To evaluate the extent of and risk factors for bone loss in a population-based cohort of patients with rheumatoid arthritis (RA) receiving conventional health care.
To evaluate the extent of and risk factors for bone loss in a population-based cohort of patients with rheumatoid arthritis (RA) receiving conventional health care.
In a longitudinal study, clinical data were collected and bone mineral density (BMD) measurements were performed at baseline and after 2 years. Dual-energy x-ray absorptiometry was used for hip and spine BMD measurements. At baseline, patients received advice about lifestyle adjustments and calcium and vitamin D supplementation; during the followup period they were treated with antirheumatic and bone-sparing drugs, according to clinical judgment.
After a mean ± SD of 2.2 ± 0.2 years, 366 (298 women, 68 men) of the 488 patients who were examined at baseline were reexamined. At that time, 47.9% were current users of corticosteroids and 37.0% were using antiresorptive drugs (hormone replacement therapy, bisphosphonates, or calcitonin). The mean BMD reduction was −0.64% in the femoral neck, −0.77% in the total hip, and −0.29% in the spine at L2-4. BMD was increased at all measurement sites in current users of antiresorptive drugs (0.16–1.64%) but was decreased in patients using calcium and vitamin D alone (−1.99% to −1.39%) and in patients not using any osteoporosis treatment (−1.20% to −0.43%). Current use of corticosteroids was independently associated with increased risk for BMD loss in the total hip (odds ratio [OR] 2.63, 95% confidence interval [95% CI] 1.38–5.00) and spine at L2-4 (OR 2.70, 95% CI 1.30–5.63), whereas current use of antiresorptive drugs was associated with decreased risk for bone loss in the total hip (OR 0.43, 95% CI 0.20–0.89).
Results of this population-based, 2-year followup study indicate that adequate management of patients with RA, addressing both the rheumatic disease and osteoporosis, protects against bone loss.
Osteoporosis occurs more frequently in patients (both male and female) with rheumatoid arthritis (RA) than in healthy individuals (1, 2). Treatment of RA with disease-modifying antirheumatic drugs (DMARDs) and corticosteroids may increase susceptibility to osteoporosis but also suppresses inflammatory activity, which is a risk factor for osteoporosis in RA. The currently accepted approach to adequate management of osteopenia and osteoporosis includes lifestyle advice, use of calcium and vitamin D supplementation, and treatment with antiresorptive drugs (e.g., hormone replacement therapy [HRT] and bisphosphonates) (3–5). Longitudinal studies have linked the bone loss in RA to disease-related risk factors such as disease activity, immobility, and use of corticosteroids (6–13). However, to our knowledge, the magnitude and predictors of bone loss have not been studied in a population-based cohort of patients with RA receiving appropriate conventional clinical care.
The primary aim of this study was to examine changes in bone mineral density (BMD) over a 2-year period in a cohort of patients with RA that was representative of the entire RA population in the geographic area and was managed in a regular clinical setting. A second goal was to explore associations between BMD changes and demographic and clinical variables at baseline and at followup.
A cohort of 394 female and 94 male patients with RA (ages 20–70 years) were recruited from the Oslo RA registry and prospectively studied for changes in BMD and evaluation of potential risk factors for bone loss. The cross-sectional baseline data from this RA cohort have been described previously (1, 2, 14). At baseline, the female patients with RA (1) were found to be representative of the entire registry population, and the male patients (2) were considered to be fairly representative. The registry, extensively described elsewhere (15, 16), comprises patients with RA who have a residential address in Oslo who fulfilled the American College of Rheumatology (ACR; formerly, the American Rheumatism Association) 1987 revised classification criteria for RA (17). The registry has been validated to be ∼85% complete (15).
The 488 patients examined at baseline were mailed invitations to receive a followup examination after 2 years. Participants underwent clinical examinations and BMD measurements at baseline and at followup (1, 2). This examination protocol has been described previously (1, 2, 14). The demographic, patient, and disease characteristics (Table 1) were recorded using a combination of self-reported questionnaires and interviews as well as clinical examinations. Specially trained research nurses under the supervision of rheumatologists performed the data collection and clinical examination of the patients. The inter-examiner agreement between physician and research nurse was acceptable for counts of 28 joints for swelling (κ = 0.64) but lower for the counts of 28 joints for tenderness (κ = 0.48) (18). The modified Disease Activity Score (DAS) was computed according to published guidelines (19).
