Between January 1990 and March 2015, 99 patients with pelvic bone and soft tissue tumors underwent periacetabular resection at Korea Cancer Center Hospital. According to the pelvic resection category described by Enneking and Dunham,12) 43 were type I + II, 28 were type I + II + III, 25 were type II + III, and 3 were type II resections. Among these, we selected 24 patients who underwent RHA and 16 patients who underwent pasteurized autograft-prosthesis composite (PPC) reconstruction. Exclusion criteria were: (1) iliac resection at the sacroiliac joint (37 patients); (2) reconstruction with saddle prosthesis (8 patients); and (3) less than 2 years of follow-up without an event (14 patients). Finally, 23 male and 17 female patients with an average age of 35.3 years (range, 15 to 68 years) were included. Follow-up duration was a minimum of 13 months (average, 101 months; range, 13 to 301 months).

The criteria for RHA were: (1) tumors located mainly in region II and III (some partially involved region I) and (2) preoperative magnetic resonance imaging (MRI) showing no sign of femoral head involvement by tumor. PPC reconstruction was indicated for tumors when at least 1 of 2 landmarks on plain pelvic anteroposterior radiographs (ilioischial and iliopectineal lines) was intact.

Of the 24 patients who underwent RHA, 18 had primary malignant bone tumors, 3 had soft tissue sarcomas, and 3 had giant cell tumors (Table 1). Of the 16 patients who underwent PPC reconstruction, 11 had primary malignant bone tumors, 3 had benign aggressive bone tumors, and 2 had metastatic tumors (Table 2). Preoperative staging included plain radiography and MRI of the pelvis, computed tomography (CT) of the chest, and whole body technetium bone scan. Staging was determined according to Enneking's criteria.24) Neoadjuvant and adjuvant chemotherapy were administered to 13 patients. Initial tumor volume, pathologic diagnosis, extent of iliac lesion on MRI, osteotomy level from iliac crest, resection type, surgical time, surgical margin, and amount of transfusion were recorded. The tumor volume was calculated using 3 parameters (height, width, and depth) by the ellipsoid formula: [V = (4π/3)×a×b×c].25) The height, width, and depth of a tumor were measured on the coronal and axial plane MRI scans. The resection and reconstruction included the following procedures. An ilioinguinal approach was used for the main incision, and a satellite incision was made from the anterosuperior iliac spine to the greater trochanter, if necessary. In 12 of the 24 patients who underwent RHA, 2 or 3 osteotomies were made to preserve the iliac bone-hip flexor or abductor continuity, and to displace the bone block-muscle complex inferiorly and laterally. Pelvic resection was classified according to the system of Enneking and Dunham.12) In 16 cases of PPC reconstruction, 6 cases were type I + II, 5 were type II + III, 4 were type I + II + III, and 1 was a type II resection. In 24 cases of RHA, 13 were type I + II + III, 6 were type II + III, 4 were type I + II, and 1 was a type II resection. The margin in the 40 patients was wide in 25, marginal in 14, and intralesional in 1, according to Enneking's criteria.24) The preparation of pasteurized bone and reconstruction using a total hip prosthesis was performed as described previously.9) In RHA, the average osteotomy length from the iliac crest was 9.5 cm (range, 4 to 14.5 cm). After tumor removal, the femoral head was pushed up to the inferior aspect of the resected ilium. In 12 cases, the femoral head was fixed to the remaining ilium with wire. In the remaining 12 cases, no fixation was undertaken. In 12 cases of RHA, previously detached iliac bone block-muscle complex was repositioned to the iliac wing with wire. Postoperatively, patients who underwent PPC reconstruction were usually immobilized for 8–10 weeks in a one-and-a-half hip spica cast, which was followed by crutch walking until radiographic union was achieved. In patients who underwent RHA, only 3 were immobilized in a hip spica cast. The remaining 21 were allowed to ambulate when drains were removed (around 2 weeks after the surgery). Monthly plain anteroposterior and bilateral oblique radiographic examinations were performed until 2 years after the index operation. In patients who underwent RHA, trimonthly pelvic CT was performed to evaluate the formation of bony neo-acetabulum. More than 3/4 of circular bony neo-acetabulum formation was defined as complete, and less than 1/2 of a circle as partial (Fig. 1).

