Bone Abstracts (2013) 2 OC18 | DOI: 10.1530/boneabs.2.OC18

Trabecular bone score applied to normal children's lumbar spine DXA scans

Judith Adams1, Elizabeth Marjanovic2, Stephen Roberts3, Zulf Mughal4 & Kate Ward5


1Manchester Royal Infirmary and Manchester Academic Health Science Centre, Central Manchester University Hospitals NHS Foundation Trust, Manchester, UK; 2Arthritis Research UK Epidemiology Unit, Institute of Inflammation and Repair, Centre for Musculoskeletal Research, University of Manchester, Manchester, UK; 3Institute of Population Health, Centre for Biostatistics, University of Manchester, Manchester, UK; 4Paediatrics and Manchester Academic Health Science Centre, Central Manchester University Hospitals NHS Foundation Trust, Manchester, UK; 5MRC Human Nutrition Research, Elsie Widdowson Laboratory, Cambridge, UK.


Trabecular bone score (TBS) extracts a texture parameter from pixel grey-level variations in DXA lumbar spine images. The TBS is claimed to be a measure of trabecular structure and was validated in an in-vitro study of vertebral bodies with micro-CT1. TBS has shown the potential for fracture pre-diction in adults2. However, data are sparse regarding the reliability and usefulness of TBS3 and the method has not previously been applied in children.

Methods: The Manchester Children’s DXA BMD reference database (Hologic QDR Discovery) was published and is widely applied in UK4. The LS-DXA (n=463; males; age mean (S.D.) range; n=252, 15.0 (5.6), 5.2–25 years; females n=211; 12.6 (3.9), 5.4–20.7 years) have been retrospectively analyzed and TBS L1–L4 extracted. The relationships between TBS and age, BMD (g/cm2), bone mineral apparent density (BMAD), height, weight and BMI were examined. Linear regression analysis explored predictors of TBS; model 1 included LSBMD, age, BMI, model 2 LSBMAD, age, BMI. Significant predictors are reported.

Results: There was a significant correlation between TBS and age in females (r=0.63, P<0.001) and males (r=0.67, P<0.001), and between TBS and BMI (females r=0.41, P<0.001; males r=0.41, P<0.001). For both males and females mean TBS was lower in the younger (<13 years) relative to the older (>13 years) age groups (P<0.001), but not significantly different between children in annually adjacent year groups. TBS was lower (P<0.001) in children with a lower (<20 kg/m2) vs a higher BMI (>20 kg/m2). Correlation between BMAD and TBS was r=0.61, P<0.01 in females and r=0.62, P<0.01 in males. Predictors of TBS were: females: i) L1–L4 aBMD (P<0.001), and BMI (P=0.002); ii) L1–L4 BMAD (P<0.001), age (P<0.001), BMI (P=0.086); males: i) L1–L4 BMD (P<0.001), BMI (P<0.001); ii) L1–L4 BMAD (P<0.001), age (P<0.001), BMI (P=0.004).

Conclusions: TBS is related to age, aBMD and BMAD in growing children. The dichotomy of the relationship with age and BMI may be due to technical limitations in the method. Our findings require further investigation in this and other populations. A critical evaluation of the tool and whether it improves our understanding of children’s bone health is required.

K Ward is funded by Medical Research Council Grant Code U10596037.

Acknowledgement for access to specialist TBS analysis software: http://www.medimaps.fr/.

References: 1. J Clin Densitom 2011 14 302–312.

2. J Bone Miner Res 2011 26 2762–2769.

3. Osteoporos Int 2012 23 1489–1501.

4. Arch Dis Child 2007 92 53–59.

Article tools

My recent searches

No recent searches.