Prevalence and predictive factors of osteoporosis in Thai systemic sclerosis

Systemic sclerosis (SSc) is recognized as a chronic inflammatory disease and several SSc-associated factors may increase the risk of osteoporosis and its related fractures. To determine the prevalence and predictive factors of osteoporosis in Thai SSc, a cross-sectional study was designed in adult SSc patients at Scleroderma clinic, Khon Kaen University Hospital. The prevalence of osteoporosis with the 95% confidence interval (CI) were determined and the odds ratio (OR) with 95%CI were assessed the clinical association with osteoporosis. A total of 205 SSc patients were recruited with the female to male ratio of 2.7:1. The majority of cases were diffuse SSc subset (83.4%) with a disease duration < 5 years (62.9%). The overall prevalence of osteoporosis was 29.3% (95%CI 23.1–36.0). After an age adjusted analysis, the respective prevalence of osteoporosis at lumbar spine (LS) in women and men was 26.3% and 10%, while the prevalence of osteoporosis at the femoral neck (FN) in women and men was 11% and 2.1%. Low BMI (≤ 18.5 kg/m2) and menopause were associated with osteoporosis at both the LS and FN. Using multivariate analysis, low BMI and menopause were associated with osteoporosis at LS (OR 7.78 and 5.32, respectively), while low BMI was also associated with osteoporosis at LS in pre-menopausal women. In conclusion, the prevalence of osteoporosis in Thai SSc was 29.3%. Osteoporosis at the LS is more common than FN in both men and women. Low BMI was associated with osteoporosis in overall SSc and pre-menopausal women, while only menopause was associated with osteoporosis at the FN.

the patients (a) who were diagnosed with overlap with other connective tissue diseases, cancer undergoing chemotherapy, chronic kidney disease, and hypothyroidism, (b) who were pregnant or breast feeding, and (c) who had been evaluated for bone mineral density (BMD) or nuclear medicine within 3 months.
Based on a previous estimate of the prevalence of osteoporosis in SSc (19.4%) 12 with a sampling variability of 5%, it was estimated that a sample size of at least 241 was needed for statistical adequacy to estimate the true prevalence of osteoporosis. After the retention period and a preliminary analysis, we found the prevalence of osteoporosis was 58 of 205 cases (29.3%) (95% confidence interval (CI) 23.1-36.0). We thus recalculated the sample size and collected 205 instead of 241 cases in order to achieve better patient care and reduce the cost of the study. This study was approved by the Human Research Ethics Committee of Khon Kaen University (HE551280). All eligible patients signed informed consent before enrolling in the study.
Demographic and clinical characteristics including age, sex, menopausal status, duration of disease, SSc subset, and family history of osteoporosis and fracture were collected. BMD at the lumber spine (LS) and femoral neck (FN) were measured using dual energy X-ray absorptiometry (DXA) by GE Lunar DPX Duo Densitometer. Osteoporosis was defined by a T-score of -2.5 SD, compared with the peak young adult mean for Thai women. The 25-hydroxy vitamin D [25(OH)D] level using direct competitive chemiluminescence immunoassay, thyroid and thyroid stimulating hormones, parathyroid hormone, calcium and phosphorus levels, and inflammatory markers [erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP)] were measured.
In this study, duration of disease was the interval between the date of osteoporosis assessment and the first SSc symptoms. Pulmonary fibrosis was defined when interstitial fibrosis was detected by either chest radiography or high resolution computed tomography (HRCT). Pulmonary arterial hypertension was defined when mean pulmonary arterial pressure was ≥ 25 mmHg as confirmed by right cardiac catheterization. Gastrointestinal involvement of SSc included any gastrointestinal symptoms (i.e., malabsorption, constipation, ileus, or pseudo-intestinal obstruction). Vitamin D insufficiency and deficiency was defined if 25(OH)D levels were < 30 and < 20 ng/ml, respectively. Statistical analysis. Demographic data are categorized and summarized using descriptive statistics. Categorical data were presented as proportions or percentages. Continuous data were presented as means with standard deviations (SD) or medians with interquartile ranges (IQR) as appropriate. The overall prevalence of osteoporosis, the prevalence of osteoporosis at the LS and FN with their 95% confidence interval (CI) were calculated. The odds ratio with 95%CI and p-value were used to determine the association between clinical characteristics and osteoporosis at the LS and FN. Continuous data were analyzed using the student t-test or Mann Whitney U-test as appropriate. The variables with a p value < 0.20 were entered into a multiple logistic regression model. The backward elimination method was applied for model fitting. Variables were tested for significance using the Wald X 2 statistic: all statistical tests were two-tailed. p values < 0.05 were considered to be statistically significant. Data analysis was performed using version 11.2 of STATA (StataCorp., College Station, TX, USA).
