Complications Associated with Locking Plate of Proximal Humerus Fractures

Abstract

Proximal humerus fractures constitute a significant percentage of fragility fractures. The growing use of locking plate has helped treat this problem, but at the same time has brought about complications. Past systematic reviews have documented these complications, however a large number of recent studies have been published since, reporting their own complication rates with different techniques. This study reviews the current complications associated with locking plate of proximal humerus fractures as well as methods to reduce them. A systematic review, following the PRISMA guidelines, was conducted in November 2013 and repeated in March 2015, using PubMed, Scopus, and Cochrane databases, to evaluate locking plate fixation (and complications) of traumatic proximal humerus fractures. Inclusion criteria included adults (>18 years), minimum of 12-month postoperative followup, articles within the last 5 years, and studies with >10 participants. Exclusion criteria included pathologic fractures, cadaveric studies, and nonhuman subjects. Eligible studies were graded using a quality scoring system. Articles with a minimum of 7/10 score were included and assessed regarding their level of evidence per the Journal of Bone and Joint Surgery and Centre for Evidence-Based Medicine guidelines. The initial query identified 51,206 articles from multiple databases. These records were thoroughly screened and resulted in 57 articles, consisting of seven Level 1, three Level 2, 10 Level 3, and 37 Level 4 studies, totaling 3422 proximal humerus fractures treated with locking plates. Intraarticular screw penetration was the most reported complication (9.5%), followed by varus collapse (6.8%), subacromial impingement (5.0%), avascular necrosis (4.6%), adhesive capsulitis (4.0%), nonunion (1.5%), and deep infection (1.4%). Reoperation occurred at a rate of 13.8%. Collapse at the fracture site contributed to a majority of the implant-related complications, which in turn were the main reasons for reoperation. The authors of these studies discussed different techniques that could be used to address these issues. Expanding use of locking plate in the proximal humerus fractures leads to improvements and advancements in surgical technique. Further research is necessary to outline indications to decrease complications, further.

Introduction

Proximal humerus fractures represent a steadily growing problem within the health-care system. Proximal humerus fractures are the third-most common type of fragility fracture, accounting for nearly 6% of all adult fractures.¹,² In addition, as the world's population has aged, the incidence of this fracture type has increased as well.³ Surgical intervention for this fracture type is around 20%, due to the increase in complications as patients age.³,⁴ Surgical fixation with locking plates is the most common type of intervention for displaced proximal humerus fractures, though other options exist, such as closed reduction with percutaneous pinning, hemiarthroplasty, proximal humeral nailing, and reverse total shoulder arthroplasty.⁴,⁵

Locking plate represents a relatively new technology that theoretically supports fixation in the setting of osteoporotic bone.⁶ Its biomechanical properties made it promising in the setting of proximal humerus fractures, where purchase in the humeral head is difficult to obtain, due to large variations in bone density and strength.⁷ Understandably, complications were highly variable as locking plates first began to be used in the proximal humerus fractures. The first systematic review in this setting noted the importance of medial calcar support and the need for more attention to technical aspects of the procedure.⁸ Sproul et al. performed another review with a focus on length of followup, to more accurately capture the time frame necessary for the development of avascular necrosis (AVN) of the humeral head. The study confirmed factors for screw cutout and had similar rates of complications.⁹

Since the publication of these two reviews, there has been an increase in the literature, regarding locking plate fixation for proximal humerus fractures. In addition, studies have attempted to curtail the complication rates mentioned earlier with augmentation strategies such as fibular strut allograft, autograft, cancellous chips, suture fixation of the rotator cuff, and defined technical steps, regarding plate and screw placement. Recent years have shown an increase in the indications for the use of locking plates, as well as reverse total shoulder arthroplasty versus hemiarthroplasty in the setting of proximal humerus fractures.¹⁰,¹¹ Given the relative infancy of locking plate fixation at the time of prior systematic reviews, and the small number of studies included within each review, a more recent systematic review of the literature is warranted. This study examines the current literature to evaluate complications experienced with locking plate in light of changes to operative technique as familiarity with this implant has increased.

