Abstract
Objectives
This study investigated serum pentosidine levels as an advanced glycation end product (AGE)-related marker of bone matrix deterioration and examined the association between b/tsDMARD use and prevalent vertebral fractures in patients with RA in clinical remission.
Methods
Seventy-six patients with RA in clinical remission (DAS28-CRP < 2.3) were included. Serum pentosidine, bone turnover markers, and bone mineral density (BMD) were assessed. Lateral thoracolumbar spine radiographs were available for 51 patients, and prevalent vertebral fractures were evaluated using artificial-intelligence–assisted morphometry, with final physician confirmation. Multivariable regression analyses evaluated factors associated with serum pentosidine levels and prevalent vertebral fractures.
Results
Patients receiving b/tsDMARDs had lower serum pentosidine levels (P = 0.004) despite comparable BMD and bone turnover marker profiles. In multivariable linear regression, DAS28-ESR was associated with serum pentosidine levels (β = 0.00581, P = 0.011), while b/tsDMARD use showed a non-significant trend toward lower serum pentosidine levels (β = − 0.00544, P = 0.091). Vertebral fractures were numerically less common in patients receiving b/tsDMARDs (2/19 [10.5%] vs 11/32 [34.4%]). In logistic regression, older age was associated with higher odds of prevalent vertebral fractures (odds ratio 1.211 per year, 95% CI 1.074–1.439; P < 0.001), while b/tsDMARD use was associated with lower odds (odds ratio 0.144, 95% CI 0.015–0.841; P = 0.030).
Conclusions
In patients with RA in clinical remission, b/tsDMARD use was associated with lower odds of prevalent vertebral fractures. Residual inflammation, reflected by DAS28-ESR and serum pentosidine levels, may be relevant to skeletal fragility beyond BMD.
|
Key Points • Lower serum pentosidine levels were observed in b/tsDMARD-treated patients, whereas DAS28-ESR was independently associated with serum pentosidine levels in multivariable analysis. • In patients with rheumatoid arthritis in clinical remission, b/tsDMARD use was associated with lower odds of prevalent vertebral fractures. • Residual inflammation and bone matrix deterioration may be related to skeletal fragility beyond bone mineral density. |
Supplementary Information
The online version contains supplementary material available at 10.1007/s10067-026-08170-3.
Keywords: Rheumatoid arthritis, Clinical remission, Bone matrix quality, Pentosidine, Vertebral fracture, B/tsDMARDs
Introduction
Rheumatoid arthritis (RA) is a chronic autoimmune disorder marked by persistent inflammation that causes irreversible destruction of bone and cartilage within the joints [1]. In addition to local joint damage and erosion, systemic bone loss is a recognized consequence of RA, primarily due to chronic inflammation [2]. This bone loss is largely driven by elevated pro-inflammatory cytokines and reduced physical activity arising from joint deformities and muscle weakness. Studies estimate that the prevalence of osteoporosis in patients with RA is approximately 27.6% [3], nearly double that of age-matched controls in the general population [4]. Furthermore, patients with RA are at increased risk of fragility fractures, particularly in the proximal femur and vertebrae [5]. Notably, such fractures may occur even in patients with normal bone mineral density (BMD) [6], suggesting that compromised bone quality plays a significant role in skeletal fragility.
Bone strength is determined by both bone density and bone quality [7]. Bone volume is regulated through the dynamic balance between osteoblastic formation and osteoclastic resorption [8]. In contrast, a key determinant of bone quality is collagen cross-linking, which occurs in two forms: physiological enzymatic cross-links, which support mechanical strength and mineralization, and non-enzymatic cross-links, such as pentosidine, formed under oxidative and glycation stress. Excessive accumulation of non-enzymatic cross-links compromises collagen elasticity, resulting in brittle bone [9]. Elevated serum pentosidine levels—indicative of deteriorated bone quality—have been linked to a significantly increased risk of fragility fractures in patients with RA [10, 11].
Previous reports have shown that the use of biologic disease-modifying antirheumatic drugs (bDMARDs) is associated with lower serum and urinary pentosidine levels [12, 13]. Nevertheless, it remains uncertain whether biological or targeted synthetic DMARD (b/tsDMARD) use is associated with prevalent vertebral fractures in patients with RA who have achieved clinical remission, and how residual inflammatory activity relates to serum pentosidine levels in this setting.
The present study aimed to examine serum pentosidine levels and prevalent vertebral fractures according to b/tsDMARD use in patients with RA in clinical remission. We also explored factors associated with serum pentosidine levels, focusing on residual inflammatory activity assessed by Disease Activity Score in 28 joints based on C-reactive protein (DAS28-CRP) and erythrocyte sedimentation rate (DAS28-ESR).
Materials and methods
Patients
This retrospective observational study was conducted at the Departments of Orthopaedic Surgery at The Jikei University School of Medicine and The Jikei University Daisan Hospital. Informed consent was obtained using an opt-out approach. Patients with RA aged ≥ 40 years who met the 2010 classification criteria of the American College of Rheumatology/European League Against Rheumatism were consecutively assessed between 1 March 2013 and 30 April 2025. The index date was defined as the visit at which serum samples were obtained for bone matrix marker measurements. Because active RA is itself strongly associated with systemic bone loss and fracture risk through inflammatory and osteoclast-mediated mechanisms, the present study intentionally restricted the cohort to patients in clinical remission. This restriction was intended to reduce confounding by overt inflammatory activity and to examine residual skeletal fragility among patients who had achieved remission. Clinical remission was defined as DAS28-CRP < 2.3 at the index visit. Patients were excluded if DAS28-CRP could not be assessed because of missing visual analogue scale data, if DAS28-CRP was > 2.3, or if they had severe chronic kidney disease (CKD), defined as an estimated glomerular filtration rate (eGFR) < 30 mL/min/1.73 m2, because advanced CKD can cause marked disturbances in bone metabolism. The patient selection process is summarized in Fig. 1.
