SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. REFERENCES
  9. Supporting Information

Objective

Obesity is a prevalent condition and a serious health concern. The relationship between obesity and rheumatoid arthritis (RA) disease activity and severity has not been adequately examined, and there are concerns that periarticular adipose tissue may reduce the utility of the joint examination.

Methods

We used a cross-sectional study to compare the performance of swollen joint count (SJC) in subjects with RA across body mass index (BMI) strata. Specifically, regression techniques tested for associations of SJC and 7 RA disease activity/severity measures (including high-sensitivity C-reactive protein level, radiographic changes, and Multidimensional Health Assessment Questionnaire scores) within BMI quartiles. We also evaluated the association of BMI with radiographic evidence of RA in multivariate analyses and the association of BMI with SJC. Clinical and laboratory data from 980 Veterans Affairs Rheumatoid Arthritis registry participants were analyzed using linear and logistic regression.

Results

Associations were evident between SJC and 6 of the 7 examined RA disease activity/severity measures. SJC predicts RA disease activity/severity in more obese subjects at least as well as in subjects with lower BMIs, and there was a trend toward better performance in individuals with higher BMIs. Subjects with higher BMIs were marginally less likely to be characterized by radiographic changes (odds ratio 0.98, P = 0.051). We found no association between BMI and SJC.

Conclusion

BMI does not obscure the relationship of SJC and objective disease activity measures. There is a borderline association of higher BMI and the likelihood of radiographic changes characteristic of RA after controlling for clinical characteristics.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. REFERENCES
  9. Supporting Information

Obesity is an extraordinarily common medical condition (1), and competes with smoking as the leading cause of morbidity and mortality among adults in the US (2, 3). The health care cost attributable to obesity approaches $150 billion annually in the US alone (4).

Prior investigators have suggested that higher body mass index (BMI) did not influence the progression to clinical rheumatoid arthritis (RA) (5), but rather that lower BMI correlates with increased radiographic progression (5–7), particularly in seropositive patients. Interestingly, the investigators were unable to identify any correlation between levels of physical function estimated by Health Assessment Questionnaire scores and BMI (5), or between the swollen joint count (SJC) and BMI (6). These preliminary epidemiologic data leave many unresolved questions that are of importance to the practicing rheumatologist.

In addition, the evaluation of obese patients with RA is complicated by practical concerns. For example, there is the question of whether SJCs are obscured by periarticular adipose tissue, or whether the latter simulates synovitis in more obese patients.

To address these questions, we performed a cross-sectional study to compare the performance of SJC in subjects with RA across BMI strata. That is, we asked whether the relationship of SJC to disease activity/severity measures remains consistent across subjects with different BMIs. We also evaluated the association of BMI with RA radiographic changes and the association of BMI with SJC. We hypothesized that increased adipose tissue, reflected by higher BMI, would render lower correlations of SJC with measures of RA disease activity/severity.

Significance & Innovations

  • The swollen joint count performs in more obese subjects at least as well as in subjects with lower body mass index (BMI), and there was a trend toward better performance in individuals with higher BMIs.

  • Subjects with lower BMIs may be more likely to demonstrate the radiographic changes of rheumatoid arthritis than subjects with higher BMIs.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. REFERENCES
  9. Supporting Information

Study design, setting, and participants.

We performed a cross-sectional substudy of the prospective longitudinal Veterans Affairs Rheumatoid Arthritis (VARA) registry. Data were derived from 6 participating US Department of Veterans Affairs Medical Centers: Denver, Colorado; Omaha, Nebraska; Salt Lake City, Utah; Washington, DC; Dallas, Texas; and Jackson, Mississippi. All subjects were adults who fulfilled the 1987 American College of Rheumatology revised classification criteria for RA (8) and were enrolled in the VARA registry.

Data sources/measurement.

