SEARCH

SEARCH BY CITATION

Keywords:

  • Rheumatoid arthritis;
  • Osteoporosis;
  • Vertebral deformities;
  • Corticosteroids;
  • Bone mineral density

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

Objective

To examine the occurrence of vertebral deformities in female patients with rheumatoid arthritis (RA), and the relationship between vertebral deformities and bone mineral density (BMD) and between vertebral deformities and clinical variables.

Methods

Lateral radiographs of the spine were obtained in 229 female patients with RA (mean age 63.4 years, range 51.4–73.6 years) recruited from a county RA register. Vertebral deformities were measured semiquantitatively by an experienced radiologist. A clinical examination including core measurements of disease activity and severity was performed, and BMD was measured at the spine (L2–L4) and hip.

Results

According to the statistical analysis, 49 patients were considered to have relevant vertebral deformities. The occurrence of vertebral deformities was independently associated with age, long-term corticosteroid use, and previous nonvertebral fracture, as well as reduced BMD. Our results failed to show any independent relationship between vertebral deformities and the activity or severity of disease.

Conclusion

Corticosteroid use is an important marker of established osteoporosis in patients with RA. Additionally, there seems to be a consistent relationship between BMD and vertebral deformities in this patient group.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

Several studies have demonstrated an increased frequency of osteoporosis in female patients with rheumatoid arthritis (RA) (1–3). However, fractures represent the clinically important aspect of osteoporosis. Vertebral deformities are associated with increased morbidity and mortality and an augmented risk of developing additional osteoporotic fractures (4, 5). Inconsistent data have been published regarding associations between clinical variables and vertebral deformities in RA, especially the relative contribution of reduced bone mineral density (BMD), corticosteroid use, immobilization, and disease activity (6–8). No studies regarding vertebral deformities have been performed in cohorts suggested to be representative of the total population with underlying RA.

The aim of this study was to describe the occurrence of vertebral deformities in a fairly representative cohort of female patients with RA, and the association between vertebral deformities and BMD and clinical and demographic variables.

PATIENTS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

Patients and setting.

The inclusion criteria for this study included enrollment in the Oslo RA Register, female sex, Caucasian race, disease duration of at least 2 years, and year of birth between 1926 and 1948. The Oslo RA Register, validated to have a completeness of 85%, consists of patients fulfilling the American College of Rheumatology (formerly, the American Rheumatism Association) 1987 revised classification criteria for RA (9), and with a residential address in the county of Oslo (10). A total of 511 patients in the register fulfilled the inclusion criteria. Three hundred fifty-eight (70.1%) of these underwent a comprehensive clinical examination, comprising commonly used assessments of disease activity, disease severity, and health status. A cohort of 244 (47.7%) agreed to participate in a substudy regarding established osteoporosis in RA. Fifteen patients were excluded because radiographs were not obtained or were not available for examination, and 229 (44.3%) were finally included in the study. The local ethics committee approved this study. The Data Inspectorate had previously approved the register of RA patients in Oslo

Data collection.

All measurements were performed during the years 1998 through 2000. The demographic, patient, and disease characteristics listed in Table 1 were recorded partly by self-reported questionnaires and partly by interview and clinical examination, performed by a specially trained research nurse in cooperation with a rheumatologist (REØ or GH). The following variables were collected: age at menopause, age at menarche, number of children, oophorectomy, use of estradiol, smoking, use of disease-modifying antirheumatic drugs (DMARDs), and use of corticosteroids. Joint pain, fatigue, and patient's and investigator's global assessment of disease activity were measured on a 100-mm visual analog scale. Self-reported physical disability was assessed by the 8-item Modified Health Assessment Questionnaire (MHAQ; range 1–4). Joint assessments included the swollen joint count (28 joints), the tender joint count (28 joints), and the deformed joint count (18 joints). The Disease Activity Score was computed using the 28-joint count (11). Patients having a rheumatoid factor (RF) titer ≥64 measured on at least one occasion during the disease course were considered to be RF positive.

