Relationship between early bone mineral density changes and long-term function and radiographic progression in rheumatoid arthritis


  • Victoria Bejarano,

    1. University of Leeds and NIHR Leeds Musculoskeletal Biomedical Research Unit, Leeds, UK
    Search for more papers by this author
    • Dr. Bejarano has received consultant fees, speaking fees, and/or honoraria (less than $10,000 each) from Abbott and MSD.

  • Elizabeth Hensor,

    1. University of Leeds and NIHR Leeds Musculoskeletal Biomedical Research Unit, Leeds, UK
    Search for more papers by this author
  • Michael Green,

    1. Harrogate District Hospital, Harrogate, and York Teaching Hospital, York, UK
    Search for more papers by this author
  • Glenn Haugeberg,

    1. Sørlandet Hospital, Kristiansand, Norway
    Search for more papers by this author
  • Andrew K. Brown,

    1. York Teaching Hospital, York, UK
    Search for more papers by this author
    • Dr. Brown has received consultant fees, speaking fees, and/or honoraria (less than $10,000 each) from Abbott, Menarini, Merck, and Pfizer.

  • Maya H. Buch,

    1. University of Leeds and NIHR Leeds Musculoskeletal Biomedical Research Unit, Leeds, UK
    Search for more papers by this author
    • Dr. Buch has received consultant fees, speaking fees, and/or honoraria (less than $10,000 each) from Abbott, Pfizer, Roche, and UCB.

  • Paul Emery,

    1. University of Leeds and NIHR Leeds Musculoskeletal Biomedical Research Unit, Leeds, UK
    Search for more papers by this author
    • Dr. Emery has received consultant fees, speaking fees, and/or honoraria (less than $10,000 each) from Abbott, Bristol-Myers Squibb, Merck, Pfizer, and Roche.

  • Philip G. Conaghan

    Corresponding author
    1. University of Leeds and NIHR Leeds Musculoskeletal Biomedical Research Unit, Leeds, UK
    • Section of Musculoskeletal Disease, Leeds Institute of Molecular Medicine and NIHR Leeds Musculoskeletal Biomedical Research Unit, Chapel Allerton Hospital, Chapeltown Road, Leeds LS7 4SA, UK
    Search for more papers by this author
    • Dr. Conaghan has received consultant fees, speaking fees, and/or honoraria (less than $10,000 each) from AstraZeneca, Bristol-Myers Squibb, Centocor, Merck, Novartis, Pfizer, and Roche.



To ascertain whether bone mineral density loss during the first year of treatment for early rheumatoid arthritis (RA) as assessed by dual x-ray absorptiometry (DXA) is associated with long-term function, quality of life, and radiographic progression.


Early RA patients, prior to commencing disease-modifying antirheumatic drug therapy, underwent evaluation with DXA scans of both hands, femoral neck, and lumbar spine at the time of diagnosis, then at 1 year and final followup at >6 years. Clinical and laboratory data and hands and feet radiographs were obtained at baseline and final followup.


Sixty-four patients were assessed. At baseline, median disease duration was 6 months, mean 28-joint Disease Activity Score was 5.85, and median Health Assessment Questionnaire score was 1.38. Median followup was 6.4 years (range 5.1–7.2 years). For erosion scores, 86% of patients with hand bone density loss above the smallest detectable change after 1 year progressed versus 55% of those with no hand bone density loss (P = 0.006). However, baseline radiographs showed a similar strength of association. Eighty-four percent of patients with erosions at baseline had erosion progression versus 57% of those with no baseline erosions (P = 0.021). Additionally, first-year bone density loss was not associated with function and quality of life status at followup.


Hand bone loss during the first year of treatment in early RA, as assessed with DXA, was associated with a high likelihood of radiographic progression; however, in the current study this was not superior to baseline radiographs. Furthermore, it was not associated with patient-reported outcomes.


