Treatment of Nonunions and Osseous Defects With Bone Graft and Calcium Sulfate

Abstract

The treatment of long bone nonunions and fractures with osseous defects is challenging. The results of 26 patients with either a persistent long bone nonunion or an osseous defect after an open fracture were reviewed. Each patient was treated with debridement of devitalized tissue, open reduction and internal fixation, and bone grafting using a mixture of autogenous iliac crest bone graft and medical grade calcium sulfate. The current study evaluated the union rate and associated complications for treatment of these injuries using this protocol. Each nonunion was confirmed intraoperatively, and healing was determined clinically by the patients’ return to full activities without pain and radiographically by the presence of bridging trabeculae. Complications included persistent nonunion (four patients), wound drainage (five patients), wound drainage and cellulitis (one patient) and cellulitis alone (one patient). Using this treatment protocol, 22 patients (85%) achieved healing after one surgery and an additional two patients (92%) achieved healing after a second surgery. Medical grade calcium sulfate increases the volume of graft material, facilitates bone formation, and is safe in the treatment of nonunions and fractures with osseous defects.

It has been estimated that 5% to 10% of all long bone fractures have delayed union or develop into a nonunion. ^2,5,7,10,18 The development of a nonunion generally is associated with pain, stiffness of surrounding joints, and disability, which often prevents employment and therefore imposes an economic burden on the patient and society. ^{3,12,13,16,19,21,22,34,36} Nonunions are thought to occur either as a result of a poor mechanical or biologic environment surrounding the fracture or a combination of the two. Unfavorable mechanical environments include instability at the fracture site, which allows excessive motion. This motion creates strain within the healing tissues preventing the differentiation of mesenchymal cells into osteoprogenitor cells. An unfavorable biologic environment also can affect bone healing; such conditions include bone loss, infection, inadequate soft tissue coverage, and diminished blood supply. Open fractures with bone loss are at risk of developing into a nonunion because of inherent instability and unfavorable biologic condition. Therefore, successful treatment of nonunions and fractures with osseous defects requires identification and treatment of these various adverse conditions.

Nonunions have been divided into two distinct groups: vital nonunions, which appear to have adequate blood supply and healing response; and nonvital nonunions, which lack adequate blood supply and are associated with a diminished healing response. ⁴³ Numerous clinical studies have documented successful treatment of nonunions using various techniques to improve the mechanical and biologic environments. ^{5,7,10,11,14,18–20,22,34,37} There are few reports of successful treatment of fractures with osseous defects after trauma. For vital nonunions, establishing fracture stability typically is sufficient to obtain union, whereas for nonvital nonunions improvement in the biologic environment also is essential for obtaining union. Unfortunately, these previous reports have focused primarily on nonunions that developed after closed treatment of a fracture or after one operative procedure. Few reports have described successful outcomes in the treatment of complicated nonunions (defined as nonunions which have failed to heal after more than one operative procedure or those associated with osseous defects).

Autogenous bone graft enhances osseous union by contributing osteoconductive and possibly osteoinductive materials to the fracture site. Unfortunately, a finite amount of bone graft is available from each individual and donor site morbidity is reported to approach 30%. ⁴⁸ These limitations have prompted the development and use of synthetic bone graft materials, including medical grade calcium sulfate, to minimize or eliminate the need for autogenous bone graft. Medical grade calcium sulfate is thought to function as an osteoconductive agent but also may have osteoinductive properties. ⁴¹ Calcium sulfate facilitates osseous healing in an osteoconductive fashion by filling the bone void and preventing fibrous tissue ingrowth. With time as the calcium sulfate is absorbed, fibrovascular tissue takes its place and eventually allows neovascularization and bone formation within the area. ^{6,8,15,23,26,32,39} Recently, calcium sulfate has been shown to be osteoinductive in an osseous defect model in adult crossbred wethers (> 45 kg). ⁴¹ The investigators found that the osteoinductive properties were related to local bone demineralization which occurs as a result of a reduction of the local pH. This demineralization of the bone leads to the release of osteoinductive molecules in the bone matrix stimulating the healing process.