Fig. 1.
STROBE-style flow diagram of patient selection. A total of 138 patients with rheumatoid arthritis were assessed for eligibility. Of these, 62 were excluded because of missing VAS data precluding DAS28 assessment (n = 22), DAS28-CRP > 2.3 (n = 38), or severe chronic kidney disease (n = 2). Seventy-six patients in clinical remission were included in the main analysis, including 48 in the non-b/tsDMARD group and 28 in the b/tsDMARD group. Among these, thoracolumbar spine radiographs were available in 51 patients, who were included in the vertebral fracture analysis. Prevalent vertebral fractures were identified in 13 patients, whereas 38 had no vertebral fractures. Abbreviations: VAS, visual analogue scale; DAS28, Disease Activity Score in 28 joints; CRP, C-reactive protein; b/tsDMARD, biological or targeted synthetic disease-modifying antirheumatic drug
A total of 76 patients were included in the main analysis. Bone matrix and turnover markers were measured in all included patients. BMD at the lumbar spine and proximal femur was assessed when available (n = 45). Among the 76 included patients, lateral thoracolumbar spine radiographs were available for 51 patients and were used for vertebral fracture assessment. The remaining 25 patients did not undergo thoracolumbar spine radiography during the study period and were not included in the vertebral fracture analysis because radiographs were unavailable. Baseline characteristics were compared between patients with and without radiographs to evaluate potential selection bias (Supplementary Table S1). b/tsDMARDs administered in this cohort at the index visit included etanercept (n = 4), certolizumab pegol (n = 11), golimumab (n = 2), tocilizumab (n = 6), sarilumab (n = 2), abatacept (n = 1), tofacitinib (n = 1), and baricitinib (n = 1). This study was conducted in accordance with the Declaration of Helsinki and approved by the Ethics Review Committee of Jikei University Hospital (Approval Nos. 35–060 (11683) and 37–050 (12687)).
Measurement of serum biomarkers and RA disease activity
Serum pentosidine, tartrate-resistant acid phosphatase-5b (TRACP-5b), and total procollagen type 1 N-terminal propeptide (P1NP) levels were measured by SRL, Inc. (Tokyo, Japan) using enzyme-linked immunosorbent assay (ELISA). Disease activity was assessed using DAS28-CRP and DAS28-ESR. Medication data included the use of methotrexate (MTX), other conventional synthetic DMARDs (csDMARDs), b/tsDMARDs, glucocorticoids, bisphosphonates, selective oestrogen receptor modulators (SERMs), active vitamin D analogues, denosumab, parathyroid hormone (PTH) analogues, and romosozumab. Glucocorticoid exposure was assessed as current glucocorticoid use, daily glucocorticoid dose, and cumulative glucocorticoid exposure. Any anti-osteoporosis medication use was defined as the use of bisphosphonates, SERMs, active vitamin D analogues, denosumab, PTH analogues, or romosozumab. Radiographic bone erosion assessments were available only in a limited subset of patients and were therefore not included in the statistical analyses because of insufficient sample size.
Evaluation of vertebral fractures in the thoracic and lumbar spine
Vertebral fractures were assessed on lateral thoracic and lumbar spine radiographs (T4–L5) by two board-certified orthopaedic surgeons each with > 15 years of experience, who were blinded to clinical variables and treatment status. Evaluation was based on the Quantitative Measurement (QM) method, aided by artificial intelligence (AI)-assisted vertebral body morphometry software (Smart QM, Shimadzu Corporation, Kyoto, Japan) [14]. The AI-assisted morphometry software was based on a previously validated algorithm for quantitative morphometry of thoracic and lumbar vertebral bodies on lateral radiographs [14]. Using the software, vertebral heights were measured, and vertebral body height ratios—central-to-anterior (C/A), central-to-posterior (C/P), and anterior-to-posterior (A/P)—were calculated. Vertebrae were classified as fractured when the C/A or C/P ratio was < 0.80 or the A/P ratio was < 0.75 [15]. Although AI-assisted morphometry was used to support quantitative measurements and improve measurement consistency, the final determination of vertebral fracture status was always made by physicians.
Inter-reader agreement for patient-level vertebral fracture status was assessed using Cohen’s kappa before adjudication. The two primary readers agreed in 50 of 51 patients, indicating excellent inter-reader agreement (Cohen’s κ = 0.947). The single discrepant case was adjudicated by a third board-certified orthopaedic surgeon, and the final determination was made by majority decision.
