Pelvic bone sarcomas account for 3% to 6% of all bone sarcomas.3,22 Wide surgical resection combined with chemotherapy, radiotherapy, or both is superior over medical treatment alone in terms of overall survival, local recurrence, and metastasis.9-11,21,23 Because of the three-dimensional (3-D) configuration of the pelvis, especially at the sacroiliac junction, tumors often grow to large sizes before diagnosis. In very large tumors, pelvic viscera and nervous structures are invaded, which make wide resection more difficult.3,11,12,22 In addition, primary tumors of iliac bone origin abutting the sacroiliac joint could invade the sacral bone and the sacral nerve roots through or around the sacroiliac joint.16 This has rendered wide resection of the sacroiliac joint challenging because an extensive procedure has been required to remove the diseased area, leading to a high rate of local complications16 and functional impairment.16 Functional outcome also depends on reconstructing the pelvis and preserving hip function.11,12,17
Difficulties in obtaining tumor-free margins and achieving function-preserving reconstruction may lead to amputation, which is considered a better strategy than internal hemipelvectomy in patients who require sacroiliac joint resection. In deciding to proceed to an internal hemipelvectomy rather than an external hemipelvectomy, the internal hemipelvectomy should at least lead to the same tumor-free margins as amputation and provide superior functional outcome with acceptable morbidity.
We wondered whether sacroiliac joint resection with limb salvage for malignant pelvic tumors can be performed with adequate (wide or marginal) surgical margins. We also asked whether the functional results were linked to nerve sacrifice or amount of bone resection. Finally we wondered whether our approach would lead to subsequent morbidity (including reinterventions related to complications).
MATERIALS AND METHODS
We retrospectively reviewed the medical records of 40 consecutive patients who had surgical resection of the sacroiliac joint for primary malignant tumors from January 1982 to June 2001 at the Bicêtre Teaching Hospital, Le Kremlin-Bicêtre, France (Table 1). There were 22 male and 18 female patients, with a mean age at diagnosis of 28 years (range, 11-66 years). The patients were followed up for a minimum of 3 years or until death, with a mean of 64 months (range, 6-180 months). Three patients were lost to followup; one after 1 year and two after 1 month.
The tumors either originated in the ilium and extended proximally to the sacrum and distally to the acetabulum, or originated in the sacrum and extended to the sacroiliac joint and ilium. Invasion of the ilium and sacrum (Table 1) was defined as cartilage or bone destruction as seen on computed tomography (CT) and/or a mass effect and abnormal signal on magnetic resonance imaging (MRI).16 Invasion of the ilium was confined to the iliac wing in 23 (57.5%) patients. The roof of the acetabulum was invaded in 10 (25%) patients, and the entire acetabulum was invaded in four (10%) patients. The lesion extended to the ischiopubic region in three (7.5%) patients. Only two patients (5%) had no infiltration of the sacral wing. In 27 (67.5%) patients, the tumor invaded the sacral wing but did not extend to the sacral foramina. Nine (22.5%) tumors invaded the ipsilateral sacral foramina, and two (5%) tumors invaded the middle of the sacrum without extending to the contralateral sacral foramina.
According to the Enneking tumor classification system,5 the stage was IB in six patients (15%), IIA in three patients (7.5%), and IIB in 31 (77.5%) patients. Thirty-nine patients had no regional or distant metastasis at surgery. One patient had intracardiac surgery with complete resection of an auricular tumor that proved to be a metastasis from pelvic chondrosarcoma. The histopathologic diagnosis of bone sarcoma was made preoperatively in all 40 patients by examination of a surgical (29 patients) or percutaneous (11 patients) biopsy specimen (Table 1). The three main tumor types were primary neuroectodermal tumor (14 patients), chondrosarcoma (11 patients), and osteosarcoma (7 patients). The indications for limb-salvage surgery were absence of metastases and of iliac vessel or sciatic nerve invasion, and anticipated feasibility of achieving tumor-free margins. All surgical procedures were done at the same institution (Bicêtre Teaching Hospital, Le Kremlin-Bicêtre, France) by the same surgeons (CC and GM).
