Article Text

Concise report
Low bone mineral density is present in newly diagnosed paediatric systemic lupus erythematosus patients
  1. Siok Hoon Lily Lim1,
  2. Susanne M Benseler1,
  3. Pascal N Tyrrell1,
  4. Martin Charron1,
  5. Elizabeth Harvey2,
  6. Diane Hebert2,
  7. Earl D Silverman1,3
  1. 1Department of Pediatrics, Division of Rheumatology, The Hospital for Sick Children, University of Toronto, Toronto, Ontario, Canada
  2. 2Nuclear Medicine, Department of Diagnostic Imaging, The Hospital for Sick Children, University of Toronto, Toronto, Ontario, Canada
  3. 3Departments of Paediatrics and Immunology, The Hospital for Sick Children, University of Toronto, Toronto, Ontario, Canada
  1. Correspondence to Professor Earl D Silverman, Ho Family Chair in Autoimmune Diseases, Department of Paediatrics, Division of Rheumatology, The Hospital for Sick Children, 555 University Avenue, Toronto, ON M5G 1X8, Canada; earl.silverman{at}


Objectives The objectives of this study were to (1) determine the prevalence of low bone mineral density (BMD) in a large prospective cohort of newly diagnosed patients with paediatric systemic lupus erythematosus (pSLE) and (2) identify risk factors associated with low BMD.

Methods Single-centre cohort study of 80 children and adolescents who underwent a dual-energy x-ray absorptiometry within 3 months of diagnosis. Low lumbar spine (LS) BMD was defined as z score ≤−2.0. BMD was correlated with baseline demographic, clinical and laboratory markers of disease activity and biochemical markers of bone health. Risk factors of BMD were evaluated with univariable and multivariable linear and logistic regression analyses.

Results Low BMD at any site was found in 15% of newly diagnosed pSLE patients. LS BMD was associated with body mass index (BMI) z score and corrected calcium (r2=0.31, p<0.0001). Hip BMD was associated with BMI z score and intact parathyroid hormone (iPTH) (r2=0.26, p=0.002). Higher BMI z score was protective against low BMD at any site (OR 0.35).

Conclusions One in six newly diagnosed pSLE patients had low BMD (at any site). Low BMI z score, low calcium and high iPTH identified children at risk for low BMD at diagnosis of pSLE.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Systemic lupus erythematosus (SLE) patients have many risk factors for osteoporosis. Inflammation during active disease mediated through cytokines, for example, tumour necrosis factor α and receptor agonist of nuclear factor κ ligand, promotes osteoclastogenesis.1 Inflammation-associated metabolic derangements such as oxidised low-density lipoprotein augment osteoclastogenesis and suppress osteoblastogenesis.2 Corticosteroids, the mainstay of treatment in active SLE, have multiple detrimental effects: inhibit osteoblastogenesis, reduce osteoblast function, increase apoptosis of osteoblasts and stimulate osteoclastogenesis.3

Childhood and adolescence are crucial periods of bone mass accrual. More than 90% of the peak bone mass should be accrued by the end of adolescence.4 Patients with paediatric SLE (pSLE) are at risk of missing this critical period of bone accrual due to disease activity, treatment, decreased physical activity and vitamin D insufficiency. Paediatric studies on bone mineral density (BMD) with rheumatic diseases demonstrated conflicting results.5,,8 The confusion in the pSLE literature was partly due to inclusion of patients with varying disease durations and use of varying definitions of ‘abnormal’ BMD.9,,13 Furthermore, the relative importance of disease activity and treatment was hard to determine as the studies were cross-sectional.

We undertook this study to (1) determine the prevalence of low BMD in a large prospective cohort of newly diagnosed patients with pSLE and (2) identify risk factors associated with low BMD.


