Dual-Energy X-Ray Absorptiometry in Predicting Mechanical Characteristics of Rat Femur

Abstract

Dual-energy X-ray absorptiometry (DXA), geometrical measurements, and mechanical testing of the rat femoral shaft and neck were performed on both femora of 51 Sprague-Dawley rats to: (i) determine the reproducibility of the DXA, geometrical, and biomechanical measurements of rat femora; (ii) determine linear and power-law ( y = ax b ) associations between the site-specific bone mineral variables and the actual mechanical characteristics of the given sites; (iii) develop, if sufficiently strong associations were found, appropriate prediction equations for the breaking load ( F) and flexural rigidity ( EI) of the femoral shaft and neck (only for F); and (iv) validate these equations in terms of accuracy of prediction. In the majority of the DXA measurements, the repeatability of the measurements was good, the CV rms varying between 1.2% and 3.9% in the bone mineral density (BMD) measurements and between 1.6% and 13.8% in the bone mineral content (BMC) measurements. DXA also proved accurate in length measurements of the rat femur (measurement error <1%). The manual (digimatic caliper-obtained) geometrical measurements of the rat femora were equally precise, the CV rms values varying between 0.2% and 5.0%. The repeatability of the biomechanical testings of these femora varied between 5.0% and 14.7%. Virtually all of the power-law and linear models explained more than 80% (at best 97%) of the variation in the F of the femoral shaft and neck, and the EI of the femoral shaft. Despite the high group-level correlations between the DXA-based predictions of bone strength and the actual breaking loads of the rat femora, and good precision of DXA, the ability of any DXA-based estimate to predict accurately the actual biomechanical characteristics of an individual bone remained relatively poor. In extreme cases, the prediction error could be tens of percent. Despite this we feel that bone strength-estimating equations can be used in the group-level analyses of experimental and clinical studies. Care must be taken, however, when choosing the most appropriate prediction method for a particular study.