Article Text

Download PDFPDF

Remission achieved after 2 years treatment with low-dose prednisolone in addition to disease-modifying anti-rheumatic drugs in early rheumatoid arthritis is associated with reduced joint destruction still present after 4 years: an open 2-year continuation study
  1. I Hafström1,
  2. K Albertsson1,
  3. A Boonen2,
  4. D van der Heijde3,
  5. R Landewé4,
  6. B Svensson5,
  7. for the BARFOT Study Group
  1. 1
    The Rheumatology Unit, Karolinska Institute, Karolinska University Hospital, Huddinge, Stockholm, Sweden
  2. 2
    Department of Internal Medicine, Division of Rheumatology, University Hospital, and CAPHRI Research Institute, Maastricht, The Netherlands
  3. 3
    Department of Rheumatology, Leiden University Medical Center, Leiden, The Netherlands
  4. 4
    Department of Internal Medicine, Division of Rheumatology, University Hospital Maastricht, and CAPHRI Research Institute, Maastricht, The Netherlands
  5. 5
    Department of Rheumatology, Lund University, Lund, Sweden
  1. Ingiäld Hafström, Rheumatology Department, R92, Karolinska University Hospital Huddinge, 141 86 Stockholm, Sweden; ingiald.hafstrom{at}


Objective: To evaluate if remission induced by low-dose prednisolone during the first 2 years of rheumatoid arthritis (RA) in the BARFOT glucocorticoid (GC) study had a sustained effect on radiological damage for a total of 4 years.

Methods: A total of 150 of 211 eligible patients with RA who had been randomised to the 7.5 mg prednisolone group (P) or no prednisolone group (NoP) in addition to the initial disease-modifying antirheumatic drugs were included. Radiographs of hands and feet were scored using the Sharp–van der Heijde scoring method. A patient was considered to be in remission if the 28-joint count disease activity score was <2.6.

Results: Mean (SD) age was 53 (14) and 57 (12) years for the patients in the P and NoP groups, respectively. 64% were female, 64% rheumatoid factor positive, and disease duration at baseline was 6 months. At 2 years the proportion of patients in remission in the P and NoP groups was 55 vs 30%, p = 0.003. Longitudinal analysis showed that over the entire course of the disease, patients on prednisolone had a higher probability of being in remission. Patients in remission at 2 years, compared with those not in remission, had significantly lower total Sharp score, erosion score and joint space narrowing score at 2 and 4 years. The changes in bone mineral density during the 4 years did not differ between those in remission and those with active disease, and were similar in the two treatment groups.

Conclusions: Prednisolone 7.5 mg daily in addition to disease-modifying anti-rheumatic drugs increases the rate of remission in patients with early RA, which has a beneficial and sustained effect on radiological damage.

Statistics from

Request Permissions

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.

Nowadays, the goal of treating rheumatoid arthritis (RA) is to induce clinical remission based on the hypothesis that the inflammatory process leads to structural damage. This relationship between inflammation and joint damage has been shown at the level of single joints and for overall disease activity.1 2 An additional proof of this connection is that early suppression of disease activity reduces radiological progression.35 On the other hand, remission of clinical disease does not preclude radiological progression6 and therapeutic interventions can affect clinical synovitis differently from that of erosive progression.7 A partial disconnect between inflammation and joint damage has thus been suggested.8

Unfortunately, it has been proven difficult to induce disease remission in many patients. To improve treatment response, several treatment strategies have been proposed. These involve the early use of disease-modifying antirheumatic drugs (DMARDs);912 to use these drugs in combinations either in fixed combinations13 or in step-up14 or step-down regimens15 and to use the new biological agents early, in particular, treatment using antitumour necrosis factor.1618 Also the use of tight control of joint inflammation has been shown to increase remission rate.19