For the postoperative histologic evaluation, the resected bone segment with the surrounding tissues was oriented by the surgeon (CC or GM), and the surgical margins were examined. The criteria described by Pring et al17 for radical margins cannot be used for pelvic tumors, for anatomic reasons. Instead, we used the criteria of Fuchs et al,6 defining the margins as adequate when resection was wide or marginal without contamination and as inadequate otherwise (intralesional, wide contaminated, or marginal contaminated).
We used the extended anterolateral approach in one patient at the beginning of our series. The other 39 patients had a combined extended ilioinguinal and posterior approach while in the lateral decubitus position.13 We first performed the ilioinguinal approach, which allowed tumor dissection from the neurovascular and visceral structures and ligation of the terminal branches of the hypogastric vessels to decrease blood loss during the posterior osteotomy of the sacrum. We then performed a modified posterior approach.13,20 The incision was divided into three segments according to the type of resection to be performed. For the first segment, we made a classic midline incision to provide exposure of the lumbar spine and upper part of sacrum. For the second segment, the incision was extended (including the previous biopsy tract) to the posterior and superior iliac spine and the posterior part of the great trochanteric process (as in the Kocher-Langenbeck hip approach). This segment exposed the posterior aspect of the sacroiliac joint and iliac wing including the sciatic notch and acetabular roof. The third segment extended longitudinally along the femoral shaft, providing access to the inferior part of the acetabulum and to the ischial region. Sacral laminectomy provided access to the spinal canal and allowed protection, ligation, and transection of the sacroiliac nerve roots if necessary. We performed a vertical sacral osteotomy from the posterior approach and traversed the sacral foramina alongside the sacral nerve roots. Protection of the large vessels, viscera, and lumbosacral trunk was necessary during the osteotomy via the anterior approach. Mobilization of the lumbosacral trunk from the sacral wing was difficult in some patients, and transection of the lumbosacral trunk was required in patients who had tumor spread through the anterior wall of the sacrum. In two patients, a computer-assisted image-guided system was used during surgery to secure the osteotomy because the tumor was too large to allow control via the anterior approach.
We performed different combinations of resections depending on the tumor location and extent of bone invasion (Fig. 1). We defined four types of sacral resection according to the location of the sacral osteotomy: Type A, ipsilateral sacral wing (two patients); Type B, ipsilateral sacral foramina (27 patients); Type C, midline (nine patients); and Type D, contralateral sacral foramina (two patients). The type of iliac resection was defined according to Enneking et al5 (Fig. 1). Twenty-three patients had Type I excision, four patients had Type I-II excision with removal of the entire acetabulum, 10 patients had Type I-IIp excision, which we defined as a Type I-II procedure with partial removal of the acetabulum, and three patients had Type I-II-III excision.
Reconstruction varied according to age of the patient, extent of bone and nerve resection, and tumor type. In one patient (Type I-IIp + B), the objective was to obtain ischiofemoral pseudarthrosis. The other patients had reconstruction to restore continuity between the spine and the pelvis. In 21 patients with Type I resection, a pedicle screw in the fifth lumbar vertebral pedicle was connected with a rod to a pedicle screw in the posterior acetabular column, and a pedicle screw inserted in the remaining sacral body was connected with a rod to a pedicle screw in the anterior acetabular column (Fig. 2). An autologous tibial graft was used to achieve fusion. Contralateral lumbosacral arthrodesis with or without instrumentation also was performed. At the beginning of the series, two other patients with Type I resections had tibial graft arthrodeses without instrumentation. In the 10 patients with Type I-IIp resection, we performed lumbariliac osteosynthesis and arthrodesis and several different procedures for the hip. Creation of a hip pseudarthrosis in the remaining acetabulum was attempted in six patients and ischiofemoral arthrodesis was done in two patients, respectively. Two other patients had sacroischiatic arthrodesis with tibial grafting combined with ischiofemoral arthrodesis (n = 1) or flail hip (n = 1). The four patients with Type I-II resection had pelvic allografting (n = 1), ischiofemoral arthrodesis (n = 2), or lumbar-iliac instrumented arthrodesis combined with a Puget procedure18 and total hip arthroplasty (THA) (n = 1). In the three patients with Type I-II-III resection, the reconstructive procedures consisted of attempted sacrofemoral pseudarthrosis (n = 1), instrumented lumboiliac arthrodesis (n = 2) combined with ischiofemoral arthrodesis (n = 1), and interposed femoral allograft (n = 1). In two patients, a vascularized omental flap was used at the end of the procedure to achieve wound closure.