Statistical analysis
Statistical analysis was conducted according to previously described methods [16]. All data are presented as mean ± standard deviation (SD) or number (%), as appropriate. Group comparisons were performed using the two-tailed Student’s t-test, Wilcoxon rank-sum test, Fisher’s exact test, or chi-squared test, as appropriate. Multivariable linear regression analysis was used to identify factors associated with serum pentosidine levels, and P values were derived from partial F-tests based on single-term deletion (drop1 in R, test = "F"). Additional sensitivity analyses for serum pentosidine levels were performed by adding cumulative glucocorticoid exposure and any anti-osteoporosis medication use to the primary linear regression model. Multivariable logistic regression analysis was performed to evaluate factors associated with prevalent vertebral fractures, and results are reported as odds ratios with 95% confidence intervals (CIs). Because the number of prevalent vertebral fracture events was limited, the primary logistic regression model included age, eGFR, and b/tsDMARD use. Exploratory sensitivity analyses were performed using separate logistic regression models to assess the robustness of the association between b/tsDMARD use and prevalent vertebral fractures while avoiding overfitting. These models additionally accounted for glucocorticoid use, daily glucocorticoid dose, cumulative glucocorticoid exposure, RA disease duration, and any anti-osteoporosis medication use. For the logistic regression models, P values were derived from likelihood ratio tests based on single-term deletion (drop1 in R, test = "Chisq"). No formal a priori sample size calculation was performed because this was a retrospective exploratory study based on available eligible patients. Because of the exploratory nature of the study, no adjustment for multiple comparisons was performed. Missing data were handled by complete-case analysis for each outcome and model. Patients without available thoracolumbar spine radiographs were not included in analyses of prevalent vertebral fractures, whereas they were retained in biomarker-related analyses when the relevant data were available. No imputation was performed. A P value < 0.05 was considered statistically significant. All analyses were performed using R version 4.5.1 (R Foundation for Statistical Computing, Vienna, Austria), with key results cross-checked using JMP version 18 (SAS Institute Inc., Cary, NC, USA). Figures were generated using GraphPad Prism version 10 (GraphPad Software, San Diego, CA, USA).
Results
Low serum pentosidine levels in b/tsDMARD-treated patients with RA in remission
To examine the relationship between b/tsDMARD use and bone status in patients with RA, individuals in clinical remission (DAS28-CRP < 2.3) were stratified according to b/tsDMARD treatment. BMD at the lumbar spine and proximal femur, along with serum bone turnover and matrix markers, was compared between groups. No significant differences in BMD were observed at either the lumbar spine (L1–L4) or proximal femur (P = 0.801 and P = 0.703, respectively) (Supplementary Fig. S1). Among biochemical markers, serum pentosidine levels—an AGE-related marker of bone matrix quality—were significantly lower in patients receiving b/tsDMARDs than in those not receiving these agents in the unadjusted comparison (P = 0.004) (Fig. 2A). In contrast, no significant group differences were observed for TRACP-5b, a bone resorption marker (P = 0.177), or total P1NP, a marker of bone formation (P = 0.708) (Fig. 2B and C).
Fig. 2.
Comparison of mean serum concentrations of bone matrix (A) and turnover (B and C) markers between b/tsDMARD and non-b/tsDMARD groups. Scatter plots with fitted regression lines and 95% confidence intervals showing associations between serum pentosidine levels and DAS28-CRP (D) and DAS28-ESR (E). Sample sizes: non-b/tsDMARD group, n = 48; b/tsDMARD group, n = 28; scatter plots, n = 76. Data are presented as mean ± SD. Statistical analyses were performed using Student’s t-test (A-C) and Pearson’s correlation test (D and E). Abbreviations: b/tsDMARD, biological or targeted synthetic disease-modifying antirheumatic drug; TRACP-5b, tartrate-resistant acid phosphatase-5b; P1NP, procollagen type 1 N-terminal propeptide; DAS28, Disease Activity Score in 28 joints; CRP, C-reactive protein; ESR, erythrocyte sedimentation rate
Determinants of serum pentosidine levels in patients with RA in clinical remission
To identify variables influencing serum pentosidine levels beyond b/tsDMARD use, patient characteristics were compared (Table 1). Significant differences were found in age (P = 0.001) and disease activity, as measured by DAS28-CRP (P = 0.009) and DAS28-ESR (P = 0.020). Additionally, the proportion of patients using csDMARDs other than methotrexate (MTX) was significantly lower in the b/tsDMARD group (P = 0.003), consistent with standard treatment sequencing in RA.
Table 1.
