Does Adding Femoral Lengthening at the Time of Rotation Hip Transposition After Periacetabular Tumor Resection Allow for Restoration of Limb Length and Function? Interim Results of a Modified Hip Transposition Procedure

Abstract

Background

Reconstruction after pelvic tumor resection of the acetabulum is challenging. Previous methods of hip transposition after acetabular resection have the advantages of reducing wound complications and infections of the allograft or metal endoprosthesis but were associated with substantial limb length discrepancy. We therefore developed a modification of this procedure, rotation hip transposition after femur lengthening, to address limb length, and we wished to evaluate its effectiveness in terms of complications and functional outcomes.

Questions/purposes

In this study, we asked: (1) What were the Musculoskeletal Tumor Society scores after this reconstruction method was used? (2) What complications occurred after this reconstruction method was used? (3) What proportion of patients achieved solid arthrodesis (as opposed to pseudarthrosis) with the sacrum and solid union of the femur? (4) What were the results with respect to limb length after a minimum follow-up of 2 years?

Methods

From 2011 to 2017, 83 patients with an aggressive benign or primary malignant tumor involving the acetabulum were treated in our institution. Of those, 23% (19 of 83) were treated with rotation hip transposition after femur lengthening and were considered for this retrospective study; 15 were available at a minimum follow-up of 2 years (median [range], 49 months [24 to 97 months]), and four died of lung metastases before 2 years. No patients were lost to follow-up before 2 years. During the period in question, the general indications for this approach were primary nonmetastatic malignant bone tumor or a locally aggressive benign bone tumor that could not be treated adequately with curettage. There were seven men and 12 women with a median age of 43 years. Nine patients underwent Zones I + II resection, eight patients had Zones I + II + III resection, and two received Zones II + III resection. After tumor resection, rotation hip transposition after femur lengthening reconstruction was performed, which included two steps. The first step was to lengthen the femur with the insertion of an allograft. Two methods were used to achieve limb lengthening: a “Z” osteotomy and a transverse osteotomy. The second step was to take the hip transposition and rotate the femoral head posteriorly 10° to 20°. The median (range) operative time was 510 minutes (330 to 925 minutes). The median intraoperative blood loss was 4000 mL (1800 to 7000 mL). We performed a chart review on the 15 available patients for clinical and radiographic assessment of functional outcomes and complications. Arthrodesis and leg length discrepancy were evaluated radiographically.

Results

The median (range) Musculoskeletal Tumor Society score was 21 points (17 to 30). Eleven of 19 patients developed procedure-related complications, including six patients with allograft nonunion, two with deep infection, two with delayed skin healing, and one with a hematoma. Two patients had minor additional surgical interventions without the removal of any implants. Local recurrences developed in four patients, and all four died of disease. All seven patients treated with a Z osteotomy had bone union. Among the eight patients with transverse osteotomy, bone union did not occur in six patients. After hip transposition, stable iliofemoral arthrodesis was achieved in seven patients. Pseudarthrosis developed in the remaining eight patients. The median (range) lower limb length discrepancy at the last follow-up visit or death was 8 mm (1 to 42 mm).

Conclusion

Although complex and challenging, rotation hip transposition after femur lengthening reconstruction with a Z osteotomy provides acceptable functional outcomes with complications that are within expectations for resection of pelvic tumors involving the acetabulum. Because of the magnitude and complexity of this technique, we believe it should be used primarily for patients with a favorable prognosis, both locally and systemically. This innovative procedure may be useful to other surgeons if larger numbers of patients and longer-term follow-up confirm our results.

Level of Evidence

Level IV, therapeutic study.

Introduction

Surgical treatment of tumors of the pelvis is challenging. For isolated resections of the ilium (Zone I), ischium, or pubis (Zone III), reconstruction may not achieve good function postoperatively [26]. For periacetabular resections (Zone II), reconstructions are usually used to maintain the femur-pelvis continuity and restore force transmission for reasonable postoperative function. Numerous reconstruction methods have been reported, including autoclaved autografts [31], vascularized fibula [7], allografts [8], saddle prosthesis [2], custom-made prostheses [35], and modular hemipelvic prostheses [17]. However, the reported proportions of complications have been high and range from 40% to 68% depending on the series and reconstruction method [6, 17-20, 27]. Complications may leave surgeons with few reconstructive options, and they often result in amputation [16].