Lately, glucocorticoids (GC) in a low dose have developed a renaissance in early RA as GC not only have a disease-modifying potency2024 but also an ability to induce remission.23 It is, however, uncertain if continuation with (low-dose) prednisolone treatment or a time-restricted initial treatment has any long-term beneficial effects. In the Arthritis & Rheumatism Council study, 7.5 mg prednisolone was given in addition to DMARDs for 2 years and then discontinued, and while the joint destruction rate then resumed during the third year, the joint damage was still significantly lower compared with patients treated with placebo.25 In contrast, the COBRA study (Combinatietherapie Bij Reumatoide Artritis) reported a long-term reduction of radiological damage in a 4–5-year follow-up in patients that received triple therapy, including prednisolone. However, prednisolone was started at a high dose, followed by tapering of the dose and withdrawal at week 35.5 None of these studies explored whether patients that achieved remission by prednisolone treatment had a more favourable disease progression than those that had not.

Therefore, in the present BARFOT follow-up study, we evaluated if remission induced by low-dose prednisolone added to the first DMARD during the first 2 years of early RA had a sustained effect on radiological damage and physical function measured over a period of 4 years.


Patients and treatment

The patients in this extension study had previously participated in the main BARFOT glucocorticoid study, which was a multicentre, open, randomised study. It evaluated treatment with 7.5 mg prednisolone + DMARD versus only DMARD over a period of 2 years in 250 patients with early RA. Radiographic progression was the primary outcome.23

Of the initial six centres, five agreed to participate in the present extension study. These centres had originally randomised 234 patients, of which 211 were still following the allocated treatment (prednisolone or no prednisolone) at the 2-year follow-up. Of these 211 patients, 150 agreed to participate in the 4-year follow-up study. At start of the first DMARD they had been randomised to either 7.5 mg prednisolone daily (P group (n = 64)) or no prednisolone (NoP group (n = 86)) for the first 2 years. At 2 years, 29 patients were scheduled to discontinue prednisolone treatment. The dose was reduced by 2.5 mg/week and reached zero after 20 weeks. The treating rheumatologists selected DMARD following the recommended treatment strategy in Sweden. Thus, most patients received methotrexate or sulfasalazine, similarly distributed among the two treatment groups. All patients were given 1000 mg calcium/day. Ethical committees approved the study and all patients signed informed consent.

Radiographic scores

Radiographs of hands, wrists and feet were obtained in the main study at baseline, 1 and 2 years, and in this follow-up study at 3 and 4 years. In the main study, two readers scored the radiographs independently in chronological order without knowledge of treatment assignment or clinical response. The interobserver intraclass correlation coefficients between these two readers were above 0.9 for the erosion, joint space narrowing and total Sharp scores. It was therefore decided that the films at 3 and 4 years could be scored by only one of the readers (KA). Knowing the scores of the 2-year time point, the films at 3 and 4 years were scored with known time sequence using the van der Heijde modification of the Sharp score.26

Disease activity

Disease activity was assessed by the 28-joint count disease activity score (DAS28).27 A patient was considered to be in remission if DAS28 was <2.6.28

Physical function

The Swedish version of the Stanford Health Assessment Questionnaire (HAQ) was used to measure physical function.29 The HAQ score ranges from 0 to 3, and a higher score indicates a higher degree of disability.

Bone mineral density

Bone mineral density (BMD) was measured by dual-energy x-ray absorptiometry with Lunar densitometer at the lumbar spine (L2–L4) with an anterior–posterior view, and at the left hip (femoral neck). BMD was expressed in gram of bone mineral per square centimetre (g/cm2) and as the number of standard deviations (SD) in comparison with healthy age- and sex-matched individuals (Z-score), obtained from Lunars combined European/US reference population (Operator manual, expert-XL, software version 1.7; Lunar Corporation, Madison, Wisconsin, USA).

Statistical analysis

The patients were analysed according to the group to which they were originally randomised. The statistical analyses were performed using the SPSS 14.0 statistical software. Radiographic damage was reported by medians with interquartile ranges.

For clinical variables, group changes were given as mean (SD). Percentages of patients in remission (DAS28 <2.6) at follow-up in the two groups were presented. To test differences between the groups the Mann–Whitney test or the independent sample t-test was used for continuous variables and the χ2 test for proportions.

