The relationship between focal erosions and generalized osteoporosis in postmenopausal women with rheumatoid arthritis

Abstract

Objective

Among rheumatoid arthritis (RA) patients who have had the disease for 10 years, more than half have focal erosions, and the risk of fracture is doubled. However, there is little information about the potential relationship between focal erosions and bone mineral density (BMD). The aim of this study was to determine whether lower BMD is associated with higher erosion scores among patients with RA.

Methods

We enrolled 163 postmenopausal women with RA, none of whom were taking osteoporosis medications. Patients underwent dual x-ray absorptiometry at the hip and spine and hand radiography, and completed a questionnaire. The hand radiographs were scored using the Sharp method, and the relationship between BMD and erosions was measured using Spearman's correlation coefficients and adjusted linear regression models.

Results

Patients had an average disease duration of 13.7 years, and almost all were taking a disease-modifying antirheumatic drug. Sixty-three percent were rheumatoid factor (RF) positive. The median modified Health Assessment Questionnaire score was 0.7, and the average Disease Activity Score in 28 joints was 3.8. The erosion score was significantly correlated with total hip BMD (r = −0.33, P < 0.0001), but not with lumbar spine BMD (r = −0.09, P = 0.27). Hip BMD was significantly lower in RF-positive patients versus RF-negative patients (P = 0.02). In multivariable models that included age, body mass index, and cumulative oral glucocorticoid dose, neither total hip BMD nor lumbar spine BMD was significantly associated with focal erosions.

Conclusion

Our results suggest that hip BMD is associated with focal erosions among postmenopausal women with RA, but that this association disappears after multivariable adjustment. While BMD and erosions may be correlated with bone manifestations of RA, their relationship is complex and influenced by other disease-related factors.

Rheumatoid arthritis (RA) is the most common form of inflammatory arthritis, affecting ∼1% of the adult population and almost 3% of persons over age 65 (1). In addition to the pain and functional limitations resulting from arthritis, RA patients experience 2 forms of disabling bone disease: focal erosions and generalized osteoporosis (OP). After 5 years of disease, 30–50% of RA patients exhibit evidence of focal erosions (2, 3). RA doubles the risk of OP and fractures compared with age- and sex-matched controls (4).

Focal erosions and OP have substantial clinical consequences and may be manifestations of a similar inflammatory cascade. Osteoblasts and osteoclasts play important roles in generalized OP, and both cell types have been implicated in focal erosions as well (5, 6). Moreover, inflammatory cytokines, such as tumor necrosis factor α (TNFα) and RANKL, play important roles in both processes (7, 8). Defining the relationship between these 2 RA-related processes may provide insight into the underlying pathophysiology of RA-related bone disease. Moreover, elucidating the process by which OP manifests in patients with RA may lead to new therapeutic strategies for both conditions, including novel drug targets and an understanding of the potential consequences of available OP treatments in patients with RA.

In this study, to better characterize the relationship between focal erosions and OP in RA patients, we recruited a cohort of postmenopausal women with RA to undergo bone mineral density (BMD) testing and hand radiography. We hypothesized that lower BMD would be associated with higher erosion scores.

PATIENTS AND METHODS

Study population.

We recruited postmenopausal women from a single-center longitudinally studied cohort of RA: the Brigham RA Sequential Study (BRASS) (9). All patients in the BRASS had been diagnosed as having RA by a board-certified rheumatologist. From this cohort, women considered to be eligible for the current study included those reporting postmenopausal status, who at the time were not taking any prescription medications for OP, including bisphosphonates, hormone replacement therapy, raloxifene, and calcitonin. Eligible women (n = 629) were sent an invitation letter allowing them to opt out of future contact. Of that group, 148 women opted out or could not be reached, leaving 481 whom we screened through telephone contact. From this group, 163 women agreed to participate and completed all parts of the study. Women who agreed to participate signed a consent form; this form and the study protocol were reviewed and approved by the Partners Healthcare System Institutional Review Board.

Data collection.

Women who agreed to participate underwent a brief interview to confirm that they were postmenopausal and were not currently taking any prescription medications for OP. The interview also determined their personal and family histories with regard to fragility fractures, use of calcium and vitamin D supplements, and intake of calcium from dietary sources. All patients who remained eligible after the brief interview were invited to undergo hip and lumbar spine dual x-ray absorptiometry (DXA), which was administered using a Hologic Discovery Bone Densitometer (Hologic, Bedford, MA). Vertebrae exhibiting artifacts from sclerosis or compression fractures were not included in readings.

