The prevalence of obesity is markedly increasing world wide, which is distressing as obesity predisposes people to an increased risk of developing several diseases such as atherosclerosis, diabetes, certain cancers and osteoarthritis. In recent years it has been recognised that adipose tissue is not inert but generates soluble mediators. It produces a variety of general proinflammatory factors, such as tumour necrosis factor α (TNFα), interleukin 6 (IL6), monocyte chemoattractant protein1, as well as adipose tissue-specific proinflammatory or anti-inflammatory cytokines that are named adipocytokines.1 Adiponectin and leptin are the most abundantly produced adipocytokines. Other adipocytokines such as visfatin and resistin are also produced by adipocytes but are mainly synthesised in the macrophages that surround the adipocytes.1 Adipocytokines not only function as hormones that influence the energy homoeostasis but also affect immune functions and inflammatory processes. Proinflammatory factors produced by adipose tissue are active participants in the development of insulin resistance and cardiovascular disease.2 3 The field of adipocytokines has attracted tremendous interest as understanding the mechanisms that couple obesity and inflammation might provide targets for the development of new treatments to reduce the morbidity and mortality that is associated with obesity.

Rheumatoid arthritis (RA) is an inflammatory disease that is associated with increased levels of proinflammatory cytokines such as IL1, IL6 and TNFα. As obesity is considered to be characterised by systemic low-grade inflammation, it can be hypothesised that obesity increases the susceptibility to RA. A recent cross-sectional study investigating whether the prevalence of RA differed between obese and non-obese subjects, showed no differences.4 This suggests that either the proinflammatory effects that are associated with obesity in general are neutralised in arthritis, or that obesity does not predispose to RA. The first aim of this study was to increase further our understanding of the effect of obesity on the development of RA, by assessing whether the body mass index (BMI) of patients with undifferentiated arthritis (UA) is associated with progression to RA.

As in RA, increased levels of inflammatory markers such as C-reactive protein (CRP) are associated with the severity of joint damage5 and proinflammatory cytokines such as IL1 and TNFα are implicated in the breakdown of bone and cartilage, it can be assumed that adipocytokines produced by adipose tissue in obese subjects influence the clinical picture of RA and the level of radiological joint destruction. Therefore, this study investigates in two independent cohorts whether the BMI of patients with RA at disease onset is correlated with the level of radiological joint destruction during 3 years of follow-up.

PATIENTS AND METHODS

Study group

In this study the Leiden Early Arthritis Cohort (EAC) was used first, to determine the effect of BMI on both the progression from UA to RA and on RA severity. Second, patients with RA included in the BeSt study were used to replicate the observed associations between BMI and RA severity.

Leiden EAC

A detailed description of the Leiden EAC can be found in the paper by van Aken et al.6 In short, the EAC was started in 1993. Patients were referred to the EAC when arthritis was suspected and included when arthritis was found at physical examination. At baseline, questionnaires were filled in, tender and swollen joint counts were determined and blood samples and radiographs of hands and feet were taken. These procedures were repeated yearly during follow-up. At present more than 2000 patients with early arthritis are included.

Two weeks after inclusion, 311 patients had a diagnosis RA according to the 1987 American College of Rheumatology (ACR) criteria and 570 patients had arthritis that could not be classified according to one of the ACR criteria and were therefore identified as UA; the remaining patients had other classified diagnoses. After 1 year of follow-up, all 570 patients with UA were examined to determine whether they had developed RA according to the ACR criteria. One hundred and seventy-seven of 570 patients with UA had progressed to RA (fig 1). The patients with UA have been extensively described by van der Helm-van Mil et al.7 From the 570 patients with UA, data on weight and height was available for 383 patients. Inherent to the design of an inception cohort the duration of follow-up differs within the study population. However, at the moment of analysis the majority of patients (94%) had been followed up for more than 1 year (mean (SD) follow-up 8 (3) years) and only 9% of the patients who were not classified as RA at 1 year developed RA later on in the disease course.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1 Flow chart of the study. BMI, body mass index; EAC, Early Arthritis Cohort; RA, rheumatoid arthritis; UA, undifferentiated arthritis.

All 488 patients who were classified as RA (311 patients already fulfilled the ACR criteria during the first visit and 177 patients fulfilled the ACR criteria during the first year after inclusion) were evaluated to study the association between BMI and level of joint destruction (fig 1). Data on BMI were available for 332 of the total 488 patients with RA. From all patients with RA, four (1.3%) had a BMI <18.5 kg/m². To reduce a confounding effect of differences in treatment strategies determined by the level of joint destruction, this study used only the patients with RA who were included in the EAC before 2002 as treatment in this longitudinal cohort study is characterised by a secular trend. The patients with RA included between 1993 and 1995 were treated initially with analgesics and subsequently with chloroquine or sulfasalazine, the patients included between 1996 and 1998, were promptly treated with either chloroquine or sulfasalazine and the patients included after 1998 were promptly treated with either methotrexate or sulfasalazine. If the initial monotherapy failed, patients were switched to another drug or a second drug was added. Since 2002, initial combination therapy and treatment with biological agents has been used more commonly, and therefore the patients who entered the EAC after 2002 were not included in the analysis of the association between body mass and RA severity.

