To date, the fixation of proximal humeral fractures with angular stable locking plates is still insufficient with mechanical failure rates of 18% to 35%. The PHILOS plate (DePuy Synthes, Switzerland) is one of the most used implants. However, this plate has not been demonstrated to be optimal; the closely symmetric plate design and the largely heterogeneous bone mineral density (BMD) distribution of the humeral head suggest that the primary implant stability may be improved by optimizing the screw orientations. Finite element (FE) analysis allows testing of various implant configurations repeatedly to find the optimal design. The aim of this study was to evaluate whether computational optimization of the orientation of the PHILOS plate locking screws using a validated FE methodology can improve the predicted primary implant stability.

The FE models of nineteen low-density (humeral head BMD range: 73.5 – 139.5 mg/cm3) left proximal humeri of 10 male and 9 female elderly donors (mean ± SD age: 83 ± 8.8 years) were created from high-resolution peripheral computer tomography images (XtremeCT, Scanco Medical, Switzerland), using a previously developed and validated computational osteosynthesis framework. To simulate an unstable mal-reduced 3-part fracture (AO/OTA 11-B3.2), the samples were virtually osteotomized and fixed with the PHILOS plate, using six proximal screws (rows A, B and E) according to the surgical guide. Three physiological loading modes with forces taken from musculoskeletal models (AnyBody, AnyBody Technology A/S, Denmark) were applied. The FE analyses were performed with Abaqus/Standard (Simulia, USA). The average principal compressive strain was evaluated in cylindrical bone regions around the screw tips; since this parameter was shown to be correlated with the experimental number of cycles to screw cut-out failure (R2 = 0.90). In a parametric analysis, the orientation of each of the six proximal screws was varied by steps of 5 in a 5×5 grid, while keeping the screw head positions constant. Unfeasible configurations were discarded. 5280 simulations were performed by repeating the procedure for each sample and loading case. The best screw configuration was defined as the one achieving the largest overall reduction in peri-screw bone strain in comparison with the PHILOS plate.

With the final optimized configuration, the angle of each screw could be improved, exhibiting significantly smaller average bone strain around the screw tips (range of reduction: 0.4% – 38.3%, mean ± SD: 18.49% ± 9.56%).

The used simulation approach may help to improve the fixation of complex proximal humerus fractures, especially for the target populations of patients at high risk of failure.