Studying bone mineral density in young people: The complexity of choosing a pQCT reference database

Many chronic illnesses affect bone health, and commonly lead to mineralization abnormalities in young people. As cortical and trabecular bone may be differentially affected in certain diseases, an imaging technique that allows for detailed study of the bone structure is required. Peripheral quantitative computed tomography (pQCT) overcomes the limitations of dual energy X-ray absorptiometry (DXA) and is perhaps more widely available for use in research than bone biopsy. However, in contrast to DXA, where there are large reference datasets, this is not the case for pQCT.

Fifty-five children and young adults aged 7 to 30 years had the non-dominant tibia scanned at the 3% & 4% sites for trabecular bone mineral density and the 38% site for cortical bone mineral density and bone mineral content. Image acquisition and analysis was undertaken according to the protocols of two of the largest reference datasets for tibial pQCT. The Z-scores generated were compared to examine the differences between protocols and the differences from the expected median of zero in a healthy population.

The trabecular bone mineral density Z-scores generated by the two protocols were similar. The same was true for cortical mineral content Z-scores at the 38% site. Cortical bone mineral density was significantly different between protocols and likely affected by differences in the ethnicity of our cohort compared to the reference datasets. Only one reference dataset extended from childhood to young adulthood. Only trabecular bone mineral density, periosteal and endosteal circumference Z-scores from one methodology were not significantly biased when tested for deviation of the median from zero.

pQCT is a useful tool for studying trabecular and cortical compartments separately but, there are variations in pQCT scanning protocols, analysis methodology, and a paucity of reference data. Reference datasets may not be generalizable to local study populations, even when analysed using identical analysis protocols.