The purpose of the current study was to evaluate the union rate and document adverse effects associated with the treatment of complicated nonunions and osseous defects using a mixture of iliac crest bone graft, medical grade calcium sulfate, and skeletal stabilization. This treatment protocol was developed to address each of the factors influencing fracture healing.

MATERIALS AND METHODS

The senior author (JB) adopted the treatment protocol evaluated in the current study in May 1996. A retrospective review was done to assess the outcomes of 26 patients treated between May 1996 and April 2000 for either a complicated long bone nonunion (n = 19) or an osseous defect (n = 9) or both, using this protocol. An osseous defect was defined as a defect greater than 2 cm involving ¾ or greater of the circumference of the bone at the level of the fracture. All patients included in the review had at least two prior surgeries to treat fractures or nonunions. Five of the nine patients with osseous defects had a circumferential defect (average, 5 cm). There were 13 men and 13 women with an average age of 45 years. Seven patients smoked more than one pack of cigarettes per day. Sixteen patients sustained isolated injuries, with the most common cause being secondary to a motor vehicle accident (Table 1).

TABLE 1: Patient Characteristics

Analysis of the Initial Fracture and Subsequent Nonunion

The initial fractures were classified according to the system of the Orthopaedic Trauma Association, ²⁵ and the extent of the open soft tissue injury according to Gustilo and Anderson. ⁹ The diagnosis of a nonunion was made preoperatively based on clinical examination and the interpretation of plain radiographs. Patients were reported to have a nonunion if there was motion at the fracture site and pain aggravated by stressing of the limb. Radiographic criteria included the absence of bone bridging the fracture site and lack of progressive healing for at least 3 months. All nonunions were confirmed at the time of surgery with motion at the fracture site and a lack of bridging bone. Patients with a residual osseous defect greater than 2 cm were not expected to heal primarily and were included in the study.

The most frequent nonunions and osseous defects were of the femur and tibia, whereas nonunions of the ulna and humerus occurred less frequently (Table 2). These fractures were treated initially with intramedullary nails in 10 patients, external fixation in nine patients, and open reduction and internal fixation in seven patients. Ten of the 26 patients with nonunions had previous attempts at union with autogenous iliac crest bone graft, whereas 16 patients were not previously treated with an iliac bone graft. Nine patients had osseous defects greater than 2 cm, with an average defect of 6 cm in length (range, 2–15 cm).

TABLE 2: Orthopaedic Trauma Association/AO Fracture Classification

The nonunions were classified as vital or nonvital based on the presence or absence of callus and the appearance of the bone ends. ⁴³ Nonunions with callus and without pencil tip-like ends were considered vital nonunions and those without callus and thin pointed tips were considered nonvital nonunions.

Twelve patients sustained open fractures: five patients had Grade II fractures, two patients had Grade IIIA fractures, and five patients had Grade IIIB fractures (Table 3). Of the points with open fractures, eight required soft tissue coverage after serial debridements of devitalized tissue, which was obtained within 7 days of injury. Five patients required rotational gastrocnemius flaps, one patient had a rectus abdominis free flap, and two patients had isolated split thickness skin grafts. On average, each patient had four surgeries (range, 2–9) before the index surgery including irrigation and debridements, previous fixation attempts, and soft tissue coverage.