|Age, years (n = 366)||55.9 ± 11.1|
|Body weight, kg (n = 366)||68.5 ± 13.0|
|Height, cm (n = 366)||168.0 ± 7.9|
|Body mass index, kg/m2 (n = 366)||24.2 ± 4.0|
|% menopausal (n = 298)||67.1|
|Age at menopause, years (n = 200)||48.25 ± 5.19|
|% current smokers (n = 357)†||33.3|
|Disease duration, years (n = 366)||13.1 ± 9.6|
|Rheumatoid factor positive (n = 341)†||51.3|
|Global assessment score, 0–100-mm VAS (n = 350)†||30.0 ± 24.6|
|M-HAQ score, range 1–4 (n = 361)†||1.59 ± 0.46|
|Joint pain, 0–100-mm VAS (n = 360)†||35.7 ± 21.3|
|Fatigue, 0–100-mm VAS (n = 358)†||42.5 ± 27.2|
|Swollen joint count (28 joints) (n = 362)†||7.9 ± 6.2|
|Tender joint count (28 joints) (n = 362)†||6.7 ± 6.5|
|Deformed joint count (18 joints) (n = 362)†||2.2 ± 3.7|
|Erythrocyte sedimentation rate, mm/hour (n = 354)†||20.6 ± 16.0|
|Disease Activity Score (n = 351)†||4.41 ± 1.42|
|% with history of nonvertebral fracture after age 25 years (n = 356)†||23.0|
|% ever used DMARDs (n = 362)†||85.1|
|% current users of corticosteroids (n = 360)†||44.2|
|% current users of hormone replacement therapy, estradiol (n = 284)†||29.9|
At both visits, the erythrocyte sedimentation rate (ESR) was measured in millimeters per hour, using the Westergren technique. At baseline, the nurses performing the BMD measurements advised the patients about lifestyle adjustments and calcium and vitamin D supplementation. During the followup period, patients received treatment with antirheumatic drugs, including DMARDs and corticosteroids, as well as bone-sparing drugs (antiresorptive treatment, calcium, and vitamin D) according to the judgment of the treating doctors. At the 2-year followup examination, notes from the hospital records were reviewed, and patients were interviewed about their use of prednisolone, DMARDs, and osteoporosis treatment (calcium, vitamin D, HRT, bisphosphonates, and calcitonin) during the followup period. We also calculated the mean cumulative doses of prednisolone for the previous 12-month and 24-month periods.
BMD in the hip (left femoral neck and total hip) and the lumbar spine at L2-4 (anteroposterior view) was measured at baseline and followup, using the same dual-energy x-ray absorptiometry equipment (Lunar Expert, Madison, WI). In 22 patients, hip measurement could not be performed because of bilateral hip replacements or technical problems, and in 12 patients the right hip was measured. Spine measurements were not available in 5 patients.
The coefficient of variation (CV) for the spine phantom, calculated from daily measurements, was 0.8% for the entire period during which measurements were obtained (from 1996 to spring 2000). The in vivo reproducibility of BMD measurements was assessed using duplicate measurements obtained in 52 healthy female hospital workers (mean ± SD age 55.6 ± 4.6 years, range 50.4-66.0). The CV was 1.6% for the femoral neck, 1.3% for the total hip, and 2.5% for the spine at L2-4. Interobserver variation among the 5 trained technicians performing the measurements and the analyses was assessed by analyzing 30 randomly selected patients with RA. The technicians were compared pairwise. Precision varied from 0.5% to 1.6% for the femoral neck, from 0.4% to 1.0% for the total hip, and from 0.7% to 1.8% for the spine at L2-4.
Changes in BMD values from baseline to followup were analyzed using the mean and SD difference in BMD (gm/cm2). The patient cohort was also analyzed according to individual bone loss. A patient was defined as a BMD loser if the loss of BMD between baseline and followup exceeded the 95% detection limit ascribed to the random measurement error for the whole BMD measurement procedure (defined as smallest detectable difference [SDD]). The SDDs at the different BMD measurement sites were estimated according to the method described by Bland and Altman (20), using the duplicate BMD measurements obtained for the 52 healthy female hospital workers described above (measured for in vivo reproducibility tests). The SDD was 0.045 gm/cm2 for the femoral neck, 0.037 gm/cm2 for the total hip, and 0.084 gm/cm2 for the spine at L2-4. For calculation of the T score, we used a large European/US reference database that has been described previously (1, 2).
