Systematic review and network meta-analysis: effect of biologics on radiographic progression in rheumatoid arthritis

As shown in Figure 1, of the 2449 nonduplicate records screened, 46 studies met the PICO eligibility criteria and were included in the systematic review. Of these 46 studies, 16 studies were included in evidence synthesis: two abatacept studies [36,37], three adalimumab studies [38–40], three certolizumab studies [41–43], one etanercept study [44], one golimumab study [45], two infliximab studies [46,47], two rituximab studies [32,48] and two tocilizumab studies [49,50]. Common reasons for exclusion from analysis include overlapping study populations and insufficient data for analysis (missing variance or population numbers). All 16 studies enrolled patients with moderately to severely active RA. Of the analyzed studies, most did not specify whether patients used biologics before study entry and were assumed to have little or no prior biologic use. Only four studies reported potential biologic use before study entry. The greatest prior biologic use was reported in LITHE at 12% [49] and very low numbers were reported in AIM at 5% [36] and GO-BEFORE at 1% [45]. There was no indication that patients enrolled in RAPID 1 were excluded for previous biologic use; however, a drug washout period was allowed [43]. Table 1 summarizes the main characteristics of these 16 studies including the intervention and change in Sharp score from baseline to year 1 for each study. Not all baseline characteristics were reported for every study. The mean age across the groups ranged from 47 to 56 years, the majority (>67%) of patients were female, and mean duration of RA ranged from approximately 6 weeks to 11 years. Additional baseline characteristics of the study population are summarized in Supplementary Table 4.

Of the 16 clinical trials identified, nine used vdHS to assess radiographic progression [38,41–47,50] and were included in the base case NMA model (Figure 2). All biologics in combination with MTX and tocilizumab without MTX showed decreased radiographic progression at 1 year compared with MTX alone, as defined by negative mean difference in vdHS change (Figure 3). Most interventions showed a statistically significant effect versus MTX alone, except for golimumab 50 mg every 4 weeks and tocilizumab 8 mg/kg every 4 weeks. Among all the included biologics, the largest statistically significant mean difference at 1 year was in adalimumab, where the increase in vdHS was 3.8 fewer units in patients administered adalimumab 40 mg every 2 weeks + MTX compared with those receiving MTX alone. The patients treated with infliximab 3 mg/kg every 8 weeks had the next highest relative effect at 3.7 fewer units versus MTX alone.

Results of the subgroup analyses are provided in Table 2. Biologics in combination with MTX and tocilizumab without MTX decreased radiographic progression at 1 year compared with MTX alone among patients with early RA and as well as those with established RA. Of the 16 studies analyzed, seven studies had data for patients with early RA. Among patients with early RA, decreases in radiographic progression were seen for the biologics + MTX and for tocilizumab without MTX. Results of the subgroup analysis showed significant mean differences at 1 year for adalimumab + MTX (-3.83), infliximab + MTX (-3.30), etanercept + MTX (-2.18), certolizumab + MTX (-1.42) and tocilizumab 8 mg every 4 weeks without MTX (-0.87). Among patients with established RA, data were available for only two biologics (certolizumab and infliximab). For both interventions, the mean difference at 1 year was greater in patients with established RA compared with those with early RA (infliximab: -5.65 versus -3.30, respectively; certolizumab: -2.42 versus -1.42, respectively).

The SMD analysis (see Table 2) included data from 16 studies (9 base case studies reporting vdHS and 7 studies reporting different modifications of Sharp scores). Results of the SMD analysis were consistent with those of the base case analysis, which showed that all biologics in combination with MTX and tocilizumab without MTX decrease radiographic progression compared with MTX alone. Most interventions showed a statistically significant effect compared with MTX alone, except for golimumab 50 mg every 4 weeks and tocilizumab 8 mg every 4 weeks without MTX.

Results of pairwise comparisons for each intervention in the base case, SMD and subgroup analyses support the results obtained with the NMA model and SMD analyses (Supplementary Table 3). The pairwise comparisons generally showed that radiographic progression in patients with moderately to severely active RA was significantly decreased at 1 year in those treated with biologics in combination with MTX compared with those treated with MTX alone. Pairwise comparisons across the biologic agents did not reveal significant differences in radiographic progression changes which conclusively favored one biologic agent versus the other biologic comparators.

