Proximal humeral fractures occur frequently, with fixed angle locking plates often being used for their treatment. However, the failure rate of this fixation is high, ranging between 10 and 35%. Numerous variables are thought to affect the performance of the fixation used, including the length and configuration of screws used and the plate position. However, there is currently limited quantitative evidence to support concepts for optimal fixation. The variations in surgical techniques and human anatomy make biomechanical testing prohibitive for such investigations. Therefore, a finite element osteosynthesis test kit has been developed and validated - SystemFix. The aim of this study was to quantify the effect of variations in screw length, configuration and plate position on predicted failure risk of PHILOS plate fixation for unstable proximal humerus fractures using the test kit.

Twenty-six low-density humerus models were selected and osteotomized to create a malreduced unstable three-part fracture AO/OTA 11-B3.2 with medial comminution which was virtually fixed with the PHILOS plate. In turn, four different screw lengths, twelve different screw configurations and five plate positions were simulated. Each time, three physiological loading cases were modelled, with an established finite element analysis methodology utilized to evaluate average peri-screw bone strain, this measure has been previously demonstrated to predict experimental fatigue fixation failure.

All three core variables lead to significant differences in peri-screw strain magnitudes, i.e. predicted failure risk. With screw length, shortening of 4 mm in all screw lengths (the distance of the screw tips to the joint surface increasing from 4 mm to 8 mm) significantly (p < 0 .001) increased the risk of failure. In the lowest density bone, every additional screw reduced failure risk compared to the four-screw construct, whereas in more dense bone, once the sixth screw was inserted, no further significant benefit was seen (p=0.40). Screw configurations not including calcar screws, also demonstrated significant (p < 0 .001) increased risk of failure. Finally, more proximal plate positioning, compared to the suggested operative technique, was associated with reduced the predicted failure risk, especially in constructs using calcar screws, and distal positioning increased failure risk.

Optimal fixation constructs were found when placing screws 4 mm from the joint surface, in configurations including calcar screws, in plates located more proximally, as these factors were associated with the greatest reduction in predicted fixation failure in 3-part unstable proximal humeral fractures. These results may help to provide practical recommendations on the implant usage for improved primary implant stability and may lead to better healing outcomes for osteoporotic proximal fracture patients. Whilst prospective clinical confirmation is required, using this validated computational tool kit enables the discovery of findings otherwise hidden by the variation and prohibitive costs of appropriately powered biomechanical studies using human samples.