Materials and Methods

Following preregistration with PROSPERO (CRD42015019038), a comprehensive search of the literature was performed in November 2013 and repeated in February 2015, to capture recent publications, utilizing the PubMed, Cochrane, and Scopus databases.¹² Database queries were performed using modifiers, limiting results to publications in the English language of the past 10 years, in studies involving human subjects. Search terms were intentionally broad to identify all relevant articles [Table 1].

Table 1.

Search terms used in the PubMed, Cochrane, and Scopus databases

The study design was conducted strictly in accordance with the PRISMA guidelines.¹² The results were subsequently filtered for duplicates, and titles and abstracts were manually screened for relevance and potential adherence to our inclusion criteria. To be included, eligible studies must have been conducted in the past 5 years (modifier in initial query was 10 years to ensure broadness of search), involving 10 or more subjects, adults aged 18 years or older, and a minimum average followup of 12 months. International studies with the English translation were included. Studies involving pathologic fractures, nonhuman subjects (in vitro studies), and cadavers were excluded. Publications with overlapping or duplicate patient populations were excluded.

Next, articles were assigned a quality score using a previously published quality scoring system, which was also used by Sproul et al.⁹,¹³ The scoring system took into account the quality of the study design, as well as the quality of its information. Two reviewers scored the articles and only studies with a minimum score of 7/10 were included. Disagreements were resolved by consensus. Finally, the studies were graded in accordance with the Journal of Bone and Joint Surgery and Centre for Evidence-Based Medicine guidelines, to universally assess the level of evidence of each study. Statistics were performed by authors with training in biostatistics. Complication rates were analyzed in a simple manner, first divided by the total number of fractures treated and followed by stratification by level of evidence. There were no comparative analyses performed due to the heterogeneity of each study.

Results

The initial query conducted through the PubMed, Scopus, and Cochrane databases identified 51,206 citations. After removing duplicates and articles with irrelevant titles and abstracts, a total of 191 full-text articles were assessed for eligibility. From these 191 articles, 57 articles were included to be a part of this systematic review. The results of screening and application of inclusion/exclusion criteria are outlined in Figure 1.¹⁴,¹⁵,¹⁶,¹⁷,¹⁸,¹⁹,²⁰,²¹,²²,²³,²⁴,²⁵,²⁶,²⁷,²⁸,²⁹,³⁰,³¹,³²,³³,³⁴,³⁵,³⁶,³⁷,³⁸,³⁹,⁴⁰,⁴¹,⁴²,⁴³,⁴⁴,⁴⁵,⁴⁶,⁴⁷,⁴⁸,⁴⁹,⁵⁰,⁵¹,⁵²,⁵³,⁵⁴,⁵⁵,⁵⁶,⁵⁷,⁵⁸,⁵⁹,⁶⁰,⁶¹,⁶²,⁶³,⁶⁴,⁶⁵,⁶⁶,⁶⁷,⁶⁸,⁶⁹,⁷⁰

Figure 1.

Flowchart showing selection of studies

There were seven Level 1, three Level 2, 10 Level 3, and 37 Level 4 studies. Level 1 and 2 studies included control groups in regard to surgical approaches, nonoperative treatment, various treatment modalities, or deferring operative techniques. Some Level 3 studies had a basis for comparison when evaluating different surgical approaches or operative techniques. Finally, Level 4 studies were case series without a basis for comparison. Some of these series investigated techniques such as strut allografts, suture fixation, bone grafting, and minimally invasive surgery.

There were a total of 3422 proximal humerus fractures that were treated with locking plate. Certain studies failed to mention or report the presence or absence of complications that were specifically being investigated. If this occurred, the study was not included in the complication's analysis. The most common complication was intraarticular screw penetration (9.5%), followed by varus collapse (6.8%), subacromial impingement (5.0%), AVN (4.6%), adhesive capsulitis (4.0%), nonunion (1.5%), and deep infection (1.4%). Reoperation occurred at a rate of 13.8%. Tables 2-9 display these rates of complications broken down by levels of evidence.

Table 2.

Rate of intraarticular screw penetration

Table 9.

Rate of reoperation

Table 3.

