A three-dimensional finite element model of femoral head osteonecrosis was used to investigate the effect of mechanical stress on collapse propensity. The effect of subchondral plate viability and the effect of core decom pression and cortical bone grafting on stress-to-strength ratios (SSR's) were studied in a typical segm ental lesion. The mechanical integrity of the subchondral plate appeared to have little effect on effective load transmission through the femoral head. Core decom pression had only a slight effect on SSR's, unless the core extended beyond the lesion midsubstance, in which c ase detrimental increases in SSR were experienced. Grafting effectively decreased SSR 's only when the graft reached the subchondral plate. When a cortical graft passed only through the lesion m idsubstance or stopped short of the subchondral plate, SSR 's were consequentially increased.
A semi-automated method developed for mapping individual patient lesions from magnetic resonance images (MRI's) onto the finite elem ent model (FEM) w as tested for inter- and intra-analyst variability. Coefficients of variation of SSR were only 4-8% in the most critically stressed bone. A series of ten necrotic hips from actual patients were modelled and resulting SSR distributions computed. These distributions were adjusted for patient relative weight, for the use of walking aids, and in one hip, for core decom pression. In the six hips for which clinical outcome was known, the depth and extent of collapse were inferred from x-ray and the rate of collapse w as estimated. A collapse severity rating was developed to index the depth, extent, and rate of collapse, as well as the existence of clinical symptoms. These clinical outcome variables were statistically compared with a number of SSR -based param eters. Peak SSR values correlated very well with collapse rate (r=0.929, P<0.05), and to a lesser extent, with collapse severity score. Based on the results of this limited study, a large, more definitive patient series can now be undertaken much in the sam e manner, but also incorporating simulation of osteotomy, core decom pression, and bone grafting to better delineate surgical indications for the treatm ent of femoral head osteonecrosis.