Table 1. Demographic and disease variables for the total sample and compared between patients with and without vertebral deformities*
VariablesAll patients (n = 229)No deformity (n = 180)Deformity (n = 49)P
  • *

    Except where indicated otherwise, values are the mean ± SD. A patient with at least 2 mild deformities or 1 moderate/severe deformity was classified as having deformities. BMI = body mass index; MHAQ = modified health assessment questionnaire.

  • Values are given without corrections for multiple analyses, because groups of variables are closely correlated. Individual values close to the limit of 0.05 should thus be interpreted with caution.

  • Ever use of ≥7.5 mg for ≥6 months.

  • §

    Reduced bone mass is defined as Z score ≤−1 SD, and osteoporosis as a T score ≤−2.5 SD.

Demographic    
 Age, years63.3 ± 6.6762.2 ± 6.567.7 ± 5.2<0.001
 Height, cm164.4 ± 6.0165.2 ± 5.9161.7 ± 5.8<0.001
 Weight, kg66.5 ± 12.666.4 ± 12.566.8 ± 12.80.86
 BMI, kg/m224.6 ± 4.524.3 ± 4.325.6 ± 5.00.08
 Current user of estradiol, %43.446.432.70.09
 Current smoker, %32.730.540.80.17
Clinical    
 Disease duration16.7 ± 10.416.1 ± 10.319.3 ± 10.30.05
 Rheumatoid factor positive, %52.549.762.50.12
 MHAQ score1.67 ± 0.541.63 ± 0.491.81 ± 0.680.08
 Disease Activity Score4.78 ± 1.224.74 ± 1.194.91 ± 1.720.40
 Erythrocyte sedimentation rate22.7 ± 18.420.7 ± 16.129.9 ± 23.50.01
 Global assessment score42.2 ± 23.741.1 ± 23.246.5 ± 25.20.16
 Tender joint count, 28 joints9.1 ± 6.79.2 ± 6.68.6 ± 7.00.55
 Swollen joint count, 28 joints7.6 ± 5.47.7 ± 5.17.3 ± 6.40.60
 Deformed joint count, 18 joints4.5 ± 5.14.0 ± 4.96.2 ± 5.40.009
 Previous nonvertebral fracture, %29.323.351.0<0.001
 Corticosteroid use    
  Current user, %49.144.167.30.004
  Ever user, %70.367.281.60.05
  Long-term ever user (≥12 months), %45.947.279.6<0.001
  Ever user of higher doses, %14.410.729.90.001
DEXA measurements§    
 BMD femoral neck, gm/cm20.806 ± 0.1520.826 ± 0.1490.729 ± 0.137<0.001
 BMD total hip, gm/cm20.836 ± 0.1580.858 ± 0.1560.752 ± 0.139<0.001
 BMD L2–L4, gm/cm21.054 ± 0.2021.080 ± 0.2010.956 ± 0.174<0.001
 Reduced bone mass femoral neck, %25.823.335.60.09
 Reduced bone mass total hip, %30.925.053.3<0.001
 Reduced bone mass of the spine, %21.018.330.60.06
 Osteoporosis of the femoral neck, %20.317.431.30.04
 Osteoporosis of the total hip, %16.613.433.30.002
 Osteoporosis of the spine (L2–L4), %21.818.334.70.01

Assessment of vertebral deformities.

Vertebral deformities (T4–L5) were scored by an experienced radiologist using a standardized semiquantitative method, as described by Genant et al (12). Deformed vertebrae were classified as grade 1 (mild), grade 2 (moderate), or grade 3 (severe), representing a reduction in any of the vertebral heights (anterior, posterior, or middle) of 20–25%, 25–40%, and >40%, respectively. For the anterior and middle heights, the corresponding posterior height was used as reference. For quality assurance, blinded rescoring was done after 6 months, involving 90 patients of whom 50% had at least 1 vertebral deformity. Kappa values were 0.75 for whether a patient was classified as having any deformity and 0.86 for whether a patient had a moderate or severe deformity.