For comparison of baseline values between patients who attended followup and those who did not attend followup, we used Student's paired 2-tailed t-test for continuous variables and Pearson's chi-square test (categorical variables). To compare baseline and followup values for patients attending followup, we applied Student's unpaired 2-tailed t-test for continuous variables and McNemar's test for categorical variables. Analysis of variance (with Bonferroni correction for multiple comparisons) was used to examine differences in mean BMD change between more than 2 groups.
Bivariate (Student's t-test, Pearson's chi-square test, and Pearson's correlation) and multivariate analyses were used to explore the relationships between the change in BMD at the different measurement sites (dependent variables) and the demographic, clinical, and treatment variables listed in Tables 1 and 2. Based on the results of the bivariate analysis and supposed clinical relevance, variables were added to the respective multiple regression models, using the entry procedure. Multivariate analyses were performed as linear regression models when BMD change was examined as a continuous variable, and as logistic regression models when BMD change was assessed as a dichotomized variable (i.e., using individual patients classified as BMD loser or non-BMD loser), according to the SDD for the BMD measurement procedure (20). All analyses were performed using the SPSS program, version 8.0 (SPSS, Chicago, IL). P values less than or equal to 0.05 were considered statistically significant.
|All patients (n = 366)||Women (n = 298)||Men (n = 68)|
|DMARDs (n = 359)†|
|Ever user, %||86.1||86.4||84.6|
|Current user, %||51.3||52.4||46.2|
|User of >7.5 mg for >6 months, % (n = 184)†||35.9||32.7||50.0|
|Current user, % (n = 355)†||47.9||47.1||51.6|
|Previous user, % (n = 355)†||22.5||23.0||20.3|
|Cumulative dose in last 12 months, gm, mean ± SD (n = 173)||2.1 ± 1.9||2.1 ± 2.0||2.1 ± 1.4|
|Cumulative dose in last 2 years, gm, mean ± SD (n = 173)||4.3 ± 2.9||4.3 ± 3.1||4.5 ± 2.5|
|Using HRT, % (n = 283)†||–||35.0||–|
|Using bisphosphonates, % (n = 348)†||6.4||7.0||4.7|
|Using calcitonin, % (n = 341)†||0.9||1.1||0|
|Using HRT combined with bisphosphonates, % (n = 283)†||3.2|
|Using any antiresorptive treatment, % (n = 359)†‡||37.0||44.1||4.7|
|Number of months used during last 2 years, mean ± SD (n = 138)||20.7 ± 6.3||21.0 ± 6.1||12.5 ± 9.0|
|Using calcium, % (n = 354)†||58.5||62.5||40.9|
|Using vitamin D, % (n = 353)†||64.9||69.6||44.8|
|Using calcium and/or vitamin D, % (n = 362)†||36.5||34.9||43.3|
|Number of months used during last 2 years, mean ± SD (n = 161)||21.9 ± 4.9||22.4 ± 4.4||18.8 ± 6.4|
The local ethics committee approved this study. The Data Inspectorate of Norway had previously approved the registry of patients with RA living in Oslo.
Of the 488 patients examined at baseline, 366 (298 women, 68 men) returned for followup (75% attendance rate; mean ± SD followup period 2.2 ± 0.2 years). The characteristics of the followup patients at baseline are shown in Table 1. We found no statistically significant differences (adjusted for multiple comparison) between patients who did attend followup and those who did not attend followup with regard to baseline variables (data not shown). Therefore, in this prospective study, as in our previous cross-sectional studies (1, 2), the patients who were examined are considered to be fairly representative of the entire background RA population within the age range studied (20–70 years at baseline).
For the 366 patients who received a followup examination, we found no statistically significant changes between baseline and followup values (adjusted for multiple comparisons) with regard to joint counts and the other disease variables listed in Table 1. At baseline and followup, the mean ± SD values for ESR were 20.6 ± 16.0 mm/hour and 21.9 ± 19.4 mm/hour, respectively, the counts of 28 joints for swelling were 7.9 ± 6.2 and 7.6 ± 5.3, respectively, the counts of 28 joints for tenderness were 6.7 ± 6.5 and 8.4 ± 6.7, respectively, the DAS were 4.41 ± 1.42 and 4.60 ± 1.31, respectively, and the modified Health Assessment Questionnaire (M-HAQ; 8 items, range of scores 1–4) scores were 1.59 ± 0.46 and 1.61 ± 0.52, respectively.
