Osteosarcoma is the most common malignant bone tumor. We see 80 to 120 new patients with osteosarcoma every year; 40% to 50% of the patients present with either locally advanced disease or with metastatic disease. For the patients suitable for limb salvage, neoadjuvant chemotherapy followed by wide resection and reconstruction of the defect has become the worldwide standard of care. However, limb salvage surgery requires high-level infrastructure: an experienced team of surgeons and oncologists, good-quality prostheses, a good tissue bank for allografts, good blood bank and blood product transfusion facilities, and good intensive care facilities. These requirements present a challenge in a developing country with severe cost constraints.
We developed an orthopaedic oncology service in November 1999 with the objective of providing quality care at an affordable cost to the Indian population. For the first time in India, two dedicated orthopaedic oncologists (MA, AP) were managing the treatment of these patients. We developed a low-cost indigenous customized megaprosthesis, chemotherapy guidelines, and limb salvage techniques, and established an oncology team. Our protocols for chemotherapy were aimed at being cost-effective and therefore excluded methotrexate. Our two protocols included a newer, more intensive one (Table 1) and an older, less intensive and much cheaper one (Table 2) intended to provide at least some chemotherapy to the patients.
We asked the following questions: What are the disease control rates of using a chemotherapy protocol without methotrexate? Does survival differ between the more intensive and expensive chemotherapy protocol and the less intensive and cheaper protocol? Does local recurrence depend on the surgical experience? Does histologic response to chemotherapy influence survival? What is the functional outcome with a majority of cases being treated with an indigenously fabricated customized endoprosthesis? What are the complications of limb salvage surgery using our approach?
MATERIALS AND METHODS
We prospectively collected data on 135 selected patients with conventional high-grade osteosarcoma but without known metastases who underwent chemotherapy and limb salvage surgery between January 2000 and February 2004. We compared the histologic response and survival with the two protocols. Further demographic details included age, gender, side, site, and function at last followup.
We estimate the number of newly diagnosed high-grade osteosarcoma patients between January 2000 and February 2004 between 400 and 450. All patients were staged by clinical assessment and with CT scans of the chest and a bone scan. The local disease was assessed with an MRI. Patients were counseled regarding chemotherapy, costs, and the options for local treatment. Only those able to take chemotherapy were offered limb salvage. Rotationplasty was considered a limb salvage procedure and offered to patients unable to afford standard indigenous or expandable prosthesis, or those unsuitable for prosthesis reconstruction because of locally advanced but without metastatic disease. Amputation was offered to those with locally advanced disease or those who refused rotationplasty or prostheses. Chemotherapy was administered to 158 patients without metastases, 23 of whom underwent amputation. The remaining 135 patients underwent limb-saving surgery and formed our study group. There were 96 male patients and 39 female patients. Ten patients were younger than 10 years of age. Most (98) were in the second decade of life (Table 3). The distal femur (72) was the most common site, followed by the proximal tibia (28) and proximal humerus (19) (Table 4). Of the 135 patients, 120 (89%) were available for followup, with a minimum of 24 months (mean, 32.4 months; range, 24-72 months). Forty-five of the 120 patients were operated on in the first 2 years of this study period from January 2000 to December 2001 while 75 were operated between January 2002 and February 2004.
If an open biopsy had been performed outside and slides were available, a review was performed by our pathologist (N. A. Jambhekar). If slides were unavailable or the patient presented without biopsy, a core needle biopsy was done. The grading was performed per the World Health Organization classification of tumors.3 The specimens were examined grossly to note the size and extent of the tumors, distance from the cut margins, and involvement of soft tissue and skin. Subsequently, the maximum surface of the entire tumor-bearing area was exposed by a coronal section. A 5-mm thick parallel slab of this coronal section was obtained by sectioning longitudinally. The tumor-bearing area was then graphically divided into squares by horizontal and vertical lines so as to obtain a grid. These squares were serially numbered to enable identification and a diagrammatic record was kept. The tumor slab was then cut with a hand saw and each section was numbered and separately processed for study. The response to chemotherapy was assessed by the criteria of Huvos.5
The chemotherapy protocol depended on what the patient could afford. For patients who could not afford the standard intensive but expensive new protocol (Table 1), a less intensive old protocol that had fewer cycles and smaller dosages and therefore smaller growth factor requirements was used (Table 2). Most patients were given three cycles of chemotherapy before surgery. More preoperative cycles were given when surgery had to be delayed. Sixty-five of these patients received the newer protocol of chemotherapy, while 55 received the less expensive older protocol.