Rheumatoid arthritis (RA) affects approximately 0.5–1% of the general population. While there have been significant therapeutic advances such as the introduction of the tumor necrosis factor (TNF) antagonists, there is still the need for simple and feasible prognostic tools that allow a rational use of expensive biologic therapies, most obviously by helping select those patients with bad prognostic factors at the time of diagnosis.

Current demographic, clinical, and laboratory variables that are associated with a bad prognosis are female sex, poor function as assessed by the Health Assessment Questionnaire (HAQ), rheumatoid factor (RF) positivity, anti–cyclic citrullinated peptide antibody positivity, presence of the shared epitope, and radiographic erosions at baseline (1–3). Hoff et al recently reported the prognostic ability of cortical hand bone mineral density (BMD) loss during the first year of disease in RA as assessed by digital x-ray radiogrammetry. Radiographic progression of erosions at both 5 and 10 years was significantly higher in those patients with early hand BMD loss compared with those without early BMD loss (4). Dual x-ray absorptiometry (DXA) is widely available and early reports from our group suggested that hand BMD loss as assessed by DXA was more sensitive to RA-related bone damage than conventional radiographic erosions (5). Therefore, the aim of this study was to establish whether DXA-measured hand BMD loss during the first year of treatment in patients with early RA would be associated with long-term radiographic progression, function, and quality of life, and therefore offer a simple and widely available tool for predicting poor prognosis.

Significance & Innovations

  • Dual x-ray absorptiometry (DXA) has been proposed as a more accurate tool to assess bone damage in rheumatoid arthritis (RA) compared with radiographs.

  • The evidence of DXA value as a prognostic tool in early disease is limited.

  • We have found a previously unrecognized association between first-year hand bone mineral density loss and long-term radiographic progression in this group of patients with early RA.

  • The first-year DXA-measured hand bone loss does not provide any more information than baseline radiographs of the hands and feet.


This research was approved by the local research ethics committee. Patients with early RA (1987 revised American College of Rheumatology classification criteria [6]) with a disease duration of <1 year and who had never taken disease-modifying antirheumatic drugs (DMARDs) were recruited to 3 different studies between 1995 and 1999. They had a DXA assessment at the time of diagnosis, at 1 year, and at final followup, and data were acquired for the hands, lumbar spine, and right femoral neck. All scans were performed by the same equipment (Lunar Expert). More than 100 patients were included in the original studies; however, an upgrade of the DXA equipment took place during the recruitment to the followup study and a decision was made to limit the enrollment to those patients who were able to have a DXA scan with the same equipment that had been used for the original studies, allowing meaningful comparisons between the measurements taken by the same machine. Standard clinical and laboratory data were collected at the DXA time points. This included 66/68 tender joint counts (TJCs) and swollen joint counts (SJCs), visual analog scale (VAS) scores for pain and patient's assessment of disease activity, HAQ score, RA Quality of Life (RAQoL) questionnaire score, C-reactive protein (CRP) level, and erythrocyte sedimentation rate. Conventional radiographs of the hands and feet were also available at baseline and final followup. They were scored by a single experienced radiologist using the Genant method (7). The reader was blind to both treatment group and chronological order.

The patients received their first treatment at the time of diagnosis and according to the 3 different study protocols. Fifty-three percent of the participants in the current study were originally enrolled in the study by Proudman et al and received either a combination of methotrexate (MTX) and cyclosporin A (CSA) or sulfasalazine (SSZ) monotherapy for 24 months (median dosages at 1 year: MTX 15 mg/week, CSA 200 mg/day, SSZ 3 gm/day [8]); the number of patients enrolled in the current study from each treatment arm was similar. Another 45% were treated with MTX monotherapy for 12 months (mean dosage 17 mg/week) (9) and 2 patients received infliximab 10 mg/kg and MTX up to 15 mg/week for up to 9 months (10). Once all of the patients completed their participation in these studies, they resumed standard clinical care with addition of other DMARDs at the discretion of their physician.

