Avascular Necrosis and Posttraumatic Arthritis After Proximal Humerus Fracture Internal Fixation: Evaluation and Management

Abstract

Purpose of Review

Avascular necrosis (AVN) and posttraumatic arthritis (PTA) are common complications following both conservative treatment and open reduction and internal fixation (ORIF) of proximal humerus fractures (PHFs). Despite the frequent utilization of ORIF, information regarding these leading causes of failure is limited. This review includes a discussion of incidence, risk factors, and evaluation of AVN and PTA following PHF. The mechanisms of treatment options and associated outcomes are also reviewed.

Recent Findings

Recent best available evidence demonstrates significant rates of AVN and PTA following ORIF of PHF. This is particularly true of complex fracture patterns. A thorough workup is required in the setting of failure caused by AVN and PTA. This includes a careful patient history, clinical exam, plain film radiographs, and CT scans. EMG and/or aspiration may also be indicated. Special consideration is given to the examination of the deltoid muscle, neurovascular status, rotator cuff function, and the possibility of infection. Biological supplementation, anatomic total shoulder replacement (aTSA), and fusion are rarely employed in the treatment of AVN and/or PTA. Due to satisfactory patient outcomes, reverse total shoulder replacement (rTSA) has increased in popularity for the elderly population, while hemiarthroplasty (HA) may be appropriate for some young, active patients.

Summary

With careful patient selection and meticulous surgical technique, AVN and PTA can be mitigated. Careful indications for ORIF may decrease the frequency of these complications. For most patients, rTSA is the optimal treatment option. Given the frequent utilization of ORIF and the higher than acceptable complication and failure rates, AVN and PTA warrant our attention.

Introduction

The frequency of proximal humerus fractures (PHF) continues to rise, mirroring our aging population. Open reduction and internal fixation (ORIF), the most commonly employed surgical choice for PHF, has demonstrated inconsistent results, with failure rates routinely around 30% and complication rates ranging from 44 to 55% among various institutions [1••, 2]. With PHF making up a third of all fractures in patients older than 65 years old, it is important that we anticipate and manage complications after ORIF [3–5].

While the vast majority of proximal humerus fractures, particularly in the elderly, can be managed nonsurgically, selected displaced fractures and delayed unions may benefit from ORIF [3]. However, the type of surgical treatment for more complex fracture types and patient characteristics remains controversial. The introduction of locking plate fixation led to an increase in frequency of ORIF [1••, 2]. While locking plates may have decreased the rate of “mechanical failure” (hardware failure, screw cutout, varus collapse, plate fracture), a large subset of postoperative complications are “biological” in nature. These complications may reflect the long-term natural history of the fractured shoulder [1••, 6]. Avascular necrosis (AVN) and posttraumatic arthritis (PTA) after ORIF are two common complications that lead to failure. This review will focus on the evaluation and management of these two common complications after ORIF.

The incidence of AVN is reported to be 6–17% among patients with PHFs treated with ORIF [1••, 2, 7, 8]. The rate of AVN increases to 30–40% in three- and four-part fractures [1••, 8]. PTA, which is even less commonly reported in studies, has an incidence that ranges from 9 to 38% [1••, 9]. In many cases, particularly when changes are “end stage,” it may be difficult to differentiate these two etiologies.

The low numbers of cases and broad range of incidence rates reported are due to a variety of factors. The follow-up time within the vast majority of studies does not extend long enough to allow for the detection of these complications [10, 11]. Studies with a longer follow-up period reported that less than half of failures were detectable at 6-month follow-up; on average, these cases were not diagnosed until 12–16 months after operative fixation [1••, 6]. Additionally, in some cases, symptoms of these complications can be mild, particularly in low demand older adults. Thus, the incidence of these long-term complications may be underestimated.

Avascular Necrosis

Risk factors for AVN following PHF have been extensively reviewed in the literature. Most risk factors may be considered non-modifiable since they are closely related to the type of fracture pattern. Complication and failure rates due to AVN increase with the complexity of fractures [8, 12]. Our study of 173 patients over the age of 60 years demonstrated a 26% failure rate in 2-part fractures, 39% in 3-part fractures, and 45% in 4-part fractures; the main complication that led to failure was AVN (52% of failures) [1••]. Boesmueller et al. demonstrate a similar pattern even among a younger patient cohort [8]. Specifically, AVN has been reported to occur in up to 77% of four-part fractures (Fig. 1) [13]. Some studies show that age had no influence on the rate of AVN, while fracture type demonstrated a positive correlation [8, 14].

