Vascularized Ulnar Bone Grafts for Limb-Sparing Surgery for the Treatment of Distal Radial Osteosarcoma^S

Introduction

The current standard of care for dogs with appendicular osteosarcoma (OSA) includes surgical removal of the tumor either via amputation or limb-sparing surgery followed by chemotherapy to control local disease and delay the development of metastasis.^1–4 In the dog, the distal radius has been determined to be the appendicular site best suited for limb-sparing surgery.^1,2,5–8 A number of limb-sparing surgical techniques have been described in the dog using cortical bone grafts, irradiated bone grafts, vascularized bone grafts, metal endoprostheses, and distraction osteogenesis with circular fixators to replace the excised bone that contained the tumor.^1,4,7–12 Problems such as local tumor recurrence, infection, implant loosening, and implant failure associated with these techniques have been well documented.^{3,5,7,8,10,11,13,14} Research on the use of vascularized bone grafts has demonstrated that vascularized bone grafts are more resistant to infection, are able to hypertrophy to fill the defect at the graft-bone junction, have accelerated bone healing, and have fewer implant failures compared with cortical allografts.^10,13–19 No studies have investigated whether there are significant benefits to using vascularized bone grafts for limb-sparing surgeries in the dog compared with any of the other available limb-sparing techniques.

This retrospective investigation includes a description of the surgical techniques involved in the use of vascularized ulnar bone grafts for limb-sparing surgery in the dog and an analysis of the historical information from the medical records of eight dogs that underwent a limb-sparing surgery. The objective of this study was to investigate whether using a vascularized ulnar bone graft, either as a vascularized free graft or a roll-in graft, would result in similar complications and complete healing of the graft with sufficient weight-bearing for ambulation compared with all other limb-sparing procedures.

Materials and Methods

The medical records of dogs with distal radial OSA that underwent limb-sparing surgery using a vascularized ulnar graft between June 2001 and May of 2008 were retrieved. Seven dogs from Michigan Veterinary Specialists (Southfield and Auburn Hills, MI) and one dog from the Veterinary Teaching Hospital at Michigan State University (East Lansing, MI) were identified. The postoperative chemotherapy data were retrieved from the dogs’ medical records as well as data regarding: signalment (breed, sex, age at presentation, and body weight); primary presenting complaint and duration of clinical signs; results of preoperative diagnostic tests (complete blood count, serum biochemistry profile, radiographic evaluation of the affected limb, thoracic radiographs, advanced imaging techniques [CT or nuclear scintigraphy] and cytology or biopsy of the affected bone); treatment information (type of limb-sparing surgery performed, type of chemotherapy used, if radiation therapy was used); results of follow-up clinical and radiographic evaluations; the number and type of postoperative complications; the development of additional medical problems; and the date of euthanasia or death, or the dog's status if the dog was still alive (Table 1).

Table 1 Data on Eight Dogs Treated with Vascularized Bone Grafts

FS, female, spayed; MN, male, neutered; OSA, osteosarcoma

View Fullscreen

Entry criteria for the study included the use of a vascularized limb-sparing technique, radiographic evidence of a lytic bone lesion in the distal radius, a confirmed histopathologic diagnosis of OSA, no evidence of distant metastasis prior to surgery, and additional follow-up information to the time of death or the end of the study. Minimum standards for preoperative staging included complete blood work, radiographs of the affected limb, and thoracic radiographs (three views). Two dogs had nuclear scintigraphy performed prior to surgery (cases 5 and 6) and two dogs had CT scans (cases 3 and 8) performed prior to surgery. Exclusion criteria for the study included the use of a vascularized limb-sparing technique for other types of primary bone tumors including fibrosarcoma or chondrosarcoma, absence of confirmed histopathologic diagnosis of OSA, evidence of pulmonary metastasis prior to surgery, early graft or implant failure within the first week leading to amputation, and a lack of additional follow-up information including the time of either death or euthanasia.

Additional preoperative imaging for staging was not performed prior to limb-sparing surgery for several reasons. Multiple radiographs were available for review of most patients which included images of other limbs, the pelvis, and the spine at the time of referral which was felt to be sufficient for preoperative screening. CT scanners and nuclear scintigraphy facilities were not readily available at all hospital locations and travel to other facilities would have been necessary. The added expense of CT or nuclear scintigraphy and additional episodes of anesthesia were not approved by all clients, especially when chemotherapy was going to be required following surgery. Four of the last five dogs included in the study had either CT or nuclear scintigraphy performed prior to the limb-sparing surgery as facilities became more readily available and the benefits of these studies became more apparent.