The surgical procedure performed depended on the extent and site of the lesion (Table 5). Allografts were used for diaphyseal defects where joint could be spared. Osteoarticular allografts were not used as we do not have the facilities. The type of prosthesis (indigenous or an import such as the Howmedica Modular Resection System [Stryker Orthopaedics, Mahwah, NJ]) depends on what the patient chooses and can afford. Ninety-two patients had wide resection and customized megaprosthesis. For the distal femur (Fig 1) or proximal tibia (Fig 2), endoprosthesis was the most common option. After surgical excision of the diseased bone 92 patients underwent reconstruction with endoprosthesis. In 38 patients with distal femur disease and 18 patients with proximal tibial disease, reconstruction was performed using an indigenous customized total knee replacement endoprosthesis (Sushrut-Adler, Muradpur, India) (Fig 3). In 11 patients with distal femur disease and four patients with proximal tibial disease, the Howmedica Modular Resection System total knee replacement endoprosthesis was used. In 14 patients with upper-end humerus disease, a customized proximal humerus prosthesis was used. In three patients with upper-end femur disease, a customized bipolar prosthesis was used. In three patients with disease involving the entire humerus, a customized total humerus prosthesis was used. In one patient with distal humerus disease, a total elbow replacement prosthesis was used. In 19 patients with distal femur disease and four patients with proximal tibial disease, rotationplasty was performed as a limb salvage procedure. In four patients with diaphyseal disease, after wide excision of the diseased segment, intercalary allograft was used for reconstruction. In two patients with diaphyseal disease, a nail and cement spacer was used for reconstruction. In two patients with upper-end humerus disease, clavicula-pro-humero reconstruction was performed10 after wide excision. In three patients with upper-end humerus disease, nail and cement reconstruction was used after wide excision of the diseased part of the bone. In four patients with upper-end fibula disease, only wide excision was performed. In one patient with tibial disease, wide excision with tibialization of the fibula was performed. Two patients, one with proximal radius disease and the other with distal ulna disease, underwent wide excision. In two patients with distal-end radius disease, wide excision with centralization of ulna was performed.
The indigenous prosthesis was a customized fixed-hinge stainless steel implant designed and fabricated locally. It is anatomic in shape and has standardized dimensions, except for the resection length and intramedullary stem diameter. We have designed this implant with dimensions to suit the majority of our patients. We cemented the prosthesis into the bone with pres-surization using a canal restrictor and a cement gun. A customized stainless steel joint was used for the proximal humerus (Fig 4) and the proximal femur. For the proximal femur, we fitted a bipolar cup onto the customized shaft. We used a Howmedica Modular Resection System when the patient could afford one. This is a fixed-hinge chromium-cobalt uncemented implant. Intercalary resections were reconstructed either with a cement spacer or with an allograft. The allograft was obtained from the in-house tissue bank.
Twenty-three patients had a rotationplasty (Table 5) after detailed counseling. We frequently performed these in young children to avoid the problems of limb length discrepancy. For the proximal radius or distal ulna or fibula, resection without any reconstruction was done. For two children with a proximal hu-merus tumor, the clavicle was rotated into the defect (claviculapro-humero operation).11
Drains were removed between day 5 and day 7 when drainage was reduced to less than 100 mL in 24 hours. For the knee prosthesis patients, mobilization began from the second postoperative day. Weight bearing ambulation on axillary crutches was started from day 4. For the proximal tibia replacement, the patellar tendon was attached to the hook on the prosthesis and a medial gastrocnemius flap was used for biological anchorage. Chemotherapy was restarted within 2 to 3 weeks of surgery after wound healing.
Patients with patellar reattachment were immobilized in a cast for 6 weeks, after which supervised mobilization was started to gain flexion and also maintain extension strength. Initially flexion was permitted only to 30° until there was no lag or only 5° extensor lag. After this, flexion was gradually increased but ensuring extension lag did not increase over 10°. Emphasis was placed on regaining extension rather than flexion. For the customized shoulders, an elbow sling was used for 4 to 6 weeks, and the patient was encouraged to use the elbow and hand. A polypropylene mesh was used to anchor the prosthesis to the glenoid to prevent subluxation.9 For the rotationplasty patients, range-of-motion exercises to maximize flexion and extension of the ankle started in the preoperative period and continued postoperatively. The ankle dorsiflexors and plantar flexors muscles were strengthened. External prostheses were fitted at 6 weeks. For the fibulectomies and the one sacroiliac resection, simple excisions without any reconstruction were performed. Patients were mobilized on crutches as soon as possible within the limits of pain. Every 3 months, the patients were assessed for pain, function, local disease control, and pulmonary metastases.
