Delayed union or nonunion of the humerus following a fracture often leads to a painful upper extremity with limited function1,2. The prevalence of humeral shaft nonunion as a complication of both nonoperative and operative treatment has been reported to range from 8% to 12%3.
Autologous cancellous bone-grafting is believed to be an important factor contributing to the high union rate achieved with plate-and-screw fixation of ununited diaphyseal fractures of the humerus4-7, but it exposes the patient to the discomforts and risks of surgery at the graft donor site8,9. It is not known if the excellent union rates obtained with plate-and-screw fixation would be diminished by the use of a bone-graft substitute.
In 2000, one surgeon (D.L.H.) at our institution who treated a large number of nonunions of the humeral diaphysis changed from using iliac crest bone graft to using demineralized bone matrix with limited alteration of other operative tactics. The results in these patients were reviewed retrospectively to determine whether the use of demineralized bone matrix, rather than autologous bone graft, decreased the rate of union after plate-and-screw fixation.
Materials and Methods
One surgeon treated ninety-eight consecutive skeletally mature patients with plate-and-screw fixation of an ununited fracture of the humeral diaphysis over an eleven-year period (1992 to 2003). Of these patients, seventy-eight met the following inclusion criteria: (1) an atrophic ununited fracture of the humeral diaphysis; (2) no clinical, radiographic, or laboratory evidence of infection; (3) a minimum of twelve months of follow-up; and (4) a minimum of six months between the injury and the diagnosis of the ununited fracture. Of the twenty patients who were excluded from the study population, sixteen were excluded because of inadequate follow-up; two, because they had been treated with both iliac crest bone graft and demineralized bone matrix and thus could not be assigned to one of the two cohorts; and two, because the duration from the injury to the diagnosis of the ununited fracture was less than six months. The ununited fractures were treated with a standardized open reduction and plate-and-screw fixation protocol and the placement of either autologous iliac crest bone graft or demineralized bone matrix allograft (Grafton DBM Flex; Osteotech, Eatontown, New Jersey) at and around the fracture site. Autologous iliac crest bone graft was utilized from 1992 to 1999. From 2000 to 2003, the treatment protocol remained essentially unaltered except that the bone graft was changed to demineralized bone matrix.
All patients had either a delayed union (no evidence of bone-healing at the fracture site by six months) or a nonunion (no evidence of fracture-healing on three consecutive monthly radiographs made more than six months after the fracture). All ununited fractures were atrophic. (None were oligotrophic or hypertrophic.) This study was approved by our institutional review board, and patient consent for the use of autograft or demineralized bone matrix was obtained accordingly.
Forty-five patients (seventeen male and twenty-eight female) with an average age of fifty years (range, fourteen to eighty-five years) were treated with iliac crest bone graft (autograft group) between 1992 and 1999. Thirty-three patients (ten male and twenty-three female) with an average age of fifty-nine years (range, twenty-two to eighty-five years) were treated with demineralized bone matrix (allograft group) between 2000 and 2003. The mean time from the fracture to the surgery was 14.0 months (range, three to 121 months) in the bone-graft group and 22.6 months (range, two to 349 months) in the group treated with demineralized bone matrix.
Thirteen patients (29%) in the autograft group and fifteen patients (45%) in the allograft group presented with delayed union. Thirty-two patients (71%) in the autograft group and eighteen patients (55%) in the allograft group presented with a nonunion.
In the autograft group, seventeen patients (38%) were initially treated nonoperatively and twenty-eight patients (62%) had undergone previous surgery (average number of operations, 1.8; range, one to six operations) before the index operation. The previous treatment included open reduction and internal fixation (eleven patients), intramedullary nailing (eleven patients), external fixation (two patients), or a combination of nailing and subsequent open reduction and internal fixation (four patients). In the allograft group, twenty-three patients (70%) were initially treated nonoperatively and ten patients (30%) had undergone previous surgery (average number of operations, 1.5; range, one to four operations). The previous treatment included open reduction and plate fixation (five patients), intramedullary nailing (four patients), and a combination of open reduction and internal fixation and use of an external fixator (one patient). Three patients treated with autograft and one patient treated with allograft had had an infection prior to presentation to our institution. None of these patients demonstrated evidence of an active infection at the time of the index procedure.
In the autograft group, eleven patients presented with a peripheral neuropathy, which involved the radial nerve only in six of them, the ulnar nerve only in three, and both nerves in two. Five of the patients in the allograft group presented with a peripheral radial neuropathy.