Patients and study design

This was a retrospective inception cohort study of newly diagnosed pSLE patients seen at the SickKids Lupus Clinic between December 2001 and December 2007. We included patients if they were 5–18 years old and underwent a dual-energy x-ray absorptiometry (DEXA) scan within 3 months from diagnosis. Eighty of the 189 newly diagnosed patients seen over the study period met entry criteria. Participants had higher disease activity at diagnosis compared with non-participants (mean SLE Disease Activity Index (SLEDAI) 11 vs 8) but did not differ in other demographic characteristics. Data were abstracted from a clinical research database (see online supplement). All patients received 800 IU/day of vitamin D supplementation and individualised calcium supplementation to meet a target of 1000 mg/day of calcium intake.

Bone health assessments

All DEXA scans were performed with a Lunar Prodigy machine (GE Lunar, Madison, Wisconsin, USA). Areal BMD was computed for the lumbar spine (LS) and the hip. LS z scores and percentage hip BMD were derived using age and gender norms supplied by the manufacturer.

Low LS BMD (z≤−2.0) was defined according to definitions endorsed by the International Society of Clinical Densitometry (ISCD) and the American Academy of Pediatrics (AAP).14 15 Hip BMD<80% was defined as low.10 Low LS BMD±low hip BMD was defined as low BMD for this study. We also reported an additional class of ‘low-normal’ LS BMD (not recognised by ISCD): −2.0<z score≤−1.0. A composite of low-normal to low LS BMD (z≤−1.0)±low hip BMD was termed ‘abnormal’ BMD.

A panel of markers of bone physiology was evaluated concomitantly with DEXA. The impact of corticosteroid exposure was also assessed (see online supplement).

Statistical analysis

Positively skewed variables were log transformed. Potential variables underwent a univariable regression screen (p<0.1) and significant variables were used to develop multivariable regression models. Linear regression was performed for BMD (LS/hip) and logistic regression for low BMD. As there were significant missing data, we first performed a complete data analysis to select a possible candidate model. The model was verified subsequently by repeating the modelling procedure following multiple imputations with a Markov Chain Monte Carlo algorithm. Ten datasets were imputed with 97% efficiency. All analyses were performed using SAS V.9.1 (North Carolina, USA).



Table 1 shows baseline demographics, clinical characteristics, treatment and laboratory assessments of participants.

Table 1

Baseline demographics, clinical characteristics, laboratory results and treatment of newly diagnosed paediatric systemic lupus erythematosus patients up to the time of the dual-energy x-ray absorptiometry study

Bone health assessments

Forty-six per cent of patients underwent DEXA within the first month and 20% between the first and second months of diagnosis of SLE. Twenty-two patients were steroid naive at the time of DEXA.

Twelve patients (15%) had low BMD (low LS BMD±low hip BMD). Twenty-six patients (33%) had abnormal BMD (LS BMD≤−1.0±low hip BMD). The distribution of LS BMD was significantly different from zero (p=0.01). See table 2 for results of BMD and markers of bone physiology.

Table 2

Bone mineral density (BMD) and markers of bone physiology and turnover

There was no difference in LS (p=0.63) or hip (p=0.89) BMD irrespective of exposure to steroids. Similarly, LS (p=0.14) and hip (p=0.15) BMD did not differ by steroid exposures grouped by timing of DEXA (see figure 1 and online supplement). Steroid-exposed patients were not more likely to have low (LS/hip) BMD (p=0.23).

Figure 1

Lumbar spine (LS) bone mineral density (BMD) z scores and the time to dual-energy x-ray absorptiometry (DEXA) scan from diagnosis. The x-axis denotes the time of the DEXA scan in each individual patient from the time of diagnosis in months. The y-axis denotes the results of each patient's LS BMD expressed as z scores. Dashed black lines at +2 and −2 denote the two SDs from a z score of 0 (limits of ‘normality’). Grey line at 0 denotes the z score of 0. Black diamonds (♦) represent steroid naive patients and open squares (□) represent patients who had received any steroid. SLE, systemic lupus erythematosus.

BMD was first modelled with linear regression. LS BMD was associated with body mass index (BMI) z score at diagnosis (β=0.06, SE=0.01, p<0.0001) and Ca (β=0.29, SE=0.10, p=0.0036) (r2=0.31, p<0.0001, 79 patients). Hip BMD was associated with BMI z score at diagnosis (β=0.05, SE=0.02, p=0.0126) and intact parathyroid hormone (iPTH) (β=(−0.06), SE=0.03, p=0.0461) (r2=0.26, p=0.002, 44 patients). The hip BMD model remained the same following multiple imputations.

