Objective To investigate the long-term effects of the tight control (TC) and conventional (CT) methotrexate-based strategies of the Computer Assisted Management in Early Rheumatoid Arthritis trial in early rheumatoid arthritis and evaluate the predictive value of an early response to treatment.
Methods Clinical and radiographic 5-year outcome was compared between initial strategies. Patients were classified according to the EULAR response criteria. The prognostic value of early response to treatment in addition to established predictors was analysed by multiple linear regression analyses.
Results 5 years of data were available for 205 of 299 patients, with no indication for selective drop-out. At 5 years there was no longer any significant difference for clinical and radiographic outcomes between treatment strategies applied during the first 2 years. Good-responders had a mean disease activity score of 2.39 (1.2) and median yearly radiographic progression rate of 0.6 (0.0 to 2.2) at 5 years; significantly lower (both p<0.02) when compared to moderate- and non-responders. Multiple regression analysis showed that early response to treatment is an independent predictor of 5-year outcome, irrespective of treatment strategy.
Conclusions The difference in disease activity between treatment strategies disappeared over the years. Good-response to treatment independently predicts significantly better 5-year clinical and radiographic outcome. The TC principle probably should be continued in the long-term.
Statistics from Altmetric.com
The ‘tight control’ principle is considered an important concept in the treatment of rheumatoid arthritis (RA). It has been found effective in the short term; however, fewer data are available on its long-term outcome.1,–,4 Next to known predictors of long-term outcome of RA, which are assessed at baseline,5,–,8 early response to treatment also could be predictive.9 Disease activity during the first 3 months of treatment has been related to clinical and radiographic outcome after the first year of treatment,6 10 11 and an early treatment response can predict the long-term work disability outcome.9
The present study investigates 5-year data of a 2-year methotrexate (MTX) based tight control treatment (TC) strategy compared to conventional MTX-based treatment (CT) strategy, and the (added) predictive value of an early response (at 6 months) to treatment for 5-year outcome with respect to disease activity and radiographic progression.
Patients and methods
Patients who participated in the Computer Assisted Management in Early Rheumatoid Arthritis (CAMERA12) trial were evaluated. Patients had been randomised to the TC strategy, based on computer-guided monthly predefined response criteria, or to the CT strategy, based on regular clinical practice with 3-monthly visits; both strategies were aimed at remission. Radiographs of hands and feet were assessed at baseline and, subsequently, yearly. In the first 2 years, medication was restricted to the protocol,12 13 thereafter it was free. The medical ethics committees of all participating hospitals approved this study and all patients gave written informed consent before entering the study.
EULAR response criteria
At 6 months, patients were classified as good-responders, moderate-responders or non-responders according to EULAR response criteria.14 These criteria are based on the actual disease activity score (DAS2815) and the change from baseline in DAS28.
DAS28 and radiographic joint damage, assessed according to the Sharp van der Heijde Score (range 0–44816) and blinded to clinical information, were investigated at 5 years of treatment. If 5-year data were not available, the mean of 4 and 6 years, or the outcome at 4 or 6 years of treatment were used, depending on data available.
DAS28 at 5 years and mean yearly radiographic progression rate over 5 years were compared between the initial treatment strategies and the early response categories and tested for significant differences with independent Student t tests and analysis of variance respectively.
The additional predictive value of early response to treatment over well-known predictors (rheumatoid factor (RF) and baseline disease activity or joint damage) was investigated with multiple linear regression analysis.
As initial response might have a different relation to outcome in the individual initial treatment strategies due to different timing and use of treatment, the interaction between treatment strategy and response to treatment was investigated.
The influence of missing data on results was investigated using multivariate imputation.
SPSS 15.0 was used for the analyses.
Baseline characteristics and demographics for all patients and for patients with available 5-year data in both treatment strategies are given in table 1.
Five years of data were available for 205 of 299 patients included (102 TC and 103 CT). For the TC compared to the CT strategy 47% versus 23% (p=0.001) were good-responders, 41% versus 44% moderate-responders (p=0.70) and 12% versus 33% non-responders (p=0.001) at 6 months of treatment respectively (table 2). No early EULAR response for eight patients could be calculated due to missing scores. For 34 and 76 patients, respectively, no DAS28 or radiographic progression at 5 years was available; for these patients data at 4 and/or 6 years were used. Baseline patient and disease characteristics were similar for patient groups including all patients, patients with available 5-year data and also for treatment strategy groups.
Treatment strategies and 5-year outcome
Figure 1A,B shows disease activity and joint damage over 5 years for the treatment strategies based on available data. The mean (SD) DAS28 values at 5 years of treatment were 2.68 (1.0) and 2.75 (1.3) for patients in the TC and CT strategies respectively (p=0.66). The median (IQR) radiographic progression rates at 5 years were 1.4 (0.1 to 3.6) and 0.8 (0.0 to 3.2) respectively (p=0.50).