Osseous healing was defined radiographically as the presence of at least three of four healed cortices, with bridging callus formation, and crossing trabeculae on anteroposterior and lateral radiographs. Clinical healing was defined as the absence of functional pain and local tenderness at the previous nonunion site.
The Mayo Elbow Performance Index10 was calculated for each patient. A score of 100 to 90 points was considered to be an excellent result; 89 to 75 points, a good result; 74 to 60 points, a fair result; and <60 points, a poor result.
Shoulder function was evaluated preoperatively and at the most recent follow-up visit with the use of a modification of the scale of Constant and Murley11,12. The maximum score on this scale is 100 points: 15 points for pain, 20 points for activities of daily living, 40 points for range of motion, and 25 points for power. Between 80 and 100 points was considered an excellent objective result; between 60 and 79 points, a good result; between 40 and 59 points, a fair result; and between 0 and 39 points, a poor result.
All complications related to the iliac crest donor site and any reference to discomfort at the iliac crest or the relevant cutaneous nerves of the harvest region were recorded. Major complications were defined as permanent symptoms, infection, or complications requiring intervention. Intermediate complications were defined as symptoms that persisted for more than six weeks but eventually resolved. Minor complications were defined as persistence of symptoms for no more than six weeks.
Data were analyzed with independent t tests as well as nonparametric Mann-Whitney, chi-square, and Fisher exact tests. The null hypothesis was that the two groups were similar. The experimental hypothesis was that the samples were from two different populations. All values represent means. A p value of <0.05 was considered to represent a significant finding.
A posterior triceps-splitting approach, as previously described13, was performed to treat fractures of the midpart or distal third of the shaft. When a patient had a distal fracture, the ulnar nerve was not routinely released or transposed but was carefully protected by subperiosteal preparation. An extended deltopectoral approach was used for proximal fractures.
After exposure of the nonunion site, multiple specimens were obtained for culture. Loose or broken implants were removed. With use of a curet and rongeur, the nonunion was radically débrided of intervening scar tissue with preservation of muscle and soft-tissue attachments to avoid devascularization of the fragments. Special caution was taken during this part of the procedure because neurovascular structures can adhere to the nonunion site. In addition, the medullary canal was opened both proximally and distally with an awl-type device or a 2.5-mm drill-bit to allow rapid neovascularization and migration of osteogenic cells. Freshening of the bone ends as well as removal of scar tissue and sclerotic bone fragments was often required.
The humerus was then reduced by gently impacting the proximal and distal fragments, thereby increasing stability and osseous contact. After adequate reduction and rotational alignment were achieved, a single 1.6 or 2.0-mm Kirschner wire was placed across the nonunion site to achieve initial stability. Occasionally, lag-screw compression was used.
Iliac crest bone graft or demineralized bone matrix was placed in the medullary canal and at the nonunion site prior to plate fixation and around the nonunion site after plate fixation. Intraoperative anteroposterior and lateral radiographs were used to confirm adequate placement of hardware and reconstitution of the osseous anatomy.
A well-padded posterior splint was worn initially, and the arm was subsequently rested in a sling for comfort and pain relief for a few days. On the first postoperative day, gentle active and active-assisted range-of-motion exercises of the shoulder and elbow were started under the supervision of a physical therapist or an occupational therapist.
Clinical and radiographic evaluation demonstrated osseous union within 4.5 months following the index surgery in 100% of the forty-five patients in the autograft group and within 4.2 months following the index surgery in 97% (thirty-two) of the thirty-three patients in the allograft group (Figs. 1-A, 1-B, and 1-C). Radiographic signs of healing correlated with clinical signs of healing, including the absence of pain or tenderness over the previous nonunion site and the absence of pain with motion. The average duration of follow-up was 32.8 months (range, twelve to 149 months) in the autograft group and 20.4 months (range, thirteen to forty-eight months) in the allograft group.
For a distal humeral nonunion, a plate was contoured to fit along the posterior aspect of the lateral column. A narrow 4.5-mm dynamic compression plate (DCP; Synthes, Paoli, Pennsylvania) was used from 1992 to 1998, after which a narrow 4.5-mm limited-contact dynamic compression plate (LC-DCP; Synthes) was employed. Recently, locked compression plates (LCP; Synthes) have been used. When necessary, an additional narrow 3.5-mm pelvic reconstruction plate (Synthes) was applied along the cortex of the medial column. The length of each plate was determined by the location and pattern of the nonunion. Engagement of at least eight cortices on both sides of the nonunion was the goal.
