To assess the association between high body mass index (BMI) and treatment response in recent-onset rheumatoid arthritis.
To assess the association between high body mass index (BMI) and treatment response in recent-onset rheumatoid arthritis.
In the Behandelstrategieën voor Reumatoide Artritis (Treatment Strategies for Rheumatoid Arthritis) study, 508 patients were randomized to initial monotherapy or combination therapy with prednisone or infliximab (IFX). The response to Disease Activity Score (DAS) ≤2.4–steered treatment (first dose and after 1 year) was compared between patients with a BMI <25 kg/m2 and ≥25 kg/m2, using relative risk (RR) regression analyses. DAS, components of DAS, and functional ability during the first year were compared using linear mixed models.
High BMI was independently associated with failure to achieve a DAS ≤2.4 on initial therapy (RR 1.20 [95% confidence interval (95% CI) 1.05, 1.37]). The effect for combination therapy with prednisone was RR 1.55 (95% CI 1.06, 2.28) and for combination therapy with IFX 1.42 (95% CI 0.98, 2.06). The RRs for failure after 1 year were 1.46 (95% CI 0.75, 2.83) and 2.20 (95% CI 0.99, 4.92), respectively. High BMI was also associated with failure on delayed combination therapy with IFX, after adjustment for selection bias related to previous failure on disease-modifying antirheumatic drugs. No significant association was observed in the initial monotherapy groups. In the first year, patients with a high BMI had higher DAS and worse functional ability, with more tender joints and a higher visual analog scale global health, but not more swollen joints and similar systemic inflammation.
High BMI was independently associated with failure to achieve low DAS on initial combination therapy with prednisone and on initial and delayed treatment with IFX. Patients with a high BMI experienced more pain, but not more swelling or systemic inflammation.
An association between treatment response to tumor necrosis factor (TNF) blockers and body mass index (BMI; kg/m2) was described in a group of patients with established rheumatoid arthritis (RA) who had failed treatment on disease-modifying antirheumatic drugs (DMARDs). Patients with a high BMI responded less well to treatment with a fixed dose of the TNF blocker infliximab (IFX) (). This finding was replicated in patients who had failed on methotrexate (MTX) and were treated with a fixed dose of adalimumab, etanercept, or IFX (). Patients with a high BMI and thus a higher fat mass might show more inflammation ([3, 4]). Yet, clinical synovitis might be less easy to assess in RA patients with a high BMI. It has also been described that patients with various conditions and a high BMI report more pain than patients with a normal or low BMI ([5-7]).
In the Behandelstrategieën voor Reumatoide Artritis (Treatment Strategies for Rheumatoid Arthritis) (BeSt) trial, a treatment to target trial in early RA patients, treatment response in terms of Disease Activity Score (DAS) and patient-reported outcomes was assessed every 3 months and yearly radiographs were taken. Because different treatment strategies were used, we were able to analyze the association between BMI and different components of treatment response not only to TNF blockers, but also to conventional DMARD monotherapy or combination therapy.
Patients from the BeSt cohort, a study originally designed to compare 4 different treatment strategies in early DMARD-naive RA patients, were analyzed. Patients were randomized to 1 of 4 treatment groups: 1) sequential monotherapy starting with MTX, 2) step-up combination therapy starting with MTX, 3) initial combination therapy, with the COBRA (Combinatietherapie Bij Reumatoïde Artritis) scheme: MTX, sulfasalazine (SSZ), and tapered high dose prednisone, or 4) a combination of MTX and IFX.
Treatment was DAS-steered and aimed at a DAS of ≤2.4, resulting in 3 monthly treatment adjustments as long as the DAS was >2.4. Thus, in groups 1–3, delayed IFX treatment was initiated if treatment had failed on at least 3 synthetic DMARDs, including MTX, SSZ, leflunomide (in arm 1) or hydroxychloroquine (in arm 2), and prednisone (in arms 2 and 3). In all arms, DMARD treatment was changed or added to at least twice in case of insufficient response (DAS >2.4) before MTX + IFX combination therapy was started. Patients treated with MTX + IFX started IFX in a dosage of 3 mg/kg/every 8 weeks, but if the DAS remained >2.4, the IFX dosage was escalated from 3 mg/kg/2 months to 6 mg, 7.5 mg, and finally 10 mg/kg if necessary. If the highest dose did not lead to a low DAS, MTX + IFX were abandoned and the next treatment initiated. At any stage of the protocol, if patients achieved a DAS ≤2.4 for ≥6 months, treatment was tapered to maintenance dose: MTX monotherapy in groups 1 and 2, SSZ monotherapy in group 3, and MTX + IFX 3 mg/kg/2 months in group 4. More details on the treatment protocol were published previously ().
