Validation of calcaneus trabecular microstructure measurements by HR-pQCT

OBJECTIVE:

Assessment of calcaneus microstructure using high-resolution peripheral quantitative computed tomography (HR-pQCT) might be used to improve fracture risk predictions or to assess responses to pharmacological and physical interventions. To develop a standard clinical protocol for the calcaneus, we validated calcaneus trabecular microstructure measured by HR-pQCT against 'gold-standard' micro-CT measurements.

METHODS:

Ten human cadaveric feet were scanned in situ using HR-pQCT (isotropic 82μm voxel size) at 100, 150 and 200ms integration times, and at 100ms integration time following removal of the calcaneus from the foot (ex vivo). Dissected portions of these bones were scanned using micro-computed tomography (micro-CT) at an isotropic 17.4μm voxel size. HR-pQCT images were rigidly registered to those obtained with micro-CT and divided into multiple 5mm sided cubes to evaluate and compare morphometric parameters between the modalities. Standard HR-pQCT measurements (derived bone volume fraction (BV/TV(d)); trabecular number, Tb.N; derived trabecular thickness, Tb.Th(d); derived trabecular spacing, Tb.Sp(d)) and corresponding micro-CT voxel-based measurements (BV/TV, Tb.N, Tb.Th, Tb.Sp) were compared.

RESULTS:

A total of 108 regions of interest were analysed across the 10 specimens. At all integration times HR-pQCT BV/TV(d) was strongly correlated with micro-CT BV/TV (r(2)=0.95-0.98, RMSE=1%), but BV/TV(d) was systematically lower than that measured by micro-CT (mean bias=5%). In contrast, HR-pQCT systematically overestimated Tb.N at all integration times; of the in situ scans, 200ms yielded the lowest mean bias and the strongest correlation with micro-CT (r(2)=0.61, RMSE=0.15mm(-1)). Regional analysis revealed greater accuracy for Tb.N in the superior regions of the calcaneus at all integration times in situ (mean bias=0.44-0.85mm(-1); r(2)=0.70-0.88, p<0.001 versus mean bias=0.63-1.46mm(-1); r(2)<0.10, p≥0.21 for inferior regions). Tb.Sp(d) was underestimated by HR-pQCT compared to micro-CT, but showed similar trends with integration time and the region evaluated as Tb.N. HR-pQCT Tb.Th(d) was also underestimated (mean bias=0.081-0.102mm) and moderately correlated (r(2)=0.55-0.59) with micro-CT Tb.Th, independently from the integration time. Stronger correlations, smaller biases and error were found in the scans of the calcaneus ex vivo compared to in situ.

CONCLUSION:

Calcaneus trabecular BV/TV(d) and trabecular microstructure, particularly in the superior region of the calcaneus, can be assessed by HR-pQCT. The highest integration time examined, 200ms, compared best with micro-CT. Weaker correlations for microstructure at inferior regions, and also with lower integration times, might limit the use of the proposed protocol, which warrants further investigation in vivo.