Greater likelihood of remission in rheumatoid arthritis patients treated earlier in the disease course: Results from the Consortium of Rheumatology Researchers of North America registry

INTRODUCTION

Remission, meaning the absence of disease activity, is the goal of therapy in rheumatoid arthritis (RA). Earlier and more aggressive use of disease-modifying antirheumatic drugs (DMARDs) is being advocated with the goal of achieving remission (1). For RA patients treated with nonbiologic DMARDs, including methotrexate, there is some evidence that drug responsiveness may be influenced by disease duration (2). Over the last decade, there has been increased attention to anti–tumor necrosis factor (anti-TNF) therapies and the potential impact they can have on remission. Anti-TNFs have demonstrated unprecedented efficacy for the treatment of RA (3-6). Clinical trials of anti-TNFs have been performed demonstrating efficacy for patients with both early and established RA (5-7). However, because direct comparisons of response rates across trials with anti-TNFs in early versus established disease are not appropriate, it is unclear if the magnitude of response or rates of achievement of remission are superior for patients with a shorter duration of disease in a more typical clinical setting. We sought to examine the relationship between disease duration and achievement of remission and whether that was influenced by the class of treatment initiated. Specifically, we sought to compare the rates of remission stratified by disease duration in those initiating nonbiologic DMARDs and anti-TNF agents.

Therefore, the aim of the present study was to investigate the a priori hypothesis that RA disease duration was an independent predictor of achieving remission within one year of treatment, regardless of DMARD class. We tested this hypothesis using a large US cohort of RA patients with a broad range of disease duration from the Consortium of Rheumatology Researchers of North America (CORRONA) registry.

PATIENTS AND METHODS

RESULTS

There were 1,646 patients who initiated a nonbiologic DMARD and 3,179 who initiated anti-TNF therapy as well as met inclusion criteria. The mean and median number of visits over the 1-year period was 2 for both of the groups. For analyzing sustained remission, we examined 54–59% of patients with at least 3 visits. The majority of patients stayed on their medications during the 1-year period (84–87% of nonbiologic DMARD initiators and 79–82% of anti-TNF initiators). Remission based on the CDAI over the 1-year followup period occurred in 18.3% of nonbiologic DMARD initiators and 17.6% of anti-TNF initiators, while sustained remission occurred in 7.2% and 7.6%, respectively, in those with 2 or more followup visits. When we limited the population to those in which a DAS28 could be calculated, DAS28 remission occurred in 14.4% and sustained remission occurred in 3.5% of nonbiologic users (n = 617). In anti-TNF initiators (n = 1,247), it was 20.0% and 7.9%, respectively.

The baseline characteristics of RA patients with early and established disease included in the evaluation of remission are shown in Table 1. Those with earlier RA were more likely to be younger and working. Among anti-TNF initiators, those with earlier disease were more likely to be receiving concomitant methotrexate. Greater disease duration was associated with more exposure to DMARDs. Among the anti-TNF initiators, there were no significant differences in the specific agent prescribed. Disease levels at baseline were generally comparable, with the exception of those with 6 to 10 years of RA who were more likely to have low disease activity as compared to the other disease duration groups for both the nonbiologic DMARD and anti-TNF initiators. Similar results of baseline characteristic comparisons were found when examining the anti-TNF initiators who had never taken biologics as well as when we compared RA patients who had 2 or more follow up visits, which was the sample used for the evaluation of sustained remission (data not shown).

The percentages of patients who achieved remission based on the CDAI over the followup period among the nonbiologic DMARD initiators was 21.3% in those with ≤5 years of disease duration as compared to 19.6% in those with 6–10 years and 13.5% in those with ≥11 years (P < 0.001) (Table 2). In the anti-TNF initiators, the rates were 22.3%, 17.7%, and 12.8%, respectively (P < 0.001). Similar results were found using the DAS28 in the anti-TNF initiators (24.1%, 21.2%, and 15.0%, respectively; P = 0.001). The DAS28 remission rates were somewhat lower among the 617 patients in the nonbiologic DMARD cohort who had the necessary data elements for the DAS28 calculation. The percentage of patients achieving sustained remission based on the CDAI was also greater for those with earlier disease duration among both the nonbiologic DMARD and anti-TNF initiators (nonbiologic DMARD initiators: 10.2%, 8.8%, and 2.5%, respectively; P < 0.001 and anti-TNF initiators: 9.7%, 9.5%, and 4.2%, respectively; P < 0.001). Similar estimates were obtained using the DAS28 among the anti-TNF initiators (8.9%, 11.6%, and 4.9%, respectively; P = 0.04).

Among the nonbiologic DMARD initiators, an increase of disease duration of 5 years was associated with a reduced likelihood of remission based on the CDAI (OR 0.91, 95% CI 0.83–0.99) in both unadjusted and adjusted analyses (Table 3). Greater disease duration was also associated with a reduced likelihood of achieving sustained remission based on the CDAI (OR 0.61, 95% CI 0.48–0.76) (Table 4). Using the DAS28 and a smaller study population, the estimates for remission and sustained remission were similar, but the 95% CIs crossed unity.