In order to further investigate associations between drug use (prednisolone), clinical status (remission) and radiographic damage (total Sharp score), longitudinal analysis was applied using generalised estimating equations (GEE). GEE is a technique that allows the analysis of longitudinal relationships (consecutive assessments in the same patient) under the adjustment for within-patient correlation. Such a technique uses all available data of the same patient without the risk of spurious associations due to high within-patient correlation. For these analyses we assumed an exchangeable correlation structure. Dependent variables in this analysis were DAS28 remission (yes versus no) and radiographic damage (total Sharp score). Independent variables were prednisolone use (longitudinal variable) and DAS28 (longitudinal variable). Adjustments were made for age, sex and disease duration. Results were expressed as odds ratios (for DAS28 remission) and estimated marginal means (for radiographic damage).

All significance tests were two-tailed and conducted at the 0.05 significance level.


Patient characteristics

At baseline, patients in the P and NoP groups had a mean (SD) age of 53 (14) and 57 (12) years, respectively (p = 0.027). Disease activity, function and total Sharp score were similar (table 1).

Table 1 Characteristics of the 150 patients at baseline at time of randomisation

The baseline characteristics for these 150 patients did not differ from the characteristics of those patients in the main study that did not participate in this extension study. Also, DAS28, HAQ score and total Sharp score at 2 years did not differ either between those patients that participated and those that did not.

Patient disposition and treatment

Of the 64 patients in the P group, 26 patients continued with 7.5 mg prednisolone/day and 31 discontinued prednisolone treatment during year 3. In the NoP group (86 patients) three started on 7.5 mg prednisolone/day during years 3–4.

At 2 and 4 years, DMARD treatments were similar in the P and NoP groups. At the 4-year follow-up 44% vs 37% of the patients in the P and NoP groups were treated with methotrexate, 13% vs 9% with sulfasalazine and 25% vs 33% with other DMARDs or combinations, while 19 vs 21% of the patients did not use DMARDs.


Remission rates in the first 2 years for the 150 patients were for the P and NoP groups at 3 months (35% vs 9% (p = 0.0005)), at 6 months (48% vs 22% (p = 0.001)), at 12 months (49% vs 42% (p = 0.36)), at 18 months (53% vs 34% (p = 0.020)) and at 2 years (55% vs 30% (p = 0.003)). GEE showed that DAS28 was more suppressed in the P group compared with the NoP group. This treatment effect sustained until 24 months, and was statistically significant at 3, 6, 12, 18 and 24 months. GEE also confirmed that the probability of being in DAS28 remission until 2 years was significantly higher in the P group as compared with the NoP group (p<0.001). Thereafter, the difference between the groups disappeared. At 4 years, 40% of all patients were in remission.

We further investigated whether the discontinuation of prednisolone in a proportion of the patients in the P group after 2 years had an impact on DAS28, using GEE with prednisolone use as a longitudinal variable. The DAS28 temporarily increased to some extent in the patients discontinuing prednisolone after 2 years, while remaining at a constant level in the patients continuing on prednisolone. The between-group difference of 0.8 DAS28 units was statistically significant (p = 0.02), but this difference completely disappeared during years 3 and 4.

Remission and radiographic damage

Patients in remission at 2 years, compared with patients not in remission, had a significantly lower mean total Sharp score, erosion score and joint space narrowing both at 2 and 4 years (table 2 and fig 1).

Figure 1

Total Sharp scores for patients in remission and not in remission, split by prednisolone use during the first 2 years. The figure shows mean total Sharp scores over time. (A) Prednisolone group; (B) no prednisolone group. The error bars show 95% confidence intervals of the mean.

Table 2 Radiological scores at 2 and 4 years as well as change in these scores from baseline to 4 years for all patients who at the 2-year control point were in remission or not in remission and separately for the two treatment groups

Furthermore, patients in remission had significantly less change in radiological scores during the 4 study years compared with those not in remission (table 2). These exploratory results were confirmed by longitudinal analysis (table 3), in which DAS28 remission proved to be significantly associated with lower Sharp scores during the entire course of observation.

