Influence of Interpersonal Geometrical Variation on Spinal Motion Segment Stiffness Implications for Patient-Specific Modeling

Study Design. A validated finite element model of an L3-L4 motion segment is used to analyze the effects of interpersonal differences in geometry on spinal stiffness. Objective. The objective of this study is to determine which of the interpersonal variations of the geometry of the spine have a large effect on spinal stiffness. This will improve patient-specific modeling. Summary of Background Data. The parameters that define the geometry of a motion segment are vertebral height, disc height, endplate width, endplate depth, spinous process length, transverse process width, nucleus size, lordosis angle, facet area, facet orientation, and the cross-sectional areas of the ligaments. All these parameters differ between patients. The influence of each parameter on spinal stiffness is largely unknown and such knowledge would greatly help in patient-specific modeling of the spine. Methods. The range of interpersonal variation of each of the geometric parameters was set at mean +/- 2SD (covering 95% of the population). Subsequently, we determined the effect of each of these ranges on the bending stiffness in flexion, extension, axial rotation, and lateral bending. Results. Disc height had the largest influence; a maximal disc height reduced the spinal stiffness to 75-86% of the mean motion segment stiffness, and a minimal disc height increased the spinal stiffness to 154-226% of the mean motion segment stiffness. Lordosis angle, transversal and longitudinal facet angle, endplate depth, and area of the capsular ligament also had a substantial influence (>5%) on the stiffness, but considerable less than the influence of the disc height. Ligament areas, nucleus size, spinous process length, and length of processes are of negligible effect (<2%) on the stiffness. Conclusion. The disc height should be accurately determined in patients to estimate the spinal stiffness. Ligament areas, nucleus size, spinous process length, and transverse process width do not need patient-specific modeling.