Is a Modular Pedicle-hemipelvic Endoprosthesis Durable at Short Term in Patients Undergoing Enneking Type I + II Tumor Resections With or Without Sacroiliac Involvement?

Abstract

Background

Functional reconstruction after Enneking Type I + II resections of the pelvis (those involving both the ilium and the acetabulum) is challenging, especially if resection of part of the sacrum is included. To assess the clinical outcomes of a newly designed modular pedicle-hemipelvic endoprosthesis, we performed a preliminary retrospective study on its clinical use in a small group of patients.

Questions/purposes

The purposes of this study were (1) to evaluate in a small case series whether the new endoprosthesis restored lower limb function and lumbopelvic stability in the short term; (2) to identify the complications associated with use of the new prosthesis; and (3) to assess the 5-year cumulative survival, the cumulative incidence of a major postoperative event, and the cumulative incidence of implant failure in this group of patients.

Methods

Between August 2012 and August 2014, our center performed 274 internal hemipelvectomies for oncologic indications. Among these, 20 were treated with the new endoprosthesis, which was designed for fixation both to the residual sacrum as well as the lumbar spine. An earlier version of the device had been removed from the market because of an unacceptable risk of serious complications. All of the 20 tumors were sarcomas necessitating en bloc resection. The implant is modular and can meet the different-sized defects in each patient. The general indication for use of the new implant was a total acetabular defect with extensive iliac involvement or total loss of the sacroiliac joint and/or hemisacrum. All 20 patients were followed up for a minimum of 24 months or until death in those patients who survived < 2 years (median, 36 months; range, 6-60 months). The clinical data were retrieved from the database and the study endpoints (function according to the Musculoskeletal Tumor Society [MSTS] score, complications, and survivorship of patients and implants) were ascertained by chart review. Lumbopelvic stability was defined as an excellent or good rating according to the International Society of Limb Salvage radiologic implant evaluation system. The cumulative survival of patients was estimated using the Kaplan-Meier approach. The cumulative incidence of major postoperative events including local recurrence, metastasis, and reoperation was estimated using a competing events analysis; the cumulative incidence of implant failure, including mechanical failure or deep infection, in patients who underwent reoperation was also estimated using a competing events analysis.

Results

In the 16 patients who survived > 12 months, the median MSTS score was 19 of 30 (range, 5–26). Radiographic assessments demonstrated lumbopelvic stability in all of the 16 patients. Twelve of 20 patients developed postoperative complications, primarily including deep infection (one), hip dislocation (two), and local recurrence (three). Major revision surgery was performed in five of 20 patients. The estimated 5-year Kaplan-Meier patient survival rate was 69% (95% confidence interval [CI], 59%-79%), whereas the cumulative incidence of major postoperative events and implant failure using the competing risk estimator was 42% (95% CI, 23%-60%) and 15% (95% CI, 4%-34%) at 5 years, respectively.

Conclusions

Preliminary results with hemipelvic reconstruction using this new endoprosthesis achieved fair functional results and the complications that were observed appeared comparable to other reconstruction options at short-term followup. Longer-term surveillance is called for to see whether this implant will be durable compared with other available reconstructive alternatives such as a custom-made megaendoprosthesis or an autograft/allograft-prosthetic composite.

Level of Evidence

Level IV, therapeutic study.

Introduction

As a result of their insidious onset, pelvic sarcomas are usually diagnosed late as large masses [7]. Wide resection in conjunction with adjuvant therapies remains the standard treatment for most pelvic sarcomas [25]. Numerous reconstruction methods including saddle prostheses, ice cream cone prostheses, custom-made and modular prostheses, autograft/allograft-prosthetic composites (APCs) as well as flail hip have been used to attempt to restore patients’ ability to stand and walk after resection of periacetabular tumors [1-3, 13, 18, 21, 24]. However, when both the ilium and acetabulum are completely excised (Enneking Type I + II resection), some of these approaches—such as flail hip and saddle or ice cream cone prostheses—may not be feasible because of the massive defect in the pelvis. Custom-made megaendoprostheses and APCs are useful under these circumstances, although complications including infection, aseptic loosening, and local recurrence are major concerns with these reconstructive methods [6, 23, 30].