Characteristics of participants stratified by antirheumatic drug use in the remission group
| Variables | Total (n = 76) |
non-b/tsDMARDs (n = 48) |
b/tsDMARDs (n = 28) |
P value |
|---|---|---|---|---|
| Age (years) | 69.5 ± 1.2 | 72.3 ± 1.3 | 64.7 ± 1.9 | 0.001 |
| Sex, female (%) | 61 (80.3) | 37 (77.1) | 24 (85.7) | 0.551 |
| Duration of RA disease (years) | 15.2 ± 1.4 | 14.9 ± 1.8 | 15.7 ± 2.3 | 0.795 |
| DAS28-CRP | 1.50 ± 0.04 | 1.57 ± 0.05 | 1.36 ± 0.05 | 0.009 |
| DAS28-ESR | 1.87 ± 0.09 | 2.03 ± 0.10 | 1.59 ± 0.15 | 0.020 |
| HbA1c (%) | 5.8 ± 0.1 | 5.9 ± 0.1 | 5.7 ± 0.1 | 0.147 |
| eGFR (mL/min/1.73 m2) | 68.9 ± 1.9 | 67.7 ± 2.5 | 71.1 ± 2.9 | 0.391 |
| Methotrexate use (%) | 47 (61.8) | 28 (58.3) | 19 (67.9) | 0.410 |
| Other csDMARD use (%) | 33 (43.4) | 27 (56.3) | 6 (21.4) | 0.003 |
| Prevalence of osteoporosis (%) | 40 (52.6) | 27 (56.3) | 13 (46.4) | 0.408 |
| Glucocorticoid use (%) | 29 (38.2) | 22 (45.8) | 7 (25.0) | 0.071 |
| Glucocorticoid daily dose (mg/day) | 1.4 ± 0.2 | 1.7 ± 0.3 | 0.9 ± 0.3 | 0.061 |
| Cumulative glucocorticoid exposure (g) | 5.8 ± 1.0 | 6.7 ± 1.3 | 4.2 ± 1.6 | 0.064 |
| Any anti-osteoporosis medication use (%) | 40 (52.6) | 28 (58.3) | 12 (42.9) | 0.192 |
| Bisphosphonate use (%) | 13 (17.1) | 9 (18.8) | 4 (14.3) | 0.757 |
| SERM use (%) | 6 (7.9) | 6 (12.5) | 0 (0.0) | 0.079 |
| Vitamin D use (%) | 14 (18.4) | 10 (20.8) | 4 (14.3) | 0.553 |
| Denosumab use (%) | 10 (13.2) | 7 (14.6) | 3 (10.7) | 0.737 |
| PTH use (%) | 2 (2.6) | 0 (0.0) | 2 (7.1) | 0.133 |
| Romosozumab use (%) | 1 (1.3) | 0 (0.0) | 1 (3.6) | 0.368 |
Data are presented as mean ± SD or number (%). Any anti-osteoporosis medication use included bisphosphonates, SERMs, active vitamin D analogues, denosumab, PTH analogues, and romosozumab. Statistical analyses were performed using Student’s t-test, Wilcoxon rank-sum test, Fisher’s exact test, or chi-squared test, as appropriate. The Wilcoxon rank-sum test was used for glucocorticoid daily dose and cumulative glucocorticoid exposure because these variables were not normally distributed
RA rheumatoid arthritis, DAS28 Disease Activity Score in 28 joints, CRP C-reactive protein, ESR erythrocyte sedimentation rate, HbA1c hemoglobin A1c, eGFR estimated glomerular filtration rate, csDMARD conventional synthetic disease-modifying antirheumatic drug, SERM selective oestrogen receptor modulator, PTH parathyroid hormone
To explore factors independently associated with serum pentosidine levels, multivariable linear regression analysis was performed using age, DAS28-CRP, DAS28-ESR, and b/tsDMARD use as explanatory variables (Table 2). DAS28-ESR was independently associated with serum pentosidine levels (β = 0.00581, P = 0.011), whereas b/tsDMARD use showed a non-significant trend toward lower serum pentosidine levels (β = − 0.00544, P = 0.091). In contrast, DAS28-CRP (β = − 0.00651, P = 0.201) and age per 10 years (β = 0.00046, P = 0.747) were not associated with serum pentosidine levels.
Table 2.
Multivariable linear regression analysis of factors associated with serum pentosidine levels
| Variables | Beta | 95% CI | P value | |
|---|---|---|---|---|
| Lower | Upper | |||
| b/tsDMARD use | − 0.00544 | − 0.0118 | 0.0009 | 0.091 |
| Age (per 10 years) | 0.00046 | − 0.0024 | 0.0033 | 0.747 |
| DAS28-CRP | − 0.00651 | − 0.0166 | 0.0035 | 0.201 |
| DAS28-ESR | 0.00581 | 0.0014 | 0.0103 | 0.011 |
CI confidence interval, b/tsDMARD biological or targeted synthetic disease-modifying antirheumatic drug, DAS28 Disease Activity Score in 28 joints, CRP C-reactive protein, ESR erythrocyte sedimentation rate
In sensitivity analyses additionally including any anti-osteoporosis medication use and cumulative glucocorticoid exposure, DAS28-ESR remained significantly associated with serum pentosidine levels, whereas any anti-osteoporosis medication use and cumulative glucocorticoid exposure were not associated with pentosidine levels (Supplementary Table S2). The non-significant trend toward lower pentosidine levels with b/tsDMARD use was attenuated after additional adjustment for cumulative glucocorticoid exposure.
Correlation analysis further evaluated the relationship between disease activity and serum pentosidine levels. No significant correlation was observed with DAS28-CRP (R = 0.071, P = 0.547) (Fig. 2D), whereas a weak but statistically significant positive correlation was identified between serum pentosidine levels and DAS28-ESR (R = 0.319, P = 0.006) (Fig. 2E).
Association between b/tsDMARD use and lower odds of prevalent vertebral fractures
Baseline characteristics were largely comparable between patients with and without available thoracolumbar spine radiographs (Supplementary Table S1). Patients without radiographs had a shorter RA duration, whereas serum pentosidine levels and b/tsDMARD use were similar between the two groups.
To assess whether b/tsDMARD use was associated with prevalent vertebral fractures, the prevalence of vertebral fractures was compared between patients with and without b/tsDMARD treatment among patients with available thoracolumbar spine radiographs. In the unadjusted two-by-two comparison, vertebral fractures were numerically less common in patients receiving b/tsDMARDs than in those not receiving these agents (2/19 [10.5%] vs 11/32 [34.4%]; P = 0.096) (Supplementary Table S3). In the same two-by-two comparisons, neither use of any anti-osteoporosis medication nor glucocorticoid use differed significantly according to prevalent vertebral fracture status (Supplementary Table S3). No significant differences in BMD at the lumbar spine or proximal femur were found (P = 0.297 and P = 0.210, respectively) (Supplementary Fig. S2), which may suggest that factors other than BMD are relevant to fracture susceptibility.
