This study compared the LCP group and the FA group involving three- and four-part PHFs in patients aged 60 years or older and evaluated the influence of fibular allograft on the radiological and clinical outcomes. PHFs treated by LCP and fibular allograft showed significantly better clinical outcomes, and a lower complication rate. The FA group also showed superior radiological results regardless of the fracture type.
The treatment of unstable and displaced PHFs in elderly patients remains a challenge . Anatomical reduction is difficult to maintain because of the poor quality of the humeral head, and many surgeons believe that LCP fixation is a promising treatment for this problem . Compared to the standard nonlocking plate, the screw fixation angle can be oriented in different directions when an LCP is used, and the locking screw provides stable fixation maintenance . However, many studies have reported variable outcomes, with high rates of complications, including screw penetration, varus collapse, and avascular necrosis of the humeral head, especially in older patients with osteoporosis or medial column comminution [22,23]. Ockert et al.  reported 10-year outcomes after operative treatment with LCP for unstable and displaced PHFs, and the majority of the patients obtained good or excellent outcomes. However, poor outcomes and complications were found in older and female patients. In recent years, many efforts have made to overcome these problems. Clinical and biomechanical studies have paid increased attention to allograft augmentation to increase the stability of locking plate fixation in PHFs. Gardner et al.  were the first to introduce this technique using fibular strut allograft to indirectly reduce the fracture and maintain the fixation; in their study, all of the seven fractures achieved complete union without any loss of reduction.
Mathison et al.  first made a biomechanical comparison between locking plate alone and locking plate with fibular allograft. They created a 10-mm wedge-shaped osteotomy at the lever of the surgical neck to simulate the comminution of the medial column. Load-displacement curve was used to test failure load and stiffness of the constructs. Their study demonstrated that the bone peg increased the failure load and the initial stiffness of the constructs. Relative to locking plate fixation alone, failure load was increased by 1.72 times and the initial stiffness was increased by 3.84 times. Chow et al.  performed a similar study to evaluate the effect of a fibular allograft. No augmented construct collapsed before 25,000 cycles, while six of the eight specimens in the non-augmented locking compression group collapsed at an average of 6,604 cycles. Neviaser et al.  retrospectively reviewed 38 patients treated by locking plate with endosteal strut augment, and they reported high clinical outcome scores and a low rate of reduction loss (2.6%), screw cut-out (0%), and avascular necrosis (2.6%). Recently, Panchal et al.  assessed the effect of intramedullary fibular allograft on the clinical and radiological outcomes in unstable PHFs with medial column disruption. According to the clinical rating scale, 26 patients had excellent or good outcomes, six patients showed fair outcomes, and only four patients experienced poor outcomes. With regard to the restoration of the humeral NSA, the result was good in 31, fair in three, and poor in two cases. When calculating the HHH, the mean loss of reduction was measured as 1.6 mm. Only one case experienced varus collapse of the humeral head, and osteonecrosis was noted in one patient. Cha et al.  compared the radiological outcomes of using only LCP and using LCP with fibular allograft in the treatment of comminuted PHFs. In the LCP group, 22 of 32 patients had a change in the NSA of more than 5°, with an average of 10.2°. Twenty patients obtained a change in HHH of more than 3 mm, with an average of 4 mm. While in the LCP with fibular allograft group, the average NSA and HHH change was 3° and 1 mm, respectively. In our study, the FA group had significantly better CMS and ASES scores, as well as shoulder range of motion, compared to the LCP group. The change in the NSA and HHH in the LCP group was also markedly higher than that in the FA group.
We considered that the fibular allograft was a reasonable option to maintain the anatomical reduction in the treatment of comminuted PHFs in the elderly patients. The fibular allograft could be used as tool to indirectly reduce the fracture. Gardner et al.  first introduced the use of screw to push the fibular allograft medially so that the graft could indirectly reduce the fractured medial cortex. Subsequently, many authors preferred to push the graft upward instead of medial because it could support the humeral head in a proper height [15, 28]. However, in some cases, the intramedullary cavity in elderly patients accompanied osteoporosis was relatively large, the graft could not be easily manipulated. So we placed a guide pin at the apex of the humeral head. Then, the fibular allograft was pushed upward in the intramedullary cavity through the guide pin as in retrograded nailing. Especially in cases with medial cortex disruption, using fibular allograft as a pillar to support the humeral head from intramedullary cavity was more helpful in maintaining reduction. The added stability provided by the fibular allograft allowed for an early rehabilitation program and reduced the complication rate. In our study, the FA group showed significant lower rates of varus malunion and screw penetration. The fibular allograft also had disadvantages, such as risk of infection, disease transmission, and high cost. The fibular allograft contains cortical bone, so it might be fractured during insertion of the screws.
This study has several limitations. First, it was limited by its retrospective design, and the number of patients was relatively small. Further study with a greater number of patients is needed. Second, the follow-up duration was rather short, as the difference in NSA and HHH might change with longer monitoring. Third, the Neer classification is the most widely used grading system for PHFs, but some studies have shown that the Neer classification only have fair to good reliability.