Intramedullary nails are ideal load-sharing devices for stabilization of most impending and complete pathologic fractures of the humerus. Antegrade and retrograde interlocking humeral nails are widely available and technically simple to use. A major advantage of these devices is that the working length of the nail spans the entire bone, especially with diffuse disease (Fig. 27.3a, b). Plates can also be used to span the majority of humerus (Fig. 27.3c, d). However nails, unlike plate constructs, are more amenable to a minimally invasive approach, which is advantageous in situations where the additional risk associated with exposure of the lesion or fracture (blood loss, nerve injury, etc.) is not warranted. Tumor debulking and cement augmentation of bone defects can still be accomplished simultaneously using additional exposures along with nailing procedures.

General indications for intramedullary nail fixation include lesions or fractures located within 2–3 cm distal of the greater tuberosity to roughly 5 cm proximal of the olecranon fossa [17]. In addition to the proximal-distal location of the lesion, 4–5 cm of intact cortical bone on either side of the nail is required for rigid fixation [18]. Proximal or distal metaphyseal defects do not preclude the use of an intramedullary nail although plate fixation or cement augmentation should be considered in these instances. Tumor debulking and cement augmentation should also complement nail fixation of diaphyseal defects >2–3 cm. Proximal and distal interlocking screws should be utilized whenever possible, especially for complete fractures [18]. When using cement augmentation, cement can be added in a more viscous state and packed around the nail after insertion or in a less mature state after reaming and immediately before the definitive device is inserted.

In appropriately selected patients, outcomes after intramedullary fixation are favorable. Durable pain relief and return to activities of daily living can be expected in >90 % of patients. Reoperation rates are less than 5 % and most commonly associated with tumor progression and prominent proximal hardware [4, 17–20] (Fig. 27.4). One retrospective case–control study demonstrated earlier functional gains and pain improvement when intramedullary fixation was augmented with cement [20], although the necessity of cement augmentation with IM nail fixation remains controversial.

There are nonetheless various pitfalls and complications associated with intramedullary humeral nails. Shoulder pain and/or decreased shoulder abduction and forward flexion is observed in 10–15 % of patients likely secondary to rotator cuff injury during insertion or a prominent proximal nail position [21–24]. This can be lessened by meticulous protection of the supraspinatus tendon during reaming and nail insertion. Ensuring the proximal nail or proximal locking bolts are not left proud will also minimize postoperative shoulder issues. In a recent systematic review comparing plate osteosynthesis and intramedullary nail fixation for non-pathologic fractures of the humerus, nail fixation was associated with a greater incidence of shoulder impingement (21/123 cases, 17 %), decreased range of motion, and hardware removal (9/69 cases, 13 %) [25]. These results may not be generalizable to patients with pathologic fractures and impending fractures given the lower functional demands and life expectancy of these patients. Regardless, patients should be counseled of the risk of shoulder impingement preoperatively.

Postoperative radial nerve palsies are also associated with intramedullary fixation of the humerus, with an incidence of 3–6 % reported in the literature [4, 24, 26]. Cadaveric studies have demonstrated a 30 % incidence of lateral-medial distal locking bolt abutment with the radial nerve after humeral nailing [27]. Although more commonly encountered reported during the treatment of femoral metastases, embolic pulmonary complications are associated with intramedullary preparation and nail insertion into the humerus [28]. Nail insertion after cement injection adds an additional risk for embolic debris and therefore low-viscosity cement combined with attentive cardiopulmonary monitoring should be employed in these cases [28, 29].

Plate fixation of humeral metastases is less commonly utilized than intramedullary nailing, mostly because these procedures are often more invasive and do not always protect the entire bone. Plate and screw constructs are ideal for joint preserving reconstructions of lesions involving the proximal metaphysis/humeral head and distal humerus where intramedullary nail fixation is unlikely to provide adequate fixation in abnormal bone [30]. For these anatomic locations a preoperative CT scan is helpful to determine the extent of bone loss, aiding preoperative decision making between plate and screw or arthroplasty options. Plate fixation also affords direct exposure of the lesion for tumor debulking, avoids violation of the rotator cuff, and permits direct fracture reduction. Because of the limited working length of plate and screw constructs, judicious tumor debulking followed by cement augmentation should be considered in all cases. Cement augmentation provides additional mechanical stability and improves the pull-out strength of orthopedic screws inserted into abnormal bone [31, 32]. Both locking and non-locking screws can be placed across a mature cement mantle. Plate constructs should be cautiously used in cases with diffuse involvement of the bone, massive segmental cortical defects, and uncontained periarticular lesions with compromised articular integrity. Extensive disease involving the humeral diaphysis with extension into the distal metaphysis creates a challenging problem when deciding on the most appropriate implant given the challenges of obtaining distal fixation with intramedullary constructs (Fig. 27.3c, d).