Treatment response (failure was defined as not achieving a DAS ≤2.4) was compared between normal weight patients (BMI <25 kg/m2) and overweight or obese patients (BMI ≥25 kg/m2) (). Both height and weight were assessed at baseline and were measured by a research nurse, using professional calibrated scales to measure weight and wall-based measure rods to measure height. Treatment response was assessed at 2 time points. First, we looked at whether or not patients achieved a DAS ≤2.4 after the first 3 months of treatment. Second, we looked at failing responses (DAS >2.4) in year 1, on treatment step 1 and 2: MTX monotherapy (15 mg/week, if necessary increased to 25 mg/week) in groups 1 and 2, on combination therapy with prednisone (MTX 7.5 mg, if necessary increased to 25 mg/week) in group 3, and on treatment steps 1, 2, or 3 (MTX + IFX increased from 3 mg, 6 mg to 7.5 mg/kg/2 months) in group 4. The different cutoff for group 4 was chosen because, based on DAS evaluations before each IFX dose, treatment could be intensified every 2 months as compared to every 3 months in the other groups. We also looked for a relation between BMI and clinical response to treatment with MTX + IFX in patients who had failed treatment on previous synthetic DMARDs in groups 1–3. After 8 years of treatment, the number of protocolized treatment steps that patients had failed on was recorded in the initial treatment groups. Radiologic damage progression was assessed using the Sharp/van der Heijde score, taking the mean of the scores of 2 independent readers, blinded for patient identity, who evaluated all the radiographs of hands and feet in nonchronological order.
Statistical analyses were performed with the software program SPSS, version 17.0, and STATA, version 12. Baseline characteristics were compared between patients with normal and high BMI using the Student's t-test, the Mann-Whitney U test, or the chi-square test. To determine whether a higher BMI was associated with impaired response to therapy according to the definitions above, a relative risk (RR) regression model was used, where the parameters were estimated using a modified Poisson regression approach with robust SEs (). These analyses give risk ratios, which are easier to interpret than odds ratios. The analyses were adjusted for sex, age, smoking habits, rheumatoid factor (RF), and baseline DAS. Then the regression analyses for treatment response were repeated and stratified for treatment group (groups 1 and 2, group 3, and group 4). The association between BMI and failure to achieve a DAS ≤2.4 on delayed IFX was examined in patients from groups 1–3 who received MTX + IFX after failing on several DMARDs. Differences in baseline characteristics in this group, associated with response to DMARDs, were observed between patients with low or normal and high BMI, indicating that there might be a selection bias. Therefore, propensity scores with age, RF, alcohol use (yes/no), treatment group, baseline erythrocyte sedimentation rate (ESR), number of swollen joints, visual analog scale (VAS) global and morning stiffness as predictors, and high BMI as outcome were calculated using logistic regression. Then, to correct for the differences between patients with normal and high BMI, a risk regression model was fitted with the weighting based on the estimated propensity score, i.e., 1/propensity score for patients with high BMI and 1/(1 − propensity score) for patients with normal BMI. Weights >5 were truncated at 5. We repeated the analyses with BMI as a (linear) continuous variable. Using the likelihood ratio test, we tested whether there was evidence of a nonlinear association by comparing the likelihoods of a model with BMI as linear covariate and the likelihood of a model where BMI was modeled with fractional polynomials. To find out whether there was a difference in disease manifestation in the first year of treatment between the BMI categories of the various DAS components or in patient-reported outcomes, linear mixed models were fitted. The following dependent variables were used in the different models: tender joint count, swollen joint count, ESR, C-reactive protein (CRP) level, patients' assessment of global health (VAS global) and of pain (VAS pain), and Health Assessment Questionnaire (HAQ) score. In each of the models, time and BMI category were entered as categorical covariates and the baseline value of the dependent variable as continuous covariate. The interaction between time and BMI was not significant in any of the analyses, therefore it was not included in the final models. The estimates were adjusted for sex, age, RF, and smoking habits.