To assure long-term stability of the instrument, anthropomorphic phantoms of the spine of known hydroxyapatite composition are scanned daily. The short-term in vivo–measurement standard deviations, estimated from duplicate scans of patients performed using the study scanner on the same day with repositioning, are 0.026 gm/cm² and 0.034 gm/cm², for posteroanterior spine and lateral spine, respectively. Long-term in vivo reproducibility, determined by calculating the residual variation around a line connecting measurements made at 0, 3, and 6 months for each patient in a prior study, is between 1% and 2% for each of the measurement sites, and the reproducibility is independent of the patient's absolute BMD.

A single experienced physician, blinded with regard to the patient's clinical status and erosion scores, evaluated the DXA scans. In a restricted sensitivity analysis, we excluded data on women whose lumbar spine DXA scan revealed substantial osteoarthritis (OA) or compression fractures. At the time of the DXA scans, patients also provided a blood specimen, which was used to measure serum calcium, parathyroid hormone, 25-hydroxyvitamin D (25[OH]D), and thyroid-stimulating hormone. The time of blood collection was not standardized.

Digitized hand radiography was performed every 24 months as part of the BRASS. The hand radiograph taken closest in time to the DXA was selected and read on a picture archiving and communication system display work station using 2K monitors for image viewing and analysis (Agfa Healthcare Informatics, Greenville, SC). The difference in months between the dates of DXA and hand radiography was controlled for in the adjusted models. Radiographs were analyzed using the Sharp method (10), by a board-certified radiologist who had been trained in these methods. The Sharp score consists of an erosion score and a joint space narrowing score, but for the current analyses, we focused only on the erosion score. The radiologist was blinded with regard to each subject's identity, BMD scores, and clinical status. Rereading of 8 radiographs by the same radiologist resulted in a coefficient of 0.97 (an intrarater reliability well within the expected range [11]).

As part of the BRASS protocol, questionnaires were completed by patients every 6 months and by their rheumatologists every 12 months. Variables assessed from these questionnaires included oral glucocorticoid use (current and past), disease-modifying antirheumatic drug (DMARD) use (current and past), Disease Activity Score in 28 joints (DAS28) (12), modified Health Assessment Questionnaire (M-HAQ) score (13), C-reactive protein level, rheumatoid factor (RF) status, anti–cyclic citrullinated peptide antibody status, and duration of RA. For variables whose values may change over time, the questionnaire completed closest to the time of the DXA scan was selected.

Statistical analysis.

The analyses focused on the relationship between erosion scores and BMD, measured at the hip and lumbar spine. Analyses assessing other relationships should be considered exploratory. We initially obtained Spearman correlation coefficients and then modeled the relationship using linear regression. Unadjusted bivariate models were examined, in which erosion score was the independent variable and BMD (total hip or lumbar spine) was the dependent variable. We then assessed confounding by testing covariates (such as disease duration, age, and body mass index [BMI]) in the bivariate linear regression models. Finally, multivariable models were constructed using a stepwise approach, keeping variables that remained significant at the P < 0.2 level. As noted above, a restricted analysis was performed by excluding the data on women whose lumbar spine DXA scan revealed substantial OA or compression fractures. In this group, multivariable models were constructed using a stepwise approach as noted above. We also explored effect modification by key variables that were found to be confounders. All analyses were conducted using SAS statistical software (SAS Institute, Cary, NC). Unless otherwise indicated, all data are reported as the mean ± SD.

RESULTS

The total study cohort consisted of 163 postmenopausal women with RA. The mean ± SD age of the women was 62 ± 9 years, and the average disease duration was 13.7 years (Table 1). Sixty-three percent of the patients were RF positive. The median M-HAQ score was 0.7 (interquartile range 0.3–1.9), and the mean ± SD DAS28 score was 3.8 ± 1.6. Almost all patients reported current use of a synthetic or biologic DMARD, and 69% reported current or past use of oral glucocorticoids. Of patients reporting glucocorticoid use, the median cumulative dose reported was 2,520 mg. In addition, 12% of the patients reported a personal history of fragility fracture (hip, spine, wrist, or upper arm), and 46% had taken an OP medication in the past, with 18% having taken a bisphosphonate.