Pseudarthrosis between the femoral head and the remaining ilium after pelvic tumor resection, also known as hip transposition, has been recommended by some authors [14, 30]. The absence of large implants may help decrease the incidence of deep infection and mechanical failures of the prosthesis. Hillmann et al. [18] reported that patients who underwent hip transposition had a lower incidence of complications after resection of the acetabulum than patients who received a prosthesis. Previous methods of hip transposition after acetabular resection had the advantages of reducing wound complications and infections of the allograft or metal endoprosthesis, but they were associated with substantial limb length discrepancy, poor ambulation, and inferior patient acceptance. We therefore developed a modification of this procedure to address limb length and wished to evaluate its effectiveness in terms of functional outcomes, complication rates, and limb length discrepancy. The modified procedure, rotation hip transposition after femur lengthening, was designed to reconstruct large defects after tumor resection in the pelvis as well as restore the limb length. This pelvic reconstruction method might not have the disadvantages associated with a prosthesis such as deep infection and long-term loosening.

In this study, we asked: (1) What were the Musculoskeletal Tumor Society (MSTS) scores after this reconstruction method was used? (2) What complications occurred after this reconstruction method was used? (3) What proportion of patients achieved solid arthrodesis (as opposed to pseudarthrosis) with the sacrum and solid union of the femur? (4) What were the results with respect to limb length after a minimum follow-up of 2 years?

Patients and Methods

Clinical Series

From 2011 to 2017, we saw 83 patients with aggressive benign or primary malignant tumors involving the acetabulum.

In general, for locally aggressive benign tumors in patients with intact bone cortices, curettage was performed. For primary malignant bone tumors in patients in whom an adequate surgical margin was not possible with the use of limb-salvage resection, hemipelvic amputation was used. If part of the acetabular columns could be saved and reconstructed with the femoral head, then total hip endoprosthesis replacement was recommended. Traditional hip transposition was used primarily for patients with high-grade sarcomas in whom it was deemed possible to achieve an adequate surgical margin. Patients were considered for rotation hip transposition after femur lengthening reconstruction if all of the following criteria were met: the pathologic diagnosis was a primary nonmetastatic malignant bone tumor or a locally aggressive benign bone tumor that could not be treated adequately with curettage, the tumor was in the acetabulum, an adequate surgical margin could likely be achieved based on preoperative imaging, and the patient was able to tolerate the procedure. Patients with metastatic carcinomas involving the pelvis or who had primary bone tumors with distant metastases and patients younger than 18 years were excluded. The latter exclusion criterion was because we wanted to avoid the issue of remaining growth in skeletally immature patients.

Based on these indications, 22% (18 of 83) were treated with hemipelvic amputation; 17% (14) were treated with curettage and underwent reconstruction with cementation, allografting, and artificial bone; 11% (9) received hemipelvectomy with traditional hip transposition; 28% (23) were treated with tumor resection and endoprosthetic replacement; and 23% (19) were treated with the procedure we are studying here: rotation hip transposition after femur lengthening. These patients were included in this retrospective study. Fifteen of the included patients had 2 or more years of follow-up, and four had died before 2 years. For the 15 living patients, the median (range) follow-up was 49 months (24 to 97 months). No patients were lost to follow-up before 2 years.

In the study group, there were seven men and 12 women with a median (range) age of 43 years (22 to 51 years) at the time of the procedure. The most common diagnosis was low-grade chondrosarcoma (Table 1). Sixteen of 19 patients were primarily treated in our institution. The remaining three patients were referred to us for the treatment of a local recurrence. This subset included two low-grade chondrosarcomas and one giant cell tumor of bone. The most predominant surgical resection type was Zone I + Zone II (9 of 19 patients), followed by Zone I + Zone II + Zone III (8 of 19 patients), according to the classification of Enneking and Dunham [11].