Rate of varus collapse

Table 4.

Rate of subacromial impingement

Table 5.

Rate of avascular necrosis

Table 6.

Rate of adhesive capsulitis

Table 7.

Rate of nonunion

Table 8.

Rate of deep infection

Discussion

Intraarticular screw penetration (9.5%)

Intraarticular screw penetration through the humeral head has been noted as a problematic complication and may lead to additional surgery to revise or remove the screw(s). Two different screw penetrations have been discussed: primary and secondary. Primary screw penetration refers to the intraoperative placement of screws into the glenohumeral joint. Secondary screw penetration refers to the screws that have violated the articular surface as a result of collapse of the humeral head due to varus collapse, AVN, or failure of fixation. Reports from the earlier literature show the prevalence of this complication to range from 0% to 23%.⁷¹ Sproul et al. demonstrated this complication to be at a rate of 7.5%.⁹

Level 1

Fjalestad et al. reported that the majority of these complications occurred in patients with Orthopaedic Trauma Association (OTA) Type C fractures, with evidence of AVN.²⁷ Another study noted that attempts to obtain maximal purchase into the humeral head led to higher rates of primary screw penetrations. The authors adjusted their surgical technique by placing screws 2 mm–3 mm away from the subchondral bone, as was done in another level 1 study.⁴⁸,⁷⁰ In comparing different plates, Voigt et al. found that polyaxial locking screws with blunted ends could be advantageous if screw penetration were to occur.⁶⁵

Level 2

Buecking et al. observed that complications pertaining to the humeral head were higher in their deltoid-splitting approach group, while complications pertaining to the humeral shaft were higher in their deltopectoral approach group.¹⁹ Konrad et al. reported screw penetration, not varus collapse or loosening, as the most common complication with locking plate in their large multicenter study of 270 patients.³⁴

Level 3

In a study comparing locking plate fixation with calcium phosphate cement augmentation versus cancellous bone chips versus no augmentation, Egol et al. demonstrated a significant decrease in intraarticular screw penetration with calcium phosphate cement augmentation. In addition, there was no association between the number of screws in the humeral head and screw penetration. However, age was associated with screw penetration as elderly patients sustained this complication more frequently.²⁶

Level 4

Little et al. described five incidences of asymptomatic screw penetration in their series evaluating 72 cases, for which endosteal augmentation was used. The authors believed that this intramedullary graft decreased the working length of the locking screws.³⁸ Ricchetti et al. reported no cases of screw penetration in their series of 54 fractures followed for 13 months. The authors described using screws 5 mm–10 mm away from the subchondral bone to decrease the risk of screw perforation both primarily and secondarily, should collapse occur. This technique has also been reported in a review article by Ricchetti et al.⁵²,⁷¹ Spross et al. noted that, by placing screws 4 mm–5 mm away from the subchondral bone, complications regarding intraarticular screw penetration decreased significantly. They also advocated the use of fluoroscopy in three planes in an attempt to avoid missing primary screw penetrations.⁶²

Varus collapse (6.8%)

Varus collapse represents one of the more important complications in the setting of locking plate in proximal humerus fractures. Varus collapse is responsible for secondary subacromial impingement and screw penetration into the articular surface of the glenohumeral joint. Sproul et al. reported this as the most common complication in their review, at a rate of 16.3%.⁹ The authors recommended that special attention should be paid to the medial column, which has led some surgeons to place inferomedial support screws, cement, or graft in hopes of lowering this complication rate.

Level 1

In comparing polyaxial versus monoaxial locking screws, Voigt et al. noted an increase in the rate of varus deformity in the group treated with monoaxial screws. The authors felt that polyaxial screws gave more options for screw placement inferomedially.⁶⁵ Zhang et al. postulated that inferomedially placed support screws would resist varus stress to the humeral head, therefore maintaining neck shaft angle in three- and four-part fractures.⁶⁹

Level 2

Evaluating a new carbon fiber–reinforced-polyetheretherketone (CFR-PEEK) locking plate, Schliemann et al. showed a lower rate of varus deformity in comparison to the control titanium locking plate. They believed that this was due to CFR-PEEK being less rigid and having a similar elastic modulus to bone.⁵⁸