Table 1. Characteristics of the 163 postmenopausal women with RA*

* Except where indicated otherwise, values are the number (%). During the study, no patients were receiving abatacept or rituximab. RA = rheumatoid arthritis; BMI = body mass index; RF = rheumatoid factor; CRP = C-reactive protein; IQR = interquartile range; DAS-CRP = Disease Activity Score using the CRP level; GC = glucocorticoid; TNF = tumor necrosis factor; DMARD = disease-modifying antirheumatic drug; M-HAQ = modified Health Assessment Questionnaire; 25(OH)D = 25-hydroxyvitamin D; OP = osteoporosis. † n = 154. ‡ Determined in patients with some prior use of oral glucocorticoids (n = 112). § n = 145.
Age, mean ± SD years	62.4 ± 9.0
BMI, mean ± SD kg/m^2	28.6 ± 6.6
Duration of RA, mean ± SD years	13.7 ± 10.9
RF positivity†	97 (63)
CRP, median (IQR) mg/dl	2.8 (1.3–6.6)
DAS-CRP, mean ± SD	3.8 ± 1.6
History of oral GC use	112 (69)
Cumulative oral GC dose, median (IQR) mg‡	2,520 (960–12,720)
Use of non–TNF antagonist DMARD during followup	116 (71)
Use of TNF antagonist during followup	72 (44)
M-HAQ score, median (IQR)	0.70 (0.3–1.9)
Exercise, mean ± SD days per week	1.0 (1–4)
History of fractures	19 (12)
Maternal history of fractures	33 (20)
Total calcium intake (dietary sources plus supplements), mean ± SD mg	929 ± 332
Use of calcium supplements	71 (44)
Serum 25(OH)D level, mean ± SD ng/dl	29 ± 11
Vitamin D deficiency (<20 ng/dl)§	29 (20)
Use of vitamin D supplements	125 (77)
Use of a bisphosphonate in the past	30 (18)
Use of a nonbisphosphonate OP medication in the past	75 (46)

Total hip and lumbar spine BMD results were available for 152 women and 144 women, respectively. The mean ± SD total hip BMD was 0.86 ± 0.12 gm/cm², and the mean lumbar spine BMD was 0.98 ± 0.15 gm/cm². Based on total hip and lumbar spine T scores, 7% and 14% of patients would have been considered as having OP, respectively. The erosion score was 36 ± 37, and 99% of women had ≥1 erosion.

The erosion score was significantly correlated with total hip BMD (r = −0.33, P < 0.0001) but not with lumbar spine BMD (r = −0.09, P = 0.27) (Table 2). This relationship is illustrated in Figure 1, where hip BMD is compared with erosion scores, demonstrating a strong negative relationship between hip BMD and quintile of erosion score. There was no correlation observed between lumbar spine BMD and erosion scores (Table 2). BMI was positively correlated with hip and lumbar spine BMD, and negatively correlated with erosion score. Duration of RA did not correlate with lumbar spine BMD, but it did correlate with total hip BMD and erosion scores. DAS28 score did not correlate with either hip or lumbar spine BMD, but did correlate with erosion score. Finally, cumulative oral glucocorticoid dose did correlate with both hip and lumbar spine BMD, as well as with erosion score.

Table 2. Spearman correlation coefficients between BMD, focal erosions, and clinical variables among the patients*

Variable	Total hip BMD	Lumbar spine BMD	Erosion score†
* BMD = bone mineral density (see Table 1 for other definitions). † Erosion component of the Sharp score. ‡ P < 0.1.
Lumbar spine BMD	0.60‡	–	−0.09
Total hip BMD	–	0.60‡	−0.33‡
Erosion score	−0.33‡	−0.09	–
Duration of RA	−0.14‡	0.0006	0.59‡
Exercise, days per week	−0.09	−0.11	−0.11
BMI	0.38‡	0.21‡	−0.24‡
Age	−0.27‡	−0.08	0.32‡
Age at onset of RA	−0.12	−0.09	−0.26‡
CRP	−0.01	−0.04	0.21‡
Total calcium intake	−0.11	−0.09	0.11
DAS-CRP	−0.05	−0.07	0.42‡
M-HAQ	−0.04	−0.02	0.11‡
Cumulative oral GC dose	−0.10	−0.003	0.26‡
Vitamin D level	−0.17‡	−0.12	0.16‡

Figure 1. Mean total hip bone mineral density in the patients by quintile of erosion score.

