Introduction

Rheumatoid Arthritis (RA) is a chronic inflammatory systemic disease affecting approximately 1% of the population with a higher frequency in women than in men [1]. Inflammation in RA primarily affects the small joints and is characterized by pain and swelling, thereby leading to chronic progressive joint destruction causing a decline in quality of life, physical function, and working ability. Cytokines such as tumor necrosis factor α (TNF) and interleukin 6 (IL-6) play a specific role in the inflammatory processes in RA [2]. As inflammation drives symptoms (pain and swelling of joints) and sequentially damage (loss of function leading to work disability), suppression of inflammation is the main target in RA. As a result of its systemic nature, the individual burden of the disease is not necessarily limited to the joints but also results from fatigue, depression, osteoporosis, and an increased risk for cardiovascular disease [3].

Management of RA has dramatically improved in the last few decades: In former times, the disease was diagnosed late and relief of symptoms or prevention of adverse events was the main goal of therapy. For much of the last century, available compounds for therapy in RA had limited efficacy but significant toxicity. Management and therapy of RA has been revolutionized by the development and approval of the first biological disease-modifying antirheumatic drugs (bDMARDs) targeting TNF at the end of the last century. Today, numerous efficacious agents with different modes of action are available and achievement of clinical remission or, at least, low disease activity is the target of therapy. Earlier treatment in the course of RA leads to a higher clinical efficacy [4] and, consequently, the 2010 American College of Rheumatology/European League Against Rheumatism (ACR/EULAR) classification criteria focus on enabling diagnosis of RA in early stages of the disease [5]. Within this phase of the disease, active inflammation instead of irretrievable structural damage is the key driver of symptoms and disability. Early therapeutic interventions aiming at a defined goal of therapy (treat to target) are supposed to halt inflammation, improving symptoms and signs, and preserving structural integrity of the joints in RA [6].

Up to now, biological DMARDs approved for therapy in RA include agents with five different modes of action: TNF inhibition, T cell co-stimulation blockade, IL-6 receptor inhibition, B cell depletion, and interleukin 1 inhibition. Furthermore, targeted synthetic DMARDs (tsDMARDs) inhibiting Janus kinase (JAK) and biosimilars also are approved for RA. Currently, five TNF inhibitors (adalimumab, certolizumab pegol, etanercept, golimumab, and infliximab) and two tsDMARDs inhibiting JAK (tofacitinib and baricitinib) are approved for therapy in RA. The present review focuses on bDMARDs and tsDMARDs regarding similarities and possible drug-specific advantages in the treatment of RA. Furthermore, compounds not yet approved in RA and biosimilars are discussed. Following the ACR and EULAR recommendations, specific treatment of the disease will be discussed with respect to the first biologic intervention and management after failure of the first bDMARD, respectively. This article is based on previous studies and does not involve any new studies on human or animal subjects performed by the authors.

Management of MTX Insufficient Responders

Efficacy of Approved Biological DMARDs

According to the recently updated EULAR recommendations on RA [7], therapy with conventional synthetic DMARDs (csDMARDs) should be started immediately after having diagnosed RA and methotrexate (MTX) should be part of the first treatment strategy. Using MTX, however, a large proportion of patients do not achieve remission or low disease activity [8, 9]. In general, bDMARDs or tsDMARDs should be considered if the first csDMARD does not achieve treatment target in the presence of negative prognostic factors or if response to two csDMARDs is insufficient in the absence of negative prognostic factors.

In the setting of MTX-IR (MTX incomplete responders), neither the EULAR nor the ACR recommendations favor the use of one specific bDMARD or tsDMARD or suggest a certain sequence of its use [7, 10]: This strategy of recommending all agents without hierarchical positioning is a novelty compared to the 2013 version of the EULAR recommendations [11] and is well in line with the published evidence (Table 1): First, head-to-head studies in MTX-IR directly comparing bDMARDs with different modes of action in combination with MTX are scarce. The Ample study investigated the clinical efficacy of abatacept versus adalimumab in an MTX-IR RA cohort. Clinical efficacy (ACR 20 response rate 64.8% in the abatacept and 63.4% in the adalimumab groups) and inhibition of radiographic progression were similar within these two agents [12]. Tofacitinib was as effective as adalimumab in a randomized study (ACR 20 response rates 51.5%, 52.6%, and 47.2% among patients receiving 5 or 10 mg of tofacitinib or those receiving adalimumab) [13]. Second, studies directly comparing TNF inhibitors are still lacking with the exception of the EXXELARATE trial investigating adalimumab and certolizumab pegol in MTX-IR. Results proved that the efficacy of certolizumab pegol was not significantly different to that of adalimumab, both in combination with MTX (ACR 20 response rates 65% for certolizumab pegol and 67% for adalimumab) [14]. Third, indirect comparisons strongly point to a similar efficacy within all bDMARDs when used in combination with MTX [15].