Survival time (ST) was defined as the time from surgery to either euthanasia or death. Local recurrence was defined as the time from surgery to the first evidence of local recurrence recorded. The metastasis-free interval (MFI) was defined as the time from surgery to the first documented evidence of either bone or lung metastasis. Only one patient, which was still alive at the time of data analysis, had no noted metastasis or local tumor recurrence and was therefore excluded from these calculations (case 8). After surgery, all patients were screened for lung metastasis by performing thoracic radiographs (three views) every three months. Mean and median ST, mean MFI, and local recurrence were calculated. Limb function was evaluated at each examination and was subjectively graded based on the use of the limb following surgery with the following scale: excellent (slight or no lameness), good (mild lameness), fair (moderate lameness), and poor (severe lameness or non-weight-bearing lameness).^3,4,10

Results

Eight dogs met the inclusion criteria (three spayed females and five neutered males) and three pure breeds of dogs were represented including two Labrador retrievers, one Great Dane, and one Saint Bernard. The remaining four dogs were mixed-breeds. Median age was 7 yr (range, 5–10 yr) and body weights ranged from 41 to 68 kg. The most common presenting complaint was forelimb lameness of 3–4 wk duration. Only one dog had an acute onset of a non-weight-bearing lameness on presentation associated with a pathologic fracture through the distal radius which was confirmed radiographically (case 1).

Preoperative blood tests were within normal limits for six of the eight dogs. One dog had a mild hypercalcemia and moderate hyperproteinemia (case 4). The mild hypercalcemia was suspected to be related to hypercalcemia of malignancy and the hyperproteinemia was suspected to be related to mild dehydration. The second dog (case 7) had mild hyperglobulinemia and mild hypoalbuminemia prior to surgery. The mild hyperglobulinemia was suspected to be caused by autoimmune effects, which have been demonstrated in some dogs with antibodies directed at the distal radial bone tumor. The cause of the hypoalbuminemia in this patient was unknown and was not noted as a consistent finding on subsequent blood work. Preoperative thoracic radiographs in all dogs showed no evidence of metastatic disease. Histopathologic examination of biopsy samples of the distal radius confirmed the diagnosis of OSA in all eight dogs.

Radiographs of all affected limbs revealed distal radial lesions consistent with a radiographic diagnosis of OSA. Specifically, an osteolytic lesion was noted in the distal one-third of the radius in all eight dogs (Figure 1). Additional radiographic findings included the following: an osteolytic-osteoproductive component in three dogs (cases 3, 5, and 8); cortical bone thinning in four dogs (cases 1, 2, 6, and 7); significant radiographic soft tissue swelling in two dogs (cases 2 and 6); periosteal bone reaction in one dog (case 7); and a pathologic fracture through the distal radial lysis in one dog (case 1). Additional radiographic images of the other limbs, pelvis, and spine were available on several patients with no documented metastatic abnormalities (cases 1, 2, 4, and 5).

View Figure

• Download as Powerpoint
• Download Figure

Additional diagnostic imaging was performed in four dogs. Two dogs had bone scans performed (cases 5 and 6) and two dogs had CT scans performed (cases 3 and 8) prior to surgery. The two dogs that had CT scans performed confirmed the presence of an osteolytic lesion in the distal radius with no evidence of ulnar invasion. One of these dogs had a full body CT scan (case 8), which revealed no evidence of thoracic metastasis and no evidence of neoplasia in other bones. Additional imaging studies were not performed in the earliest cases (1, 2, 4, and 7) due to a combination of factors including the added expense of additional testing, the need for a separate anesthesia, and the fact that these additional imaging studies had to be performed at a distant, nonaffiliated facility.

Six dogs had evidence of degenerative joint disease (osteoarthritis, OA) in more than one joint on plain radiographs, CT scans, or bone scan images. Three dogs had radiographic evidence of moderate to severe OA in both stifle joints (cases 1, 2, and 4), a fourth dog had mild OA in both stifle joints as well as more significant OA noted in several other joints (case 5), the fifth dog had mild OA noted in the right elbow with a concurrent fragmented medial coronoid process (case 3), and the sixth dog had mild OA at several sites (case 6). The fourth dog had a bone scan performed which demonstrated evidence of OA at multiple sites. Radiographs were taken of the spine and suspicious joints to confirm the absence of metastatic bone lesions in this fourth dog. Marked OA was identified in both shoulder joints, mild OA was found in the articular facets of the lumbosacral vertebra, and mild OA was noted in the right elbow and both stifle joints of this patient. One other dog had a bone scan performed (case 6) which revealed mild changes at several other joints and vertebra suggestive of OA which had been noted on previous radiographic images. Both of the bone scans confirmed the presence of an active lesion in the distal radius.