The number of treatment steps on which patients had failed after 8 years was compared using the Mann-Whitney U test.
Patients with a BMI ≥25 kg/m2 were older than patients with a BMI <25 kg/m2 (56 years versus 53 years; P = 0.03) and were less often smokers (31% versus 41%; P = 0.01) (Table 1). No other significant differences in baseline characteristics were observed. A BMI ≥30 kg/m2 was observed in 15% of all patients.
|BMI <25 kg/m2 (n = 216)||BMI ≥25 kg/m2 (n = 292)||P|
|Female, no. (%)||155 (72)||188 (64)||0.08|
|Age, years||53 ± 15||56 ± 13||0.03|
|BMI, kg/m2||23 ± 2||29 ± 3||< 0.001|
|Symptom duration, median (IQR) weeks||23 (13–57)||23 (14–47)||0.7|
|ACPA positive, no. (%)||131 (65)||160 (59)||0.2|
|Rheumatoid factor positive, no. (%)||149 (69)||180 (62)||0.09|
|Disease Activity Score||4.4 ± 0.8||4.4 ± 0.9||0.4|
|Health Assessment Questionnaire score||1.4 ± 0.6||1.4 ± 0.7||0.4|
|CRP level, median (IQR) mg/liter||20 (8–55)||21 (9–50)||0.96|
|ESR, median (IQR) mm/hour||38 (20–56)||34 (18–56)||0.4|
|Tender joint count, median (IQR)||13 (9–17)||13 (9–19)||0.3|
|Swollen joint count, median (IQR)||13 (10–19)||14 (9–18)||0.8|
|VAS global health||51 ± 20||54 ± 20||0.09|
|VAS physician||58 ± 18||57 ± 18||0.6|
|VAS pain||54 ± 21||55 ± 22||0.3|
|Smoking, no. (%)||88 (41)||89 (31)||0.02|
High BMI was an independent predictor of failing (not achieving a DAS ≤2.4) on the first treatment step with an RR of 1.20 (95% confidence interval [95% CI] 1.05, 1.37) (Table 2). A minor effect was observed for failing on treatment steps in year 1 (steps 1 and 2 in groups 1–3 or steps 1, 2, and 3 in group 4) with an RR of 1.15 (95% CI 0.92, 1.43). Analyses were repeated with BMI as a continuous variable, and these results confirm the findings of the dichotomized analyses. High BMI was again an independent predictor of failing on the first step (RR 1.03; 95% CI 1.01, 1.06) and for failing on treatment steps in year 1 (RR 1.02; 95% CI 1.01, 1.04) (Table 3). Adding nonlinear terms in BMI to the model did not improve the fit (P = 0.99).