In the autograft group, a posterior approach was used in thirty-three patients (73%), an extended deltopectoral approach was used in eleven patients (24%), and a prior olecranon osteotomy was used in one patient. In the allograft group, a posterior approach was performed in twelve patients (36%) and an extended deltopectoral approach was used in twenty-one patients (64%). A single-plate construct was used to stabilize the nonunion in twenty-two patients (49%) in the autograft group, double-plate fixation was used in twenty-two patients (49%), and fixation without a plate but with a tension band and a lag screw was used in one patient, as previously described14. Interfragmentary compression lag screw fixation was performed in forty-four patients (98%) in the autograft group. In the allograft group, fracture fixation was performed with a single plate in fifteen patients (45%), a double-plate technique in seventeen patients (52%), and tension-band fixation in one patient. A lag screw was applied in all patients in that group.
In the early postoperative period, assessment of peripheral nerve function revealed a transient radial neurapraxia in five patients in the autograft group and in two patients in the allograft group. All patients in the autograft group recovered completely within two to sixteen weeks, and both patients in the other group recovered completely within sixteen weeks.
Staphylococcus aureus grew in the broth used for routine culture of the intraoperative specimens from one patient in the autograft group. There was no macroscopic evidence of superficial or deep infection, and the patient was treated with oral antibiotics for six weeks and remained asymptomatic. Two patients underwent a reoperation for complete removal of implants: one of these removal procedures was performed because of symptoms, and the other was done because of patient preference. One patient had a prominent screw removed with the use of local anesthesia, one patient with poor elbow function underwent manipulation while under anesthesia, and one patient with a deep wound infection at the iliac crest donor site was treated with irrigation and débridement.
In the allograft group, one plate broke after eight months, indicating loss of fixation and failure of the index operation. Revision surgery was carried out with removal of the hardware and osteosynthesis with use of a double-plate construct. Intraoperatively, the nonunion site was seen to have not completely healed. Uneventful healing was noted five months after the revision. In a second patient, with a proximal humeral nonunion that had been successfully treated with tension-band and screw fixation, the Kirschner wire broke, resulting in intermittent soft-tissue irritation. Following removal of the tension band, the patient was asymptomatic.
As evaluated with use of a modification of the scale of Constant and Murley11, excellent shoulder function was observed in 73% (thirty-three) of the forty-five patients in the autograft group and in 73% (twenty-four) of the thirty-three patients in the allograft group. Good shoulder function was observed in 22% (ten) of the patients in the autograft group and in 27% (nine) of the patients in the allograft group. Fair shoulder function was observed in 4% (two) of the patients in the autograft group, and no patient had poor shoulder function.
According to the Mayo Clinic Performance Index for the elbow, an excellent result was observed in 64% (twenty-nine) of the forty-five patients in the autograft group and in 85% (twenty-eight) of the thirty-three patients in the allograft group. A good result was observed in 33% (fifteen) of the patients in the autograft group and in 12% (four) of the patients treated with allograft. A fair result was seen in one patient in the autograft group and in one patient treated with allograft.
No differences between the group treated with autograft and the group treated with allograft were detected with regard to gender (p = 0.4), mechanism of injury (p = 0.3), affected side (p = 0.9), type of fracture (open or closed) (p = 0.2), number of previous operations (p = 0.2), location of the nonunion (p = 0.1), type of nonunion (p = 1.0), number of plates used for fixation (p = 0.9), use of a lag screw (p = 1.0), healing of the nonunion (p = 1.0), time to union (p = 0.4), follow-up time (p = 0.4), arm pain (p = 1.0), or shoulder score (p = 0.9).
Differences between the groups were detected for age (mean age, fifty years for the autograft group and fifty-nine years for the allograft group; p = 0.04), initial nonoperative treatment (performed in 38% of the patients in the autograft group and 70% of the patients in the allograft group; p = 0.006), surgical approach (posterior in 73%, deltopectoral in 24%, and olecranon osteotomy in 2% of the patients in the autograft group and posterior in 36% and deltopectoral in 64% of the patients in the allograft group; p = 0.005), and elbow score (excellent for 64%, good for 33%, and fair for 2% of the patients in the autograft group and excellent for 85%, good for 12%, and fair for 3% of the patients in the allograft group; p = 0.04).
One obvious difference between the two groups was the additional surgery for harvesting of the iliac crest autologous bone graft. The overall morbidity rate was 44% (Table I), with the major complications (permanent or requiring surgical intervention) obviously being the most important. There was one infection that required irrigation and débridement at eleven days as well as intravenous antibiotics for two weeks. The infection then fully resolved. Three patients had permanent pain and dysesthesias, and one patient had an insufficiency pelvic fracture and secondary heterotopic ossification.
