous study in early RA, lower disease activity at withdrawal (deep remission) was associated with discontinuation without disease flare, 7,8  Nevertheless, the data presented here implicate disease activity and RF in the maintenance of disease control after discontinuation of CZP treatment.
In CZP-treated patients, both DAS28-ESR and RF were reduced at Week 52 compared to baseline. Both DAS28-ESR at baseline and at reported. In that report, disease activity and RF were associated with low disease activity 2 years after discontinuation of adalimumab. 15 Taken together, immunological status PT and the good clinical status achieved with anti-TNFs may play a crucial role in long-term disease management in addition to the short-term control of disease activity.
CZP + MTX patients who entered the PT period of the C-OPERA study, maintained a very high radiographic non-progression rate, despite some patients withdrawing from the study due to disease flare.
Radiographic non-progression during the first year of treatment was 91.7% and was maintained after CZP discontinuation (94.4%). 5 Due to the small number of patients with radiographic progression in this group, the association between factors at Week 52 and cRRP was not analyzed.
The results of these post-hoc analyses suggest that patients with low disease activity and low RF may be able to consider discontinuation of CZP, whereas patients with high disease activity and/or high RF should consider continuing CZP treatment.

| CON CLUS IONS
The results of this post-hoc analyses indicate that disease activity and joint damage at the start of treatment are predictive of longterm outcomes in patients treated with optimized MTX plus CZP.
Therefore, patients with poor prognostic factors may benefit from initial aggressive treatment with CZP and MTX. In addition to deep remission, low RF at discontinuation may be predictive of remission maintenance following discontinuation of CZP.

ACK N OWLED G EM ENTS
The authors thank the patients, the investigators and their teams who took part in this study. The authors acknowledge Pauline Ralston, UCB Pharma, for statistical support. The authors also acknowledge Arianna Psichas, PhD, and Simon Foulcer, PhD, from Costello Medical, UK, for medical writing and editorial assistance in preparing this manuscript for publication, based on the authors' input and direction.

DATA AVA I L A B I L I T Y S TAT E M E N T
The datasets generated and analyzed for this article were derived from study data available in anonymized format upon reasonable request via the Clinical Study Data Request platform (www.clini calst udyda tareq uest.com).