Table 1 Results of randomized controlled trials (RCTs) of biologicals and small molecule JAK inhibitors compared to placebo in MTX insufficient responders

In general, bDMARDs are used in combination with MTX. However, more than one-third of patients are intolerant to MTX [16] and adherence is often poor, especially when administered orally [17]. Consequently, about 30% of patients in clinical practice are treated with bDMARDs in monotherapy [18]. Next to MTX, various csDMARDs have been analyzed for their efficacy in RA patients. However, the combination of these other csDMARDs with bDMARDs has not been tested extensively: In the REACT [19] and the oral Sync studies [20], the combination of adalimumab or tofacitinib, respectively, with different csDMARDs including sulfasalazine, leflunomide, antimalarials, azathioprine, and gold demonstrated proven efficacy. Furthermore, the combination of TNF inhibitors with leflunomide was as effective and equally well tolerated as TNF inhibitors plus MTX [21]. Similarly, disease activity and functional disability did not differ significantly between tocilizumab plus MTX versus tocilizumab plus leflunomide [22].

Special focus has to be given to the clinically relevant issue of using bDMARDs or tsDMARDs in monotherapy: Tocilizumab was the first biological agent to show statistically significant clinical efficacy superior to MTX when used in monotherapy, although approximately two-thirds of patients in the MTX group reached and maintained a 20 mg MTX/week dose by week 8 [23]. The randomized controlled ADACTA trial evaluated adalimumab versus tocilizumab in RA patients with an insufficient response or intolerance to MTX—both agents were used in monotherapy. This superiority study showed a significantly greater clinical improvement with tocilizumab compared to adalimumab (ACR 20 response rates 65.0% vs. 49.4%) [24]. In an open-label study, tocilizumab was similarly effective, when used as monotherapy or in combination with csDMARDs [25]. The ACT-RAY was a double-blind study to compare adding tocilizumab versus switching to tocilizumab monotherapy in MTX-IR. In this study, no clinically relevant superiority of the addition of tocilizumab to MTX over the switch to tocilizumab monotherapy was proven, but there was a modest difference in the percentage of patients with DAS28 remission and in the inhibition of radiographic progression favoring the addition strategy [26, 27]. Adding tocilizumab to MTX in patients with active disease despite MTX was clinically and radiographically superior compared to switching from MTX to tocilizumab in another two randomized trials [28, 29]. Regarding monotherapy with tsDMARDs, ACR 20 response rates were similar for baricitinib plus placebo compared to baricitinib plus MTX (nearly 80% in both groups) in a recently published RCT [30], and the effect of tofacitinib was irrespective of the dosage of background MTX [31]. To sum up, the use of tocilizumab or of tsDMARDs is a reasonable evidence-proven strategy in case of monotherapy in RA, especially in patients who do not tolerate or cannot be treated with MTX.

Safety Issues Regarding Approved Biological Compounds in RA

Compared to csDMARDs, biological drugs in RA are associated with an increase in the number of serious infections of six per 1000 patients treated each year [32]. Increasing age, comorbidity, glucocorticoid use, and previous history of serious infections were significant predictors of future infections in different databases, including the German RABBIT register. The overall relative risk of treatment with TNF inhibitors compared with that of csDMARDs was 1.8 in the RABBIT cohort, meaning that the absolute risk increase depends on baseline risk in different groups of patients [33]. For daily practice, an online calculator has been developed [34]. The rates of serious infections in RCTs of newer biologicals appear to be in general consistent with rates seen in the anti-TNF registers of between two and six per 100 patient years of follow-up [35]. The ACR 2015 guideline for the treatment of RA has addressed the issue of safety [10]. Supported by low-level evidence, the ACR recommends to use csDMARDs in combination over TNF inhibitors in patients with previous serious infections and to use abatacept over TNF inhibitors within these high-risk patients, mainly driven by the results of the ATTEST trial, demonstrating less serious infections with a lower need of hospitalizations during therapy with abatacept compared to infliximab [36, 37]. These recommendations are conditional, meaning that the special recommendation is applicable to the majority of patients, but some may not want to follow the recommendation.