|Crude RR (95% CI)||Adjusted RR (95% CI)a|
|Fail on initial treatment step (all)||1.20 (1.04, 1.38)b||1.20 (1.05, 1.37)b|
|Fail on first dose MTX monotherapy||1.10 (0.96, 1.25)||1.10 (0.97, 1.25)|
|Fail on initial dose MTX + SSZ + prednisone||1.57 (1.02, 2.41)b||1.55 (1.06, 2.28)b|
|Fail on initial dose MTX + IFX||1.37 (0.93, 2.02)||1.42 (0.98, 2.06)|
|Fail in year 1|
|All||1.13 (0.89, 1.43)||1.15 (0.92, 1.43)|
|Groups 1 and 2||1.04 (0.82, 1.31)||1.05 (0.84, 1.30)|
|Group 3||1.37 (0.68, 2.75)||1.46 (0.75, 2.83)|
|Group 4||2.12 (0.93, 4.83)||2.20 (0.99, 4.92)|
|Crude RR (95% CI)||Adjusted RR (95% CI)a|
|Fail on initial treatment step (all)||1.025 (1.011, 1.040)b||1.023 (1.010, 1.037)b|
|Fail on first dose MTX monotherapy||1.016 (1.002, 1.030)b||1.016 (1.003, 1.029)b|
|Fail on initial dose MTX + SSZ + prednisone||1.052 (1.014, 1.092)b||1.049 (1.014, 1.085)b|
|Fail on initial dose MTX + IFX||1.029 (0.99, 1.071)||1.028 (0.99, 1.064)|
|Fail in year 1|
|All||1.032 (1.008, 1.056)b||1.029 (1.005, 1.055)b|
|Groups 1 and 2||1.022 (1.002, 1.043)b||1.020 (0.998, 1.043)b|
|Group 3||1.063 (0.97, 1.165)||0.99 (0.99, 1.160)|
|Group 4||1.039 (0.97, 1.114)||1.041 (0.98, 1.112)|
After 8 years of DAS-steered treatment, the median (interquartile range) number of treatment steps patients had failed on was 1 (0–3) for patients with a BMI <25 kg/m2 and 2 ([1-4]) for patients with a BMI ≥25 kg/m2 (P < 0.001). The percentage of patients who after 8 years were no longer treated according to protocol due to failing on all treatment steps was not different (26% versus 22%; P = 0.4).
In groups 3 and 4, a higher risk of impaired response to therapy for patients with a high BMI was found with RRs of 1.55 (95% CI 1.06, 2.28) and 1.42 (95% CI 0.98, 2.06) for response to the first dose. For group 3, the RR for response to the first 2 treatment steps in year 1 was 1.46 (95% CI 0.75, 2.83). The effect of impaired response in patients with a high BMI was stronger in group 4 (RR 2.20; 95% CI 0.99, 4.92). In groups 1 and 2, no significant association between treatment response and BMI was observed.
For patients initially treated with MTX + IFX in group 4 (n = 120), demographic or disease characteristics between patients with a high and low BMI were similar at baseline (data not shown). In contrast, patients with a BMI ≥25 kg/m2 who received MTX + IFX in groups 1–3 were less often positive for anti–citrullinated protein antibody and RF (57% versus 83% and 66% versus 90%; P = 0.004 and 0.002, respectively) (see Supplementary Table 1, available in the online version of this article at http://onlinelibrary.wiley.com/doi/10.1002/acr.21978/abstract). They were older than patients with a BMI <25 kg/m2 (mean age 51 years versus 46 years; P = 0.02). There were 32 patients with a BMI >30 kg/m2. Of these, only 9 patients (28%) responded well to medication after 1 year. Of the patients in groups 1–3 with a BMI <30 kg/m2, 89 of 193 (46%) responded well.
However, in crude analyses no association was seen between BMI and response to treatment in patients from groups 1–3 who received delayed MTX + IFX (RR 1.11; 95% CI 0.71, 1.73) for response to first dose, and a trend was seen for response after 1 year (RR 1.56; 95% CI 0.80, 3.04). After adjusting for the misbalance in the baseline characteristics using propensity weighting, the RR of failure to the first dose changed to 1.37 (95% CI 0.81, 2.31), and the RR of failure after 1 year changed to 2.09 (95% CI 0.97, 4.49).
In year 1 and adjusted for baseline differences, patients with high BMI had higher disease activity (difference in DAS 0.30 [95% CI 0.15, 0.45]), a higher HAQ score (difference 0.14 [95% CI 0.05, 0.23]), and a higher VAS pain (difference 6.2 mm [95% CI 3.0, 9.4]). For DAS components, a difference was found in tender joints (difference 1.4 [95% CI 0.6, 2.2]) and patient's assessment of global health (VAS difference 4.9 mm [95% CI 1.9, 7.8]), but not for swollen joints (difference 0.6 [95% CI −0.02, 1.2]) (Table 4 and Figure 1). Radiologic damage progression in year 1 and over 8 years followup was similar in patients with high or low/normal BMI (Figure 2).