At final follow-up, 28 patients were event-free, 6 died of their disease, 5 had no evidence of disease, and 1 was alive with disease. Nine patients (22.5%) developed local recurrence, and their surgical margin status was marginal in 6, wide in 2, and intralesional in 1. Functional results were assessed by the Musculoskeletal Tumor Society (MSTS) system.26) Scores were recorded at the last visit for patients with intact PPC and who underwent RHA. For PPC reconstruction patients whose graft was removed, the functional scores before and after graft removal were recorded. Final limb length discrepancy on radiographs and the shoe heel height to compensate for limb shortening were recorded. Reconstruction failure was defined as the removal of the composite or iliac wing-femoral head coaptation due to complications. Time to failure (months) was defined as the time elapsed between the first surgery and the date of reconstruction removal. A major complication was defined as the one that eventually necessitated removal of the graft or prosthesis as a revision procedure. A minor complication was defined as a problem other than those described above, which necessitated an additional surgical procedure or conservative management. Demographic and treatment variables in the 2 study groups were compared using the t-test and Fisher exact test. Analyses were performed using SPSS ver. 13.0 (SPSS Inc., Chicago, IL, USA), and p-values less than 0.05 were considered significant.

Compared to the 16 cases of PPC reconstruction, the 24 cases of RHA showed lower major and minor complication rates (p < 0.001), shorter surgical time (p < 0.001), and superior MSTS score (p < 0.001) (Table 3). No patients who underwent RHA experienced disruption of the femoral head-iliac wing articulation. However, 11 of the 16 PPCs (69%) were removed at an average of 66 months (range, 1 to 218 months). Causes of PPC failure were graft fracture in 8, infection in 2, and nonunion in 1. The 11 failed PPCs were converted to pseudoarthrosis in 8 and a saddle prosthesis in 3. Minor complications in patients who used PPC included wound infection, plate failure, and dislocation. Overall, 16 patients who used PPC underwent 17 major and 24 minor additional procedures after the index operation (Table 4). The average functional score of the 11 failed PPC patients was 22.4 (74%) until removal of the construct, and their average score deteriorated to 12.4 (41%) after removal. Mean leg length discrepancy of the 11 patients with failed PPC was 5.4 cm (range, 3 to 7 cm). One each RHA case showed flap necrosis and chronic pain. Another 2 patients who had RHA underwent re-exploration for suspicious recurrent lesions; however, these proved to be new bones in neo-acetabulum formation.

Of the 24 patients who underwent RHA, circular bony neo-acetabulum on CT was identified in 7 and partial neo-acetabulum in 9; the remaining 8 patients did not show a bony acetabular structure (Table 5). Average time to bony neo-acetabulum formation was 7 months (range, 4 to 13 months) (Fig. 2). Excluding 2 patients who had a hip spica cast postoperatively, all 13 patients with an osteotomy > 9 cm from the iliac crest showed partial or complete bony neo-acetabulum formation, while only 3 of 9 patients with osteotomy level < 9 cm demonstrated bony neo-acetabulum (Fig. 3). The average MSTS functional score in 9 patients with < 9 cm of the remaining iliac wing was 21 (70%), while that of 15 patients with > 9 cm of the ilium was 25 of 30 points (83%). Average limb shortening in 24 patients who underwent RHA was 3.7 cm (range, 1.3 to 6.4 cm).

Excision of periacetabular tumors usually leaves a large skeletal defect, and attempts at reconstruction by arthrodesis or pseudoarthrosis often result in considerable limb shortening and poor function.17) In this regard, anatomical reconstruction of the hip and hemipelvis by biological or mechanical means were suggested to provide improved functional outcomes and walking ability.11114272829) However, most reconstructions had high complication and failure rates. Therefore, a strategy of resection alone has been revisited.67232830) In our comparative study of anatomical reconstruction and RHA, we confirmed that RHA is a reliable primary procedure for periacetabular tumors, with low complication rates, good functional results, and short surgical time. Moreover, in patients who underwent RHA, less iliac wing resection and early postoperative mobilization seemed to facilitate early stable bony neo-acetabulum formation.