The median hip and lumbar spine BMD values were calculated for selected categories of patients (Table 3). The total hip BMD values were significantly lower in RF-positive patients (0.83 gm/cm²) than in RF-negative patients (0.89 gm/cm²) (P = 0.02). However, the lumbar spine values were almost identical in RF-positive patients (0.96 gm/cm²) and RF-negative patients (0.97 gm/cm²). Patients reporting current use of a TNF antagonist had significantly lower total hip BMD, and those reporting past bisphosphonate use had lower BMD at the total hip and lumbar spine. However, both of these findings are likely due to confounding by indication. The erosion scores were higher in RF-positive versus RF-negative patients.

Table 3. Association between discrete patient variables and bone manifestations of RA*

Variable	Total hip BMD		Lumbar spine BMD		Erosion
	Median (IQR)	P	Median (IQR)	P	Median (IQR)	P
* P values were calculated by Kruskal-Wallis test. See Table 1 for definitions.
RF
Positive	0.83 (0.76–0.91)	0.02	0.96 (0.88–1.03)	0.4	32 (15–71)	0.0001
Negative	0.89 (0.79–0.95)		0.97 (0.88–1.07)		13 (7–26)
CRP
Elevated	0.83 (0.77–0.95)	0.9	0.96 (0.86–1.11)	0.4	29 (13–70)	0.04
Normal	0.85 (0.77–0.92)		0.97 (0.89–1.05)		20 (7–46)
Maternal history of fracture
Yes	0.84 (0.74–0.93)	0.20	0.99 (0.88–1.06)	0.6	24 (7–44)	0.8
No	0.87 (0.78–0.93)		0.96 (0.88–1.06)		22 (8–58)
Prior fragility fracture
Yes	0.80 (0.71–0.93)	0.16	1.00 (0.89–1.03)	0.9	24 (7–68)	0.7
No	0.86 (0.77–0.93)		0.97 (0.88–1.07)		22 (8–51)
TNFα antagonist use
Yes	0.83 (0.77–0.90)	0.04	0.97 (0.88–1.03)	0.8	47 (21–83)	<0.001
No	0.88 (0.78–0.96)		0.97 (0.88–1.07)		13 (5–27)
DMARD use
Yes	0.85 (0.76–0.93)	0.5	0.97 (0.89–1.07)	0.21	24 (9–68)	0.07
No	0.86 (0.79–0.93)		0.96 (0.87–1.03)		20 (5–34)
History of oral GC use
Yes	0.84 (0.76–0.92)	0.14	0.97 (0.88–1.06)	0.9	16 (7–44)	0.19
No	0.87 (0.80–0.96)		0.97 (0.88–1.06)		24 (9–60)
History of bisphosphonate use
Yes	0.78 (0.74–0.88)	0.009	0.90 (0.84–1.02)	0.08	47 (23–70)	0.0005
No	0.87 (0.79–0.94)		0.97 (0.89–1.07)		19 (7–38)
History of nonbisphosphonate OP treatment
Yes	0.86 (0.77–0.92)	0.8	0.97 (0.87–1.03)	0.6	22 (9–59)	0.5
No	0.85 (0.77–0.94)		0.97 (0.89–1.08)		20 (7–51)
25(OH)D level
<20 ng/dl	0.87 (0.78–0.97)	0.2	1.01 (0.89–1.07)	0.2	21 (7–30)	0.3
≥20 ng/dl	0.85 (0.77–0.92)		0.96 (0.88–1.05)		23 (9–59)

In multivariable models that included age, BMI, and cumulative oral glucocorticoid dose, neither hip nor spine BMD was significantly associated with focal erosions (Table 4). In the full multivariable model, the only variable that remained an independent predictor of total hip and lumbar spine BMD was BMI (a higher BMI was associated with higher BMD). Several variables modified the effect of the relationship between the erosions and total hip BMD, including age, duration of RA, BMI, and cumulative oral glucocorticoid dose. These effects are explored in Table 5, where the parameter estimate for erosion scores in the fully adjusted model on all patients is compared with the fully adjusted parameter estimate in models stratified for one variable at a time. The relationship between erosion score and hip BMD was strengthened among younger patients (patients <62 years old versus patients ≥62 years old [62 years was the median age]), patients with shorter disease duration (<5 years versus ≥5 years), patients with a higher BMI (≥28 kg/m² versus <28 kg/m²; [28 kg/m² was the median BMI]), and patients with lower cumulative oral glucocorticoid intake (<960 mg versus ≥960 mg [960 mg was the median dose]). The sensitivity analysis, in which we removed data on patients with <4 vertebral bodies, yielded slightly different results for the relationship between lumbar spine BMD and erosions (Table 4). Notably, the relationship between lumbar spine BMD and erosions appeared to be stronger in this restricted analysis.