Fig. 1.

Grashey (A) radiograph of a 66-year-old woman with a four-part PHF dislocation; (B) this radiograph was taken 6 months following ORIF. There is evidence of avascular necrosis and destruction of the joint; (C) removal of hardware was completed; (D) given the severe avascular necrosis and posttraumatic arthritis, a reverse total shoulder arthroplasty was performed. ORIF, open reduction and internal fixation; PHF, proximal humerus fracture

The intraoperative perfusion of the humeral head has been studied in an attempt to predict the risk of postoperative AVN. Some studies have indicated that the blood flow from the posterior circumflex vessels is compromised in the setting of a shortened metaphyseal head extension. In studies by Hertel and colleagues, a short calcar segment of < 8 mm, disruption of the medial hinge, and increasing number of fracture fragments (parts) were associated with higher rates of AVN [7, 12, 15]. In the setting of a humeral head dislocation, the periosteum (and vessels) are stripped; a displacement of > 4 mm has been found to be associated with higher complication rates [7].

Posttraumatic Arthritis

Intra-articular impaction injury in PHF is relatively rare. Consequently, literature on isolated PTA after ORIF is limited. Instead, PTA after ORIF is often related to postoperative sequelae that lead to chondral damage. Such causes include malunion and nonunion causing joint surface incongruity, screw perforations, recurrent instability, and/or late-stage AVN [16]. Failure rates due to nonunion are reported to range from 5 to 15%, while malunion causing failure ranges from 2 to 31% [1••, 9, 17]. If bone fails to heal normally, joint surface incongruity and bone loss may occur leading to arthritic changes within the joint [1••, 16, 18].

Recent studies report intra-articular screw penetration as a cause of 1–10% of failures [1••, 10, 19]. Prominent screws within the joint may lead to rapidly progressive and severe arthritis [20]. While primary screw penetration is caused by misplacement of the screws intraoperatively, secondary screw penetration refers to the screws that invade the articular surface due to humeral head collapse. Humeral head collapse can be caused by AVN, varus collapse, or general failure of fixation (Fig. 2) [10].

Fig. 2.

A 51-year-old woman with failed ORIF of a 3-part PHF (surgical neck and greater tuberosity) as well as a bony Bankart lesion that was treated with Latarjet (Grashey (A) and axillary lateral (B) views). Images show evidence of humeral head collapse and screw prominence, likely secondary to avascular necrosis. The patient had concomitant deltoid weakness and an EMG was performed to assess the axillary and suprascapular nerves (which were normal); (C) a reverse total shoulder arthroplasty with hardware removal was performed. ORIF, open reduction and internal fixation; PHF, proximal humerus fracture.

While posttraumatic sequelae following ORIF can lead to chondral damage, they are rarely distinguished as isolated arthritis. One reason for the lack of effort to distinguish this complication may be because in most cases, it does not alter treatment choice, which is most commonly shoulder arthroplasty.

Evaluation

Symptoms of AVN and PTA are variable. Symptoms are often insidious and may include pain, decreased range of motion, crepitus, and weakness (Fig. 3). Evaluation of the failed PHF fixation begins with a careful history and physical exam.

Fig. 3.

A 61-year-old female with a comminuted PHF dislocation (Grashey (A) view); (B) she was treated with ORIF; (C) after 16 months, she returned with significant pain and limited range of motion. The radiograph shows evidence of avascular necrosis and posttraumatic arthritis. Given her mild symptoms, she is being monitored without revision. ORIF, open reduction and internal fixation; PHF, proximal humerus fracture

During the patient interview, it is important to inquire about pre-fracture shoulder pain and management (injections, physical therapy, surgery, etc.) to identify the patient’s baseline. Preoperative and postoperative details should be evaluated. This should include imaging, neurovascular status, shoulder pain, and pain management. Time from fracture to surgery, particularly when prolonged by more than 3 months, is important. Information regarding postoperative wound healing and drainage, operative notes, and any findings at external facilities should also be reviewed [21].

