Introduction:​ Stemmed tibial implants are used in total knee arthroplasties requiring additional support due to deficient proximal bone. Surgeries involving intramedullary stems have a clinical track record with cement fixation. In recent years in the United States, these stemmed implants have also been implanted with hybrid fixation, where only the metaphyseal region is cemented and a canal filling stem is used. Overall, hybrid fixation has shown 5-20% failure rates up to 6 years post-operatively, but long-term results are not yet available. To better understand the implant and surgical parameters which may affect the long-term clinical outcome, both experimental and computational studies have been performed to investigate initial post-operative implant motion. However, to-date an experimentally validated specimen-specific computational model has not been previously created.

Methods:​ In-vitro testing was performed on seven matched-pair tibias with modular stemmed tibial implants with hybrid fixation. Motion of the tray and stem, along with the bone, were measured using an optical motion capture system in which the precision, repeatability, and accuracy were measured to be less than 10 ^m for the current test set-up. In-vitro data provided analysis of both the rigid body assumption of the stemmed implant, and assessment of relative implant-bone motion. A specimen-specific finite element (FE) model was created from pre- and post-operative computed tomography (CT) data. The model was tuned to the in-vitro data using both one-factor-at-a-time and factorial analyses. The tuned FE model was used to investigate the effect fixation and bone densities have on initial post-operative implant motion.

Results:​ In-vitro tray-stem motion was found in the anterior-posterior and medial-lateral directions. Implant-bone motion was greatest in the proximal-distal direction, with the tray having more relative motion than the stem tip. In addition, high variability in relative motion was measured both between and among specimen. No correlations were found between implantbone motion and stem size, tibial tray coverage, age, gender, or bone density. Sensitivity studies on the FE model found the bone's deflection in bending was significantly affected by the definition of bone density and the empirical relationship used to define elastic modulus. The distal and proximal constraints of the FE model were found to have the largest effect on the motion of the bone, tray and stem. The tuned FE model found decreased implant-bone motion with cement fixation, compared to hybrid fixation, as well as normal bone density, compared to osteoporotic bone density.

Conclusions:​ A tuned specimen-specific FE model of a modular stemmed implant in a tibial bone was created. The CT-to-FE protocol provides a method for accurate FE model creation and parameter assessment through sensitivity studies. With additional tuning, the current FE model can be used to evaluate both implant and surgical parameters to provide predictions for long-term clinical results.