Patients with high-grade tumors (n = 19) received chemo-therapy and radiotherapy after the surgical resection.
All postoperative complications were recorded. Pseudarthrosis was suspected clinically if the patient had growing pain and progressive increasing impairment for walking. Radiographically, pseudarthrosis is difficult to assess in this region but was considered present if loosening or rupture of the device was present.
At the last followup, patients were seen at the hospital out-patient clinic or interviewed by telephone. Local recurrences and deaths were recorded. In recurrence-free survivors, function was assessed using the 1993 MSTS score.4 Twenty-eight (73.5%) patients were available for functional evaluation. A numerical value (0-5) was assigned to each of six categories: pain, disability, emotional acceptance, need for external supports, walking ability, and gait. The values were summed, and the result was expressed as the percentage of the highest possible score (30 points). Given the extent of the surgical procedure and the long rehabilitation period, function was evaluated at least 1 year postoperatively after consolidation of a pseudarthrosis, if relevant. This criterion left 28 patients available for functional assessment, which was performed every 6 months until the last followup. The mean followup was 70 months (range, 12-180 months).
Adequate margins were achieved in 28 (74%) patients, including 20 (52%) with wide margins and eight (21%) with marginal margins (Table 1). Of the 10 (26%) patients with inadequate margins, two (5%) had wide contaminated margins (wide resection of osteosarcomas with malignant thrombi in the drainage veins [Batson plexus]), four (10%) had marginal contaminated margins, and four (10%) had intralesional margins. Local recurrence developed in two of the 28 patients (7%) with adequate margins compared with seven of the 10 (70%) patients with inadequate margins.
Functional results differed according to the type of bone and nerve resection (Table 2). After Type B sacral resection, leaving the sacroiliac roots intact combined with Type I iliac resection (n = 8), the mean MSTS score was 81% (range, 26-96%). However, in the three patients treated with the same type of bone resection (sacral Type B and iliac Type I) and with resection of the lumbosacral trunk (union of the L4 to L5 anastomosis to the L5 nerve root), the mean MSTS score was only 48% (range, 40- 60%). When the sacral resection was Type C (with removal of the S1, S2, and S3 roots) and iliac resection was Type I without lumbosacral trunk resection (n = 4), the mean MSTS score was 65% (range, 26-83%). In patients with Type B sacral resection and Type I-II iliac resection (n = 4), the mean score was 61%. The mean score was 25% in patients (n = 3) with Type I-II-III iliac resection. However, when all sacral resection types were combined, the MSTS score was 32% in patients who required lumbosacral trunk removal (n = 4) compared with 71% in patients with lumbosacral trunk preservation and the same iliac resection types (n = 12). When all sacral resection types were combined and the lumbosacral trunk was preserved, the MSTS score was 40% after removing the acetabulum (n = 11) and 73% after preserving the acetabulum.