Limb-sparing surgery using a vascularized ulnar bone graft was successfully performed in all eight dogs. In four dogs (cases 1, 4, 7, and 8) the ulna was used as a vascularized free graft (Figure 2) whereas an ulnar roll-in graft was used in the remaining four dogs. Seven of the eight procedures were performed by the same surgeon. A bivalve cast (cases 1, 4, 7, and 8) or a palmar splint was applied in four dogs each (cases 2, 4, 5, and 6) following surgery. The splint and bivalve casts were maintained on the dogs’ limbs for a range of 1.5–14 wk, with an average time of 7.2 wk. Bandage changes were performed at least every other week, but there was significant variability in the frequency of these changes.

View Figure

• Download as Powerpoint
• Download Figure

Five of the eight patients appeared to have problems with the external coaptation devices during the postoperative period. One dog developed a small pressure sore over the caudal accessory bone which required removal of the splint 3 wk after surgery (case 5). Four dogs developed significant swelling of the limb requiring more frequent bandage changes for the first few weeks postoperatively (cases 2, 3, 6, and 7). Three of these four dogs developed swelling in the distal limb and paw within a few days of surgery and swelling was noted around the elbow, proximal to the bivalve cast in case 6, about 6 wk after surgery. In most dogs, the swelling and bandage irritation resolved with more frequent bandage changes; however, two dogs required early removal of the cast or splint to resolve this postsurgical complication (cases 5 and 6).

The most common postoperative complication encountered in this study was infection. Five of the eight dogs (62.5%) developed some form of infection following surgery (cases 1, 2, 3, 6, and 7). Four of these dogs had a mild superficial skin infection early in the recovery period characterized by a moist dermatitis and moderate erythema of the skin, which improved following the administration of oral antibiotics (cases 1, 3, 6, and 7). Three of the five dogs developed deeper infections (cases 2, 3, and 7) several weeks to two months postoperatively, which was identified by the appearance of purulent exudate with dehiscence along the incision or the appearance of a draining tract near the original incision. Draining tracts suggestive of implant infection developed in two of these three dogs >6 mo after the initial surgery (cases 3 and 7). Positive cultures were obtained in these two dogs by using a needle to collect a sample of fluid from near the bone plate. Staphylococcus aureus was identified in case 3, with sensitivity to marbofloxacin^z whereas S. intermedius was isolated from case 7, with sensitivity to amoxicillin trihydrate/clavulanate potassium.

Antibiotics were administered for variable periods of time throughout the study. All dogs were prescribed either cephalexin or amoxicillin trihydrate/clavulanate potassium postsurgically for a minimum of 3 wk. The three dogs which did not have any evidence of infection (cases 4, 5, and 8) were prescribed antibiotics for the shortest period of time (range, 3–4 wk). The four dogs that demonstrated superficial signs of infection (cases 1, 3, 6, and 7) were administered antibiotics for a total of 6–12 wk following surgery: cephalexin was administered for a total of 8 wk in one patient (case 6), and 12 wk in another patient (case 1). Amoxicillin trihydrate/clavulanate potassium was administered to case 7 for 6 wk. Case 2 was initially prescribed cephalexin for 2 wk, but the antibiotic was changed to amoxicillin trihydrate/clavulanate potassium for 4 wk due to worsening of the infection while on the initial antibiotic.

Cases 2 and 7, which also developed deep infections, were treated with multiple courses of antibiotics during their recovery. As previously stated, both of these dogs had cultures performed of their implant sites to help guide their therapies and received the appropriate antibiotic for a total of 7–8 wk. These latter two cases eventually required removal of some of the implants to resolve their infections. Case 7 had three loose bone screws removed from the bone plate 18 mo after surgery only to require the removal of another loose bone screw 3 mo later. Eventually, the draining tracts in this dog healed with no further complications. In case 3, the bone plate was removed 9 mo following the original surgery. This dog fractured the ulnar bone graft 7 days after removal of the bone plate. The fracture was treated by applying a type II external skeletal fixator (ESF) which was left in place for 14 wk. The initial fracture healed; however, the limb eventually refractured 9 mo after removal of the ESF due to recurrence of the tumor at the graft site (23 mo postsurgically). This was the only noted dog in which recurrence of the tumor caused a pathologic fracture in the ulnar graft.

