Discussion

Overview of Key Findings

We conducted a systematic review and network meta-analysis (NWMA) to update the 2012 review of the comparative effectiveness of drug therapies for rheumatoid arthritis (RA);¹ in this report we focused solely on early RA in adults (within 1 year of diagnosis). The objective was to evaluate the comparative effectiveness and harms of monotherapies, combination therapies, and different treatment strategies. These therapies include several categories of drugs: (1) corticosteroids; (2) two classes of disease-modifying antirheumatic drugs (DMARDs)— conventional synthetic (cs) and targeted synthetic (ts) DMARDS; (3) two classes of biologic DMARDs—tumor necrosis factor (TNF) and non-TNF biologics; and (4) biosimilars. The drug classes and constituent drugs and their abbreviations/acronyms can be found in Table 1.

A total of 41 randomized controlled trials (RCTs) and 8 observational studies comprised the evidence base of this updated review. Table 12 summarizes our findings about benefits and harms and gives the strength of evidence grades (SOE, in bold) for three Key Questions (KQs) addressed by this report. Studies (n=2) or study outcomes rated high risk of bias were excluded from analyses and used only in sensitivity analyses for the network meta-analysis. Given that there were sparse data available about subgroups (KQ4), we present this information after the table. SOE grades reflect the level of certainty about conclusions drawn from findings; they are high, moderate, low, or insufficient. Detailed assessment of the SOE for KQ outcomes can be found in Appendix E.

Of specific interest are the following outcomes related to efficacy—disease activity, radiographic changes, functional capacity, and remission—and the following outcomes related to harms—overall discontinuations, discontinuations attributable to adverse effects and serious adverse events. The study population included patients with moderate to high disease activity.

Existing comparative evidence for our review was diverse. It included comparisons of monotherapies, combination therapies, triple therapy (methotrexate [MTX], sulfasalazine [SSZ], hydroxychloroquine [HCQ]), and treatment strategies. Additionally, the drug classes spanned corticosteroids, csDMARDs, tsDMARDs, TNF biologic DMARDs, and non-TNF biologic DMARDs. No studies on the use of biosimilar DMARD agents in early RA were included in this report because they did not fit the inclusion criteria.

For corticosteroids and csDMARDs, the evidence allowed us to draw some conclusions for early RA. Corticosteroids, in combination with MTX, led to higher remission rates than MTX alone for MTX naïve patients with moderate to severe disease; results were mixed, however, for radiographic progression, health-related quality of life (HRQOL), and functional capacity. There were no significant differences in serious adverse events and discontinuations attributable to adverse events between these two treatment regimens. The corticosteroids used were heterogeneous and included varying doses of prednisone (PRED), prednisolone, and methylprednisolone regimens.

Studies of csDMARD therapies mainly examined SSZ and MTX. Comparisons of combination therapy with monotherapy found no differences in disease activity, functional capacity, serious adverse events, or discontinuations attributable to adverse events.

Although several biologic agents are available, the head-to-head evidence remains limited. Moderate strength of evidence supports combination therapy of adalimumab (ADA) plus MTX versus ADA only for several outcomes; specifically, ADA plus MTX led to higher American College of Rheumatology (ACR) response rates, higher remission rates, and less radiographic progression than ADA monotherapy. There were no significant differences in serious adverse events or discontinuations attributed to adverse events between these two medication regimens. Our NWMA also found significantly higher ACR50 response rates and less radiographic progression following use of ADA plus MTX versus ADA monotherapy. The data showed that both TNF biologics (ADA, etanercept [ETN], or infliximab [IFX]), but not non-TNF biologics (abatacept [ABA] or tocilizumab [TCZ]), in combination with MTX have higher ACR50 treatment response than biologic monotherapy. The results of comparative NWMA for overall discontinuation and discontinuation attributed to adverse events had confidence intervals that were too wide to support firm conclusions.

The evidence comparing TNF biologics (ADA, certolizumab pegol [CZP], ETN, or IFX) plus MTX with MTX monotherapy generally showed higher remission rates, better functional capacity, and less radiographic progression for the combination medications. Serious adverse events or discontinuations did not differ significantly. Similar findings were also noted for non-TNF biologics (ABA, rituximab [RIT], or TCZ) in combination with MTX. Head-to-head evidence for biologics is limited to one trial,⁸ which found no significant differences in disease activity and remission with RIT compared with TNF biologics (ADA or ETN).

Combination therapies with csDMARD triple therapy (MTX plus SSZ plus HCQ) compared with TNF biologics (either ADA or IFX) plus MTX found no significant differences in remission or radiographic changes. Rates of adverse events did not differ. In terms of treatment strategies, the BeSt study^79–91 assessed several treatment strategies for early RA; the investigators included sequential monotherapy, step-up combination therapy, combination therapy with tapered PRED, and combination therapy with IFX. Over the long term (i.e., 10 years), radiographic progression, remission, and functional capacity did not differ across the arms of the trial.

Subpopulation data were limited to post hoc analyses. For most comparisons, we did not find eligible evidence on the benefits and harms among subpopulations.

