Reproducibility of trabecular bone score with different scan modes using dual-energy X-ray absorptiometry: a phantom study.

Abstract

The trabecular bone score (TBS) accounts for the bone microarchitecture and is calculated on dual-energy X-ray absorptiometry (DXA). We estimated the reproducibility of the TBS using different scan modes compared to the reproducibility bone mineral density (BMD). A spine phantom was used with a Hologic QDR-Discovery A densitometer. For each scan mode [fast array, array, high definition (HD)], 25 scans were automatically performed without phantom repositioning; a further 25 scans were performed with phantom repositioning. For each scan, the TBS was obtained. The coefficient of variation (CoV) was calculated as the ratio between standard deviation and mean; percent least significant change (LSC%) as 2.8 × CoV; reproducibility as the complement to 100 % of LSC%. Differences among scan modes were assessed using ANOVA. Without phantom repositioning, the mean TBS (mm(-1)) was: 1.352 (fast array), 1.321 (array), and 1.360 (HD); with phantom repositioning, it was 1.345, 1.332, and 1.362, respectively. Reproducibility of the TBS without phantom repositioning was 97.7 % (fast array), 98.3 % (array), and 98.2 % (HD); with phantom repositioning, it was 97.9 %, 98.7 %, and 98.4 %, respectively. LSC% was ≤ 2.26 %. Differences among scan modes were all statistically significant (p ≤ 0.019). Reproducibility of BMD was 99.1 % with all scan modes, while LSC% was from 0.86 % to 0.91 %. Reproducibility error of the TBS was 2-3-fold higher than that of BMD. Although statistically significant, differences in TBS among scan modes were within the highest LSC%. Thus, the three scan modes can be considered interchangeable.