Risk of bias assessments were conducted on the 16 trials included in the analysis [32,36–50]. In the case of the five conference abstracts [36,42,43,49,50] respective full-text publications were used for the assessments [51–55]. A summary of the risk of bias assessment is presented in Table 3. In the 16 trials, the risk of selection bias was mainly low or unclear for random sequence generation and allocation concealment. The risk of performance bias, pertaining to blinding of participants and personnel, was mainly unclear as study authors simply described the study as ‘double-blind’ with no further details on how blinding was done and maintained throughout the trial. The risk of detection bias (pertaining to blinding of outcome assessors) and attrition bias (pertaining to incomplete outcome data) was mainly low. In the ATTRACT trial, the risk of attrition bias was assessed as high [47]. In this trial, no methods were employed to impute missing data; participants with missing data were excluded from analysis. Additionally, discontinuations were greater in the methotrexate alone group (50%) compared with all the groups that received infliximab plus methotrexate (21%), with lack of efficacy as the reason for discontinuation also being greater in the group that received methotrexate alone (36 vs 12%) [47]. All trials had low risk of reporting bias (pertaining to selective reporting of outcomes) and other sources of bias.

As shown in Table 4, radiographic data (spanning 2–10 years of follow-up) were found in 17 unique trials [8–32]. Direct comparisons of these trials are difficult because the study designs are different. For example in some studies, patients could cross over to treatment with the biologic agent (early escape) if defined study criteria were met during the double-blind treatment phase of the study [21–25], and in other studies patients crossed over to the treatment with the biologic when they entered the open-label extension phase of the study after completing the double-blind phase [8–12,14–16]. Despite the differences in study design, results of these studies demonstrate sustained slowing of radiographic progression for the biologics (abatacept, adalimumab, certolizumab, etanercept, golimumab and infliximab) in combination with MTX and for tocilizumab ± MTX.

Two trials (PREMIER and DE019) assessed the effect of a biologic agent on radiographic progression in RA over a 10-year period of time [12,56]. The PREMIER study examined the effects of adalimumab with and without methotrexate in patients with early RA [12]. In PREMIER, patients were randomly assigned to receive adalimumab + MTX, adalimumab alone or MTX alone for 2 years. Following the 2-year double-blinded period, patients were given adalimumab for up to 8 additional years, beginning as monotherapy; MTX could be added at the investigator's discretion. Results of PREMIER demonstrated that radiographic progression was prevented to a greater extent in patients initially randomized to adalimumab + MTX than in those initially receiving adalimumab alone or MTX alone (vdHS: 4.0, 8.8 and 11.0 at year 10, respectively) [12]. DE019 examined the long-term effectiveness of adalimumab in patients with established RA. In DE019, patients received adalimumab 20 mg weekly + MTX, adalimumab 40 mg every 2 weeks + MTX, or MTX alone during the 1-year double-blind treatment phase. The patients who completed DE019 could receive open-label adalimumab 40 mg every 2 weeks + MTX for an additional 9 years. Inhibition of radiographic progression was sustained in patients who received ADA + MTX [56]. It was noted that patients who initially received adalimumab + MTX had significantly lower mean change in vdHS at year 10 than those who received MTX alone (0.7 vs 6.2, respectively; p = 0.001) [56] suggesting that it is possible to prevent irreversible damage in patients with established RA. Results of a longitudinal, integrated analysis of the data from PREMIER and DE019 showed that treatment of RA with ADA ± MTX dramatically slowed annual radiographic progression over 10 years [13].

Our systematic literature search identified 16 studies that matched the predefined criteria for PICO to evaluate the effects of biologics + MTX compared with MTX alone on 1-year radiographic progression using total Sharp scores in patients with RA; all 16 were included in evidence synthesis [32,36–50] and nine were included in the base case NMA model [38,41–47,50]. Findings of the current study support the value of biologics in the inhibition of radiographic progression among patients with RA. Some biologics like adalimumab have up to 10 years of radiographic inhibition evidence in patients with early and established RA reported in the literature [13,56]; future studies are required to confirm these long-term findings for other biologics.

