Purpose: Optimal fixation for comminuted proximal humerus fractures is controversial. Complications using locked plates have been addressed by anatomic reduction or medial cortical support. The current study measured relative mechanical contributions of varus malalignment and medial cortical support.

Method: Forty synthetic humeri were divided into three groups, osteotomized, and fixed at 0, 10, and 20 degrees of varus malreduction with locked proximal humerus plates (AxSOS, Global model, Stryker, Mahwah, NJ, USA). This simulated mechanical medial support with the cortex intact. Axial, torsional, and shear stiffness were experimentally measured. Half of the specimens in each of the groups underwent a second osteotomy to create a segmental defect which simulated loss of medial support with the cortex removed. Axial, torsional, and shear stiffness experiments were repeated, followed by shear load to failure in 20 degrees of abduction.

Results: For isolated malreduction with the cortex intact, the repair construct at 0 degrees showed statistically equivalent or higher axial, torsional, and shear stiffness than other groups assessed. Subsequent removal of cortical support in half the specimens resulted in a drastic effect on axial, torsional, and shear stiffness at all varus angles. Repair constructs with the cortex intact at 0 and 10 degrees resulted in mean shear failure forces of 12965.4 N and 9341.1 N, respectively. These were statistically higher (p< 0.05) compared to most other groups tested. Specimens failed mainly by plate bending as the femoral head was pushed down medially and distally.

Conclusion: Anatomic reduction with the medial cortex intact was the stiffest construct after a simulated two-part fracture. This study also supports the practice of achieving medial cortical support by fixing proximal humeral fractures in varus if necessary. This may be preferable to fixing the fracture in anatomic alignment when there is a medial fracture gap.