To solve these problems, we previously combined the pedicle screw–rod system with a modular hemipelvic endoprosthesis (Modular Hemipelvic Prosthesis II; Chunli Co, Beijing, China) and achieved acceptable complication rates and favorable functional outcomes at a minimum followup of 15 months [33]. However, further followup of our patients after we reported these preliminary data [33] identified a number of major complications. These included rod breakage, prosthetic dislocation, and pedicle screw loosening, leading to substantial deterioration of lower limb function. As a result, we devised a new generation of modular pedicle-hemipelvic endoprostheses with the aim of decreasing the complication rate and increasing the durability of the prosthesis. The newer design (Modular Pedicle-hemipelvic Endoprosthesis; Chunli Co) is characterized by its enhanced fixation to the remaining sacrum aside from the simple connection to the lumbar spine like in the older version. We designed the new device in our center to achieve both fixation to the residual sacrum and connection to the lumbar spine. Additionally, the porous structure was three-dimensional (3-D)-printed on the medial surface to facilitate osseointegration and long-term stabilization. A double-axle component was also first introduced so that surgeons were able to adjust the position and angle of the acetabulum more conveniently during installation. Today, it is commercially available nationwide.

In the current study, we evaluated the newer design of this implant and asked: (1) In the setting of a small case series, did the new modular pedicle-hemipelvic endoprosthesis restore lower limb function and lumbopelvic stability in the short term? (2) What were the complications associated with use of the new prosthesis? (3) What was the 5-year Kaplan-Meier patient cumulative survival and what were the cumulative incidence of major postoperative events and implant failure at 5 years calculated by competing events analysis in this group of patients?

Patients and Methods

Between August 2012 and August 2014, our center performed a total of 274 internal hemipelvectomies for oncologic indications. Of those, 95 were treated with hemipelvic endoprosthetic reconstruction. We generally used this method in patients whose acetabulum was totally deficient. In contrast, lumbopelvic fixation was used for reconstruction of major iliac defects if the acetabulum was intact. Of the 95 patients who received hemipelvic endoprosthetic reconstructions, 20 patients were reconstructed with the new pedicle-hemipelvic endoprosthesis. The general indication for use of the new implant was a total acetabular defect with extensive iliac involvement or total loss of the sacroiliac joint, which resulted in a lack of remaining bone for fixation of conventional modular hemipelvic endoprostheses [18]. The current report is a retrospective study of this group of patients performed with approval from the institutional review board. The last generation of pedicle-hemipelvic endoprostheses [33] has been discontinued since July 2012 because of an unacceptable risk of serious complications and was not included in this study.

There were 14 male and six female patients with a median age of 36 years (range, 12-62 years). All 20 patients had primary pelvic sarcomas (Table 1): seven were osteosarcomas, five were chondrosarcomas, two were Ewing sarcomas, two were undifferentiated pleomorphic sarcomas, and four other histologic types. With regard to staging [10], 18 patients were Stage IIB and two patients were Stage III. Resections were primarily planned based on Enneking and Dunham’s classification (Fig. 1) [9]. Four patients underwent Type I + II resections, one had Type I + II + III resection, nine had Type I + II + IV resections, and six had Type I + II + III + IV resections.

Table 1.

Demographics and clinical outcomes of patients

Fig. 1.

This figure shows Enneking and Dunham’s classification of pelvic resections.

The new prosthesis has three components: the sacroiliac holder, the double-axle component, and the acetabular component (Fig. 2A). The L-shaped sacroiliac hold is made of 3-D-printed titanium alloy and is characterized by its minor bottom fitting the narrow space below the sacroiliac joint (Figs. 2A, 3A). Its lateral piece has multiple polyaxial screws for lumbar connection along with several sacral screw holes for sacroiliac fixation (Fig. 2B). To achieve long-term compressive osseointegration between the holder and the remaining bone, the medial surface of the holder is printed with a porous structure. The double-axle component is adjustable in either a sagittal or transverse plane (Fig. 2C). After the acetabular anteversion and abduction angle are established, it can be fixed in place by tightening two screws within the gears. Multiple sizes of double-axle components are available for reconstructing varying sizes of pelvic bones. In the acetabular component, screw holes are available for fixation to the ischiopubic bones.

