Introduction

Systemic lupus erythematosus (SLE) is an autoimmune disease characterized by immunological hyperactivity and multi-system organ damage [1]. Multi-systemic organ damage in SLE, measured by Systemic Lupus International Collaborating Clinics (SLICC)/ACR Damage Index (DI), includes vertebral fractures among its components [2]. Low bone mineral density (BMD) has been widely analyzed in SLE patients. Its prevalence using WHO criteria (T score less than −1.0 SD) is found between 13% and 74% and osteoporosis (T score less than −2.5) between 3% to 42%. However, the possible associations with low BMD have controversial results. On the other hand, only few studies have evaluated vertebral fractures [3, 4].

Vertebral fractures are the hallmark of osteoporosis but it is not the only factor since bone quality also contributes. They are the most common osteoporotic fractures with prevalence estimates of 35% to 50% among those older than 50 years [5]. Only about 30% of vertebral fractures are clinically recognized [6]. Subjects with vertebral fractures have an increased risk for new vertebral and nonvertebral fractures [7, 8]. Furthermore, vertebral fractures provoke disability that conduce to a reduced quality of life [9, 10]. Long-term studies have shown an increased mortality after diagnosed with vertebral fractures [11, 12]. Several diagnosis methods could detect vertebral fractures, but only few have been standardized for trials. One of them is the semiquantitative method described by Genant et al. [13], which is shown to be reproducible for prevalent and incident vertebral fractures. Finally, vertebral fractures may be attributable to several factors, and one of them is osteoporosis (based on a standard definition of osteoporosis), which explains only 10% to 40% [14]. Corticosteroids (CTS), which are frequently administrated to SLE patients, are other major risk factor for fractures since CTS have larger effects on trabecular bone. The relationship between BMD and vertebral fractures is less clear in patients with osteoporosis caused by inflammatory and/or CTS use [15–17].

The aim of the current study was to analyze the role of traditional and SLE-related risk factors in the development of vertebral fractures.

Materials and methods

A cross-sectional study was performed in 210 women with SLE, who met at least four of the American College of Rheumatology (ACR) classification criteria for SLE [18] and were recruited from our cohort for this case–control study. The study was performed from 2006 to 2008 at the Systemic Autoimmune Disease Research Unit, Hospital General Regional No. 36 IMSS, Puebla, Mexico and at the Osteoporosis Clinic, Puebla, Mexico. The local ethics committee approved the study and all patients provided informed consent for their participation. Patients were included in the case study if they were ≥18 years with vertebral fractures and for controls if they were ≥18 years without vertebral fractures. Exclusion criteria for both case and controls were pregnancy, renal impairment (creatinine >2 mg/dl), and untreated thyroid disease. All participants underwent the following procedures: first, a structured interview conducted by one rheumatologist in order to collect traditional and potential SLE-related risk factors for fractures, then BMD measurement, and finally a spine morphometry. When assessments of vertebral fractures were interpreted, we fixed the case and controls.

Traditional risk factors for vertebral fractures

Study participants provided the following information, using a questionnaire and/or patients’ charts: age at study visit, self-designated race/ethnicity, premenopausal and postmenopausal status, first grade family history of fractures, history of fractures after 45 years, current smoking status, use of caffeine, exercise status, use of hormone preparations, calcium and vitamin D supplements, use of osteoporosis medication(s), thyroid hormones, diuretics, anti-epileptics, and anticoagulants. Body weight, height, and body mass index (BMI) were also assessed. Menopause was defined as amenorrhea of ≥6 months (except for women who had undergone a hysterectomy), and serum follicle-stimulating hormone levels of ≥ 30 IU/l were determined if age was less than 50 years, but there was history of irregular menses or partial hysterectomy.

Potential SLE-related risk factors

The factors were as follows: age at SLE diagnosis, duration of SLE, and cumulative disease damage as measured by the Systemic Lupus International Collaborating Clinics/American College of Rheumatology Damage Index (SLICC/ACR DI) [2]. A modified DI score was derived excluding the osteoporosis/fracture item (1 point) collected. Disease activity was scored using the Systemic Lupus Erythematosus Disease Activity Index (MEX-SLEDAI) [19] validated in the Mexican population. History of CTS use was assessed evaluating cumulative CTS dose.