Table 1.

Patient demographics

Characteristic	Value
Number of patients	19
Age in years, median (range)	43 (22-51)
Gender
Male	7
Female	12
Tumor diagnosis
Low-grade chondrosarcoma	8
Dedifferentiated chondrosarcoma	4
Conventional osteosarcoma	2
Undifferentiated pleomorphic sarcoma	2
Giant cell tumor of bone	1
Low-grade intramedullary osteosarcoma	1
Epithelioid osteoblastoma	1
Surgical resection type
Zone I + II	9
Zone II + III	2
Zone I + II + III	8
Follow-up in months
Median	49
Minimum, maximum	24, 97

Six of eight patients with high-grade sarcomas received neoadjuvant chemotherapy and surgery, followed by adjuvant chemotherapy. The remaining two patients received neoadjuvant chemotherapy and surgery but refused to undergo adjuvant chemotherapy. No patient received adjuvant radiotherapy of the pelvic resection site.

Surgical Technique

All patients underwent preoperative intestinal cleansing and general anesthesia. The operations were performed in a floating position (this means patient position can alternate from either semisupine or semilateral without needing resterilization or redraping). The procedure uses the T-incision surgical approach described by Lackman et al. [23]. It includes two incisions: one is a routine ilioinguinal incision, which begins at the posterior iliac spine, runs along the iliac crest and inguinal ligament, and ends at the pubic symphysis, and the other is a longitudinal incision that begins laterally centered over the palpable proximal femoral shaft and extends superiorly over the greater trochanter and joins with the transverse incision to form a T-shaped incision. The turning point of the T incision is located much more laterally than the universal incision, and the vertical limb runs straight down the lateral thigh rather than turning posteriorly [11]. This approach allows extensive visualization of the inner and outer aspects of the hemipelvis. The tumor was resected with the assistance of a computer navigation system (Stryker OrthoMap Navigation System) in six patients, as reported previously [24, 34], and without navigation assistance in 13 patients. If freehand surgery could be performed easily without compromising the surgical margin, no computer navigation was used. Otherwise, computer navigation was used to precisely resect the tumor to achieve an adequate surgical margin. Because this decision was arbitrary and surgeon-dependent, the impact of navigation use on the surgical margin was not analyzed, nor was it the purpose of this report.

The reconstruction procedure included two steps (Fig. 1A-E). The first step was to lengthen the femur and the second step was to transpose the hip and proximal femur with femoral head rotation. There were two ways of achieving length. One was a “Z” or reverse “Z” osteotomy, and the other was a transverse osteotomy. A Z or reverse Z osteotomy started 3 to 5 cm caudal to the lesser trochanter (Fig. 2A). The length of the osteotomy depended on the size of the tumor resection and the resulting length that needed to be restored. The gaps resulting from the osteotomy were filled with cortical allograft strips (Fig. 2B). The Z or reverse Z osteotomy was designed to increase contact between the allograft and proximal femur, thereby increasing the likelihood of bone healing. However, because of the overall long surgical time, in some patients, a transverse osteotomy was used to reduce the surgical time and potential intraoperative complications. If Z osteotomy is not possible, autograft might be useful to graft the gaps of the transverse osteotomies. We did not use supplemental autografts because of concern of the length of the operation, but it may be worth considering depending on the general status of the patient during the procedure. Before and during the operation, anesthesiologists evaluate the patients and determine whether they can tolerate a longer procedure. If the anesthesiologists agree, we believe these grafts may be useful. After Z or reverse Z or transverse osteotomy, two parts of the femur were separated into a distance to lengthen the femur. The distance was determined by the vertical distance between the upper edge of the acetabulum and the upper third of the sacroiliac joint. The femur and allograft were firmly fixed with an intramedullary rod, plate, screws, and wires. To avoid coronal transposition of the lower limb, the proximal part of the femur was posteriorly rotated 10° to 20° to be fixed to the sacroiliac bone. The lengthening and rotation created some overlaps and gaps between the two femur fragments. The gaps were filled with allograft to facilitate healing. We attempted to achieve arthrodesis between the femoral head and remaining ilium or sacrum by removing articular cartilage from the proximal femur and fixing it to the sacroiliac bone with wires. In addition, patients were treated with an immobilizing brace for 3 months postoperatively.