Among the anti-TNF initiators, disease duration was associated with remission using both the CDAI (OR 0.88, 95% CI 0.83–0.94) and the DAS28 (OR 0.90, 95% CI 0.83–0.99) (Table 5). Greater disease duration was also associated with a reduced likelihood of achieving sustained remission based on the CDAI (OR 0.85, 95% CI 0.75–0.97). Using the DAS28, the estimate was similar but the 95% CI crossed unity (OR 0.89, 95% CI 0.75–1.06) (Table 6).

As a sensitivity analysis, we examined the relationship between disease duration and achievement of remission, as well as sustained remission, in those who had never taken biologics using logistic regression models (there were too few patients to examine rates in those initiating their first nonbiologic DMARD). There was a significant relationship between disease duration and likelihood of remission using the CDAI (OR 0.85, 95% CI 0.79–0.938 for a 5-year increase in RA disease duration); however, the 95% CI crossed unity when using the DAS28 outcome definition (OR 0.97, 95% CI 0.86–1.09). This is likely related to the greater than 50% decrease in the sample size (1,217 versus 579) when the analyses were limited to those who had never taken biologics. Similar findings were demonstrated when examining the likelihood of sustained remission using the CDAI (OR 0.85, 95% CI 0.73–0.98) and DAS28 (OR 0.94, 95% CI 0.78–1.14).

DISCUSSION

In this large US-based cohort of RA patients, we report on remission rates achieved using nonbiologic DMARDs and anti-TNF agents based on RA disease duration through two well-recognized remission definitions. Interestingly, unadjusted CDAI remission and CDAI sustained remission rates were similar in the nonbiologic DMARD and anti-TNF initiators, particularly those with ≤5 years of disease duration. In both unadjusted and adjusted analyses, initiation of nonbiologic DMARDs and anti-TNFs in those with earlier disease was associated with a greater likelihood of remission and sustained remission using the CDAI. This was also seen using the DAS28 remission definition in anti-TNF initiators. Similar results were found when the analyses were limited to patients who had never taken biologics at the time of anti-TNF initiation.

Our results build upon the work of others who reported disease duration to be an important prognostic predictor of RA drug response. A meta-analysis of randomized controlled trials of nonbiologic DMARDs, including methotrexate, demonstrated greater ACR response rates for patients with a shorter disease duration (2). However, remission outcomes were not examined and patients in the meta-analysis had markedly greater disease activity based on tender and swollen joint counts, ESR, and M-HAQ scores than our US-based cohort. In a post hoc analysis of etanercept trials, patients with early RA had greater improvement in functional status as measured by the HAQ than those with established disease (13). In a post hoc analysis of the DE019 trial, which demonstrated the efficacy of adalimumab in patients with RA, there was a trend toward superior clinical, functional, and radiographic outcomes in patients with early versus established disease (14). These studies support our findings in a large observational cohort that disease duration is an important predictor of remission outcomes.

To our knowledge, this is the first large clinical observational study showing greater remission rates associated with anti-TNF use stratified by disease duration. The largest published clinical study of remission and anti-TNF therapy was the Research in Active RA (ReAct) trial, an open-label trial of adalimumab (15). The ReAct study population of 6,610 RA patients had remission rates of 27% by the CDAI and 38% by the DAS28 (15). The baseline level of disease activity in ReAct was greater than in the CORRONA cohort. Interestingly, disease duration was not found to be a predictor of DAS28 remission. Similarly, in the British Society for Rheumatology Biologics Register, disease duration was neither a predictor of DAS28 remission nor a response for patients prescribed etanercept (n = 1,267) or infliximab (n = 1,612), and they examined the influence of disease duration based on 10-year increments (16). A third observational study from Italian investigators of 1,257 RA patients treated with anti-TNFs failed to identify disease duration as a predictor of remission, using 12 years as the cut point (17).

Our findings differ from the results of the 3 European clinical studies, and a number of differences may explain these discordant results. First, the study population of anti-TNF–treated patients in the CORRONA US-based registry included a large percentage of patients with less severe RA, with baseline joint counts and disease activity markedly lower than observed in the European studies. The many societal and health access reasons for this disparity have been discussed in detail (11). Briefly, the use of biologic drugs is much greater in the US than in European societies because in Europe certain minimal disease characteristics are mandated in one-payer systems. Finally, the primary definition of remission is also different between studies. Since our results were similar using the CDAI and the DAS28 definitions, this gives us additional confidence in our findings.