Replication cohort

The BeSt study is a multicentre, randomised controlled trial comparing the efficacy of four different treatment strategies in 508 patients with active recent-onset RA. A detailed description of the study design is provided elsewhere.8 The four treatment strategies were sequential monotherapy, step-up combination therapy, initial combination therapy with prednisone (COBRA) and initial combination therapy with infliximab. The first two treatment strategies resemble the treatment of the EAC before 2002 and the patients treated accordingly (n = 247) were used for replication. Inclusion criteria for the BeSt study were fulfilling the 1987 ACR criteria for RA, being disease-modifying antirheumatic drug (DMARD)-naïve, symptom duration <2 years and ⩾6 tender joints and ⩾6 swollen joints, and either an erythrocyte sedimentation rate >28 mm/1st h or a visual analogue scale global health >20 mm on a scale of 0–100 mm. At inclusion, the median symptom duration was 25 weeks (interquartile range 14–55) and the median disease duration 2 weeks (interquartile range 1–4). Treatment response was measured every 3 months using the Disease Activity Score in 44 joints (DAS44). If patients had a DAS >2.4, the next step in each treatment strategy was taken.

In the patients randomised to sequential monotherapy, initial treatment consisted of methotrexate (15 mg/week). In patients with a DAS >2.4, methotrexate was increased to 25–30 mg/week. Subsequent treatment steps were sulfasalazine monotherapy, leflunomide monotherapy, methotrexate with infliximab, and next different conventional DMARDs.

The patients assigned to step-up combination therapy also started with methotrexate (15 mg/week). If the response was insufficient, first methotrexate was increased to 25–30 mg/week, next sulfasalazine was added, followed by the addition of hydroxychloroquine and then prednisone. Failing this combination, patients switched to methotrexate with infliximab and next to different conventional DMARDs. Data on BMI were available for the whole of this replication cohort. Four patients (1.6%) had a BMI <18.5 kg/m².

Radiological joint destruction

In both cohorts, radiographs of hands and feet were made at baseline and yearly thereafter. The radiographs were scored for erosions and joint space narrowing according to the Sharp–van der Heijde method.9 The doctors who scored the radiographs were unaware of the clinical data and the study question. In the EAC the radiographs were scored by one reader. In the BeSt study, radiographs were scored by two readers and for each radiograph the mean score of the two readers was used. The interobserver intraclass correlation coefficient was 0.96.

Statistical analysis

The BMI was calculated by dividing the body weight (kg) by height squared and was normally distributed in both cohorts. To analyse whether BMI is associated with progression to RA, the patients with UA were divided in three groups according to their baseline BMI: BMI <25 kg/m² (normal weight), 25 to <30 kg/m² (overweight) and ⩾30 kg/m² (obesity). Differences in patient characteristics between these three groups were compared by the χ² test for proportions and by the Mann–Whitney test for continuous variables. The correlation between the BMI and the Sharp–van der Heijde score (both continuous variables) was calculated using the Pearson correlation coefficient. Linear regression analysis with a backward selection procedure was used to identify the variables that were independently associated with the level of joint destruction. In this analysis the Sharp–van der Heijde score at 3-year follow-up was entered as dependent variable and age, gender, CRP, number of swollen joints, presence of baseline erosions, anti-cyclic citrullinated peptide (anti-CCP) antibodies and BMI as possible explanatory variables.

The statistical package for the social sciences (SPSS) version 12.0 (SPSS, Chicago, IL, USA) was used to analyse the data. A p value ⩽0.05 was considered significant.

RESULTS

BMI and progression from UA to RA

To assess whether obesity is a risk factor for the development of RA, the progression to RA in patients with UA was assessed in relation to the baseline BMI. The baseline characteristics of the patients with UA whose BMI was missing were not significantly different from those whose BMI was known (data not shown), except for a lower frequency of anti-CCP positivity in patients with missing data on BMI (13% vs 25%). Of the 383 patients with UA with known BMI, 140 (37%) had progressed to RA after 1 year. At baseline, of the 383 patients with UA, 184 patients had a BMI <25 kg/m², 143 a BMI 25 to <30 kg/m² and 56 a BMI ⩾30 kg/m². In these three categories progression to RA was present in respectively 68 (37%), 53 (37%) and 19 patients (34%), showing no differences (p = 0.91). Assessing the percentage RA development in anti-CCP-positive and anti-CCP-negative patients with UA separately disclosed no significant differences between the BMI categories (data not shown). Also after correction for age, gender, CRP, number of swollen joints and anti-CCP antibodies in a logistic regression analysis, the BMI was not associated with the development of RA. In conclusion, obesity does not influence the risk of progressing from UA to RA.

