Proximal humerus fractures (PHFs) remain challenging to the treating orthopedist in both their initial diagnosis and classification and in their corresponding management. They account for approximately 5% of all fractures, and their incidence is expected to increase secondary to an aging population and the associated osteoporosis. PHFs are the third most common type of fragility fracture after distal radius and hip fractures in patients aged 65 years and older. In general, PHFs occur more frequently in females than males (3 : 1 ratio). Nonoperative versus operative management of these injuries depends on the mechanism of injury, the patient’s physiologic age, including activity level, and fracture pattern. Although the incidence of PHFs in the elderly did not change from 1999 to 2005, the rate of surgical treatment has increased significantly. In a study evaluating the trends and variance in the incidence and surgical treatment of PHFs in the elderly, Bell et al. noted marked regional variation in the rates of surgical treatment, highlighting the need for better consensus regarding optimal treatment and additional research to determine which fractures are best treated operatively.
Nearly three-fourths of all PHFs occur in patients older than 60 years and are generally the result of low-energy trauma, such as a fall from standing height. The majority of these injuries are nondisplaced or minimally displaced and have a good overall prognosis with nonsurgical management despite short-term impairment. Specific risk factors associated with PHFs in the elderly include low bone density, impaired vision and balance, lack of hormone replacement therapy, previous fracture, three or more chronic illnesses, and smoking.
In contrast, younger patients generally sustain PHFs following high-energy trauma such as motor vehicle collisions, seizures, or electrical shock. These injuries tend to be more severe regarding soft tissue compromise, neurovascular injury, and fracture displacement and often require operative intervention.
There remains a wide variation in treatment recommendations for PHFs, with low-quality evidence to support one modality over another. This issue may be partially attributed to the difficulty with radiographic interpretation and the subsequent classification of these injuries.
To appropriately manage PHFs, it is crucial to understand the complex anatomy of the shoulder girdle. Stability and function of the glenohumeral joint is provided by the interaction of mechanisms that promote a near global range of motion (ROM) and purposeful function. External loads that are transferred to the shoulder girdle are initially offset by joint surface anatomy, joint volume, atmospheric pressure, and joint fluid cohesion and adhesion. Moderate and large loads are counterbalanced by the deltoid and rotator cuff and by the capsulolabral and bone structures, respectively. PHFs alter these complex interactions, resulting in pain, decreased ROM and stiffness, and disability.
The proximal humeral anatomy comprises four main parts: the humeral head, greater tuberosity (GT), lesser tuberosity (LT), and humeral shaft. Following a fracture of the proximal humerus, displacement of each “part” occurs in a predictable manner and is based on the deforming forces created by the tendinous insertions of the pectoralis major, subscapularis, supraspinatus, and infraspinatus ( Fig. 6-1 ).
The normal dimensions and relationships of the bony anatomy of the proximal humerus have been extensively studied. The articular head is spherical and has a diameter of 37 to 57 mm. The most superior portion of the articular surface of the humeral head averages 8 mm above the GT, and the humeral version averages 29.8 degrees (range, 10 to 55 degrees). The head is inclined approximately 130 degrees with respect to the humeral shaft.
The bicipital groove lies between GT and LT, and serves as a pathway for the long head of the biceps as it traverses from its intraarticular origin (i.e., superior glenoid–labral complex) into the proximal arm. The distal aspect of the groove is internally rotated with respect to the proximal portion.
The anatomic neck of the proximal humerus is located at the junction of the articular surface and the tuberosities. The surgical neck represents an indistinct region (i.e., meta-diaphyseal junction) below the tuberosities but above the humeral shaft. There are prognostic implications regarding the precise location of a PHF. For example, a fracture involving the anatomic neck is prognostically worse than fractures involving other regions of the proximal humerus with respect to the potential disruption of the vascular supply to the humeral head and the subsequent development of avascular necrosis (AVN; Fig. 6-2 ).
The GT, which is located in a posterior-superior location with respect to the humeral shaft, serves as the attachment site for the supraspinatus, infraspinatus, and teres minor tendons of the rotator cuff. The LT, located on the anterior aspect of the proximal humerus, serves as the attachment site for the subscapularis tendon.
The glenoid is a convex structure of shallow depth shaped like an inverted pear. It articulates with the humeral head, and serves as the attachment for the labrum and joint capsule.
The acromion, the coracoacromial ligament, and the coracoid process form the coracoacromial arch, a rigid bony-ligamentous structure that imparts stability to the shoulder girdle. The rotator cuff, subacromial bursa, and subdeltoid bursa pass underneath the coracoacromial arch. Displaced PHFs can impede normal movement of these structures, causing impingement and disruption of normal glenohumeral motion. In PHFs (displaced and nondisplaced fractures), the subdeltoid and subacromial bursae can become thickened and fibrotic, forming adhesions that can similarly limit normal glenohumeral motion. Early ROM exercises after a fracture have been hypothesized to decrease the formation of such adhesions.
The proximal humerus receives its blood supply from the anterior and posterior humeral circumflex branches from the third division of the axillary artery. The posterior humeral circumflex artery travels with the axillary nerve, enters the quadrilateral space posteriorly, and anastomoses with a branch of the anterior circumflex to supply the posterior cuff. The anterior humeral circumflex artery (AHCA) arises from the axillary artery at the inferior border of the subscapularis, and provides vascular inflow to the humeral head by way of its terminal anterolateral branch known as the artery of Laing (also known as the arcuate artery ). The ascending branch of the AHCA courses parallel to the lateral aspect of the long head biceps tendon, and enters the humeral head at the interface of the bicipital groove and GT ( Fig. 6-3A ). Injury to the arcuate artery may result in osteonecrosis of the humeral head ; however, additional extraosseous collateral branches can permit humeral head perfusion despite complete ligation of the arcuate artery. A recent anatomic study has demonstrated that the posterior humeral circumflex artery provides nearly 64% of the blood supply to the humeral head, which may in part explain the relatively low rates of AVN following displaced PHFs ( Fig. 6-3B ).
Vascular injuries following PHFs are uncommon, occurring in 5% to 6% of cases. When vascular injuries do occur, they typically involve the axillary artery at the level of the surgical neck just proximal to the trifurcation of the anterior and posterior circumflex and subscapular arteries (i.e., the third part of the axillary artery). Most axillary artery injuries occur in patients older than 50 years of age, suggesting that comorbid conditions (e.g., arteriosclerosis) may play a role in this increased incidence. A careful neurologic examination must be performed in addition to the vascular assessment secondary to a high correlation between axillary artery and brachial plexus injuries.
The brachial plexus (C5-T1 nerve roots, with small contributions from the C3 and C4 nerve roots) provides innervation to the shoulder complex. The most commonly injured nerve following a PHF is the axillary nerve, and it is most susceptible to injury during anterior fracture-dislocations. The axillary nerve arises from the posterior cord of the brachial plexus (C5 and C6 nerve roots), and travels posterior to the surgical neck through the quadrilateral space along with the posterior humeral circumflex artery. The posterior trunk of the nerve innervates the posterior deltoid and teres minor muscles; a terminal branch of the posterior trunk forms the superior lateral cutaneous nerve of the arm that provides sensation to the overlying deltoid region. The anterior trunk of the nerve continues along the deep surface of the deltoid muscle, and provides branches to the middle and anterior deltoid heads (see Fig. 6-2 ). Lastly, the axillary nerve supplies an articular branch. Previous anatomic studies documenting the course of the axillary nerve have demonstrated that the mean distance from the proximal aspect of the humerus to the nerve is 6.1 cm (range, 4.5 to 6.9 cm), and the mean distance from the surgical neck of the humerus to the nerve is 1.7 cm (range, 0.7 to 4.0 cm).
The suprascapular nerve, the second most commonly injured nerve, originates from the upper trunk (C5 and C6 nerve roots), and provides motor innervation to the supraspinatus and infraspinatus muscles. This nerve is most susceptible to tractional injury at two locations: the upper trunk where the nerve originates, and the suprascapular notch where the nerve passes below the transverse scapular ligament.
The musculocutaneous nerve originates from the lateral cord of the plexus (C5-C7 nerve roots), passes through the conjoint tendon at a variable distance from the coracoid process (3.1 to 8.2 cm), and terminates at the lateral antebrachial cutaneous nerve, supplying sensation to the anterolateral forearm. Injury to the musculocutaneous nerve is a rare event, but can occur with blunt trauma and traction injuries to the shoulder, resulting in the loss of active elbow flexion.
Electromyographic studies should be completed between 3 and 6 weeks if clinical recovery is not apparent by physical examination. Documentation of motor denervation without improvement on additional studies may warrant surgical exploration for complete injuries at 3 to 6 months. Early consultation with a peripheral nerve surgeon is paramount in the proper and timely management of such injuries. Hems and Mahmood retrospectively evaluated 100 patients with injuries to the infraclavicular brachial plexus and noted axillary nerve injury in 15 patients (i.e., 15%) with a PHF due to medial displacement of the humeral shaft. They concluded that specific injury patterns increase the likelihood of nerve injury and that surgical neck fractures require urgent surgical management to decompress the terminal branches of the plexus.