Complications were common. They were dominated by infections in the early postoperative period and mechanical failure at a distance after surgery. Early complications that required surgical revision occurred in 16 (40%) patients. Complications included deep infection (n = 15) and skin necrosis (n = 1). Skin necrosis occurred in the only patient who had surgery using the anterolateral approach alone. All 16 patients recovered after surgical débridements without removal of foreign material and with appropriate multidrug antimicrobial therapy. No infections occurred in the two patients treated with vascularized omental flaps. Mechanical failure with implant loosening in the lumbar-iliac reconstruction occurred in five patients (17%) who required revision surgery (instrumentation and autograft). Lumbosacral destabilization (L5-S1 spondylolisthesis or lumbar scoliosis) developed in two patients after lumbar-iliac reconstruction without contralateral instrumented lumbosacral arthrodesis and in the patient who did not have reconstruction. Pseudarthrosis occurred in seven patients, of whom six required revision surgery with autologous bone grafting. Fusion was obtained in these six patients. In two patients with partial acetabular resection (Type I-IIp), femoral head necrosis developed after 24 months, causing functional impairment. Prosthetic hip reconstruction was performed in one of these patients, and partial head resection without hip reconstruction was done in the other patient. There were 29 reoperations for infection or mechanical problems. At the last followup, 17 (42.5%) patients were alive, 20 (50%) had died, and three (7.5%) were lost to followup.
Primary bone malignancies of the sacroiliac joint spread to the nerves and hip before diagnosis.1,10,16 As a result, resection with wide or marginal margins requires a complex surgical approach, and reconstruction raises major challenges. Therefore the risk and benefits of limb-sparing surgery should be compared with those of amputation. To assist surgeons in evaluating possible benefits from internal hemipelvectomy, we sought to determine whether acceptable (wide or marginal, and uncontaminated) margins can be achieved in most patients who have an internal hemipelvectomy with sacroiliac joint resection for a malignant pelvic tumor. We also evaluated surgical margins and local recurrence rates, functional outcomes, and complication rates.
Limitations of the study include the retrospective design and the small number of patients in each resection reconstruction group; the small numbers precluded statistical tests. However, our study design and numbers of patients were sufficient for determining whether internal hemipelvectomy produced adequate margins. In addition, primary bone malignancies involving the sacroiliac joint are rare, and achieving larger group sizes would require a multicenter study, which would have to overcome the major obstacles raised by heterogeneity in management practices. Given the limited number (2) of surgeons, the approach was relatively uniform.
Surgical margins in the pelvic area are classified as adequate (wide or marginal, without contamination) or inadequate (other situations).6,17 We achieved adequate margins in 74% of patients, which is within the 25% to 88%6,10,17,21,22 range reported for pelvic sarcomas in any location. Adequate margins are more difficult to obtain when the sacroiliac joint is involved10,15,23 because the tumor is often large, usually extracompartmental (92.5% of the patients in our series), and occasionally responsible for distant tumor thrombi in the Batson plexus.22 Batson plexus thrombi cannot be detected preoperatively and resulted in wide contaminated resection in two of our patients.
The high rate of adequate margins supports the use of the internal hemipelvectomy in patients with pelvic malignancies involving the sacroiliac joint. However, anticipated ability to achieve adequate margins was among the criteria used to select patients for an internal hemipelvectomy. Local recurrence and complication rates must be considered when comparing this procedure with amputation. Local recurrence is more common with iliosacral tumors than with tumors at other pelvic sites.14 None of the previous studies focused specifically on malignancies involving the sacroiliac joint. Local recurrence rates for pelvic malignancies at any location ranged from 20% to 38%.10,11,14,17,21-23 Our recurrence was rare (7%) when margins were adequate and common (70%) when margins were inadequate. Other groups reported recurrences in 9% to 38%9,22of patients with adequate margins and 80%10 of patients with inadequate margins.