TABLE 3: Associated Soft Tissue Injuries

Operative Procedure

The index surgeries were done by one of two surgeons (WR, JB), at an average of 11 months (range, 1–48 months) from the time of the initial fracture. For patients with an osseous defect, the average time from fracture to bone grafting was 3.5 months (range, 2–7 months). At the time of the index surgery, each patient had correction of deformity (alignment, rotation, and length), revision internal or external fixation, and autogenous bone graft supplemented with calcium sulfate pellets (Osteoset®, Wright Medical Corp, Memphis, TN). Calcium sulfate was added to the autogenous bone graft as an extender, and because it had been used successfully in other types of bone healing situations by previous investigators. In each case, the iliac crest bone graft was harvested with an acetabular reamer. This technique for harvesting iliac crest bone graft was reported originally by Sanders and DiPasquale. ²⁸ It is done by making an incision approximately two finger-breadths posterior to the anterior-superior iliac spine and extending it posteriorly along the iliac crest for approximately 7 cm. The interval between the external oblique and the abductor musculature is identified. A subperiosteal exposure of the outer table of the iliac wing is done and an acetabular reamer (diameter, 42 mm) is placed along the outer table and the cortical and underlying cancellous bone is collected. Per the current iliac crest protocol, hemostasis was obtained with electrocautery, the area was irrigated with normal saline, and the bone surface was packed for 5 minutes with thrombin (Jones Pharmaceuticals Inc, St Louis, MO) soaked Gelfoam (Pharmacia & Upjohn Corp, Kalamazoo, MI) and a dry lap sponge. Once hemostasis was obtained, the Gelfoam was removed and the wound was closed in layers with a ⅛-inch closed wound suction evacuator drain (C. R. Bard, Inc, Covington, GA). The amount of autogenous iliac crest bone graft harvested was not standardized but typically reflected the amount safely obtainable from one incision along the anterior iliac crest. ⁴⁴ This bone graft slurry (average volume, 20 cc) was mixed with approximately 15 cc of the large (4.8-mm diameter) calcium sulfate pellets (average volume, 15 cc). The volume of the mixture ultimately was determined by the size of the defect and area of the nonunion. In general, the mixture contained one part bone graft and one part calcium sulfate. In patients with an osseous defect, the nonunion was stabilized and the void was filled completely with an iliac crest bone graft and calcium sulfate mixture (Figs 1–3). In nonunions without osseous defects, the medullary canals were packed a few centimeters in both directions with the bone graft mixture. The remaining grafting material was placed around the nonunion site before wound closure. When necessary, soft tissue coverage was achieved with a gastrocnemius flap, vascularized muscle transfer, or split thickness skin graft.

Fig 1A–B.: (A) Anteroposterior and (B) lateral radiographs show a Grade IIIA open tibia and fibular fracture.

Fig 2A–B.: (A) Anteroposterior and (B) lateral radiographs were obtained after irrigation and debridement, local flap coverage, and stabilization with an external fixator. Autogenous bone graft and calcium sulfate mixture was added to the osseous defect of the diaphysis.

Fig 3A–B.: (A) Anteroposterior and (B) lateral radiographs obtained 1 year after the index surgery show complete healing.

Closed wound suction evacuator drains were not used in the first 11 patients but were placed before closure in the subsequent 15 patients.

Perioperative intravenous antibiotics, Ancef (G.C. Hamford Manufacturing Company for Schein Pharmaceuticals, Inc, Syracuse, NY) or vancomycin, were administered ½ hour before starting the index procedure and continued for 48 hours postoperatively. Postoperatively patients were immobilized with plaster splints and Red Cross cotton (Johnson & Johnson, Arlington, TX) for the initial 2 postoperative weeks. Afterwards, patients with shaft or distal tibial nonunions wore a pneumatic fracture-walking splint (EBI a Biomet Co, Marlow, OK) while proximal tibia and femur nonunions were not immobilized. In each patient, active assisted range of motion of the neighboring joints began 2 weeks after surgery. Patients were touch-toe weightbearing for the initial 6 weeks and partial weightbearing (50 lb) for the second 6 weeks. At 12 weeks, patients were advanced to full weightbearing status and began a rehabilitation program.