The VARA registry consists of a longitudinal clinical repository and biorepository of serum and plasma established in 2003, with collection currently occurring at 11 sites geographically distributed across the US. Clinical characteristics and outcomes of participating veterans are similar to clinical practice (9), as well as other RA registries (10). Data were derived from questionnaires completed by subjects at the time of registry enrollment and at subsequent scheduled clinical encounters. Health information collected by providers was also entered into the VARA clinical database by study staff at the time of outpatient clinical encounters. These data were supplemented by inpatient and outpatient administrative data.

Ethics.

Each participating site received institutional review board approval prior to initiation of the study. Patients consented and completed agreements to disclose health information at the time of enrollment. An independent scientific ethics advisory committee approved this substudy.

Variables/definitions.

RA patient characteristics.

Patients described their educational history and tobacco use, and submitted written estimates for their pain (0–10 scale) and overall functional status according to the Multidimensional Health Assessment Questionnaire (MD-HAQ; 0–3 scale) (11). Clinicians recorded the presence of erosions or unequivocal periarticular osteopenia on radiographs (ever), the presence of rheumatoid nodules (ever), 28 tender joint count (TJC) scores, 28 SJC scores, erythrocyte sedimentation rate (ESR; mm/hour) results determined by the local clinical laboratory, and provider assessments of global disease severity (0–10 scale). Given the longitudinal nature of the registry, we calculated mean values for all RA patient characteristics for each subject.

Biologic measures.

Serum collected at registry enrollment was analyzed to determine anti–cyclic citrullinated peptide IgG antibody (anti-CCP) concentrations using a second-generation enzyme-linked immunosorbent assay kit (Diastat, Axis-Shield Diagnostics; positivity at >5 units/ml). Rheumatoid factor (RF; positivity at ≥15 IU/ml) and high-sensitivity C-reactive protein (hsCRP; mg/liter) levels were determined by nephelometry (Siemens Healthcare Diagnostics).

BMI.

A single random BMI measurement acquired by clinical staff during usual care was abstracted from the administrative data for each subject, since we did not have access to repeat BMI measurements for each individual. To account for the difference in timing between the BMI and SJC measures, we determined the median date on which SJC was recorded for each subject, and then calculated the difference in time between this median SJC and the date BMI was measured (the “measurement time gap”). This measurement time gap was included in the analyses, as described below.

Statistical methods.

To address the issue of bias, comparisons were made between those registry participants with available BMIs (i.e., subjects used in the analysis) and those without BMIs. This was performed using the Student's t-test for continuous variables and the chi-square test for dichotomous variables.

The cohort was divided into quartiles based on patient BMI. For each BMI quartile, linear and logistic regression was used to determine the association of mean SJC with a number of disease activity/severity measures, including MD-HAQ score, radiographic changes, nodules, hsCRP level, ESR, RF, and anti-CCP. Because we were interested in the biologic relationship of high BMI and disease activity/severity, as well as the technical challenge of performing accurate joint counts in overweight/obese patients, we selected more objective measures as our outcomes. For that reason, we did not focus on TJC or patient outcomes such as pain or patient global assessments of well-being.

To determine whether SJC is better associated with disease activity/severity measures in subjects with lower BMIs, we used the Cuzick nonparametric test for trend across ordered groups to compare the number of RA disease activity/severity measures with a statistically significant association in each BMI quartile (12). To assess whether variations in the time between SJC measurement and BMI may have affected the outcome, we performed a series of univariate linear and logistic regressions evaluating the relationship of BMI to each of the 7 disease activity/severity characteristics listed above, with and without controlling for the measurement time gap. Lastly, we employed multivariate logistic regression to evaluate the association of BMI and the presence of radiographic damage, controlling for age, sex, current tobacco use, and RF positivity. The threshold for significance was established at P values less than 0.05 and confidence bands set at 95%. Analyses were conducted using Stata, version 11 (StataCorp).