After approval of TNF inhibitors, reports of new onset or deterioration of heart failure were reported for infliximab dosed at 10 mg/kg [38]. This could not be reproduced with etanercept [39]. In parallel, a recent review shows that the rate of heart failure is not increased in patients receiving TNF inhibitors. Furthermore, risk of symptomatic congestive heart failure was not increased in the high risk group of patients with established heart failure during therapy with TNF inhibitors [40].

When initiating a biological DMARD, special focus relates to the increased risk of opportunistic infections [41] and to check for latent tuberculosis. Risk for reactivation of tuberculosis is higher in patients receiving anti-TNF monoclonal antibodies compared to etanercept. One reason therefore might be the importance of membrane-bound TNF in protecting against tuberculosis, which is neutralized by the TNF antibody but not by the TNF receptor fusion molecule [42, 43]. In contrast, increased risk of reactivation was not seen during therapy with rituximab in RA. In mice, abatacept did not impair the ability to control a chronic Mycobacterium tuberculosis infection [44]. Tuberculosis (TB) was rarely seen during therapy with tofacitinib and it was speculated that some of the observed TB cases were instances of newly acquired infection during the trial, given that nearly all the cases occurred in regions of high TB endemicity where exposure would be more likely [45]. No increased risk for tuberculosis reactivation was seen with tocilizumab in RCTs [46]. In general, prior to treatment with all approved bDMARDs in MTX-IR, the well-known screening algorithm should be applied [10].

With respect to vaccination, treatment with rituximab and with MTX in monotherapy is associated with impaired vaccine responses, and temporary MTX discontinuation improves the immunogenicity of seasonal influenza vaccination in patients with RA [47, 48]. In contrast, tofacitinib had less impact on vaccination responses [49], whereas TNF inhibitors and tocilizumab did not reduce vaccination response [47, 50].

A special safety issue during therapy with tocilizumab is the risk of lower intestinal perforation (LIP). Recently, real-life data from Europe found a significantly higher rate of LIP in tocilizumab-treated patients compared to patients treated with csDMARDs and other bDMARDs. Importantly, risk for LIP was not attenuated after adjustment for concomitant glucocorticoid use in Cox regression analysis [51]. In our opinion, a medical history positive for diverticulitis contradicts the use of tocilizumab, knowing that most cases with LIP did not have a history of diverticulitis. Patients should be informed about this risk and the fact that negative markers of inflammation cannot be interpreted during therapy with tocilizumab.

Regarding malignancies, cancer, lymphoma, melanoma, and non-melanoma skin cancer do not occur more frequently in patients on TNF inhibitors compared to patients on conventional DMARDs [52, 53]. However, rituximab is the first choice in the case of certain comorbidities, e.g., concomitant multiple sclerosis or past or present lymphoproliferative disorders.

As 30–40% of RA patients show an inadequate response to bDMARDs, trials investigating the combination of two bDMARDs with different modes of action have been done. However, the combination of abatacept plus TNF inhibitors [54], of an interleukin-1 receptor antagonist plus TNF inhibitors [55], and of rituximab plus TNF inhibitors [56] demonstrated a lack of added benefit and at least the first two combinations significantly increased the rates of adverse events including serious infections.

Management in TNF Insufficient Responders (TNF-IR)

There is still ambiguity how to manage RA patients with an inadequate response to the first TNF inhibitor. Keeping in mind the growing number of modes of action and consequently bDMARDs, this is a central question. Observational data indicate an advantage in changing the mode of action in TNF-IR [57, 58]. The ACR consequently recommends the use of a non-TNF biological in case of an insufficient response to the first TNF inhibitor [10]. In primary non-responders to the first TNF inhibitor, data from one randomized trial and several observational studies point to less clinical response after switching to a second TNF inhibitor in primary TNF inhibitor non-responders compared to secondary non-responders or patients stopping the first TNF inhibitor because of adverse events [59,60,61,62]. Within primary non-responders, disease might be mediated by cytokines other than TNF, and consequently these patients could experience greater benefit from compounds using modes of action other than blocking TNF. In contrast, EULAR recommends to employ a second TNF inhibitor or an agent with a different mode of action without a hierarchical ranking, if one TNF inhibitor has failed [7]. Randomized controlled head-to-head studies once more are scarce, with the exception of the EXXELARATE study corroborating this recommendation: About 60% of patients switching to a second TNF inhibitor after primary failure to the first TNF inhibitor achieved a DAS28 reduction of at least 1.2 points [14]. Recommendation differs in patients failing TNF inhibition in monotherapy: As stated above, the use of tocilizumab or tsDMARDs is our recommended strategy in TNF-IR on monotherapy with bDMARDs.

