Introduction

Bone fractures are a major complication of rheumatoid arthritis (RA). We previously reported that despite an improvement in the disease activity of patients with RA, the incidence of fractures remained unchanged and the frequency of non-vertebral fractures in postmenopausal women with RA was high [1]. The occurrence of new fractures in patients with RA significantly reduces their quality of life. In order to reduce fracture risk in this patient population, it is important to identify predisposing risk factors associated with RA.

In the general population, vitamin D deficiency has been associated with an elevated risk for bone fractures, falls, and a poor response to antiresorptive therapy [2]. The results of several longitudinal studies have shown that a vitamin D deficiency and/or low levels of serum 25-hydroxy vitamin D (25[OH]D) are associated with an elevated risk for bone fractures [3,4,5,6,7]. We and others have previously reported that a vitamin D deficiency has been common among patients with RA [8,9,10,11,12,13]. Reportedly, low levels of vitamin D have been associated with various complications, such as disease activity and physical disability [9,10,11, 14], as well as cardiometabolic intermediates [15] in patients with RA. Longitudinal studies using cohort data have not addressed the association between vitamin D deficiency and fractures in patients with RA, although results of a cross-sectional study indicate that a significant association exists between vitamin D deficiency and osteoporotic fractures in patients with RA in France [16].

Using data from an observational cohort study at our institute (Institute of Rheumatology Rheumatoid Arthritis, or IORRA), we evaluated various clinical characteristics of osteoporosis in Japanese patients with RA [1, 12, 13, 17,18,19]. We previously reported clinical risk factors for fractures or falls [17,18,19] and factors associated with vitamin D deficiency and decreasing serum 25(OH) levels [12, 13, 19] in Japanese patients with RA. However, we did not evaluate vitamin D deficiency as a potential risk factor for fractures in our previous studies [17, 18].

In this study, we aimed to investigate the association between vitamin D deficiency and the occurrence of new fractures in Japanese patients with RA using our IORRA cohort study. Optimal serum levels of 25 (OH) D may depend on age, race, and ethnicity [20, 21]. A serum 25(OH)D level of < 20 ng/mL has been reported to increase fracture risk in Japanese [7] and Caucasians [22]. In this study, vitamin D deficiency was defined as a serum 25(OH)D level of < 20 ng/mL in accordance with the 2011 US Endocrine Society guideline [23] and the result of the public research in Japan [2], as we employed it our previous study [12, 13, 19]. Previous studies investigated the relationship between vitamin D deficiency [4, 5, 7] and fractures in women, and the focus of our current study is also women. This decision was prompted by the results of our previous cross-sectional study which revealed that vitamin D deficiency was more common in women [12] and our previous longitudinal study which showed that female gender was a significant risk factor for new fractures [17].

Methods

Patients

In October 2000, at the Institute of Rheumatology, Tokyo Women’s Medical University, the IORRA cohort was established as a single, institute-based, large, observational cohort of Japanese patients with RA. Details regarding the purpose and methodology of this study have been reported previously [1, 12,13,14,15,16,17,18,19]. Briefly, all Japanese patients diagnosed with RA at our institute were registered in the IORRA study and were asked to complete and submit a survey biannually. The IORRA study was approved by the Ethical Committee of the Tokyo Women’s Medical University (2922-R16). In this study, we analyzed female patients over the age of 50 years who participated in the IORRA study—between the 22nd survey in 2011 and the 34th survey in 2017—and documented their serum 25(OH)D in the 22nd survey. We extracted cases of self-reported new fractures from our cohort dataset and validated them against image findings and descriptions in medical records.

Primary and secondary outcomes

The primary outcome of this study was an association between vitamin D deficiency and the occurrence of new clinical and osteoporotic fractures. The secondary outcomes were associations between vitamin D deficiency and the occurrence of new osteoporotic and clinical vertebral and non-vertebral fractures.

Fracture assessments

Clinically recognized incidents of vertebral and non-vertebral fractures were enumerated from self-reports, as documented in a questionnaire of the IORRA study. Every 6 months, from April 2011 until September 2017, participants were asked about fractures at the ankle, arm, clavicle, elbow, foot, hand, hip, knee, leg, nose, pelvis, rib, shoulder, thoracic spine, lumbar spine, and wrist. Verification of self-reported fractures and fracture sites was achieved by a review of radiology reports or medical records. We excluded patients with self-reported fractures that we could not verify in this manner, as well as fractures resulting from a traffic accident or high energy trauma. We defined osteoporotic fractures as vertebral, proximal humerus, distal radius, proximal femur, pubic, and ischium fractures. Only the first fracture event reported by the patient was assessed in this study. Asymptomatic vertebral fractures were excluded given that routine thoracic and lumbar spine radiographs were not obtained for spinal morphometry.

25(OH)D measurement

In 2011, fresh serum samples, collected from patients who participated in our vitamin D study, were evaluated using the 25(OH)D 125I radioimmunoassay kit (DiaSorin, Italy) for quantitative determination of 25(OH)D [8, 9, 18]. Vitamin D deficiency was defined as a serum 25(OH)D level of < 20 ng/mL in accordance with previously reported deficiency thresholds [2, 20, 21].