Bone mineral density measurements

BMD was measured by dual energy X-ray absorptiometry using the same equipment Lunar DPX (Lunar Radiation Corp, Madison, WI, USA). Measurements were made at the hip (total hip and femoral neck) and lumbar spine (L1–L4; anteroposterior view) by a trained technician. All measurements were performed in accordance with standard instrument procedures and matched to gender, race, and weight. All BMD measurements were expressed in grams per square centimeter and as the number of standard deviations (SD) above or below the mean results of young female adults, T score. For patients with severe avascular necrosis of joint replacement on one hip, the BMD of the other hip was used. The precision (coefficient of variation in percentage) of the bone densitometer in measuring BMD at the hip and at the lumbar spine was 1.7% and 1.5%, respectively. On the basis of the WHO criteria for osteoporosis, patients were determined normal, osteopenia, or osteoporosis [20] in at least one region measurement.

Assessment of vertebral fractures

Lateral radiographs of the thoracic and lumbar spine (focusing on T8 and L3) were performed according to a standard protocol. They were performed by a trained operator and only one experienced radiologist scored the films. Standardized semiquantitative method described by Genant et al. [13] was used for scoring spine radiographs. For binary fracture vs. nonfractured comparison, a vertebral body was considered to be fractured if there was at least 20–25% reduction in anterior, middle, and/or posterior height. For quality assurance, consistency was determined by a blinded intraobserver. The kappa value for whether an SLE patient was classified as having any vertebral fracture was 0.73.

Statistical analysis

Data were expressed as mean ± standard deviation and percentages. Comparison between women in the fracture and no fracture groups was made by the Student’s t test (two-tailed) for continuous variables and the chi-square test for categorical data. Univariate logistic regression was used to screen traditional fracture risk factors, BMD, and SLE-related factors to identify the most relevant (p < 0.2) factors associated with fractures for inclusion in a multiple logistic regression model. Significant factors selected by univariate analysis were evaluated for multicollinearity. Results of the multiple logistic regression model are presented as adjusted odds ratios with corresponding 95% CI, excluding zero which indicated significant differences. Statistical analysis was completed using statistical software package SPSS version 10 (Chicago, IL, USA) for Windows Xp.

Results

We studied 210 women with SLE according to ACR criteria: 53 had at least one vertebral fracture, which were assigned to the case group, and 157 patients without vertebral fractures as controls. Vertebral fracture characteristics are shown in Table 1. Most of these fractures were grade 1 according to the Genant method. Concerning the location, thoracic spine was the most affected site.

Table 1 Assessment of vertebral fractures
Full size table

In relation with demographic, clinical, and treatment variables, we found the mean age of 50 ± 14 years for the fracture group and 41 ± 13 years for the control group (p = 0.001). According to recommendations, most of our patients were taking calcium supplements and vitamin D during the study period, but there were no differences among groups. Thirty of these were postmenopausal women, 7.5% of which had received hormonal replacement therapy. Twenty-four percent of the vertebral fracture group had never administered biphosphonates. The general characteristics of the patients studied are shown in Table 2. There were no differences in body mass index between patients with vertebral fractures and those without vertebral fractures, despite the fact that BMI has been associated with bone loss. Both groups were similar in terms of lifestyle factors, but the vertebral fracture group was more likely to have personal fracture history after they were 45 years old. The majority of postmenopausal women were in the case group compared with the controls (64.1% vs. 44.5%, p = 0.017). When we analyzed the prevalence of vertebral fractures in the premenopausal subgroup, it was found at 17.7%. Moreover, a higher proportion of women with vertebral fractures were postmenopausal and had taken biphosphonates. Regarding clinical variables, disease duration was similar between both groups. These patients had during the study period low level of disease activity measured by MEX-SLEDAI, but most of them have disease damage being higher in the vertebral fracture group (p = 0.014). The majority of patients had taken CTS and antimalarial drugs. There was a significant difference between cases and controls with cumulative CTS dose (p = 0.015) and it could influence the disease damage results. No significant differences were found regarding the proportion of women on antimalarials and daily dose of prednisone or its equivalent during the study period.

Table 2 Demographic, clinical, and treatment variables in women with SLE with and without vertebral fractures
Full size table

BMD measurements between cases and controls are shown in Table 3. The prevalence of normal BMD in our cohort was 40%. Osteopenia and osteoporosis depending on the WHO criteria were 55.2% and 16.7%, respectively. Osteopenia was not different in cases and controls; however, there was a difference in osteoporosis. When stratification of data by prevalence of vertebral fracture in relation to BMD was done, we found 22.6% in the group with normal BMD, 21.6% in the group with osteopenia, and 60% in the osteoporosis group. On the other hand, the proportion of normal BMD among patients with vertebral fractures was 35.8%.

Table 3 BMD variables in cases and controls
Full size table

Univariate analyses of traditional osteoporosis risk factors and SLE-related factors were performed to identify the most relevant factors associated with vertebral fractures. Of all the possible risk factors studied, only seven factors (age during the study period, postmenopausal status, personal history of fractures, chronic damage, biphosphonates use, cumulative CTS dose, and BMD at the total hip) were significantly associated with vertebral fractures. A multivariate regression analysis was made including significant univariate factors associated with vertebral fractures. The results of the multiple logistic regression model are shown in Table 4. Only three factors were significantly associated with vertebral fractures: age during the study period, CTS cumulative dose, and BMD at the total hip.