Findings in Relationship to What Is Already Known

We conducted a systematic review and NWMA to update the 2012 review of the comparative effectiveness of drug therapies for RA;¹ in this report we focused solely on early RA in adults (within 1 year of diagnosis). All of the early RA studies included patients with moderate to high disease activity. In a clinical setting, patients with early RA may present with varying levels of severity. Also, the studies did not consistently parse out which patients had tried one or more therapies and which ones were treatment naïve.

Our results go further than treatment recommendations for early RA from the ACR and the European League against Rheumatism (EULAR) and support additional therapies for patients with moderate to high levels of disease activity. The ACR and EULAR task force both support a treat-to-target approach over a nontargeted approach with the goal of achieving remission or low disease activity.^{42, 43} The BeST and FIN-RACo trials used a treat-to-target approach, and their results support the ACR and EULAR recommendations in this respect.^{22, 79}

The ACR guidelines recommend csDMARD monotherapy (MTX preferred) instead of double or triple csDMARD therapy in patients who have never taken a csDMARD.⁴² If disease activity remains moderate or high, despite csDMARD monotherapy, then the ACR recommends double or triple csDMARD therapy or a TNF or non-TNF biologic DMARD (with or without MTX). Our evidence was insufficient to support one DMARD over another (e.g., csDMARDs, biologic DMARDs). However, we found that when biologics were used in combination with MTX therapy, patients achieved lower disease activity, higher functional capacity, and higher remission rates than with monotherapy alone. The difference between the results of our findings and the ACR guidelines may be due to a few reasons. First, all of our studies included patients with moderate to high disease activity at baseline. Patients with early RA in a clinical setting may present with less disease severity and prior history with MTX could vary. Additionally, this report assessed comparative effectiveness based on current available evidence and included secondary longer time points when available. Clinical practice guidelines use systematic reviews as evidence and if evidence is not enough they may consider other resources. The ACR based their recommendations on a consideration of the balance of relative benefits and harms of the treatment options as well as expert opinion and preferences.

Although the evidence for the effectiveness of MTX plus biologics in early RA is favorable, it is not the standard of care for a number of reasons. First, some data indicate that certain patients will do well on MTX monotherapy, but no information is available about how to identify or predict these patients. Second, many insurers require MTX failure as a prerequisite to add a biologic (probably based on the effectiveness of MTX). Third, patients may be wary of a combination therapy approach in early disease (e.g., cost, side effects, injections). Additionally, patients must balance the potentially higher efficacy of multiple drugs with the burden and potential for increased risk.

The EULAR task force advocates starting with csDMARDs as first-line therapy in the absence of poor prognostic factors (e.g., high disease activity, early joint damage, autoantibody positivity) in early RA.⁴³ When poor prognostic factors are present, the task force advocates for adding a TNF or non-TNF biologic to a csDMARD. This guideline group regards all biologic DMARDs as similarly effective and safe after csDMARD monotherapy failure. Our findings harmonize with EULAR’s guidelines recommending combination therapy with a biologic as first-line therapy for patients with poor prognostic factors. The evidence we found comparing combinations of biologics and MTX with either biologic or MTX monotherapies (N=10 studies) in patients with early RA and poor prognostic factors reported that patients receiving combination therapies achieved higher remission rates.^{12–15, 17, 32–34, 37, 41} However, we had no available studies that specifically examined the effect of therapies in patients with early RA and less severe disease activity to patients with early RA plus poor prognostic factors.

Applicability

Although we derived our evidence primarily from RCTs that typically enrolled a discrete population and were conducted under ideal situations, the findings from observational and efficacy trials were generally consistent. However, the observational and noncontrolled studies reported higher discontinuation rates. For example, one observational study of MTX versus MTX plus SSZ in a SSZ-resistant population had overall discontinuation rates ranging from 33.9 percent to 50.0 percent at 1 year due to either side effects or lack of response.²⁶ A second observational study of MTX versus SSZ reported similar reasons for discontinuation.²⁸ Discontinuation rates from clinical trials were generally lower than 20 percent. The higher discontinuation rates in observational studies may reflect real-world settings as compared with the tighter adherence in a controlled clinical trial. The observational studies in this report describing harms were rated as medium to high risk of bias. Higher quality observational studies may affect the estimates of these results.

The range of mean (or median) disease durations across all 49 included studies was 2 weeks to 12 months. All our included studies enrolled patients with moderate to high disease activity at baseline as measured with mean or median Disease Activity Scale (DAS) 28 scores, ranging broadly from 3.4 to 7.1 (DAS ranges from 0 to 10; 3.2 is a threshold for low disease activity; more than 5.1 is considered high disease activity). More than one-half of the patient population were women; the mean age range was 46 to 64 years. Study durations ranged from 6 months to 15 years.

In addition, trials comparing corticosteroids used varying doses and tapering strategies. Similarly, MTX dosing ranged from 7.5 mg per week to 25 mg per week. This degree of heterogeneity did not allow for suitable evidence comparison, but it may be typical of common clinical practices.