Given that the number of patients was predetermined, a sample size calculation did not take place. Nonparametric statistical analyses, including Spearman's rank correlation, Pearson's chi-square test, and binary logistic regression, were used.

The smallest detectable change (SDC) in DXA values applied in this study was calculated previously: the SDC for mean hands BMD was 0.0071 gm/cm2, for femoral neck was 0.0539, and for lumbar spine was 0.0598 (5).


DXA data were available for 64 patients. The mean age at baseline was 54 years and 58% of the patients were women. Sixty-six percent were RF positive. This was an early disease cohort with severe disease activity (baseline median disease duration 6 months and mean 28-joint Disease Activity Score [DAS28] 5.85) and significant functional and quality of life impairment (median HAQ score 1.38, median RAQoL score 17) (Table 1). The median followup time was 6.4 years (range 5.1–7.2 years). At followup, the cohort still had moderate disease activity (DAS28 3.33, HAQ score 1.00, RAQoL score 11) (Table 1).

Table 1. Demographics and disease characteristics at baseline and followup*
  • *

    IQR = interquartile range; RF = rheumatoid factor; CRP = C-reactive protein; DAS28 = 28-joint Disease Activity Score; HAQ = Health Assessment Questionnaire (range 0–3); RAQoL = Rheumatoid Arthritis Quality of Life questionnaire (range 0–24).

  • Fifty-six percent of patients in low disease activity range, including 37% in remission.

Age, mean ± SD years54.1 ± 12.7
Women, no. (%)37 (58)
Disease duration, median (IQR) months6.0 (4.0–8.0)
RF positivity, no. (%)42 (66)
CRP level, median (IQR) mg/liter22.0 (7.3–49.3)7.5 (5.0–17.8)
DAS28, mean ± SD5.85 ±1.113.33 ± 1.44
HAQ score, median (IQR)1.38 (0.88–1.63)1.00 (0.38–1.59)
RAQoL score, median (IQR)17 (7–18)11 (4–20)


As shown in Table 2, more patients had loss of BMD that exceeded the SDC over the first year in their hands than in the lumbar spine or femoral neck (66%, 28%, and 17%, respectively). In comparison, the proportion with a negative 1-year change in BMD at the hands, lumbar spine, and femoral neck was 67%, 63%, and 50%, respectively. Similarly, between baseline and final followup, more patients had BMD loss exceeding the SDC in their hands compared to the lumbar spine or femoral neck (82%, 29%, and 42%, respectively), while a negative change in BMD was now seen in 82% in the hands, 59% in the lumbar spine, and 74% in the femoral neck. During the first year there was no association between change in disease activity and hand BMD loss (▵CRP level ρ = 0.136, ▵SJC ρ = −0.081). Furthermore, at 12 months neither the TJC68 (ρ = −0.108, P = 0.394), SJC66 (ρ = −0.162, P = 0.200), nor VAS for disease activity (ρ = −0.007, P = 0.955) was related to first-year hand BMD loss. However, CRP level was weakly associated (ρ = −0.316, P = 0.016). Final followup CRP level was not substantively related to change in BMD from baseline to final followup (ρ = −0.050, P = 0.692), but showed a weak relationship with radiographic change over the same period (ρ = −0.328, P = 0.008).

Table 2. Dual x-ray absorptiometry data*
 Baseline12-month changeFollowup change
  • *

    BMD = bone mineral density; SDC = smallest detectable change.

Hand BMD, mean ± SD gm/cm20.431 ± 0.071−0.164 ± 0.024−0.034 ± 0.035 (n = 50)
 No. (%) > SDC 42 (66)41 (82)
Spine BMD, mean ± SD gm/cm21.199 ± 0.163−0.023 ± 0.058−0.029 ± 0.099 (n = 51)
 No. (%) > SDC 18 (28)15 (29)
Right femoral neck BMD, mean ± SD gm/cm20.966 ± 0.129−0.005 ± 0.054−0.050 ± 0.072 (n = 50)
 No. (%) > SDC 11 (17)21 (42)

Radiographic progression.