Implant problems were identified in three other dogs with no concurrent evidence of infection or tumor recurrence (cases 1, 4, and 8). A small halo of lucency was evident around a single proximal radial bone screw in cases 1 and 4. The lucency resolved in each of these patients with time. The third dog developed a fracture in the radius, proximal to the ulnar graft, at the level of the second bone screw several months after surgery. Despite the development of a radial bone fracture in this dog, there was radiographic evidence of ulnar graft healing and graft hypertrophy at the time of the fracture. The fracture in the radius healed uneventfully using a bivalve cast for support for 2 mo.

Radiographic evidence of progressive healing at the bone-graft interface was documented in all dogs. Complete healing at both ends of the ulnar graft occurred in all of the patients between 8 and 20 wk postoperatively (Figure 3). There did not appear to be any significant difference in the rate of healing at either the proximal bone-graft interface or the carpal arthrodesis in most of the dogs. There was only one case (case 4) in which the carpal arthrodesis site demonstrated complete healing faster than the proximal ulnar bone graft site at 8 wk.

View Figure

• Download as Powerpoint
• Download Figure

Radiographs were used more frequently than other modalities to evaluate healing of the ulnar grafts in this study; however, the first two dogs in this study (cases 4 and 7) had either a Doppler evaluation or scintigraphy performed within 1 wk of their vascularized free graft surgery to monitor the integrity of the blood supply to the ulnar bone graft. In one dog (case 7), a Doppler evaluation was performed daily for the first 2 days following surgery, which showed good pulses in the blood vessels supplying the ulnar graft. This patient also had nuclear scintigraphy performed 5 days following surgery, which demonstrated an adequate blood supply to the ulnar graft. One other dog (case 4) had nuclear scintigraphy performed 7 days following the limb-sparing surgery, which demonstrated adequate blood supply to the healing ulnar graft (Figure 4) Serial radiographs were used in the remaining dogs to document graft survival. The appearance of a periosteal reaction within 4 wk, callous formation at the bone-graft interface by 6 wk, and early hypertrophy of the graft segment suggested successful vascular transfer.^14,15,18

View Figure

• Download as Powerpoint
• Download Figure

Hypertrophy was defined as a nonpainful, uniform thickening of the proximal aspect of the antebrachium with a concurrent uniform increase in bone opacity along the ulnar segment on the radiographic image (Figure 5). The degree of hypertrophy was graded as mild, moderate, or severe based on the thickness of the graft or the thickening at the bone-graft interface. In most dogs described in the literature, hypertrophy gradually increased with time.^18,19 Hypertrophy of the bone grafts occurred in all eight dogs included in this study. Some dogs required up to 12 mo for remodeling of the bone graft (Figure 6). Extensive hypertrophy of the ulnar bone graft was noted in three dogs (cases 1, 5, and 6) with one of these patients developing hypertrophy as early as 6 wk postoperatively (case 5). Hypertrophy was differentiated from local tumor recurrence based on a combination of physical and radiographic changes. At no time in this study was hypertrophy mistaken for recurrence of the bone tumor when considered in combination with physical and radiographic changes.

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

After bone healing was complete and short-term complications were resolved, limb function was considered excellent in 25% of the dogs (cases 1 and 4), good in 62.5% of the dogs (cases 2, 5, 6, 7, and 8), and fair in the remaining case (see Supplementary Video I).^3,4 All eight dogs had functional use of the limb within a few days to weeks following surgery which subjectively appeared to improve in the majority of dogs as the ulnar graft healed and hypertrophied.

A return of a non-weight-bearing lameness on the operated limb with subsequent swelling was noted in cases 3 and 5, which were found to have local recurrence of the tumor at the surgical site. Neither of these patients were those who had either presurgical fracture through the bone tumor or postoperative fracture at the proximal aspect of the bone graft. Case 3 developed pulmonary metastatic lesions and local tumor recurrence at the surgical site 560 days postoperatively. Case 5 developed a swelling at the surgical site which was presumed to be local tumor recurrence 280 days after surgery. Both of these dogs had advanced diagnostics performed prior to their limb-sparing surgery including a preoperative biopsy. In cases 3 and 5, the preoperative CT scans were clear of ulnar invasion and the preoperative bone scan performed on case 5 had no noted involvement of the ulna.