The study was designed by the authors and approved by the Human Research Ethics Committee of Khon Kaen University as per the Helsinki Declaration and the Good Clinical Practice Guidelines (HE551280). All eligible patients signed informed consent before enrollment. The sponsor had no role in the study.

Results
A total of 205 cases were recruited for the final analysis. There were 149 women and 56 men with a ratio of 2.7 to 1. The mean age and BMI at the time of study were 53.6 ± 10.6 years (range, 27-81) and 21.1 ± 4.0 kg/m 2 (6.7-31.4), respectively. The majority (62.9%) had a duration of disease < 5 years and most (83.4%) were the dcSSc subset. Approximately one-fifth of the cases (n = 44) had a modified Rodnan skin score (mRSS) of more than 20 points.
The overall median 25(OH)D level for all patients was 31.5 ng/ml (IQR 25.2-37.1). Vitamin D insufficiency and deficiency were 29.3% and 13.2%, respectively. Women had a significantly lower vitamin D level, and a higher rate of vitamin D insufficiency and deficiency than men (p < 0.001, p = 0.004, p = 0.006, respectively) ( Table 1).
The overall prevalence of osteoporosis was 29.3% (95%CI 23.1-36.0). The prevalence of osteoporosis at the LS in SSc patients was 28.3% (95% CI 22.2-35.0), and it was significantly higher in women than in men (32.2% vs. 19.4%) ( Table 1). While, the prevalence of osteoporosis at FN was lower (8.8%, 95% CI 5.3-13.5) and trend to be higher in women than in men without significant difference (11.2% vs. 3.2%, p = 0.06). The prevalence of osteoporosis at both sites (LS and FN) was 7.8% (95%CI 4.5-12.4) and there was no significantly difference between women and men (6.8% vs 1.0%, p = 0.16). In this study, the age-adjusted prevalence of osteoporosis in women and men was 26.3% and 10% at LS and 11.0% and 2.1% at FN. The clinical characteristics including the vitamin D level and BMD, classified by sex, were shown in Table 1.
In this study, low BMI (≤ 18.5 kg/m 2 ) and menopause were associated with osteoporosis at both LS and FN. However, other factors including SSc subset, severity of skin tightness, glucocorticoid used, vitamin D insufficiency or deficiency, smoking history, previous fracture and family history of fracture were not associated with osteoporosis at any sites (Tables 2 and 3 www.nature.com/scientificreports/ A further subgroup analysis in pre-menopausal women with low BMI was performed. The prevalence of osteoporosis at LS and FN was 13.9% (95%CI 3.2-24.7) and 2.3% (95%CI 0.3-7.0), respectively. We found that only BMI ≤ 18.5 kg/m 2 was associated with osteoporosis at LS but not FN. And there was no association between other clinical parameters and osteoporosis in pre-menopausal SSc women.

Discussion
In the present study, the overall prevalence of osteoporosis in our Thai SSc patients was 29.3%, which is higher than previous reports from Taiwan and China (17-19%) 12,18 and some Caucasian populations (3.3%) 14 but less than Brazilians and Moroccans reports (31.7-33%) 15,22 . The high prevalence of osteoporosis in Thais, Brazilians, and Moroccans might be explained by the higher proportion of the dcSSc subset compared to more prevalent lcSSc subset reported by Western countries. A previous study demonstrated a lower BMD in dcSSc than in lcSSc and intermediate SSc; probably because of extensive skin tightness and early internal organ involvement. The high disease activity might be related to limited daily activities which is a well-known risk factor for osteoporosis in the general population 14 . In contrast to our finding, the BMD at both LS and FN were comparable between dcSSc and lcSSc. Moreover, dcSSc subset was not associated with increased risk of osteoporosis in Thai SSc patients. This finding was due to a limited number of lcSSc in our study. The prevalence and associated factors for osteoporosis in SSc in the literature and our study are presented in Table 4.