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. REFERENCES
  9. Supporting Information

There were no differences between subjects with available BMIs (n = 980) and those without BMIs (n = 372) with regard to sex, current tobacco use, average MD-HAQ score, TJC, SJC, hsCRP level, ESR, pain score, and provider global assessments (P > 0.1 for each). There were slight differences in age (67.5 years versus 65.0 years; P < 0.01), education (12.7 years versus 13.2 years; P < 0.01), RF positivity (81% versus 70%; P < 0.01), and anti-CCP IgG antibody positivity (76% versus 60%; P < 0.01).

The study cohort was comprised of 980 individuals. Demographic characteristics for the cohort are shown in Table 1. In general, subjects were elderly men (mean ± SD age 67.47 ± 11.06 years), reflecting the national composition of US veterans. Nearly one-half of the subjects had rheumatoid nodules at some time during their course and ∼80% exhibited positive RF serology, while 76% were anti-CCP positive. The median duration of RA at the time of BMI measurement was 11.1 years (interquartile range 5.96–20.16 years).

Table 1. Demographics and clinical characteristics of subjects with rheumatoid arthritis (n = 980)*
 NValue
  • *

    Values are the mean ± SD (range) unless otherwise indicated. hsCRP = high-sensitivity C-reactive protein; MD-HAQ = Multidimensional Health Assessment Questionnaire; ESR = erythrocyte sedimentation rate; anti-CCP = anti–cyclic citrullinated peptide antibody.

Age, years98067.47 ± 11.06 (25–96)
Men, %98090.51
Education, years84512.73 ± 2.64 (1–20)
Current tobacco use, %98026.73
hsCRP level, mg/liter97912.76 ± 19.87 (0–163)
Radiographic changes present, %93837.42
Rheumatoid nodules present, %87448.97
MD-HAQ score (range 0–3)9780.97 ± 0.53 (0–2.85)
ESR, mm/hour97826.44 ± 19.41 (0–119.33)
Provider global assessment (range 0–100 mm)92233.82 ± 17.36 (0–100)
Tender joint count (range 0–28)9804.41 ± 4.84 (0–28)
Swollen joint count (range 0–28)9803.42 ± 3.3 (0–25)
Patient pain score (range 0–10)9774.42 ± 2.09 (0–9.8)
Rheumatoid factor positive, %97980.90
Rheumatoid factor, IU/ml979348.42 ± 732.04 (9.4–8,720)
Anti-CCP IgG antibody positive, %97576.41
Anti-CCP IgG antibody, units/ml975274.12 ± 447.96 (0–5,429.6)

The distribution of BMIs for the cohort is shown in Supplementary Figure 1 (available in the online version of this article at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)2151-4658). The mean BMI was 28.2 kg/m2. The cutoffs for the quartiles were 24 kg/m2 (25%), 27.3 kg/m2 (50%), and 31.4 kg/m2 (75%).

Table 2 shows the associations between provider-assessed SJC and objective RA disease activity/severity measures, separated according to patient BMI quartile. These results, for example, demonstrate a persistent relationship between SJC and hsCRP level across all strata of patient BMI. Therefore, for each additional swollen joint detected by a clinician, the hsCRP level is increased by a mean of 0.62–1.55 mg/liter, depending on the BMI quartile. The differences between BMI categories, however, were not statistically different, since 95% confidence intervals (95% CIs) overlap. Similarly, associations were detectable within various strata for all examined RA disease activity/severity measures, with the exception of ESR.

Table 2. Association of swollen joint count and various rheumatoid arthritis disease activity/severity characteristics, stratified by patient BMI quartile*
BMI quartileCoefficient/ORSEP95% CI
  • *

    Linear regression was used for continuous dependent variables (high-sensitivity C-reactive protein [hsCRP], Multidimensional Health Assessment Questionnaire [MD-HAQ], erythrocyte sedimentation rate [ESR], rheumatoid factor, and anti–cyclic citrullinated peptide [anti-CCP]) and logistic regression was used for dichotomous outcome variables (presence of radiographic changes and presence of nodules). BMI = body mass index; OR = odds ratio; 95% CI = 95% confidence interval.