Low BMD was the outcome of interest in logistic regression (table 3 shows tested risk factors). None of the multivariable logistic models performed better than the univariable model containing BMI z score at diagnosis (OR 0.35, 95% CI 0.18 to 0.69, 80 patients). For every 1SD decrease in the BMI z score, the odds of low BMD increased by 65%.

Table 3

Univariate analysis of putative clinical and laboratory predictors of bone mineral density (BMD)


This report demonstrated that 15% of 80 pSLE patients, within 3 months of diagnosis, had low BMD (definition endorsed by both the ISCD and the AAP).14 15 Low BMD at the LS and/or hip was associated with low BMI, low Ca and high iPTH. In addition, we found that 33% of patients had abnormal LS BMD (−2.0<z score≤−1.0). Although abnormal BMD is not defined by ISCD guidelines, a previous study of paediatric rheumatology patients found that patients with fractures had a mean LS BMD in this range.6 As such, it is important to report on these patients as they probably represent a group at risk of fractures.

In this study, we found that low BMI z score was associated with low BMD, and for every SD decrease in the BMI z score, the odds of having a low BMD increased by 65% (OR 0.35). At the LS, low corrected calcium and BMI were associated with low BMD. At the hip, iPTH was identified to be a significant risk factor during univariate analysis. Although there was significant missing data, multiple imputations verified the model associated with hip BMD. However, as iPTH (probably a reflection of low vitamin D or calcium status) was measured in only 55% of the patients, we suggest caution in interpreting high iPTH as a risk factor for low BMD until this is confirmed in future studies.

There were some limitations to our study. We did not have information regarding our patients' genetic data that may influence BMD. The patients included had more active disease (higher SLEDAI) than those who were excluded. This was probably due to these patients returning more frequently so that DEXA could be performed within the time constraints of this study.

Although fractures are the hallmark of osteoporosis and prerequisites for diagnosis of ‘osteoporosis’,16 we did not systematically take x-rays of all patients. However, all patients with an LS BMD z score ≤−2.0 and/or back pain had spinal x-rays and none had a fracture. A prospective study of paediatric patients with rheumatic diseases found that seven out of nine vertebral fractures (found in 7% of patients) were symptomatic.6 As none of our patients with back symptoms or low BMD had a vertebral fracture on spinal x-ray, it is very unlikely that we missed more than one (or any) spinal fracture.

Weight-bearing physical activities are felt to be important for bone mass accrual,17 18 although the effect of physical activity has been conflicting among children with rheumatic diseases. In a group of children with different rheumatic diseases, the level of physical activities was not different in children with low BMD compared with those with normal BMD.5 Among juvenile arthritis patients, physical activities were important in increasing volumetric BMD or bone mineral content.19 20 These findings could be confounded by a greater decrease in weight-bearing activities if patients had significant lower limb arthritis. This effect may not be significant in our population who did not have significant lower limb arthritis. We did not have information regarding our patients' premorbid activity levels.

In this large inception cohort of pSLE patients, we found reduced BMD (at any site) in every one of six newly diagnosed pSLE patients. Steroid exposure within 3 months of diagnosis did not appear to alter BMD significantly. We suggest that this finding supports the hypothesis that the detrimental effect of systemic inflammation on BMD is more important than short-term exposure to corticosteroids even at high dose. We identified low Ca, high iPTH and low BMI z score at diagnosis to be associated with low BMD at the LS and/or hip. We recommend that all newly diagnosed patients with pSLE should have a baseline BMD, especially in patients with risk factors identified in this study. Longitudinal follow-up of this cohort of patients will clarify their future outcomes.


Supplementary materials


  • Competing interests None.

  • Ethics approval This study was conducted with the approval of the Hospital for Sick Children.

  • Provenance and peer review Not commissioned; externally peer reviewed.