Fig. 2.

A-C Prosthetic design. (A) AP view: the endoprosthesis consists of three components: the sacroiliac holder, the double-axle component, and the acetabular component. (B) Medial view: gears within the double-axle component are adjustable in either sagittal or transverse planes to gain appropriate acetabular positioning. Screw holes are available in the acetabular component for its fixation to the ischiopubic bones. (C) Lateral view: multiple polyaxial screws were designed for connection to lumbar pedicle screws.

Fig. 3.

A-C Illustration of prosthetic installation. (A) When the osteotomy is through the sacroiliac joint and the remaining bone is sufficient for prosthetic fixation, the sacroiliac holder is directly fixed to the sacrum wing. (B) When the osteotomy is through the ipsilateral sacral wing, the femoral head (red) is trimmed into a cube shape and grafted in the gap between the prosthesis and the residual sacrum. (C) When the osteotomy is through the midline of the sacrum, the implant could be installed slightly medial because the bone graft (red) may not be sufficient to fill the massive gap within the lumbosacral region.

Surgical Technique

Before the surgery, radiographic images of the patients were carefully reviewed to determine the osteotomy sites precisely. If the osteotomy was planned to be through the sacroiliac joint or ipsilateral sacral wing, patients were positioned in lateral decubitus. The incision is started at the midline of the lower back and progresses over the iliac crest; the extended anterior hip approach (Smith-Petersen) was used for hip exposure. If the osteotomy line was planned to be medial to the ipsilateral sacral foramina, patients were first positioned prone to open the sacral canal and the sacral osteotomy was performed posteriorly. Then patients were positioned in lateral decubitus to make a sacral osteotomy anteriorly and for subsequent procedures. Ipsilateral lumbar pedicle screws could be inserted at L2 to L5 either in the lateral decubitus or prone position, whereas contralateral lumbopelvic fixation, if necessary to enhance lumbopelvic stability, could only be made first in the prone position. Normally, we prefer to trim the harvested femoral head into a cube shape and graft it in the gap between the prosthesis and the residual sacrum if part of the sacrum is also excised (Figs. 3B-C, 4A-B). Tumor invasion to the hip cavity or the femoral head must be determined preoperatively to avoid tumor seeding if an intraarticular approach is used. Otherwise, extraarticular resection is used and bulk allograft bone could serve as an alternative instead of the patient’s femoral head to augment the sacral fixation.

Fig. 4.

A-C Patient presentation. (A) A 17-year-old male patient received reconstruction with the new modular pedicle-hemipelvic endoprosthesis after en bloc resection of a sacropelvic tumor. The harvested femoral head was grafted in the gap between the prosthesis and the residual sacrum (blue arrow). The yellow line indicates the posterior midline of the sacrum. (B) One year after surgery, bone fusion was achieved between the femoral head autograft and the residual sacrum (red arrow), and bone resorption was not observed in the interface between the sacroiliac holder and the grafted femoral head (green arrow). (C) This patient underwent external hemipelvectomy for local recurrence. Hard-tissue slicing was performed after external hemipelvectomy as a result of local recurrence, and the micro-CT scan of the bone-prosthesis interface showed osseointegration within the 3-D-printed porous structure.

The positioning of the acetabulum must be carefully planned. When the tumor is excised en bloc, the pelvic specimen is usually placed at its original site in a sterile plastic bag to determine the true acetabular position. Then, surgeons can select the double-axle component of the correct size and assemble it with the other components. Sacral screws are inserted through the screw holes in the sacroiliac holder into the residual sacrum, and the acetabular position and angle are fixed by tightening the gears of the double-axle component. Subsequently, the implant is stabilized by connecting the polyaxial screws in the sacroiliac holder to the preinserted lumbar pedicle screws with multiple rods (Fig. 4A).