BMI and RA severity

To investigate whether obesity influences the amount of joint damage in RA, the radiological determinant of joint destruction measured by the Sharp–van der Heijde method was studied in patients with RA from the EAC. Table 1 presents the characteristics of the patients with RA in each BMI group. The characteristics of the patients with RA whose baseline BMI was missing were not significantly different from those whose BMI was known (data not shown).

Assessment of the correlation between baseline BMI and the Sharp–van der Heijde score after 3-years’ follow-up showed an inverse correlation (r = −0.15̧ p = 0.025). Subsequent investigation of whether the BMI correlated with the total erosion score or with the total joint space narrowing score, or both, showed an inverse correlation for both variables (r = −0.14, p = 0.04 for erosions and r = −0.13, p = 0.05 for joint space narrowing). Given the recent evidence indicating that the pathophysiology of anti-CCP-positive RA differs from that of anti-CCP-negative RA,10 we assessed whether the association between BMI and joint destruction differed between patients with RA with and without anti-CCP antibodies. Intriguingly, the correlation between BMI and total Sharp–van der Heijde score after 3 years was only present in anti-CCP-positive patients with RA (r = −0.19, p = 0.04) and completely absent in anti-CCP-negative patients with RA (r = −0.03, p = 0.80). In all patients with RA, as well as in the anti-CCP-positive and anti-CCP-negative subsets, a linear regression analysis was performed to correct for factors known to be associated with RA severity, showing that the BMI was independently associated with a lower level of joint destruction in anti-CCP-positive RA but not in anti-CCP-negative RA (table 2).

Figures 2A–C depict the levels of joint destruction of the whole cohort as well as the anti-CCP-positive and -negative subgroups during 3 years of follow-up.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2 (A) Sharp–van der Heijde scores during 3 years of follow-up in patients with rheumatoid arthritis (RA) from the Leiden Early Arthritis Cohort (EAC) for the three different body mass index (BMI) categories. Presented are the mean (SEM). The numbers of patients in the BMI categories <25, 25 to <30 and ⩾30 kg/m² were 156, 146 and 30, respectively. (B) Sharp–van der Heijde scores during 3 years of follow-up in anti-cyclic citrullinated peptide (anti-CCP)-positive patients with RA from the Leiden EAC for the three different BMI categories. Presented are the mean (SEM). The numbers of patients in the BMI categories <25, 25 to <30 and ⩾30 kg/m² were 79, 78 and 5, respectively. (C) Sharp–van der Heijde scores during 3 years of follow-up in anti-CCP-negative patients with RA from the Leiden EAC for the three different BMI categories. Presented are the mean (SEM). The number of patients in the BMI categories <25, 25–<30 and ⩾30 kg/m² were 60, 52 and 12, respectively.

Replication of the association between BMI and RA severity

To replicate the findings on the correlation between BMI and joint damage, an independent cohort of 247 patients with RA treated by comparable strategies were studied. Table 3 presents the baseline characteristics of these patients.

Assessing the correlation between baseline BMI and the Sharp–van der Heijde score after 3-years’ follow-up showed an inverse correlation (r = −0.27̧ p<0.001). Both the total erosion score and the total joint space narrowing score showed an inverse correlation with the level of joint destruction (r = −0.24, p<0.001 and r = −0.27, p<0.001, respectively). Separate analyses in the anti-CCP-positive and anti-CCP-negative patients showed only a significant correlation in the anti-CCP-positive patients with RA (r = −0.32, p<0.001 versus r = −0.06, p = 0.63 in the anti-CCP-negative group). In linear regression analyses, the BMI was independently associated with the amount of joint destruction in the anti-CCP-positive subgroup, in contrast to the anti-CCP-negative subgroup (table 4). Figures 3A–C depict the levels of joint destruction of the whole replication cohort, as well as the anti-CCP-positive and -negative subgroups, during 3 years’ follow-up.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3 (A) Sharp–van der Heijde scores during 3 years of follow-up in patients with rheumatoid arthritis (RA) from the replication cohort for the three different body mass index (BMI) categories. Presented are the mean (SEM). The numbers of patients in the BMI categories <25, 25 to <30 and ⩾30 kg/m² were 108, 98 and 35, respectively. (B) Sharp–van der Heijde scores during 3 years of follow-up in anti-CCP-positive patients with RA from the replication cohort for the three different BMI categories. Presented are the mean (SEM). The numbers of patients in the BMI categories <25, 25 to <30 and ⩾30 were 64, 55 and 19, respectively. (C) Sharp–van der Heijde scores during 3 years of follow-up in anti-CCP-negative patients with RA from the replication cohort for the three different BMI categories. Presented are the mean (SEM). The numbers of patients in the BMI categories <25, 25–<30 and ⩾30 kg/m² were 34, 38 and 12, respectively.

