Abstract
Introduction and Objective
Curative resection of proximal humerus tumours is now possible in this era of limb salvage with endoprosthetic replacement considered as the preferred reconstructive option. However, it has also been linked with mechanical and non-mechanical failures such as stem fracture and aseptic loosening. One of the challenges is to ensure that implants will endure the mechanical strain under physiological loading conditions, especially crucial in long surviving patients. The objective is to investigate the effect of varying prosthesis length on the bone and implant stresses in a reconstructed humerus-prosthesis assembly after tumour resection using finite element (FE) modelling.
Methods
Computed tomography (CT) scans of 10 humeri were processed in Mimics 17 to create three-dimensional (3D) cortical and cancellous solid bone models. Endoprostheses of different lengths manufactured by Stryker were modelled using Solidworks 2020. The FE models were divided into four groups namely group A consisting of the intact humerus and groups B, C and D composed of humerus-prosthesis assemblies with a body length of 40, 100 and 120 mm respectively and were meshed using linear 4-noded tetrahedral elements in 3matic 13. The models were then imported into Abaqus CAE 6.14. Isotropic linear elastic behaviour with an elastic modulus of 13400, 2000 and 208 000 MPa were assigned to the cortical bone, cancellous bone and prosthesis respectively and a Poisson's ratio of 0.3 was assumed for each material. To represent the lifting of heavy objects and twisting motion, a tensile load of 200 N for axial loading and a 5 Nm torsional load for torsional loading was applied separately to the elbow joint surface with the glenohumeral joint fixed and with all contact interfaces defined as fully bonded. A comparative analysis against literature was performed to validate the intact model. Statistical analysis of the peak von Mises stress values collected from predicted stress contour plots was performed using a one-way repeated measure of analysis of variance (with a Bonferroni post hoc test) using SPSS Statistics 26. The average change in stress of the resected models from the intact state were then determined.
Results
The validation of the intact humerus displayed a good agreement with literature values. The peak bone stress occurred distally above the coronoid and olecranon fossa closer to the load application region in the intact and resected bone models with a significant amount of loading borne by the cortical bone, while the peak implant stress occurred at the bone-prosthesis contact interface under both loading conditions. Based on the results obtained, a statistically significant difference (p =.013) in implant stress was only seen to occur between groups B and C under tension. Results illustrate initiation of stress shielding with the bone bearing lesser stress with increasing resection length which may eventually lead to implant failure by causing bone resorption according to Wolff's law. The peak implant stress under torsion was 3–5 times the stress under tension. The best biomechanical behaviour was exhibited in Group D, having the least average change in stress from the intact model, 5% and 3.8% under tension and torsion respectively. It can be deduced that the shorter the prosthesis length, the more pronounced the effect on cortical bone remodelling. With the maximum bone and implant stresses obtained being less than their yield strength, it can be concluded that the bone-implant construct is safe from failure.
Conclusions
The developed FE models verified the influence of varying the prosthesis length on the bone and implant stresses and predicted signs of stress shielding in longer endoprostheses. By allowing for 2 cm shortening in the upper extremity and post-surgical scarring, it is beneficial to err towards a shorter endoprosthesis.