A severe, non-weight-bearing lameness appeared in three other cases (1, 2, and 4) several months following their limb-sparing surgeries. The lamenesses were in either the same or a different limb and all three dogs had radiographic evidence of distant bone metastasis. Case 4 developed radiographic evidence of a bone tumor at the distal tibia 1,092 days postoperatively, which was not definitively diagnosed and was treated conservatively using pain medications until the dog was euthanized 4 mo later. Case 1 developed an osteosarcoma of the bone in the left proximal humerus 1,512 days postoperatively which was managed using palliative radiation therapy over a 7 mo time period. Case 2 developed a bone lesion at an unidentified location which was also treated with a course of radiation therapy 8 mo (224 days) after the limb-sparing surgery.

All eight dogs were treated with some form of chemotherapy following their limb-sparing surgery. Chemotherapy protocols were not standardized and were dictated by the oncologist managing the case. Chemotherapy protocols used either a single agent or a combination of agents including doxorubicin^aa, cisplatin^bb, carboplatin^cc, or cyclophosphamide^dd. Only one dog received a single dose of doxorubicin prior to limb-sparing surgery (case 3). All other dogs were started on chemotherapy approximately 14 days postoperatively. Chemotherapy protocols were individualized for each patient with two dogs avoiding doxorubicin therapy due to pre-existing cardiovascular disease (cases 5 and 7). Chemotherapy was used more than once in some patients due to the development of metastasis, local tumor recurrence, or the development of additional neoplasias. Metronomic maintenance chemotherapy (cyclophosphamide and deracoxib^ee) was used in case 6 after completing the induction protocol using doxorubicin and carboplatin, which resulted in severe hemorrhagic cystitis causing discontinuation of the cyclophosphamide.

Most of the chemotherapeutic agents used in these patients were well tolerated with gastrointestinal side effects reported most commonly. These side effects were frequently controlled with antiemetics, antibiotics, and a change or reduction in chemotherapeutic dosages depending on the severity of the reactions. Overall, initial chemotherapy was well tolerated in seven of the eight patients with only one patient discontinuing the planned induction chemotherapy protocol due to severe side effects perceived by the owners (case 8).

Seven of the eight patients were euthanized by the time this manuscript was prepared (cases 1–7). In these dogs, STs ranged from 9 to 61 mo (252–1,708 days). Mean and median STs were 29.3 mo (820 days) and 25 mo (700 days), respectively. The mean MFI for the seven dogs was 33.67 mo (942 days) with a range of 8–54 mo. Local recurrence occurred at 10 and 20 mo (280–560 days) following limb-sparing surgery in two dogs (cases 3 and 5, respectively).

Discussion

The vascularized ulnar bone grafts used in this study achieved complete radiographic bony union at both ends with concurrent graft hypertrophy. This resulted in weight-bearing on the operated limb that could be classified as good to excellent in 87.5% of this study population following recovery. Both the ulnar roll-in and vascularized free graft techniques were successfully used to replace the resected radial bone segment affected with OSA.

Complications were anticipated as they have been previously reported in the veterinary literature with allograft limb-sparing surgery. The observed complications were manageable and no life-long therapeutic treatments were required in any patient. The frequency and types of complications reported following limb-sparing surgery for radial OSA in the dog include implant infection (40–70%), tumor recurrence (15–40%), and implant loosening or implant failure (10–60%).^{2,7,8,10–13,17} All of these complications were observed in this study. Specifically, tumor recurrence occurred in 25% (2/8) of the dogs, metastasis in 50% (4/8) of the dogs, implant loosening in 37.5% (3/8) of the dogs, implant failure in 12.5% (1/8) of the dogs, and infection in 62.5% (5/8) of the dogs. Similar to the reported veterinary literature, infection was the most common complication encountered in this study.

Postoperative infection is a significant problem in limb-sparing surgery for distal radial OSA. The reported implant infection rate is 40–60% when cortical allografts are used and 55% when an endoprosthesis is used.^{6,8,10,11,13,17} Once infection has become deep-seated, antibiotic therapy cannot be stopped without recurrence of the infection until the implants are removed.^7,13 Infection during limb-sparing surgery likely results from the long surgery times, extensive soft tissue dissection required to remove the tumor and bone, the loss of blood supply to the distal limb after the removal of a large mass, the application of implants, and limited soft tissue coverage over the implant during healing.^{3–5,7,11,13}

On initial evaluation, the dogs in this retrospective study appeared to have a relatively high rate of overall infection. Although antibiotics were administered to all patients postoperatively, 62.5% of the patients demonstrated evidence of surgical site infection which required antibiotic administration for a longer time period than expected (>4 wk). Infection in these patients was defined as purulent discharge exuding from a healing or previously healed incision. Three of these patients (37.5%) required intermittent long-term antibiotic administration (>8 wk) with two of these dogs undergoing implant removal to help resolve the infections. Although some dogs received several courses of antibiotics, none of these patients were administered life-long antibiotic therapy as has previously been described following implant infection with limb-sparing surgery. Only two of the cases with infection required removal of hardware to resolve the infection, which was unexpected. This unique observation may reflect either an error in defining deep-seated infections in this study population or a possible benefit of vascularized grafts. It is possible that vascularized bone grafts may be less susceptible to infection than other limb-sparing surgeries and that an intact blood supply may make clearing infections possible during healing.

