Role of endocortical contouring methods on precision of HR-pQCT-derived cortical micro-architecture in postmenopausal women and young adults

A Summary Precision errors of cortical bone micro-architecture from high-resolution peripheral quantitative computed tomography (pQCT) ranged from 1 to 16 % and did not differ between automatic or manually modified endocortical contour methods in postmenopausal women or young adults. In postmenopausal women, manually modified contours led to generally higher cortical bone properties when compared to the automated method. Introduction First, the objective of the study was to define in vivo precision errors (coefficient of variation root mean square (CV%(RMS))) and least significant change (LSC) for cortical bone micro-architecture using two endocortical contouring methods: automatic (AUTO) and manually modified (MOD) in two groups (postmenopausal women and young adults) from high-resolution pQCT (HR-pQCT) scans. Second, it was to compare precision errors and bone outcomes obtained with both methods within and between groups. Methods Using HR-pQCT, we scanned twice the distal radius and tibia of 34 postmenopausal women (mean age +/- SD 74 +/- 7 years) and 30 young adults (27 +/- 9 years). Cortical micro-architecture was determined using AUTO and MOD contour methods. CV%(RMS) and LSC were calculated. Repeated measures and multivariate ANOVA were used to compare mean CV% and bone outcomes between the methods within and between the groups. Significance was accepted at P < 0.05. Results CV%(RMS) ranged from 0.9 to 16.3 %. Within-group precision did not differ between evaluation methods. Compared to young adults, postmenopausal women had better precision for radial cortical porosity (precision difference 9.3 %) and pore volume (7.5 %) with MOD. Young adults had better precision for cortical thickness (0.8 %, MOD) and tibial cortical density (0.2 %, AUTO). In postmenopausal women, MOD resulted in 0.2-54 % higher values for most cortical outcomes, as well as 6-8 % lower radial and tibial cortical BMD and 2 % lower tibial cortical thickness. Conclusions Results suggest that AUTO and MOD endocortical contour methods provide comparable repeatability. In postmenopausal women, manual modification of endocortical contours led to generally higher cortical bone properties when compared to the automated method, while no between-method differences were observed in young adults.