Table 4. Adjusted analysis for hip and lumbar spine BMD*

Variable	Total hip BMD as outcome (n = 152)				Lumbar spine BMD as outcome (n = 144)
	Partial model		Full model		Partial model		Full model A		Full model B
	Parameter estimate	P	Parameter estimate	P	Parameter estimate	P	Parameter estimate	P	Parameter estimate	P
* In the full model, all variables listed in Table 1 were tested for entry and remained if P < 0.2. The difference in months between the date of dual x-ray absorptiometry (DXA) scanning and hand radiography was controlled for in the partially and fully adjusted models. Full model A includes all patients with lumbar spine bone mineral density (BMD) measurements. Full model B excludes any woman whose DXA scan was limited by osteoarthritis or a fracture. See Table 1 for other definitions. † Variable forced into all models.
Erosion score†	−0.0047	0.070	−0.0044	0.10	0.00007	0.9	−0.0017	0.6	−0.0068	0.13
Age, per decade†	−0.0038	0.0001	−0.0040	0.0001	−0.0012	0.4	−0.0023	0.1	−0.0020	0.9
BMI†	0.0073	<0.0001	0.0069	<0.001	0.0052	0.006	0.0040	0.05	0.0044	0.10
Cumulative oral GC dose†	−0.0000011	0.17	−0.0000008	0.34	0.0000002	0.9	−0.0000003	0.6	−0.00058	0.6
Prior fragility fracture	–	–	−0.046	0.17	–	–	–	–	–	–
Any DMARD use	–	–	–	–	–	–	0.055	0.06	0.075	0.04
Prior bisphosphonate use	–	–	−0.05	0.03	–	–	–	–	–	–
Prior nonbisphosphonate OP treatment	–	–	0.03	0.02	–	–	–	–	–	–
CRP, per 1-unit increase	–	–	–	–	–	–	–	–	0.0032	0.03

Table 5. Modification of the effect of erosions on total hip BMD by selected patient characteristics*

	Total hip BMD as outcome
	Parameter estimate	P
* BMD = bone mineral density (see Table 1 for other definitions). † As shown in Table 4.
Fully adjusted model†	−0.0044	0.10
Stratified models
Patient age
<62 years	−0.006	0.08
≥62 years	−0.004	0.4
Disease duration
<5 years	−0.016	0.3
≥5 years	−0.005	0.1
BMI
<28 kg/m^2	−0.001	0.8
≥28 kg/m^2	−0.008	0.04
Cumulative GC dose
<960 mg	−0.010	0.02
≥960 mg	0.0004	0.9

DISCUSSION

The skeletal system is a major target organ for RA, but relatively little is known about how focal erosions relate to generalized OP. We studied this relationship in a cohort of postmenopausal women with a median duration of RA of 14 years. Total hip BMD correlated with erosion scores in unadjusted analyses; however, this relationship was not significant after controlling for other variables. This relationship appeared stronger among certain subgroups of patients, including patients with a shorter duration of disease, higher BMI, lower cumulative oral glucocorticoid dose, and lower 25(OH)D levels.

Our findings suggest that the relationship between focal erosions and generalized OP is complicated and modified by many aspects of RA and other factors. The fact that the relationship was stronger among patients with a shorter disease duration suggests that, with longer disease duration, other variables dilute the relationship between focal erosions and total hip BMD. These other variables might include the use of specific DMARDs (nearly universal in our study cohort), disease activity, and markers of inflammation. While we have good information on these factors for adjustment in a regression model, our information on these longitudinal variables is not perfect. Similar to disease duration, the cumulative oral glucocorticoid dose appears to blunt the relationship between focal erosions and BMD over time. Prior studies have shown that higher BMI is associated with a lower erosion score (14). This is supported by our findings (Table 2). Moreover, the relationship between total hip BMD and erosion score was most evident in patients with a higher BMI.

We found that the relationship of erosions with total hip BMD was stronger than that with lumbar spine BMD. There are several potential explanations for this apparent discrepancy across anatomic sites. It is possible that the process of inflammation underlying RA affects hip BMD more than lumbar spine BMD. On the one hand, because the vertebral bodies have a greater trabecular bone content than the total hip and because the trabecular bone is more metabolically active, one might anticipate that cytokine perturbations related to RA would affect lumbar spine more dramatically than total hip BMD. On the other hand, since total hip BMD may more closely relate to joint mobility and overall functional status than does lumbar spine BMD, it is possible that the effects of RA would be more apparent at the hip than at the lumbar spine. Furthermore, it is possible that DXA artifacts, known to affect the lumbar spine BMD more than the proximal femur, dilute the relationship between BMD and erosions differentially by anatomic site. This concept is supported by findings of the sensitivity analysis in which patients with <4 lumbar vertebral bodies were excluded; this analysis showed a strengthened relationship.

