gms | German Medical Science

Objectives: While pedicle screw rod instrumentation is still the gold standard for the treatment of a diverse range of spinal degenerative disorders, anchorage in the elderly spine with poor bone quality remains challenging. Yet, there are no models available to assess the specific influence of screw design parameters at the pedicle screw-bone interface. In our study, we aimed to develop numerical micro finite element (μFE) models based on μCT scans to investigate the screw bone interface on a micro scale level.

Methods: 24 vertebrae extracted from 5 fresh frozen human lumbar spines were divided into two groups based on CT-based bone mineral density (BMD) (group A: high BMD, n=12, uncemented fixation; group B: low BMD, n=12, cemented fixation). Two screw types were implanted in the pedicles of each vertebra (left: Aesculap Ennovate pedicle screw, Pentacore® design; right: Aesculap S4 pedicle screw, conical screw design). Specimens were then scanned with a μCT at 56 μm and images converted to µFE models of sub-regions around each screw with a mean of 6 million nodes. These μFE models were assigned linear isotropic materials and solved under static 0.1mm displacement simulating axial screw pull-out. The predicted pull-out stiffness was then calculated as a ratio of reaction force to applied displacement. In parallel, physical tests were performed regarding pullout and torsion strength on sub groups of group A and B (n=6). Correlations were then established between the experimental strength, the predicted pull-out stiffness, and the local apparent bone density (BV/TV) calculated from the μCT scans.

Results and Conclusion: Experimental pull-out and torsional strengths were moderately to strongly correlated to BV/TV around the implanted screw for both screw type (pull-out, R²>0.81; torsion, R²>0.65). Numerical predictions of pull-out stiffness also correlated well with experimental pull-out and torsional strengths (R²>0.73 and R²>0.69), with stronger correlations for the Pentacore screw design (R²>0.87). Predicted ratios of left and right screw fixation stiffnesses were not significantly different with the experimental ratios in both loading modes (p>0.482), showing similarities in structural effects of screw design for the experimental and μFE models. These predictions also showed differences in bone tissue stresses around the two types of screws depending on local bone density and proximity to the cortical bone. The use of μFE models including a detailed geometrical representation of bone microstructure and interfaces allowed structural predictions that correlated well with experimental measurements. Such models showed how fixation stiffness and bone tissue stresses are influenced not only by bone quality but loading and implant design. This approach will be extended to simulate torsional loads and other physiological loads such as compression bending as well as variations in cementing techniques, and how these could potentially minimize the risks of anchorage failures.