Table 3 Longitudinal analysis investigating the relationship between DAS remission and radiographic damage in patients with rheumatoid arthritis, randomised to receive prednisolone or not

In a post-hoc analysis the patients in remission were subdivided into those with prednisolone during the first 2 years and those without; then it was found that patients in the P group, who were in remission, had significantly less change in radiological score after 4 years compared with patients not in remission. The patients in the NoP group, who were in remission, showed the same trend, but the differences were not statistically significant (table 2).

The longitudinal analysis gave some support to these findings. Although the interaction between DAS28 remission (yes versus no) and prednisolone use (yes versus no) was not statistically significant (p = 0.153), radiological progression in patients in remission was dependent on prednisolone use. The difference in marginal means (3.0 in the NoP group and 1.5 in the P group) was statistically significant (p = 0.011).

Radiographic progression

We further analysed radiographic progression over the 4 years by GEE, irrespective of remission state (fig 2). Patients randomised to the NoP group had a steeper increase in total Sharp score during the first year (p = 0.019). Thereafter, the difference obtained at year 1 remained during years 2, 3 and 4, but this trend was not statistically significant (p = 0.079). We could not confirm a significant effect of stopping prednisolone after 2 years on radiographic progression.

Figure 2

Radiographic progression over 4 years for patients treated and not treated with prednisolone during the first 2 years. Symbols represent estimated marginal means, calculated by generalised estimating equations. Error bars represent standard errors of the marginal means.

Functional capacity

As reported in the main study HAQ scores improved significantly more during the first 2 years in the P group compared with the NoP group.23 This significant difference between the P and NoP groups was still present at 4 years (p = 0.034). Patients in remission at 2 years had a significantly lower HAQ score at both 2 and 4 years, irrespective of whether they were treated with prednisolone or not.

Bone mineral density

Baseline BMD and Z-scores at L2–L4 and femur neck are given in table 1 and show that the Z-score at L2–4 was significantly lower in the P group compared with that in the NoP group. The changes in these measures did not differ significantly between the treatment groups, neither for the first 2 years nor for the whole 4-year period. The changes in bone mineral did not differ between patients in remission at 2 years and those that were not, irrespective of treatment group (table 4).

Table 4 Change in bone mineral density (BMD) and Z-scores from baseline to 4 years for all patients who at the 2-year control point were either in remission or not, and separately for the two treatment groups

Adverse events

In the present study the incidence of serious adverse events was low and occurred during the first 2 years of treatment.23 Thus no incidence of hypertension, diabetes or weight gain was reported during years 2–4.


This study shows that patients randomised to prednisolone 7.5 mg daily in combination with DMARDs had a significantly higher remission rate compared with those randomised to no prednisolone. Furthermore, patients on prednisolone who were in remission after 2 years benefited from sustained reduced joint damage even at 4 years, which was an effect not seen in patients who were not in remission after 2 years. A post-hoc analysis suggested that the beneficial effect on joint damage of being in remission could be dependent on prednisolone use.

Irrespective of remission state, patients on prednisolone showed a lower increase in total Sharp score after 1 year compared with patients not on prednisolone, and this benefit in total Sharp score was maintained during the four study years. Furthermore, patients in remission after 2 years also had a significantly lower HAQ score, which was not dependent on prednisolone treatment.

There are three earlier reports concerning the effect of early prednisolone treatment on long-term joint damage. In the Arthritis & Rheumatism Council study patients taking 7.5 mg prednisolone for 2 years and then stopped, resumed the same radiographic progression rate as those not receiving prednisolone, but the damage was still less after a total of 3 years25 and thus in line with the results of the present study.

In the COBRA study the initial intensive treatment, which included prednisolone 60 mg/day, a dose that was tapered and stopped after week 28, resulted in a sustained lower rate of joint damage for more than 4 years longer, independent of subsequent antirheumatic treatment or disease activity after the first year.5 This effect was not seen in the group of patients receiving DMARD monotherapy only.