Followup

All patients were available for followup at an average of 16 months (range, 7–30 months). Each patient had serial radiographs and was examined at regular intervals at least until healing was complete or the diagnosis of a persistent nonunion was made. Radiographs were taken at regular intervals to assess healing and each patient was followed up at least until healing was complete or the diagnosis of a persistent nonunion was made. Radiographic healing was defined as the presence of bridging trabeculae crossing the nonunion site and was confirmed by stability of the fixation with time. Clinical union was based on history and physical examination with pain-free weightbearing without the use of assistive devices or functional use of the upper extremity.

RESULTS

Twenty-two patients (85%) had evidence of healing of the nonunion at an average of 4 months (range, 2–8 months), whereas persistent nonunion developed in four patients. All 26 patients have returned to full independent activities of daily living, including 20 who have returned to unrestricted leisure activities and preoperative employment status. Four patients have sought vocational rehabilitation and two remain unemployed.

When the two groups of patients were assessed separately, 15 of the 17 patients with nonunions achieved healing and seven of the nine patients with osseous defects achieved healing after one procedure.

As described above, 22 patients achieved successful healing after the first attempt at skeletal stabilization and bone grafting for a persistent nonunion or osseous defect. The average time to union for the whole group was 4 months (range, 2–8 months), whereas the average time for union for patients with a previous infection also was 4 months. Patients with an osseous defect achieved healing at an average of 5 months (range, 3–8 months), and patients with a persistent nonunion achieved healing on an average of 3 months (range, 2–5 months). Five patients, including four with previous osseous defects, had computed tomography (CT) evaluation of the affected areas after the index procedure and the CT scan confirmed successful osseous union in each patient. Each of the 22 patients who had successful union achieved healing with an acceptable alignment. Acceptable tibia and femoral alignment was defined as alignment in the sagittal plane of less than 5° varus or less than 10° valgus and less than 15° procurvatus or recurvatum. Average knee range of motion (ROM) for patients with tibial and femoral nonunions was 100° (range, 0°–120°) whereas the average ankle ROM of patients treated for a tibial nonunion was 5° dorsiflexion to 35° plantar flexion. The five patients with humeral nonunions achieved healing with an acceptable alignment and have recovered functional ROM of their ipsilateral shoulders and elbows.

Four patients had the initial treatment fail and a persistent nonunion developed. Of the four patients with persistent nonunions, none used tobacco or had evidence of infection. Two of the four (one femoral and one proximal tibia) achieved union after a second grafting procedure with iliac crest and calcium sulfate pellets. The other two patients have persistent proximal ulna nonunions and have refused additional treatment.

The first patient who had a persistent nonunion develop sustained a Grade II, open tibia fracture (41-C), after a fall which initially was treated with external fixation. He had a 3-cm bone defect, required a gastrocnemius rotational flap, and ultimately had a nonvital nonunion develop after six failed surgeries. He had revision external fixation with bone grafting according to the authors’ protocol; however, the tibia was left in varus malalignment and a persistent nonunion developed. A second surgery was done and tibial alignment was corrected, additional bone graft was added, and the patient achieved healing.

The second patient who had a persistent nonunion develop sustained a closed femur fracture after a fall (33-A). There was no bony defect and the initial treatment included open reduction and internal fixation with a plate and screw construct. He had a vital nonunion develop which was treated with revision open reduction and internal fixation, calcium sulfate pellets, and iliac crest bone graft. Unfortunately, postoperative alignment of the femur remained in varus and the nonunion persisted. A second surgery to correct the varus malalignment was done and the persistent nonunion healed.

The remaining two patients in whom the initial treatment failed had high-energy proximal ulna fractures. The first patient did not have a bony defect and was treated with open reduction and internal fixation. The other patient had a 5-cm bony defect after a gunshot wound and was treated initially with external fixation. Both patients had internal fixation and grafting per the authors’ protocol but had a persistent nonunion develop. The patient with the 5-cm defect required a rotational flap and also had prolonged postoperative wound drainage after the index surgery, but at the followup there was no evidence of persistent infection. Both patients have been able to do activities of daily living and have elected not to have revision surgery.

