Risk factor of osteoporosis in systemic lupus erythematosus patients: a cross-sectional study in a tertiary hospital in Indonesia

Abstract

Introduction

Osteoporosis and fragility fracture are more frequently found in systemic lupus erythematosus (SLE) patients than in normal population. This study aims to determine the risk factors in SLE patients in a tertiary hospital in Indonesia.

Methods

This is a cross-sectional study involving patients with SLE according to the 1997 American College of Rheumatology (ACR) criteria or 2019 ACR/EULAR classification criteria, from the rheumatology clinic in Dr. Soetomo General Academic Hospital. Subject characteristics data were extracted from medical records, and dual X-ray absorptiometry was performed on all the subjects to measure the lumbar spine and femoral neck bone mineral density (BMD), Z-score and T-score.

Result

Data from 152 subjects were collected. 27% of subjects had osteoporosis and the mean age was 34.63±10.42-year old. Lower body mass index (BMI) (B 0.074; p 0.004), higher disease activity (B 0.054; p 0.04), corticosteroid use (B −0.018; p 0.036) and older age (B −0.027; p 0.003) were significantly associated with lower T-score. In particular, higher age was associated with lower femoral neck BMD, while menopause was correlated to lower lumbar spine and total hip BMD. The use of azathioprine (AZA) (OR 0.225; CI 0.051 to 0.990; p 0.048) and hydroxychloroquine (HCQ) (OR 0.293; CI 0.115 to 0.746; p 0.01) was associated with a lower incidence of osteoporosis.

Conclusion

The prevalence of osteoporosis in the SLE population was 26.97%. Ageing, menopause, lower BMI, higher current disease activity and higher cumulative dose of corticosteroids were associated with low BMD in our study. Risk factor for osteoporosis in our study was lower BMI, menopause and the use of high dose steroids. The use of AZA and HCQ appeared to be protective for osteoporosis.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Osteoporosis and low bone mass are more prevalent in systemic lupus erythematosus (SLE) patients than in the general population.

• Traditional risk factors for osteoporosis in SLE patients include cumulative glucocorticoid dose, low body mass index (BMI), age and menopause. Several studies showed that disease activity markers such as erythrocyte sedimentation rate and complement levels are associated with osteoporosis.

WHAT THIS STUDY ADDS

• This is the first study to elaborate on the risk factors of osteoporosis and low bone mass in SLE population in Indonesia.

• In our study, increasing age, menopause, higher disease activity, lower BMI and higher cumulative dose of corticosteroids were associated with low bone mineral density (BMD). Other than that, treatment with azathioprine and hydroxychloroquine (HCQ) was negatively correlated with osteoporosis.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• This study highlights the importance of BMD measurement and monitoring in SLE patients undergoing treatment with glucocorticoids.

• This study highlights the potential role of HCQ and azathioprine for prevention or mitigation of osteoporosis in SLE patients.

Introduction

Systemic lupus erythematosus (SLE) is a chronic multisystem autoimmune disease characterised by a relapsing and remitting course, leading to prolonged inflammation and tissue/organ damage.^{1 2} The incidence of SLE has been increasing, and the improvements in survival rates have resulted in an increased prevalence of complications and comorbid conditions. Among these, osteoporosis, a common clinical manifestation in chronic inflammatory diseases, has become increasingly prevalent in patients with SLE.³

Since the 1990s, the prevalence of osteoporosis in SLE patients has been reported to vary from 1.4% to 68%. A large population-based study involving 7732 SLE patients and 28 079 age-matched and sex-matched controls found that the incidence of osteoporosis was 2.53 times higher in SLE patients. Additionally, the incidence of symptomatic fractures was elevated, ranging from 1.2 to 4.7 times higher compared with the control group.^{4 5} Another study from Korea reported a 2.964-fold increased risk of osteoporotic fracture in SLE patients compared with age-matched and sex-matched non-SLE controls. Notably, male or middle-aged SLE patients exhibited a relatively higher fracture risk.⁶ In cross-sectional studies, the prevalence of osteopenia ranges from 11% to 62% in the lumbar spine (LS) and from 6% to 74% in the hip. In comparison, the prevalence of osteoporosis ranges from 4% to 42% in the LS and from 3% to 42% in the hip. These wide variations across studies are likely attributable to differences in ethnicity, age, sex, study design, disease severity, glucocorticoid use and disease duration.⁷