Level 3

Lin et al. reported low rates of varus collapse in both their minimally invasive and deltopectoral approach groups. However, they attributed a slightly higher rate within the minimally invasive group due to a false sense of security with progression through weight-bearing exercises because of faster wound healing and smaller scars.³⁷

Level 4

Ricchetti et al. discussed additional contouring of the locking plate as a method to reduce the incidence of varus malunion as it aids in obtaining an anatomic neck–shaft angle. In addition, they placed bone graft for complicated three- and four-part fractures.⁵² Kim et al. performed a study using autologous iliac bone impaction graft with locking plate of four-part fractures and reported 0 incidences of varus collapse in 21 cases over a 27.5-month followup period. The authors believed that their meticulous attention to restoring the medial calcar, obtaining sufficient screw purchase in the inferomedial aspect of the humeral head, and suturing the rotator cuff to the plate led to such positive results.³² Badman et al. reported on 81 proximal humerus fractures, a majority of which were three-part fractures. The authors focused on supplemental suture fixation of the rotator cuff in the five cases of varus collapse and also used structural allograft in the setting of severe osteopenia.¹⁶

Subacromial impingement (5.0%)

Subacromial impingement can be the result of poor intraoperative plate positioning or the sequelae of humeral head collapse. Impingement is frequently symptomatic and may require plate removal. Sproul et al. reported this complication at a rate of 4.8%.⁹ Increased attention to plate placement and preventing varus collapse are the methods surgeons are using to decrease this complication.

Level 1

Only one study mentioned a case of subacromial impingement. Olerud et al. reported one patient in their treatment group of 27 cases with three-part fractures, who subsequently required plate removal. The authors recommended averting humeral head collapse to prevent this complication.⁴⁸

Level 2

Konrad et al. attributed their cases of impingement to placing the plate too superiorly, leading to five cases in their study population of 270.³⁴

Level 3

Lin et al. attributed their cases of impingement to varus collapse, while Jung et al. attributed their one case to intraoperative error.³¹,³⁷ Jung et al. described their operative positioning of the plate following reduction as caudal to the superior end of the greater tuberosity and lateral to the bicipital groove.³¹ Bachelier et al. instead specified plate position 1 cm caudal to the superior aspect of the greater tuberosity.¹⁵

Level 4

Sahu reported no case of impingement, taking the shoulder through a range of motion arc before closure of the wound to detect any symptoms of impingement.⁵⁵ Osterhoff et al. described the majority of their 10 cases of impingement to be strongly associated with medial calcar comminution.⁴⁹ Ricchetti et al. positioned the locking plate 5 mm–10 mm lateral to the bicipital groove and 15 mm–20 mm caudal to the tip of the greater tuberosity. Two patients in their series of 54 cases had postoperative subacromial impingement symptoms.⁵² Finally, Aggarwal et al. described provisionally fixing the plate with K-wires and placing the shoulder through a range of motion arc under fluoroscopy. This technique resulted in five cases of impingement in their series of 47.¹⁴

Avascular necrosis (4.4%)

AVN has been a historic concern with proximal humerus fractures. Sproul et al. reported this rate to affect 10.8% of patients.⁹ This has even led to studies attempting to discover predictors of necrosis, following a proximal humerus fracture. Hertel et al. determined that the most relevant factors included integrity of the medial hinge, length of the dorsomedial metaphyseal extension, and different fracture types.⁷² Furthermore, growing indications for reverse total shoulder arthroplasty and recent favorable studies in comparison to hemiarthroplasty may have artificially decreased the rate of AVN currently being reported.⁷³ In addition, AVN may present later in followup and inadequate lengths of followup would, in turn, under report this complication. Recent studies have attempted to use deltoid-splitting or minimally invasive approaches with the belief that less soft tissue disruption in proximity to the humeral head would preserve its blood supply. Finally, there is growing belief that asymptomatic cases of AVN can potentially over-report this serious complication.