Axillary Nerve Injury

The clinical evaluation should focus on several critical factors. With incidences ranging between 6.2 and 67%, nerve dysfunction after PHF is relatively common. The axillary nerve is the most commonly injured nerve [22]. Axillary nerve function can be assessed by evaluating the contractility and strength of the deltoid muscle. Inspect the patient for signs of deltoid atrophy, which can be indicated by an increased acromion prominence. Examination should include active abduction of the arm against resistance [23], as well as active extension against resistance. It is important to note that the lack of sensation to light touch over the lateral aspect of the shoulder is sometimes unreliable in detection of an axillary nerve injury [24].

If the dysfunction is related to the injury, it is commonly a neurapraxia and is typically self-limited. However, if the recovery does not occur within 3 months, an electromyography (EMG) is indicated [14]. Conversely, if there is a risk of iatrogenic injury to the nerve, some have recommended that early exploration improves the chances of nerve recovery [25]. Prior to considering any surgical management, it is crucial to maintain a low threshold for EMG evaluation and consultation with a brachial plexus/peripheral nerve specialist.

Wound Healing

Incision healing is another factor that must be evaluated. Noting all previous surgical scars is an important part in operative planning. Extensive scarring within the area of interest serves as an distinguishable hint that the normal anatomy has been altered (Fig. 4) [26]. Importantly, incisional healing should be evaluated for any signs of infection.

Fig. 4.

A 71-year-old female with failed ORIF (Grashey (A) and axillary (B) views). She developed persistent instability which was addressed unsuccessfully by her previous surgeon via capsular repair, Bankart repair, and screw removal. Images show subluxation, avascular necrosis, and intra-articular screw penetration; (C) while substantial scarring and poor bone quality complicated her reconstruction, final images demonstrate acceptable position and tuberosity union following reverse total shoulder arthroplasty. ORIF, open reduction and internal fixation; PHF, proximal humerus fracture

Infection

In all cases of failed ORIF for PHF, infection must be ruled out. The classic inflammatory signs of infection may not always be present, given the indolent nature of some shoulder infections. Signs of infection may be subtle, with postoperative pain or stiffness sometimes being the only symptoms [27]. During the patient interview, it is important to obtain information that could indicate an underlying infection. It is also necessary to inquire about the history of postoperative wound drainage, surgical irrigation and debridement, and prolonged postoperative antibiotics [21].

Infection may dramatically alter surgical plans [28, 29]. Thus, in virtually all cases of failed ORIF for PHF, ultrasound or fluoroscopically guided aspiration of the joint for microbiology, cell count, and differentiation is performed [1, 28, 30]. Additionally, cultures must be held long enough (typically a minimum of 2 weeks) to exclude an infection by Cutibacterium acnes [21].

Imaging

In all cases of ORIF failure, evaluation of Grashey (internal rotation and external rotation), scapular y, and axillary views is necessary. Images should be reviewed for healing, hardware failure, bony deformities, and glenoid changes [26]. X-rays are useful in the staging of osteonecrosis, which has been well defined in the literature [31–33]. After an initial stage that is not seen with plain radiographs, sclerotic changes are identified in stage II. Stage III is characterized by subchondral collapse and a change in the spherical shape of the humeral head. The humeral head is further collapsed and degenerative changes within the glenoid fossa are seen in stage IV and stage V, respectively [32, 33].

Evidence of PTA on X-ray may include narrowing joint-space, osteophytes, sclerosis, and articular cartilage loss which are best seen with the axillary view. In the vast majority of cases, computed tomography (CT) scans with metal suppression and three-dimensional reconstruction are performed to enhance radiographic detail [21, 26]. This will allow a more accurate prediction of bone union. Sagittal reconstructions of CT scans are utilized to assess the rotator cuff. With 40% of proximal humerus fractures associated with rotator cuff tears, it is important to consider rotator cuff integrity [34]. In addition, CT may aid in the localization of focal articular defects and subchondral cystic changes [30].