Similarity was assessed by examining clinical and study-level sources of variation including differences across studies in early versus established RA and type of Sharp score used. Results in patients with early RA were slightly better for some biologics suggesting that patients who initiated treatment early may experience less radiographic progression. However, results of this subgroup analysis should be interpreted with caution because of the limited number of studies available. Although, all the studies we selected for analysis measured radiographic progression in patients with RA, there are several versions of the Sharp radiographic scoring method (modified van der Heijde, modified Genant and others), which have different scales [57,58]. As a result, the base-case model included only those studies that reported the modified vdHS as this was the most commonly used outcome measure across the various biologics. In order to include all available trials in a single analysis, we performed a sensitivity analysis by standardizing the means of radiographic progression scores across the multiple radiographic progression scoring systems (i.e., SMD analysis). The findings of the SMD analysis revealed similar trends as the base case analysis, in other words, all biologics in combination with MTX decreased radiographic progression compared with MTX alone.

We qualitatively reviewed data on long-term radiographic progression because randomized clinical trials assessing radiographic progression beyond the 1-year time frame for all of the biologics are lacking. Although our qualitative assessment shows that within each of the reviewed randomized clinical trials all the examined biologics inhibited radiographic progression beyond the 1-year time point, there may be differences in the rates of radiographic progression between these agents. It is possible that these variable inhibition rates may result in differences in the magnitude of the efficacy of the various biologics beyond 1 year. Future randomized clinical trials should consider long-term, open-label extension studies to confirm that the shorter-term findings of the efficacy of biologics in inhibiting radiographic progression are maintained over a longer treatment period.

The clinical relevance of inhibiting radiographic progression among patients with moderately to severely active RA is clear. A published systematic review of studies examining the association between joint damage and functional disability among patients with RA showed that an increase in joint damage is associated with an increase in disability over time [59]. Thus, preserving the structural integrity of the joints is important, not only to prevent loss of physical function, but also to reduce pain and fatigue, all of which can improve attendance at work, productivity at work and performance of various tasks at home, as well as participation in social and leisure activities [60]. Results of our base case analysis showed that adalimumab + MTX and infliximab + MTX decreased radiographic progression at 1 year to a greater extent than the other biologics (certolizumab pegol, etanercept, golimumab or tocilizumab) + MTX compared with MTX alone. Similar results were seen in the subgroup population with early RA.

As imaging technology, such as MRI, has become more recognized as an important tool in the assessment of patients with RA, its use in the early detection and prognosis of RA is being evaluated [61]. Studies have shown that progressive erosive disease can occur in 19–40% of patients who are in clinical remission [62–64]. These findings have raised the issue of whether remission criteria should be extended to include imaging criteria in treat-to-target guidelines. Future studies are needed to examine the relationship between imaging results and synovial pathobiology in patients with RA before and after therapeutic interventions [61] to determine criteria for imaging remission despite clinical remission. The patients with RA who attain remission show better function, health-related quality of life and productivity even when compared with a low-disease activity state [65]. From a cost perspective, those with higher states of disease activity incur higher direct and indirect costs, making remission a worthwhile goal to reduce the burden to both patients and society [65].

Other recent network meta-analyses have assessed the radiographic progression benefits associated with biologics in RA patients who were naive to MTX and in those who had an inadequate response to MTX or other DMARDs [66–68]. Hazelwood et al. [68] found that MTX in combination with adalimumab, etanercept, certolizumab or infliximab was statistically superior to MTX alone in preventing joint damage at 1 year in MTX-naive patients, though not in patients with a previous inadequate response to MTX. Conversely, Singh et al. [66] found no benefit of using biologics + MTX versus MTX alone to prevent radiographic progression in MTX-naive patients, but did find a statistically significant benefit in favor of biologics among patients who had an inadequate response to MTX or other DMARDs [67]. In our study, patients were predominantly MTX/DMARD-naive and biologic-naive. All studies had to have radiographic progression assessed at a minimum of 1 year (using total Sharp scores) and patients had to receive treatment for ≥4 weeks. Results of our analyses generally agree with those of Hazlewood et al. [68]. In our base case analysis (Figure 3), we found that radiographic progression at 1 year was statistically significantly inhibited in those on adalimumab, certolizumab, etanercept, infliximab and tocilizumab in combination with MTX compared with those on MTX alone. In our SMD analysis (Table 2), radiographic progression at 1 year was significantly inhibited with all biologics, except golimumab, in combination with MTX compared with MTX alone.