Clinical characteristics were then compared between patients with and without vertebral fractures (Table 3). Significant differences were observed in age (P < 0.001), eGFR (P = 0.033), glucocorticoid daily dose (P = 0.029), and cumulative glucocorticoid exposure (P = 0.031). Logistic regression incorporating age, eGFR, and b/tsDMARD use as explanatory variables showed that older age was associated with higher odds of prevalent vertebral fractures, whereas b/tsDMARD use was associated with lower odds (Table 4). The odds ratio for age per year was 1.211 (95% CI, 1.074–1.439; P < 0.001), and for b/tsDMARD use was 0.144 (95% CI, 0.015–0.841; P = 0.030).
Table 3.
Characteristics of participants stratified by vertebral fracture status
| Variables | Total (n = 51) |
with vertebral fracture without vertebral fracture | P value | |
|---|---|---|---|---|
| (n = 13) | n = 38) | |||
| Age (years) | 69.4 ± 1.3 | 77.3 ± 2.1 | 66.7 ± 1.3 | < 0.001 |
| Sex, female (%) | 39 (76.5) | 10 (76.9) | 29 (76.3) | > 0.999 |
| Duration of RA disease (years) | 17.2 ± 1.9 | 20.8 ± 5.0 | 16.0 ± 1.8 | 0.262 |
| DAS28-CRP | 1.48 ± 0.05 | 1.55 ± 0.09 | 1.45 ± 0.05 | 0.363 |
| DAS28-ESR | 1.74 ± 0.11 | 2.00 ± 0.23 | 1.65 ± 0.12 | 0.182 |
| HbA1c (%) | 5.8 ± 0.1 | 5.8 ± 0.2 | 5.8 ± 0.1 | 0.894 |
| eGFR (mL/min/1.73 m2) | 70.4 ± 2.0 | 63.2 ± 4.0 | 72.9 ± 2.1 | 0.033 |
| Pentosidine (μg/mL) | 0.0545 ± 0.0020 | 0.0585 ± 0.0046 | 0.0532 ± 0.0021 | 0.274 |
| TRACP-5b (mU/dL) | 278.6 ± 23.2 | 235.8 ± 40.6 | 293.2 ± 27.5 | 0.291 |
| total P1NP (μg/L) | 51.7 ± 3.8 | 41.8 ± 6.8 | 55.1 ± 4.4 | 0.132 |
| Methotrexate use (%) | 32 (62.7) | 6 (46.2) | 26 (68.4) | 0.152 |
| Other csDMARD use (%) | 21 (41.2) | 8 (61.5) | 13 (34.2) | 0.109 |
| b/tsDMARD use (%) | 19 (37.3) | 2 (15.4) | 17 (44.7) | 0.096 |
| Prevalence of osteoporosis (%) | 28 (54.9) | 9 (69.2) | 19 (50.0) | 0.336 |
| Glucocorticoid use (%) | 17 (33.3) | 6 (46.2) | 11 (28.9) | 0.315 |
| Glucocorticoid daily dose (mg/day) | 1.1 ± 0.3 | 2.3 ± 0.6 | 0.7 ± 0.2 | 0.029 |
| Cumulative glucocorticoid exposure (g) | 6.4 ± 1.4 | 12.9 ± 3.5 | 4.2 ± 1.3 | 0.031 |
| Any anti-osteoporosis medication use (%) | 28 (54.9) | 9 (69.2) | 19 (50.0) | 0.336 |
| Bisphosphonate use (%) | 7 (13.7) | 3 (23.1) | 4 (10.5) | 0.352 |
| SERM use (%) | 4 (7.8) | 1 (7.7) | 3 (7.9) | > 0.999 |
| Vitamin D use (%) | 10 (19.6) | 3 (23.1) | 7 (18.4) | 0.702 |
| Denosumab use (%) | 9 (17.6) | 4 (30.8) | 5 (13.2) | 0.208 |
| PTH use (%) | 2 (3.9) | 0 (0.0) | 2 (5.3) | > 0.999 |
| Romosozumab use (%) | 1 (2.0) | 0 (0.0) | 1 (2.6) | > 0.999 |
Data are presented as mean ± SD or number (%). Any anti-osteoporosis medication use included bisphosphonates, SERMs, active vitamin D analogues, denosumab, PTH analogues, and romosozumab. Statistical analyses were performed using Student’s t-test, Wilcoxon rank-sum test, Fisher’s exact test, or chi-squared test, as appropriate. The Wilcoxon rank-sum test was used for glucocorticoid daily dose and cumulative glucocorticoid exposure because these variables were not normally distributed
RA rheumatoid arthritis, DAS28 Disease Activity Score in 28 joints, CRP C-reactive protein, ESR erythrocyte sedimentation rate, HbA1c hemoglobin A1c, eGFR estimated glomerular filtration rate, TRACP-5b tartrate-resistant acid phosphatase-5b, P1NP procollagen type 1 N-terminal propeptide, csDMARD conventional synthetic disease-modifying antirheumatic drug, b/tsDMARD biological or targeted synthetic disease-modifying antirheumatic drug, SERM selective oestrogen receptor modulator, PTH parathyroid hormone
Table 4.