Table 4 Multiple regression analyses of vertebral fractures
Full size table

Discussion

Long-term complications have taken importance in SLE patients due to the improved life expectancy in those patients. There has been a lot of information about BMD in SLE patients, but few studies have analyzed the presence of vertebral fractures, even when these are risk factors for incident vertebral and nonvertebral fractures, which decrease quality of life and increase mortality rates. On the other hand, until now the risk factors associated with osteoporosis and vertebral fractures in SLE patients are controversial.

In our study, the prevalence of vertebral fractures was 25.2%, which is higher than other studies [3] using also a standardized method for vertebral fractures screening in both premenopausal and postmenopausal women. As it has previously been found, the majority of vertebral fractures were located in the thoracic spine and they were grade 1 [3, 4]. When the prevalence is only evaluated in premenopausal women, we found that 17.7% of them had vertebral fractures. This prevalence is lower than previously found [4] in SLE patients. Recently, the first study population in Latin American was published [21], where 11.18% prevalence of radiographically vertebral fractures was found in an age-stratified random sample of 1,922 women aged 50 years and older. There are other risk factors influencing our relatively young cohort with higher vertebral fracture prevalence. Therefore, it was necessary to examine the risk factors in this group of women.

There had been a lot of risk factors associated with osteoporosis and fractures, even in SLE patients. In order to get a good association between risks factors and vertebral fractures in our study, we chose those SLE patients who had consistently been related with low BMD. We found a lack of association between asymptomatic vertebral fractures and risk factors such as low body mass index, low dietary calcium intake, and smoking, which is consistent with the view that in patients given long-term CTS therapy, medication together with the disease could be by far the most important risk factors.

Chronic administration of CTS is associated with bone loss and increased fracture risk, particularly in the trabecular bone of the vertebrae. However, in SLE patients, this correlation is controversial when bone loss was analyzed attributing this to the differences in the patient population and methodological strategies used. Although we found a significant association of asymptomatic vertebral fractures with CTS, it was limited. Bultink et al. [3] found a relationship only with the use of intravenous methylprednisolone in 90 patients, and Borba et al. [4] did not find any correlation with CTS. A possible explanation for the weak association in our study could be attributed to a “threshold effect”, since according to the inclusion criteria, almost all patients were on current CTS treatment with a cumulative dose of ≥1,350 mg of prednisone equivalent. Besides, the cross-sectional design and the lack of a disease-related control group did not allow to distinguish CTS-related fractures from CTS unrelated ones. It cannot be excluded that a correlation would be apparent if only the CTS user were considered.

Another point relates to BMD values obtained by DXA. We only obtained evidence for their predictive role of prevalent asymptomatic vertebral fractures with low BMD at the total hip, but not with low BMD of the spine. Our findings are consistent with those by Ørstavik et al., who reported that in rheumatoid arthritis patients currently given CTS, neither bone ultrasound measurements nor DXA could discriminate between subjects with and without vertebral fractures [22]. Together with previous reports, our results lend support to the view that the particularly high risk of fractures in postmenopausal women given CTS therapy is accounted for by deterioration of the bone quality much more than diminution of bone mass. However, since we did not exclude fractured vertebrae from DXA analysis, the lack of association between lumbar spine BMD and asymptomatic vertebral fracture prevalence may also be accounted for by the increased BMD in patients with asymptomatic vertebral fractures. The explanation for this dissociation between bone quantity and quality involves the lifespan of osteocytes. It has been demonstrated that chronic CTS administration to mice and humans increases the prevalence of osteocyte and osteoblast apoptosis and an osteocyte accumulation exists [23]. This disruption could result, therefore, in defective bone material properties and increased fragility of bone, independent of any changes in bone mass. In a clinical study, Angeli et al. [24] provided that BMD determination was ineffective for predicting the number or severity of the fractures; however, vertebral fractures are strong risk factors for additional fractures, independent of BMD.

In any case, an important suggestion would be to perform vertebral fracture assessment in all patients who receive chronic CTS therapy. Although BMD measurements might be clinically useful in the follow-up of patients, they cannot completely be used to identify those at risk of CTS-induced osteoporosis.

In summary, the high prevalence of vertebral fractures in our study, both in premenopausal and postmenopausal CTS users with SLE, indicates that the evaluation for vertebral fractures in SLE patients should be performed, since they may be asymptomatic, and the high incidence of fractures worsens quality of life and increases mortality.