Surgical Treatment and Outcomes

Bone Grafting and Core Decompression

Bone grafting and core decompression are joint sparing approaches that can be used to treat AVN. Core decompression promotes healing of necrotic bone. By decreasing the intraosseous pressure, this intervention restores blood flow to the fragment. Decompression may be appropriate if it is performed at earlier stages of the disease, prior to collapse [29, 35, 36•]. Laport et al. demonstrated that after an average follow-up of 10 years since receiving core decompression, 91% (30 of 33 patients) with stage I or II had positive results. Seventy percent of patients with stage III had successful results [37]. However, successful results were not attained by patients with stage IV disease, with only 14% of shoulders experiencing successful results. Furthermore, most studies focus on idiopathic or steroid-induced AVN, and it is possible that these results do not apply to AVN after trauma.

Core decompression and/or bone grafting can be completed for lower stage AVN. Arthroscopic and fluoroscopic assistance can aid in the localization of the lesions, which should only be done in “precollapse” stages. Careful attention should be paid to avoiding injury to the biceps tendon and axillary nerve with these “percutaneous techniques.”

There are limited reports on vascularized grafts for proximal humerus. Some studies have investigated the use of autologous stem cells in the treatment of AVN within the proximal humerus. Heringou et al. compared simple core decompression and core decompression with cell therapy. Among 30 shoulders that were treated with core compression with transplanted mesenchymal stem cells, only 3 (10%) shoulders progressed to humeral head collapse, compared to 25 shoulders (74%) of 34 shoulders after simple core decompression [36•].

This treatment option could avoid or postpone the collapse or arthroplasty [29, 36•]. There are limited studies on these techniques in the setting of previous ORIF. Furthermore, in many cases, AVN is not recognized until the humeral head has collapsed. Therefore, core decompression with or without grafting should only be considered to relieve pain in young patients who have localized AVN that is still in its early stages. An intact rotator cuff and limited degeneration of glenoid cartilage or chondral changes is also required [29]. Further data is needed to evaluate the role of these techniques in the setting of ORIF of PHF.

Arthroplasty

While rTSA has largely replaced hemiarthroplasty (HA) as the main treatment choice, HA is still an option for select patients with failed ORIF due to AVN and/or PHF. In comparison to the elderly population, younger patients are more likely to have intact rotator cuff function, superior bone quality, and increased potential for tuberosity healing [38]. Most failures of HA are related to poor tuberosity healing or rotator cuff dysfunction. Arthritic changes to the glenoid would also typically preclude a successful outcome with hemiarthroplasty. Thus, only young patients (< 50 years old) with the following criteria may benefit from hemiarthroplasty: a functional deltoid and rotator cuff, humeral cartilage articulation that is either unsalvageable or unlikely to be reconstructed, and tuberosities that are likely to undergo anatomic healing or are already healed in a good position [21, 38, 39].

While HA allows for the preservation of the glenoid bone stock, anatomical total shoulder arthroplasty (aTSA) provides better pain relief and function when compared to HA for AVN and glenohumeral arthritis. In a study comparing HA and aTSA performed for posttraumatic AVN, reoperation rates were higher among the patients who underwent HA (24% vs 10%) [40]. While it may not apply to the post-ORIF cohort, the revision rate following HA and aTSA for osteoarthritis in general is reported to be 13% and 7% respectively [41]. Bartelt et al. reported a 72% survivorship for HA and 92% survivorship for aTSA at 10 years [42]. Future studies that focus on the post-ORIF cohort are necessary.

Despite these appealing features, glenoid component loosening is a common cause of failure of aTSA [43, 44]. Studies report glenoid loosening occurring at rates of 15.7–38% among patients who received aTSAs [45, 46]. By avoiding the placement of a glenoid component, HA mitigates this frequent complication that occurs in young and active patients. In comparison to the elderly, young patients experience inferior postoperative outcome following rTSA [47–49]. Thus, despite the low likelihood of returning to baseline, HA remains an alternative for certain young patients with AVN or PTA following ORIF. Importantly, these patients must be counseled on the risks and benefits.

In summary, aTSA can be considered in very rare cases of end-stage arthropathy after failed ORIF of PHF, and minimally displaced PHFs treated nonsurgically. As discussed, aTSA is associated with high complication rates especially among younger, more active patients. The overall success of aTSA depends heavily on whether the rotator cuff is functional.