SLE patients are at risk of low bone mass due to prolonged inflammation, glucocorticoid use, vitamin D deficiency, premature ovarian failure, increased damage and traditional risk factors such as age, gender, body mass index (BMI), reduced physical activity, menopausal status and ethnicity.38,10 SLE patients with osteoporosis have high morbidity and mortality due to fragility fractures, especially when hip fractures occur. These fractures are not only clinically significant but also associated with substantial healthcare costs, contributing to a considerable economic burden. Therefore, osteoporosis and fragility fractures in the SLE population represent serious complications that warrant focused attention and proactive management.^{4 7} Limited data are available on the incidence of osteoporosis among Indonesian patients with SLE. Therefore, this study aimed to evaluate the risk factors associated with osteoporosis in SLE patients treated at Dr. Soetomo General Academic Hospital, Surabaya.

Methods

We conducted a review of medical records from patients with SLE who attended the Rheumatology Outpatient Clinic at Dr. Soetomo General Academic Hospital, a tertiary referral and teaching hospital located in Surabaya, Indonesia, from August 2023 to August 2024. Inclusion criteria included adults ≥18-year old who were diagnosed with SLE, based on either the 1997 American College of Rheumatology (ACR) diagnostic criteria or the 2019 ACR/European League Against Rheumatism classification criteria.^{11 12} Exclusion criteria were incomplete data from medical records. All patients. Samples were acquired using the total sampling method. The minimum number of samples required for linear regression with power (1-β) of 0.8 and α of 0.05 was calculated using G*power,¹³ which was 98.

Medical records of SLE patients were reviewed, and data were extracted on the following parameters: age at diagnosis and at the time of dual-energy X-ray absorptiometry (DXA), sex, BMI at the time of DXA, menopausal status, disease duration, current and previous clinical manifestations (including mucocutaneous, musculoskeletal, renal, central/peripheral nervous system, serositis, cardiorespiratory and haematologic involvement), initial disease activity assessed using the Systemic Lupus Erythematosus Disease Activity Index (SLEDAI), pre-existing comorbidities, glucocorticoid use (daily and cumulative dose since diagnosed with SLE), both current and previous immunosuppressant use (including hydroxychloroquine (HCQ), methotrexate, azathioprine (AZA), calcineurin inhibitors (CNI which include cyclosporine, tacrolimus and voclosporine), mycophenolic acid analogue (which include mycophenolate mofetil and mycophenolic acid) and cyclophosphamide. We categorised high corticosteroid use as ≥30 mg/day of prednisone for more than a month.¹⁴

The subjects were then contacted to undergo DXA examination to measure their LS and femoral neck (FN) bone mineral density (BMD). A DXA examination was performed using GE-Lunar Prodigy DXA at the diagnostic centre of Dr. Soetomo General Academic Hospital. Collected data included BMD of the LS and FN sites, along with their corresponding T-scores.

The T-score obtained from a BMD test compares an individual’s bone density to the average peak bone density of a healthy young adult of the same sex, whereas the Z-score reflects the difference between an individual’s BMD and the average BMD of healthy individuals matched for age, sex and ethnicity.^{15 16} We classified osteoporosis as defined by WHO with a BMD T-score of −2.5 or lower, measured by DXA.¹⁵ Most patients at Dr. Soetomo General Academic Hospital were of Indonesian Javanese ethnicity. However, the Z-score database in the DXA machine was not tailored to this population, as it relied on normative reference values provided by the manufacturer. Therefore, in this study, we primarily used direct BMD values, T-scores and the WHO classification of osteoporosis to categorise patients’ BMD results.