Level 1

Comparing the minimally invasive approach to the deltopectoral approach, Liu et al. reported one case of AVN in the latter group and zero in the former. The authors believed that the minimally invasive approach decreased soft tissue stripping and preserved the blood supply around the proximal humerus.³⁹ Zhang et al. reported only one case of AVN in their study that focused on medial support screws using a deltopectoral approach. In their opinion, preventing medial collapse also aided in preventing AVN.⁶⁹ Interestingly, the findings from Fjalestad et al. showed that nonoperatively treated patients had a higher rate of AVN than those in the operative group. All patients had displaced three- and four-part fractures.²⁷

Level 2

Buecking et al. reported no case of AVN and no difference between deltoid-splitting and deltopectoral approaches. Followup, however, was only for 1 year.¹⁹ Schliemann et al. reported a lower incidence of AVN in patients treated with their CFR-PEEK implant compared to conventional locking plate. Their followup was for a minimum of 2 years.⁵⁸

Level 3

Martetschlager et al. reported higher rates of AVN in patients treated with a deltopectoral approach compared to a minimally invasive deltoid-splitting approach. With a mean followup of nearly 4 years, AVN was diagnosed in six of 33 patients in the deltopectoral approach group and one of 37 patients in the deltoid-splitting approach group.⁴¹ Wu et al. reported similar findings over a mean followup of 2.5 years in comparing a minimally invasive approach to a deltopectoral approach.⁶⁷

Level 4

Using a minimally invasive plating technique and a mean followup of nearly 3 years, Chen et al. reported only one case of AVN in their series of 64 cases.²¹ Little et al. reported low rates of AVN as well by using a deltoid splitting approach and a medial strut allograft.³⁸ On the other hand, Spross et al. reported 20 cases of AVN in their large case series of 294 followed for 1 year. The patients were treated using a deltopectoral approach. The authors determined, however, that fracture type influenced whether AVN occurred or not, with fracture dislocations having the highest rate.⁶²

Reoperations (13.8%)

Reoperations are a very important measure of how successful the index operation was and also highlight the most significant complications. Reoperations also highlight possible improvements in surgical decision-making or technique to avoid certain complications. Even in regard to AVN, more meticulous soft tissue management, attempts at minimally invasive techniques, and consideration of arthroplasty as primary surgery have led to a decrease in reoperation. It is also important to make a distinction between “planned” versus “unplanned” operations as many patients do request to have hardware removed. Hardware removal has been associated with a very low complication rate and high patient satisfaction as indicated in a recent case series.⁷⁴ This should be differentiated from the need to undergo an arthroplasty procedure due to failed primary open reduction and internal fixation (ORIF).

Level 1

Cai et al. reoperated on three of 12 patients following locking plate. The patients originally had four-part fractures and reoperations were during the 2^nd year of followup. Plates were removed for fixation failure and revision internal fixation for nonunion.²⁰ Zhu et al. performed five screw revisions due to primary screw penetration.⁷⁰ Voigt et al. attributed the majority of reoperations in their study due to secondary displacement of the greater tuberosity.⁶⁵

Level 2

Buecking et al. reported a large number of reoperations: three screw revisions, 18 plate removals, four revision ORIF, and seven arthroplasties in their study population of 90.¹⁹ Seventeen of the plate removals were at the request of the patient and the rest were due to screw perforation, implant loosening, or infection. Schliemann et al. performed seven plate removals with arthrolysis in two of those cases.⁵⁸

Level 3

Kralinger et al. reported mechanical failure as a strong predictor of reoperation in their study consisting of majority three- and four-part fractures. Two revision arthroplasties, six capsular releases, six revisions of internal fixation, 14 plate removals, and one hematoma evacuation were performed.³⁵ Sanders et al. discussed screw revisions and plate removals secondary to intraarticular screw penetration and impingement, respectively, as a major reason for their 50% reoperation rate.⁵⁶

Level 4

Ockert et al. noted an unplanned reoperation rate of 14% and a planned reoperation rate (due to impingement, patient request, or range of motion deficit) of 40% in its series of 43 patients followed for 10 years.⁴⁶ Kim et al. reported 2 implant removals for cultural reasons in their case series of 21 four-part fractures followed for 27.5 months.³² Finally, Schliemann et al. reported impingement and screw penetration as the primary reason their revision rate was close to 30%.⁵⁷