Several prior studies have examined the bone manifestations of RA. The COBRA (Combinatietherapie Bij Reumatoïde Artritis) randomized controlled trial tested a treatment regimen of high-dose glucocorticoids, methotrexate, and sulfasalazine versus sulfasalazine alone (15). Levels of bone resorption markers, such as urinary pyridinoline and deoxypyridinoline, were correlated with erosion scores, but no analyses were performed testing the correlation between erosions and BMD. While several studies have demonstrated a significant positive correlation between hand BMD and erosions, the relevance of these findings to the current study is not clear because of the difference in anatomic sites (7, 16, 17). A cohort study similar to the current one was conducted in Canada using 204 patients with early-onset RA. BMD at the hip and BMD at the spine were significantly correlated with Larsen scores (18) at baseline and after 2 years of followup in the women in the study, but not the men (19). A very recent post hoc analysis of the BeSt (Behandelstrategieën voor Reumatoïde Artritis) study (20) of patients with an RA duration of <2 years showed that progression of erosion scores was associated with reduction in BMD (21).

There were some important limitations to our study. First, the population for the current study was not extremely large. While post hoc power calculations suggests that the study had >80% power to detect a clinically important correlation between BMD and erosions, subgroup analyses were extremely limited. Second, the DXA scans and hand radiographs were not obtained simultaneously. The mean lag time between DXA scanning and radiography was 3.4 months, which was included as a covariate in our models. However, this lag time may have diluted the relationship between BMD and erosions.

Third, it is possible that the relationship between BMD and erosions was blunted in our population because of intensive treatment and supplemental vitamin D use. While it is true that our subjects received aggressive care, our cohort does not differ from several other cohorts in terms of vitamin D status and DAS28 scores (22–25). We did include the use of TNF antagonists and other DMARDs as covariates, since these agents have been found to affect BMD (26). Fourth, some patients had received medications for OP in the past. Restricted analyses excluding the data on women who had used bisphosphonates (n = 30) showed very similar results.

Fifth, our analysis included many clinical and some biologic variables, but data on some potential confounders such as tobacco use were not accurately collected in our study database. Finally, only one radiologist, trained in musculoskeletal radiology, read all the radiographs. This precluded consensus readings for radiographs that were difficult to interpret. However, the radiologist's intrarater reliability was excellent (r = 0.97), radiographs were obtained at a single center on very similar imaging processors without any radiographs requiring exclusion, and standard methods for interpretation (e.g., an erosion atlas) were used.

While there are limitations to the current study, it is important to note that this is one of the only studies to date that has focused on the relationship between two skeletal manifestations of RA. The patients included in the study were very well characterized from a clinical standpoint. Also, the vast majority of DXA measurements were conducted with a single bone densitometer, with very careful quality controls in place.

In conclusion, we found that hip BMD correlated with erosion scores among postmenopausal women with RA. However, this relationship was not statistically significant after adjustment for clinical factors. It appears that the relationship between BMD and erosions is stronger in patients with early RA. In conjunction with the findings from the BeSt trial, showing a longitudinal relationship between erosions and BMD in patients with <2 years of disease, this suggests that early RA would be an important population for future studies (27). It may be that the presumed association between erosions and BMD is most relevant with regard to patients with severe or early untreated RA. This information is likely to become increasingly important as more bone-directed treatments become part of RA management paradigms. Thus, it may become possible to treat multiple skeletal manifestations of RA with a single agent. Data from trials of denosumab (a monoclonal antibody directed against RANKL) suggest that it may be effective at improving BMD and reducing progression of erosion (28, 29). In addition to the clinical benefit of elucidating this relationship, further studies that include more biologic information may clarify important roles of inflammation in both focal erosions and OP in RA.

AUTHOR CONTRIBUTIONS

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. Solomon 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. Solomon, Finkelstein, Shadick, Winalski, Weinblatt, Gravallese.

Acquisition of data. Solomon, Finkelstein, Shadick, LeBoff, Winalski, Glass, Weinblatt, Gravallese.

Analysis and interpretation of data. Solomon, Finkelstein, Shadick, LeBoff, Stedman, Brookhart, Weinblatt, Gravallese.