Head-to-head trials assessing the comparative effect of biologics on radiographic progression are limited. The current study, using Bayesian NMA, provides a means of making comparisons among biologics to fill this gap in the literature and is among the first to review the radiographic efficacy of biologics. NMA relies on the following assumptions: there is similarity across studies in the patients included and common comparators, the treatment effects are transitive, and that there is consistency of evidence [69–72]. We were limited to the aggregate (study-level) reported data in our assessment of differences across studies. The availability of patient-level data for the entire evidence base would allow for a more complete exploration of homogeneity and similarity. We are also limited in our ability to perform subgroup analysis of biologic-naive versus biologic-experienced patients. Although many of the studies had low levels of patients with previous biologic use, outcomes are not reported for this subgroup of patients. Results of some studies may not have been published yet. We attempted to address this limitation by including conference abstracts for the data that were analyzed, but this was not an exhaustive process. Finally, while the relative effect was the largest for adalimumab, the CI was wide suggesting that additional studies are needed to decrease the uncertainty and confirm the findings of this analysis.

Results of this meta-analysis demonstrate that in patients with RA, biologics and MTX provide additional inhibition of radiographic progression as compared with MTX alone. Some biologics like adalimumab have published evidence of inhibition of radiographic progression in patients with early and established RA for up to 10-years of follow-up. Longer-term studies are needed to provide further evidence of the inhibition of radiographic progression beyond 1 year.

The current treat-to-target guidelines for RA recommend clinical remission be the therapeutic target as defined by ACR-EULAR remission criteria [73] because clinical trials and practice and most of the inflammatory rheumatology literature addressing radiographic changes, disability and quality of life among patients with RA is based on clinical observations and not on observations using imaging techniques [74]. However, it is recognized that inflammation may still be present in patients who are in clinical remission and there is discussion about expanding remission criteria by adding imaging criteria to the clinical composite indices in the ‘treat-to-target’ recommendations. In order to include radiographic progression as a target, decisions need to be made about the level of inflammation that can be tolerated, what imaging method should be used (radiography, CT scan, MRI and ultrasound), what joints should be assessed, and which cut-off point should be used [75]. We expect that the relationship between imaging technology and synovial pathobiology will be defined as future studies use this technology to examine synovial tissue along with joint damage at different disease stages before and after treatment with established and newly emerging biologic therapies.

In addition, a greater number of clinical trials in RA in the future are likely to have longer-term open-label follow-up periods due to the growing interest in understanding the efficacy of such therapies beyond the typical time-frame of a clinical trial. Radiographic data are also more likely to be gathered through real-world sources such as patient registries. We also expect to see more meta-analyses published in RA on the topic of radiographic progression as well as other patient-centric outcomes such as pain, fatigue, physical functioning, work productivity and disease remission.

The design, study conduct, and financial support for the study/trial were provided by AbbVie. AbbVie participated in the interpretation of data, review and approval of this publication. E Murray is an employee of Doctor Evidence and received funding from AbbVie to conduct this research. A Ellis and Y Butylkova were employees of Doctor Evidence at the time this study was conducted and received funding from AbbVie to conduct this research. Y Butylkova is a current employee of Amgen. M Skup, J Kalabic and V Garg are employees of AbbVie and may hold stocks or stock options in AbbVie.

Medical writing assistance was provided by Joann Hettasch, of Fishawack Facilitate Ltd, Conshohocken, PA, USA, and was funded by AbbVie.

This work is licensed under the Attribution-NonCommercial-NoDerivatives 4.0 Unported License. To view a copy of this license, visit http://creativecommons.org/licenses/by-nc-nd/4.0/

To view the supplementary data that accompany this paper please visit the journal website at: www.futuremedicine.com/doi/full/10.2217/cer-2017-0106