Simulated increased soft tissue thickness artefactually decreases trabecular bone score: a phantom study

BACKGROUND: Trabecular bone score (TBS), which has been proposed to be used in complementary with bone mineral density (BMD) to improve the assessment of fracture risk, is negatively associated with body mass index (BMI). The effect of soft tissue, which is expected to be thicker in subjects with high BMI, on TBS was studied using three scan types: Hologic with fast array mode (Hfa), Hologic with high definition mode (Hhd), and GE-Lunar iDXA. METHODS: A spine phantom provided by Hologic for routine quality control procedure was scanned using three scan types: Hfa, Hhd, and iDXA. The phantom was scanned with an overlying soft tissue equivalent material (bolus used in radiotherapy) of 0 (without), 1, 2.5, 3.5, 5 and 7.5 cm thick. For each setting, 30 acquisitions were performed in the same way as for the quality control procedure. TBS was calculated using TBS iNsight® software version 2.1 on the same regions of interest as those used for lumbar spine BMD. RESULTS: Mean ± SD TBS of the phantom (without overlying soft tissue) were 1.379 ± 0.018, 1.430 ± 0.009, and 1.423 ± 0.005 using Hfa, Hhd, and iDXA, respectively. A one-way repeated measures ANOVA showed that there were statistically differences in TBS due to different thicknesses of soft tissue equivalent materials for all three scan types (p < 0.001). A Tukey post-hoc test revealed that the decrease in TBS was statistically significant (p < 0.001) when the soft tissue thickness was 1 cm (−0.0246 ± 0.0044, −0.0319 ± 0.0036, and −0.0552 ± 0.0015 for Hfa, Hhd, and iDXA, respectively). Although to a lesser degree, the effects were also statistically significant for BMD (p < 0.05): an increase for Hfa and Hhd but a decrease for iDXA. However, these changes did not exceed the least significant change (LSC) derived from patients. CONCLUSIONS: Increased soft tissue thickness results in lower TBS value. Although BMD is also affected, it is unlikely to pose a clinical problem because the change is unlikely to exceed the patient-derived LSC.