Univariate statistical analyses were performed to assess associations between patient variables and T-score of the spine, FN and the lowest T-score among spine and FN. An independent two-sample t-test was used to compare T-scores for dichotomous variables. For continuous variables, correlation analysis was conducted to examine the strength of association with T-scores.

Multivariate linear regression analysis was performed using a backward variable selection method to identify variables associated with changes in T-scores, while adjusting for potential confounders. Logistic regression was also performed to identify independent risk factors for osteoporosis. Partial correlation coefficients, reflecting the adjusted strength of association between T-scores and other variables, are also reported. A p value of ≤0.05 was considered statistically significant.

Results

Subjects’ characteristics

There were 213 patients fulfilling the inclusion criteria. Sixty-one patients were excluded due to incomplete characteristic data. 152 SLE patients were included in the study. Basic patient characteristics are presented in table 1. Of these, three patients (2%) were male, and the mean age was 34.63±10.42 years. Twenty patients (13.2%) were postmenopausal, and only one patient (0.6%) had experienced an osteoporotic fracture (hip fracture). The average BMD of the LS and FN was 1.02±0.17 g/cm² and 0.83±0.17 g/cm², respectively. Based on their lowest T-score, 41 patients (26.97%) were classified as having osteoporosis, while 76 patients (50%) were categorised as having osteopenia.

Table 1. Subject characteristics included in this study.

Variable	Mean±SD or N (%)
Sex
Women	149 (98)
Men	3 (2)
Age	34.63±10.42
BMI	22.38±3.48
Postmenopause	20 (13.2)
Duration of Illness (years)	4.99±3.43
Initial SLEDAI	8.43±3.09
Current SLEDAI	2.51±3.367
Total steroid consumed (grams of methyl prednisolone)	10.68±10.41
Comorbidities
Diabetes mellitus	3 (2)
Autoimmune thyroid disease	7 (4.6)
Atopy	2 (1.3)
Other systemic autoimmune diseases	26 (17.1)
Tuberculosis	3 (2)
Chronic viral hepatitis	3 (2)
Manifestation
Lupus nephritis	49 (32.3)
NPSLE	9 (5.9)
Cutaneous lupus	36 (23.7)
Haematologic manifestation	12 (7.9)
Musculoskeletal manifestation	64 (42.1)
Treatment
CYC	20 (13.2)
MPAA	65 (42.8)
CNI	42 (27.6)
AZA	33 (21.7)
MTX	27 (17.8)
HCQ	95 (62.5)

AZA, azathioprine; BMI, body mass index; CNI, calcineurin inhibitor; CYC, cyclophosphamide; HCQ, hydroxychloroquine; MPAA, mycophenolic acid analogue; MTX, methotrexate; NPSLE, neuropsychiatric SLE; SLEDAI, SLE Disease Activity Index.

Proportion of subjects with osteopenia and osteoporosis based on WHO criteria for BMD

The proportion of osteoporosis and osteopenia in the subjects at any BMD measurement site is presented in table 2. The proportion of subjects with osteoporosis among postmenopausal women was significantly higher than that among premenopausal women (55% vs 22.7%, p=0.032).

Table 2. The proportion of subjects with osteopenia and osteoporosis based on the WHO Classification for BMD.

	Osteopenia (T score between −1 and −2.5)				Osteoporosis (T score ≤−2.5)
	Total (%)	FN (%)	LS (%)	TH (%)	Total (%)	FN (%)	LS (%)	TH (%)
Premenopause	67 (50.8)	68 (51.5)	63 (47.7)	44 (33.3)	31 (23.5)	23 (17.4)	18 (13.7)	8 (6.1)
Postmenopause	8 (40)	9 (45)	7 (36.8)	14 (70)	10 (50)	7 (35)	7 (36.8)	1 (5)
Total	75 (49.3)	77 (50.7)	70 (46.1)	58 (38.2)	41 (27)	30 (19.7)	25 (16.4)	9 (5.9)

BMD, bone mineral density; FN, femoral neck; LS, lumbar spine (L1–L4); TH, total hip.

