Currently, there is little information concerning periprosthetic humeral fractures after shoulder arthroplasty. The few reports in the literature have included a limited number of patients1-6. The prevalence of these fractures has been reported to be between 1.6% (seven of 436)1 and 2.4% (six of 252)5. Six of seven fractures in the report by Boyd et al.1 did not heal with nonoperative treatment. All seven fractures were centered about the tip of the prosthesis. Campbell et al.4 reported on five fractures that healed with nonoperative treatment. Four of those fractures were distal to the tip of the humeral stem. Worland et al.5 reported on six patients who sustained a periprosthetic humeral fracture after shoulder arthroplasty. One fracture, which occurred distal to the tip of the humeral stem, healed with nonoperative treatment. The remaining five fractures, which occurred about the stem, either had immediate operative treatment or required operative treatment after the failure of nonoperative treatment. Because of the limited amount of information that is currently available in the literature, we reviewed our experience to determine the etiology of these fractures, the results of their treatment, the risk factors for an unsatisfactory outcome, and the rates of reoperation in order to propose an algorithm for their treatment in the future.
Materials and Methods
The present study was carried out under a protocol approved by the institutional review board at our institution. Three thousand and ninety-one shoulder arthroplasties were performed at our institution between 1976 and 2001; 56% of the procedures were performed in women. There were 2692 primary shoulder arthroplasties and 399 revision procedures. The average age at the time of the arthroplasty was 63.2 years. The average duration of follow-up was 6.7 years. Thirty-five patients were not followed at our institution, and 171 patients were followed for less than three months.
Nineteen patients subsequently sustained a postoperative humeral shaft fracture. Three patients were excluded from our study because a complete series of radiographs was not available for review. According to the clinic notes, all three of these patients had a distal humeral shaft fracture and all three fractures united with nonoperative treatment. The clinical course of the remaining sixteen patients was studied by means of a retrospective review of the charts, operative reports, and radiographs.
The study group included six men and ten women. The average age of these patients at the time of the arthroplasty was sixty-three years (range, thirty-seven to seventy-six years). The fracture involved the right humerus in seven patients and the left humerus in nine. The initial arthroplasty was performed because of osteoarthritis in five patients, rheumatoid arthritis in five, a failed arthroplasty in two, avascular necrosis in one, posttraumatic arthritis in one, and a nonunion of a proximal humeral fracture in two. None of these patients had an ipsilateral elbow arthroplasty.
Ten patients had a total shoulder arthroplasty, and six had only a humeral head replacement. The decision to perform hemiarthroplasty as opposed to total shoulder arthroplasty was based on the condition of the articular cartilage of the glenoid and the integrity of the rotator cuff. One total shoulder arthroplasty and one humeral head replacement were revision procedures. The humeral components included seven Neer prostheses (3M, St. Paul, Minnesota) and nine Cofield prostheses (Smith and Nephew, Memphis, Tennessee). The glenoid components included six Cofield and four Neer implants. The decision to use cement fixation as opposed to press-fitting of the components was based on the quality of the cancellous bone. The humeral component was cemented in five cases, and the glenoid component was cemented in six.
Two patients with substantial osteolysis and loose components had moderate to severe pain before the fracture. The remaining patients had no pain or occasional pain. Before the fracture, the average active shoulder abduction was 117° (range, 35° to 160°) and the average external rotation was 42° (range, 25° to 70°).
The average time from the arthroplasty to the fracture was forty-nine months (range, one to 146 months). Four patients sustained the injury during the recovery period (less than six months after the index arthroplasty), at an average of sixty-six days (range, thirty-six to 102 days) postoperatively.
Twelve fractures occurred as the result of a fall on the affected upper extremity. The remaining four fractures occurred as the result of other mechanisms: one occurred without any apparent trauma, one occurred during a motor-vehicle accident, one occurred as the result of forceful use of the upper extremity, and one occurred during manipulation of the shoulder with the patient under general anesthesia six weeks after the arthroplasty.