The data presented in this systematic review not only support data from past reviews but also present potential solutions proposed by investigators, in the hopes of decreasing the complication rate associated with locking plate of proximal humerus fractures. Recent reviews have emphasized the importance of AVN and fracture dislocation patterns negatively impacting outcome. Complex, intraarticular fracture patterns have high complication rates when treated with locking plate. Brorson et al. also noted that the methodological quality of studies is lacking.⁷⁵ Tepass et al. noted that three- and four-part fractures actually had better outcomes when treated with head preserving surgery compared to a hemiarthroplasty and that there were an increase in the number of complications as the fracture complexity increased.⁷⁶ Finally, in a review specifically looking at referrals for complications, Jost et al. discussed the importance of making the primary surgery the definitive surgery. A majority of the patients received arthroplasty as a revision surgery, secondary to complications from locking plate. In these patients, primary reduction was not achieved, indicating that the more complex fracture patterns may not necessarily be amenable to locking plate.⁷⁷

None of the articles presented in this review were in the most recent comprehensive systematic review, as Sproul et al. completed their literature search in 2009. In addition, none of the articles from the previous systematic reviews are in this review as we only included the most recent articles. We repeated our queries to capture the most recent literature and data, noting that there were quite a few articles we would not have been able to include. More experience with locking plate in treating proximal humerus fractures and the application of newer techniques has definitely adjusted complication rates. Moreover, the complication rates may not have been entirely accurate in the previous reviews as they were analyzing a smaller number of total cases. Thus, one of the aims of this review was to encompass as many recent articles without sacrificing quality, which we accomplished by including only high scoring articles into this study.

It is also worthwhile to discuss the fact that there is a large amount of literature describing nonoperative treatment of proximal humerus fractures. Concerning the number of complications as well as costs associated with surgical treatment, there have also been studies comparing operative versus nonoperative treatment of these fractures. Handoll et al. found in their Proximal Fracture of the Humerus: Evaluation by Randomization trial that surgical treatment does not result in improved outcomes in most patients and that it is not cost effective.⁷⁸ This lends support to the argument that every fracture should be treated on a case-by-case basis. The osteoporotic nature of some of these fractures leaves it incredibly difficult to treat, and though locking plate has been promising in theory, not every plate is the same and cannot replicate force distributions of the proximal humerus. The most used plate seemed to be the Synthes PHILOS plate; however, not every article in our review reported which plate they used. The design of the plate cannot prevent varus collapse and subsequent cutout of the screws. Biomechanical studies have shown this and the importance of a medial buttress to prevent those complications.⁷⁹

There are a number of limitations to this review. First, minimum followup was set to 12 months, which may have underestimated the incidence and prevalence of AVN. Not every article commented on every complication analyzed in this report. This could have led to either under-reporting or over reporting of results. Next, the heterogeneity of the articles, whether it was based on level of evidence, type of fracture, approach, or specific technique used, could not be fully accounted for. In addition, we did not perform a metaanalysis of the complication rate. Our aim was to provide a broad overview of complications with proposed methods to decrease complication rate. A meta-analysis was not our goal as we cannot equally compare each study, such as comparing a three part fracture with a four-part fracture. An attempt was made to highlight the most important conclusions from each article. Finally, giving more weight to articles with higher levels of evidence may have skewed the results as well.

Conclusion

The points of consideration from this review, in regard to the major complications associated with locking plate in proximal humerus fractures, are as follows:

Screw penetration

Greater care with fluoroscopy, use of at least two perpendicular planes to confirm screw is not within the glenohumeral joint. Placement of screws that are too short of subchondral bone should be avoided.

Varus collapse

Ensure the medial column is intact (medial hinge). Consideration of the use of strut allograft, bone graft, suture augmentation, and plate contouring. Consider placement of inferomedial support screws.

Subacromial impingement

Ensure the plate does not sit too proximally, AVN, Consider fracture type to stratify risk of AVN, Careful soft-tissue dissection, Consider minimally invasive techniques.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.