|Unadjusted difference (95% CI)||Adjusted difference (95% CI)a|
|DAS||0.25 (0.10, 0.40)||0.30 (0.15, 0.45)|
|HAQ score||0.13 (0.04, 0.21)||0.14 (0.05, 0.23)|
|VAS global||4.4 (1.5, 7.3)||4.9 (1.9, 7.8)|
|ESR, mm/hour||0.9 (−1.3, 3.1)||1.3 (−0.9, 3.5)|
|CRP, mg/liter||0.1 (−2.2, 2.3)||0.7 (−1.5, 2.9)|
|TJC||1.1 (0.4, 1.9)||1.4 (0.6, 2.2)|
|SJC||0.5 (−0.1, 1.1)||0.6 (−0.02, 1.2)|
|VAS pain||5.4 (2.3, 8.6)||6.2 (3.0, 9.4)|
In this DAS-steered treated cohort with early RA patients, high BMI was associated with failure to achieve a low DAS (≤2.4) on antirheumatic therapy, including after adjustment for confounders. This was most noticeable in patients who were treated with initial combination therapy with MTX, either combined with prednisone and SSZ, or with IFX. The association between high BMI and failure of treatment remained if the dose of MTX or IFX was increased. After stratification for initial therapy (initial monotherapy with MTX in groups 1–2, initial combination therapy with MTX, SSZ, and prednisone in group 3, or MTX and IFX in group 4), patients with a high BMI who were treated with initial combination therapy were more likely to show a decreased response to treatment than patients with a normal BMI. This association was still seen after 1 year, after failure on the initial treatment had led to dose increases (of MTX in group 3 and of IFX in group 4), but less so in group 3 than in group 4. High BMI was also associated with failure to achieve a low DAS on delayed treatment with IFX in patients who had failed treatment on at least 3 conventional DMARDs. Due to more failure to achieve a low DAS on treatment, patients with high BMI went through significantly more treatment steps over 8 years of DAS-targeted treatment than patients with a low/normal BMI. Failure to achieve a low DAS depended mainly on the pain and joint tenderness scores, which were higher in the patients with a high BMI, whereas joint swelling and laboratory parameters of inflammation were similar in patients with high or low/normal BMI.
Recently, Klaassen et al reported that patients with a high BMI responded less well to delayed treatment with fixed-dose IFX after failure on a median of 2 DMARDs (). It has been suggested that this may be due to high levels of proinflammatory cytokines produced by adipocytes ([3, 4]). Our results confirm that patients with a high BMI fail treatment more often on IFX, also as initial treatment, and also if the dosages are increased up to 10 mg/kg/every 8 weeks. Thus, a failure to respond on IFX in patients with higher BMI is not due to underdosing, which is also theoretically unlikely, since IFX is dosed per kilogram and the drug remains mainly in the intravascular space (), the volume of which can increase with higher BMI (). However, our data also show patients with a high BMI fail more often on treatment with a combination of MTX, SSZ, and prednisone and on subsequent treatment steps during 8 years of DAS ≤2.4–targeted treatment. Only in patients treated with initial MTX monotherapy, patients with higher BMI did not fail to achieve a low DAS more often than patients with low/normal BMI. This might be related to the fact that, in general, failure on initial MTX monotherapy was more common than on initial combination therapy, which makes it harder to analyze the role of individual risk factors.
Rather than being the result of a high ESR or swollen joint counts, the higher DAS scores in patients with higher BMI appear to depend on pain. Higher pain scores and worse global health were also reported in patients with a high BMI in a large Swedish cohort (). In that study, patients with a BMI ≥30 kg/m2 also had a higher ESR and CRP level at followup. We found no association between a high BMI and higher parameters of inflammation or more joint swelling, but there were very few patients with a BMI ≥30 kg/m2.