For proximal lesions, a deltopectoral approach with a distal anterolateral extension provides adequate exposure while a triceps-sparring or -splitting posterior approach should be used for distal lesions. Distal lesions of the humerus have the highest incidence of mechanical failure and revision surgery (30 %); therefore dual plating with tumor debulking and cement augmentation is recommended to provide maximal stability and longevity [3]. Locking plates compared to non-locking fixation has been shown to provide superior screw fixation in the setting of poor bone quality, which has expanded the indications of these devices to include patients with metastatic bone lesions [30, 33, 34]. Contrary to this, satisfactory results using non-locking fixation and cement augmentation in the humerus are possible [5] and should not be abandoned, especially as government and hospital cost-containment pressures increase. In either setting, plates should span the defect by at least two cortical diameters, permit three bicortical screws on either side of the lesion, and, when possible, cover as much of the entire length of the bone permitted by the surgical approach [2, 5, 30]. Percutaneous fixation to limit surgical exposure can be used, when safe, in order to extend the length of the construct.

Like intramedullary fixation, outcomes after plate and screw fixation are favorable; pain relief can be expected in 85–95 % of patients, and the majority of patients will resume activities of daily living with the affected extremity [5, 18, 34, 35]. In patients surviving more than 1 year, revision surgery is required in about 10–15 % of patients for adverse events such as infection, mechanical failure, and tumor progression [3–5, 35]. In the context of humeral metastases, plate and screw reconstructions are associated with increased blood loss, longer operative times, and a higher incidence of iatrogenic radial nerve injuries compared to the results of IM nail fixation [18, 36, 37]. An iatrogenic radial nerve palsy , even if transient, can be a significant functional impairment in this patient population, especially when survival is limited. This limited data however should be interpreted with caution as high-quality, prospective, controlled studies directly comparing fixation techniques are lacking.

Endoprosthetic reconstructions of the proximal and distal humerus using modular tumor prostheses are valuable treatment options and should be considered when traditional internal fixation methods are unlikely to achieve durable stability and pain reduction. Indications for prosthetic reconstruction of the humerus include lesions of the humeral head with joint destruction and articular compromise, large segmental cortical defects, revision of failed intramedullary nail and/or plate and screw stabilizations, and defects of the distal humerus. In this context, proximal humerus resections are reconstructed using an endoprosthetic hemiarthroplasty [3, 38, 39] whereas distal humerus resections are coupled to a total elbow arthroplasty [3, 40, 41]. Because of pre- and postoperative radiation, systemic chemotherapy, and general poor bone quality, cemented fixation should be used whenever possible.

For proximal humerus reconstructions, a deltopectoral approach provides reliable access and visualization. Division of the rotator cuff insertion is frequently required and creative, although largely ineffective measures are often employed to reapproximate these tissues to the prosthesis. When possible, securing the native joint capsule around the prosthesis using a purse-string suture is thought to augment joint stability. Otherwise, a delicate balance of humeral head retroversion, head size selection, and rotator cuff tendon approximation are essential for long-term stability. Depending on the length of the bone resection needed, detachment of deltoid insertion is sometimes required. In these instances, the deltoid should be tenodesed to the pectoralis major tendon [38]. Deciding on whether to use a standard hemiarthroplasty implant, reverse shoulder or humeral megaprosthesis may depend on a number of factors including the amount of proximal bone loss, life expectancy, implant cost and access, and the potential for adequate soft tissue coverage and capture. To date, no literature has supported the use of one construct over another and shoulder stability can be adequately achieved with either. Proponents of a reverse total shoulder or allograft prosthetic composite argue improved shoulder function but the use of these somewhat more complicated reconstructions should be evaluated in the context of the patients’ overall condition.

