Reliability of TBS: experience from a network of tertiary care centres

Abstract

Introduction Trabecular bone score (TBS), a textural estimate of bone microarchitecture, is increasingly used as part of fracture risk assessment. It is derived from lumbar spine dual-energy X-ray absorptiometry (DXA) scans and provides additional information for 10-year absolute fracture risk calculation using the FRAX tool. The precision of TBS measurements is important for determining if the change in TBS over time is significant enough to indicate a real biological change clinically. Objectives This study aims to calculate the precision, as expressed by the root-mean-square percent coefficient of variation (RMS-%CV), and the least significant change (LSC) of TBS across three academic hospitals in Toronto, Canada. Methods Participants were prospectively recruited from Toronto General Hospital (TGH), Toronto Western Hospital (TWH) and Princess Margaret Cancer Centre (PMCC), all part of the University Health Network (UHN) Centre of Excellence in Skeletal Health Assessment (CESHA). Four ISCD-certified technologists each performed DXA scans on 30 participants, using four Hologic DXA machines (one Horizon A and three Discovery As), according to standard protocol for precision testing. At each centre, the participants were scanned twice at the lumbar spine by the same operator on the same day. Both scans were performed within a few minutes of each other, with only time for the participant to get off the table and be repositioned. TBS was analyzed from each of the two lumbar spine DXA images using TBS iNsight version 3.0. As a result, two TBS values were generated per participant. This study was done as part of quality control and written informed consent was obtained prior to the scans. RMS-%CV and LSC were computed for test-retest TBS values at individual centres, and combined for participants across all centres. Results A total of 120 participants (35 men, 85 women) completed this precision study. Mean age was 54 (age range= 20-90), mean BMD at the lumbar spine and total hip were 0.954 (SD± 0.171) g/cm2 and 0.879 (SD±0.164) g/cm2, and mean T-scores were -0.9 (SD±1.5) and -0.5 (SD± 1.2), respectively. The RMS-%CV for lumbar spine and total hip BMD were 0.95% and 1.04%, respectively. The RMS-%CV and LSC for TBS are shown in table below: Conclusion In summary, TBS as determined from lumbar spine DXA images has a precision error of 2.85% compared to 0.95% for lumbar spine BMD. BMD laboratories should perform precision testing for TBS in addition to BMD. Clinicians should interpret changes in TBS values and the calculated fracture risk keeping this level of precision in mind.