Prediction of structural failure of tibial bone models under physiological loads: Effect of CT density–modulus relationships

Although finite element (FE) models can provide distinct benefits in understanding knee biomechanics, in particular the response of the knee to implants, their usefulness is limited by the modelling assumptions and input parameters. This study highlights the uncertainty of material input parameters derived from the literature and its limitation on the accuracy and usefulness of FE models of the tibia. An FE model of the intact human knee and a database of knee forces (muscles, ligaments and medial and lateral tibio-femoral contacts) were developed for walking and stair-descent activities. Ten models were constructed from ten different combinations of apparent bone density to elastic modulus material property relationships, published in the literature. Some of the published material property relationships led to predictions of bone strains in the proximal tibia which exceeded published failure criteria under loads imposed by normal activities. These relationships appear not to be applicable for the human tibia. There is a large discrepancy in proposed relationships that cover the cancellous bone density range. For FE models of the human tibia, the material relationship proposed by Morgan et al., which assumed species and anatomic site dependence, produced the most believable results for cancellous bone. In addition to casting doubt on the use of some of the published density–modulus relationships for analysis of the human proximal tibia, this study highlights the need for further experimental work to characterise the behaviour of bone with intermediate densities.