Fig. 1.

A-E (A) A reverse Z osteotomy of the femur is shown. It started from 3 to 5 cm down the lesser trochanter. (B) The proximal femur was moved up after the osteotomy. (C) The femoral head should be posteriorly rotated 10° to 20°. The gaps resulting from the osteotomy were filled with cortical allograft strips. (D) The femur and allograft were fixed with an intramedullary rod, plate, screws, and wires. The cartilage of the femoral head was removed and fixed to the sacroiliac bone with one or two wires. (E) After transverse osteotomy, the femur and segment allograft were fixed together with internal fixation.

Fig. 2.

A-B (A) An intraoperative image of the Z osteotomy is shown. (B) The proximal femur was moved up, and the gaps were filled with cortical allograft strips.

Other Surgical Details and Oncologic Outcomes

The median (range) operative time was 510 minutes (330 to 925 minutes). The median intraoperative blood loss was 4000 mL (1800 to 7000 mL). According to Enneking’s definition of surgical margins [9], wide surgical margins were achieved in 15 of 19 patients. One patient had a contaminated wide margin, meaning that it was recognized intraoperatively and the intralesional margin was re-resected to achieve wide margins. Marginal margins were achieved in the remaining four patients.

Local recurrence developed in four patients (two with wide margins, one with a contaminated margin, and one with a marginal margin); three of these patients died before the minimum follow-up of this series (2 years) and one died 29 months postoperatively. These patients were included for purposes of assessing surgical complications but not leg length differences, radiographic endpoints, or MSTS scores. Among these four patients, the primary diagnoses were dedifferentiated chondrosarcoma in three and undifferentiated pleomorphic sarcoma in one. The intervals between resection and recurrence were 9, 11, 12, and 18 months, respectively. There was no local recurrence in patients with low-grade sarcomas or locally aggressive benign tumors. Three of four patients who had a local recurrence had a diagnosis of lung metastasis simultaneously and died of disease within 1 year. One patient developed a lung metastasis 3 months after local recurrence, and this patient died after another 8 months. The remaining 15 patients were alive at the last follow-up evaluation.

Rehabilitation and Follow-up Protocols

Patients were encouraged to perform lower-limb muscle exercise starting the day after the procedure. Patients were asked to wear a hip brace and remain in bed until 3 months postoperatively. After 3 months, patients were allowed to begin weightbearing walking with 10 to 20 kg on the surgical side. The weightbearing gradually transitioned to full weightbearing approximately 5 to 6 months postoperatively.

Patients were scheduled to have follow-up visits every 3 months during the first year, every 6 months for 2 years, and then yearly. At each follow-up visit, patients received a physical examination and a radiographic evaluation.

Functional and Complication Evaluation

To answer the first question on functional outcome, we performed a chart review on the 15 available patients for assessment of MSTS scores [10] at the final follow-up visit. We recorded the subsets of the MSTS score, including pain, function, emotional acceptance, support, walking, and gait. To answer the other three questions, we recorded complications related to the reconstruction procedure, including skin problems and deep infections. In a radiographic review, two authors (HX, YL), evaluated allograft healing, iliofemoral arthrodesis, and lower limb-length discrepancy. Successful allograft healing and arthrodesis was defined as defect bridging on AP view radiographs with new bone formation with uniform radiodensity. Allograft nonunion and pseudarthrosis was defined as no new bone formation or uniform radiodensity in the bridging of the osteosynthesis sites or between the femoral head and sacrum. The lower limb length discrepancy was measured by full-length standing radiographs from the pelvis to the ankles of both lower limbs in all 19 patients.

Ethical Approval

Ethical approval for this study was obtained from Beijing Ji Shui Tan Hospital.