Multivariable logistic regression analysis of factors associated with prevalent vertebral fractures
| Variables | Odds ratio | 95% CI | P value | |
|---|---|---|---|---|
| Lower | Upper | |||
| Age (per 1 year) | 1.211 | 1.074 | 1.439 | < 0.001 |
| eGFR | 0.963 | 0.893 | 1.026 | 0.250 |
| b/tsDMARD use | 0.144 | 0.015 | 0.841 | 0.030 |
CI confidence interval, eGFR estimated glomerular filtration rate, b/tsDMARD biological or targeted synthetic disease-modifying antirheumatic drug
Because glucocorticoid exposure differed between patients with and without vertebral fractures, exploratory sensitivity analyses were performed using separate logistic regression models to avoid overfitting. The association between b/tsDMARD use and lower odds of prevalent vertebral fractures remained consistent after additionally accounting for glucocorticoid use, daily glucocorticoid dose, cumulative glucocorticoid exposure, RA disease duration, and any anti-osteoporosis medication use (odds ratio range, 0.096–0.149; all P < 0.05) (Supplementary Table S4).
Discussion
The assessment of BMD, bone turnover, and matrix markers in patients with RA in clinical remission revealed no significant differences in BMD or bone turnover markers between those with and without b/tsDMARD treatment. In contrast, serum pentosidine levels—an AGE-related marker reflecting bone matrix quality—were lower in the b/tsDMARD-treated group in the unadjusted comparison. These findings suggest that mechanisms beyond BMD, potentially including bone matrix deterioration, may be relevant to skeletal fragility in RA remission; however, causal inferences cannot be made from this retrospective observational study.
Previous studies have demonstrated that serum pentosidine levels are influenced by factors such as age, renal function, and glucose tolerance [17, 18]. In the present study, eGFR and glycated hemoglobin A1c (HbA1c) did not differ significantly between groups, although the mean age was significantly lower in those receiving b/tsDMARDs. This age disparity may reflect a clinical tendency to avoid prescribing b/tsDMARDs to older adults because of concerns regarding adverse effects. To account for this potential confounding, multivariable regression analysis including age, DAS28-CRP, DAS28-ESR, and b/tsDMARD use was performed. In this model, b/tsDMARD use showed only a non-significant trend toward lower serum pentosidine levels after adjustment, whereas DAS28-ESR remained independently associated with pentosidine. This association remained consistent in sensitivity analyses additionally including cumulative glucocorticoid exposure and any anti-osteoporosis medication use. These results suggest that residual inflammatory activity may be related to bone matrix deterioration even in clinical remission.
Multivariable regression analysis identified DAS28-ESR as independently associated with serum pentosidine levels. While DAS28-ESR significantly correlated with serum pentosidine levels, DAS28-CRP did not. This discrepancy may stem from the nature of inflammation captured by each index, with CRP reflecting acute inflammation and ESR indicating subacute to chronic inflammatory activity [19]. Although both DAS28-CRP and DAS28-ESR are commonly used to evaluate disease activity in RA, they differ substantially. DAS28-CRP values are typically lower than DAS28-ESR scores, which may lead to underestimation of disease activity and overestimation of therapeutic efficacy [20, 21]. In this study, remission was defined using DAS28-CRP, yet it is plausible that some patients fulfilling this criterion still exhibited residual inflammatory activity. Prior research has suggested that serum pentosidine levels reflect chronic inflammation in patients with RA, as accumulation of advanced glycation end products is promoted by sustained inflammatory states [22, 23]. These observations are consistent with the possibility that lower serum pentosidine levels observed in b/tsDMARD-treated patients may partly reflect suppression of residual chronic inflammation, particularly when discrepancies exist between DAS28-CRP and DAS28-ESR. Effective suppression of disease activity and appropriate antirheumatic therapy may be important for maintaining bone health and bone-related outcomes in patients with RA, who face an increased risk of osteoporosis [24]. Therefore, DAS28-ESR may complement DAS28-CRP in capturing residual disease activity, although ESR should be interpreted with consideration of potential confounders such as age.
In the unadjusted comparison, vertebral fractures were numerically less common in patients receiving b/tsDMARDs, and multivariable logistic regression showed that b/tsDMARD use was associated with lower odds of prevalent vertebral fractures after adjustment for age and eGFR. The difference between the unadjusted and adjusted analyses suggests that age and renal function may have confounded the unadjusted comparison, although the small number of fracture events warrants cautious interpretation. No significant differences in BMD at the lumbar spine or proximal femur were observed between patients with and without vertebral fractures, which may suggest that factors other than BMD are relevant to fracture susceptibility. Sensitivity analyses additionally accounting for glucocorticoid exposure, RA disease duration, and any anti-osteoporosis medication use showed consistent results, although residual confounding cannot be excluded. Collectively, these findings are consistent with the possibility that mechanisms beyond changes in bone mass, potentially including differences in bone matrix quality, may be relevant to skeletal fragility in RA remission; however, causal relationships and effects on future fracture outcomes cannot be inferred from this observational study with cross-sectional assessment of prevalent vertebral fractures.