All patients were followed until union. A fracture was thought to have healed when there was both clinical and radiographic evidence of union. Clinical union was defined as a painless fracture site on examination. Radiographic union was defined as evidence of bridging bone on two views without evidence of hardware failure. Subsequent follow-up after union was performed at regular intervals as part of routine ongoing follow-up after the shoulder arthroplasty. Patients who were unable to return for evaluation were sent a questionnaire for the evaluation of function and satisfaction. Additionally, patients were asked to have a local orthopaedic surgeon send the results of a clinical examination and recent radiographs. Ten of the sixteen patients had the latest follow-up examination at our institution. The other six patients chose to send us their most recent radiographs along with a clinical evaluation from their local orthopaedic surgeon.
The results at the time of the latest follow-up were graded according to the systems of Neer et al.7 and Cofield8. The result was considered to be excellent if the patient had no or slight pain, had external rotation to at least 45° and active abduction to at least 140°, and was satisfied with the procedure. The result was considered to be satisfactory if the patient had no or slight pain or moderate pain only with vigorous activity, had external rotation to at least 20°, had active abduction to at least 90°, and was satisfied with the procedure. If any of these criteria were not met, the result was considered to be unsatisfactory.
Radiographic Review and Results
The prefracture radiographs were studied to evaluate osteopenia, cortical erosion over time, cortical penetration at the time of reaming during the arthroplasty, alignment of the prosthesis, and the presence of osteolysis.
Osteopenia was graded on the basis of the ratio of the combined width of the mid-diaphyseal cortices to the diameter of the diaphysis, as described by Campbell et al.4. The bone was graded as normal if the ratio was >50%, mildly osteopenic if the ratio was 25% to 50%, and severely osteopenic if it was <25%. Before the injury, nine patients had mild osteopenia and seven had severe osteopenia.
In three patients, the humeral stem had gradually shifted in position as a result of endosteal erosion of the lateral cortex. The humeral component had been inserted without cement in these patients. In one patient, a portion of the medial humeral cortex had been reamed at the time of the index arthroplasty. Two patients had severe osteolysis with component loosening. Additionally, three patients had complete radiolucent lines in all zones on radiographs but were asymptomatic before the injury.
The fractures were classified on the basis of their location with respect to the distal tip of the humeral implant, according to the system of Wright and Cofield6. Type-A fractures are located at the tip of the prosthesis and extend proximally, type-B fractures lie at the tip of the prosthesis without extension or with only minimal extension proximally but can have a variable amount of extension distally, and type-C fractures are located distal to the tip of the prosthesis. Fractures also were classified as comminuted or not comminuted and were categorized according to fracture pattern as transverse, oblique, or spiral. In addition, the amounts of angulation and displacement were assessed. Angulation was classified as mild (≤15°), moderate (15° to 30°), or severe (>30°). Similarly, displacement was classified as mild (within one-third of the diameter of the humeral shaft), moderate (one-third to two-thirds of the diameter of the humeral shaft), or severe (beyond two-thirds of the diameter of the humeral shaft).
Six fractures were spiral, four were oblique, and six were transverse. Six fractures were type A, six were type B, and three were type C. The remaining fracture was a transverse fracture that was located in the proximal diaphyseal area in a patient with severe osteolysis and a loose implant. Because it was proximal to the tip, this fracture could not be classified with the system of Wright and Cofield6. Neither of the fractures that occurred in patients who had had the arthroplasty because of a proximal humeral nonunion involved the site of the original nonunion.
Angulation was mild or absent in ten patients, moderate in two, and severe in four. All six patients who were treated nonoperatively had no or only mild angulation. Five fractures were nondisplaced, six were mildly displaced, and five were severely displaced. Only one of the six patients who was treated nonoperatively had severe displacement.