BMD measurements.

The BMD measurements of the hip (total hip and femoral neck) and the lumbar spine (L2–4, anterior–posterior) were performed by 4 trained technicians using the same dual-energy x-ray absorptiometry equipment (Lunar Expert, Madison, WI). In 12 patients, no hip measurement was performed due to bilateral hip replacements. BMD values were compared with a pooled European/US reference database, including data for T score estimation and equations for computing age- and weight-adjusted Z score estimations. Detailed descriptions of the reference database and the equations have been published previously (3).

The spine phantom coefficient of variation (CV%) calculated from daily measurements for the whole measurement period was 0.7%. The in vivo reproducibility of BMD measurements was assessed on the basis of duplicate measurements of 42 healthy hospital workers (mean age 55.1 years, range 50.4–66.0 years). The CV% was 1.7% for the femoral neck, 1.3% for the total hip, and 2.4% for L2–4. Analyzing the results from 30 randomly selected patients with RA assessed the interobserver variation between the 4 trained technicians performing the measurements. The readings of each technician were compared in a pairwise manner. For BMD measurements at the femoral neck, the interobserver variation ranged from 0.5% to 1.6%, at the total hip from 0.4% to 0.9%, and at the lumbar spine from 0.9% to 1.3%.

Statistical analysis.

The presence of ≥2 mild deformities or at least 1 moderate/severe deformity was required in order to classify an individual patient as having a vertebral deformity. Osteoporosis was defined as a T score ≤2.5 SD, and reduced bone mass was defined as a Z score ≤ −1 SD below the age- and weight-adjusted reference values. Patients classified as having or not having vertebral deformities were compared regarding demographic and clinical variables and BMD measurements, using 2-sided t-tests for continuous variables and chi-square tests for counts. Possible predictors of vertebral deformities were subsequently entered into a logistic regression analysis, applying the presence of vertebral deformities as a dependent variable. The inclusion criteria for independent variables in the logistic regression analysis were statistical significance in the bivariate analysis and a supposed clinical relevance for the variable. Statistical analyses were performed using the SPSS (Chicago, IL) program, version 10.0.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

Patient characteristics.

Table 1 shows the demographic and clinical characteristics of the patients included in the analyses. This group of RA register patients (n = 229) was slightly younger than the RA register patients who were not included (n = 282) (mean difference 1.5 years; 95% confidence interval [95% CI] 0.38, 2.62). There were no statistically significant differences regarding disease duration (mean difference 1.13 years, 95% CI −0.67, 2.93) or RF positivity (mean difference 4.7%, 95% CI −4.0, 13.2). Among patients who met for clinical examination (n = 358), no statistically significant differences were found between those finally included in the study (n = 229) and those who did not undergo BMD measurements for the following clinical variables: MHAQ score (mean difference 0.02, 95% CI −0.09, 0.13), disease activity score (mean 0.01, 95% CI −0.29, 0.30), current use of prednisolone (mean difference 7.3%, 95% CI −1.2, 16.0), or ever use of DMARDs (mean difference 4.7%, 95% CI −3.9, 13.5). Thus, it is considered that the patients included in the present study are fairly representative of the entire underlying patient population regarding demographic and disease characteristics

Frequency of vertebral deformities.

The total number of deformities among the vertebrae evaluated was 168; of these, 87 were classified as moderate or severe. Twenty of 3,206 vertebrae (0.6%) could not be evaluated. Figure 1 shows the distribution of deformities across different vertebrae of the spine and displays that most vertebral deformities were identified in the middle thoracic and the thoraco-lumbar region. Seventy-three patients (31.9%) had ≥1 mild, moderate, or severe deformities, and 39 (17.0%) had ≥1 moderate/severe deformities. Forty-nine patients (21.4%) had at least 2 mild or one moderate deformities and according to our criteria were classified as subjects with definite vertebral deformities.

thumbnail image

Figure 1. Distribution of vertebral deformities along the spine.