Mechanism of Injury
The most common cause of PHFs is a fall on an outstretched hand from standing height (or less), particularly in patients older than 60 years of age. * Less commonly, high-energy trauma such as a motor vehicle accident, fall from a height, and seizure; these injury mechanisms are more typical in younger patients. These injuries tend to be more severe with regard to the fracture pattern and soft tissue compromise. In the setting of metastatic bone disease or primary malignancy, a pathologic fracture can occur with minimal trauma.
* References .
A thorough history and physical examination should always be performed on a patient with a PHF. The history should include patient age, hand dominance, occupation, the mechanism of injury, concomitant injuries, history of malignancy, premorbid level of function, and the ability to participate in a structured rehabilitation program. A review of systems should include queries regarding loss of consciousness, neck pain, ipsilateral elbow or wrist pain, and paresthesias of the hand and upper limb. On physical examination, the surgeon should look for swelling, ecchymosis, soft tissue injuries, and gross deformity. Most patients present with guarding, holding the arm in internal rotation (i.e., forearm overlying the abdomen). Any attempt at active or passive movement elicits significant pain. Palpation of the shoulder reveals crepitus. A careful neurovascular assessment should include evaluation of the brachial plexus, axillary nerve, and axillary artery. Posterior fracture-dislocations can demonstrate flattening of the anterior aspect of the shoulder, with an associated posterior prominence; anterior fracture-dislocations manifest with the opposite findings. The surgeon should assume a vascular injury, even in the presence of a benign examination, in four-part PHFs with axillary dislocation of the humeral head.
Radiographic evaluation of the shoulder is often difficult due to patient discomfort, but is critical for treatment decision-making. The shoulder trauma series includes a true anteroposterior (AP) of the glenoid (Grashey view), an axillary lateral view, and a scapular Y lateral view (also known as the transscapular lateral view ).
The true AP view may be taken with the patient’s arm in a sling and with the patient standing, seated, or prone (i.e., the most comfortable position). To obtain this view, the unaffected shoulder must be rotated away 30 to 40 degrees, thus allowing the injured side to rest upon the x-ray plate ( Fig. 6-4A ). Additional AP views (e.g., internal rotation and external rotation AP views) can be obtained to better reveal LT or GT fractures, respectively.
The scapular Y lateral (transscapular lateral) view is obtained by placing the anterior aspect of the affected shoulder against the x-ray plate, with the unaffected side rotated out 40 degrees ( Fig. 6-4B ). This view is essential for assessing associated fractures of the scapular spine and acromion, acromioclavicular (AC) joint injuries (e.g., type IV injuries), and posterior-superior displacement of GT fractures.
The axillary view is essential in evaluating the relationship of the humeral head to the glenoid, displacement of the GT, and the glenoid articular surface. This view can be obtained in the seated, standing, or prone positions. The arm is gently held in 30 degrees of abduction, with an x-ray plate placed on the posterosuperior aspect of the shoulder. The x-ray tube is directed cephalad from a level slightly below the plane of the patient. Due to the nature of the injury, the patient may be unable or unwilling to participate in the abducted axillary view. The Velpeau axillary view is an excellent alternative and is preferred in the trauma setting ( Fig. 6-4C ; also see Fig. 6-8B ). This view is obtained by directing the x-ray beam from superior to inferior as the patient leans backward over the x-ray cassette with his or her arm in a sling.
Magnetic resonance imaging (MRI) is rarely indicated in the setting of an acute injury. If a pathologic fracture is a possibility, then MRI may be useful for staging of the disease. Computed tomography (CT) analysis may be helpful in evaluating tuberosity displacement, the degree of comminution, and glenoid articular surface involvement. This imaging modality improves the surgeon’s ability to make treatment decisions and aids in preoperative surgical planning. A recent review by Berkes et al. did not demonstrate improved interobserver and intraobserver reliability when comparing two- (2D) and three-dimensional (3D) CT imaging with regard to classification and treatment decision-making except among surgeons with limited experience.
A functional classification scheme should be easy to use, reproducible (i.e., both interobserver and intraobserver reliability), and direct appropriate management. The system must be comprehensive enough to include all variables (patient and fracture variables), but at the same time be specific enough to allow accurate diagnosis and guide treatment.
The most commonly used system is Neer’s four-part fracture classification, which was initially reported in 1970. Prior to Neer’s system, several others had attempted to classify these injuries according to the fracture location (Kocher [ Fig. 6-5 ]), fracture pattern (Codman [ Fig. 6-6 ], De Anquin and De Anquin, and De Palma and Cautilli), and injury mechanism (Watson-Jones, Dehne).
Jakob and Ganz from the AO (Arbeitsgemeinschaft für Osteosynthesefragen) group proposed a classification scheme involving 27 subgroups based on articular involvement, location, and the degree of comminution and dislocation, with special emphasis on the integrity of the vascular supply. This system helped distinguish valgus-impacted four-part PHFs from other four-part injuries with partial preservation of the vascular inflow to the articular segment through the medial capsule. The complexity of this system has generally precluded its routine use, despite similar interobserver and intraobserver reliability when compared with Neer’s classification system.
The interobserver and intraobserver reliability of the Neer classification system has been previously challenged; however, it remains the most reliable and commonly used system for assessing PHFs. The classification scheme is based on four segments or parts: the humeral shaft, articular surface, GT, and LT, with emphasis on identifying fracture patterns that are associated with vascular injury to the humeral head ( Fig. 6-7 ). Displacement occurs when a segment or part (articular surface, GT or LT, shaft) has separated more than 1 cm or angulated more than 45 degrees. If this criterion is not met, the fracture is considered minimally displaced, although multiple segments may be involved.
Neer additionally focused on fracture-dislocations of the proximal humerus, head-splitting fractures, and impression fractures of the humeral head. Fracture-dislocations are classified according to the direction of the humeral head dislocation (anterior or posterior) in conjunction with additional displacement of the fracture segments. Head-splitting and impression fractures have specific features of articular surface involvement and have been subdivided according to the percentage of involvement (<20%, 25% to 40%, and >45%) to direct treatment. Classification of these injuries depends on proper radiographs (i.e., shoulder trauma series) and adequate knowledge of the proximal humerus anatomy, including the rotator cuff insertions.
Some authors have attempted to improve the reliability of the Neer classification with CT imaging, but no evidence of decreased variability has been demonstrated with this modality. The same authors have concluded that classification of PHFs remains a difficult task, with reproducibility depending on the surgeon’s experience and level of expertise. Shrader et al. assert that the problem lies not with an understanding of the classification scheme but with the comprehension of the complex imaging. Similarly, Bruinsma et al. found poor interobserver reliability with the use of 3D CT analysis of PHFs among shoulder and elbow specialists and orthopedic traumatologists.
Hertel et al. modified Codman’s original classification to evaluate the predictors of humeral head ischemia after intracapsular fractures of the proximal humerus. According to the authors, the most relevant predictors of ischemia were the length of the dorsomedial metaphyseal extension, integrity of the medial hinge, and basic fracture pattern. The combination of an anatomic neck fracture, short calcar segment (<8 mm of metaphyseal extension), and disrupted medial hinge demonstrated a positive predictive value of 97% when determining humeral head ischemia. Fracture displacement proved less important with regard to humeral head viability. These earlier findings have since been reevaluated by Bastian and Hertel and questioned by Campochiaro et al. Hertel reported on the development of AVN in 51 patients with intracapsular fractures of the proximal humerus. Of the 10 humeral heads that were initially deemed “ischemic,” only two developed collapse at the last follow-up, indicating the possibility of revascularization. In contrast, four of the 30 well-perfused humeral heads developed structural changes indicative of AVN. Campochiaro et al. reported a 3.4% rate of AVN in 267 PHFs that were treated with plate and screw fixation, despite “poor quality of reduction” in 50% of cases. The authors concluded that “Hertel’s criteria are important in the surgical planning but are not sufficient.” These authors also concluded that 3D CT analysis of the medial calcar should be performed to more accurately apply Hertel’s criteria.
In a recent study evaluating the reliability of classification systems, Majed et al. determined that the Codman-Hertel classification scheme achieved the highest interobserver score, followed by the Neer and AO/Orthopaedic Trauma Association classification systems. Despite these findings, the Neer classification remains the most commonly used system by orthopedic surgeons for treating PHFs.