We (MA, AP, CA, MS) evaluated function at followup using a modified Enneking system.2 Local and chest radiographs were performed every 3 months and a chest CT scan was taken every 6 months.
Descriptive statistics were given by frequency and percentages. We obtained survival using the Kaplan Meier method.6 The disease-free survival was calculated from time of surgery to the first adverse event (local recurrence or systemic metastasis) or to the last followup. We used a log-rank test to compare survival between patients having the older and newer chemotherapy protocols (significance at p < 0.05). Statistical analysis was performed with SPSS Inc (Chicago, IL) v.14 software.
Seventy-three of the available 120 patients (61%) were disease-free at followup (Fig 5). The likelihood of being disease-free was similar with the older and newer protocols (32 of the 55 patients or 58% who received the old protocol, and 41 of the 65 patients or 63% who received the new protocol) (Fig 6). Because these percentages were similar, we combined the results of the two groups and presented them as one group. Most events occurred between 12 and 24 months from surgery (Fig 5).
Local recurrence was seen in 18 of 120 (15%). Seventeen of these 18 patients developed a local recurrence with pulmonary metastases. One patient had an isolated local recurrence that was excised and the patient is now disease-free 4 years after excision. Patients operated on from January 2000 to December 2001 had a higher rate of local recurrence (10 of 45 patients or 22%) than those operated on between January 2001 and February 2004 (seven of 75 or 9.3%). This difference was statistically significant (p = 0.049). Thirty patients (25%) developed pulmonary metastases without local recurrence. Some of these patients also had other sites of metastatic disease, including the brain (two), skin (one), and lymph nodes (two). No patient developed isolated bone metastases.
Only the group with a 100% necrosis had an improvement (p = 0.016) in survival compared with the other groups (Figs 7, 8). Thirty-four patients (28%) showed a Grade IV or complete response to chemotherapy. Another 29 patients (24%) showed a Grade III response or greater than 90% necrosis. Overall, 52% showed a good response to chemotherapy. Patients with Grade IV response (100% necrosis) had the best survival, with 27 of the 34 patients (79%) disease-free. Of the 29 patients with Grade III response (90-99% necrosis), 19 (65%) were disease-free. Of the 42 patients with Grade II response (50-89% necrosis), 21 (50%) were disease-free. Of the 15 patients with Grade I response (<50% necrosis), six (40%) were disease-free. Most patients reported a substantial decrease in pain and in the size of the swelling after starting chemotherapy. Two patients who were considered nonsalvageable and one considered borderline before chemotherapy were considered salvageable after chemotherapy. No patient had an increase in tumor size.
The average functional score was 25.5 of 30 (85%) for the knee prosthesis and 20 of 30 (66%) for the shoulder. The function in the shoulder group was scored low due to the inability to perform overhead activities resulting in severe function restriction.
Forty-three of the original 135 patients (32%) had a complication, half of whom needed another operation for the management of the complication. Ten of the patients had more than two additional procedures. Three patients had superficial flap necrosis, which resolved with nonoperative treatment. Eight patients (7%) had infection after surgery. In three patients the infection was controlled with a wound wash and intravenous antibiotics for 6 weeks. Two patients with knee megaprosthesis subsequently underwent rotationplasty after repeated wound washes failed to control the infection. To eradicate the infection in these patients, the implant and the bone cement in the medullary canal of the bones were removed, and the patients were given antibiotic cement spacers and intravenous antibiotics. Two patients who had infection following knee megaprosthesis opted for an amputation after two wound washes failed to control the infection. One patient with total elbow replacement also ended up with implant removal after two wound washes failed to control the infection.
There were two cases of intraoperative vascular injury to the popliteal vessel. One was repaired with sutures and the other needed a vessel graft. The patient with a vessel graft had a vessel that was stuck to the tumor and injured on attempted separation. She subsequently developed local recurrence and pulmonary metastases. One patient who had rotationplasty underwent amputation on the first postoperative day due to limb ischemia. One patient with rotationplasty had venous stasis postoperatively. The venous circulation was restored after exploration of the wound.
