The primary objective of this study was to determine whether the stability of a dedicated external fixation system for a three-part proximal humerus fracture provides sufficient stability for early postoperative rehabilitation. The external fixator limited the amount of fragment displacement at the lower physiological load level of 160 N to 1 mm or less. At higher load levels, fragment displacement exceeded 2 mm (240 N) or even resulted in fixation failure (320 N). Fixation with locking plates limited the amount of fragment displacement to less than 1 mm for both load levels and did not fail to stabilize the fracture up to the 320 N load level. These findings suggest that dedicated external fixation systems should be considered semi-rigid constructs with less load-bearing capacity compared to plate fixations. None of the fixations with the external fixator failed al 160N suggesting that external fixation provides sufficient stability for limited rehabilitation exercises during the first post-op weeks.
The fixator and plate differed in their response to loading concerning the different fracture displacements. For the displacements between the head fragment and shaft, the fixator allowed higher axial displacements compared to the plate. On the medial side, both systems had similar axial displacements. The rotation in the frontal plane was greater for the fixator with the head rotating in valgus, while the plate rotated slightly in varus. This difference may be due to the different pivot points of the two systems (see Fig. 11). The pivot point of the plate is on the lateral side, while the pivot point of the fixator is more medial, where the oblique pins connect the shaft to the head. The pivot point of the fixator also depends on the height of the fracture and the position/angle of the pins. This tendency for valgus displacement of the external fixator compared to varus displacement of the plate may have implications in clinical practice, because varus displacement is less tolerated than valgus displacement, as it increases the risk of screw cut-out and glenoid erosion.
For the displacements between the head and the greater tuberosity fragment, we found no differences between the two fixation systems, with both systems providing sufficient stabilization of the tuberosity.
Migration of external fixator pins is a common complication, particularly with thin wire pins [19–22]. In this study, pin migration was observed for both the fragment head pins and the shaft head pins. The shaft head pins migrated into the bone, whereas the fragment head pins migrated out of the bone. However, the pin migrations were small and did not result in penetration of any articular surface. In contrast to clinical observations [23, 24] we did not observe any association of fracture displacement or stability with bone mineral density. This may be related to our sample, which was restricted to specimens from elderly females with a limited range of bone mineral densities.
The mechanical performance of dedicated external fixation systems has previously been the subject of investigation in a similar experimental setup [2]. Although the results are not directly comparable due to the different test parameters (number of cycles, cycle frequency, fracture gap configuration) and failure criteria, it is possible to identify some common characteristics. Keeping in mind the different failure criteria, some of these results can be compared to our study. The mean load to failure of the fixator was higher in the Harbrecht study but also had twice the standard deviation. The number of samples failing before reaching the maximum cyclic load appears to be comparable between our study and that of Harbrecht [2]. For the fracture gap movement, only the medial displacements and varus/valgus rotation at 235 N (Harbrecht step 6 of cyclic loading) could be compared to our 240 N values. In the comparison of the results of the two studies, only the fixator rotation had significant differences. The other values did not reveal any significant differences. This may be due to the similarity of the results (especially with the plate) but also due to the small sample size and high standard deviations. The different results of fixator rotation could be related to different clamping angles (20° frontal vs. 25° frontal + 8° sagittal) or more likely due to different load applicators used. Harbrecht used a 1-axis rotatory applicator while we used a 2-axis linear applicator.
In another study, Castoldi et al. [25] evaluated different shaft-to-head pin configurations. They found that external fixation with four fully threaded wires provided the greatest stiffness and strength among the different constructs. They also found that this construct was mechanically similar to a locking plate. In their clinical experience, this has been able to reduce osteoporotic-related complications in comparison to standard percutaneous K-wire fixation [26].
Jabran et al. reviewed biomechanical studies of proximal humerus plates [15]. Most studies tested humeri without considering tendon or muscle forces in a two-part or three-part fracture in axial loading, bending, torsion, or a combination. Although the studies used various testing methods, the tendencies of lateral and medial fracture gap motion and head rotation were comparable to the findings from our study. The medial fracture gap stiffness was around 200 N/mm [27, 28].
The external fixator has the usual advantages and disadvantages such as reusable parts with possible cost reduction [29], ease of use, and minimal invasiveness. However, treatment success is highly dependent on patient compliance and acceptance [16, 30–33]. Pin hygiene can affect pin-tract infection and pin loosening. Patient comfort may be limited because the fixation is outside the body and there is a risk of getting stuck or hitting an obstacle. Nevertheless, this method often works well clinically [16, 26, 34]., With early passive motion starting in week 1 followed by active assisted motion in week 2, the recommended rehabilitation is much more conservative compared to locking plate fixation. Faster shoulder mobilization would be desirable for patients but would require greater mechanical rigidity of the fixation construct.
Limitations
Our study has the usual limitations of benchtop biomechanical studies. Soft tissues such as muscles and ligaments which may further stabilize the fracture were not considered. Fracture healing with callus formation accommodating greater loads in advanced healing stages could not be mimicked. The assumption that load increases linearly in cycles over 6 weeks is highly simplified and does not account for any unexpected peak loads.