All fractures united (see Appendix). Six patients (Cases 1-6) were treated successfully with nonoperative methods, including a hanging arm cast (two patients), a shoulder spica cast (one), and a coaptation orthosis (three). The fractures in this group included three type-A, one type-B, and two type-C fractures. These fractures healed at an average of 180 days (range, forty-nine to 332 days).
Five patients (Cases 7-11) underwent operative treatment after an average of 123 days (range, forty-nine to 173 days) of unsuccessful nonoperative treatment. The fractures in this group included four type-B fractures and one type-A fracture. One patient had sustained a type-B fracture through both the bone and the cement mantle (Figs. 1-A, 1-B, 1-C, 1-D). Three patients with a type-B fracture and a well-fixed humeral component were treated with open reduction and internal fixation with use of a plate, screws, cables, and an iliac crest bone graft or an allograft. Two patients, including one with a type-B fracture and one with a type-A fracture, were found to have a loose prosthesis at the time of surgery. Both patients were treated with revision arthroplasty with use of a long-stem humeral component. One of the two fractures (a type-A fracture that was revised with a long-stem ingrowth component and an allograft) went on to nonunion. The fracture eventually healed following a vascularized free fibular transfer procedure (Figs. 2-A through 2-E).
Five patients (Cases 12-16) had immediate operative treatment. Two patients underwent open reduction and internal fixation. One of these two patients (Case 12) had sustained a radial nerve palsy after a type-C fracture. She underwent open reduction and internal fixation with screws, cables, and a dynamic compression plate. The radial nerve was explored and was found to be intact, without any obvious injury. The fracture healed, and radial nerve function recovered within three months. The second patient (Case 16) was treated with open reduction and internal fixation with use of screws, cables, and a humeral strut graft because of a type-A fracture through both the bone and the cement mantle (Figs. 3-A, 3-B, and 3-C).
Two patients had severe preinjury osteolysis and a loose prosthesis. One of these patients (Case 14) had a type-B fracture. The second patient (Case 13) had a transverse fracture through the proximal diaphyseal area in a region of severe osteolysis. Both patients were treated with revision arthroplasty with use of a cemented long-stem component and cancellous allograft. The patient with the type-B fracture required a custom (tumor) prosthesis because of the extensive bone loss in the proximal fragment.
The fifth patient who had immediate operative treatment (Case 15) sustained a type-A fracture during manipulation under anesthesia. The fracture occurred through the site at which the humeral cortex had been perforated during the index arthroplasty. The presence of the tip of the humeral component at the fracture site precluded fracture reduction. The patient was treated with open reduction and revision with a long-stem humeral component and Parham bands.
In the case of the patient who received a free fibular transfer (Case 10), it took 1116 days for the fracture to heal from the time of the first operative treatment. With the exclusion of this patient and the one who received the custom prosthesis (in whom there was no fracture left to heal), the remaining eight fractures that were treated operatively healed at an average of 278 days (range, 135 to 558 days) after surgery.
Among the patients who were treated nonoperatively, four had healing with bridging callus and two had direct bone-to-bone healing. Among those who were treated operatively, two had healing with bridging callus, four had mostly direct bone-to-bone healing with minimal callus, and two had direct bone-to-bone healing. One patient had healing with the incorporation of a transferred free fibular graft (Fig. 2-E). One patient had healing with direct bone-to-bone contact without incorporation of the strut graft (Fig. 3-C).
The average duration of follow-up was sixty-seven months (range, four to 191 months) from the time of the fracture. At the time of the latest follow-up, the average active abduction was 107° (range, 0° to 160°) and the average external rotation was 43° (range, 0° to 90°).
Radiographs that had been made at the time of the most recent follow-up were available for ten of the sixteen patients. A humeral periprosthetic radiolucent line was present in two of the seven patients who had had a total shoulder arthroplasty: the line was 1 mm thick and incomplete in one patient, and it was 2 mm thick and completely surrounded the component in the other. A humeral periprosthetic radiolucent line was present in two of the three patients who had had a hemi-arthroplasty: the line was 1 mm thick and incomplete in one patient, and it was 1 mm thick and completely surrounded the component in the other.