For hip reconstruction, the polyethylene acetabular cup is incorporated with the acetabular component with antibiotic-impregnated polymethylmethacrylate (PMMA) cement. The acetabular component is always completely embedded in the bone cement to prevent infection. In patients in whom the femoral head is used as a bone graft, the surgical procedure is largely the same except for a few more steps of graft trimming and positioning (Figs. 3B-C). The gluteus maximus can cover the posterior portion of the prosthesis, whereas the sartorius, anterior femoral rectus, and tensor fascia lata are reattached to the abdominal muscles with acceptable tension. Flaps such as a rectus abdominis myocutaneous flap was not used in this group of patients because the soft tissue tension was normally acceptable after resection of the iliac wing and the blood supply was usually not severely impaired as a result of abundant vascular anastomosis surrounding the pelvis. To restore hip abduction force, the gluteus medius is tightly stitched with the lateral portion of the sacroiliac holder.

Postoperatively, we maintained the patients in a slight hip abduction position to facilitate early healing of the abductor muscles and prevent hip dislocation. Anticoagulants were given to patients after 24 hours or when daily drainage reduced to 250 mL and usually stopped within 4 weeks after the surgery. Third-generation cephalosporins or penicillins supplemented with a β-lactamase inhibitor (eg, 3 g sulperazone or 4.5 g Tazocin) were systematically administered to patients every 12 hours starting from 4 hours after surgery until the removal of wound drainage tubes, normally within 14 days, with additional oral prophylaxis for 7 days. Ankle pumps and quadriceps sets are encouraged right after surgery. Patients are allowed to sit on a high chair and stand with a walking aid after 4 weeks. Full weightbearing is not encouraged until 4 months after surgery.

Complications were divided into five typical groups as described previously [34]: Type A = local wound-related complications; Type B = implant failure; Type C = systemic complications; Type D = other complications; and Type E = local tumor recurrence. Nerve lesions are regarded as complications only when directly related to the use of instrumentation or a prosthesis.

The patients were followed up every 3 months for the first 3 years and every 6 months afterward, primarily through outpatient visits. At each followup, the oncologic status, Musculoskeletal Tumor Society (MSTS) functional scores [8], and complications were all evaluated by senior surgeons and documented in the records by our study coordinator (YS). The radiographic images at each followup were also collected and archived. All 20 patients were effectively followed for a minimum of 24 months or until death in patients who survived < 2 years (median, 36 months; range, 6-60 months).

To answer the research questions (function according to the MSTS score, complications, and survivorship of patients and implants), we retrieved all the clinical data from the database and performed a chart review. Surgical margins were determined using the operative record and histologic report. Limb discrepancies were measured from the postoperative full-length lower limb radiographs. Lumbopelvic stability was assessed according to the International Society of Limb Salvage (ISOLS) radiologic implant evaluation system [20]. Bone remodeling was evaluated using AP radiographs. No change from the discharge radiograph was defined as an excellent rating. Osteopenia or bone resorption was defined as a good or fair rating. Other parameters including interface, anchorage, and implant problems were also examined accordingly.

Statistical Analysis

Survivorship of patients was estimated using the Kaplan-Meier approach using the SPSS 21 software package (IBM, Armonk, NY, USA). The cumulative incidence of major postoperative events including local recurrence, metastasis, and reoperation was estimated using a competing events analysis, because death was a competing risk against those events. Similarly, the cumulative incidence of implant failure including mechanical failure or deep infection that needed reoperation was also estimated using competing events analysis. R package 3.3.2 (R Development Core Team) was used for competing events analyses.

Results

Function and Lumbopelvic Stability

In the 16 patients who survived > 12 months, the median MSTS score was 19 (range, 5–26 of 30) of the maximal possible score. Four were excellent (23-30), seven were good (16-22), three were fair (8-15), and two were poor (0-7). Thirteen of the 16 patients were able to stand after rehabilitation, whereas four, six, and three of the 13 patients were able to walk without support, walk with one crutch, and walk with two crutches, respectively. Of the two patients whose lower limb function was poor, both had chondrosarcoma and received early external hemipelvectomy as a result of local recurrence.