In conclusion, in two independent cohorts of patients with early RA the BMI is independently and inversely correlated with the degree of joint damage in anti-CCP-positive patients.

DISCUSSION

This study shows a paradoxical effect of a high BMI in RA: although obesity is generally considered to be a risk factor for proinflammatory disorders such as atherosclerosis and diabetes, in RA a high BMI correlated with a less destructive disease course. Additionally, obesity was not associated with the risk to progress from UA to RA.

We observed no association between BMI and RA development, which corresponds to a similar finding in a recent cross-sectional study.4 This suggests that in obese patients with UA the net result of the secreted inflammatory mediators by the adipose tissue is balanced and/or does not affect the susceptibility to RA. Conversely, when patients have developed RA obese patients have a less severe disease course. A similar observation was made in a small cohort of 54 patients with RA.11 In this study patients with a baseline BMI <27 kg/m² developed more radiological progression.11 Indirect evidence that body mass modifies the severity of RA can also be derived from a recent study of Escalante et al.12 These authors showed that obesity in RA is associated with decreased mortality. This effect was not explained by a high mortality in lean patients (due to rheumatoid cachexia). Moreover, this protective effect disappeared after correction for RA severity and comorbidity,12 indicating that the protective effect was mediated by these two variables. Unfortunately, the authors did not correct for RA severity and comorbidity separately. Nevertheless, these data are also supportive of a protective effect for obesity in RA.

In the present study the correlation between BMI and joint damage was only observed in anti-CCP-positive patients. This may indicate that mediators excreted by the adipose tissue differentially affect the mechanisms driving erosiveness and cartilage degradation in anti-CCP-positive and anti-CCP-negative patients with RA. However, a more likely explanation is that an effect is more easily seen when the variation in joint destruction is larger. Since the anti-CCP-positive patients with RA had a more severe disease course, this may have increased the ability to detect significant differences. In contrast, the anti-CCP-negative subgroup had minimal joint destruction.

Generally in RA, less joint damage correlates with better physical functioning measured by a low Health Assessment Questionnaire score. We did not observe any relation between the BMI categories and Health Assessment Questionnaire scores (data not shown). This may be because a BMI ⩾30 kg/m² itself is associated with a decreased physical functioning in healthy obese people13 and obese patients with RA.14 Therefore, the current paper does not indicate that a high BMI has a beneficial effect on the level of disability.

Joint destruction in RA is composed of bone erosions and cartilage degradation (radiologically visualised as joint space narrowing), produced by osteoclasts and fibroblast-like synoviocytes. Both cell types are activated by proinflammatory cytokines such as IL1 and TNFα. This raises the question what mechanisms underlie the beneficial effect of obesity on bone and cartilage breakdown in RA. Factors with anti-inflammatory activity are likely mediators. Although not statistically significant, patients with a BMI ⩾30 kg/m² tended to have a lower CRP and a smaller number of swollen joints. The linear regression analyses controlled for the effects of potential confounders such as measures expressing the disease activity (number of swollen joints and CRP) on the rate of joint destruction, and showed that the BMI was independently associated with a less destructive disease course. Nevertheless, the possibility that the favourable effect of obesity on the level of joint destruction, on CRP concentration and on the number of swollen joints originates from the same casual or anti-inflammatory pathway(s) cannot be excluded.

The only (known) adipocytokine with anti-inflammatory activities is adiponectin, raising the interest of this adipocytokine in the light of the current observations. Adiponectin influences both the innate and adaptive immune system as it reduces the levels of interferon γ and TNFα, increases the level of IL10 and has a negative effect on T-cell proliferation and B-cell lymphopoiesis.1 Synoviocytes express adiponectin, and in RA the synovial fluid level of adiponectin is increased and negatively associated with the local leucocyte count,15 which suggests that adiponectin may counteract the local inflammatory process in RA. An alternative anti-inflammatory mediator is oestrogen; obesity is generally associated with an increased level of bioavailable oestrogen, and oestrogens have fairly potent anti-inflammatory properties and positively influence bone turnover.16 Alternative explanations might also be proposed but the current study did not provide data explaining the observed associations.

In conclusion, our study showed in two independent cohorts that in RA the presence of a high BMI is independently associated with a less severe disease outcome. Determining the biological mechanism underlying this correlation is of great interest as it may result in new therapeutic options for RA.