BIOMECHANICAL ANALYSIS OF SCREW LOAD SHARING IN PEDICLE FIXATION OF THE LUMBAR SPINE

Segmental fixation of the spine by means of rods or plates and pedicle screws posteriorly usually results in a stable and rigid construct. T he magnitude of the stresses on the instrumentation particularly at th e bone-screw and rod (or plate) interfaces will depend on the load dis tribution between the bone and the implant as well as the number of si tes of bone purchase of the implant. If a fusion is to be obtained in the case of a degenerative spine, the role of the instrumentation is t o prevent translation of the motion segments, thereby allowing the com pressive loads to be transmitted through the vertebral bodies and the degenerated discs. In the case of a fracture, the instrumentation is m ade to withstand the bulk of the loads since the structural integrity of a motion segment may have been lost. This study was undertaken to e valuate the effect of different constructs on the stresses in Cotrel-D uboussett (C-D) pedicle (tulip) screws close to their junction with th e rod bridging the motion segments to be immobilized. In order to mini mize the effect of anatomical and material property variation between spines, adult porcine spines were used, thus providing a reproducible experimental model. Fresh 3-year-old porcine spines were potted in hol ders after the soft tissues were removed (with the exception of the li gamentous structures). The potted spine was then fixed into the loadin g frame of a materials testing machine (MTS 858 Bionix Test System) an d cycled 500 times with an axial load of 380 N. Following this pre-cyc ling, 6.5-mm diameter C-D tulip screws mounted with strain gauges (Mic ro-Measurements CEA-06-125UN350) were inserted and aligned. The constr ucts tested were an intact spine with an eight-screw assembly, a spine with an L4 corpectomy with an eight-screw assembly and a spine with a n L4 corpectomy with a four-screw assembly. Each of these constructs w as loaded with and without the presence of cross-linking devices. For each construct, five spines were tested. A computerized data acquisiti on system converted the strain gauge data to a bending moment for each screw. Statistical tests at the 95% level of significance show that t he presence of cross links did not affect the mean bending moments wit hin a given construct. Although the bending moments were greater in th e distal screws than in the proximal screws in the four-screw corpecto my model, the overall bending moments of the corpectomy models were gr eater than those encountered in the intact models by a factor of 3. In the four-screw corpectomy model, the bending moments on distal screws were larger than on the proximal screws. The proximal screws accounte d for approximately 12% of the total bending moment, whereas the corre sponding fractions for the distal screws was approximately 75%. In the corpectomy models, the loads of the intermediate screws were distribu ted proportionate to the number of distal screws; thus failure is less likely to occur in vivo with four distal screws than with two screws since the proportionate load increased from 60 to 190 Nm, respectively .