We found that a low BMI (BMI ≤ 18.5 kg/m 2 ) is increased risk of osteoporosis in Thai SSc patients including pre-menopausal women and men. This finding is consistent as in general population 22 . Despite low BMI is a strong predictor of osteoporosis in our study, the exact mechanism for the development of osteoporosis has not been elucidated. Patients with SSc are prone to having low BMI due to decreased dietary intake, esophageal hypomotility, and intestinal malabsorption 15,16,23 . It has been accepted that low BMI is a symbolic clinical sign of malnutrition 23 . Around one-third of the SSc patients are at risk of malnutrition particularly in the patients with shorter duration of disease, more disease severity and having gastrointestinal involvement 24 . Moreover, lean mass and fat mass are decreased in SSc patients, which contributes to a low BMI 18,25 . Loss of lean mass in patients with SSc is related to physical inactivity due to joint contracture, arthritis, muscle wasting, and malnutrition, while low fat mass is associated with a low estrogen level 26 . Previous research demonstrated that low BMI is related to a low level of insulin-like growth factor (IGF-1) and low estrogen levels 27 ; both of which are well-known associated factors with osteoporosis. Therefore, the SSc patients who are at risk of malnutrition or low BMI even in premenopausal women should be screened for early identification, closed monitoring, awareness of the complication and giving the proper management for osteoporosis.
Our study did not demonstrate the associations among intestinal involvement, vitamin D deficiency, and osteoporosis; all of which may be the result of the malabsorption problem occasionally found in the dcSSc   www.nature.com/scientificreports/ subgroup of SSc. However, our unpublished data demonstrated that several physical limitations were associated with malnutrition status in Thai SSc (viz., daily activity, self-feeding capability, hand grip ability, painful digital ulcers, and narrow mouth aperture). Physical limitations and self-feeding capability might thus increase the risk of malnutrition, leading to osteoporosis in Thai SSc patients. The results from current study provides clinical clues that may help to evaluate osteoporosis in SSc patients. Further research is needed to determine whether improvement of nutritional status at disease onset can decrease the prevalence of osteoporosis in SSc patients. We did not found the associations between disease activities and osteoporosis risk in the present study including severe skin tightness, dcSSc subset, internal organ involvement, or inflammatory markers (ESR, CRP). The findings in our study are inconsistent with a case-control study which found that digital ulcers, presence of anti-centromere antibody, low lean mass and high number of previous fracture were an independent risks of low trabecular volumetric bone mineral density at tibia 28 . The authors also proposed the possible role of vasculopathy on the bone demineralization process 28 . The dissimilar findings with ours might be explained by the different method and different site of bone mineral density testing. According to the pathogenesis of SSc, the inflammatory process was not remarkable, thus the inflammatory marker was not associated with osteoporosis in our patients unlike other rheumatic diseases (i.e., rheumatoid arthritis) 29 .
Not only BMD assessment but also bone quality assessment has been evaluated in SSc. Poor bone quality evaluated by trabecular bone score (TBS) was reported among SSc patients. The prevalence of low TBS seemed to be more frequently found in SSc than in other connective tissue disease 30 . Moreover, there was a negative correlation between TBS and Dickkopf-1 serum levels which was a natural inhibitor of the Wnt signaling pathway promoting osteoclastogenesis 31 . According to the poor bone quality detection among SSc, therefore both of bone mass and bone quality should be cooperatively evaluated and further study of bone quality and the risk of fracture among SSc patients is suggested.
It has been accepted that treatment options for oral bisphosphonates may be limited due to esophageal disease, whereas treatment options for raloxifene and hormone therapy may be limited due to potential thrombotic complications. To avoid the gastrointestinal tract, intravenous zoledronate, denosumab, and teriparatide are better options for treating osteoporosis in SSc patients. However, the treatment option of osteoporosis was not www.nature.com/scientificreports/ our objective of the study, so we did not evaluate the treatment of osteoporosis in our SSc patients. The further research is needed to assess the outcome.
There are a number of limitations to the present findings. First, we did not perform tests for gastrointestinal malabsorption in severely malnourished SSc patients even though such a condition can be related to development of osteoporosis. Secondly, we did not check the hormonal status in pre-menopausal women even though low estrogen is associated with early development of osteoporosis. Third, we did not use a food intake questionnaire for evaluating nutritional status. Fouth, we did not exclude low BMI. However, this study was conducted in a single center with a large number of SSc patients. The current study included several important parameters, including vitamin D level, parathyroid hormone, thyroid and thyroid stimulating hormones, and inflammatory markers, which might associate with development of osteoporosis in SSc patients and also excluded the comorbid conditions that might confound BMD measurement (i.e., cancer undergoing chemotherapy, chronic kidney disease, hypothyroidism). Moreover, the findings from this study, including the prevalence and factors associated with osteoporosis in SSc patients, can be used for developing treatment guidelines to prevent osteoporosis.
In conclusion, the overall prevalence of osteoporosis in Thai patients with SSc was 29.3%, with women having a higher prevalence than men. Osteoporosis of the lumbar spine is more common than osteoporosis of the femoral neck in both gender. Low BMI (≤ 18.5 kg/m 2 ) and menopausal status are associated with osteoporosis at the lumbar spine and femoral neck.