There were three excellent, four satisfactory, and nine unsatisfactory results at the time of the latest follow-up. The reasons for the unsatisfactory results included pain (one patient), poor motion (five patients), and both pain and poor motion (three patients). The details are presented in Table I. In the group of patients who had an unsatisfactory result, two patients had had a tuberosity nonunion prior to the fracture and one patient had lost upper extremity function as a result of a cerebrovascular accident.
One patient (Case 11) who was treated with open reduction and internal fixation and an iliac crest bone graft had a postoperative infection with Propionibacterium acnes. The infection responded to irrigation, débridement, and intravenous administration of antibiotics, and the fracture healed.
Periprosthetic humeral shaft fractures are relatively uncommon. The fractures in the present study represented 0.61% of all shoulder arthroplasties done at our institution. The few reports on these fractures in the current literature have included a limited number of patients1,2,5,6. This experience has indicated that these fractures can be difficult to treat successfully.
The location and configuration of the fracture has an important effect on outcome. Fractures located distal to the tip of the prosthesis (type-C fractures) are similar to closed fractures of the humeral shaft and respond favorably to nonoperative treatment. Two of the three type-C fractures in the present study healed successfully with nonoperative treatment (Figs. 4-A and 4-B). The fractures in the three patients who were excluded from the present study because of a lack of radiographs healed with nonoperative treatment. Based on the information in their medical histories, these three patients probably had type-C fractures. Previous studies have also suggested that type-C fractures respond favorably to nonoperative treatment. In the study by Campbell et al.4, four of the five fractures that were treated successfully with functional bracing were located distal to the tip of the prosthesis. Worland et al.5 reported on a similar fracture that was treated successfully with nonoperative means.
However, fractures that are located at the tip of the prosthesis (type-A and type-B fractures) behave differently. The literature includes reports of several such fractures that did not heal with nonoperative treatment: one such case was described by Bonutti and Hawkins2, and five such cases were reported by Boyd et al.1 in a series of seven fractures. In one case described by Worland et al.5, the implant proved to be unstable after a trial of nonoperative treatment and therefore operative treatment was needed.
In the present study, four of the five type-B fractures that were treated nonoperatively failed to heal and eventually required surgery. However, among the four type-A fractures that were treated nonoperatively, three went on to heal without operative treatment whereas one required multiple procedures, including a free fibular transfer, before healing was achieved. It is possible that the latter type-A fracture failed to respond to nonoperative treatment because of the presence of a loose humeral component. Specific indications for surgery that we identified included a loose prosthesis, fracture through both the bone and the cement mantle, difficulty of fracture reduction because of an interposed prosthesis, and recent surgery on the contralateral upper extremity.
In this small series, fractures that were treated nonoperatively healed sooner than those that were treated operatively. We identified some possible causes for the longer time to union following operative treatment. We occasionally used cancellous allograft, and it is possible that the use of an autogenous iliac crest bone graft, as recommended by Wright and Cofield6, would have led to quicker healing. In one case in which the time to union was 558 days, healing was complicated by a postoperative infection.
It is difficult to correlate the outcome with the type of fracture or the type of treatment because of the limited number of patients in the present study. However, the primary reason for an unsatisfactory result appeared to be restriction in motion.
Similar to Cameron and Iannotti3, we identified some factors that might increase the risk of periprosthetic fracture: osteopenia, cortical thinning due to osteolysis, and reaming of the medial cortex at the time of the arthroplasty.