Baseline erosion score was 1.55 (interquartile range [IQR] 0–4.53) compared with a followup score of 2.85 (IQR 0.13–8.54). Similarly, the total score progressed from 2.73 (IQR 0–6.08) to 4.40 (IQR 0.63–16.74). Seventy-five percent of patients had deterioration in their erosion score and 78% had deterioration in their total radiographic score (Table 3).

Table 3. Radiographic data*
 BaselineFollowup change
  • *

    Values are the median (interquartile range) Genant scores of hands and feet radiographs unless otherwise indicated. JSN = joint space narrowing.

Erosion score1.55 (0–4.53)2.85 (0.13–8.54)
 Deterioration, no. (%) 48 (75)
JSN score0 (0–1.91)0.48 (0–5.49)
 Deterioration, no. (%) 32 (50)
Total score2.73 (0–6.08)4.40 (0.63–16.74)
 Deterioration, no. (%) 50 (78)

Relationship between BMD and radiographic progression.

When the patients were grouped according to their change in BMD in the first year (loss or no loss exceeding SDC), there was an association between hand BMD deterioration and progression of erosions. Only 12 (55%) of 22 of those with no hand BMD loss had erosion progression versus 36 (86%) of 42 of those with hand BMD loss (χ2 = 7.48, P = 0.006). However, baseline radiographs showed a similar strength of association. Of 21 patients who had no radiographic erosions present at baseline, 12 (57%) became erosive at followup, compared to 36 (84%) of 43 patients with erosions at baseline who had an increased erosion score (χ2 = 5.32, P = 0.021). When BMD loss and erosions at baseline were entered into a binary logistic regression model of the odds of erosion progression, although the model as a whole was significant (χ2 = 8.52, 2 df, P = 0.014), neither variable made a statistically significant contribution independently of the other, and the odds ratios (ORs) for each were comparable (BMD loss OR 3.53, 95% confidence interval [95% CI] 0.93–13.48; P = 0.065 and baseline erosion OR 2.21, 95% CI 0.58–8.45; P = 0.249). Erosion progression was not associated with first-year BMD loss in the lumbar spine (33 [71.7%] of 46 versus 15 [83.3%] of 18; χ2 = 0.93, P = 0.336) or femoral neck (39 [73.6%] of 53 versus 9 [81.8%] of 11; χ2 = 0.57, P = 0.566).

Patient-reported outcomes.

HAQ and RAQoL at followup were not substantively associated with first-year BMD loss at any site (all Spearman's ρ = <|0.2|).

Medication effects.

At followup, 25 patients were receiving MTX monotherapy, 5 were receiving SSZ, 1 was receiving leflunomide, 1 was receiving hydroxychloroquine, 18 were receiving combination therapy including any of the above, 1 was receiving the MTX-CSA original combination, and 8 were receiving anti-TNF agents. Five were receiving no DMARDs. Of those taking combination therapy, 6 were receiving prednisolone between 5 and 7 mg/day. Four (6%) had bisphosphonates and 6 (9%) had hormone replacement therapy (HRT). There were no differences in loss of BMD from 1 year to followup for those taking prednisolone, anti-TNF agents, bisphosphonates, or HRT; however, the numbers were very small and no firm conclusions can be made. Table 4 shows the medication use at followup according to hand BMD loss during the first year. The use of HRT, monotherapy, biologic therapy, and no need for DMARDs was similar. The number of patients requiring combination treatment was also comparable; however, more patients in the hand bone loss group included prednisolone in their therapy. This was parallel with more prescriptions of bisphosphonates in the same group.

Table 4. Medication at followup by first-year hand bone loss*
 BMD loss <SDCBMD loss >SDC
  • *

    BMD = bone mineral density; SDC = smallest detectable change; DMARDs = disease-modifying antirheumatic drugs; HRT = hormone replacement therapy.