The ability to remove the stabilizing implant is considered an additional advantage of both ulnar roll-in grafts and vascularized free grafts as these vascularized grafts incorporate into bony defects within 2 to 4.5 mo and may hypertrophy to fill the defect within 1 yr.^18,21,22 The length of time to complete graft incorporation depends on the size of the original defect, the type of graft used, and the size of graft used to fill the defect.^12,20–22 Successful implant removal following vascular grafts is described in the literature; however, graft fracture following implant removal can occur even when significant hypertrophy is present.^1,20,21

Deep-seated bone infections may weaken the bone, prevent or delay healing at the graft site, or cause loosening of implants. Chronic or deep infections can be devastating for patients that have undergone a limb-sparing surgery as premature removal of implants may be required which may predispose the graft to fracture. Infection leading to implant failure and possible loss of the limb defeats the purpose of performing a limb-sparing surgery. This possible complication should be discussed with owners prior to performing any limb-sparing surgery for the treatment of distal radial OSA and prior to removing any implants after healing at the graft site.

Based on the dogs included in this study, implants should be removed with caution. The removal of loose bone screws appeared to be well tolerated and resolved persistent infection; however, removal of the entire bone plate was catastrophic. To prevent failure at the graft site or screw holes, the authors recommend following any plate removal due to implant infection with some form of temporary stabilization such as an ESF, splint, or bivalve cast until the infection resolves. Radiographic evaluation of the graft site should be used to help determine when there has been enough hypertrophy, bone healing, and resolution of infection to allow the stabilizing external device to be removed in these situations.

Previous reports in the veterinary literature have documented that dogs that develop an infection following limb-sparing surgery had longer STs than those whose implants did not get infected.^5,7,9,10 This finding, although intriguing, does not make infection an acceptable complication. This phenomenon has yet to be definitively explained. There did not appear to be a direct correlation between infection and prolonged survival in this study population as two of the three dogs with the longest recorded STs (cases 1 and 4) did not develop signs of infection at the graft site. Overall, there was a high incidence of infection in this study (62.5%) which might have been complicated by the retrospective design. The data were collected over a period of 8 yr and the researchers were limited by the information documented in the records from multiple locations. The majority of dogs in this study were not directly observed by the primary author and all of the case details were extrapolated from the medical records. Radiographs and cultures were not performed at every appointment when an infection was suspected, which possibly contributed to an over-interpretation of infection throughout this study.

Infection did not appear to adversely affect the overall healing in either limb-sparing surgery as healing of the carpal arthrodesis, radius-graft interface, and hypertrophy of the ulnar graft was noted in all cases in a timely manner. Hypertrophy of vascularized bone grafts can be demonstrated radiographically.^15,18 This phenomenon has been found to be unique to the healing of vascularized bone grafts and may be suggestive of graft viability.^18,19,21 Surviving osteocytes in the graft respond to external stresses in a similar manner to normal bone following Wolff's law, which causes thickening of the cortices in response to an increase in stress on the bone. Varying degrees of graft hypertrophy were noted in the majority of dogs (7/8) in this study suggesting that the vascularized grafts were viable. The possibility still exists that some of the blood vessels supplying the ulnar grafts in these dogs were compromised or became thrombosed, converting portions of the vascularized grafts into avascular grafts.¹⁵ Previous studies comparing vascular and avascular grafts determined that avascular or devitalized grafts have irregular remodeling with significant gaps in some areas undergoing resorption. Some areas of the avascular graft are replaced via creeping substitution while other areas may revascularize.^15,18,22 None of the radiographs in this study demonstrated loss of bone in the ulnar graft site suggesting that none of the ulnar grafts underwent devitalization. All of the dogs in this study had radiographic bony union of their grafts within 2 to 5 mo and most demonstrated hypertrophy and remodeling in their grafts by 12 mo postsurgically. These changes suggest that the grafts maintained their blood supply during healing because avascular grafts may take longer than 1 yr for graft incorporation.^18,20