The average BMD measurements across various skeletal sites are presented in table 3. The mean LS BMD values for subjects classified as osteoporotic, osteopenic and normal according to WHO classification were 0.89±0.19, 1.04±0.11 and 1.13±0.13 g/cm², respectively. The corresponding mean FN BMD values were 0.69±0.11, 0.82±0.12 and 1.01±0.16 g/cm², respectively. For total hip (TH), the values were 0.97±0.12, 0.85±0.09 and 0.69±0.10, respectively. BMD values at LS, TH and FN sites were higher in premenopausal subjects compared with postmenopausal subjects, although statistical significance was observed only for FN BMD (0.830±0.138 vs 0.787±0.14 g/cm², p=0.113 for TH BMD; 1.022±0.145 vs 0.991±0.270 g/cm², p=0.105 for LS BMD; 0.827±0.170 vs 0.771±0.134 g/cm², p=0.03 for FN BMD).

Table 3. Average BMD and T-score of study subjects.

	BMD (g/cm2)		TH	T-score		TH
	L1–L4	FN		L1–L4	FN
Premenopause	1.022±0.145	0.827±0.170	0.840±0.138	−1.204±1.263	−1.413±1.173	−0.854±1.122
Postmenopause	0.991±0.270	0.771±0.134	0.787±0.14	−1.980±1.364	−1.937±0.994	−1.395±1.174
Total	1.02±0.168	0.826±0.166	0.833±0.139	−1.310±1.299	−1.486±1.160	−0.928±1.14

BMD, bone mineral density; FN, femoral neck; TH, total hip.

Out of the three male subjects, 2 (66.67%) were found to have osteoporosis and the other one (33.33%) had osteopenia, according to the WHO classification. The mean LS BMD of the men (0.895±0.041 g/cm²) was lower compared with the women (1.021±0.168 g/cm²) but not statistically significant (p 0.085). There was no significant difference between the FN BMD of the men (0.818±0.085 g/cm2) and women (0.826±0.167 g/cm²) from this study (p 0.984).

Factors associated with BMD change

Bivariate analysis was conducted to examine the relationship between clinical characteristics and BMD measured by DXA at FN, TH and LS as shown in table 4. Age was significantly associated with lower FN BMD and overall T-score, but not with LS or TH BMD. Cumulative steroid dose showed a negative correlation with both FN and LS BMD, as well as with the overall T-score, but not TH BMD. In contrast, a history of high-dose steroid use was significantly associated only with LS BMD. Additionally, menopause was found to be associated with lower overall T-scores and FN BMD, but not with LS nor TH BMD. While initial disease activity did not show any correlation with BMD, the SLEDAI index during the study period showed correlation with overall T-score, TH and FN BMD, but not LS BMD.

Table 4. Association between BMD and T-score and clinical characteristics.