It is possible that we underestimated joint swelling in patients with a high BMI (). The higher tender joint counts in patients with a high BMI might still reflect more local inflammation. We previously reported that local joint tenderness is a predictor of local joint damage after 1 year, independent of swelling (). This in fact supports the practice of using a composite score such as the DAS as treatment target, not merely joint swelling. We found no differences in joint damage progression after 8 years of DAS-targeted treatment in patients with high or low/normal BMI. This may be due to more treatment adjustments (because of higher DAS) in patients with high BMI, or there may be another reason why patients with high BMI appear to be protected against joint damage progression ([15, 16]). It may also be that the pain experienced by patients with high BMI does not reflect inflammation. We did not do routine assessments of fibromyalgia features, but we cannot exclude that a fibromyalgia component was present in part of these patients. Self-reported pain, especially musculoskeletal pain, is higher in patients with a high BMI, in particular with a BMI ≥30 kg/m2, and they are more likely to report pain in multiple locations ([5, 6]). The mechanism of the relationship between obesity and pain is unclear, but it is suggested that disturbances in neurotransmitters and hormones might be, at least partially, responsible (). This relation between BMI and pain may also influence the association between high BMI and functional disability, which was found in this cohort. Pain and body size itself may both interfere with the daily activities that are listed in the HAQ ().
In conclusion, in the DAS ≤2.4–targeted BeSt study we found that RA patients with a higher BMI fail more often than patients with low/normal BMI to achieve a low DAS on antirheumatic treatment. This resulted in more treatment adjustments over time. The higher DAS scores were mainly dependent on joint tenderness and self-reported pain and well-being, and were associated with less functional ability, but not with more damage progression over time.
In treatment to target strategies, finding a high DAS based on inflammation or on noninflammatory pain may have different therapeutic consequences. Additional research, including advanced imaging techniques and biomarker studies, may further elucidate the relation between BMI and failure to treatment, thus helping us to decide how we can best treat our individual patients.
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 submitted for publication. Dr. Heimans 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. Han, Kerstens, Huizinga, Lems, Allaart.
Acquisition of data. Van den Broek, Riyazi, Han, Kerstens, Lems, Allaart.
Analysis and interpretation of data. Heimans, van den Broek, le Cessie, Siegerink, Huizinga, Allaart.
Schering-Plough and Janssen had no role in the study design or in the collection, analysis, or interpretation of the data, the writing of the manuscript, or the decision to submit the manuscript for publication. Publication of this article was not contingent upon approval by Schering-Plough and Janssen.
We would like to thank all the patients, as well as the following rheumatologists (other than the authors), who participated in the Foundation for Applied Rheumatology Research (all locations are in The Netherlands): W. M. de Beus, MD (Medical Center Haaglanden, Leidschendam); C. Bijkerk, MD, A. J. Peeters, MD (Reinier de Graaf Gasthuis, Delft); M. H. W. de Bois, MD, G. Collée, MD (Medical Center Haaglanden, The Hague); F. C. Breedveld, Prof (Leiden University Medical Center, Leiden); J. A. P. M. Ewals, MD, H. M. J. Hulsmans, MD (Haga Hospital, The Hague); A. H. Gerards, MD, P. A. H. M. van der Lubbe, MD (Vlietland Hospital, Schiedam); J. H. L. M. van Groenendael (Franciscus Hospital, Roosendaal); J. M. W. Hazes (Erasmus University Medical Center, Rotterdam); M. H. de Jager, MD (Albert Schweitzer Hospital, Dordrecht); J. M. de Jonge-Bok, MD (retired); M. V. van Krugten, MD (Walcheren Hospital, Vlissingen); H. van der Leeden, MD (retired); M. F. van Lieshout-Zuidema, MD, J. Ph. Terwiel, MD (Spaarne Hospital, Hoofddorp); A. Linssen, MD (retired); C. Mallée, MD, K. S. S. Steen, MD, S. ten Wolde, MD (Kennemer Gasthuis, Haarlem); E. T. H. Molenaar, MD (Groene Hart Hospital, Gouda); H. C. van Paassen, MD, J. M. G. W. Wouters, MD, D. van Zeben, MD (Sint Franciscus Gasthuis, Rotterdam); H. K. Markusse, MD (deceased); R. M. van Soesbergen, MD (retired); P. B. J. de Sonnaville (Admiraal de Ruyter Hospital, Goes); I. Speyer, MD, M. L. Westedt, MD (Bronovo Hospital, The Hague); A. E. Voskuyl, MD (VU Medical Center, Amsterdam). We would also like to thank all other rheumatologists and trainee rheumatologists who enrolled patients in this study, and all research nurses for their contributions.