The ultimate goal of a proximal humerus endoprosthetic reconstruction is to obtain a stable shoulder, providing a platform for independent elbow and hand function. Preservation of elbow and hand function and pain reduction are principal advantages of these reconstructions. Consequently, patient satisfaction is generally favorable with these procedures. However, because the rotator cuff insertion is sacrificed with these resections, suboptimal shoulder function is common postoperatively. Despite deltoid and axillary nerve preservation, resultant forward flexion and abduction are unlikely to exceed 90 degrees. [38, 39, 42]. Patients should be counseled that a reasonable postoperative expectation is for the ipsilateral hand to reach the mouth and face [39]. Proximal migration of the prosthesis or glenohumeral instability is observed in a 20–30 % of cases [38, 39]. Because of the inherent instability of the glenohumeral articulation, most centers advocate 4–6 weeks of restricted motion in a shoulder immobilizer to allow sufficient time for soft tissue healing. Because of rotator cuff deficiency and limited overhead mobility following standard endoprosthetic reconstructions of the proximal humerus, some authors have advocated using a reverse total shoulder arthroplasty (RTSA) [43, 44]. With these implants, the center of joint rotation is moved inferior and medial, which improves deltoid biomechanics and enables greater potential for abduction and forward flexion beyond 90 degrees. Intraoperative and postoperative complications are more common with RTSA as compared to primary shoulder arthroplasties [45]; however outcomes in metastatic patients are lacking and warrant further investigation.

Metastatic lesions of the distal humerus are relatively uncommon, although complications and revisions are proportionately more common in these cases [3]. Distal humeral resections coupled to a hinged or semi-constrained total elbow prosthesis facilitate complete tumor removal and rapid restoration of elbow function [40, 41]. A total elbow arthroplasty is often sufficient for smaller lesions of the trochlea and capitellum, where larger, more destructive lesions of the distal humeral metaphysis should be reconstructed with a modular endoprosthesis or allograft prosthesis composite (Fig. 27.5). A midline posterior approach to the elbow can be used for the majority of these cases. The ulnar nerve should be dissected and mobilized prior to exposure of the joint. Joint exposure can be accomplished by a variety of techniques such as the Bryan-Morrey posteromedial approach [46], working on either side of the triceps [47], an osteo-anconeus flap [48], and triceps-splitting approach [47], depending on local anatomy and surgeon preference.

With these procedures, patients can expect a substantial improvement in pain and elbow motion. Postoperative elbow motion in the sagittal plane is sufficient for most activities of daily living and coordinated positioning of the hand towards the mouth and face. A major limitation of total elbow arthroplasties is diminished lifting capabilities. Most surgeons advocate permanent lifting restrictions of 5–10 lbs. Early complications can be expected in 25–30 % of cases, the most common complication being iatrogenic injury to the ulnar nerve followed by infection [40, 41, 49]. Other causes of revision include triceps avulsion, local disease progression, and peri-prosthetic fractures. Implant instability is uncommon.

In situations where extensive diaphyseal bone loss is initially identified or can be expected after tumor debulking or resection, reconstruction using a cemented intercalary endoprosthesis may provide some appealing benefits (Fig. 27.6). The reconstruction allows for a limited exposure directly over the affected area of bone loss and intramedullary stem insertion. This mitigates the need for extensive exposure that may be required for long plate fixation or violation of the shoulder for proximal nail insertion. Early reports of these devices in the USA were complicated by a high rate of transient nerve palsies (likely secondary to distraction needed for implant coupling), peri-prosthetic fractures, and failure at the implant coupling interface [50]. Newer implant designs have mitigated some of these complications although aseptic loosening in one study was reported in 3/11 (27 %) patients [8]. In a separate report from Europe, the authors reported one case of aseptic loosening in eight patients at a mean follow-up of 29 months [51]. Based on these findings, the authors propose a narrow indication for these implants limited to patients with limited life expectancy and proximal or distal bone stock to allow for a minimum of 5 cm of intramedullary fixation [8].