Patients with secondary inefficacy to TNF inhibitors may have lost response because of the development of antidrug antibodies; these patients would therefore be expected to exhibit a clinically relevant response to an antigenically distinct treatment. Data from a randomized placebo-controlled trial demonstrated efficacy and safety of certolizumab in RA patients with secondary inadequate response or intolerance to the previous TNF inhibitors [63]. Importantly, patients developing antibodies to a first TNF inhibitor are at higher risk for developing antibodies to a subsequently employed TNF inhibitor. Therefore, immunogenicity might be the cause leading to the lower clinical efficacy of the second TNF inhibitor observed in TNF-IR [57, 64]. Within this context, changing the mode of action in patients failing treatment with TNF inhibitors is worth considering, especially as abatacept [65], golimumab [61], rituximab [66,67,68], tocilizumab [69, 70], tofacitinib [71], and baricitinib [72] demonstrated clinical efficacy in patients with a previously inadequate response to at least one TNF inhibitor (Table 2). However, direct head-to-head comparisons using compounds with different modes of action in TNF-IR are still lacking. Moreover, we still do not have any data exploring the efficacy and safety of TNF inhibitors used after non-TNF-inhibiting bDMARDs have failed, of a second IL-6 receptor inhibitor after tocilizumab has failed, and of a second JAK inhibitor after the first has failed. In summary, selection of the next bDMARD in TNF-IR is still undetermined, whereas EULAR and ACR both recommend switching to a non-TNF biological if a second TNF inhibitor fails.

Table 2 Results of randomized controlled trials of biologicals and small molecule JAK inhibitors compared to placebo in TNF insufficient responders

Despite observational data and indirect comparisons pointing to a lower risk of infection during treatment with rituximab [73, 74], there was no consensus for making recommendations regarding the use of the CD20 antibody over TNF inhibitors in the setting of previous serious infections. A pooled analysis of safety data reported a rate of serious infections with rituximab that was comparable to that with MTX and reactivation of tuberculosis does not occur more frequently in RA treated with rituximab [75].

Furthermore, the frequency of serious infections was similar in TNF-IR randomized to abatacept compared to the placebo group (10.5% and 11.3%) [65]. In the few head-to-head trials, incidence of serious infections was not statistically significantly different between two TNF inhibitors (EXXELERATE, 2.2% in certolizumab pegol vs. 2.0% in adalimumab treated patients) [14], and a TNF inhibitor compared to tocilizumab and baricitinib [24, 76], whereas abatacept had a more acceptable safety profile than infliximab as already mentioned [36]. In contrast, a recent meta-analysis suggests an increased risk of adverse events for certolizumab pegol during the first months of treatment [77].

In general, RCTs in TNF-IR ask for a washout phase of at least 4 weeks after the last injection or infusion of TNF inhibitors. However, the need for a washout phase has been rebutted in the ARRIVE, ACT-SURE, and EXXELARATE trials [14, 69, 78]. In our clinic, we have generally abandoned the use of a washout period. We directly start the new compound at the date scheduled for the next infusion or injection of the formerly used agent in TNF-IR. In our hands, no increase of adverse events was found (unpublished data).

Practicability

In 2017, treating physicians have the possibility to choose not only among biological compounds with different modes of action but also with different routes of administration. Abatacept and tocilizumab are approved for use via subcutaneous injection or intravenous access. Head-to-head studies comparing the intravenous and subcutaneous route of administration of these two agents did not show any difference in clinical efficacy and safety, except that injection site reactions were more common with the subcutaneous access [24, 79]. Abatacept has demonstrated that switching from weekly subcutaneous to intravenous abatacept and back is effective and safe and may be used to bridge 4 weeks, e.g., to cover a vacation [80]. In real life, about 50% of patients with subcutaneous injections do not feel confident about injecting themselves [81]. Data investigating compliance with the orally administered JAK inhibitors are lacking. In our clinic, we observed a sinusoidal course in disease activity among patients treated with tofacitinib in real life, raising the issue of poor compliance. This might be the case especially in phases of good control of inflammation (manuscript under preparation).