At followup, 37.0% of the patients (44.1% of females and 4.7% of males) were currently using any potent antiresorptive osteoporosis treatment, 36.5% were using calcium and/or vitamin D alone (34.9% of females, 43.3% of males) (Table 2), and 47.9% were taking prednisolone (47.1% of females and 51.6% of males). The mean ± SD cumulative prednisolone doses in patients who ever used corticosteroids were 2.1 ± 1.9 gm for the last 12 months and 4.3 ± 2.9 gm for the last 24 months. This is equivalent to prednisolone use of 5.8 mg per day in the 12-month period and 5.9 mg per day in the 24-month period. Current use of DMARDs at followup was reported by 51.3% of all patients and 86.1% of those who ever used corticosteroids. Further details about drug treatments are shown in Table 2.
Among patients who used corticosteroids and had a baseline T score less than or equal to −1 SD at the spine and/or hip (the ACR-recommended threshold for treatment of corticosteroid-induced osteoporosis) (3, 21), 83.0% (90.7% of females, 51.9% of males) were using calcium and/or vitamin D, and 46.7% were using antiresorptive drugs (55.0% of females, 11.5% of males) at followup. Among all patients (regardless of their use of corticosteroids) who had a baseline T score less than or equal to −2.5 SD at the spine and/or hip (the World Health Organization [WHO] definition for osteoporosis in females , here also applied to males), 81.3% (83.8% of females, 66.7% of males) were currently using calcium and/or vitamin D at followup, and 61.0% (67.1% of females, 25.0% of males) were using antiresorptive drugs.
As illustrated in Figure 1, the mean reduction in BMD over 2 years ranged from −0.29% to −0.77% at the different measurement sites, from −0.11% to −0.57% in females, and from −0.94% to −1.95% in males. Whereas mean BMD increased at all measurement sites in patients using antiresorptive drugs (0.16–1.64%), BMD decreased in patients treated with calcium and/or vitamin D alone (−1.99% to −1.39%) and in those not treated for osteoporosis (−1.20% to −0.43%) (Figure 2). Among patients using HRT, the change in BMD was −0.08% for the femoral neck, 0.42% for the total hip, and 0.91% for the spine at L2-4; among those using bisphosphonates, these changes were −0.51%, 0.27%, and 2.62%, respectively. Calcitonin was not explored separately, due to the low number of patients who were using it. A particularly high increase in BMD (6.98% for the femoral neck, 4.49% for the total hip, and 8.96% for the spine at L2-4) was demonstrated in a subgroup of 9 female patients who were using HRT and bisphosphonates concurrently.
Among patients who were using antiresorptive drugs concurrently, BMD loss was arrested at the hip and was increased at the spine in current users of corticosteroids. In contrast, BMD increased at all measurement sites in patients who were not receiving corticosteroids (Figure 3). Among patients who were not using antiresorptive drugs, BMD was reduced in both current and noncurrent users of corticosteroids (Figure 3). In general, patients treated with antiresorptive drugs were significantly more affected by RA and were more often users of corticosteroids compared with those who were not being treated with antiresorptive drugs (data not shown). Similarly, current users of corticosteroids at followup were significantly more severely affected by RA compared with those not currently using corticosteroids (data not shown). The proportions of patients classified as BMD losers (according to BMD loss exceeding the measurement error) were 18.0% for the femoral neck, 19.4% for the total hip, and 13.0% for the spine at L2-4; the proportions of patients gaining BMD at these sites were 9.6%, 10.7%, and 11.1%, respectively.
We found several differences between patients classified as BMD losers (patients with bone loss exceeding the SDD) and non-BMD losers (those with bone loss not exceeding the SDD) with regard to baseline demographic and clinical variables, but the differences were not consistent across all 3 measurement sites. The use of high-dose, long-term prednisolone (>7.5 mg for >6 months) was significantly associated with loss of BMD in the femoral neck (55.2% of BMD losers versus 31.7% of non-BMD losers; P = 0.04). For the total hip, different variables related to corticosteroid use were associated with BMD loss: use of long-term high-dose corticosteroids (57.9% of BMD losers versus 29.1% of non-BMD losers; P = 0.001); current use of corticosteroids (60.6% versus 42.8%; P = 0.009); mean ± SD cumulative dose of prednisolone over the last 12 months (2.7 ± 1.8 gm versus 1.9 ± 2.0 gm; P = 0.02); and mean ± SD cumulative dose of prednisolone over the last 24 months (5.2 ± 3.1 gm versus 4.0 ± 2.0 gm; P = 0.03). In contrast, current use of HRT, both at baseline (17.8% of BMD losers versus 33.5% of non-BMD losers; P = 0.04) and at followup (21.3% versus 40.0%; P = 0.02), and use of antiresorptive osteoporosis treatment (19.4% versus 40.7%; P = 0.001) were associated with protection against loss of BMD for the total hip.