Download figure to PowerPoint

Frequency of osteoporosis and reduced bone mass.

Mean BMD values at the 3 measurement sites are shown in Table 1. The frequency of osteoporosis at the femoral neck was 20.3%, at the total hip was 16.6%, and at the spine (L2–4) was 21.8%. The corresponding numbers for reduced bone mass (Z score ≤−1 SD) were 25.8%, 30.9%, and 21.0%, respectively.

Comparison of patients with or without vertebral deformities.

Table 1 gives bivariate comparisons of demographic and clinical variables between patients classified as having or not having vertebral deformities. Vertebral deformities were statistically significantly associated with older age, longer disease duration, higher erythrocyte sedimentation rate (ESR), number of deformed joints, history of previous nonvertebral fracture, and current or long-term use of corticosteroids. Vertebral deformities were also negatively associated with height, but there were no statistically significant differences between the 2 groups regarding body mass index (BMI) or weight. Patients with vertebral deformities had lower BMD at all 3 measurement sites, more frequently had osteoporosis at all measurement sites, and reduced bone mass of the total hip.

Possible risk factors for vertebral deformities.

In logistic regression analysis, the presence of vertebral deformities was independently associated with age, long-term corticosteroid use (>12 months), a history of nonvertebral fracture, and low bone mass at the total hip (Table 2). ESR and the presence of deformed joints were entered into the model but were not selected. In other models, substituting bone mass of the total hip with the other measurement sites, it appeared that femoral neck and spine L2–4 were not significantly associated with vertebral deformities. However, when replacing the dichotomized Z scores with BMD as a continuous variable, the BMD value of any of the 3 measurement sites was independently associated with vertebral deformities (data not shown).

Table 2. Possible risk factors for vertebral deformities*
 βSEOR95% CI
  • *

    Logistic regression (forward stepwise) analysis (n = 211). SE = standard error; OR = odds ratio; 95% CI = 95% confidence interval; ESR = erythrocyte sedimentation rate; SD = standard deviation.

  • 12 months or more.

Age0.150.361.171.09–1.25
Long-term corticosteroid use1.490.454.151.70–10.07
Previous nonvertebral fracture1.020.412.781.24–6.22
Deformed jointsNot selected   
ESR >20 mm/hourNot selected   
Z score ≤−1 SD total hip1.180.413.261.45–7.35
Constant−13.072.50  

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

This is, to our knowledge, the largest, single-center study on vertebral deformities in female patients with RA, examining their association with clinical and demographic variables as well as BMD measurements. We found that the presence of vertebral deformities was independently associated with age, long-term corticosteroid use, and previous nonvertebral fractures. The results also imply that reduced bone mass is weakly but independently associated with vertebral deformities in patients with RA.

We chose a standardized, semiquantitative method for identifying vertebral deformities (12). There is no consistence regarding which grade of change should be defined as a deformity, and the number of deformities found will differ according to the method and cut-off chosen (13, 14). For the statistical analyses, we chose to define clinically relevant vertebral deformities as at least 2 mild or 1 moderate/severe deformity, an approach that has also been used in a previous study of vertebral deformities in RA (4). A semiquantitative method will detect more deformities than does quantitative morphometry if a reduction of 3 SD is chosen for the latter method, and it may be questioned if some of these deformities may be considered clinically important (15). However, several studies have demonstrated that mild deformities are also associated with morbidity and increased risk of future additional fractures (8, 9).

Some preliminary analyses were done before deciding on the models used for logistic regression analysis. Especially, we explored the significance of adding weight, height, or BMI, because these factors have been shown to relate to both BMD and fracture risk. As shown in Table 1, there was no significant relationship between vertebral deformities and either weight or BMI. The strong negative association between height and vertebral deformities remained when examined by logistic regression but had no influence on the significance of the other variables in the equation (data not shown). We thereby decided to exclude this value, because height reduction may be a consequence of deformities by themselves.