No studies have definitively identified the length of time required for clinical union following the use of a vascularized ulnar graft in limb-sparing surgery. It has been suggested that it may take several years for the classic cortical allograft to be replaced by bone and it may never fully incorporate. Clinical union has been identified to occur anywhere from 11 mo to 3 yr following cortical allograft surgery and appears to be dependent on several factors including the size of the graft incorporated, presence of infection, stability of the implant, and type of preparation used on the bone for grafting.^{3,11,15,18,20} Additionally, bone transport osteogenesis may require anywhere from 3.3 to 7.3 mo (94–205 days) for complete distraction of the radial segment with removal of the circular fixator.¹² In this study, complete healing of the ulnar graft was demonstrated in 2 to 5 mo (56–140 days) with remodeling continuing until the ulnar graft hypertrophied to fill the original defect in the radius. Such rapid healing may lead to improved functional use of the limb without the need for bulky external coaptation after the initial 2 mo of healing.¹⁰

All patients in this study recovered well from their limb-sparing surgery and maintained good functional use of the limb both during recovery and throughout the remainder of their life until recurrence or the appearance of other tumors caused additional lameness. Many owners consider the ability to ambulate normally vital to a good quality of life for their dog. Dogs with distal radial OSA were generally considered good candidates for limb-sparing surgery if they could have problems ambulating following limb amputation due to their large size, the presence of concurrent neurologic disease, the presence of severe degenerative joint disease in other limbs, or owner reluctance to amputate.^1,2,5,8,13 Half of the dogs in this study were giant breed dogs or giant mixed-breeds, more than half of these dogs had evidence of significant degenerative joint disease/OA in at least one other limb (5/8), and some clients specifically requested the limb-sparing surgery as an alternative to amputation. All of these patients had some degree of gait abnormality following surgery; however, most of these dogs had good to excellent use of the limb with full to near full weight bearing (6/8) following surgery (see Supplementary Video II).

The mild gait changes observed in these dogs postsurgically were thought to be directly related to the arthrodesis of the carpal joint that was performed as part of the limb-sparing surgery in this study.^8,10 Following a carpal arthrodesis, many dogs demonstrate mild circumduction of the forelimb, but learn to compensate with minimal lameness.^4,10 Arthrodesis of the carpal joint was performed with this limb-sparing surgery to help increase the overall stability of the construct (i.e., graft, bone plate, and radius), to allow for early ambulation during graft incorporation with removal of the cast or splint after 8 wk, and to improve the eventual incorporation of the graft into the radial defect.

The fact that the ulnar graft hypertrophies to fill the original defect in the distal radius makes implant removal theoretically possible; however, based on the experience in this study population, implant removal should be performed with caution even with radiographic evidence of complete carpal arthrodesis and graft incorporation.²⁰ Plate removal in patients without an underlying infection may be possible, but failure may still occur and should be discussed as a potential complication with owners prior to implant removal. Overall, the patients in this study had good functional use of their limbs following limb-sparing surgery together with the carpal arthrodesis based on observations made by clinicians and owners alike (see Supplementary Video III).

Vascularized autografts appear to be a good option for distal radial limb-sparing surgery in dogs with OSA. Similar complications occurred following limb-sparing surgery as other described techniques.^10,14,15 It has been demonstrated that bone grafts in which the blood supply can be maintained heal faster than cortical allografts due to an increased survival of osteocytes, increased osteogenic potential of these osteocytes, less bone necrosis requiring replacement, and maintenance of a more organized overall structure.^3,15 The ability to maintain the blood supply during healing may be the primary factor responsible for the improved healing and quicker remodeling, which makes vascularized bone grafts appealing for use in limb-sparing surgery.^10,17

Two vascularized techniques were used in this study to reconstruct the radial bone defect. There are currently no published recommendations for choosing one limb-sparing technique over the other and no studies have demonstrated a significant advantage of one particular surgical technique. The authors of this study based their recommendations for performing a particular limb-sparing surgery on several factors. The vascularized free graft technique was time consuming and costly due to longer surgery times and additional surgical equipment required to complete the anastomosis (e.g., an operating microscope) as well as having a surgeon willing to perform the intricate microvascular repair.^10,14,18 The ulnar roll-in technique was less time consuming, less costly, and did not require specialized equipment for surgical dissection, but did not allow removal of a portion of the ulna adjacent to the radial bone tumor, increasing the possible risk of local tumor recurrence following surgery. The degree of owner commitment, surgeon preferences, availability of equipment, possible involvement of the ulna, and the presence of a fracture were all factors that influenced the type of surgery recommended for each patient.