Radiographic assessments were performed for the 16 patients who were still alive 12 months after surgery. At the last time of followup (Table 1), all of the patients were rated as excellent with regard to anchorage, implant body, and articulation. In four patients who did not have resection of the sacrum, bone remodeling was excellent in three patients and good in three patients. In 12 patients who had resection of the sacrum, no evidence of resorption was detected between the residual sacrum and the grafted femoral head yet, although minor bone absorption was found in six of 12 graft-implant interfaces, whereas the other six graft-implant interfaces were radiographically osseointegrated (Fig. 4B). Microscopic bone ingrowth was proved using micro-CT scanning in one patient who underwent removal of the implant as a result of local recurrence and external hemipelvectomy (Fig. 4C). Neither prosthetic dislocation nor bone fracture was seen in any of the patients. No signs of screw loosening or rod breakage were discovered.

Complications

Twelve of 20 patients experienced postoperative complications. Major revision surgery was performed in a total of five patients.

Type A Complications

Wound-related complications were the most common complication type we observed. Delayed wound healing occurred in five patients, four with superficial necrosis and one with a hematoma; all were treated by local débridement under local anesthesia. Deep infection occurred in one patient whose implant was removed.

Type B Complications

Hip dislocation occurred in two patients; both were treated with open reduction and this did not recur after hip stabilization using abduction orthoses for 3 months. We have followup on one patient who is still alive 42 months since the dislocation and on the other patient who did not survive 8 months since the dislocation. Aseptic implant failure, screw breakage, or implant loosening has not been observed in any patient as yet.

Type C Complications

Deep vein thrombosis occurred in two patients; both were treated with anticoagulation therapy. No symptomatic pulmonary embolism or cardiovascular failure occurred in any patients.

Type D Complications

The limb discrepancy of the involved limb was 5 mm (median; range, 0-12 mm) shorter compared with the contralateral limb as assessed on radiographs. Among the 12 patients whose tumor had sacral invasion, three had invasion to the sacral foramina and ipsilateral S1-S5 nerve roots were sacrificed after posterior laminectomy and sagittal osteotomy of the sacrum. All of the three patients experienced minor sciatic nerve palsy as expected preoperatively. Among those three patients, we found that the impact of the surgery to bowel and bladder function was minor. All of them had recovered to largely normal function as before the surgery, although they had constant hypoesthesia of the perineal area. No nerve damage was observed as directly related to the prosthesis or pedicle screws.

Type E Complications

Among the 12 patients in whom wide resection was not achieved, two had intralesional margins resulting from unexpected rupture of the tumor during excision and three had marginal but contaminated margins according to the histologic reports, whereas adequate margins without contamination were achieved in the other seven patients. Local recurrence was confirmed in a total of three patients: two who had inadequate margins (one intralesional, one marginal but contaminated), whereas wide resection was obtained in the other patient. These were treated in two patients with external hemipelvectomy, whereas the other patient with recurrent osteosarcoma was treated with radiotherapy.

Survival of Patients/Incidence of Major Postoperative Events and Implant Failure

The estimated 5-year Kaplan-Meier patient cumulative survival rate was 69% (95% confidence interval [CI], 59%-79%), whereas the cumulative incidence of a major postoperative event and implant failure using the competing risk estimator was 42% (95% CI, 23%-60%) and 15% (95% CI, 4%-34%) at 5 years, respectively (Fig. 5A-B). Reoperation resulting from implant failure was performed in three patients, of whom two underwent open reductions of dislocated hips and the other one had implant removed because of deep infection. Other major postoperative events included local recurrence in three patients and distal metastasis in seven patients. Lung metastasis developed at a median of 8 months (range, 3–18 months) in seven patients, of whom six died of disease and one is still alive after resection of metastatic lung nodules.

Fig. 5.

A-B The cumulative incidence of a major postoperative event and implant failure. (A) The cumulative incidence of a major postoperative event with 95% CI was estimated using a competing risks method. (B) The cumulative incidence of implant failure with 95% CI estimated using the competing risks method.