Two patients with indigenous distal femur megaprostheses underwent revision for loosening, one at 2 years and one at 3 years. One was successfully revised, and one had an infection after revision and was converted to a rotation-plasty in a two-stage procedure. Two patients had an intraoperative insertion fracture of the femur during reaming (one Howmedica Modular Resection System and one indigenous prosthesis). Two patients with a Howmedica Modular Resection System implant had an undisplaced oblique insertion fracture of the tibial shaft from a large tibial component of a distal femur megaprosthesis. One female patient with a knee megaprosthesis implanted for a distal femur osteosarcoma had a local recurrence of disease which was intraosseous and just above the intramed-ullary stem. A wide excision with total femur replacement was performed. She subsequently developed infection, which repeated wound washes failed to control. A hip disarticulation was finally performed on this patient. Three patients with indigenous knee megaprostheses had breakage of the implant. All were successfully revised. We had one breakage of a Howmedica Modular Resection System distal femur implant, which was revised and became infected postoperatively. The implant was removed and a rotationplasty subsequently performed after control of infection. One patient with a distal femur Howmedica Modular Resection System implant had a spiral insertion fracture at the tip of the stem, which was noticed postoperatively and fixed with a cerclage wire and plate. This implant was revised at 2-year followup for loosening of the femoral intramedullary stem.
Two patients with claviculo-pro-humero reconstruction had fracture of the clavicle. Two patients with knee megaprostheses had common peroneal nerve palsy. One patient recovered completely. One patient with a proximal humeral prosthesis developed radial nerve palsy but also recovered. One patient who had an intercalary resection for diaphyseal disease of the femur and reconstruction with an allograft and long plate developed nonunion at the allograft-host bone junction. Bone grafting of the junctions was done. However, this patient subsequently had implant failure with breakage of the allograft. He was operated on again and had reconstruction with a live fibula with allograft composite. One patient with rotationplasty had nonunion with plate loosening. Repeated attempts at bone grafting and bone marrow injection failed to secure union. Finally, the loose plate was removed and intramedullary nailing was done. The bones subsequently united.
In a developing country like India, amputation is not easily accepted; apart from being socially looked down upon, it is difficult for an amputee to lead a normal daily life. Medical insurance is available only to a small percentage of the population, with most of the population economically constrained. Few public hospitals provide quality care at highly subsidized costs, but they tend to be overcrowded. Alternative medicine is still preferred by many, and it is often too hard to seek quality medical care in a far-off city for an average rural person. Limb salvage therefore remains an unlikely option and amputation the most practical alternative for a large majority.
We note several limitations. The study was performed over a period of 4 years during which time the team became more experienced, as evidenced by our selection of cases and our technical abilities when performing a limb salvage surgery. Our higher local recurrence in the initial period is a testimony to this. Our group of patients ranged from those with rural and poor socioeconomic backgrounds to very urban and well-to-do backgrounds. Often our patients did not receive chemotherapy on schedule due to nonavailability of funds. Surgery was repeatedly delayed owing to an inability to accommodate patients in a timely fashion, due to a long waiting list and inadequate operating room availability. Tumor size varied from very small to very large, a factor not considered an independent variable in this study. There was a higher percentage of local recurrence than reported in the literature, often due to the selection of large tumors for salvage (initial enthusiasm coupled with surgical inexperience). Our local recurrence rate in the first 2 years of this series was 22% (10 of 45) but decreased to 9.3% (7 of 75) in the following 2 years. Three patients with 100% necrosis on the more expensive newer treatment protocol had a local recurrence (all in the first half of the series), followed by pulmonary metastases in two patients.
With economic growth, especially in cities, infrastructure has developed and some of the hospitals are providing a high standard of care at a reasonable cost. Our hospital is one such center where the only costs to a patient are the prosthesis, the consumables, and the drugs. Chemotherapy drugs are also now available at a much lower cost than in the past, which has led to the development of low-cost nonmethotrexate-based protocols for osteosarcoma. In November 1999, orthopaedic oncology became a separate service run by dedicated orthopaedic oncologists. Because availability of a megaprosthesis is the backbone of limb salvage surgery, our first target was to develop a low-cost megaprosthesis.