This study has several limitations. First, there are many confounding factors in relatively small comparative cohort groups. We acknowledge the heterogeneity due to factors such as the amount of bone and soft tissue resection, differences in postoperative management, use of chemotherapy, and the nonrandomized choice of reconstruction type. In addition, because we compared our recent cases of RHA with past PPC reconstruction cases, improvement in surgical skill may have influenced the complication rate. However, between 2 groups, no differences were found in tumor size, pathologic diagnosis, resection type, local recurrence, or metastasis rate. Furthermore, the high proportion of male patients in the RHA group may be related with the superior functional outcome. However, this factor cannot offset the time-related failure pattern in the PPC group.

Our comparative study and previous reports show that RHA is a valuable primary procedure after periacetabular resection, with much reduced complication rates or need for further surgery (Table 6). Chronologically, patients with pelvic resection face 2 major complications: infection and mechanical failure. Infection is a devastating event that may lead to removal of the reconstruction hardware or hindquarter amputation. However, pelvic reconstruction was also reported as an independent contributory factor to infection.28) Lower rates of infection in the resection alone group may be explained by the shorter operative time, no foreign body, and reduced dead space by permitting proximal migration of the femoral head. Ensuing problems are mechanical, and include nonunion or fracture of biologic material and loosening or breakage of the prosthesis.728) These late mechanical complications also necessitate the removal of the construct in a substantial proportion of patients, and the functional results of failed cases after intervention are worse than those of primary RHA. Conceptually, because either RHA or failed reconstruction is a pseudoarthrosis, failed reconstruction is assumed to have a functional score similar to that of resection alone. However, 2 factors are related to superior outcome of RHA. One critical factor is the integrity of the femoral head. Patients with failed reconstruction invariably lose the femoral head, and this leads to an additional shortening of around 5 cm (the usual height of the femoral head), compared to RHA patients. Moreover, this loss of femoral head precludes the development of neo-acetabulum. In this regard, in patients who can accept initial shortening of the affected limb, RHA would be a valuable procedure with long-term durability and low risk of complications.

RHA after periacetabular resection is not new. As early as 1978, one study reported the procedure as a satisfactory substitute for hindquarter amputation in 5 patients with chondrosarcoma.2) Since then, several series have reported the usefulness of RHA after pelvic resection with some variation in technique and concept.182030) However, for better results, there are points to consider with regard to optimal indications, surgical technique, and postoperative care. To create a bony or fibrous “neo hip joint,” the femoral head should not be involved by the tumor and a substantial portion of the iliac wing should be saved. Patients with less iliac wing resection (preferably iliac osteotomy level > 9 cm from the iliac crest) show minimal limb shortening and an increased percentage of bony neo-acetabulum formation. At the time of surgical approach, the origins of hip abductors and hip flexors are detached inferiorly and laterally through osteotomies made along the iliac crest and anterior iliac spine. This approach seems to facilitate repair after resection and functional recovery. To maintain the iliac wing-femoral head contact and to minimize the external rotation of the femoral head, a single wire was tied between the femoral head and ilium in half of our patients. However, as the case number increased, we found that this wire fixation was not necessary. In a previous report, to control the location of the femoral head postoperatively, skin traction with rotation-proof shoes or skeletal traction was applied (ambulation was started 4–6 weeks later).23) In our series, except for the 3 early cases with postoperative hip spica cast, all patients were encouraged to ambulate around 2 weeks after surgery. Early active exercise seemed to promote the formation of a neohip joint. In active young patients, walking without aid and neo-hip joint formation was observed around 6 months postoperatively; however, at older ages, independent walking took up to 1 year. The average MSTS functional score in patients with neo-acetabulum formation was 25.3, while that of patients without neo-acetabulum was 20.1.

In conclusion, our comparative study confirmed that RHA for periacetabular tumors can be an excellent alternative to anatomical reconstruction. RHA offers a short surgical time, low complication rates, and functional results comparable to those of other reconstruction methods. However, this procedure is indicated for patients who can accept some shortening of the limb, and the tumor should be confined to the periacetabular area.