Reconstruction procedures increase the rates of infection and mechanical complications.3,8,10,12,14,15,22 Reconstruction after limb salvage procedure is not always performed.6,22 Function was poor in the three patients treated without lumbosacral stabilization, at the beginning of our series. Others also reported poor function after wide re-section without reconstruction, most notably in patients with flail hips11,12,14,17,22 with considerable limb shortening. Without reconstruction, progressive scoliosis may develop.6,13 Stability of the pelvic girdle was provided by the sacroiliac region22 because the major direction of the weightbearing force started from the lumbar spine to the sacral wing via the sacroiliac joint, the roof of the sciatic notch, and then to the acetabulum.11 We think restoration of posterior pelvic stability is mandatory and can be achieved with lumbar-pelvic instrumentation and fusion.
The status of the lumbosacral trunk was the main determinant of postoperative function in all resection categories. Therefore, when lumbosacral trunk removal is required, an internal hemipelvectomy with limb salvage may not be the best surgical option. Conversely, unilateral removal of the sacral plexus for the vertical osteotomy of the sacrum did not adversely impact function. However, impairment of anorectal and bladder function has been reported to be mild or transient.7,20 We evaluated function using the 1993 MSTS score, which does not consider anorectal or bladder function. Another determinant of function in our study was the status of the acetabulum, as partial or total resection of the acetabulum was associated with poorer function. Functional results were not different between partial and total resection of the acetabulum. However, partial resection is easier for reconstruction and allows a subsequent THA if needed.
Complication rates also affect choices among surgical options. Postoperative infection was the main complication of sacroiliac resection. The 37.5% infection rate was consistent with the previously reported range of 14% to 60%.10,11,14,17,22 The infection rate was greater after sacral osteotomy and reconstruction,6 which require a lengthy surgical procedure and leave a poorly vascularized region with a large residual cavity. To fill the cavity, limit hematoma formation, and improve blood supply, we started using a vascularized omental flap with the goal of decreasing the infection rate. The high infection rate casts doubt on the wisdom of complex reconstruction in the same step as tumor resection. Hip reconstruction can be performed as a second step11,14 to reduce the risk of infection, thereby allowing immediate completion of chemotherapy. The choice between immediate and delayed reconstruction is controversial.14 Mechanical failure occurred in 12% of patients and pseudarthrosis occurred in 17%. The reconstruction subgroups were too small for a meaningful comparison of the mechanical failure risk. External hemipelvectomy has been recommended for pelvic bone malignancies and has a 53% complication rate,2,8 similar to our 52% overall complication rate (infection plus mechanical failure). In addition, the emotional acceptance and function of patients are poor after an external hemipelvectomy.2 An external hemipelvectomy generally is reserved for patients who are severely debilitated and for patients with very large tumors that cannot be resected with adequate margins.
Sarcomas that involve the sacroiliac joint have a poorer prognosis than other pelvic sarcomas.3,9,14,22,23 However, we found adequate margins could be obtained during internal hemipelvectomy in approximately 75% of patients. Infectious complications were common but consistently responded to treatment without removal of fixation devices. Mechanical problems occurred in 27% of patients. Our findings linking functional outcomes to the type of resection and reconstruction should help surgeons in deciding the appropriateness of an internal hemipelvectomy.
We thank Lorna Saint Ange for editing.