Unfortunately, the rotator cuff (more specifically the tuberosities) is commonly abnormal in PTA and AVN especially among older patients [50, 51].

rTSA is the most common surgical option for PTA and AVN after nonsurgical and surgical management of PHF (Fig. 5). rTSA is a valid treatment option in primary and secondary treatment of PHF [9, 28, 39, 52••,53, 54, 55•]. However, there are only a few studies that evaluate the outcome of rTSA specifically after failed ORIF of a PHF. Grubhoefer et al. evaluated 44 patients who received rTSA following a failed ORIF for a complex PHF. They reported an improvement of a mean Constant score from 26 to 55 points, with 80% self-reporting their outcome as either excellent or good.

Fig. 5.

A 57-year-old woman with a 2-part PHF (severe translation of the humeral shaft) and minimally displaced fractures of the greater and lesser tuberosity (Grashey (A) and axillary (B) views); (C) following hardware removal for a failed ORIF with fibular strut allograft, the patient developed progressive avascular necrosis with dissociation of the humeral head from the shaft; (D) a reverse total shoulder arthroplasty was completed. ORIF, open reduction and internal fixation; PHF, proximal humerus fracture

Many studies have suggested primary rTSA as a superior treatment for patients who are not ideal candidates for ORIF. Improved functional outcomes have been shown among patients who received primary rTSA for PHF [2, 9, 39, 54]. They also note that patients who were initially treated with rTSA had fewer complications and better outcomes than those who had salvage rTSA following sequela or failed ORIF treatment [9, 39, 52••,53, 54, 55•].

The most common contraindication for rTSA after PHF is axillary nerve injury. In these cases, careful preoperative discussion with a nerve or brachial plexus specialist and the patient should be completed. In some cases, a radial-to-axillary nerve transfer can be performed to restore deltoid function [56]. There are some advocates for reconstruction of the anterior deltoid via a pedicled pectoralis major transfer [57, 58]. In addition, infection needs to be treated prior to rTSA. In most cases, this is through a 2-stage approach (resection and spacer with delayed arthroplasty).

Fusion

While the vast majority of patients with AVN or PTA after PHF are treated with arthroplasty options, there are some cases which may benefit from a fusion. Fusion of the humeral head to both the glenoid and acromion will provide pain relief at the expense of motion limitation. Fusion is typically only considered if the patient has a nonfunctional (and nonreconstructible) deltoid [21].

Conclusion

AVN and PTA are leading causes of failure following ORIF for PHF. As discussed, these complications are associated with certain fracture types and other postoperative sequela. To substantially decrease the risk of AVN and PTA, careful consideration of surgical treatment option should be completed. ORIF should be avoided in certain fracture types; due to the high risk of AVN, ORIF should not be performed on elderly patients with three- and four-part fractures with or without an associated head dislocation.

Salvage of failed ORIF of PHF due to AVN and/or PTA is complex. A thorough workup includes an assessment of the possibility of infection, axillary nerve function, and rotator cuff integrity, which may drastically alter surgical options. Furthermore, to decide on the appropriate surgical management, it is critical to consider patient demographics and goals.

By restoring blood flow to the fragment, core decompression, and/or bone grafting may promote necrotic bone healing. While this treatment has shown promise in postponing collapse or arthroplasty, this treatment option is limited to only young patients with lower stages of AVN. Conversely, HA is a more common treatment that is also reserved for younger patients who are more likely to have an intact rotator cuff, better bone quality, and a higher likelihood for tuberosity healing.

While aTSA may provide superior pain relief and function when compared to HA, glenoid loosening is a frequent complication that limits this treatment option to very rare cases of end-stage arthropathy among older adults. Fusion of the humeral head is another option that is rarely utilized due to the resulting motion limitation.

Given the limited indications of these treatments after AVN/PTA following ORIF for PHF, rTSA is the most reliable option for most patients. While revision rTSA provides improved function, primary rTSA for fracture has superior outcomes and lower complication rates. Thus, in older patients at high risk for AVN and PTA, rTSA should be considered as treatment of acute PHF, while HA may remain a treatment option following risk-benefit analysis of a young patient.

Declarations

Human and Animal Rights and Informed Consent

This article does not contain any studies with human or animal subjects performed by any of the authors.

Conflict of interest

Jonathan D. Barlow receives royalties and acts as a consultant for Stryker. Alexandra Cancio-Bello declares that she has no conflicts of interest. No funding was received.