Pathologic fractures have previously been identified as a contraindication for limb-sparing surgery due to the increased seeding of tumor cells into the surrounding tissues.⁸ Case 1 in this study presented with a pathologic fracture though the radial bone tumor. The vascularized free grafting procedure was performed despite the possible risks. This dog did not suffer from local tumor recurrence, but did develop a metastatic bone lesion in the ipsilateral proximal humerus 54 mo after the limb-sparing surgery. This patient recorded the longest MFI and the longest ST (61 mo) within this study. Although the survival of one case does not warrant a change in the criteria for suitable limb-sparing candidates, it indicates that a vascularized free graft may be an option worthy of consideration in patients with fractures through the distal radial tumor when a limb-sparing surgery is desired with a minimal increased risk of local tumor recurrence.

Based on the results of this study, it appears that dogs treated with either of the vascularized limb-sparing procedures described suffered from continued disease progression with subtle differences in the location of tumor regrowth. Following ulnar roll-in surgery, dogs may be more prone to local tumor recurrence. In contrast, dogs having vascularized-free graft surgery may be more prone to the development of metastatic lesions. The small sample size and the retrospective study design may have influenced the perceived rate of tumor recurrence or metastasis and are significant deficiencies of this study. In addition, the rate of disease progression may have been influenced by the retrospective study design due to a lack of control in the level of preoperative staging. The frequency of local tumor recurrence following ulnar roll-in surgery demonstrates the importance of additional preoperative diagnostic imaging such as a CT scan to better evaluate the extent of tumor involvement prior to pursuing a limb-sparing surgery. Although such imaging did not prevent the local recurrence of tumor in one case, the information provided by a CT scan does help guide surgeons in the decision making process to determine whether limb-sparing surgery is a viable option for patients. Although most studies have focused on the importance of identifying the proximal extension of tumor in the medullary cavity, evaluation of the caudal radial cortex and cranial ulnar cortex are important when ulnar grafting surgeries are considered.²³ Additionally, the incidence of metastatic bone lesions noted in patients who lacked additional preoperative imaging demonstrates the importance of preoperative staging in the form of a CT scan, bone scan, or survey bone radiographs. Although metastatic bone lesions were identified more than 39 mo following limb-sparing surgery in all except one dog, the lack of consistent survey radiographs prior to surgery makes it difficult to determine when these bone lesions began or the rate at which they developed. An additional prospective investigation with stricter preoperative staging should be pursued to better compare the use of these two limb-sparing techniques and to assess the possible trends in metastasis or local recurrence.

The small sample size of this retrospective study was an overall weakness and limits the ability to make significant claims regarding the advantage of one of these two limb-sparing techniques over the other or over other available limb-sparing techniques. Additional large-scale studies need to be performed to compare the healing, function, and survival rates following each of these surgeries. The ulna appears to be a good donor site that is relatively easy to access following tumor removal from the radius and appears to hypertrophy well in response to weight-bearing stresses to provide a strong repair. Overall, the ulnar graft techniques used in this study demonstrated excellent healing properties, resulted in similar long-term complications as other limb-sparing options, and allowed all dogs to bear weight on the operated limb with acceptable limb function for the remainder of their lives.¹⁰

Conclusion

The grafting techniques used in this study resulted in bony healing and good to excellent limb function. Complications were expected with this complex surgery and were not obviously different from other limb-sparing techniques such as cortical allografts and metal endoprostheses. Unlike limb-sparing surgery using cortical allografts and endoprostheses, initial observations with vascularized ulnar grafts suggest that infection can be clinically resolved with the administration of antibiotics. Although removal of the bone plate is possible to resolve persistent infections, based on the outcomes in this study, implant removal should not be routinely performed as the hypertrophied graft does not appear to be strong enough to support the weight of the patient without added stabilization. Additional research is needed to evaluate the potential cause of this failure more fully before routine plate removal can be recommended. The mean and median STs were 29.3 mo (820 days) and 25 mo (700 days), respectively, for vascularized ulnar graft limb-sparing surgeries. Overall, this retrospective study acts as a preliminary clinical investigation inviting more specific research to be performed in the area of vascularized grafts for the replacement of distal radial OSA. Specific healing times for ulnar vascularized grafts should be determined, the time until possible implant removal should be elucidated, and the extent of radial bone that could ultimately be replaced by the ulna should be investigated in greater depth.