The study has several limitations. First, no significant difference was observed in serum pentosidine levels between patients with and without prevalent vertebral fractures. This may be due to the relatively small number of fracture cases included. Previous studies have shown that urinary pentosidine levels, which correlate with serum concentrations, are associated with increased fracture risk in general populations [25, 26], suggesting a potential link between pentosidine levels and vertebral fracture risk. However, the present study does not establish the incremental predictive value of serum pentosidine beyond BMD or established fracture risk factors. Second, vertebral fractures were assessed only in the subset of patients with available thoracic and lumbar spine radiographs. Although serum pentosidine levels and b/tsDMARD use were similar between patients with and without radiographs, the latter group had a shorter RA duration, which may have introduced selection bias and could influence the estimated prevalence of vertebral fractures. Therefore, the fracture-related findings should be interpreted cautiously and warrant confirmation in larger prospective cohorts with systematic imaging. Third, the observational design may be subject to confounding by indication and residual confounding. Although sensitivity analyses accounted for available fracture-related variables, including age, eGFR, glucocorticoid use, daily glucocorticoid dose, cumulative glucocorticoid exposure, RA disease duration, and any anti-osteoporosis medication use, several important fracture-related factors—such as menopause status, fall risk, prior fracture history, osteoporosis severity, and FRAX-related variables—were not systematically collected. In addition, individual bone-active drugs could not be evaluated separately because of the limited sample size and small number of prevalent vertebral fracture events. Fourth, the sample size was limited, and the vertebral fracture analysis included only 51 patients with 13 prevalent vertebral fracture events. No formal a priori sample size calculation was performed, and multiple comparisons were performed without adjustment for multiplicity. Therefore, the possibility of unstable estimates and type I error cannot be excluded, and non-significant trends, including the association between b/tsDMARD use and serum pentosidine levels after adjustment, should be interpreted cautiously. Finally, because this study included only patients with RA who had achieved clinical remission based on DAS28-CRP, the findings may not be generalizable to the broader RA population. The cross-sectional assessment of prevalent vertebral fractures also limits causal inference. Future prospective studies with systematic imaging, incident fracture assessment, and comprehensive evaluation of fracture risk factors are required to confirm these findings.
In patients with RA in clinical remission, b/tsDMARD use was associated with lower odds of prevalent vertebral fractures. Serum pentosidine levels were independently associated with DAS28-ESR, suggesting that residual inflammatory activity may be related to bone matrix deterioration even when remission is defined by DAS28-CRP. These findings support the concept that mechanisms beyond BMD, potentially including bone matrix quality, may influence fracture susceptibility in RA.
Supplementary Information
Below is the link to the electronic supplementary material.
Acknowledgements
The authors thank H. Takizawa for their assistance.
Authors' contributions
Y.Y. designed the experiments. Y.Y., K.M. and A.Z. contributed the development of methodology and the collection of samples. Y.Y analysed data. K.M and M.S. supervised the experiments. Y.Y. wrote the manuscript with support from K.M., A.Z. and M.S. All authors have critically reviewed and approved the final manuscript to be published.
Funding
This work was supported in part by JSPS KAKENHI Grant numbers 22K09411 (K.M.) and 24K19636 (Y.Y.) from the Ministry of Education, Culture, Sports, Science and Technology.
Compliance with ethical standards
Ethics approval and consent to participate
The study was conducted in accordance with the Declaration of Helsinki and was approved by the Ethics Review Committee of Jikei University Hospital (Approval Nos. 35–060 (11683) and 37–050 (12687)). Informed consent was obtained using an opt-out approach.
Conflict of interest
M.S. has a patent application pending related to the AI software used in this study. The software was developed in collaboration with Shimadzu Corporation. The other authors have disclosed no conflicts of interest.
Footnotes
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Smolen JS, Aletaha D, Barton A et al (2018) Rheumatoid arthritis. Nat Rev Dis Prim 4:1–23. 10.1038/nrdp.2018.1 [DOI] [PubMed] [Google Scholar]
- 2.Maeda K, Yoshida K, Nishizawa T et al (2022) Inflammation and bone metabolism in rheumatoid arthritis: molecular mechanisms of joint destruction and pharmacological treatments. Int J Mol Sci 23:2871. 10.3390/ijms23052871 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Moshayedi S, Tasorian B, Almasi-Hashiani A (2022) The prevalence of osteoporosis in rheumatoid arthritis patient: a systematic review and meta-analysis. Sci Rep 12:1–11. 10.1038/s41598-022-20016-x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Tanaka Y (2021) Managing osteoporosis and joint damage in patients with rheumatoid arthritis: an overview. J Clin Med 10:1241. 10.3390/jcm10061241 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Van Staa TP, Geusens P, Bijlsma JWJ et al (2006) Clinical assessment of the long-term risk of fracture in patients with rheumatoid arthritis. Arthritis Rheum 54:3104–3112. 10.1002/art.22117 [DOI] [PubMed] [Google Scholar]
