This article is more than 5 years old. Information may no longer be current.
Bone mineral density score disparity highlights need to redefine osteoporosis in children
Defining pediatric osteoporosis based on lumbar spine-bone mineral density z score may be problematic due to differences in reference databases and variability in prevalent vertebral fracture association and prediction performance across databases, according to a presentation at the American Society for Bone and Mineral Research 2014 Annual Meeting.
“There is marked disparity among bone mineral density z scores generated by different reference databases,” Leanne Marie Ward, MD, FRCP, FAAP, of the University of Ottawa, Canada, told Endocrine Today. “This fact alone makes the use of a lumbar spine-bone mineral density (LS-BMD) threshold to define osteoporosis challenging in children.”
Leanne Marie Ward
The association between vertebral fracture (VF) and LS-BMD z score across reference databases does not appear to be influenced by the choice of reference data, according to researchers.
Ward and colleagues looked at 186 children recently diagnosed with leukemia who underwent LS-BMD testing. Four reference databases were used to generate z scores. Prevalent VF was assessed using semi-quantitative technique.
Logistic regression was used to evaluate relationships between VF (sustained in 15.6% of children) and LS-BMD z score, either as a continuous factor or a binary risk factor at ±2.0 standard deviation (SD). Area under the curve (AUC) and net reclassification improvement (NRI) were used to assess prediction performance.
Between 15% and 48% of children had LS-BMD z scores ≤–2.0. Based on the –2 SD cut point, the relationship between LS-BMD z score and VF generated ORs from 2.3 (95% CI, 1–5.3) to 5 (95% CI, 2–12.4), AUCs from 0.60 (95% CI, 0.51–0.70) to 0.66 (95% CI, 0.56–0.76) and NRIs from -0.64 (95% CI, –1.02 to –0.26) to –0.08 (95% CI, –0.35 to 0.18).
Conversely, with LS-BMD z score used as a continuous risk factor, ORs narrowly ranged from 2.4 (95% CI, 1.6–3.7) to 2.6 (95% CI, 1.6–4.1), AUCs from 0.71 (95% CI, 0.59–0.82) to 0.72 (95% CI, 0.6–0.83) and NRIs from –0.17 (95% CI, -0.57 to 0.22) to –0.06 (95% CI, –0.45 to 0.33).
“This issue merits consideration regardless the fracture type to which BMD z scores are being associated,” Ward said. — by Allegra Tiver
For More Information: Ward LM. Abstract 1094. Presented at: American Society for Bone and Mineral Research 2014 Annual Meeting; Sept. 12-15, 2014; Houston.
Disclosures: This study was funded by the Canadian Institutes for Health Research.
Perspective
Back to Top
Dr. Ward’s research provides an important cautionary tale about the limitations of DXA as a way to diagnose pediatric osteoporosis. It’s one of many tools we use to assess bone health in pediatrics, and I agree that DXA cannot be used to reliably diagnose vertebral fractures in this population. Fortunately, a low bone density is no longer needed to establish the diagnosis of osteoporosis in children or teens once a vertebral fracture has occurred.
The recent guidelines for assessing bone health in youth from the 2013 Pediatric Position Development Conference have been published and include information on the indications for densitometry, interpretation of data and fracture prediction.
These guidelines revised the criteria for the diagnosis of osteoporosis in childhood. Documentation of one or more vertebral fractures occurring without major trauma is sufficient to establish the diagnosis; measuring BMD can add to the assessment of these patients but is not required. Alternatively, children with low bone density and two low trauma long bone fractures by age 10, or three by age 19, can be diagnosed with osteoporosis
A DXA bone density test has proven to be helpful in identifying adults at risk for hip or spine fractures in the elderly. Determining how good it is for predicting fractures in young people, however, has been more challenging. One key problem is the reliability of the reference data for what is “normal.” Adult norms have been derived from very large cohorts of healthy individuals, which allows for really good comparisons. Most pediatric normative data to date have been relatively small, although a recent study funded by the National Institutes of Health has improved this situation. Depending upon which reference data are used as the “yardstick” for what is normal can give quite different estimates of how a low patient’s bone density is for age.
Dr. Ward’s group looked at children with leukemia who had had vertebral fractures. Low trauma spine fractures are abnormal in childhood and a clinical indication that bone strength is reduced. The investigators measured bone density in the spine by DXA and compared the standard deviation scores for the results using different data sets. They found marked differences in these scores depending upon which data set was used.
The take-away point was that you can’t use bone density to reliably predict vertebral fractures in children.
Since the Ward team only looked at bone density in relation to one type of osteoporotic fracture, the study does not tell us how well DXA might predict fracture at other sites. For example, children with cerebral palsy or other disorders that reduce their mobility, are more likely to fracture in the lower extremity than the spine. In these children, evaluating bone density in the femur is a better predictor of their fracture risk than is the spine bone density.
Dr. Ward and her team have provided a valuable reminder that DXA is only part of a comprehensive skeletal assessment in pediatrics. Their work underscores the limitations of the normative pediatric data for DXA but beyond that, bone quality can be reduced to the point bones will fracture before the bone density result falls into the clearly abnormally low range. The new guidelines for diagnosis of pediatric osteoporosis reflect this important observation. More work will be needed to optimize the normative DXA data for children and teens and to assess their predictive value for fractures at all skeletal sites.
Ideally, physicians will develop the means to more accurately identify patients at risk for fracture before the bone breaks actually occur. This will likely require more careful observation studies in young patients with chronic diseases that we know are threatening to their skeletons. Bone densitometry will be part of this predictive model but likely will require consideration of clinical factors such as nutrition, activity, medications and disease course as well. Most importantly, we must track the occurrence of low trauma fractures at the spine and other sites as the “gold standard” indication that bone quality and/or quantity has been compromised.
For More Information: Gordon CM, Leonard MB, Zemel BS. 2013 Pediatric Position Development Conference: Executive summary and reflections. J Clin Densitometry. 2014;doi:17: 219-224.