For the spine at L2-4, differences between BMD losers and non-BMD losers were as follows: mean ± SD age 50.0 ± 11.4 years versus 56.8 ± 10.7 years (P < 0.001), mean ± SD M-HAQ score 1.44 ± 0.40 versus 1.61 ± 0.47 (P = 0.02), mean ± SD swollen joint count 5.6 ± 4.9 versus 8.3 ± 6.3 (P < 0.001), mean ± SD tender joint count 3.8 ± 4.7 versus 7.1 ± 6.6 (P < 0.001), mean ± DAS 3.77 ± 1.25 versus 4.52 ± 1.40 (P = 0.001), users of HRT 22.5% versus 39.0% (P = 0.04), and users of antiresorptive treatment 21.7% versus 39.4% (P = 0.02).
Bivariate analyses, applying BMD changes as continuous dependent variables, revealed no major differences from these results (data not shown).
In the multivariate statistical analyses (in accordance with the selection criteria described), the following variables were included in the models: age, sex, body mass index, M-HAQ, ESR, current use of antiresorptive osteoporosis treatment (HRT, bisphosphonates, calcitonin, or a combination of HRT and bisphosphonates), current use of prednisolone at followup, and baseline BMD. In this model, current use of corticosteroids was independently associated with an increased risk for BMD loss for the total hip (odds ratio [OR] 2.63, 95% confidence interval [95% CI] 1.38–5.00) and for the spine at L2-4 (OR 2.70, 95% CI 1.30–5.63), whereas antiresorptive treatment for osteoporosis was associated with a decreased risk of bone loss for the total hip (OR 0.43, 95% CI 0.20–0.89) (Table 3). No major differences from these results were shown in the linear regression model using BMD change as a dependent variable.
|Femoral neck (n = 322)||Total hip (n = 323)||Spine L2-4 (n = 338)|
|Beta||SE||OR (95% CI)||Beta||SE||OR (95% CI)||Beta||SE||OR (95% CI)|
|Age||−0.0027||0.0162||1.00 (0.97–1.03)||−0.0230||0.0153||0.98 (0.95–1.01)||−0.0262||0.0173||0.97 (0.94–1.01)|
|Sex||−0.0927||0.4169||0.91 (0.40–2.06)||0.5421||0.3763||1.72 (0.82–3.60)||−0.7387||0.5322||0.48 (0.17–1.35)|
|Body mass index||0.0044||0.0386||1.00 (0.93–1.08)||0.0144||0.0394||1.01 (0.94–1.10)||−0.0416||0.0482||0.96 (0.87–1.05)|
|Modified Health Assessment Questionnaire score||0.4404||0.3562||1.55 (0.77–3.12)||0.2420||0.3491||1.27 (0.64–2.53)||−0.5550||0.4435||0.57 (0.24–1.37)|
|Erythrocyte sedimentation rate||0.0108||0.0094||1.01 (0.99–1.03)||0.0092||0.0094||1.01 (0.99–1.03)||−0.0136||0.0130||0.99 (0.96–1.01)|
|Current antiresorptive treatment at followup†||−0.3426||0.3528||0.71 (0.36–1.42)||−0.8557||0.3773||0.43 (0.20–0.89)||−0.6694||0.4247||0.51 (0.22–1.18)|
|Current use of prednisolone at followup||0.2256||0.3300||1.25 (0.66–2.39)||0.9665||0.3272||2.63 (1.38–5.00)||0.9946||0.3747||2.70 (1.30–5.63)|
|Bone mineral density at baseline for the different measurement sites||2.5496||1.1205||12.80 (1.42–115.11)||1.4754||1.1058||4.37 (0.50–38.19)||2.3053||0.9247||10.03 (1.64–61.41)|
This 2-year observational study exploring BMD changes in a large cohort of patients with RA shows that bone loss can be prevented in this population, which is known to be at high risk for fractures (23–26). In the present study, BMD increased (0.16–1.64%) at all measurement sites in patients using antiresorptive drugs but decreased in those using calcium and vitamin D supplementation only (−1.99 to −1.39%) as well as in patients receiving no treatment for osteoporosis (−1.20 to −0.43%).