Statistical Analysis

For quantitative variables, we reported the median and range. Categorical variables are reported as counts. We performed a nonparametric test to evaluate the difference between the comparisons. We performed an independent- sample t-test to compare the limb length discrepancies between two timepoints. We performed all statistical analyses using a commercially available software package (IBM SPSS Statistics for Mac, version 25, IBM Corp). A p value of < 0.05 was considered significant in all tests.

Results

Functional Outcomes

The median (range) MSTS score of the 15 surviving patients was 21 points (17 to 30), which was 70% of the maximal possible score. Thirteen patients were pain free, and the remaining two patients had modest pain when walking. The median functional activity score was 60% (3 of 5). The median emotional acceptance score was 100% (5 of 5). The median support score was 60% (3 of 5). Six of 15 patients used a cane or crutch for walking and one patient used two crutches. The remaining eight patients could walk without support at the latest follow-up examination. All 15 patients could walk outside, but 14 patients had less walking ability than they did preoperatively. Only one patient had no alteration in gait, and the median gait score was 60% (3 of 5).

Complications

The most common complications were related to infection, wound healing, and allograft nonunion. Surgery-related complications were seen in 11 of 19 patients. Additional surgical interventions were performed in two patients. Two patients had deep infections and were treated successfully with debridement, irrigation, and systemic antibiotics without the removal of any implants. Two patients with delayed wound healing and one with a hematoma were treated nonoperatively; none of these were related the superior corner of the inferior flap. Four of five patients with wound complications received chemotherapy, but with the numbers we had, we could not correlate wound complications with chemotherapy use. The 15 surviving patients were evaluated radiographically for allograft healing. Seven patients had either a Z or reverse Z osteotomy and eight had a transverse osteotomy. In each patient with the Z-type osteotomy, there was one anastomotic junction. However, there were two junctions in each patient with a transverse osteotomy. There were seven anastomotic junctions in seven patients with a Z or reverse Z osteotomy, and all of these junctions united radiographically by 15 months postoperatively (Fig. 3). For patients with a transverse osteotomy, 8 of 16 anastomotic junctions failed to achieve bone union by 24 months postoperatively (Fig. 4). Six patients had allograft nonunion. No patients with nonunion received chemotherapy, but all had transverse osteotomy. We did attempt to stimulate union by using a bone graft at the nonunion sites. In addition, since all patients were asymptomatic, they declined further operations to achieve union.

Fig. 3.

A-C A 22-year-old woman with a recurrent giant cell tumor of bone in the acetabulum was treated with our technique. (A) A radiograph shows that the acetabulum, part of the ilium, and the pubis were involved. (B) After tumor resection, the femur was lengthened until the femoral head reached the remaining ilium and sacrum. The gaps were filled with cortical allograft strips after a Z osteotomy. (C) After 97 months, the allograft showed good healing with host bone, and iliofemoral arthrodesis was seen.

Fig. 4.

A-B A 21-year-old woman with pelvic osteosarcoma is shown. (A) A radiograph taken after neoadjuvant chemotherapy shows that the left ilium and acetabulum were involved. (B) The patient received a transverse osteotomy and segment allograft. A radiograph at 24 months of follow-up showed no bone union at both anastomotic junctions. There was an iliofemoral arthrodesis.

Iliofemoral Arthrodesis and Pseudarthrosis

After hip transposition, stable iliofemoral arthrodesis was achieved in seven of 15 surviving patients. Pseudarthrosis developed in the remaining eight patients. With the numbers we had, we could not detect a difference in the median MSTS score between patients in whom arthrodesis was achieved and those in whom it was not. The median MSTS scores of patients who underwent iliofemoral fusion and those with pseudarthrosis were 22 points and 21 points, respectively (p = 0.27). Only 1 of 7 patients with pseudarthrosis complained of mild-to-moderate walking pain.

Lower Limb Length Discrepancy

All 19 patients were examined for lower limb discrepancy, which was evaluated with full-length standing radiographs from the pelvis to the ankles of both limbs, immediately after the procedure and at the latest follow-up visit (Fig. 5). The median (range) lower limb discrepancy immediately after resection was 11 mm (2 to 38 mm). The median discrepancy at the latest follow-up examination or death was 8 mm (1 to 42 mm). With the numbers we had, we could not show a difference between the two timepoints (independent-sample t-test: t = 0.605; p = 0.55).