1. Abdelwahab IF, Miller TT, Hermann G, Klein MJ, Kenan S, Lewis MM. Transarticular invasion of joints by bone tumors: hypothesis. Skeletal Radiol
2. Apffelstaedt JP, Driscoll DL, Spellman JE, Velez AF, Gibbs JF, Karakousis CP. Complications and outcome of external hemipelvectomy in the management of pelvic tumors. Ann Surg Oncol
3. Campanacci M, Capanna R. Pelvic resections: the Rizzoli Institute experience. Orthop Clin North Am
4. Enneking WF, Dunham W, Gebhardt MC, Malawar M, Pritchard DJ. A system for the functional evaluation of reconstruction procedures after surgical treatment of tumors of the musculoskeletal system. Clin Orthop Relat Res
5. Enneking WF, Spanier SS, Goodman MA. A system for the surgical staging of musculoskeletal sarcoma. Clin Orthop Relat Res
6. Fuchs B, O'Connor MI, Kaufman KR, Padgett DJ, Sim FH. Iliofemoral arthrodesis and pseudarthrosis: a long-term functional outcome evaluation. Clin Orthop Relat Res
7. Gunterberg B, Kewenter J, Petersen I, Stener B. Anorectal function after major resections of the sacrum with bilateral or unilateral sacrifice of sacral nerves. Br J Surg
8. Ham SJ, Schraffordt Koops H, Veth RP, van Horn JR, Eisma WH, Hoekstra HJ. External and internal hemipelvectomy for sarcomas of the pelvic girdle: consequences of limb-salvage treatment. Eur J Surg Oncol
9. Kawai A, Healey JH, Boland PJ, Lin PP, Huvos AG, Meyers PA. Prognostic factors for patients with sarcomas of the pelvic bones. Cancer
10. Kawai A, Huvos AG, Meyers PA, Healey JH. Osteosarcoma of the pelvis: oncologic results of 40 patients. Clin Orthop Relat Res
11. Kollender Y, Shabat S, Flusser G, Flusser G, Isakov J, Neuman Y, Cohen I, Weyl-Ben-Arush M, Ramo N, Meller I. Internal hemipelvectomy for bone sarcomas in children and young adults: surgical considerations. Eur J Surg Oncol
12. Langlais F, Lambotte JC, Thomazeau H. Long-term results of hemi-pelvis reconstruction with allografts. Clin Orthop Relat Res
13. Missenard G, Dubousset J, Genin J. Wide resection of the sacroiliac joint: technique, reconstruction, anatomical and functional results. Rev Chir Orthop
14. O'Connor MI, Sim FH. Salvage of the limb in the treatment of malignant pelvic tumors. J Bone Joint Surg Am
15. Ozaki T, Flege S, Kevric M, Lindner N, Maas R, Delling G, Schwarz R, von Hochstetter AR, Salzer-Kuntschik M, Berdel WE, Jurgens H, Exner GU, Reichardt P, Mayer-Steinacker R, Ewerbeck V, Kotz R, Winkelmann W, Bielack SS. Osteosarcoma of the pelvis: experience of the Cooperative Osteosarcoma Study Group. J Clin Oncol
16. Ozaki T, Rodl R, Gosheger G, Hoffmann C, Poremba C, Winkel-mann W, Lindner N. Sacral infiltration in pelvic sarcomas: joint infiltration analysis II. Clin Orthop Relat Res
17. Pring ME, Weber KL, Unni KK, Sim FH. Chondrosarcoma of the pelvis: a review of sixty-four cases. J Bone Joint Surg Am
18. Puget J, Uthéza G. Reconstruction of the iliac bone using the homolateral femur after resection for pelvic tumor. Rev Chir Orthop
19. Nakai S, Yoshizawa H, Kobayashi S. Anorectal and bladder function after sacrifice of the sacral nerves. Spine
20. Simpson AH, Porter A, Davis A. Cephalad sacral resection with a combined extended ilioinguinal and posterior approach. J Bone Joint Surg Am
21. Sucato DJ, Rougraff B, McGrath BE, Sizinski J, Davis M, Papandonatos G, Green D, Szarzanowicz T, Mindell ER. Ewing's sarcoma of the pelvis: long-term survival and functional outcome. Clin Orthop Relat Res
22. Wirbel RJ, Schulte M, Mutschler WE. Surgical treatment of pelvic sarcomas: oncologic and functional outcome. Clin Orthop Relat Res
23. Yang RS, Eckardt JJ, Eilber FR, Rosen G, Forscher CA, Dorey FJ, Kelly CM, al-Shaikh R. Surgical indications for Ewing's sarcoma of the pelvis. Cancer