  • P < 0.05.

hsCRP level, mg/liter    
 11.320.460.0040.42, 2.21
 21.550.480.0010.61, 2.49
 30.620.260.0190.10, 1.13
 40.670.300.0270.08, 1.26
Radiographic changes present    
 11.080.040.0571.00, 1.16
 21.090.050.080.99, 1.20
 31.090.050.0421.00, 1.18
 41.110.040.0061.03, 1.20
Rheumatoid nodules present    
 11.040.040.3220.96, 1.13
 21.130.070.0441.00, 1.26
 31.070.050.1490.98, 1.17
 41.090.040.0371.00, 1.17
MD-HAQ score    
 10.000.010.649−0.01, 0.02
 20.030.010.0110.01, 0.05
 30.040.01< 0.0010.02, 0.06
 40.030.010.0020.01, 0.05
ESR, mm/hour    
 10.040.040.273−0.03, 0.11
 20.070.050.122−0.02, 0.16
 30.020.030.586−0.05, 0.08
 40.050.030.126−0.01, 0.11
Rheumatoid factor, IU/ml    
 10.000.000.1350.00, 0.01
 20.010.000.0010.00, 0.01
 30.000.000.0680.00, 0.01
 40.000.000.0160.00, 0.01
Anti-CCP, units/ml    
 1−5.169.770.598−24.40, 14.08
 2−1.968.540.819−18.78, 14.87
 317.458.280.0361.14, 33.76
 411.657.980.146−4.07, 27.37

The highest BMI quartile manifested an association with 5 of the 7 disease activity/severity measures, the second and third quartiles demonstrated an association with 4 measures, and the patients in the lowest BMI quartile only demonstrated an association with hsCRP. The number of associations for each stratum trended toward significance according to the Cuzick nonparametric test for trend across ordered groups (P = 0.1). We found no association of BMI (kg/m2) with SJC using univariate linear regression (coefficient = −0.21 [95% CI −0.55, 0.01], P = 0.23). The inclusion of the measurement time gap had no effect on the relationship of BMI to the 7 examined disease activity/severity characteristics; the coefficients and P values for each of these disease activity/severity characteristics remained unchanged, regardless of whether the measurement time gap was included.

Finally, our multivariate analysis examining the association of BMI and radiographic damage (Table 3) revealed a borderline association between higher BMI and lower risk of radiographic damage (odds ratio [OR] 0.98, P = 0.051). For each point increase in BMI, the odds of radiographic damage may decline by 2%. Restricting the analysis to anti-CCP–positive patients produced virtually identical results (OR 0.97, P = 0.050).

Table 3. Results of multivariate logistic regression to evaluate the association of BMI and the presence of radiographic changes, after controlling for age, sex, tobacco use, and RF*
 ORSEP95% CI
  • *

    BMI = body mass index; RF = rheumatoid factor; OR = odds ratio; 95% CI = 95% confidence interval.

  • P < 0.05.

BMI, kg/m20.980.010.0510.95, 1.00
Age, years1.020.010.0261.00, 1.03
Male sex0.770.190.2710.48, 1.23
Current tobacco use0.780.130.1330.56, 1.08
RF positivity1.380.250.0750.97, 1.96

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. REFERENCES
  9. Supporting Information

We hypothesized that increased adipose tissue, reflected in higher patient BMIs, would diminish the correlations of SJC with measures of RA disease activity/severity. Our data, however, suggest that SJC demonstrates associations with RA disease activity/severity in subjects with elevated BMIs at least as well as in patients with lower BMIs. The fact that SJC performs at least as well in subjects in higher BMIs (and that there is a trend toward superior performance compared to subjects with low BMIs) should be reassuring to clinicians relying on SJC as part of a comprehensive assessment strategy. It also raises the question of why the SJC did not perform better in subjects with low BMIs. We can speculate that lower BMIs may reflect a population with “rheumatoid cachexia,” where systemic disease involvement or concurrent comorbid illness (e.g., cancers) predominates over synovial disease (13).