- 6.Haugeberg G, Uhlig T, Falch JA et al (2000) Bone mineral density and frequency of osteoporosis in female patients with rheumatoid arthritis: Results from 394 patients in the Oslo County rheumatoid arthritis register. Arthritis Rheum 43:522–530. 10.1002/1529-0131(200003)43:3<522::AID-ANR7>3.0.CO;2-Y [DOI] [PubMed] [Google Scholar]
- 7.NIH Consensus Development Panel on Osteoporosis Prevention, Diagnosis, and Therapy (2001) Osteoporosis prevention, diagnosis, and therapy. JAMA 285:785–795. 10.1001/jama.285.6.785 [DOI] [PubMed] [Google Scholar]
- 8.Yamashita Y, Hayashi M, Liu A et al (2025) Fam102a translocates Runx2 and Rbpjl to facilitate Osterix expression and bone formation. Nat Commun 16. 10.1038/s41467-024-55451-z [DOI] [PMC free article] [PubMed]
- 9.Knani I, Bouzidi H, Zrour S et al (2018) Increased serum concentrations of Nɛ-carboxymethyllysine are related to the presence and the severity of rheumatoid arthritis. Ann Clin Biochem 55:430–436. 10.1177/0004563217733500 [DOI] [PubMed] [Google Scholar]
- 10.Okano T, Inui K, Tada M et al (2017) High frequency of vertebral fracture and low bone quality in patients with rheumatoid arthritis—results from TOMORROW study. Mod Rheumatol 27:398–404. 10.1080/14397595.2016.1213943 [DOI] [PubMed] [Google Scholar]
- 11.Mochizuki T, Yano K, Otani N et al (2025) Association between vertebral fractures and comorbidities in patients with rheumatoid arthritis: a cross-sectional study. J Bone Miner Metab 43:402–410. 10.1007/s00774-025-01597-9 [DOI] [PubMed] [Google Scholar]
- 12.Kageyama Y, Takahashi M, Ichikawa T et al (2008) Reduction of oxidative stress marker levels by anti-TNF-α antibody, infliximab, in patients with rheumatoid arthritis. Clin Exp Rheumatol 26:73–80 [PubMed] [Google Scholar]
- 13.Kageyama Y, Takahashi M, Nagafusa T et al (2008) Etanercept reduces the oxidative stress marker levels in patients with rheumatoid arthritis. Rheumatol Int 28:245–251. 10.1007/s00296-007-0419-1 [DOI] [PubMed] [Google Scholar]
- 14.Arakawa S, Shinohara A, Arimura D et al (2025) An automated algorithm for quantitative morphometry of thoracic and lumbar vertebral bodies in lateral radiographs. JBMR Plus 9. 10.1093/jbmrpl/ziaf017 [DOI] [PMC free article] [PubMed]
- 15.McCloskey EV, Spector TD, Eyres KS et al (1993) The assessment of vertebral deformity: a method for use in population studies and clinical trials. Osteoporos Int 3:138–147. 10.1007/BF01623275 [DOI] [PubMed] [Google Scholar]
- 16.Yamashita Y, Hayashi M, Saito M et al (2022) Osteoblast lineage cells-derived Sema3A regulates bone homeostasis independently of androgens. Endocrinology 163:1–7. 10.1210/endocr/bqac126 [DOI] [PubMed] [Google Scholar]
- 17.Nishizawa Y, Miura M, Ichimura S et al (2019) Executive summary of the Japan Osteoporosis Society Guide for the use of Bone Turnover Markers in the diagnosis and treatment of osteoporosis (2018 Edition). Clin Chim Acta 498:101–107. 10.1016/j.cca.2019.08.012 [DOI] [PubMed] [Google Scholar]
- 18.Sugiyama S, Miyata T, Ueda Y et al (1998) Plasma levels of pentosidine in diabetic patients: an advanced glycation end product. J Am Soc Nephrol 9:1681–1688. 10.1681/asn.v991681 [DOI] [PubMed] [Google Scholar]
- 19.Gabay C, Kushner I (1999) Acute-phase proteins and other systemic responses to inflammation. N Engl J Med 340:448–454. 10.1056/nejm199902113400607 [DOI] [PubMed] [Google Scholar]
- 20.Fleischmann RM, Van Der Heijde D, Gardiner PV et al (2017) DAS28-CRP and DAS28-ESR cut-offs for high disease activity in rheumatoid arthritis are not interchangeable. RMD Open 3:2–6. 10.1136/rmdopen-2016-000382 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Inoue E, Yamanaka H, Hara M et al (2007) Comparison of Disease Activity Score (DAS)28- erythrocyte sedimentation rate and DAS28- C-reactive protein threshold values. Ann Rheum Dis 66:407–409. 10.1136/ard.2006.054205 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Miyata T, Ishiguro N, Yasuda Y et al (1998) Increased pentosidine, an advanced glycation end product, in plasma and synovial fluid from patients with rheumatoid arthritis and its relation with inflammatory markers. Biochem Biophys Res Commun 244:45–49. 10.1006/bbrc.1998.8203 [DOI] [PubMed] [Google Scholar]
- 23.Hein GE, Köhler M, Oelzner P et al (2005) The advanced glycation end product pentosidine correlates to IL-6 and other relevant inflammatory markers in rheumatoid arthritis. Rheumatol Int 26:137–141. 10.1007/s00296-004-0518-1 [DOI] [PubMed] [Google Scholar]
- 24.Okabe H, Maeda K, Yamashita Y et al (2026) Biologic therapy is associated with faster bone union after the Sauvé-Kapandji procedure in patients with rheumatoid arthritis. Rheumatol Ther. 10.1007/s40744-026-00847-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Hagino H, Moriwaki K, Wada T et al (2023) Urinary pentosidine level is associated with the risk of fracture in community-dwelling older adults: a prospective observational study. Osteoporos Int 34:1703–1709. 10.1007/s00198-023-06816-5 [DOI] [PubMed] [Google Scholar]
- 26.Tanaka S, Saito M, Hagino H et al (2022) Association of urinary pentosidine levels with the risk of fractures in patients with severe osteoporosis: The Japanese Osteoporosis Intervention Trial-05 (JOINT-05). JBMR Plus 6:1–10. 10.1002/jbm4.10673 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.