Many custom-made implants in the Indian market have been designed without adequate attention to the biomechanical demands. These implants, made with stainless steel and small diameter stems, were inherently weak. Many of these implants broke or became dislocated if the patient survived. Since the survival of patients with these implants was poor because of lack of chemotherapy protocols, breakage was not a major problem; few patients survived long enough for implant problems to develop. Limb salvage was a palliative procedure, providing patients with better quality of life while they were alive. Because the limb salvage program was poorly developed, most manufacturers were not interested in investing in the technology necessary to make prostheses comparable to the standard joint implants because they would not generate enough money. We therefore developed a customized joint made in 316L stainless steel. We found an orthopaedic implant manufacturer willing to devote unused machine time toward making a few of these joints. With this, our limb salvage program developed and our prostheses improved, and we are now doing about 80 to 90 of these implants annually. Our current design (Restor system, Sushrut Adler, Muradpur, India) is modular with the stem in titanium to reduce the incidence of breakage. This is noteworthy since two of the seven revisions in this series for loosening and breakage ended with an infection and subsequent rotationplasty.
Our experience with the Howmedica Modular Resection System implant was limited to 15 cases (11 distal femur and four proximal tibia). One implant broke at 5 years. In addition to the high cost, the minimum intramedullary size of 11 mm is a constraining factor for our patient population. We also had one femoral insertion fracture and two tibial insertion fractures. We are now using the Global Modular Replacement System (Stryker Orthopaedics), which has 9- and 10-mm stems and has a rotating hinge unlike the Howmedica Modular Resection System.
Our surgical morbidity has remained in the acceptable range, comparable with other series.1,7,10 The infection rate in our series was 6.7% (8 of 120); half were delayed infections presenting a few months or a year after surgery. Often infection is contracted during postoperative chemotherapy when medical attention is not sought in time during neutropenia. Patients often prefer to get chemotherapy from local physicians who may not be as well trained as those in a major medical center. Sometimes our patients are not able to afford the expensive higher antibiotics. Two of our patients with infections had amputations and two had a rotationplasty.
Our initial patients showed a high local recurrence rate of 15% (18 of 120). This high rate may have been due to our inexperience in surgical technique and case selection. The rate has now fallen steeply as evidenced by fewer recent local recurrences. The tumors in our patients have been large, and often we have been forced into limb salvage by a patient who is unwilling to have an amputation. Similar findings have been reported in a Brazilian osteo-sarcoma study.8 Most of our patients with a local recurrence have also had pulmonary metastases (17 of 18). This may suggest either aggressive or advanced disease at presentation. Good response to chemotherapy is associated with lower local recurrence rates.4 Local recurrence occurred in 3 of 34 (9%) patients in the group with 100% necrosis, 4 of 29 (14%) in the group with 90% to 99% necrosis, and 11 of 57 (19%) of those with less than 90% necrosis (Table 6). Though not statistically significant due to the small numbers, it points to the trend of less local recurrence with better response to chemotherapy. Size has not been recorded and studied in this analysis and therefore cannot be assessed as an independent factor. We believe the high local recurrence rate is due to a combination of factors like large tumor size, surgical inexperience, and poor necrosis.
We have endeavored to develop affordable, robust, and functional indigenous customized megaprostheses for this purpose. The joints are anatomic and provide valgus and hyperextension. They have a well-formed trochlear notch for smooth patellar tracking. However, because they are made of stainless steel, they are certain to fail at some point. This was seen in the three breakages in our series even after relatively short followup. The indigenous pros-theses provide satisfactory cosmesis and function. Our functional evaluation showed 85% function, comparable with the studies from the West.1,7,10 We have not replaced the patella in any of our patients. We are now developing a new rotating hinge design made from titanium, and this should reduce the breakage problem. This has increased the cost by 20% but should greatly reduce the risk of breakage.
Rotationplasty, which allows the ankle to function as the knee, provides an excellent alternative, especially in children. It compensates for growth, is a permanent procedure, and provides excellent function with an extension prosthesis. Children are even able to participate in recreational sports. It is also very affordable, unlike joint replacement, and costs $400 in our setup including the cost of the external prosthesis. The cost factor is of great importance in developing countries such as ours. An expandable prosthesis would cost approximately $20,000, making it practically out of reach for most of our patients. Even an indigenous nonexpandable prosthesis was out of reach of many of our patients at a cost of $800 (compare with $8000 for Howmedica Modular Resection System). The cost of any revision surgery would again be unaffordable for many of our patients. In this situation, our patients have accepted the cosmetic disfigurement of rotationplasty as the function has been good. In the rural setup, the ability to squat and load the limb is a major advantage over a protected life for a megaprosthesis patient. As in the West, very few of our well-to-do patients agree to rotationplasty. With the help of a prosthesis manufacturer in South Africa (ISIQU Orthopaedics, Capetown, South Africa), we have been able to get an expandable prosthesis for $5000. As a result, many more patients opt for this implant, unlike the current series.
