High-resolution peripheral quantitative computed tomography reliability and accuracy of bone density and morphology properties in children

Abstract

Volumetric bone density, microarchitecture, and strength measures using HR-pQCT are valuable measures of bone health in pediatrics. Our cross-sectional study evaluated bone measure reproducibility in pediatric participants using repeat HR-pQCT (XtremeCT II, Scanco Medical) scans of non-dominant distal tibia and radius of 30 healthy children and adolescents (7-17 yr, 47% female) by 2 technicians. Additionally, we examined HR-pQCT and micro-CT of 26 cadaveric distal tibia specimens to evaluate agreement between the modalities. All HR-pQCT scans were analyzed using manufacturer-provided software (Image Processing Language, Scanco Medical) and a fully automated, previously validated, in-house algorithm, offering measures of bone microstructure (eg, trabecular plate-rod distribution, transverse trabeculae, trabecular bone strength) currently unavailable in the manufacturer-provided software. Root-mean-squared percent coefficient of variation (RMS-%CV) assessed precision and least significant change. Intraclass correlation coefficients (ICCs) using absolute-agreement, 2-way random-effects models assessed reliability. Pearson's correlation coefficients and ICC assessed linear relationships and agreements between HR-pQCT and micro-CT, respectively. In children and adolescents, RMS-%CV of HR-pQCT-derived trabecular bone measures at distal tibia and radius ranged from 0.8-4.4 to 0.8-6.0, respectively. Most cortical bone results were in a similar range. Both computational algorithms showed ICC > 0.90. RMS-%CV and least significant change values were lower for tibia than radius. ICCs were lower for cortical than trabecular outcomes. Most cadaveric results showed ICC > 0.83, other than trabecular bone thickness and modulus (ICC = 0.7). Pearson's r (>0.86) suggested strong correlations for almost all parameters. HR-pQCT and micro-CT results using in-house algorithm did not differ significantly, while manufacturer-provided algorithm results showed lower yet still moderate reliability (ICC > 0.55). Reliability of the second-generation HR-pQCT in pediatric participants is excellent - better in tibia vs radius. Our results support the high reproducibility of XtremeCT II scans and thus the use of this clinical imaging modality in studies of pediatric bone health.