On the basis of our experience, we can make the following recommendations. For a type-C fracture with a well-fixed humeral component, which is similar to a closed humeral fracture, a trial of nonoperative treatment is recommended if an acceptable closed reduction can be obtained with the use of an orthosis. A trial of nonoperative treatment may be considered for a well-aligned type-B fracture with a well-fixed humeral component. However, as noted in this series, four of five type-B fractures that were treated nonoperatively failed to heal and eventually required surgery. Therefore, operative intervention should be considered for a type-B fracture that has not progressed toward union by three months. Our current practice is to use a low-contact dynamic compression plate or strut graft with screw fixation in the distal portion and cerclage fixation in the proximal portion9. While we are unaware of any data on the effect of autograft or allograft in the treatment of periprosthetic humeral fractures, our current recommendation is to use autograft in order to maximize the healing potential. Our experience has shown this to be a good alternative to revision of a well-fixed prosthesis.
A type-B fracture that is associated with a loose humeral component should be revised with a cemented long-stem implant that is supplemented with iliac crest bone graft. Currently, we are unaware of any information in the literature on the use of cortical strut allograft or plate fixation with cerclage wires and screws to provide adjunctive fixation for a long-stem implant. However, such adjuncts have been reported to improve fixation in the setting of revision total hip arthroplasty in patients with a periprosthetic fracture10.
Our data do not clearly indicate that operative treatment is needed for type-A fractures unless the humeral component is loose. Three of six type-A fractures healed with nonoperative treatment, one failed to heal with nonoperative treatment, and two were treated with immediate operative fixation. Because of the substantial overlap between the length of the fracture and that of the humeral stem, the humeral component has a high likelihood of being loose. Type-A fractures that are associated with loose components should be treated operatively. All loose components should be revised with a cemented long-stem component spanning beyond the fracture, and iliac crest bone graft is recommended to augment healing. Supplementary fixation with an allograft or with a plate and screws and cables can be used to obtain secure fixation.
A table presenting detailed information on all of the patients in the study is available with the electronic versions of this article, on our web site at www.jbjs.org (go to the article citation and click on “Supplementary Material”) and on our quarterly CD-ROM (call our subscription department, at 781-449-9780, to order the CD-ROM).
The authors did not receive grants or outside funding in support of their research or preparation of this manuscript. They did not receive payments or other benefits or a commitment or agreement to provide such benefits from a commercial entity. No commercial entity paid or directed, or agreed to pay or direct, any benefits to any research fund, foundation, educational institution, or other charitable or nonprofit organization with which the authors are affiliated or associated.
A commentary is available with the electronic versions of this article, on our web site (http://www.jbjs.org) and on our quarterly CD-ROM (call our subscription department, at 781-449-9780, to order the CD-ROM).
Investigation performed at the Mayo Clinic, Rochester, Minnesota
1. Jr, Thornhill TS, Barnes CL. Fractures adjacent to humeral prostheses. J Bone Joint Surg Am.
2. , Hawkins RJ. Fracture of the humeral shaft associated with total replacement arthroplasty of the shoulder. A case report. J Bone Joint Surg Am.
3. , Iannotti JP. Periprosthetic fractures of the humerus and scapula: management and prevention. Orthop Clin North Am.
4. , Moore RS, Iannotti JP, Norris TR, Williams GR. Periprosthetic humeral fractures: mechanisms of fracture and treatment options. J Shoulder Elbow Surg.
5. , Kim DY, Arredondo J. Periprosthetic humeral fractures: management and classification. J Shoulder Elbow Surg.
6. , Cofield RH. Humeral fractures after shoulder arthroplasty. J Bone Joint Surg Am.
7. 2nd, Watson KC, Stanton FJ. Recent experience in total shoulder replacement. J Bone Joint Surg Am.
8. . Total shoulder arthroplasty with the Neer prosthesis. J Bone Joint Surg Am.
9. , Duncan CP. Cortical onlay allograft struts in the treatment of periprosthetic femoral fractures. Instr Course Lect.
Copyright 2004 by The Journal of Bone and Joint Surgery, Incorporated
10. , Beard DM, Pupparo F, Ries M, Wiedel J. One-stage revision of periprosthetic fractures around loose cemented total hip arthroplasty. Am J Orthop.