Variables	Overall T-score	FN BMD	LS BMD	TH BMD
BMI	ρ 0.128 p 0.125	ρ 0.116 p 0.159	ρ 0.091 p0.27	ρ 0.115 p 0.057
Age	ρ −0.221 p 0.008	ρ −0.185 p 0.023	ρ −0.048 p0.558	ρ −0.128 p 0.115
Menopause	p 0.00	p 0.03	p 0.105	p 0.113
Yes	−2.51±0.97	0.77±0.13	0.99±0.27	0.787±0.14
No	−1.72±1.03	0.83±0.17	1.02±0.14	0.84±0.138
Duration of Illness	ρ −0.89 p 0.286	ρ −0.135 p 0.098	ρ −0.115 p 0.167	ρ −0.095 p 0.243
Initial SLEDAI	ρ 0.064 p 0.443	ρ 0.009 p 0.917	ρ −0.049 p 0.551	ρ 0.039 p 0.638
Current SLEDAI	ρ 0.184 p 0.027	ρ 0.186 p 0.024	ρ 0.056 p 0.499	ρ 0.179 p 0.029
Steroid cumulative dose	ρ −0.167 p 0.04	ρ −0.186 p 0.023	ρ −0.193 p 0.018	ρ −0.148 p 0.07
History of high-dose steroid (≥30 mg/day for more than 30 days)	p 0.316	p 0.367	p 0.043	p 0.579
Yes	−1.97±1.27	0.81±0.19	0.99±0.16	0.825±0.159
No	−1.78±0.93	0.83±0.14	1.04±0.17	0.838±0.123
CYC	p 0.508	p 0.993	p 0.907	p 0.344
Yes	−1.85±1.05	0.79±0.21	1.01±0.153	0.860±0.147
No	−1.68±1.16	0.83±0.16	1.02±0.171	0.829±0.138
MPAA	p 0.76	p 0.211	p 0.223	p 0.39
Yes	−1.82±1.08	0.80±0.16	0.99±0.14	0.821±0.137
No	−1.86±1.04	0.84±0.16	1.04±0.18	0.841±0.141
CNI	p 0.084	p 0.143	p 0.374	p 0.017
Yes	−2.06±1.06	0.794±0.143	1.00±0.21	0.789±0.132
No	−1.77±1.05	0.84±0.173	1.03±0.15	0.849±0.139
AZA	p 0.29	p 0.12	p 0.39	p 0.015
Yes	−1.70±1.09	0.87±0.14	1.04±0.17	0.885±0.139
No	−1.88±1.05	0.81±0.17	1.01±0.17	0.818±0.137
MTX	p 0.09	p 0.112	p 0.101	p 0.219
Yes	−0.154±0.90	0.87±0.18	1.07±0.15	0.863±0.112
No	−1.89±1.08	0.82±0.16	1.01±0.17	0.826±0.144
HCQ	p 0.158	p 0.054	p 0.575	p 0.181
Yes	−1.76±1.03	0.84±0.156	1.02±0.14	0.844±0.136
No	−1.98±1.10	0.81±0.18	1.02±0.20	0.813±0.144
Comorbidities
Diabetes mellitus	p 0.70	p 0.36	p 0.79	p 0.837
	−1.80±1.69	0.88±0.106	1,02±0.28	0.816±0.161
	−1.89±1.12	0.82±0.17	1.01±0.17	0.833±0.14
Autoimmune thyroid disease	p 0.99	p 0.402	p 0.846	p 0.989
	−1.86±2.44	0.79±0.32	1.03±0.30	0.831±0.224
	−1.89±1.05	0.82±0.16	1.01±0.16	0.832±0.136
Other autoimmune diseases	p 0.21	p 0.28	p 0.667	p 0.233
	−2.16±0.92	0.79±0.18	0.99±0.14	0.802±0.123
	−1.82±1.15	0.82±0.16	1.02±0.17	0.838±0.142
TB infection	p 0.91	p 0.782	p 0.849	p 0.557
	−1.77±1.04	0.86±0.16	1.02±0.07	0.785±0.206
	−1.89±1.12	0.82±0.17	1.01±0.17	0.833±0.139
Chronic viral hepatitis	p 0.10	p 0.737	p 0.235	p 0.674
	−2.90±0.79	0.80±0.14	0.89±0.18	0.795±0.09
	−1.86±1.12	0.82±0.17	1.01±0.17	0.829±0.141

Statistically significant values are highlighted in bold.

AZA, azathioprine; BMD, bone mineral density; BMI, body mass index; CNI, calcineurin inhibitor; CYC, cyclophosphamide; FN, femoral neck; HCQ, hydroxychloroquine; LS, lumbar spine (L1–L4); MPAA, mycophenolic acid analogue; MTX, methotrexate; NPSLE, neuropsychiatric SLE; SLEDAI, SLE disease activity index; TH, total hip.

