Factors associated with the achievement of biological disease-modifying antirheumatic drug-free remission in rheumatoid arthritis: the ANSWER cohort study

Intensive treatment strategies utilizing biological disease-modifying antirheumatic drugs (bDMARDs) have revolutionized rheumatoid arthritis (RA) treatment. Remission or low disease activity is now a realistic goal for most patients. After achieving remission, it would be advantageous if remission could be maintained without using bDMARDs (bDMARDs-free remission (BFR)) because of the associated cost-effectiveness and prevention of adverse events. It would be of clinical importance to determine which bDMARD is advantageous for achieving BFR and in what conditions BFR could be successfully maintained in daily clinical practice [1].

Discontinuation of bDMARDs after remission has been attempted in previous studies, including prospective uncontrolled trials and randomized controlled trials (RCTs) [2‐17]. For example, discontinuation of infliximab (IFX) was attempted in patients with established RA, and low disease activity was maintained in 43% of patients at 1 year after discontinuation in the RRR study [5]. Similarly, remission was maintained in 58% of patients with established RA at 6 months after discontinuation of adalimumab (ADA) in the HONOR study [8]. The remission maintenance rate after discontinuation of a soluble tumor necrosis factor (TNF) receptor, etanercept (ETN), was low (28%) compared to that in the ETN continuation group (50 mg/week; 59%) or the ETN reduction group (25 mg/week: 69%) at 1 year in the PRESERVE study [11, 12]. However, the ENCOURAGE study showed that 54% of patients maintained clinical remission after discontinuation of ETN [13]. Certolizumab pegol (CZP) (Fab fragments against TNF fused with polyethylene glycol) was discontinued after achieving remission in early RA patients, and 42% of patients remained in remission 1 year after discontinuation in the C-OPERA study [14]. Discontinuation of bDMARDs has also been attempted for non-TNF inhibitors. For example, abatacept (ABT) was withdrawn along with concomitant methotrexate (MTX) treatment after achieving remission, and “drug-free remission” was maintained in 15% of patients in the AVERT study [15]. “Drug-free remission” was also maintained after discontinuation of the anti-interleukin (IL)-6 receptor antibody tocilizumab (TCZ) in 9% of patients in the DREAM study [16] and 14% of patients in the ACT-RAY study [17], respectively, after 1 year.

However, the results of these clinical trials cannot be compared because each clinical trial was conducted under different conditions, with different patient backgrounds (early or established RA), study designs (prospective uncontrolled trials or RCTs), protocols (bDMARD free or drug free), and failure outcomes (remission, low disease activity, or restart of bDMARDs) [1, 18]. BFR achievability may vary depending on the type of bDMARDs, which have different modes of action (TNF inhibitors (TNFi), CTLA4-Ig (ABT), and IL-6R inhibitors (IL-6Ri)). In addition to the typical classification of bDMARDs according to target molecules, TNFi can be classified into two groups: fully functional monoclonal antibodies with an immunoglobulin Fc portion (TNFi(mAb)), such as IFX, ADA, and golimumab (GLM); and soluble TNF absorption molecules (TNFi(R/P)), such as soluble TNF receptor (ETN) or Fab fragments against TNF fused with polyethylene glycol (CZP). It is possible that TNFi(mAb) (IFX, ADA, and GLM) might be more advantageous for achieving BFR than TNFi(R/P) (ETN and CZP) because anti-TNF monoclonal antibodies have higher cytotoxic activity against transmembrane TNF-expressing cells via complement-dependent and antibody-dependent cell-mediated cytotoxicity and inhibit granulomatous inflammation [19, 20]. However, this hypothesis cannot be tested by clinical trials that use totally different protocols. Therefore, the data from these clinical trials are not sufficient for determining how and when BFR can be successfully achieved in typical clinical practice.

Observational data from registries of patients in typical clinical practice could potentially contribute to answering these questions and providing real-world data that could be applied in daily clinical practice [21]. The Kansai Consortium for Well-being of Rheumatic Disease Patients (ANSWER) cohort was an observational multicenter registry of patients with RA in the Kansai district in Japan [22]. The data of patients at six universities (Kyoto University, Osaka University, Osaka Medical University, Kansai Medical University, Kobe University, and Nara Medial University) and associated hospitals were included. From 2011 to 2016, 4461 patients with RA were registered, and 52,654 serial disease activities were available from the database.

With the aforementioned in mind, the aim of this study was to determine which bDMARD is advantageous for achieving BFR and in what conditions BFR can be successfully achieved in typical clinical practice by utilizing the data from this multicenter observational cohort.

In this study, we analyzed favorable conditions for BFR achievement after bDMARD discontinuation in typical clinical practice using a multicenter RA registry in Japan (the ANSWER cohort). We found the following: BFR was achieved in 21.5% of patients at 1 year after bDMARD discontinuation in typical clinical practice; TNFi(mAb) or CTLA4-Ig was advantageous for achieving BFR compared with TNFi(R/P) or IL-6Ri; and sustained remission, Boolean remission, and glucocorticoid-free medication at the time of bDMARD discontinuation were important factors associated with a decreased risk of BFR failure. These findings will help decision-making in daily clinical practice when considering bDMARD discontinuation.