Early response to treatment and 5-year outcome
Figure 1C,D shows disease activity and joint damage over time for good-responders, moderate-responders and non-responders based on available data. They had mean (SD) DAS28 at 5 years of 2.39 (1.2), 2.69 (1.1) and 3.11 (1.2) respectively. At 5 years, good-responders had lower disease activity compared to moderate-responders and non-responders (p=0.09 and p=0.001 respectively), and moderate-responders lower values compared to non-responders (p=0.046) (figure 1C). Patients with a good-response, moderate-response or non-response had median (IQR) radiographic progression rates of 0.6 (0.0 to 2.2), 1.5 (0.2 to 3.6) and 2.5 (0.5 to 6.2) respectively. The differences between good-responders and moderate-responders and non-responders were statistically significant (p=0.013 and p=0.001 respectively), in contrast to the difference between moderate-responders and non-responders (figure 1D).
In multiple regression analyses, early response was significantly related to 5-year DAS28, independently of established predictors: the explained variance (R2) of the model increased from 0.037 to 0.091 when including early response. Standardised βs indicated that the influence of early response on DAS28 was stronger compared to the predictive value of RF and baseline DAS28 (table 3).
Early response was also significantly related to the mean yearly radiographic progression rate next to the established predictors. The R2 of the model increased from 0.208 to 0.242 when including early response. However, according standardised βs the influence of early response on outcome was smaller than that of baseline joint damage (table 4).
The interaction between treatment strategy and early response on 5-year outcome also was investigated. In both strategies, good-response and moderate-response were associated with less 5-year disease activity and radiographic progression when compared to non-response in the CT strategy. The effect of early response to treatment on DAS28 and radiographic progression rate was not significantly different between the treatment strategies. However, moderate-responders and non-responders in TC seemed to have worse outcome and good-responders seemed to have a slightly better outcome when compared to the corresponding responder groups of CT. Adding established predictors and covariates to the models did not alter the results above (see supplementary file).
The possible influence of missing data on the results was investigated using multivariate imputation analysis. Using this imputation method, data for 270 (139 TC, 131 CT) and 282 (143 TC, 139 CT) patients were available for analysis regarding 5-year outcome. Results of these analyses were similar to those of the non-imputed dataset (data not shown).
The difference in disease activity between the TC and CT strategies of CAMERA slightly decreased over the following years and was absent after 5 years. A possible explanation might be that the TC principle was abandoned with free medication prescription after the 2-year trial. Patients in the TC strategy might have been treated less tightly and those in the CT strategy more tightly after 2 years, diluting the difference in effect. For instance, in both groups about 13% of patients used a biological during follow-up. However, when comparing the follow-up of CAMERA to other TC strategies similar disease activity and remission values were seen.17 18 The TC principle probably should be continued in the long term and the strategy should be intensified, for example, by adding biologics to the strategy when the early response to treatment after 6, or even 3, months is insufficient. This is in line with recently published recommendations for daily clinical practice.19
Remarkably, joint damage was still minor and not different between both treatment strategies after 5 years. However, when classifying patients according EULAR response at 6 months of treatment, there were clear differences in 5-year outcome in favour of good-responders, for radiographic outcome and disease activity. This implies that early response to treatment is very important in defining an optimal treatment strategy and optimising treatment after initial non-response in early RA.
The early response to treatment is also influenced by treatment strategy. Non-responders in the TC strategy might be ‘real’ non-responders, whereas the group of non-responders in the CT strategy might be composed of non-responders and ‘late-responders’.
Limitations of this study are that only two-thirds of the patients were available for analysis after 5 years of treatment. However, there seemed to be no selective drop-out, suggesting that the available patients were a good representation of the total CAMERA population. Early response to treatment was determined after 6 months of treatment to allow for a relevant period to optimise MTX treatment for the CT strategy (with only 3-monthly visits). However, for patients in the TC strategy (with monthly visits), this time period already might have been set, for example, after 3 months of treatment. Preferably, the response to treatment should be identified as early as possible in the disease course.
In conclusion, although no differences in disease activity and progression of joint damage were seen between the treatment strategies of CAMERA after 5 years of treatment, a good-response to treatment predicted better 5-year outcome, irrespective of the treatment strategy used. These results support the continuing use of the TC principle over time and the use of more adequate medication schemes early in the disease.
The authors thank all participating rheumatologists and research nurses of the Utrecht Rheumatoid Arthritis Cohort study group for data collection, A W J M Jacobs-van Bree for data entry, and A A van Everdingen, MD, PhD for scoring radiographs.
Competing interests None.
Patient consent Obtained.
Ethics approval This study was conducted with the approval of the METC UMC Utrecht, the Netherlands.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.