Methods of Treatment
The vast majority of PHFs are minimally displaced or angulated, and do not require surgical intervention. Early protection with gradual mobilization is the guiding principle with nondisplaced injuries. The patient generally wears a sling with an axillary pad for comfort and to prevent skin irritation. The patient is encouraged to perform active finger, hand, wrist, and elbow exercises. The sling is to be used for comfort; most patients are able to wean from the sling between weeks 3 and 6 (often dependent on fracture pattern and patient factors). By 7 to 10 days, gentle passive and active-assisted ROM and pendulum exercises may be initiated as tolerated under the supervision of a physical therapist. Delay of motion beyond 2 weeks has deleterious effects on shoulder ROM, pain, and function. Serial biplanar radiographs should be obtained to monitor possible interval fracture displacement or angulation. ROM gains are generally noted between weeks 3 and 8. By 6 weeks, patients may begin to use the affected arm for activities of daily living (ADLs) as tolerated to improve muscular strength and endurance. Formal strengthening exercises are initiated once there is radiographic evidence of progressive fracture healing and consolidation and when the ROM is approximately 75% to 80% that of the contralateral side.
Approximately 20% of PHFs are comminuted or displaced, and require surgical intervention by means of closed reduction with percutaneous fixation, open reduction and internal fixation (ORIF), or shoulder arthroplasty (i.e., hemiarthroplasty or reverse total shoulder arthroplasty [RTSA]). When deciding whether to pursue operative treatment of a PHF, one must collectively consider both the patient (premorbid health status, physiologic age, activity level, and ability to comply with postoperative rehabilitation) and fracture-specific characteristics (severity of the fracture including the degree of displacement and comminution, bone quality, and rotator cuff status). Patients with low demands and significant medical comorbidities (e.g., dementia) should be managed nonoperatively. The ideal goal of treatment, whether operative or nonoperative, is to reestablish a functional pain-free ROM. It is important to note that most of the published treatment algorithms for PHFs are generally based on case series (level IV evidence) in which a single treatment option has been isolated and assessed for efficacy.
Treatment Considerations According to Fracture Subtype
Greater Tuberosity Fractures
Isolated displaced fractures of GT account for a small percentage of PHFs. Chun et al., in their review of two-part PHFs, reported that 26 of 141 (18%) were displaced GT fractures. As an uncommon event, these injuries may be overlooked or trivialized by orthopedic surgeons and other healthcare providers ( Fig. 6-8 ).
Nature of the Injury
The amount of tuberosity displacement guides the decision-making regarding surgical intervention. In view of Neer’s criteria regarding displacement, most authors agree that the shoulder has little tolerance for GT displacement. McLaughlin suggested that displacement of 5 mm or more can cause impingement and rotator cuff dysfunction. Furthermore, Park et al. have suggested that fractures in manual laborers or athletes with as little as 3 mm of displacement should be surgically reduced. The direction of displacement of the tuberosity is just as important as the amount of displacement. Posterior displacement can affect external rotation, but is generally better tolerated than superior displacement, which results in subacromial impingement and weakness with active elevation.
Different mechanisms of injury have been proposed for GT fractures, including impaction (direct fall onto the shoulder or a fall on an outstretched hand forcing the shoulder into hyperabduction, causing compression of GT against the acromion; see Fig. 6-8 ) or avulsion and shearing (anterior glenohumeral joint dislocation causing the tuberosity to shear across the glenoid rim; Fig. 6-9 ), although the pathomechanics have not been clearly elucidated.
Peripheral nerve injury is the most commonly associated injury, occurring in nearly 33% of patients following a shoulder fracture-dislocation. Nearly 50% of patients older than 50 years of age with an anterior shoulder fracture-dislocation have been found to have a peripheral nerve injury. Isolated axillary nerve injuries are the most common nerve injury associated with two-part GT fracture-dislocations.
The initial evaluation of a patient with an injured shoulder has been previously discussed including a detailed history and physical examination followed by plain radiographs (i.e., shoulder trauma series). Advanced imaging such as CT and/or MRI are not routinely used; however, 3D CT may be utilized to delineate GT displacement ( Fig. 6-10 ).
Nonoperative treatment is reserved for nondisplaced and minimally displaced GT fractures and for low-demand elderly patients. This treatment includes a brief period of sling immobilization for 1 to 2 weeks. Passive ROM exercises are started as early as 7 to 10 days postinjury, starting with pendulum exercises and advancing to stick exercises below the horizontal plane for the first 4 weeks. Passive ROM exercises in all planes as tolerated are encouraged after 4 weeks when there is radiographic evidence of fracture healing and stability without interval displacement. By 6 weeks, patients are permitted to begin active-assisted and active ROM exercises in the form of ADLs, and progressive strengthening exercises may be initiated once there is radiographic evidence of progressive fracture healing and the patient has regained near-normal motion ( Fig. 6-11 ).
In 2013, Rath et al. reported the short-term clinical outcomes of 69 patients following nonoperative treatment of minimally displaced GT fractures (<3 mm). The average Constant score improved from 40 points to 95 points. Of interest, the mean duration of pain relief and improvement in ROM from the time of injury was 8.1 months. Hébert-Davies et al. recently reported a 26% incidence of GT fracture migration in patients younger than 70 years of age when the injury resulted from a fracture-dislocation. Displacement of the GT was 5.6 times more likely with dislocation than without. This reinforces the need for early radiographic surveillance and patient counseling that surgery may be required even when the initial postreduction radiographs reveal a near-anatomic fracture position.
Operative intervention is recommended for GT fractures with more than 5 mm of displacement to prevent fracture malunion and subsequent pain, stiffness, and rotator cuff dysfunction. Patients are placed in the semisitting (“beach chair”) position under general anesthesia. An interscalene nerve block may be considered for postoperative pain control, which is performed in the recovery room after a neurovascular examination has been performed on the operative extremity. An arm-positioning device attached to the beach chair may be used to support the operative limb; alternatively, a padded Mayo stand may be used. A C-arm image intensifier is used to obtain AP and axillary views ( Fig. 6-12 ). The C-arm may be positioned from the head of the bed or from the opposite side of the bed (i.e., surgeon preference) to ensure that proper orthogonal views are obtained.
Surgical fixation of displaced GT fractures may be performed using either open, arthroscopic, or arthroscopic-assisted techniques. Regardless of the surgical technique used, the principles are the same for fracture mobilization and reduction, secure fixation, and repair of concomitant pathology such as rotator cuff tears to provide an environment that will permit early ROM. Surgical exposure for open techniques can proceed by way of a superior deltoid-splitting or deltopectoral approach. Some authors have advocated the use of the deltoid-splitting approach due to the relative ease of visualization of the GT fragment. Fracture displacement must be carefully reviewed on preoperative imaging to ensure that the deltoid split is properly positioned to permit adequate mobilization, reduction, and fixation of the GT fragment. If necessary, an acromioplasty can also be performed with this technique to improve surgical exposure; however, this is rarely required. The deltopectoral approach avoids detachment of the deltoid, and allows exposure of the surgical neck when there is a concomitant proximal humeral shaft fracture; however, this surgical approach should be used with caution in isolated two-part GT fractures, especially when there is posterior fracture displacement.
Multiple open techniques for isolated GT fixation have been described and utilize either all-suture or all-screw and/or plate constructs depending on the degree of comminution present. Traditionally, tension banding and cancellous screw fixation techniques have been used for displaced, noncomminuted GT fractures. Recent biomechanical evidence reveals that tension band constructs provide significantly higher load to failure when compared with two parallel cancellous screws used alone. Isolated screw fixation is also not recommended because the tuberosity can theoretically fragment or displace around the screw. Screws may be used as a post when placed distal to the fracture when using sutures for tension band fixation. All-suture techniques have also been described using heavy (e.g., No. 5) nonabsorbable sutures placed at the rotator cuff tendon-bone interface, proximally, and through bone tunnels, distally (i.e., transosseous suture construct). Suture constructs used in isolation may fail by suture cutout. More recently, a hybrid fixation construct has been described for noncomminuted and comminuted GT fractures that uses a combination of cancellous screws and suture anchors that are placed distal to the fracture to act as a secondary tension band ( Fig. 6-13 ). Together, each individual construct neutralizes displacing forces created by the other construct and prevents fracture malreduction and subsequent malunion. From a biomechanical standpoint, this hybrid construct may offer fracture stability that theoretically exceeds any single construct and permits early rehabilitation.
Some authors have advocated percutaneous reduction and fixation to minimize soft tissue dissection, although fixation with a single screw or pin remains biomechanically inferior.
Arthroscopic techniques have been described more recently. They have the additional advantage of addressing concomitant pathology such as rotator cuff and labral tears that may be present in patients with an anterior fracture-dislocation. The GT fracture fragment can be visualized and reduced using a grasper device and internally stabilized with pins or screws. Some authors have advocated the use of a suture bridge and double row techniques solely through arthroscopic means with mixed results. There are limitations to this technique, specifically in the following settings: (1) subacute presentation and severely displaced GT fractures may preclude reduction due to early fracture healing and soft tissue contracture; (2) large GT fracture fragments may be difficult to visualize and reduce; (3) all-arthroscopic constructs may be technically challenging for those surgeons without specialized training in arthroscopy. Yin et al. have appropriately emphasized that the surgeon must comprehend the “personality” of the fracture pattern, as this often dictates the surgical approach (open vs. arthroscopic) and method of fixation.