Multivariate linear regression analysis was performed using variables that showed significant associations in the bivariate analysis (table 5). The adjusted R² for the model was 0.119 (F=5.792; p<0.001). Age (β=−0.326, p=0.003) and total cumulative steroid dose (β=−0.206, p=0.032) were significantly associated with lower overall BMD T-scores. In contrast, BMI (β=0.194, p=0.004) and current SLEDAI (β=0.166, p=0.044) were positively associated with higher BMD T-scores.

Table 5. Multivariate linear regression.

Dependent variables	Variables	Coefficients	P value	95% CI for coefficients
Lowest T-score	Age	−0.027	0.003	−0.044 to −0.009
	BMI	0.074	0.004	0.024 to 0.124
	Current SLEDAI	0.054	0.04	0.001 to 0.106
	Total cumulative dose of steroids	−0.018	0.036	−0.034 to −0.001
BMD FN	Age	−0.03	0.012	−0.06 to −0.01
	Total cumulative dose of steroids	−0.003	0.033	−0.05 to −0.001
	BMI	0.097	<0.001	0.044 to 0.151
BMD L1–L4	BMI	0.081	0.009	0.021 to 0.142
	Menopause	−0.869	0.006	−1.485 to −0.253
	Total cumulative dose of steroid	−0.019	0.057	−0.05 to 0.001
BMD TH	BMI	0.010	0.001	0.004 to 0.017
	Menopause	−0.067	0.028	−0.130 to −0.008
	Total cumulative dose of steroid	−0.003	0.006	−0.005 to −0.001
	Use of azathioprine	0.067	0.012	0.005 to 0.113
	Current SLEDAI	0.005	0.099	−0.001 to 0.012

BMD, bone mineral density; BMI, body mass index; FN, femoral neck; LS, lumbar spine (L1–L4); SLEDAI, SLE disease activity index; TH, total hip.

Bivariate analysis of demographic variables and osteoporosis classification revealed that subjects classified as having osteoporosis, compared with those without, had significantly lower BMI (21.34±3.95 vs 22.61±3.58, respectively, p 0.043) and a higher proportion of postmenopausal status (26.83% vs 8.11%, respectively, p 0.02). Among the medications administered, a history of high-dose corticosteroid use (> 30 mg/day of prednisone equivalent for more than 30 days) was more prevalent in osteoporotic subjects compared with those without (56.1% vs 39.6% p 0.008). In contrast, current use of AZA and HCQ was more prevalent in subjects without osteoporosis compared with those with osteoporosis (27% vs 7.32% p 0.009 for AZA and 67.6% vs 48.8% p 0.034 for HCQ).

Logistic regression analysis performed using variables that showed significant associations in the bivariate analysis indicated that menopause (OR=9.01, 95% CI 2.62 to 30.98, p<0.01) and a history of high-dose corticosteroid use (OR=5.93, 95% CI 2.21 to 15.92, p 0.01) were independently associated with osteoporosis. In contrast, higher BMI (OR 0.853 CI 0.74 to 0.992 p 0.034) and current use of AZA (OR 0.225 CI 0.051 to 0.990, p=0.048) and HCQ (OR 0.293 CI 0.115 to 0.746 p=0.01) were found to be protective against osteoporosis. Further details on the logistic regression done in this study may be found in online supplemental materials.