We reported the successful open reduction and internal fixation of atrophic delayed unions and nonunions of the humerus. Hypertrophic nonunions and infected nonunions were not included in this study since they present different treatment challenges.
The efficacy of autologous bone-grafting in the treatment of delayed union and nonunion has been confirmed4-6,15. The histocompatibility and lack of immunogenicity of an autograft16,17 complement its osteoinductive and osteoconductive properties4. Furthermore, there is no risk of an autologous graft transmitting occult disease or infection4,18. There is reasonable surgical access to the iliac crest and an ample amount of graft material.
Of concern has been the substantial rate of morbidity at the autologous bone-graft donor site, as we observed in the present study. Forty-four percent (twenty) of the forty-five patients in the bone-graft group had morbidity, ranging from minor to major complications. It was troubling that some patients reported painful irritation and sequelae at the iliac crest even after the original debilitating humeral nonunion had healed and was completely asymptomatic. We therefore initiated this study to compare the effectiveness of autograft with that of demineralized bone matrix allograft, and we found results regarding fracture-healing to be comparable.
The limitations of this study include differences between the cohorts in terms of the number of patients, the initial fracture treatment, and the time from the fracture to the treatment of the ununited fracture as well as the sequential nature of the two study groups. More patients in the group treated with demineralized bone matrix had received nonoperative treatment, and these patients had, on the average, a longer duration from the fracture to the definitive treatment. In addition, the fractures in the initially treated (bone-graft) group were stabilized with 4.5-mm compression plates and lag screws, whereas locked compression plates, which became available later, were preferentially utilized in the group treated with demineralized bone matrix. As these plates were applied in the compression mode, their use may have had little relevance, but the plates employed in the two groups were definitely different. It is important to point out that careful and precise preparation of the nonunion site and use of reliable internal fixation techniques may be more important than the type of bone graft for the successful treatment of an ununited humeral fracture. It is also possible that bone graft is not necessary at all to achieve healing.
The excellent results observed in this study evoke the question of why the ununited fractures healed irrespective of the type of bone graft that was utilized. First, the surgical and plate-fixation techniques are of immense importance. Fixation must be performed with the aim of achieving a high degree of cortex-to-cortex stability. This is accomplished with interfragmentary lag-screw compression and rigid plate fixation. Occasionally, double-plate fixation is required in osteoporotic bone or if micromotion is noted at the bone interface following stabilization with a single plate. It is also important to correct angular deformities and achieve good axial and joint alignment. A stable construct provides an advantageous mechanical environment and is a prerequisite for successful bone-healing. A potential advantage of the LCP locked plates is the ability to use both bicortical and unicortical locking screws, which have increased holding power, especially in osteoporotic bone19.
The second critical principle of our operative concept is enhancement of the local biology of bone, soft tissue, and vascularity. Our protocol mandates radical débridement. We have found it essential to resect the nonunion and to remove all fibrous tissue. Gentle dissection is required to preserve soft-tissue attachments to osseous fragments and any residual blood supply. The medullary canal of both the proximal and the distal fragment is often sealed by fibrous tissue or a sclerotic osseous end-cap. We believe that the canal should be drilled to allow rapid neovascularization and migration of osteogenic cells. Careful intrafocal débridement is a key to ensuring optimal cortex-to-cortex stability, which facilitates neovascularization and migration of osteogenic cells and prepares the host environment for successful graft integration.
The third treatment principle is the application of a substance with osteoinductive properties to stimulate new bone formation and promote healing. One of us (D.L.H.) and colleagues previously showed that the local environment of atrophic nonunions is not deprived of growth factors20. However, augmentation with osteoinductive bone graft results in graftmediated release of growth factors and stimulates differentiation of precursor cells and osteoblasts21. The present study demonstrated that either autogenous bone graft or demineralized bone matrix can be effective in this regard.
The clinical benefits of the use of demineralized bone matrix instead of bone graft are evident since similar healing rates and clinical outcomes were found in our two treatment groups. Use of autologous bone graft requires an additional surgical site and increased surgical time, and it is associated with multiple complications8,22 that are not seen with the use of demineralized bone matrix.
