Prediction of the compressive strength of human lumbar vertebrae.

The axial compressive strength of 98 motion segments of human thoracolumbar spines was measured in vitro under conditions simulating the in vivo environment. In addition, the density of the trabecular bone in the midplane of the vertebrae was determined by quantitative computed tomography; the areas of the vertebral endplates were measured by computed tomography as well as by photographic methods. The results of this experiment show that the compressive strength of human thoracolumbar vertebrae increases in proportion to bone density as well as to the size of the endplates. With knowledge of bone density and endplate area, compressive strength can be predicted with an error of estimate of 1 kN. Since both parameters (bone density as well as endplate area) can be determined in vivo by CT, the data from this experiment allow for an in vivo prediction of the strength of vertebral bodies by noninvasive means. The strength prediction may be employed to quantify the risk of fracture in physically very demanding tasks, to support expert opinion in trauma cases or to assist in therapeutic decisions in severe cases of osteoporosis. In the spine specimens investigated, compressive strength increased in the cranio-caudal direction. Bone density and endplate area data, the predictors of compressive strength, were available from 53 human thoracolumbar spines. It was therefore investigated how these parameters vary in the cranio-caudal direction. Bone density is practically constant in the thoracolumbar spine, while endplate area increases from T10 to L5. The observed cranio-caudal increase in compressive strength is therefore due to the increase in geometric dimensions and not to an increase in bone density. As this pattern was found to be extremely uniform in all spines investigated, a strength prediction of all vertebrae of an individual spine can be based on the density and area measurement of only one vertebra, say of L3; strength of the adjacent vertebrae may then be extrapolated with high accuracy.