Because our registry only collects 28 joint counts, our findings do not provide any insight into the question of whether more extensive joint counts perform differently relative to subjects' BMIs. The difficulty of assessing metatarsophalangeal joints and potential variations in adiposity between these joints and metacarpophalangeal joints may lead to divergent results if more extensive joint counts had been performed. However, a study based on 2 cohorts that collected 44 and 53 joint counts found an inverse association of BMI and the Sharp/van der Heijde score, which is consistent with our findings (5).

There is increasing interest in the interplay of adipose tissue and the immune system (14, 15). Studies have found that synovial adipocytokines are associated with the presence of RA and synovial leukocyte count (16), and in some studies, plasma levels of adipocytokines differ between RA patients and healthy controls (17). The clinical implications of these findings have not previously been well articulated.

Our analyses did suggest a weak inverse association of BMI and radiographic damage, which parallels the findings of prior studies (5, 6). Although these earlier studies were unable to correlate BMI with the presence of erosions, they did identify an association between BMI and radiographic progression. Our cross-sectional study found no association of BMI and SJC, which is also consistent with the study by Westhoff et al (6).

Since BMIs are not normally distributed in the VARA registry or the general population, classification of our study cohort based on commonly defined, but somewhat arbitrary, World Health Organization (WHO) categories (underweight, normal, overweight, and obese) would have resulted in groups of differing sizes. These unequally-sized strata would have critically threatened our results by biasing our analyses; larger strata would be more likely to demonstrate statistical relationships, simply based on increased power. Similarly, if our registry population had been artificially sampled to produce equal numbers of subjects in each WHO category, such a distribution might introduce a collection bias and not reflect the real-world BMIs. We therefore elected to define our strata in a manner that ensured similarly sized groups (i.e., by BMI quartiles).

The limitations of our study should be acknowledged. First, cross-sectional studies cannot establish causality, only associations. Second, although reflective of obesity, BMI is not an optimal measure of adiposity. Use of alternate and perhaps more optimal approaches, such as bioimpedance measurement, was not available for this cohort. Similarly, imaging modalities, such as musculoskeletal ultrasound, may improve the assessment of synovitis and inform future studies. Third, our sample size limited our ability to evaluate the putative role that medication exposures (e.g., glucocorticoids) and comorbidities (e.g., diabetes mellitus) may have played in this study. Fourth, the absence of BMI measurements in some VARA registry participants excluded them from our analysis, and this may have led to bias. The clinical differences among those included and excluded, however, were generally small. Lastly, our study was performed in a cohort primarily comprised of elderly men, which may affect the generalizability of our findings. It is important to note, however, that many of our findings corroborate the work of earlier investigators.

Our study was based on prospectively collected data and took place within a comprehensive health care system, thereby increasing the likelihood that the relevant clinical data were captured. Additionally, the sample size for this study compares favorably with those of prior investigations, allowing us the opportunity to control for clinical characteristics that might have biased or confounded earlier studies. The enrollment of subjects from multiple sites may enhance the generalizability of our findings.

The results of this study have potential clinical implications, by reassuring clinicians that SJCs perform in patients with high BMIs as well as in patients with low BMIs. In addition, the inverse association of BMI and radiographic damage may prompt clinicians to perform more aggressive radiographic surveillance of joint disease in individuals with lower BMIs.

AUTHOR CONTRIBUTIONS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. REFERENCES
  9. Supporting Information

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. Caplan 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. Caplan, Bright.

Acquisition of data. Caplan, Davis, Kerr, Lazaro, Khan, Richards, Johnson, Cannon, Reimold, Mikuls.

Analysis and interpretation of data. Caplan, Davis.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. REFERENCES
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. REFERENCES
  9. Supporting Information

Additional Supporting Information may be found in the online version of this article.

FilenameFormatSizeDescription
ACR_21734_sm_SupplFig1.doc35KSupplementary Figure 1

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.