Monotherapy, no. (%)11 (50)21 (50)
Combination therapy, no. (%)5 (23)8 (19)
Combination therapy including prednisolone, no. (%)1 (5)5 (12)
No DMARDs, no. (%)2 (9)3 (7)
Biologic therapy, no. (%)3 (14)5 (12)
HRT, no. (%)2 (9)4 (10)
Bisphosphonate, no. (%)0 (0)4 (10)


In this study we have identified an association between hand BMD loss in the first year of RA and the likelihood of radiographic progression at followup. This would suggest that those patients whose BMD deteriorates over the first year of disease have a high chance of subsequent radiographic progression. There is no doubt about the need for good reliable prognostic tools for RA at the time of presentation to guide treatment decisions and help utilize the expensive biologic therapies more cost effectively. DXA is a simple, reproducible, and widely available technique. However, DXA performed just marginally better than baseline radiographs alone. Eighty-four percent of those with erosions at baseline had radiographic progression at followup. Our cohort had a high percentage of patients with erosions at baseline compared with the average early RA population. This is explained by the fact that more than one-half of the patients had been recruited to studies that selected patients for bad prognosis (8, 10). However, this may overestimate the ability of the baseline radiographs to predict the followup radiographic progression. Equally, the results of this study may not be applicable to patients with milder disease at presentation. While it has been documented that radiogrammetry performs better than DXA when detecting hand bone loss in RA (11), this study did not include a comparison of these techniques.

Deodhar et al reported a group of 40 patients with early RA who were followed up for 5 years with serial radiographs and DXA (12). They found no association between bone loss in the first 6 months and final Larsen radiographic score; however, patients with high bone loss at 6 months had a worse functional status as assessed by the Duruoz Hand Index compared to those with low bone loss. Our data do suggest an association between early hand bone loss and radiographic progression, and this discordance may be due to the different sample size, disease characteristics, or therapeutic approaches. Their cohort had less functional disability at baseline (HAQ score 1.1 versus 1.38 in our group); however, they had a higher inflammatory response (CRP level 43.5 versus 22.0 mg/liter, respectively). Other measures of disease are not comparable. Proportionally, Deodhar et al reported less use of MTX or combination therapy and there were no patients receiving anti-TNF agents. This may represent a milder disease cohort.

There was no association between initial BMD loss, function, and quality of life outcomes (HAQ, RAQoL). Similarly, Deodhar et al found no association with the HAQ. Although clearly the HAQ is related to structural change in the long term, there are suggestions that other aspects such as inflammation may influence the scores, especially early in the disease course (13). It is important to note that a proportion of patients did progress radiographically without an initial BMD loss. This suggests that there may be a “threshold” effect of inflammation on measurable bone loss.

The main limitations to this study are the size of the cohort and the lack of first-year radiographs. The followup radiographs include the progression in the first year, and this may overestimate the value of DXA in the first year by taking into account a damage that could have been measured simply by a radiograph at 1 year. Moreover, the true predictive value of DXA would address the radiographic progression from 1 year to the final followup, excluding the radiographic progression in the first year. Additionally, an ideal prognostic tool would be able to provide the required information within a much shorter frame of time. BMD loss over the first few months of the disease may provide more helpful results. Furthermore, it may be relevant to investigate its prognostic ability to predict progression from undifferentiated inflammatory arthritis to RA. The relationship between bone loss and disease course was not examined in this study as serial disease activity measures over the long-term followup period were not available.

In summary, there is an association between hand bone loss in the first year of RA, as assessed by DXA, and long-term radiographic progression; however, this tool performs just marginally better than baseline radiographs and, based on these results, it does not add to the management of early RA.


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. Conaghan 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. Bejarano, Green, Haugeberg, Brown, Emery, Conaghan.

Acquisition of data. Bejarano, Green, Brown, Buch, Emery, Conaghan.

Analysis and interpretation of data. Bejarano, Hensor, Green, Haugeberg, Brown, Emery, Conaghan.