Six patients had wound drainage, which persisted for more than 48 hours after the index surgery. Of these six patients, none had a deep infection and no patient required an additional surgery. Each of these six patients achieved healing of the nonunions after the index surgery. In these patients, drainage persisted from the time of surgery for an average of 10 days (range, 7–30 days). The drainage was typically serosanguinous with a tinge of white discoloration. This drainage was not cultured because of the lack of local or systemic signs of infection and the surgeon’s confidence that this drainage was related to the calcium sulfate pellets. Five of six patients with prolonged wound drainage required soft tissue coverage (split thickness skin graft, one; free flap, one; rotational flap, three) at the time of the initial injury or after subsequent debridements for infection, and each had sustained an open fracture before the nonunion developed. Once drainage ceased all wounds healed without evidence of infection. One patient had coincidental cellulitis and wound drainage during the postoperative period. One patient had cellulitis alone. The cellulitis began on average 11 days postoperatively (range, 9–14 days) and persisted for 6 days.

None of the 26 patients experienced either a minor (hematoma) or major complication (infection requiring debridement) at the iliac crest bone graft site. At the most recent followup 12 patients described only mild discomfort at the site and none of the patients was limited functionally as a result of the bone graft harvest.

DISCUSSION

This review of 26 patients who had operative treatment of a nonunion or osseous defect associated with a fracture showed the success of a protocol involving operative stabilization, correction of deformity, and bone grafting using a mixture of iliac crest bone graft and medical grade calcium sulfate. Using this protocol, 85% of these patients (22 of 26) achieved healing after one surgery, and two additional patients achieved union after a second surgery for an overall success rate of 92%. Seven of the nine patients (77%) who had an osseous defect greater than 2 cm achieved healing of their defects after one surgery.

Long bone nonunions are thought to occur as a result of multiple factors including infection, soft tissue interposition, and poor mechanical or biologic environments or both. The soft tissue envelopes, the presence or absence of infection, bony stability, the presence of an osseous defect, and apparent vascularity of the fracture fragments, influence treatment of nonunions. Successful treatment includes addressing each of these factors and often includes the use of a bone graft. Ring et al ²⁷ reported successful treatment of infected tibial nonunions using an approach that combined debridement of nonviable tissue, application of vascularized soft tissues, bone grafting, and skeletal stabilization. Using a protocol similar to this plus the addition of calcium sulfate, the current authors also had success in the treatment of these complicated conditions.

Autogenous iliac crest bone grafts have been used in the treatment of nonunions and osseous defects. Based on some reports iliac crest bone graft contributes osteoconductive and possibly osteoinductive materials to the nonunion site. ^{4,5,7,10,11,14,18–20,22,37,38} The disadvantages if iliac crest bone graft include the finite amounts available and donor site morbidity. ⁴⁸ Medical grade calcium sulfate, if found to facilitate healing of persistent nonunions, could be used to expand bone graft volumes and potentially eliminate the need for autogenous bone grafting in this setting.

Calcium sulfate in the form of plaster of Paris (CaSO₄) has been used for numerous years in the treatment of bone defects. Its first reported internal use dates to 1892; in that series it was used successfully as a bone void filler in the treatment of tuberculous osteomyelitis. ⁶ Currently, calcium sulfate is available as medical grade calcium sulfate pellets. The crystallization process used in the manufacturing of this material is done in a highly controlled microenvironment to produce regularly shaped crystals. These uniform crystals assure predictable resorption, generally within 30 to 60 days implantation depending on the location and volume implanted. The rate of pellet absorption was designed to coincide with the ingrowth of vascular tissue into the osseous defect and nonunion site. ^8,23,26,39

Absorption of the calcium sulfate pellets is done by osteoclasts, in the same fashion as bone is absorbed. Osteoblasts have been shown to bind to this material during this process and during remodeling of the surrounding bone. ³² This calcium sulfate does not incite an inflammatory or foreign body reaction by the host, and no appreciable increased serum calcium levels have been detected with resorption of this material. ^26,31 Previous animal studies have shown the efficacy of calcium sulfate as a bone graft materials. ¹⁵ Specifically, in a rat femoral defect model, Huff and Grisoni ¹⁵ found a comparable healing rate when calcium sulfate alone was compared with corticocancellous allograft. There currently are no animal or human studies reporting the efficacy of using calcium sulfate in the treatment of complicated long bone nonunions.