Statistical analysis

Clinical characteristics were summarized according to serum 25(OH)D levels (< 20 and ≥ 20 ng/mL), and compared using Fisher’s exact test for categorical variables and Student’s t-tests for continuous variables. Accumulated incidence rates for any clinical fractures were compared using the Kaplan–Meier method and the log-rank test between groups categorized by serum 25(OH)D levels < 20 and ≥ 20 ng/mL. To determine the associations between the occurrence of new fractures in female Japanese patients over the age of 50 years with RA and patients’ clinical characteristics, Cox proportional hazard analyses with baseline and time-dependent covariates were conducted using IORRA data and employing an open cohort design. The hazard ratio (HR) and associated 95% confidence interval (CI) for each variable were estimated. All potential risk factors from among the reported factors available from our IORRA cohort database were selected. Confounding factors were selected based on findings in our previous studies [1, 12, 13, 17,18,19]. Continuous variables included age (per 10-year increase), body mass index (BMI), duration of RA, the 28-joint disease activity score (DAS28), disability measured using the Japanese Health Assessment Questionnaire Disability Index (J-HAQ) [24], daily prednisolone dosage, and serum 25(OH)D level. Categorical variables included a self-reported history of prior fracture, use of methotrexate (MTX), biological disease modifying anti-rheumatic drugs (bDMARDs), glucocorticoids, bisphosphonates, and active vitamin D3. Time-dependent variables were age, BMI, disease duration of RA, history of fractures, DAS28, J-HAQ, daily PSL dosage, bisphosphonate use, and active vitamin D3 use. The proportional hazards assumption was checked using a complementary plot of log–log survival. P < 0.05 was considered significant. All statistical analyses were performed using EZR (Saitama Medical Center, Jichi Medical University, Japan), which is a graphical user interface for R (The R Foundation for Statistical Computing, Vienna, Austria).

Results

Altogether, approximately 6000 Japanese patients with RA from the IORRA cohort were enrolled in the observational period for this study; of them, 2567 were female, over the age of 50 years, and included in this study. Demographic and clinical variables, as well as the use of medication at the time of enrollment or baseline in this study, are summarized in Table 1. Patients with a vitamin D deficiency were younger, had a higher J-HAQ, received a higher daily PSL dosage, had lower serum 25(OH) levels, and were more likely to take MTX, bDMARDs, and glucocorticoids compared with those without a vitamin D deficiency.

Table 1 Baseline demographic and clinical characteristics of Japanese postmenopausal patients with RA
Full size table

During a median (25% tile–75% tile) follow-up of 6.5 (5.3–7.5) years, among the included cases, the occurrence of new fractures was validated in 205 patients. Limited to new osteoporotic fractures, they were validated in 139 patients; these comprised 63 vertebral fractures and 76 non-vertebral fractures (Fig. 1).

Fig. 1
figure 1

Study flow diagram. The study population comprised 2567 Japanese female patients with RA over the age of 50 years, with and without new clinical and osteoporotic fractures. Patients participated in surveys between 2011 and 2017

Full size image

The rate of new fractures among all included cases was calculated for two groups according to serum 25(OH)D levels (i.e., < 20 and ≥ 20 ng/mL; Fig. 2). The group with serum 25(OH)D < 20 ng/mL had a higher fracture rate than those without a vitamin D deficiency in the group including all clinical fractures and the osteoporotic fracture group. The any clinical fracture incidence rate over 5 years was 8.4% for patients with serum 25(OH)D < 20 ng/mL and 6.6% for those with serum 25(OH)D ≥ 20 ng/mL. The any osteoporotic fracture incidence rate over 5 years was 5.9% for the patients with serum 25(OH)D < 20 ng/mL and 3.8% for those with serum 25(OH)D ≥ 20 ng/mL.

Fig. 2
figure 2

Accumulated incidence of any clinical fractures. The accumulated incidence rates of any clinical and osteoporotic fractures were compared using the Kaplan–Meier method and the log-rank test among the groups categorized by serum 25(OH)D levels. Dotted line: < 20 ng/mL (vitamin D deficiency), solid line: ≥ 20 ng/mL. a Any clinical fracture (*p = 0.0407). b Any osteoporotic fracture (*p = 0.0443)

Full size image

The results of the Cox proportional hazards analysis are summarized (Table 2). A vitamin D deficiency was significantly associated with the occurrence of any new clinical fractures; for any fracture, the adjusted hazard ratio (aHR) was 1.44, 95% confidence interval (CI) 1.02–2.05 (p = 0.0365). Likewise, vitamin D deficiency was significantly associated with the occurrence of new osteoporotic fractures. For any fracture, aHR was 1.75, 95% CI 1.14–2.69 (p = 0.0109); for vertebral fractures, aHR was 2.11, 95% CI 1.10–4.02 (p = 0.0243) and for non-vertebral fracture, aHR was 1.53; 95% CI 0.857–2.72 (p = 0.152). The significance of the increase in any osteoporotic fracture remained after adjusting for age, DAS28, J-HAQ, and daily PSL dosage.