Postoperative rehabilitation begins the day after surgery, with pendulum, passive forward elevation to the horizontal plane, and adducted external rotation exercises (i.e., stick exercises). Between 4 and 6 weeks postsurgery, passive ROM exercises are permitted in all planes as tolerated. Internal rotation and adduction maneuvers are avoided until 6 weeks postsurgery. After 6 weeks, active ROM exercises are initiated in the form of ADLs, with concurrent passive stretching exercises in all planes as tolerated. Rotator cuff–strengthening exercises are postponed until 10 to 12 weeks postsurgery when there is radiographic evidence of fracture healing. All patients need to be informed that clinical improvement may not be maximized until 1 year postsurgery.
Stiffness, malunion, and nonunion remain the most common complications that can occur after nonoperative or operative treatment of GT fractures. Shoulder stiffness may be treated early with a dedicated passive stretching program; however, the patient should be informed of the possible need to perform an arthroscopic capsular release with or without an acromioplasty to address concomitant impingement.
Osteotomy and mobilization of a malunited GT fragment has unpredictable results, , often requiring a rotator cuff interval slide(s) to reduce the fragment and the rotator cuff. Fixation of the mobilized tuberosity fragment can prove tenuous due to significant osteopenia. In the setting of nonunion, the tuberosity may be markedly displaced, making mobilization and repair difficult secondary to scarring.
There are limited published studies reporting the surgical results of displaced GT fractures. Paavolainen et al. reported good results in the surgical treatment of six displaced fractures with screw fixation. Chun et al. treated 10 GT fractures with ORIF (eight with screws). At a mean follow-up of 5.1 years, the results were graded as excellent in one patient, good in seven, and fair in three according to Neer’s criteria. The mean active forward elevation and external rotation were 118 and 35 degrees, respectively. Flatow et al. reported their experience with surgical fixation of 16 displaced GT fractures using heavy nonabsorbable sutures. At 4.5 years of follow-up, the authors reported six excellent and six good results according to Neer’s criteria. Park et al. reported 78% “excellent” and 11% “good” results in 13 fractures treated with suture fixation with a mean follow-up of nearly 4.5 years.
Lesser Tuberosity Fractures
Isolated LT fractures without an associated posterior shoulder dislocation or surgical neck fracture are rare. Because of the subscapularis insertion, the LT fragment will displace medially in the event of a fracture. The mechanisms of injury include sudden resisted abduction and external rotation with eccentric loading of the subscapularis and extension and external rotation of an axially loaded humerus. Diagnosis of these injuries remains challenging, often leading to delayed treatment. Patients may have soft tissue swelling and ecchymosis in the axilla associated with internal rotation weakness, evident with a positive Napoleon sign and modified belly-press test. Plain radiographs may not reveal the LT fracture. Supplementary imaging in the form of a CT scan may confirm the surgeon’s clinical suspicion. If the fragment is small, minimally displaced, and does not block internal rotation, a short period of immobilization in slight external rotation (to approximate the fragment to the donor site) followed by early passive and active-assisted ROM exercises is appropriate ( Fig. 6-14 ). Unfortunately, there are no fracture displacement guidelines as there are for GT fractures to assist the surgeon in treatment decision-making for LT fractures. In 2009, Robinson et al. reported the surgical results of isolated displaced LT fractures in 17 patients. The estimated overall incidence reported in this study was low (0.46 per 100,000). Surgical management routinely resulted in restored shoulder function, with a low rate of complications.
More commonly, isolated LT fractures result from a posterior fracture-dislocation of the shoulder. Surgical intervention is recommended for displaced large tuberosity fragments and for those with articular surface involvement. The tuberosity may be fixed either anatomically or, if present, into the base of the humeral head defect (reverse Hill-Sachs lesion) if the shoulder is unstable after open reduction (with or without cancellous bone graft to backfill the defect). In 2015, Liu et al. reported their results in 22 of 29 patients available for follow-up who underwent ORIF of an LT fracture associated with a locked posterior dislocation. The mean patient age was 41.7 years, and the majority were male (21 of 22). The length of time until the initial surgery was the only prognostic factor to correlate with an inferior outcome; 9 of 22 patients (41%) in this study were misdiagnosed at the time of their initial presentation, leading to a delay in surgical treatment.
Open reduction may proceed using a deltopectoral approach with the LT fragment stabilized with screws, sutures, and/or suture anchor fixation, or a combination of constructs (i.e., hybrid fixation) ( Fig. 6-15 ). One must assess the medial wall of the bicipital groove and possible associated pathology involving the long head biceps tendon (e.g., tears or medial subluxation of the tendon caused by disruption of the biceps pulley); a biceps tenodesis may be required, and should be discussed with the patient preoperatively. Arthroscopic fixation of a displaced LT fragment has been described by Scheibel et al. utilizing a standard posterior viewing portal, anterior and anterosuperior portals, suture relay, and suture anchor fixation of the tuberosity fragment. Rotator interval closure should be considered in cases of persistent posterior subluxation after the primary fracture fixation is completed; during interval closure, the arm should be held in a neutral position to avoid excessive loss of external rotation postoperatively.
Surgical Neck Fractures
Surgical neck fractures account for the majority (60% to 65%) of PHFs. Nearly 80% of these fractures are minimally displaced and warrant nonoperative management. Surgical indications include fracture displacement, polytrauma, ipsilateral upper extremity injuries, vascular compromise, open fractures, and patient compliance with a postoperative rehabilitation program. Malunion can be tolerated if the tuberosity and articular surface relationships are not distorted.
As Iannotti et al. have suggested, there are two distinct patient populations: young, male patients with high-energy trauma, and elderly, female patients with low-energy trauma. These should be treated differently, even for patients with the same fracture pattern. The young patient typically has good bone quality and can generally comply with postoperative therapy, allowing the surgeon to employ surgical methods such as ORIF with a plate or intramedullary (IM) nail fixation. The elderly patient and patients with dementia may not fully comprehend or appreciate his or her role in the postoperative rehabilitation process. Bone quality in elderly patients tends to be poor, with evidence of comminution after minor trauma; fixation options may be limited, and in some cases of surgical neck nonunion, hemiarthroplasty, or RTSA may be viable options.
The amount of displacement is critical when making decisions regarding nonoperative versus operative treatment for surgical neck fractures ( Fig. 6-16 ; also see Fig. 6-32 ). In the elderly patient, bony contact of fracture segments may be all that is necessary for a functional result. In active patients, less than 50% shaft diameter displacement and less than 45 degrees of angulation may be tolerated. Varus deformity, valgus deformity, comminution, and 100% displaced surgical neck fractures are considered unstable and require surgical intervention.
Management of surgical neck fractures depends on the fracture displacement, bone quality, and the functional demands and mental status of the patient. Nonoperative treatment is generally reserved for patients with minimally displaced or less than 50% displacement who can participate in a rehabilitation program. However, some patients, when treated nonoperatively, do not have good outcomes. Chun et al. reported only 55% good or excellent results in 56 surgical neck fractures treated nonoperatively with a mean forward flexion arc of 104 degrees. Displaced fractures in active patients should undergo ORIF or closed reduction and IM nailing. This latter treatment option works well when the surrounding soft tissue envelope is compromised and concern for postoperative infection exists ( Fig. 6-17 ). In a multicenter retrospective observational study, Hatzidakis et al. found that patients with displaced two-part surgical neck fractures that were managed with a locked angular-stable IM had reliable fracture healing, favorable clinical outcomes, and little residual shoulder pain. An alternative option, although technically demanding, involves percutaneous fixation with multiple threaded K-wires after closed reduction. Biomechanical studies have indicated improved fixation rigidity and increased load to failure with a configuration containing parallel descending and ascending pins. The main advantage of this technique is the minimal soft tissue dissection required, which can theoretically minimize the risk of iatrogenic injury to the vascular supply of the articular segment. Disadvantages of this approach include hardware migration, failure of fixation prior to fracture healing, pin tract–related infection, and the necessity for secondary removal of the hardware.
Most recently, in 2015, Rangan et al. performed a randomized controlled trial (RCT) comparing operative to nonoperative management of displaced surgical neck fractures. A total of 231 patients (114 operative and 117 nonoperative) were included in the study; the mean age of the patients was 66 years (range, 24 to 92 years). At the final 2-year follow-up, there was no difference in the patient-reported clinical outcomes using the Oxford Shoulder Score (OSS) and the 12-Item Short Form Health Survey. The authors further questioned the growing trend of operative intervention of these injuries in view of these results. However, surgeons should be aware of the limitations that exist with multicenter (and multisurgeon) studies before these data are universally applied to surgical practice.