Also, at the 3-year follow-up in the Utrecht study of patients with early RA who initially were treated for 2 years with 10 mg prednisone without DMARDs, there was persisting inhibition of the radiological progression compared with the patients who were placebo-treated.30 In that study, prednisone was tapered and stopped if possible, but these patients still took that drug during 35% of the 3-year follow-up period.

The data in our study are compatible with an initial effect of prednisolone on radiographic progression (first year). Afterwards, the initially gained benefit was maintained irrespective of prednisolone use.

The differences in these studies concerning the prolonged effect of prednisolone might depend on different treatment strategies. However, the patients were not separated into those who achieved remission and those not. The possible effect on joint damage of being in remission could therefore not be evaluated.

The importance of going into remission has also been shown in the FIN-RACo trial; here, 42% of patients with early RA treated with a combination of DMARDs and 5 mg prednisolone were in remission after 2 years,11 using a modification of the American College of Rheumatology preliminary remission criteria.31 Patients in remission had a significantly lower Larsen score at 2 years compared with those not in remission. However, the possible influence of prednisolone treatment was not addressed.

Remission rates may vary in studies according to which remission criterion is applied and differences in patient selections,32 eg, male sex, absence of rheumatoid factor and a high disease activity have been associated with remission.3335 In the present study, such an association was only found for absence of rheumatoid factor in the whole patient cohort, but no differences in these variables were found at baseline between the treatment groups and could thus not explain the differences in remission rate.

The association of remission with radiological joint damage suggests a causal link between inflammation and structural damage. The data proposing that this link was dependent on prednisolone use are of interest suggesting that additional factors restricted to prednisolone play a role once disease activity is not driving progression any more. So far these factors are unknown. One possible explanation might be effects on the OPG:RANKL ratio as shown in the COBRA trial.36 The OPG:RANKL ratio and the first year time-averaged erythrocyte sedimentation rate independently predicted the 5-year radiographic progression of joint damage.

It is important to mention that clinical remission does not necessarily imply absence of radiographic progression. Some patients in complete clinical remission, for example, have been shown to have radiographic progression anyway,6 while on the other hand radiographic progression was completely absent in patients treated with tumour necrosis factor blockers but without any clinical response.37 38 These findings underscore that clinical disease activity and radiographic progression may be disconnected under certain circumstances.

The functional capacity, measured by HAQ, was also much improved after 4 years in patients randomised to prednisolone for the first 2 years compared with those that were not. However, the significant difference between patients in remission or those that are not may depend on the fact that HAQ is influenced by several conditions of which disease activity is the most important in early disease.39 Later on, the relationship between disability and structural damage becomes apparent, and this relationship is also close in early disease after correction for disease activity.40 The lower HAQ score in the present study in the P group might thus be a combined reduction of disease activity and joint damage.

In the present study the incidence of serious adverse events was low and all occurred during the first 2 years of treatment.23 This is in agreement with a recent review by da Silva et al.41 The fact that there was no difference in BMD between the treatment groups might be explained by the more efficient suppression of the inflammatory process accomplished by prednisolone.

Some limitation in the present study should be considered. Only 60% of the patients from the main BARFOT GC study participated in this extension study suggesting the possibility of selection bias, which reflects difficulties in long-term studies in “real life”. However, the baseline characteristics of the patients continuing in the extension study were similar to those of the patients who withdrew.

In summary, 7.5 mg prednisolone in addition to DMARDs in patients with early RA suppresses disease activity better and more patients go into remission than treatment with DMARDs alone. Furthermore, patients in remission after 2 years’ treatment with prednisolone had significantly less joint destruction after both 2 and 4 years. Therefore, remission should be the treatment goal, and treatment with low-dose prednisolone in combination with DMARDs should be considered in early RA.


We acknowledge research nurse Siv Norén for skilful data monitoring. This study has been supported by grants from the Swedish Rheumatism Association, King Gustaf V 80 years foundation, the Gorthon Foundation in Helsingborg and Stiftelsen för Rörelsehindrade i Skåne.



  • Competing interests: None.