Much of the orthopaedic literature addressing long bone nonunions has focused on the treatment of tibial nonunions. The treatment options include continued closed treatment with weightbearing, electrical stimulation, ultrasound, fibular osteotomy, open reduction and internal fixation with plates and screws, intramedullary nailing, external fixation, and bone transport. ^{5,7,12,14,16,21,24,29–31,33,35,37,39,40,42,45–47} Autogenous cancellous bone grafting in conjunction with skeletal stabilization is an acceptable treatment modality with success rates ranging from 87% to 100%. ^{5,7,10,11,14,17–20,22,34,37}

However, the successful treatment of persistent nonunions occurring after multiple previous attempts at union, and fractures or nonunions associated with osseous defects, can be particularly challenging. The treatment of tibial nonunions after intramedullary nailing includes continued observation, dynamization, exchange nailing, and bone grafting, whereas the treatment of tibial nonunions after treatment with external fixation typically includes continuation of external fixation and the addition of a posterolateral bone graft. In these cases healing rates have approached 90%. ^1,2 Alternatively, Wiss et al ⁴⁶ reported a 92% union rate with compression plating and iliac crest bone grafting for tibial nonunions after failed external fixation. In addition to the difficulty in obtaining osseous union, many persistent nonunions are accompanied by a tenuous soft tissue envelope. Infection also may be present and the endosteal or periosteal vascular supply or both may be lacking or inadequate to support osseous union. Ring et al ²⁷ reported success in treating infected tibial nonunions using a combined approach of debridement of nonviable tissue, application of a well-vascularized soft tissue envelope when necessary, bone grafting, and skeletal stabilization.

The treatment algorithm used by the current authors for complicated nonunions and fractures and nonunions associated with osseous defects included improvement of the soft tissue envelope, eradication of infection, correction of alignment, stabilization of the fracture or nonunion, and the addition of iliac crest bone graft mixed with medical grade calcium sulfate. This approach seems promising, in that union was obtained by 22 of 26 patients (85%), after the index surgery and two additional patients with nonunions achieved healing after a second surgery (92%). When a closed wound suction evacuator was added to the protocol, no patient experienced wound drainage or cellulitis, which were the primary complications observed early in this series. It is possible that the drain diminished the size of the initial hematoma and therefore less fluid was available to leak from the wound.

The cause of the four persistent nonunions that did not heal after the first surgical attempt is thought to be multifactorial; however, the most clinically evident cause was failure to adequately realign the limb. This seems most likely to be the cause, because after revision open reduction and internal fixation, repeat bone grafting with calcium sulfate, and correction of the alignment the two initially malaligned nonunions healed. One of the patients with a persistent nonunion had prolonged wound drainage; however, this patient currently has no evidence of infection, is asymptomatic, and has declined additional surgery.

Limitations of the current study include its retrospective and nonrandomized nature. Also, the patient population was heterogeneous; although this may be a limitation in certain circumstances, the heterogenicity of this group may make these findings more generalizable. Although the amount of autogenous iliac crest bone graft harvested was not standardized, it typically reflected the maximum amount safely obtainable from an iliac crest. Also, determining the exact timing of bony union and the amount of the defects filled with bone is difficult. Despite these shortcomings, this protocol using calcium sulfate as a bone graft extender seems safe and efficacious in the treatment of complicated nonunions and nonunions associated with osseous defects.