Fig. 5.

A-B Radiographs of the total lower limb length of the same patient as in Fig. 4 are shown. (A) There was a lower limb length discrepancy of 7 mm immediately postoperatively. (B) The lower limb length discrepancy was 14 mm at 97 months of follow-up.

Discussion

Advances in medical oncology, surgical techniques, and reconstruction options have enabled limb-sparing surgery with reconstruction to become the preferred choice of treatment over external or internal hemipelvectomy [5, 20, 33]. Although there are many reconstruction methods for pelvic defects after tumor resection, the ideal option is not currently known. It is generally accepted that complications have been common regardless of the chosen reconstruction method. We developed a novel reconstruction method for pelvic tumors that involves the acetabulum, and our early results show reasonable function based on an MSTS functional assessment and complications that are comparable with those of other resection techniques [4, 8, 29]. In addition, we restored limb length in most patients.

Limitations

We acknowledge that this research has several limitations. First, this was a retrospective analysis of patients who had received this reconstruction technique. A prospective study with a comparison cohort would have been preferable to demonstrate how this technique compares with other reconstruction options, but because these are rare procedures and patients are highly selected for the approach, this type of study is difficult to perform in a single institution. Nevertheless, we think our findings can provide initial evidence that this reconstruction method has some merit and bears further study. In addition, the typical limitations of retrospective research were not so severe here; for example, our indications were relatively consistent over the study period, and no patients were lost to follow-up before 2 years. We acknowledge that our findings would be improved if we could compare our patients with those who had a standard hip transposition without lengthening or other types of reconstruction, but in our center, we do not have sufficient numbers of patients to do so. Additionally, because of the way in which we selected patients for this study, these patients would likely not be directly comparable. Further, our series of 19 patients was small. More patients will be needed to verify the types and proportions of complications, functional outcomes, and correlation between surgical margins and local recurrences, but we feel this preliminary presentation of this technique will encourage other surgeons to attempt it in appropriate patients. Finally, although the median follow-up of 49 months was longer than that of many of other reports, it was insufficient to address the long-term complications and functional outcomes associated with this technique. Longer follow-up will be needed to evaluate the effects of this technique on the lower spine and ipsilateral knee.

Functional Outcomes

The median MSTS score of our patients was 21 points, which is comparable to the scores of other studies of limb-salvage methods [4, 8, 29]. There are several pelvic reconstruction methods, each of which has its own indications. The reported mean MSTS score of one study that examined the use of a saddle prosthesis, titanium pedestal cup, extracorporeally irradiated autograft, hip transpositions with mesh pouch, and other reconstructions was 65%, ranging from 53% to 87%, mostly depending on the resection type [33]. Gebert et al. [15] reported on 62 patients treated with hip transposition. The main problem for these patients was limb length discrepancy. The mean MSTS score was 62%. To eliminate the limb length discrepancy, Rödl et al. [30] used a unilateral external fixator to lengthen the femur in four patients at a mean of 5.7 years after pelvic surgery. Two patients had lengthening procedure-related complications. The MSTS score was 73%, better than the score before the lengthening procedure. Our MSTS score was similar. In the present study, the limb was lengthened with an allograft after tumor resection. However, our patients lost the ability to internally or externally rotate their hip, which compromised the overall functional outcomes to some extent. Additionally, eight patients received Types I + II + III resections. Although we tried to correlate the outcomes and the zonal resections, no connections could be found between them. More patients with longer follow-up are needed to see if there is an association by type of resection. We believe that rotation hip transposition after femur lengthening could be a reasonable reconstruction option for tumors involving the acetabulum, even if the desired resection margins must be achieved in the entire hemipelvis.