As stated previously, subpopulation studies of differences in benefits and harms were mostly lacking. The data were sparse for any comparative differences in serious infections and malignancies in this early RA population. The evidence was limited to post hoc subgroup analyses from studies comparing TNF biologics with csDMARDs.

Contextual Questions

During the review process, we flagged studies for their relevance to the contextual questions during the review process and we also supplemented this evidence base with a targeted literature search.

Limitations

Our review update has some limitations. No consensus exists on the definition of early RA. Moreover, criteria used in the literature for defining populations with early RA are variable. A recent task force of RA experts recommended defining early RA as no more than 1 year of diagnosed disease duration.⁴³ For this review, we defined populations with early RA as having a diagnosed disease duration limited to 1 year or less and included mixed population studies if >50 percent of the study populations had an early RA diagnosis. It is possible that patients described in this way may have longer disease (symptoms).

Additional evidence on treatment comparisons might be gained by expanding the definition to 2 years. However, requiring a diagnosed disease duration of 1 year or less is in line with current clinical practice. In reviewing our literature, we identified but excluded 7 studies (reported in 10 articles) of adults with a duration of RA between 1 and 2 years from diagnosis. On brief review of the 7 studies, findings did not differ from the current report.

For several of the studies evaluating corticosteroids, drug dosing was heterogeneous. This factor limited our ability to draw conclusions from comparisons of these agents. Similarly, in csDMARD comparisons, MTX dosing varied from 7.5 mg to 25 mg weekly.

Few data were available about subgroups that are of interest to this field; typically, we found only limited data on age. Evidence was limited for the tsDMARD class and nonexistent for biosimilars in the early RA population. Although existing evidence of biologics in combination with MTX shows that this regimen can improve disease activity, we do not know whether starting treatment with a biologic rather than a csDMARD improves long-term prognosis of RA.

Because of a lack of head-to-head trials, we often had to rely on results from the NWMAs to estimate the comparative effectiveness of interventions of interest for treating patients with early RA. Network (sometimes referred to as indirect or mixed) meta-analyses are an important analytic tool in the absence of direct head-to-head evidence, but they also have limitations. The “transitivity assumption” relies on the premise that any patient in the network would be equally likely to have received any of the treatments in the network. It is difficult to assess this assumption when no direct head-to-head studies are available and estimates are based exclusively on indirect comparisons. In the case of our NWMAs, most comparisons were based on a “star network” with MTX as the common comparator. A star network indicates a dearth of head-to-head studies directly comparing interventions. Most effect estimates, therefore, were derived from indirect comparisons rather than mixed treatment comparisons. Although we carefully assessed the clinical heterogeneity of all trials included in the network meta-analyses to ensure that they were as homogenous as possible, we were not able to statistically assess the assumption of homogeneity (and transitivity) for most comparisons. Furthermore, NWMAs often yield estimates with wide confidence intervals that encompass clinically relevant benefits or harms for both drugs (or combinational therapies) that are being compared. Such inconclusive results should not be misinterpreted as evidence for no difference in benefits or harms. In general, these limitations are reflected in the strength of evidence ratings.

Research Needs

Future research should help clinicians and researchers draw stronger conclusions on the comparative effectiveness and harms of medications for patients with early RA. Multiple established therapies exist for early RA, but comparative evidence is badly needed. Studies comparing therapy options in patients diagnosed with early RA who have different degrees of disease activity or poor prognostic factors would be helpful in the clinical setting.

Also, at least some, or perhaps many therapies for early RA may be initially effective, but longer-term effects have not been well studied. Studies with longer treatment periods and followup of 5 years or longer would provide better information on adherence and adverse events. Registry data have the potential to include real-world populations with data on long-term effects and follow-up. They would also yield insights as to whether starting with a biologic improves long-term prognosis of RA.

Most studies that we used for this review evaluated csDMARD and biologic medications. FDA has approved several biosimilars, but because they have not been studied specifically among early RA patients, we could not include any studies of them in this review. Per FDA guidance, efficacy outcomes for the biosimilars are based on extrapolation from studies in several indications and may not be specifically studied in RA, either early or late. Four^{7, 18, 29, 41} of 39 studies reporting radiographic outcomes described MRI findings. This is an evolving technology.

Analyses of subpopulations based on age and coexisting medical conditions (hepatitis C, congestive heart failure, diabetes, and cancer) would also be helpful for clinicians and patients newly diagnosed with RA. Currently, treatment selection based on benefits and harms is difficult in these populations. Additionally, patient-centered research is needed with appropriate use of patient-reported outcomes and other patient-generated health data so that results are truly reflective of patient preferences and desires.

Conclusions

For patients with early RA, qualitative and network meta-analyses suggest that the combination of MTX with TNF or non-TNF biologics improves disease activity and remission when compared with monotherapy with a biologic or csDMARD. This comprehensive review found similar adverse events and discontinuation rates for csDMARDs, TNF biologics, and non-TNF biologics in studies ranging in length from 6 months to 15 years.