In three-part fractures, cleavage lines occur through the surgical neck and GT or LT. The degree of displacement of the segments largely depends on the deforming forces of the rotator cuff muscles. The GT is more commonly involved than the LT, and is usually displaced in a posterior and superior position by the pull of the attached supraspinatus, infraspinatus, and teres minor. The humeral head is pulled into internal rotation (i.e., retroversion) secondary to the pull of the subscapularis on the intact LT, while the humeral shaft is displaced anteromedially due to the pull of the pectoralis major. If the fracture involves LT, the subscapularis pulls this segment medially. The intact GT and articular segment are pulled into adduction and external rotation, and the humeral shaft is similarly pulled in an anteromedial direction ( Fig. 6-18 ). Surgical treatment options include closed reduction and percutaneous fixation, ORIF, closed reduction and IM nail fixation with or without suture supplementation, hemiarthroplasty, and RTSA. The latter arthroplasty options are reserved for comminuted osteoporotic fractures that are unsuitable for osteosynthesis in elderly lower-demand, yet physiologically active patients.
In 1970, Neer reported his early experience in 39 patients with three-part PHFs treated by closed reduction. Only three patients demonstrated a satisfactory result by his criteria. Poor results were due to malreduction, nonunion, humeral head resorption, and osteonecrosis. He concluded that nonoperative intervention for these injuries was inadequate in active patients. More recent reports by Lill et al. and Zyto et al. suggest that good functional outcomes can be achieved after nonoperative management of these injuries. Zyto et al. prospectively evaluated 40 elderly patients (mean age, 74 years) with displaced three- and four-part PHFs randomized to nonoperative management or tension band fixation. At 3 to 5 years of follow-up, they found no differences in functional outcome between the two methods, despite radiographic evidence revealing improved position of the humeral head in the surgical patients. In 2011, Olerud et al. performed a randomized study in 60 patients (mean age, 74 years; range, 56 to 92 years) comparing ORIF using a locking plate to nonoperative management of displaced three-part PHFs. At the final 2-year follow-up, the results for ROM, function, and health-related quality of life (HRQoL) scores were all in favor of the locking plate group; however, critical appraisal of the differences between the groups reveals only marginal improvements in the operative group. Specifically, the mean forward elevation in the locking plate group was 120 degrees (vs. 111 degrees in the nonoperative group) and the mean abduction was 114 degrees (vs. 106 degrees). Functional scores including the Constant score was 61 in the operative group (vs. 58), and the Disabilities of the Arm, Shoulder and Hand (DASH) score was 26 (vs. 35). The small positive results in the operative group were also tempered by a 13% complication rate requiring a “major” revision surgery, and 17% of the patients underwent a “minor” reoperation (total reoperation rate of 30%).
Despite the potential risk for repeat surgery, the majority of displaced three-part PHFs in young adults require operative fixation due to the residual humeral deformity and functional deficits that can prevent the patient from returning to premorbid level of activity.
Nonoperative management of four-part PHFs should only be employed in the medically unfit and/or low-demand elderly patients. Because of the poor outcomes and high incidence of complications associated with nonoperative treatment (e.g., osteonecrosis, malunion, nonunion, and post-traumatic arthritis), most fractures are treated surgically ; however, patients should be properly counseled regarding possible risks and complications associated with both forms of treatment. Operative options include percutaneous K-wire fixation, ORIF ( Fig. 6-19 ), hemiarthroplasty, and RTSA. Combined or hybrid surgical techniques may be required for complex fracture patterns involving both the proximal humerus and humeral shaft (i.e., segmental fractures) ( Fig. 6-20 ). Closed reduction and percutaneous wire fixation is a viable option for acute injuries (<7 to 10 days) with good bone stock and minimal comminution. Open reduction is generally performed for fractures not amenable to reduction by closed means, for comminuted fractures, and for injuries that are 10 days to 4 months old.
The valgus-impacted four-part fracture is an uncommon but important subtype to identify. This injury has a more favorable prognosis when compared with other multi-part PHFs secondary to the integrity of the medial capsular blood supply. The fracture pattern is characterized by impaction of the lateral aspect of the humeral articular surface due to a fracture of the anatomic neck ( Fig. 6-21 ). The articular surface faces superiorly toward the acromion rather than the glenoid. The tuberosities typically displace into valgus secondary to impaction of the articular surface on the humeral metaphysis, but often do not displace superiorly. The prevalence of osteonecrosis approaches 5% to 10%, much less than that of the corresponding standard four-part PHFs (approximately 20% to 30%).
Jakob et al. achieved a satisfactory outcome in 74% of patients with valgus-impacted fractures who underwent closed reduction or limited ORIF. The major reason for failure was AVN, which occurred in five cases (26%). Resch et al. performed limited ORIF on 22 patients with four-part valgus-impacted fractures, demonstrating no evidence of osteonecrosis at a mean follow-up of 36 months (minimum, 18 months). The results were graded as excellent in 12 patients (54%), particularly in cases in which anatomic reduction was maintained.
Decision-making regarding the appropriate treatment for the aforementioned two-, three-, and four-part fracture subtypes remains challenging, especially in the elderly population ( Fig. 6-22 ). Several studies have attempted to collate the available data through meta-analysis. In 2012, Handoll et al. performed a review of 23 randomized trials involving management of PHFs. Their analysis did not provide fracture-specific management recommendations, except for the early mobilization of nondisplaced fractures. Fjalestad et al. performed an RCT comparing nonoperative treatment and ORIF in 50 elderly patients with displaced three- and four-part fractures. The authors did not find a difference in functional outcome at the 1-year follow-up between the treatment groups. Similarly, Li et al. failed to demonstrate a significant difference in clinical outcomes when comparing operative and nonoperative management of three- and four-part fractures in elderly patients. The authors cautioned against the extrapolation of the results of this study, as the analysis was based on limited data in a specific patient demographic.
Proximal Humerus Fracture-Dislocations
Two-part fracture-dislocations are amenable to ORIF due to the integrity of the vascular supply that is maintained by the soft tissue attachments to one or both tuberosity segments. Three- and four-part fracture-dislocations with or without articular surface involvement similarly require operative intervention ( Fig. 6-23 ). Repeated attempts at closed reduction or delayed ORIF can result in an increased incidence of myositis ossificans. The results of ORIF for four-part fracture-dislocations are poor; however, this remains the standard of care in young physically demanding patients. Arthroplasty serves as an appropriate treatment method for such injuries in older patients (i.e., over the age of 65 years) ( Fig. 6-24 ). Closed management should only be considered in the medically unfit patient.
Operative Techniques: General Considerations
Once faced with the decision for operative management of a PHF, the surgeon must decide to use one of the multiple surgical techniques available, which is dependent on the fracture subtype and surgeon experience.
Percutaneous Treatment of Proximal Humerus Fractures
The goal of surgical fixation of PHFs is twofold: to restore the anatomic relationship between the articular surface and the tuberosities, and to preserve the vascularity of the articular fragment. Closed reduction with percutaneous fixation minimizes soft tissue dissection, and can preserve vascular inflow for anatomic healing. Diminished adhesion formation and improved cosmesis are additional benefits of this approach. The caveat to this technique is that it involves an increased level of technical skill and ancillary support from operating room personnel. Percutaneous fixation of PHFs was initially used by Bohler to treat pediatric injuries, and was subsequently applied to adult surgical neck fractures. Because of the growing emphasis on biologic fixation, some authors have advocated percutaneous treatment of three- and four-part fractures. The learning curve is substantial, but proponents of this method cite a decreased incidence of osteonecrosis because the region of the ascending branch of the AHCA is undisturbed (see Fig. 6-3A ).
Historically, percutaneous and mini-open reduction of PHFs resulted in good outcomes and a low prevalence of osteonecrosis with short-term follow-up. However, more recent data from a multicenter case series with intermediate follow-up demonstrated an increased prevalence of osteonecrosis and posttraumatic osteoarthritis compared with that initially proposed. Some patients with these complications presented as late as 8 years postoperatively.
The indications for closed reduction and percutaneous pinning ideally include two-part (e.g., surgical neck), three-part, and valgus-impacted four-part PHFs without comminution in compliant patients with good bone stock. Weekly radiographic evaluation and shoulder immobilization for 4 to 6 weeks are required. Severe comminution and osteopenia are absolute contraindications to this technique. Additionally, numerous attempts to reduce fracture fragments should be avoided; in this situation, the surgeon should be prepared to convert to an open procedure. Fracture-dislocations can prove extremely difficult to manage with this method, and thus it is generally not recommended.
Specific technical steps are required to ensure a positive outcome from this intervention. Obtaining good biplanar images after patient positioning and before draping is crucial to the success of this procedure. Neuromuscular relaxation is required in the form of general anesthesia, muscle relaxants, and a regional block (e.g., interscalene block) to facilitate the reduction maneuver. A surgical assistant who can maintain reduction during the fixation process is also recommended.
The patient is placed in the beach chair or semi-Fowler’s position that allows access to the entire shoulder. Commercially available beach chair positioner attachments and mechanical arm holders will help significantly. Reduction instruments include bone elevators and hooks to manipulate the fragments into position. Terminally threaded 2.5-mm pins and 4.0-mm cannulated screws are commonly used for this technique. The C-arm image intensifier is placed parallel to the table to allow AP and axillary views (see Fig. 6-12 ).
A provisional reduction maneuver should be performed to determine the feasibility of a closed reduction. With two-part surgical neck fractures, there is apex anterior angulation at the fracture site. The arm should first be placed in 20 to 30 degrees of abduction. Longitudinal traction is then applied to the arm, while the humeral shaft is directed lateral to the humeral head. Concomitant pressure against the anterior surface of the shoulder will help reduce the apex anterior angulation ( Fig. 6-25 ).
A single pin is placed on the skin of the anterior shoulder in a distal-to-proximal direction from the lateral humeral cortex to the humeral head. An AP image confirms the inclination, and a skin marker is used to document pin positioning. A 1-cm incision is made on the lateral aspect of the shoulder at the level determined by the C-arm image. Blunt dissection is performed down to the bone with a hemostat. The hemostat is then used to palpate the anterior and posterior cortices of the humerus. The initial pin is then guided under power at the previously determined angle. Initial horizontal pin placement can prevent the pin from sliding across the cortex.
While the assistant maintains the reduction, the pin is advanced under fluoroscopy to engage the humeral head. Because of the approximate 30 degrees of retroversion of the humeral head, fluoroscopy must be used to confirm appropriate pin placement. A second pin is then advanced parallel to the first with approximately 1.5 to 2 cm of separation. Several authors have demonstrated that parallel pin placement increases torsional stability and rigidity in biomechanical testing when compared with converging pins. A third pin is then placed from the anterior humeral shaft into the humeral head. A fourth pin from anterior to posterior may be necessary to augment the fixation ( Fig. 6-26 ). An alternative surgical technique involves placement of two K-wires, one in the humeral head and another in the shaft. The K-wires are then utilized as joysticks to achieve fracture reduction, with subsequent placement of terminally threaded pins or cannulated screws.
Three-Part and Four-Part Fractures
Three- and four-part PHFs add a level of complexity that proves difficult to manage with this technique, even in the most experienced hands.
With a three-part valgus-impacted fracture, the humeral head is tilted into valgus while the GT remains at the correct height. After positioning the humeral shaft under the humeral head, a small incision is made laterally to allow placement of an elevator into the fracture site under the humeral head. Under fluoroscopy, the humeral head is then levered upward and into varus alignment. As the humeral head is reduced, the GT will be pulled below the articular surface due to the integrity of the rotator cuff insertion. A 2.5-mm threaded pin or 4.0-mm cannulated screw is then advanced through a separate superior incision to capture the GT and the humeral head. Cannulated screws are usually placed to hold the tuberosity and articular surface reduction, while additional pins or screws are inserted in an antegrade fashion to stabilize the tuberosity to the humeral shaft.
Four-part fractures, except for the valgus-impacted subtype, generally require open reduction and internal fixation. On the infrequent occasion that percutaneous fixation is attempted, the same steps to reduce the GT, articular surface, and humeral shaft are employed. The LT may be reduced by internally rotating the arm and by placing a hook on the fragment through a laterally based incision. Once reduced, the fragment is held in place with a 4.0-mm cannulated screw. Once satisfied with fracture reduction and fixation, the threaded pins are trimmed below the skin, which is closed. Tenting of the skin from the pin can occur as shoulder swelling diminishes. In some cases, the pin(s) require trimming on an outpatient basis.
Most patients are admitted postoperatively for 24-hour observation and continued parenteral antibiotics. The shoulder is placed in a shoulder immobilizer for comfort. Hand, wrist, and elbow active ROM exercises are encouraged immediately. Pendulum exercises may be initiated at 1 week following outpatient radiographs that confirm fracture reduction and hardware stability. Passive forward flexion and external rotation exercises are initiated during the third postoperative week. Weekly physical examination and radiographs are obtained, and the pins are generally removed at 4 to 6 weeks postoperatively. Active-assisted shoulder exercises are then performed under the supervision of a physical therapist. Radiographic union should be apparent by 2 or 3 months. Shoulder strengthening may be initiated by 10 to 12 weeks postoperatively.
Resch et al. reported good to very good functional results (Constant score, 91% in three-part and 87% in four-part fractures) in 27 patients treated with percutaneous fixation of three- and four-part PHFs (9 three-part, 18 four-part) without evidence of humeral head necrosis at 24 months of follow-up. A follow-up study by Resch et al. on the treatment of four-part fractures indicated good to very good results, with a Constant score of 87% and an osteonecrosis rate of 11% in 18 patients.
Keener et al. reported the results after percutaneous reduction and fixation of two-, three-, and four-part PHFs in 35 patients from three institutions. The mean follow-up was 35 months, with mean visual analog pain, American Shoulder and Elbow Surgeons (ASES), and Constant scores of 1.4, 83.4, and 73.9, respectively. Fracture type, age at presentation, malunion, or osteoarthritis had no significant influence on the measured outcomes. The authors concluded that this technique remained a viable and predictable option in the treatment of PHFs.
Open Reduction and Internal Fixation
Numerous techniques of internal fixation for two-, three-, and four-part PHFs have been described in the literature. These methods include plate and screw fixation, blade plates, tension banding, antegrade or retrograde nailing, or a combination of internal fixation methods with sutures, cables, and metallic implant fixation. Although the ideal construct for these injuries has not been defined, current advances in locking plate technology have provided surgeons with another viable option in the treatment of multipart PHFs, particularly in patients with suboptimal bone stock.
The osseous architecture of the humeral head varies with respect to the density of the cancellous bone stock. In elderly patients, this proves particularly problematic when the goal is to achieve rigid internal fixation capable of withstanding a progressive rehabilitation program. Attempts at rigid fixation with prior implants have indicated a high risk of failure due to osteonecrosis, nonunion, malunion, screw cutout, and screw loosening. Blade plate fixation has been used to overcome concerns regarding early hardware failure, but this challenging technique only allows a single point of proximal fixation in the humeral head. In addition to the soft tissue dissection required with this technique, some authors have reported unacceptable complication rates related to blade plate protrusion into the glenohumeral joint (eight of 36, or 22% of patients in one study). Using a three-part cadaveric model, Weinstein et al. demonstrated increased torsional stiffness of the locking plate construct when compared with the blade plate construct. Percutaneous pinning and IM fixation can preserve the soft tissue envelope and decrease the risk of osteonecrosis, but the indirect reduction techniques can make anatomic alignment of the fracture fragments difficult to achieve. At this time, there is no consensus regarding the optimal form of fixation for these injuries.
Saudan et al., Frankhauser et al., and Bjorkenheim et al. have reported early favorable results using a similar titanium implant. All authors reported excellent results, with union rates greater than 90%. Rose et al. reported their 1-year experience with a stainless steel proximal humeral-locking plate of similar design. They followed 16 patients (nine with three-part fractures, five with two-part fractures, and two with four-part fractures) until union or revision at a mean follow-up of 12 months. Of the four nonunions, three were three-part fractures with extensive metaphyseal comminution in smokers. Patients with united fractures demonstrated mean forward elevation and external rotation of 132 degrees and 43 degrees, respectively. The authors were encouraged by the early positive findings, but they emphasized that there are limitations with this method based on the degree of comminution and patient factors (e.g., smoking). Longer term follow-up will be necessary with this surgical technique to better elucidate long-term outcomes and negative sequelae.
Nonlocking and Locking Plate Fixation
Based on a current review of the literature, Nho et al. recommended buttress plate fixation in young patients with diaphyseal cortices greater than 3.5 mm. However, metaphyseal comminution prevented the use of a nonlocking device. The proximal humeral-locking plate has been developed to overcome some of the limitations of conventional plating and to address issues specifically related to osteoporotic bone and metaphyseal comminution. The stability of this implant is based on the fixed-angle relationship between the screws and the plate. The threaded screw heads lock into the corresponding threaded plate holes, preventing toggle, pullout, and sliding. The stability of the fracture-and-plate construct does not depend on any single screw because of the fixed-angle construct. As a single unit, the plate is better able to withstand bending and torsional forces. Precontoured proximal humeral-locking plates provide a mechanical advantage in fractures with metaphyseal comminution ( Fig. 6-27 ). Proximal humeral-locking plates have also evolved to include peripheral holes for rotator cuff fixation and polyaxial screws that allow variable angle screw placement. Both implant traits help mitigate, but unfortunately, have not eliminated, issues related to PHFs with osteoporotic bone.
Intramedullary Nail Fixation
Early reports on IM fixation of two- and three-part fractures by Lin et al., Adedapo et al., and Rajasekhar et al. suggested that this is a reliable technique, with union rates of 100% (21 patients), 97% (29 patients), and 100% (23 patients), respectively. Agel et al. placed emphasis on the appropriate location of nail insertion medial to the GT, and suggested that lateral metaphyseal comminution can contribute to implant failure. Of the 20 fractures treated by these authors, only 65% achieved union at the latest follow-up (range, 4 to 9 months), and 15% (three cases) demonstrated proximal screw loosening. These authors further concluded that IM fixation of PHFs is not effective in the following circumstances: (1) if the insertion site is incorrect, (2) if the insertion site is violated by fracture, or (3) if the fracture contains extensive lateral metaphyseal comminution.
Park et al. performed IM fixation augmented with a tension band and a locking suture technique, and noted fracture union in 25 of 26 patients (95.8%) at a mean of 8.7 weeks (range, 7 to 12 weeks). At a mean follow-up of 39 months (range, 24 to 59 months), ASES scores averaged 85 (range, 40 to 100), with no statistical difference evident when comparing patients younger or older than 65 years of age (range, 27 to 79 years old). The authors concluded that open IM nailing remained a viable technique for PHFs in elderly patients, and good functional outcomes could be expected at midterm follow-up.
Koike et al. reported their results with the use of a commercially available IM nail in 54 patients. Using Neer’s criteria, the majority of the injuries were classified as two-part fractures (29 patients or 54%). All surgically managed fractures achieved union, with 43 patients (79%) demonstrating “satisfactory or excellent” results. Residual varus deformity and GT malunion was reported in eight shoulders (15%).
Popescu et al. prospectively evaluated the functional and radiologic outcomes of using an IM nail for fracture fixation in 29 patients (21 proximal and eight humeral shaft fractures). Eighteen patients were older than 70 years of age. All fractures healed at a mean of 2.7 months of follow-up. Complications were limited to a single case of AVN and one case of proximal screw migration. The authors concluded that the proximal locking mechanism of this device reduced the risk of screw failure. There was an additional trend towards better functional results in patients younger than 70 years of age, but this finding did not reach statistical significance.
More recently, Hatzidakis et al. utilized a medial starting point for nail placement to avoid injury to the rotator cuff and reported favorable outcomes in 38 patients with two-part fractures at 1-year follow-up. The study group demonstrated a mean Constant score of 71, a Constant pain score of 13 (range, 0 to 15, with 15 indicating no pain), mean forward flexion of 132 degrees, and a neck shaft angle at union greater than 125 degrees in all but one patient.
Lopiz et al. compared two IM nail constructs in a cohort of 52 prospectively evaluated patients and concluded that an articular entry straight nail resulted in similar outcomes (i.e., union rate) compared to curvilinear implants, with a reduced impact on the rotator cuff and with fewer complications. The authors reported a 42% reoperation rate in the curvilinear nail group (vs. 11% in the straight nail group); 73% of the patients in the curvilinear nail group demonstrated rotator cuff disease–related symptoms (vs. 35% in the straight nail group). Seven of the 11 cases that required revision surgery in the curvilinear nail group were related to screw migration. The remaining four patients required hardware removal, with one case requiring conversion to an RTSA.
Locking Plate Versus Intramedullary Fixation
Kitson et al. investigated the biomechanical behavior of a locking plate versus a locking IM nail for the treatment of three-part PHFs. Paired cadaveric specimens were tested with four directional loads: flexion, extension, varus, and valgus. When cantilever bending loads were below the failure threshold, the nail construct demonstrated significantly higher stiffness results in all directions. The authors also noted different modes of failure for each implant. In all samples, the nail construct failed at the bone-screw interface, whereas all of the plates failed due to bending of the implant at the level of the surgical neck fracture. These authors cautioned extrapolating these data to the clinical setting because the results of controlled osteotomies cannot be equated to PHFs with cortical and metaphyseal comminution.
In contrast, Sanders et al. performed a similar cadaveric evaluation and found that the locking plate construct demonstrated greater stiffness in valgus loading compared with the nail construct (420 vs. 166 N/mm). All other loading vectors showed no statistical differences between the two constructs. Failure in the locking plate occurred in a similar manner through the surgical neck osteotomy site as described by Kitson et al. Both studies used similar biomechanical testing techniques; however, it remains unclear whether Sanders et al. performed their testing using a locking version of the IM nail. Both studies were limited by sample size, and thus, conclusions must be applied to clinical practice with caution.
Boudard et al. recently performed a retrospective observational study comparing locked plate fixation and IM nail fixation for nonosteoporotic three- and four-part fractures of the proximal humerus. A total of 63 patients (33 with plate and 30 with nail fixation) were available for a 1-year clinical and radiographic follow-up. The mean Oxford, Constant, relative Constant, and QuickDash scores did not differ between the groups. Patients with four-part fractures and GT comminution demonstrated statistically significant lower functional scores. Complication rates did not differ between the two surgical groups, although the plate group had three infections. The authors concluded that internal fixation was appropriate for three- and four-part fractures. Although they could not recommend one modality over the other, they suggested that the locking plate technique was “more aggressive.”
In 2016, Gracitelli et al. performed a prospective, single-center, RCT comparing the clinical and radiographic outcomes of PHFs (two- and three-part) treated with either a locking IM nail (third-generation, curved nail with an articular entry point) or locking plate. There were 36 patients randomized to each treatment group; baseline characteristics such as age, smoking status, and fracture pattern and displacement were well balanced between groups. There were no statistically significant differences between treatment groups for the clinical outcome scores (Constant-Murley, DASH, and VAS) at the 12-month follow-up; however, there were significantly more complications in the IM nail group. Specifically, 11 patients (34%) from the nail group and seven (21%) from the plate group had one or more complication. Reoperations were performed in six patients (19%) in the nail group and in one patient (3%) in the plate group.
In the 1950s, Neer initially developed the technique of humeral head arthroplasty (HA) for treating PHFs. Before this method was developed, these injuries were treated with benign neglect or humeral head resection. By 1970, Neer reported uniformly poor results after ORIF for four-part fractures and fracture-dislocations, and suggested HA for these fractures. Because of the high risk of humeral head osteonecrosis with three- and four-part PHFs, Neer believed that even with anatomic reconstruction with internal fixation, posttraumatic osteoarthritis would develop, resulting in poor outcomes for the patients. Neer reiterated that HA and stable fixation of the tuberosities could provide a reasonable result for the patient.
Clinical studies regarding hemiarthroplasty for PHFs are difficult to directly compare as there remains heterogeneity regarding implant design, surgical technique, rehabilitation, and the outcome instruments used to evaluate patient outcome. In view of these limitations, the literature demonstrates that approximately 90% of patients treated with HA demonstrate minimal pain despite a wide range of function, ROM, and muscle strength. Factors that predict a poor outcome after HA for fractures include tuberosity malposition, tuberosity resorption and associated superior migration of the humeral prosthesis, stiffness, persistent pain, and poor initial positioning of the implant (excessive retroversion, decreased height); women older than 75 years are also at risk for poor outcome. Regarding the timing of surgery (i.e., the time between injury and definitive surgery), most authors report poorer outcomes with surgical intervention delayed longer than 2 to 3 weeks, particularly as it relates to functional results. Robinson et al. reported satisfactory results in patients treated with primary HA with a mean prosthetic survival of 6.3 years. In this study, the majority of patients were pain free at their last follow-up, despite a wide variation in motion, function, and muscle power. Poorer results were noted in elderly patients, especially in the settings of neurologic deficit or instability with retracted tuberosities. Good functional outcomes at 1 year could be anticipated in younger patients who were nonsmokers without postoperative complications or neurologic deficits and with satisfactory radiographs at 6 weeks.
Bicknell et al. performed a pilot study using cadaveric specimens to assess the feasibility of a computer-assisted system in the treatment of four-part fractures. Seven pairs of specimens underwent an HA by way of computer-assisted and traditional methods to anatomically reconstruct the following seven characteristics: humeral head version, humeral head inclination, humeral head offset, humeral length, medial articulation point, GT position, and LT position. The differences between the intact and reconstructed proximal humeral anatomic values were improved in five of seven characteristics, but only humeral head offset demonstrated statistically significant improvement with the computer-assisted method. Although computer-assisted HA is promising, the authors emphasized the need for further investigation and refinement of this technique to allow reproducibility and to facilitate its use in the clinical setting.
Several retrospective and prospective studies have confirmed the viability of HA for the treatment of three- and four-part PHFs that were not amenable to ORIF. Antuña et al. reported their results in 57 patients with a mean age of 66 years and a mean follow-up of 10.3 years (range, 5 to 22 years). Utilizing Neer’s criteria, less than 50% of the patients reported satisfactory results (22/57) despite significant pain relief. Fialka et al. demonstrated improved Constant scores at 1 year of follow-up when utilizing a fracture-specific humeral stem; however, nearly 30% of the patients demonstrated tuberosity resorption. Hertel et al. reported their results on 49 patients, with a mean follow-up of 5 years (range, 3.3 to 7.3 years). The mean Constant and Subjective Value Scores were 70 (range, 39 to 84) and 90 (range, 40 to 100), respectively. Boileau et al. demonstrated improved Constant scores in 30 patients at a mean follow-up of 45 months. The absolute mean Constant scores were 68 and 93 when adjusted for age and gender, respectively. Male patients and those individuals under the age of 75 years fared better with regards to functional results after HA. Nearly all patients reported “satisfactory” results after surgery (29 of 30). In 2011, Olerud et al. performed a prospective study of 55 patients (mean age, 77 years; range, 58 to 92 years), who were randomized to either HA or nonoperative treatment for a displaced four-part PHF. At the final 2-year follow-up, the authors concluded that HA resulted in improved quality of life (as per the EQ-5D) and pain relief; however, there were no significant differences regarding the Constant score or ROM. Both groups achieved a mean forward elevation of 90 to 95 degrees despite the use of a fracture-specific stem and reproducible surgical technique. The reoperation rate was low (three patients in the HA group and one patient in the nonoperative group). More recently, Cai et al. randomized 32 patients (mean age, 71.9 years; range, 67 to 86 years) to either HA or internal fixation using a locking plate for a displaced four-part fracture. A control group (i.e., nonoperative treatment) was not included in this study. Twenty-seven patients (84%) were available for the final 2-year follow-up. The functional outcomes (Constant and DASH scores), health-related quality of life, and pain scores favored HA over internal fixation, although most outcomes did not reach statistical significance. ROM parameters were also better for the HA group with regards to forward flexion and abduction.
Reverse Total Shoulder Arthroplasty
Due to growing concerns regarding the potential for poor outcomes following HA related to tuberosity healing or implant malposition, significant interest has developed amongst shoulder surgeons in performing an RTSA for displaced three- and four-part PHFs in the elderly population. The main attraction to this form of arthroplasty is the ability to achieve functional shoulder elevation and abduction regardless of tuberosity healing, position, and degree of comminution; however, the repair and union of the GT fragment(s) during RTSA have been demonstrated to improve functional outcomes including external rotation and patient satisfaction compared to outcomes after tuberosity resection, nonunion, or resorption. An additional advantage of an RTSA is that the medial glenoid erosion observed following HA may be prevented through placement of the glenosphere component.
Over the last 10 years, several studies have demonstrated that RTSA is a viable surgical option for acute three- and four-part PHFs and fracture sequelae. Most published studies represent case series (level IV evidence) or systematic reviews; high-level evidence is limited. Boulahia et al. reported their initial results (mean follow-up, 35 months) of a mixed patient population (16 patients) with cuff tear arthropathy and fracture sequelae who underwent RTSA. The mean preoperative-to-postoperative forward elevation increased from 70 to 139 degrees, with a corresponding increase in the Constant score from 31 to 59. The authors did note a revision rate of 12.5%.
Boileau et al. reviewed their results using the Delta III prosthesis (Depuy) in the management of three distinct patient groups: cuff tear arthropathy (21 patients), implant revision (19 patients), and fracture sequelae with associated arthritis (5 patients). Although all groups demonstrated significant increases in forward elevation (55 to 121 degrees) and Constant score (17 to 58), the authors did not observe an improvement in external rotation. More importantly, they reported a 20% (one of five cases) complication rate with the fracture sequelae group (one implant was revised to a hemiarthroplasty due to glenoid fracture). These authors emphasized that functional results are less predictable and complication rates are higher in revision and fracture sequelae cases compared to cuff tear arthropathy cases.
Cazeneuve and Cristofari retrospectively analyzed 23 patients who underwent primary RTSA using the Delta III prosthesis for acute three- and four-part PHFs (18 patients) and fracture-dislocations (five patients). Tuberosity reduction was possible in only five patients. There were 16 patients (70%) available for final clinical and radiographic follow-up; seven patients died prior to the final follow-up. The mean Constant score was 60 (83 for the contralateral, uninjured side) at a mean follow-up of 86 months. Significant gains in forward elevation, abduction, and pain relief were reported. Complications included regional pain syndrome (two patients), infection (one patient, requiring revision), and anterior dislocation (one patient, requiring revision). Based on their results, the authors recommended RTSA for elderly patients with unmanageable tuberosities (e.g., tuberosity comminution or high-risk of tuberosity resorption postoperatively or preexisting rotator cuff pathology) to attain gains in pain relief and forward elevation.
Bufquin et al. reported their results in 41 patients (mean age, 78 years; range, 65 to 97 years) who underwent RTSA for displaced three- and four-part PHFs. The mean follow-up was 22 months (range, 6 to 58 months). The clinical outcome was satisfactory, with a mean active forward elevation of 97 degrees (range, 35 to 160 degrees) and abducted external rotation of 30 degrees (range, 0 to 80 degrees). The Constant and modified Constant scores were 44 and 66%, respectively. Tuberosity displacement occurred in 19 patients (53%) and scapular notching in 10 (25%). The authors concluded that “satisfactory” mobility could be obtained even with tuberosity displacement.
Boyle et al. performed a retrospective case-control study (level III evidence) comparing HA (313 patients) to RTSA (55 patients) for three- and four-part PHFs. Compared with HA, patients who underwent RTSA were significantly older (mean age, 79.6 vs. 71.9 years) and more often women (93% vs. 78%). The 6-month OSS was 28.1 for RTSA and 27.9 for HA. The RTSA group had a significantly better 5-year OSS than the HA group (41.5 vs. 32.3). The authors did not observe a significant difference between the RTSA and HA groups in the revision rate per 100 component years (1.7 vs. 1.1) and in 1-year mortality (3.5% vs. 3.6%). They concluded that patients with acute PHFs who undergo RTSA achieve superior 5-year functional outcomes compared with patients who undergo HA.
Cuff and Pupello performed a prospective cohort (i.e., nonrandomized) study involving 43 patients with three- and four-part PHFs treated with either HA (23 patients) or RTSA (24 patients) with a minimum 2-year follow-up. ASES, Simple Shoulder Test, and patient satisfaction scores were significantly lower in the HA group. Interestingly, tuberosity union rates were higher in the RTSA group (83% vs. 61%). Forward elevation was better in the RTSA group (139 degrees vs. 100 degrees), and three patients from the HA group required revision to RTSA (13% conversion rate). However, analysis of the outcomes relative to tuberosity healing yielded similar results between the HA and RTSA groups. The authors concluded that RTSA resulted in consistently better clinical outcomes and similar complication rates compared with HA for the treatment of comminuted PHFs in the elderly.
In 2013, Mata-Fink et al. performed a systematic review of the literature comparing RTSA to HA. Fifteen studies met the inclusion criteria for the meta-analysis, including 377 patients treated with RTSA and 504 patients treated with HA. The RTSA group had improved forward elevation and functional outcome scores compared with the HA group but diminished external rotation. Complication rates were similar between the two groups. The authors suggested that RTSA is a reasonable alternative for treating older adults (>60 years) with PHFs, but additional research and longer term follow-up are necessary. Sebastia-Forcada et al. performed a prospective study of 62 patients, with equal numbers of patients randomized to either RTSA or HA. RTSA patients had significantly higher Constant scores (56.1 vs. 40.0), forward elevation (120.3 vs. 79.8 degrees), and abduction (112.9 vs. 78.7 degrees), but no difference in internal rotation (2.7 vs. 2.6 degrees). The DASH score was higher in the HA patients (17 vs. 29), denoting greater disability in the HA group. In the HA group, 56.6% of tuberosities healed versus 64.5% in the RTSA cohort. Six patients (19.4%) within the HA group had proximal humeral head migration that required revision to an RTSA. The functional outcome was independent of tuberosity healing in the RTSA group. Scapular notching was observed in one RTSA patient. One patient developed a hematoma, and another developed an infection requiring two-stage revision to another RTSA. Most recently, in 2015, Ferrel et al. performed a systematic review comparing the results of RTSA to HA (30 studies; 1346 patients), and noted improved forward elevation (118 degrees vs. 108 degrees) but inferior external rotation (3 degrees vs. 20 degrees) with RTSA. No significant clinical differences in either the ASES (RTSA 64.7 vs. HA 63) or Constant score (RTSA 54.6 vs. HA 58) were identified. Furthermore, the authors found an increased rate of complications with RTSA (9.6%) but a lower revision rate (0.93%) at the short- to mid-term follow-up.