Discussion

When the acetabulum and the ilium are completely resected, the major defect leads to severe limb length discrepancy and loss of lumbopelvic stability if no reconstruction is performed [28]. Despite that, some studies have shown reasonable function in patients with no reconstruction [12, 17] and no studies have definitively shown superior function with the existing reconstruction options [3, 18, 21, 33]. The available reconstructive approaches have not been shown to offer favorable function in the long term [1, 23] and are associated with major complications such as implant failure and deep infection [6, 24]. The pedicle-hemipelvic endoprosthesis evaluated in the current report was designed to improve patients’ lower limb function by improving durability with its enhanced fixation to the residual sacrum and potential bone-prosthetic fusion with the help of a 3-D-printed porous surface, whereas the risk of infection was expected to be reduced with the compact design and minimized metal exposure to the surrounding soft tissue. In our study, we demonstrated that in the short term, patients could achieve acceptable lower limb function with use of the new prosthesis. Additionally, the complications, the rate of major postoperative events and implant failure, appear to be reasonable and similar to findings of others [3, 7, 18].Further evaluation and comparison with other reconstructive approaches are strongly needed.

This study had a number of limitations. First, this is a retrospective case series without a control group. That being said, we cannot directly compare our results with alternatives. Also, studies of this design typically suffer from selection bias, particularly when evaluating a new device. Certainly we endeavored to choose our patients well, but we note that all of these patients had severe pelvic bone loss; we report here only on patients with a minimum defect that combined Enneking Type I + II (iliac and complete acetabular bone loss). Another limitation of retrospective case series is assessment bias. The treating team in this study also entered the patients’ data into the medical records, which was the source of the data for the current report. However, we supplemented provider-reported outcomes (like the MSTS score) with more objective endpoints like complications and revision surgery. The types of resections and reconstructions were also heterogeneous. Although all patients had Type I and II resections, the involvement of sacral and sacroiliac involvement varied so we could not fully assess the success of this prosthesis based on the extent of the sacral resection. In addition, a minimum of 2-year followup is insufficient to validate the durability of this prosthesis, and we continue to follow these patients. These devices are subject to high in vivo loads, and it therefore seems likely that in time, more of them will develop loosening because some of these patients are debilitated from disease; infections may also occur later. Revising these devices would be difficult in all patients and perhaps impossible in some; removing these devices might leave a patient more disabled than if he or she had never undergone reconstruction. However, because many of these patients have limited life expectancies, we consider these risks worth taking in well-selected and fully informed patients, because this reconstruction offers the opportunity to achieve a higher level of function than might otherwise be achieved. Last, the ISOLS radiologic implant evaluation system used to determine prosthetic stability has not been validated for pelvic reconstructions, although to our knowledge, it was the most appropriate system for our purposes.

We used our newly designed endoprosthesis and achieved favorable functional outcomes (MSTS score 65%) in a small group of patients with short followup. This appears to be better in comparison to the previous generation of pedicle-hemipelvic endoprostheses, which yielded an MSTS score of 58% [33], although we cannot directly compare the two groups of patients to show a statistical or clinically important difference. The new prosthesis offers enhanced modularity and more precise positioning of the acetabular cup compared with the previous version. Furthermore, the immediate lumbopelvic stability provided by the new system has allowed patients to start rehabilitation earlier with more active exercises. Multiple approaches have been previously implemented for this difficult reconstruction (Table 2). Custom-made endoprostheses have been designed to accommodate the unique defect of each individual patient. Ozaki et al. [23] reported a series of eight patients undergoing custom-made endoprosthetic reconstruction after hemipelvic resections, which yielded a poor average MSTS score of 39%. In another small group of nine patients, Müller et al. [22] reported that only one patient had achieved an excellent functional outcome after a mean followup of 62 months. We note also that if the osteotomy site differs from what was planned before surgery, a custom-made prosthesis cannot perfectly match the remaining bone, which may lead to inappropriate installation and subsequent mechanical failure of the prosthesis. Recently, the use of intraoperative navigation has improved the precision of osteotomy and facilitated the installation of custom-made endoprostheses [5, 32]. APCs have also been used for reconstruction of this type of massive defect (Table 2). They have been reported to achieve favorable postoperative function with an average MSTS of 70% to 77% [3, 4, 6, 30]. Nevertheless, a well-maintained bone bank is necessary, and adherence to guidelines is required to avoid disease transmission. The reported incidence of infections and fractures may also adversely affect function in the long term [23]. Other methods such as the new saddle prosthesis and the modular hemipelvic endoprosthesis [2, 15] are more suitable for reconstruction of simple acetabular defects with minor iliac destruction, because they rely on fixation to the remaining iliac bone.