Compounds Not Yet Approved in the Treatment of RA

Sarilumab inhibits IL-6-mediated signalling by specifically binding to both soluble and membrane-bound IL-6 receptors. Administered subcutaneously every 2 weeks, sarilumab significantly improved disease activity in MTX-IR and TNF-IR [82, 83]. Moreover, sarilumab was superior to adalimumab (both in monotherapy) by improving clinical disease activity in RA patients unable to continue MTX (ACR 20 response rates 71.1% in sarilumab-treated patients vs. 58.4% in adalimumab-treated patients) [84]. Similarly, HAQ-DI score and patient-reported outcomes improved at a statistically significant greater extent in the sarilumab group compared to the adalimumab group. The incidence of infections was similar between the two groups. Treatment with the IL-12/23 antibody ustekinumab and the IL-23 antibody guselkumab did not significantly reduce disease activity in rheumatoid arthritis patients with an inadequate response to MTX (ACR 20 response 53.6% and 41.3% vs. 40.0% in placebo) [85]. The data for the compounds secukinumab and brodalumab blocking the cytokines IL-17 and IL-23 in RA are far disappointing [86, 87], whereas ixekizumab improved RA signs and symptoms in patients naive to or with an inadequate response to TNF inhibitors [88]. It is therefore questionable if IL-17 and IL-23 blocking agents will be competitive against the already approved and available compounds in RA.

Biosimilars

EULAR mentions that biosimilars approved by the European Medical Agency (EMA) or the US Food and Drug Administration (FDA) have similar efficacy and safety as the respective biological originator, and should be preferred if they are indeed appreciably cheaper than the originator or other biological compounds [7]. Biosimilar infliximab is already widely available, whereas biosimilar etanercept and biosimilar rituximab are approved merely by the FDA at the moment, but the EMA recommended approval recently [89]. Furthermore, a substantial pipeline of biosimilars is in development, including biosimilar adalimumab, which is recommended for approval [90].

Tapering and Withdrawal

If a patient is in persistent remission after having tapered steroids, tapering bDMARDs can be considered [7]. Reducing from 50 to 25 mg etanercept injected once per week or increasing the time interval between the etanercept or adalimumab injections was non-inferior to usual continued medication with respect to the preservation of clinical remission or low disease activity and the rate of clinically relevant flares [91, 92]. In contrast, abrupt cessation of bDMARD therapy led to flares in many if not most of the patients, not all of these regaining their former state of remission or low disease activity after restart of bDMARDs. In the ACT-Ray study, 50% of patients discontinued tocilizumab following sustained clinical remission after 1 year. Subsequently, 84% of these patients experienced a flare-up with recurrent response to the reintroduced drug [93]. In the CERTAIN double-blind randomized trial, patients who achieved remission during treatment with certolizumab pegol and csDMARD stopped treatment. Only three of the 17 prior certolizumab pegol patients maintained remission until week 52 [94]. In line with EULAR recommendations, in cases of persistent remission, tapering the csDMARD could be considered. However, this recommendation is a matter of debate, as many rheumatologists never would stop a csDMARD and leave patients without any disease-modifying therapy [7].

Conclusions

In general, important steps for controlling the progression of the disease have to be taken long before choosing a bDMARD: Early identification of RA represents the crucial step. Therefore we have to emphasize the importance of awareness for RA within the population and general practitioners. Perhaps even more challenging is the next step: Patients suspected of suffering from RA have to be seen by rheumatologists quickly, and if diagnosed, treatment with preferentially MTX should be started immediately. Initiating therapy within 3 months of disease onset played the most important role in achieving remission [95]. Even the worst prognostic factors lost significance when patients were treated aggressively and promptly in the BeST trial [96]. Treatment decisions have to be based on measures of disease activity and should be aimed at reaching remission or low disease activity, with respect to patient factors. Frequent follow-up visits especially in active disease and the overarching principle of treat-to-target are particularly important for successful therapy of RA. However, predictors for differential response to the different bDMARDs and tsDMARDs are missing. We still do not know which patient will best respond to which drug in which stage of disease and pretreatment. To provide better justification for the decision to prefer a specific bDMARD in different situations, future randomized head-to-head studies are urgently required.