Complications

With a median follow-up of 49 months, 7 of 15 patients experienced major complications including infection and nonunion, but only two patients were treated with minor additional surgical interventions without the removal of reconstruction implants. Six of 8 patients with transverse osteotomy had nonunion of allograft, and no patient with Z or reverse Z had nonunion. Insufficient allograft–host bone contact may contribute to the nonunion of patients with transverse osteotomy. Although we worried about the hardware breakage, no patients were interested in undergoing a second procedure to treat the nonunion since it was asymptomatic. One of the concerns of this technique was the potential risk of bone necrosis in the proximal femoral fragment, which could affect the union rate. To decrease this risk, the soft tissue around the bone was retained as much as possible intraoperatively. However, normal uptake was observed on bone scans at follow-up visits (used to monitor for bone metastases), indicating the bone was alive postoperatively. To reduce the wound complications, we now recommend that vacuum-assisted closure drains be applied immediately after the surgery in the future, although we have no evidence to show that they would reduce wound complications in our patients. Most published studies report a high incidence of complications; for example, Wilson et al. [33] reviewed 22 studies with 801 patients who had pelvic tumors, revealing a mean nononcologic complication proportion of 49% and nononcologic reoperation proportion of 37%. Four of our patients experienced local recurrences, and all of them developed metastases and died. Local recurrence and surgery-related complications are the two major concerns that affect the outcome of pelvic reconstruction after tumor resection. Local recurrence rates after pelvic tumor resection have ranged from 22.1% to 43.7% [1, 3, 4, 12, 17, 21, 25, 28, 32]. It is extremely difficult to achieve a clear margin when treating recurrent pelvic tumors, even with amputation and removal of all implants. In many patients with local recurrence, a lung metastasis develops at the same time or soon after. Thus, pelvic tumor resection is sometimes called a one-shot surgery, indicating that the patient does not have a good likelihood of a second resection opportunity. Because patients with a high likelihood of local recurrence would be unlikely to benefit from this reconstruction method and because this procedure is complex, we believe it is best reserved for patients with an anticipated local recurrence of less than 10%. In the current series, 3 of 4 patients who had a local recurrence had dedifferentiated chondrosarcoma. We now believe that this procedure should not be considered for patients with dedifferentiated chondrosarcoma.

Iliofemoral Arthrodesis and Pseudarthrosis

In our study, 7 of 15 patients achieved stable iliofemoral arthrodesis. According to a study with long-term follow-up data, patients with primary solid fusion did better functionally than patients with pseudarthrosis [13]. In the current series, two measures were taken to obtain arthrodesis, including intraoperative wires and a postoperative hip brace. In addition, the femoral head was rotated to have wider contact with the sacroiliac region to achieve a sound arthrodesis. However, it was still difficult to achieve this goal. Interestingly, in a recent report, pseudarthrosis provided good functional results [22]. In our opinion, the use of an implant such as the MUTARS (Implantcast) attachment tube is unnecessary to achieve hip arthrodesis or pseudarthrosis as in a previous report [14]. This avoids potential complications associated with a unilateral external fixator and MUTARS implant.

Lower Limb Length Discrepancy

We achieved reasonable limb lengths in most patients. The median discrepancy at the latest follow-up examination or death was 8 mm, which we believed was acceptable for our patients. Traditional hip transposition has the advantage of being a simple surgery with shorter surgical time. However, if the patients survive, lower limb length discrepancy becomes an obvious problem. In a report with four patients [30], the lower limb length discrepancy was 6, 11, 12, and 12 cm after hip transposition. Three patients were corrected by distraction osteogenesis. The postoperative lower limb length discrepancy was 0, 3.5, 6, and 6 cm.

Conclusion

In summary, we believe that rotation hip transposition after femur lengthening reconstruction is a reasonable option for patients with tumors involving the acetabulum as long as negative margins can be achieved and local recurrence can be minimized. We believe the procedure may not be appropriate for patients with dedifferentiated chondrosarcoma because of our experience with local recurrence in those patients. Because of the challenges of this procedure (long surgical time, substantial blood loss, and a high degree of complexity), the candidate patients for this technique should be those with nonmetastatic primary resectable tumors with a reasonable expectation of survival. During the femur lengthening process, we recommend a Z osteotomy, if possible, rather than a transverse one because the former resulted in better allograft healing.