A few earlier studies have described possible predictors of vertebral deformities in RA, and especially the relationship to use of corticosteroids (6–8). In 2 of 3 studies, there was no significant relationship between vertebral deformities and various measurements of corticosteroid use (6, 8). However, in a recent Italian multicenter study, Sinigaglia et al (7) found a significant relationship between cumulative corticosteroid dose and vertebral deformities. Furthermore, several larger epidemiologic studies on patients receiving corticosteroids for various reasons recently demonstrated an increased, dose-dependent risk of both vertebral and nonvertebral fractures (16, 17).

Of all the parameters tested in our study, except for age, long-term use of corticosteroids showed the strongest association with vertebral deformities. We chose long-term use as the variable of interest because reliable data on cumulative dose were not available. Because vertebral deformities are to be regarded as an irreversible event, long-term use will also include patients with a history of previous treatment that could have contributed to deformities, even though the corticosteroid-induced effect on BMD has been shown to be partly reversible (18). Exchanging long-term use with prednisolone (doses >7.5 mg for longer than 6 months) did not significantly change the results. The dose of 7.5 mg is considered by many clinicians and researchers to be the threshold for deleterious effects on bone metabolism and is used as a cut-off for recommending therapeutic interventions to prevent corticosteroid-induced bone loss (19). In our study, however, a much smaller number of patients had used corticosteroids in such high doses, compared with the number of long-time users (33 versus 124 patients), reflecting that low doses of corticosteroids over long periods of time are frequently used by patients with RA. Our data support that there is no “safety threshold” for steroid dose, but that the risk of fractures increases over the whole range of doses.

Exposure to corticosteroids could reflect long-term disease severity and activity rather than being a risk factor in itself. We explored this by adding various disease variables to the multivariate analyses. Unfortunately, radiographic damage scores for our sample of patients are not available. The deformed joint count is considered an alternative damage variable that could reflect the suggested common disease mechanisms for osteoporosis and joint destruction (20, 21). However, neither disease activity variables nor the presence of deformed joints was independently associated with vertebral deformities.

We observed a strong and independent association between a history of previous nonvertebral fracture and vertebral deformities. Previous fractures were self-reported, and no attempt was made to verify this information. This method has, however, been proven to be reliable in at least 2 previous studies (22, 23).

Our data, suggesting that BMD measurements are independently, but weakly, associated with vertebral deformities, is consistent with other data from samples of patients with RA or corticosteroid-treated patients (6, 7, 24). The fact that BMD values of the total hip showed a relationship to vertebral deformities that was an equal to or stronger than that of BMD of the spine (L2–4) might be due to degenerative or arthritic changes in the spine of patients with RA. This observation is supported by an earlier study in patients from the Oslo RA register showing no reduction of BMD values in the spine compared with controls (3). A possible limitation to this part of our study was that we did not exclude fractured vertebrae from the BMD analysis. However, such adjustments are rarely done in a clinical setting.

In conclusion, the clear and consistent relationship between the use of corticosteroids and vertebral deformities demonstrated in this study should alert physicians to be especially aware of this complication in patients with RA. This is important, taking into account that corticosteroids are effective both as symptom-modifying and disease-modifying agents in RA. A history of earlier nonvertebral fractures is also relevant when considering the risk of having vertebral deformities, as well as reduced BMD. Our study was, however, unable to show any consistent relationship between various clinical variables and vertebral deformities in RA. Studies with a longitudinal design are required to examine whether the associations identified in this cross-sectional study also can serve as risk factors for future fractures and vertebral deformities.

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

We thank Petter Mowinckel for statistical advice. We gratefully appreciate our technicians Margareth Sveinsson, Ingerid Müller, Sidsel Arnkværn, and Anne Katrine Kongtorp for expert technical assistance.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES