Vascular supply of the humeral head is provided by the anterior and posterior circumflex arteries, both originating from the axillary artery. Whereas at the beginning the role of the anterior circumflex artery was overestimated [12, 27], the importance of the posterior circumflex artery’s contribution was highlighted later by several authors [20, 34, 39]. Decisive for the development of posttraumatic avascular head necrosis (PAVN) is the vascularity of the articular segment. In vivo measurement of this parameter is technically difficult and allows only assessment of the situation directly after the trauma. Posttraumatic processes of revascularization cannot be determined or quantified in a sufficient manner despite modern imaging tools. In 2004, Hertel et al. [33] published a pioneering work regarding this topic and identified 3 primary predictors of humeral head ischemia in 100 acute fractures after an average time period of 4 days after the trauma. These predictors were a calcar segment <8 mm, a disrupted hinge (>2 mm), and a fracture line at the anatomic neck (fracture types 2, 9, 10, 11, and 12 according to the classification of Hertel [33]). The combination of all 3 predictors was associated with a risk for humeral head ischemia of 0.97.

Thus, the risk for development of PAVN is mostly determined by the type of fracture. An incidence between 3 and 14 % has been reported for three-part fractures and 13–34 % for four-part fractures [18, 43, 56].

Avascular necrosis has been classified by Ficat and Arlet [24] and was modified by Cruess [17] for the humeral head:

Stages I to III affect only the subchondral bone and cartilage, whereas the soft tissue envelope including capsule and rotator cuff is not involved. Due to the collapse of the head in advanced cases (stages IV and V), the center of rotation (COR) of the humeral head is shifted medially (medialization of COR) associated with capsular restriction and shortening of rotator cuff muscles adapted to the changed position of COR. This has to be taken into consideration when inserting a humeral head component of normal size and even more when total shoulder arthroplasty is planned with insertion of a glenoid component. Extended soft tissue management with bifocal capsular release on the humeral as well as on the glenoid side is of utmost importance in these cases to avoid overstuffing followed by pain and postoperative stiffness. Furthermore, release of the subscapularis tendon represents a key step of the procedure. A 270° tenolysis according to Matsen is performed mobilizing the tendon from the conjoint tendon, beneath the coracoid, and from the glenoid and scapular neck. Hereby, care must be taken not to harm the axillary nerve. In cases with malunion of the lesser tuberosity, correction osteotomy should be performed with lateralization of the lesser tuberosity.

Gerber et al. [26] stated that in cases with humeral head necrosis, the deformity rather than the necrosis itself creates disability underlining the importance of anatomic configuration of the proximal humerus with acceptable tuberosity alignment.

Treatment strategies are first of all stage dependent. Quite often a discrepancy is seen between advanced radiological changes with collapse of the head and low pain and moderate functional restriction and vice versa. Thus, the indication for surgical treatment has to be set individually depending on patients’ complaints. However, range of shoulder motion should not be limited too much and for too long time in order to avoid soft tissue shortening and degenerative alterations as atrophy and fatty infiltration of the rotator cuff muscles. A functional and compensated soft tissue envelope represents a prerequisite for a good clinical outcome after posttraumatic shoulder arthroplasty.

If radiological signs of PAVN are present and the full spectrum of nonoperative treatment is exhausted, humeral head replacement represents the only sensible treatment solution. As standard implant stemmed humeral prosthesis has been established. A key aspect for adequate shoulder function is a correct COR. In order to restore anatomic glenohumeral joint kinematics, it is necessary to restore the COR, which in PAVN with loss of sphericity and possible additional tuberosity malunion has changed. Out of this rationale, it is important to use a shoulder prosthetic design of the fourth generation with 3D adaptability (inclination, retroversion, and eccentricity). In order to choose the right head size, preoperative radiographs of the uninjured shoulder are recommended to be made. If the surgeon is in doubt regarding the head size, a smaller one should be chosen in order to avoid joint overstuffing, especially if total shoulder arthroplasty is carried out. A major problem of stem designs in PAVN is that implantation of the humeral component refers always to the humeral shaft limiting the possibility of COR restoration in severely distorted anatomy. One option is to choose a smaller stem size and to implant it cemented. Nevertheless, some varus or valgus malalignment (Fig. 16.3) has to be tolerated, and sometimes it is even impossible to restore the COR due to severe malunion. In a recent retrospective study, Moineau et al. [40] followed up (mean follow-up time of 52 months) 55 patients undergoing anatomic shoulder arthroplasty (80 % TSA) without greater tuberosity osteotomy for type 1 fracture sequelae. The Constant score improved from 32 to 69 points, the Simple Shoulder Value was 81 %, and the revision rate was 7 %. As negative prognostic factors, proximal humeral deformity, specifically varus more than valgus malunion of the greater tuberosity, and fatty infiltration of the rotator cuff muscles have been identified. The authors concluded that anatomic shoulder arthroplasty for PAVN with acceptable deformation of the proximal humerus provides good and predictable outcomes. When negative predictive factors are present, reverse shoulder arthroplasty may be more appropriate, especially in elderly patients. The findings of Moineau et al. are consistent with those reported in the literature [8, 11, 19, 29, 36], including the author’s own experience [58].

Out of the complex posttraumatic situations requiring independency from the humeral shaft, stemless designs were developed. Usually, the metaphyseal bone at the resection plane corresponding to the anatomic neck is of good, sclerotic bone quality allowing for stable fixation of various stemless designs. In the author’s practice, the Eclipse prosthesis is used (Arthrex, Naples, Florida). This stemless implant offers a dual primary fixation mechanism. First, the so-called trunnion, a certain kind of baseplate, has cortical support on the resection plane, and a second stabilizing mechanism is provided by the cage screw, which is inserted through a central hole in the baseplate. Both, the trunnion and the cage screw, are coated with hydroxyapatite allowing for a reliable bony ingrowth. Finally, the head component is set on the top of the baseplate, which completes the three-component implant (Fig. 16.4). The plane of resection is defined using a 135° inclination resection guide. The amount of retrotorsion should be decided individually and ranges between 20° and 40°. During the resection using an oscillating saw, care must be taken not to harm the rotator cuff insertion which in some cases overlaps the resection plane due to tuberosity malunion and collapse of the articular segment. Small bone cysts, if encountered, can be filled with autologous or allogeneic bone material and usually do not set primary fixation stability of the cage screw at risk. However, if the metaphyseal bone is weak and/or soft with insufficient primary fixation of the cage screw, a stemmed prosthesis should be implanted. Thus, both arthroplasty systems should be at disposal in the surgical unit, when posttraumatic arthroplasty is performed. In a retrospective multicenter study, the outcome of the stemless Eclipse prosthesis has been evaluated [14]. From a total of 233 patients, 70 were posttraumatic cases type 1 with an average age of 59 years and a follow-up period of 27 months; 71 % of the patients had previous surgery. The age- and sex-related Constant score increased from 48 % preoperatively to 72 % postoperatively. Regarding the subgroups, the pain score improved from 7 to 12 points, activities of daily living from 8 to 13, range of motion from 16 to 21, and strength remained unvaried at 8 points. Range of motion increased in all planes significantly with flexion from 91° to 109°, abduction from 67° to 106°, and external rotation from 11° to 32°. Previous surgery seemed to negatively influence the final outcome (79 % according to the age- and sex-related Constant score for primary surgery and 69 % for revision surgery, respectively) without achieving statistical significance (P = .18). In comparison to the 100 cases operated on for primary osteoarthritis (53 % Constant score preoperatively to 84 % postoperatively), the posttraumatic results were moderately inferior (Fig. 16.5).

If PAVN is in an early stage with preserved sphericity of the articular segment (stages I to III, which is rarely the case), humeral head resurfacing can be performed. Initially designed for primary osteoarthritis, these cup prostheses replace only the cartilage and the subchondral bone. Therefore, it should be warranted that the necrotic process has ended without any further progression and loss of epiphyseal bone stock. Implants of the first generation were nonanatomic due to a geometric mismatch creating a glenohumeral overstuffing. With introduction of the new cup generation, this problem can be avoided. As main disadvantages, the risk for secondary glenoid erosion and the difficulty of glenoid component implantation have to be mentioned, which in PAVN is not always necessary.

In a retrospective study, Raiss et al. [48] compared the functional results after cementless humeral replacement between posttraumatic (n = 8) and nontraumatic (n = 9) osteonecrosis of the humeral head. They included even stage 4 osteonecrosis with the presence of bone loss of as much as 31 % of the humeral head. After an average follow-up period of 3 years, significant improvement was observed for both groups (average age 48 years) with superior benefit in favor to the nontraumatic group (34 preoperatively to 70 points postoperatively in the Constant score for the nontraumatic group and 28–52 points for the PAVN, respectively). Signs of implant loosening were not observed in this short-term follow-up period. In 2010, comparable results were published by Pape et al. [47] reporting on 28 shoulders undergoing humeral head resurfacing after PAVN. The average age of the patients was 60 years with an average follow-up of 31 months. Three types of PAVN were distinguished according to the inclination of the impacted head: type 1, an impacted fracture with the head in an anatomic position; type 2, a valgus impacted fracture; and type 3, a varus impacted fracture. The overall Constant score improved from 23 to 55 points, with the best results for valgus impacted fractures. No loosening was reported with one patient suffering from mild secondary glenoid erosion. One patient required revision arthroplasty using a reverse design due to anterosuperior instability and persistent pain.

If PAVN is combined with severe tuberosity malunion or nonunion, shoulder arthroplasty using a reverse design should be considered. It is well documented in the literature that combined humeral head replacement and greater tuberosity osteotomy is associated with poor and unpredictable functional outcomes [9, 36, 45]. Thus, a constrained prosthetic design should be preferred in cases with PAVN and severely distorted anatomy of the tuberosities with secondary functional cuff deficiency.

These severe shoulder injuries are rare but still occur due to the fact that mainly posterior shoulder dislocations or fracture-dislocations represent the most commonly misdiagnosed dislocations of the body with devastating consequences. Main reasons are high-energy trauma or seizures [53]. The purpose is to primarily recognize such dislocations, to reduce them, and to reconstruct bony impression defects such as Hill-Sachs or Malgaigne lesions and bony glenoid rim lesions in order to prevent recurrence. Patients with type 2 fracture sequelae are younger [4, 31], and the main objective of shoulder arthroplasty is pain relief [4]. Anatomic shoulder arthroplasty (HSA or TSA) is recommended in type 2 fracture sequelae [8, 9]. Besides management of bony lesions, soft tissue balancing represents a challenging but key step procedure regarding the functional outcome. Capsule and rotator cuff muscles have been adapted to the chronic dislocated position of the joint and have to be addressed adequately and accurately during arthroplasty. In chronic posterior shoulder dislocation with fixed glenohumeral internal rotation and static decentration, the anterior soft tissue structures are shortened, which means to perform an extended anterior capsular and subscapularis tendon release together with tendon lengthening. In addition, a posterior capsular plication sometimes might be necessary in order to re-center the humeral head. However, the risk for recurrent instability of the shoulder arthroplasty might be given, so that a constrained prosthetic design becomes necessary in some cases.

Boileau et al. [9] reported on shoulder replacement in 9 patients with type 2 fracture sequelae, 7 had locked posterior dislocation, 1 locked posterior fracture-dislocation, and 1 locked anterior fracture-dislocation. After a mean delay of 19 months, 8 had HSA and 1 TSA with partial greater tuberosity in only one patient. The Constant score improved from 22 to 71 points, which was the highest improvement compared to other types of fracture sequelae. Although the overall clinical results are good, an increased complication rate up to 32 % has to be considered. Most frequent complications are infection, restriction of range of motion, persistent posterior instability, and heterotopic ossifications [4]. Similar good results have been reported for the outcome after reverse shoulder arthroplasty in type 2 fracture sequelae. Gwinner et al. [30] published their results of reverse shoulder arthroplasty for fracture sequelae treatment including 4 type 2 sequelae with combined rotator cuff pathology. The Constant score improved from 19 to 75 points (27–101 % age- and sex-adjusted) which was again the best result of all types of fracture sequelae treated with a reverse design. Neyton et al. [44] observed comparable results (16–57 points according to the Constant score), which are equivalent to anatomic shoulder arthroplasty. Thus, reverse shoulder arthroplasty is recommended by those authors in chronic locked anterior and posterior dislocations or fracture-dislocations with associated rotator cuff deficiency.

Humeral surgical neck nonunions are rare, but they cause considerable disability for the patient in terms of pain and functional impairment. Epidemiological data indicate a prevalence of proximal humeral nonunions of 1.1 % (11 out of 1.027 fractures) or up to 8 % if metaphyseal comminution is present and 10 % if there is between 33 and 100 % translation of the surgical neck [16]. In that study, the high percentage of fractures (89 %) treated nonoperatively with a nonunion rate of 0.8 %, in contrast to a 3.4 % nonunion rate in the operatively treated group, should be noted. However, other authors have reported a 20 % rate of proximal humerus nonunions with nonoperative fracture treatment identified as the principal cause [61]. Surgical treatment of fracture sequelae may be challenging due to patient-associated factors, such as advanced age, osteoporosis, medical illness, reduced compliance, alcoholism, and nicotine abuse. In addition, fracture and fracture care-associated factors, such as comminution, severe bone loss, cavitation of the humeral head resulting in a small articulating fragment, insufficient reduction and stabilization, and contracture of the glenohumeral joint following long-term disuse, are of major relevance. Several surgical techniques have been described for the treatment of humeral surgical neck nonunions, including reconstructive procedures using intramedullary nails [63], blade plates [25, 50, 57], locking plates [1], Rush rods with tension-band wiring [23, 41], and intramedullary bone pegging with internal plate fixation [60]. On the other hand, shoulder arthroplasty has been described using anatomic humeral head replacement [2, 8, 23, 32], as well as reverse shoulder arthroplasty [37, 44, 62].