The magnitude of bone loss demonstrated in the current study was generally less than that shown in previously published longitudinal studies of patients with RA (6–13). In the preceding studies, annual BMD changes ranged from +0.2% to −2.0% in the lumbar spine and from −0.3% to −3.2% in the femoral neck (6–13). Also, patients who were receiving treatment for osteoporosis were either excluded from those studies (7, 8, 10, 11, 13) or comprised only a minority of the study population (6, 12), or information about their osteoporosis treatment was not reported (9). In accordance with results from several other longitudinal studies in patients with RA, we found that more BMD loss occurred at the hip than at the spine (6, 8, 10, 12). This variation may be explained by synovitis of the hip joint; subsequent impaired functional capacity may lead to impaired loading of the lower limbs in patients with RA. Alternatively, the observed difference may only be artificial, caused by increased osteoarthritis in the spine with aging.
In a Finnish population, the annual rate of BMD loss was estimated in cross-sectional studies to be −0.2% for the lumbar spine and −0.3% for the femoral neck in men (27), and −0.7% and −1.3%, respectively, in women ages 39 years and older (28). However, Melton et al (29), examining the rate of bone loss in men and women older than age 50 years, demonstrated that compared with longitudinal data, cross-sectional data overestimate the rate of bone loss (29). Using cross-sectional data, BMD for the total hip in women appeared to decline by −0.8% per year but by only −0.2% per year using longitudinal data; in men, the comparable figures were −0.4% and +0.03% (29). Compared with these longitudinal rates of annual BMD loss, the rates for our female patients with RA were slightly increased, but those for our male patients with RA were increased severalfold.
Previous controlled clinical trials showed that HRT and bisphosphonates are effective for the prevention and treatment of both primary osteoporosis and glucocorticoid-induced osteoporosis (30–33). In randomized, controlled trials of bisphosphonates for the prevention of glucocorticoid-induced osteoporosis, the annual increase in BMD among patients in the treatment group was 0.6–2.9% for the lumbar spine and 0.2–1.0% for the femoral neck, whereas among patients in the placebo group (calcium, alone or in combination with vitamin D), loss of BMD was −0.4% to −3.2% and −1.2% to −3.1%, respectively (31–33). The same pattern was observed in our study, supporting the ineffectiveness of calcium and/or vitamin D alone to stop bone loss. Our treatment strategy was based mainly on recommendations given in consensus statements published in the mid 1990s (3, 4), recommending HRT as the first treatment choice for female patients with RA. However, later guidelines for the prevention and therapy of glucocorticoid-induced osteoporosis recommended bisphosphonate therapy as the treatment of choice (5, 21).
Several authors have reported that patients using long-term corticosteroid therapy are receiving inadequate treatment for the prevention of bone loss (34–36), and that is especially true for male patients (34, 35). In the study by Buckley et al (35), 54% of the postmenopausal women and only 5% of the men who were receiving long-term corticosteroid treatment were also receiving antiresorptive treatment. Our patients with RA who were currently using corticosteroids at followup were, to some extent, treated with antiresorptive drugs more frequently (55% of women and 11% of men) than were patients described in these previous studies (34–36). However, our results (especially for male patients) are far from satisfactory, considering the recommendations of the ACR for prevention and treatment of glucocorticoid-induced osteoporosis, which were first presented in 1996 (3) and updated in 2001 (21). In our study, the greatest rate of BMD increase was observed in patients using HRT and bisphosphonates simultaneously. The additive effect of bisphosphonates and HRT on hip and spine BMD has previously been shown in postmenopausal osteoporosis (37–39) but not in randomized controlled trials in glucocorticoid-induced osteoporosis. Our data suggest that the combination of HRT and bisphosphonates may be considered in postmenopausal patients with RA who have severe or established osteoporosis.
We also explored the change in BMD between baseline and followup, using the WHO definition for osteoporosis (22). BMD increased in patients with a T score less than or equal to −2.5 SD at baseline, was unchanged in patients with a T score greater than −2.5 SD and less than −1 SD, and declined in patients with a T score greater than −1 SD (data not shown). The fact that BMD increased in patients with the lowest BMD (T score less than or equal to −2.5 SD) reflects a trend toward treating patients who have the lowest BMD levels.