Discussion

We reported a study highlighting the risk factors of low BMD and osteoporosis in a tertiary hospital in Indonesia. To our knowledge, this is the first study to describe the risk factors for osteoporosis in Indonesian SLE patients and compare it to previous studies from around the world. In this study, we found that the proportion of subjects with osteopenia (T-score between −2.5 and −1) and osteoporosis (T-score ≤−2.5) was 49.3% and 27%, respectively. Overall, 68% (104 out of 152) of the subjects had either osteoporosis or low BMD for age. Previous studies have shown that low bone mass and osteoporosis are more prevalent among patients with SLE compared with the general population. A retrospective study conducted in Saudi Arabia reported that 53.5% of SLE patients had low BMD, defined as either osteoporosis or low bone mass for age.¹⁷ In that study, the prevalence of low bone mass for age and osteoporosis was 29.9% and 13.2%, respectively.¹⁷ The proportion of subjects with osteoporosis in this study was higher than that reported in the Saudi Arabian study and more comparable to findings from a study conducted in Finland by Almehed et al.¹⁸ This difference may be explained by the varying definitions of osteoporosis used; for example, the Saudi study applied the T-score threshold of <−2.5 strictly to postmenopausal women. The proportion of subjects with osteoporosis in our study was lower than that reported in studies from Korea (32.9%) and China (40%),^{19 20} although the mean age of our study population (34.63±10.42 years) was considerably lower than those in the Korean (53.7±6.8 years) and Chinese (46.2±12.9 years) studies.^{19 20}

In this study, we found that the proportion of osteoporosis was higher at the FN than at the LS and TH BMD (18.7% vs 16.7% vs 5.9%). Several studies have reported that in SLE or glucocorticoid-induced osteoporosis, the LS is more commonly affected than the FN.^{17 20 21} We suspect that the differing pattern observed in this study may be due to the younger mean age of our subjects compared with those in previous studies. Research involving younger and juvenile SLE populations has shown that FN BMD tends to be lower than LS BMD.²² This may reflect incomplete attainment of peak bone mass and skeletal architecture in younger individuals. For this reason, Z-scores are preferred over T-scores when evaluating bone density in younger patients.^{15 23}

In this study, the risk factors associated with lower T-scores and BMD included menopause, older age, higher disease activity, lower BMI and higher cumulative steroid dose. Numerous studies have demonstrated the effect of BMI on BMD and the prevalence of osteoporosis.^{24 25} Higher body mass directly increases loading on bones, affecting bone remodelling patterns. Additionally, higher body mass is associated with higher oestrogen and testosterone, which in turn contribute to higher bone density.²⁶

Steroids are also a well-established risk factor for reduced bone mass. They increase the expression of RANK-L, the activity of osteoclasts and decrease osteoblast survival.²⁷ Even a low dose of 2.5 mg of prednisone has been associated with an increased risk of vertebral fractures.²⁸ Consistent with our findings, previous studies have demonstrated a dose-dependent effect of steroids on BMD.^{29 30}

In this study, overall T-score was associated with SLEDAI score during the period of the study but not with the initial SLEDAI. A number of studies also showed similar association,^{3 31 32} while some showed no correlation between these two variables.³³ The effect of disease activity on BMD is complex and may depend on the degree of inflammation, frequency and duration of ‘flare up’ episode, organ damage and the accumulation of steroid dose used.³ Further research should include these factors to elucidate the relationship between disease activity and osteoporosis.

In this study, menopause was identified as an independent risk factor for lower LS and TH BMD, but not for FN BMD. In contrast, increasing age was an independent risk factor for lower FN BMD but not LS or TH BMD. Menopause is a well-established risk factor for reduced BMD in both SLE and non-SLE populations, primarily due to the decrease of sex hormones, including oestrogen.^{23 34} During menopause, both trabecular and cortical bone experience a reduction in bone mass, with the more metabolically active trabecular bone being affected to a greater extent than cortical bone.^{35 36} Ageing is associated with a relatively balanced loss of both cortical and trabecular bone; however, in very advanced age, cortical bone loss tends to become more predominant.^{34 36 37} The lumbar vertebral body contains a higher proportion of trabecular bone and a thinner cortical shell. In contrast, the FN has a higher proportion of cortical bone compared with the lumbar vertebrae.^{37 38} TH BMD is the weighted average BMD of FN, trochanter and intertrochanter regions, with the latter two containing higher proportion of trabecular bone. One previous study also showed that trochanteric BMD starts showing age-related bone loss only after menopause.³⁹ We postulate that the discrepancy in the effect of age and menopause on FN, TH and LS BMD in this study may be attributed to differences in the underlying pathomechanisms of bone loss.