Table 2.

Comparison of clinical outcomes with other studies

Our complication rate appeared to be favorable to other types of reconstructions, although 12 of our 20 patients experienced complications. Of the major complications, one patient had a deep infection, two had hip dislocations, and there were three local recurrences. Deep infection, implant failure, and local recurrence, all of which result in major second operations and tremendous deterioration of patients’ lower limb function and quality of life, are three of the most disastrous complications in pelvic tumor surgery [3, 24]. The infection rate is fairly high (15%-25%) after custom-made endoprosthetic reconstruction or APC reconstruction (Table 2). In our prosthesis, we attempted to reduce the infection rate by three measures: first, minimizing the volume of the prostheses by keeping the necessary structural parts and eliminating the redundant parts that were merely useful for anatomic reconstruction. Second, the implant is made of 3-D-printed titanium alloy, which has good biocompatibility. Third, the acetabular component is entirely embedded in antibiotic-impregnated PMMA cement to reduce metal exposure and fill the dead space, which may allow for bacterial growth. Another approach such as coating the prostheses with silver or iodine has been used for antibacterial purposes without evident toxicologic side effects [29, 31]. However, loss of coating material may decrease the antiinfection effects in the long term. As for implant failure, the risk of screw breakage and implant loosening is high after reconstruction using pelvic megaendoprostheses and APCs (Table 2) attributable primarily to the great shearing force at the sacroiliac joint and the concentrated stress on the fixation screws [16, 24]. Our prosthesis intended to solve this problem by alleviating the stress concentrated on the sacroiliac screws with a pedicle screw and rod system. In addition, osseointegration may be achieved with the 3-D-printed porous surface and multiple sacroiliac screws exerting compressive force on the remaining bone. Hence, lumbopelvic stability is expected in the long term. With regard to local recurrence, although there was a greater proportion of tumors invading both iliac and acetabular regions in our patients, we achieved a favorable incidence of local recurrence in comparison to other studies (Table 2), suggesting the adequate surgical margins were not compromised.

Our patient survival rate of 69% appeared to be comparable to that reported by others (Table 2), although because the types of tumors, their locations, marginal status, adjuvants, and other factors vary, survival rates are not directly comparable. The 5-year cumulative survival rate of patients with pelvic sarcomas has been reported to be 38% to 67% [11, 14, 26, 27]. Sacroiliac invasion, recurrence, and metastasis correlate with poor survival of the patients [11, 19]. We observed three local recurrences in 20 patients, all of whom had sacral extension. We believe that the value of this prosthesis is that it allows reconstruction of resections that include the sacrum making it possible to achieve wider margins. External hemipelvectomy has been recommended in patients with sacral involvement as a result of concerns of compromised surgical margins and lack of a reconstructive solution in a limb salvage procedure [19]. Because our new prosthesis is applicable for reconstruction of extensive sacroiliac defects, our preliminary results suggest using a limb-sparing procedure in selected patients in whom adequate margins are likely to be obtained. Our cumulative incidence of implant failure of 15% at 5 years is encouraging. In other studies, 42% to 87% of implants were noted to survive at different endpoints (Table 2). We acknowledge that more patients and longer followup are still needed to evaluate the durability of this prosthesis. Although we have observed no fixation failures to date, some of the patients may experience extensive revision or external hemipelvectomy if device failure or loosening occurs. In the next generation of pedicle-hemipelvic endoprostheses, we would like to adopt a constrained acetabular component to reduce the probability of hip dislocation, although we have not tested this concept.

We observed fair restoration of function and lumbopelvic stability with expected cumulative incidence rates of major postoperative events and implant failure at short-term followup in this small series using a new pedicle-hemipelvic endoprosthesis for reconstruction after Enneking Type I + II oncologic resections. However, more patients and longer followup are needed to determine the durability of the prosthesis in comparison to other options, because in time, more implant loosening and infection may occur and deteriorate the functional outcome and quality of life of the patients. Nevertheless, we believe that this prosthesis offers potential advantages worthy of further study as an alternative to custom-made megaendoprostheses or APCs for reconstruction of massive pelvic defects.