Table 2 New fracture risk for vitamin D deficient postmenopausal Japanese patients with RA using the Cox model
Full size table

Discussion

We evaluated the association between vitamin D deficiency and the occurrence of new fractures in Japanese patients with RA (Table 2). Previously, several longitudinal studies reported the association between vitamin D deficiency and new bone fractures in the general population [3,4,5,6,7]. To the best of our knowledge, this is the first longitudinal study to show a significant correlation between vitamin D and fractures in Japanese patients with RA.

Our results show that vitamin D deficiency was significantly associated with new clinical fractures. In our previous study conducted using our cohort data from 2011 to 2013, 73.3% and 68.2% of the total number of enrolled Japanese patients with RA in 2011 and in 2013, respectively, had a serum 25(OH)D level of < 20 ng/mL [13]. These results indicate that a majority of Japanese patients with RA were vitamin D deficient. In addition, 8.8% of patients in this cohort had serum 25(OH)D levels that decreased by > 5 ng/mL over 2 years. We propose that measures aimed at preventing the loss of vitamin D in this group can be expected to reduce the risk of new fractures. Vitamin D supplementation has been reported to be effective, not only to counteract a vitamin D deficiency, but also to improve disease activity and limit functional disability in patients with RA [25,26,27,28]. Thus, Japanese patients with RA could mitigate a vitamin D deficiency by taking vitamin D supplements and by making certain they have adequate exposure to sunlight; this in turn has the potential to reduce the risk of new clinical fractures.

In this study, we did not observe any significant associations between a vitamin D deficiency and non-vertebral fractures after adjusting for age, BMI, disease duration of RA, history of fractures, DAS28, J-HAQ, daily PSL dosage, bisphosphonate use, and active vitamin D3 use, although a vitamin D deficiency was significantly correlated with any clinical fractures (Table 2). Tamaki et al. [7] reported significant associations between vitamin D deficiency and any fractures among Japanese community-dwelling women aged ≥ 50 years [7]. Low levels of serum 25(OH)D were reported to be associated with lower limb and vertebral fractures [5], and long bone fractures [6] in Japanese postmenopausal women. Future studies with a larger number of cases and longer follow-up periods are needed to make more conclusive associations between a vitamin D deficiency in patients with RA and non-vertebral fractures.

In this study, older age, low DAS28, high J-HAQ, and a high dosage of PSL were significantly associated with the occurrence of new osteoporotic fractures in Japanese patients with RA. We and others have already reported these associations [12, 13], and we reconfirm them in this study. There was an inverse result between DAS28 and J-HAQ with respect to their association with new osteoporotic fractures (Supplementary Table 1). In our previous studies, vitamin D deficiency was significantly associated with J-HAQ but not DAS28 [12, 13]. Several previous studies, including ours, reported that the HAQ disability score is significantly inversely associated with 25(OH)D levels [9, 11,12,13,14]. Furthermore, the inverse relationship between 25(OH)D levels and DAS28 was reported previously [9, 11, 14]. Those reports suggest that disability, rather than disease activity, may be a primary determinant of serum 25(OH)D levels in patients with RA. The results of our current study suggest that disability may cause decreasing serum vitamin D levels followed by the occurrence of new osteoporotic fractures in Japanese patients with RA.

This study has several limitations. First, we recognize that those data for patient characteristics and medications were obtained via self-reported questionnaires; as such, some degree of under or over reporting is likely. Second, we measured serum 25(OH)D levels once only; we did not take into consideration seasonal variations in serum 25(OH)D values as previously reported in longitudinal studies [3,4,5,6,7,8,9]. However, according to our previous Japanese cohort study, the mean annual rate of change in 25(OH)D values across all subjects was minimal, or 7.6% [29], and serum changes in 25(OH)D values have been shown to be no more than 5 ng/mL over 2 years in approximately 70% of the participants [19]. Furthermore, we did not evaluate the use of vitamin D supplements or changes in dietary intakes, as those data were not collected for this study. A third limitation concerns the relatively small number of patients sustaining new fractures. A fourth consideration relates to data undergoing corrections every 6 months, which meant that there were variations in the period between when the fracture occurred and correction of the data. Fifth, since the 25(OH)D assay kit which we used measured 25(OH)D with an old method, any vitamin D deficiency that we diagnosed could be under- or over-estimated. Sixth, although we selected confounding factors according to our previous studies [1, 12, 13, 17,18,19], any other confounding factors that we were unable to measure could have affected the results. Lastly, IORRA is a single-institution cohort study; therefore, our results may not be generalizable to all Japanese patients with RA, even though we estimate that our patients represent 1% of all patients in Japan with RA. We could not compare the characteristics between our patients with RA and all patients with RA in Japan given that our IORRA cohort is the largest database of its kind for Japanese patients with RA, and limited data exist in the literature summarizing the general characteristics of all patients with RA in Japan.

Conclusion

We investigated the association between vitamin D deficiency and the occurrence of new fractures in Japanese patients with RA in the IORRA cohort. The results of this study suggest that vitamin D deficiency appears to be a significant risk factor for new fractures in this population and screening patients with RA for serum 25(OH)D levels could potentially reduce the risk for new fractures.