Interestingly, we found approximately 50% fewer patients in sustained remission compared with remission at one visit during the followup period. This suggests that many patients move in and out of remission, no matter how the remission is defined. It is likely that patients have ongoing disease activity even when they are characterized as being “in remission.” Nevertheless, we observed that disease duration was an independent predictor of sustained remission. To achieve and sustain remission in patients, treating rheumatologists will need to be diligent in detecting and documenting low levels of clinical disease activity. Sustained remission is probably preferable, as it has been demonstrated that longer sustained remission, variously defined, was associated with less radiographic or functional deterioration and that shorter times in remission were associated with more radiographic progression (18-21). However, radiograph progression may occur despite being in clinical remission (22). Brown et al demonstrated synovitis by ultrasound in some patients in clinical remission, and this led to progression of radiographic joint damage (23). Therefore, all of the clinical implications of patients who go in and out of remission versus those who have a sustained remission have not really been uniformly established. It is presently unclear if disability or quality of life issues are substantially different between these two therapeutic responses.

It is not surprising that prior anti-TNF use decreased the likelihood of achieving remission, as the decision to discontinue is often based on response. Others have found that the effectiveness of anti-TNF agents was consistently superior for patients who had never taken biologics versus biologic-experienced patients (24-26). Similar findings were demonstrated in randomized controlled trials of new biologic agents, suggesting that patients with a prior inadequate response to anti-TNF agents may represent a more treatment-resistant population (27). Interestingly, patients with greater functional impairment (indicated by a greater M-HAQ score), disability, and greater fatigue at baseline also have a decreased likelihood of achieving remission. Taken together with our observation of an enhanced clinical response in patients with earlier disease, it follows that patients with a high M-HAQ score may have more advanced disease, and therefore, the anti-TNF agent was less likely to be effective. The same observation was noted in the nonbiologic DMARD initiators. However, alternate hypotheses include the possibility that even earlier interventions will not be as effective in patients who already have established functional decline, as they may have more treatment-resistant disease and have an irreversible component to their functional disability that would be unresponsive even to effective therapies for RA. It is also possible that the components of the M-HAQ can be related to alternative disease states, including secondary osteoarthritis, which would not necessarily be expected to respond to more aggressive interventions for RA.

Our study has certain limitations. Currently, there are several approaches to assess clinical remission. We used the CDAI as our primary remission measure, principally because this allowed for the inclusion of the greatest number of patients with complete measures. The CDAI reflected the impact of disease duration on both remission and sustained remission in both the unadjusted and adjusted models, while the DAS28 showed effects only for remission (not sustained remission). This difference may have simply been the result of having fewer patients in the DAS28 group for which sustained remission could be assessed (CDAI group n = 1,830 versus DAS28 group n = 602 among the anti-TNF initiators). Also, the CDAI is independent of the ESR, leaving it more freely responsive to the purely clinical measures. It is reassuring that the CDAI remission criteria are more stringent than the DAS28, in that these require less residual disease activity (28). Our registry does not mandate radiographic imaging at predefined time periods; thus, our analyses could only focus on the attainment of clinical remission with nonbiologic DMARDs and anti-TNF therapies without examination of radiographic progression. Lastly, the patients with a longer disease duration were more likely to be switching nonbiologic DMARD therapy rather than the first-time DMARD users, and there were insufficient numbers of patients starting their first nonbiologic DMARD in order to analyze their results separately. Therefore, it is difficult to disentangle the effect of disease duration from treatment-resistant disease suggested by nonbiologic DMARD switching.

The strengths of our study include a very large number of “real-world” patients from a large US-based registry that includes detailed clinical data measurements derived from rheumatologists at the time of a clinical encounter. Many of the reports from registry data are derived from Europe, where the overall disease activity tends to be substantially greater than that found within the US (11). Therefore, US registry data are important for comparisons of clinical outcomes across different societal and political contexts.

In conclusion, the results of this US-based study demonstrate that for every increase of disease duration of 5 years, the likelihood of remission and sustained remission are reduced by 10–15% in the adjusted models. Given that it represents a state with the least likelihood of disease progression, remission should be the therapeutic goal for patients and providers. To achieve this goal, our data lend further evidence that early and aggressive treatment of RA is an appropriate strategy in order to achieve remission as soon as possible after the onset of the disease. Patients and providers should strive for tight control through serial assessment of disease activity, and by basing treatment decisions on these findings.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Greenberg had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Furst, Pangan, Harrold, Reed, Greenberg.

Acquisition of data. Kremer, Greenberg.

Analysis and interpretation of data. Furst, Pangan, Harrold, Chang, Reed, Greenberg.

ROLE OF THE STUDY SPONSOR

Abbott, as part of its subscription contract, proposed the research study and was involved in the early design, as well as manuscript review and comment for this study. The final study design, analysis plan, reporting of study results, writing of the manuscript, and decision to publish the manuscript were those of the non-Abbott authors. A number of other companies, including Amgen, BMS, Centocor, Genentech, Lilly, and Roche, have also supported the Consortium of Rheumatology Researchers of North America registry through subscription contracts. These companies were not involved in the study design, data collection, data analysis, writing, approval, or decision to publish this manuscript.