In this study, BFR was achieved in 21.5% of patients at 1 year after bDMARD discontinuation; the rate of BFR was lower than the rates reported in previous clinical trials [1, 3‐6, 14]. The low BFR achievability in this study may be because of the stricter BFR protocol used (maintaining remission at every visit) or the diverse patient backgrounds encountered in daily clinical practice (longer disease duration, fewer bDMARD-naive patients, more patients with comorbidities, etc.). The results of this study suggest that maintaining BFR after bDMARD discontinuation is more difficult in typical clinical practice than has been reported by clinical trials.

The present results showed that TNFi(mAb) is more advantageous for achieving BFR than TNFi(R/P). This is the first study to show a substantial difference between TNFi(mAb) and TNFi(R/P) with respect to the achievability of BFR in the same observational cohort. TNFi(mAb) not only binds to soluble TNF-α but also to transmembrane TNF-α, the binding of which induces outside-to-inside signaling, leading to apoptosis of the pathogenic cells bearing transmembrane TNF-α [19]. Therefore, TNFi(mAb) but not TNFi(R/P) may not only neutralize soluble TNF but also inhibit the granuloma formation of TNF-expressing cells. The latter might create favorable conditions for successful BFR achievement after TNFi(mAb) discontinuation [19]. In addition, TNF-α inhibition might expand or restore the suppressive function of regulatory T (Treg) cells that are important for the maintenance of immunological tolerance [25‐27]. Transmembrane TNF-α might be involved in this process because ADA but not ETN drives regulatory T-cell expansion via TNF-receptor 2 expressed by Treg cells [28].

CTLA4-Ig provided better survival for BFR, followed by TNFi(mAb) in the unadjusted model (Fig. 1), whereas it was almost equal to TNFi(mAb) in the adjusted model (Fig. 2). Since CTLA4-Ig targets CD4 T cells upstream of the pathological condition of RA, it might be easier to maintain a good condition even after discontinuation of bDMARDs. In fact, CTLA4-Ig reduces the number of follicular helper T cells and consequently reduces the number of switched memory B cells and autoantibodies, which may favor BFR achievement [29, 30].

It is known that IL-6 inhibition increases Treg and reduces effector T cells, which can create favorable conditions for immunological tolerance [30]. However, the BFR rate after IL-6Ri use was not as high as that associated with TNFi(mAb) or CTLA4-Ig use. Since the IL-6 signal is restored after discontinuation of TCZ, it is possible that a Treg-dominant condition might be reversed after withdrawal of IL-6Ri. Alternatively, DAS28-CRP remission might not be suited as the cutoff value considering BFR after TCZ therapy because TCZ masks CRP production and DAS28-CRP remission by TCZ can be overestimated.

Even using any bDMARDs, BFR can be achieved only in 21.5% of patients at 1 year after bDMARD discontinuation. This result suggests that immunological tolerance that could lead to long-lasting BFR has not yet been established after current bDMARD therapies, and there are still unmet needs for an RA “cure.”

This study demonstrated that BFR can be successfully achieved after achieving sustained and strict remission at the time of bDMARD discontinuation (Table 3). This result is mostly consistent with previous recommendations and the consensus for bDMARD discontinuation [1]. The importance of minimal disease activity for > 6 months has been indicated in clinical trials [1]. The importance of achieving more stringent remission than DAS28 remission before withdrawing bDMARDs has been indicated by previous studies; for example, using a lower DAS28-CRP cutoff value or the absence of Doppler signals on an ultrasonogram [5, 8, 31]. This study showed that sustained and stringent remission at the time of bDMARD discontinuation is important for successfully achieving BFR, not only in clinical trials but also in real-world clinical practice.

This study also showed that no glucocorticoid use at the time of bDMARD discontinuation is important for achieving BFR. The importance of tapering the glucocorticoid dose before bDMARD discontinuation has been suggested in the EULAR recommendations, which state the following: “If a patient is in persistent remission after having tapered glucocorticoids, one can consider tapering bDMARDs” [32]. However, the clinical evidence supporting this recommendation is insufficient. The present study strongly suggests that the glucocorticoid dose should first be tapered when considering bDMARD discontinuation because the use of glucocorticoids at the time of bDMARD discontinuation was associated with failure of BFR in the real-world observational cohort (Table 3).

The present study has several limitations. First, the number of patients was small. Even including a multicenter cohort, serial disease activity at every visit was available only in limited cases. Therefore, the present results need to be confirmed by future studies including a larger number of participants. Second, due to the small number of study participants, all patients who met the inclusion and exclusion criteria were included regardless of the reasons for drug discontinuation. The reason for drug discontinuation may have affected the BFR survival time, although we adjusted for discontinuation due to remission in multivariate analysis. Third, since this study had a retrospective design and used an observational cohort from daily clinical practice, the unknown background factors (e.g., the use of csDMARDs other than MTX or disease status at the initiation of bDMARDs) may have affected the results. Finally, the radiographic progression of joint destruction was not evaluated in this study. Notably, bDMARDs have strong protective activity against bone destruction; therefore, radiographic destruction can be inhibited by bDMARD use, even though disease activity cannot be fully controlled [33, 34]. Future studies should address whether radiographic remission can be maintained if strict BFR is maintained after bDMARD discontinuation.

This study investigated the real-world conditions affecting BFR achievement in patients with RA. Although BFR is difficult to achieve in typical clinical practice, after strained and strict remission without glucocorticoid use, bDMARDs can be successfully withdrawn while retaining remission after discontinuation. Furthermore, TNFi(mAb) or CTLA4-Ig may be more advantageous for achieving BFR than TNFi(R/P) or IL-6Ri.