Factors associated with BMD change in RA are difficult to examine, due to their complexity and interrelationships (e.g., measures of disease activity are suppressed and functional capacity is increased by using corticosteroids and DMARDs in RA) (40). This issue is further complicated by the use of potent drugs for osteoporosis that effectively increase bone mass (30, 32, 33). Despite these methodologic problems, our study demonstrated that corticosteroids (a well-known risk factor for bone loss) (41, 42) are associated with bone loss. However, the results for all measurement sites were not consistent. The negative effect of corticosteroids on bone may be balanced by reducing the impact of other factors contributing to RA, such as inflammation and immobility (6, 40). In the study by Ferraccioli et al (40), effective control of systemic inflammation in patients with RA who were aggressively treated with prednisone (5 mg daily), methotrexate (10 mg/week), and cyclosporine (3–5 mg/kg) led to a 3.9% increase in spine BMD. Van Schaardenburg et al (12), in their randomized study, found greater BMD loss in elderly patients with RA receiving prednisone compared with patients treated with chloroquine and healthy controls. Verhoeven and Boers (43) concluded, after reviewing the literature, that the use of short-term (≤1 year) low-dose corticosteroid treatment (<10 mg prednisone/day) has limited influence on BMD in RA.
In contrast to previously reported results, we did not find consistent relationships between bone loss and disease measures. Gough et al (6), examining patients with early RA, found that both level of disability and C-reactive protein level predicted BMD loss, but use of prednisolone did not. In a study by Mazzantini et al (11), patients with active RA lost significantly more BMD at the lumbar spine (−5.5%) than did those with less active disease (−1.1%). Our data may, however, suggest a role of inflammation in the pathogenesis of osteoporosis in RA. In bivariate analysis using cutoffs for SDD other than the 95% detection limit, ESR was significantly associated with bone loss for the total hip (data not shown), and the multivariate linear regression model revealed a possible association between elevated ESR and BMD reduction at the hip (P = 0.07).
The strength of our observational population-based study is that it provides insight into what takes place in the real world of patients with RA regarding the magnitude of BMD loss and the attitude of physicians toward the prevention and treatment of osteoporosis. Our study also had obvious limitations. Compared with a longitudinal observational study, a randomized controlled study design gives stronger evidence for treatment effects, but most researchers will agree that the 2 designs provide complementary information. The detection of potential risk factors associated with BMD loss in patients with RA should, therefore, be interpreted with caution, because no causality is established from multivariate statistical analyses, even when adjusting for confounders.
For the SDD estimation of the BMD measurement, using the approach described by Bland and Altman (20), we ideally should have used duplicate BMD measurements obtained from a representative sample of the RA patients who were examined. Ravaud et al (44) found that the cutoffs for SDD were inversely related to BMD level. The mean BMD level in our RA population was approximately 9–15% lower (10–15% in women, 2–12% in men) than that in the healthy female hospital workers used for the SDD estimation in our study (data not shown). However, we did not find any major differences between associations when different cutoffs (e.g., 75%, 90%, and 97.5% limits of agreement) were applied. The short (2-year) observational period in our study is one of the reasons why as many as 70% of the patients with RA did not undergo a change in BMD exceeding the SDD with 95% limits of agreement. By increasing the length of the observational period and by using the mean values of duplicate BMD measurements obtained from patients at baseline and followup, the number of patients with changes in BMD exceeding the SDD would have increased. However, such an approach would have limited the generalizability of our results.
In conclusion, the results of this study add evidence to the importance of treating patients with RA in accordance with current guidelines for the prevention of further bone loss. Only the group of patients with RA who were using antiresorptive drugs did not lose BMD during followup, emphasizing the potency of these drugs in protecting against bone loss. The male patients with RA, who had a higher rate of BMD loss than the female patients, had been treated less aggressively with antiresorptive drugs than had the female patients. Therefore, more attention should be focused on treatment of osteoporosis in male patients with RA. Whether these results also can be translated into effects on fractures (23–26) remains to be studied.
We thank Petter Mowinckel for statistical advice. We are grateful to and appreciative of our technicians Ingerid Müller, Sidsel Arnkværn, Margareth Sveinsson, Anne Katrine Kongtorp, and Espen Haavardsholm for expert technical assistance.