For the proximal humerus, both the deltoid and the rotator cuff are excised for wide margin and the prosthesis therefore functions as a mere spacer. A polypropylene mesh is used to create an artificial capsule to prevent subluxation.9 Though shoulder function is poor, the elbow and hand retain useful function, better than after a shoulder disarticulation or forequarter amputation and external prosthesis.
Limb salvage has now come of age in developing countries such as ours. In the recent past, osteosarcoma has been treated with pessimism because of the nonavailability of effective chemotherapy in our country. With the advent of the new chemotherapy protocol and the availability of affordable indigenous joints, limb salvage and long-term survival have become a reality. The 2-year results with a survival rate of 61% appear similar to those from other studies,8,12 although inferior to those from developed centers in the West.1,4,7,10 These are the early results in a new unit and are likely to show improvement with time. The surgical results too have been comparable in functional outcome and morbidity to those reported in the literature.1,7,10 It has become possible now to get results comparable with those in the West at a fraction of the cost. Even very low cost chemotherapy appears effective. The large numbers of patients provided a sizeable experience in a short time, and accelerated developing local technology to produce a megaprosthesis. A high-quality rotating hinge megaprosthesis currently being developed indigenously should help reduce long-term complications.
The authors thank Dr. Nirmala A. Jambhekar, Professor of Pathology at the Tata Memorial Hospital, for evaluating the biopsies and specimens of the cases in this series. We wish to thank her for graciously agreeing to step down as an author in order to comply with CORR specifications. The authors thank Anagha Kakade, who at very short notice helped us organize and evaluate the data in the SPSS packages for statistical evaluation. The authors thank Mr. M. N. Bobde and Mrs. Saraswati in procuring patient charts for evaluation.
1. Bacci G, Ferrari S, Bertoni F, Ruggieri P, Picci P, Longhi A, Casadei R, Fabbri N, Forni C, Versari M, Campanacci M. Long term outcome for patients with non-metastatic osteosarcoma of the extremity treatment at the istituto ortopedico rizzoli according to the istituto ortopedico rizzoli/osteosarcoma-2 protocol: an updated report. J Clin Oncol
2. Enneking WF, Dunham W, Gebhardt MC, Malawar M, Pritchard DJ. A system for the functional evaluation of reconstructive procedures after surgical treatment of tumors of the musculoskeletal system. Clin Orthop Relat Res
3. Fletcher CDM, Unni KK, Mertens F. Pathology and Genetics of Tumours of Soft Tissue and Bone. World Health Organisation Classification of Tumours
. Lyon, France: IARC Press; 2002.
4. Grimer RJ, Taminiau AM, Cannon SR. Surgical Subcommittee on behalf of the European Osteosarcoma Intergroup: surgical outcomes in osteosarcoma. J Bone Joint Surg Br
5. Huvos AG. Bone Tumors: Diagnosis, Treatment, and Prognosis. 2nd ed
. Philadelphia, PA: Saunders; 1991.
6. Kaplan EL, Meier P. Nonparametric estimation for incomplete observation. J Am Stat Assoc
7. Lindner NJ, Ramm O, Hillmann A, Roedl R, Gosheger G, Brink-schmidt C, Juergens H, Winkelmann W. Limb salvage and outcome of osteosarcoma: the University of Muenster experience. Clin Orthop Relat Res
8. Petrilli AS, de Camargo B, Filho VO, Bruniera P, Brunetto AL, Jesus-Garcia R, Camargo OP, Pena W, Pericles P, Davi A, Prospero JD, Alves MT, Oliveira CR, Macedo CR, Mendes WL, Almeida MT, Borsato ML, dos Santos TM, Ortega J, Consentino E. Brazilian Osteosarcoma Treatment Group Studies III and IV. Results of the Brazilian Osteosarcoma Treatment Group Studies III and IV: prognostic factors and impact on survival. J Clin Oncol
9. Puri A, Agarwal M. Use of polypropylene mesh to stabilize skeletal reconstructions after resection for bone tumors. J Surg Oncol
10. Renard AJ, Veth RP, Schreuder HW, van Loon CJ, Koops HS, van Horn JR. Function and complications after ablative and limb-salvage therapy in lower extremity sarcoma of bone. J Surg Oncol
11. Tsukushi S, Nishida Y, Takahashi M, Ishiguro N. Clavicula pro humero reconstruction after wide resection of the proximal humerus. Clin Orthop Relat Res
12. Yip KM, Leung PC, Kumta SM. Osteosarcoma in Hong Kong. Clin Orthop Relat Res