In conclusion, we successfully incorporated augmentation with demineralized bone matrix allograft into a standard concept for the treatment of atrophic ununited fractures of the humerus with radical débridement of the fracture site, rigid plate fixation, and lag-screw compression. Demineralized bone matrix proved to be effective for augmenting the surgical repair, and the time to healing was comparable with that following augmentation with autograft. We believe that demineralized bone matrix can be used for standard graft augmentation in the treatment of humeral nonunions and delayed unions, to avoid the harvest of iliac bone graft and its associated donor-site morbidity.▪
The authors did not receive grants or outside funding in support of their research for 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 () and on our quarterly CD-ROM (call our subscription department, at 781-449-9780, to order the CD-ROM).
Investigation performed at the Orthopaedic Trauma Service, The Hospital for Special Surgery, New York, NY
1. , Zych GA, Hutson JJ, Malinin TI. Salvage of humeral nonunions with onlay bone plate allograft augmentation. Clin Orthop Relat Res. 2001;386: 203-9.
2. , Kloen P, Campbell D, Schwartz M, Liew A, Myers E, Helfet DL. Open reduction and internal fixation of humeral nonunions: a biomechanical and clinical study. J Bone Joint Surg Am. 2002;84: 1315-22.
3. , Verheyen CC, Besselaar PP. Humeral shaft nonunion: evaluation of uniform surgical repair in fifty-one patients. J Orthop Trauma. 2002;16: 108-15.
4. . Bone grafts. The basic science rationale for clinical applications. J Bone Joint Surg Am. 1987;69: 786-90.
5. , White GM, Mick CA, Brooker AF Jr, Weiland AJ. Nonunion of the humeral shaft. Clin Orthop Relat Res. 1987;219: 206-13.
6. , Kloen P, Anand N, Rosen HS. Open reduction and internal fixation of delayed unions and nonunions of fractures of the distal part of the humerus. J Bone Joint Surg Am. 2003;85: 33-40.
7. , Kloen P, Anand N, Rosen HS. ORIF of delayed unions and nonunions of distal humeral fractures. Surgical technique. J Bone Joint Surg Am. 2004;86 Suppl 1: 18-29.
8. , Smith WJ, Chambers HG, Bucknell AL, Davino NA. Complications of iliac crest bone graft harvesting. Clin Orthop Relat Res. 1996;329: 300-9.
9. , Senunas LE, DeSilva GL, Greenfield ML. Autogenous iliac crest bone graft. Complications and functional assessment. Clin Orthop Relat Res. 1997;339: 76-81.
10. , Adams RA. Semiconstrained arthroplasty for the treatment of rheumatoid arthritis of the elbow. J Bone Joint Surg Am. 1992;74: 479-90.
11. , Murley AH. A clinical method of functional assessment of the shoulder. Clin Orthop Relat Res. 1987;214: 160-4.
12. , Perey BH, Jupiter JB. The functional outcome of operative treatment of ununited fractures of the humeral diaphysis in older patients. J Bone Joint Surg Am. 1999;81: 177-90.
13. , Hotchkiss RN, Weiland AJ. Alternative operative exposures of the posterior aspect of the humeral diaphysis with reference to the radial nerve. J Bone Joint Surg Am. 1996;78: 1690-5.
14. , Rubel IF, Helfet DL. A two-tension-band technique for treatment of nonunions of the surgical neck of the humerus. Tech Shoulder Elbow Surg. 2001;2: 187-93.
15. , Jupiter JB, Gasser S. Indirect reduction and tension-band plating of tibial nonunion with deformity. J Bone Joint Surg Am. 1992;74: 1286-97.
16. , Kopta JA. The immunology of bone and cartilage transplantation. Orthopedics. 1991;14: 987-96.
17. , Friedlaender GE. Immunologic aspects of bone transplantation. A rationale for future studies. Orthop Clin North Am. 1987;18: 227-33.
18. , Villar RN. The bacteriology of bone allografts. J Bone Joint Surg Br. 1992;74: 398-9.
19. , Gautier E, Muller M, Helfet DL, Wagner M. First clinical results of the Locking Compression Plate (LCP). Injury. 2003;34 Suppl 2: B43-54.
20. , Doty SB, Gordon E, Rubel IF, Goumans MJ, Helfet DL. Expression and activation of the BMP-signaling components in human fracture nonunions. J Bone Joint Surg Am. 2002;84: 1909-18.
21. . Bone repair techniques, bone graft, and bone graft substitutes. Clin Orthop Relat Res. 1999;360: 71-86.
22. , Kelly CM. The effect of allomatrix injectable putty on the outcome of long bone applications. Orthopedics. 2003;26(5 Suppl): s567-70.