In terms of treatment, we also found that the use of HCQ and AZA was associated with a lower incidence of osteoporosis. Several in vitro studies showed that HCQ could inhibit osteoclast activity by interfering in its lysosome activity.^{40 41} One animal study showed that concentrating HCQ to the bone by conjugation to hydroxyalkylil bisphosphonate (HABP) (a bisphosphonate with minimal antiresorptive activity) increased osteoblast activity compared with either HABP or HCQ alone.⁴² HCQ-HABP conjugate also had osteoclastic inhibition effect although this was only achieved at very high HCQ concentration. These effects seem to be mediated by HCQ preventing the degradation of TRAF3, a signalling molecule that inhibits osteoclast formation and promotes osteoblast survival.⁴² The study also gave insight that for the bone protection effect of HCQ to take place, high doses of HCQ may be required and thus increasing the risk of adverse effects. Therefore, it may be beneficial if HCQ could be concentrated to the bone, such as through the conjugation with bisphosphonate in order to minimise the systemic side effects.⁴² In humans, five studies demonstrated that HCQ use is associated with improvement in bone mass.43,46 Another study found that HCQ use is linked to a reduction in urinary C-terminal peptide.⁴¹ Two studies in that review reported that HCQ use is associated with lower BMD,^{47 48} while one study found no association between HCQ use and BMD.⁴⁹ Overall, the evidence appears to support a beneficial role of HCQ in bone health by reducing inflammation, modulating the RANK/RANKL/OPG pathway and influencing the autophagic activity of osteoclasts.⁴³

The role of AZA in modulating BMD remains controversial. Animal studies have reported that AZA is associated with reduced BMD.⁵⁰ Other studies, however, found no significant effect of AZA on bone density, although its steroid-sparing properties may offer a protective benefit.^{51 52} In this study, AZA was prescribed for non-severe SLE cases without renal or neurological involvement. Although disease activity was not significantly associated with osteoporosis, it showed a positive correlation with high-dose steroid use. This relationship may help explain the observed effect of AZA in our findings.

In terms of comorbidities, we found that diabetes mellitus, tuberculosis infection, other autoimmune disease and chronic hepatitis were not associated with lower BMD. These comorbidities were known to cause chronic inflammation, which may worsen BMD loss in SLE patients.53,55 However, the number of the comorbidities was low (3 subjects (2%) for diabetes mellitus, tuberculosis infection and chronic hepatitis; 2 subjects (1.3%) for atopy; 26 subjects (17.1%)) and may explain the negative findings of this study.

Several limitations were that the cross-sectional design of this study could not fully establish causality between the identified risk factors and osteoporosis. Furthermore, the effect of osteoporotic treatment on the BMD of our subjects could not be assessed since the majority of the subjects were newly diagnosed with osteoporosis/osteopenia and had just started treatment during the period of our study. Furthermore, while our study captured one subject with hip fracture, we did not screen the subjects for asymptomatic lumbar fracture as an outcome of this study. Future prospective studies are necessary to evaluate the long-term impact and to monitor the dynamic changes in BMD over time as well as the incidence of morphometric fracture in this population.

Finally, despite our effort to include all of the SLE patients currently treated in our hospital, there is a significant number of patients excluded due to incomplete data (61 out of 213 or 28.6%), which limits the ability of this study to predict the prevalence of osteoporosis in the general population. Further studies with larger sample sizes are needed in order to find osteoporosis risk factors unique to SLE.

Conclusion

In this study, we found that the key risk factors for reduced BMD included lower BMI, higher current disease activity, corticosteroid use, older age and menopause. Age was significantly associated with FN BMD, while menopause was linked to lower LS BMD. The use of AZA and HCQ appeared to be associated with a lower incidence of osteoporosis. These findings underscore the importance of regular BMD monitoring in SLE patients to help prevent fragility fractures and guide appropriate management strategies.

Data availability statement

Data are available upon reasonable request.