What Are the Functional Results and Complications From Allograft Reconstruction After Partial Hemipelvectomy of the Pubis?

Abstract

Background

In patients undergoing hemipelvectomies including resection either of a portion of the pubis or the entire pubis from the symphysis to the lateral margin of the obturator foramen while sparing the hip (so-called Dunham Type III hemipelvectomies), reconstructions typically are not performed given the preserved continuity of the weightbearing axis and the potential complications associated with reconstruction. Allograft reconstruction of the pelvic ring may, however, offer benefits for soft tissue reconstruction of the pelvic floor and hip stability, but little is known about these reconstructions.

Questions/purposes

(1) What is the postoperative functional status after allograft reconstruction of Type III pelvic defects? (2) What are the rates of hernia, infection, and hip instability?

Methods

In this case series, we reviewed all patients with Type III pelvic resections (with or without anterior acetabular wall resections) who underwent allograft reconstruction between 2005 and 2013 at one center (N = 14). During the period in question, reconstruction was the general approach used in patients undergoing these resections; during that time, three other patients were treated without reconstruction as a result of either surgeon preference or the patient choosing to not have reconstruction after a discussion of the risks and benefits. Of the 14 patients treated with reconstruction, complete followup was available at a minimum of 1 year in 11 (other than those who died before the end of the first year; median, 19 months; range 16–70 months among those surviving), one was lost to followup before a year, and two others had partial telephone or email followup. Patient demographics, disease status, functional status, and complications were recorded. For a portion of the cohort (four patients) later in the series, we used a novel technique for anterior acetabular wall reconstruction using the concave cartilaginous surface of a proximal fibula allograft; the others received either a long bone (humerus or femur) or hemipelvis graft. Seven patients died of disease; two had local recurrence, and five died of metastatic disease.

Results

All patients remained ambulatory Pain at 12 months after surgery was reported as none in five, mild in two, moderate in two, and severe in one. Operative complications included infection in two, symptomatic hernia in one, hip instability in one, dislocated total hip arthroplasty on the first postoperative day in one, and graft failure in one.

Conclusions

Allograft reconstruction after Type III pelvic resections can provide functional reconstruction of the pelvic ring, pelvic floor, and, in certain patients with partial anterior acetabular resections, the resected anterior acetabulum. This has implications in preventing the occurrence of hernia and hip instability in this patient population that is classically not reconstructed, although longer-term outcomes in a larger number of patients would help to better delineate this because infection, hernia, hip instability, and graft nonunion still remain concerns with this approach. The most important unanswered question remains whether, on balance, any benefits that may accrue to these patients as the result of reconstruction are offset by a relatively high likelihood of undergoing secondary or revision surgery.

Level of Evidence

Level IV, therapeutic study.

Introduction

Surgical treatment of malignant bone tumors of the pelvis is a complex procedure that typically involves wide resection of the affected portions of the pelvis. According to the Enneking and Dunham classification of these resections, Type III internal hemipelvectomy involves resection of either a portion of the pubis or the entire pubis from the symphysis to the lateral margin of the obturator foramen while sparing the hip [5]. These resections are uncommon, accounting for only 9% [1] and 12.6% [2] of cases in two large, recently published series of internal hemipelvectomies. The resulting defect after a Type III resection classically is not reconstructed [4, 5, 8, 12, 17] given the preserved continuity of the weightbearing axis through the proximal femur, acetabulum, sciatic buttress, and spine. In addition, allograft or autograft pelvic ring reconstruction is traditionally thought to add complexity and risk to the procedure, including infection, nonunion, and fracture. The infection rate, for instance, in a previously published series of patients at our institution who underwent allograft reconstruction after internal hemipelvectomy for malignant pelvic bone tumors was 20% [11], although reported rates of infection with pelvic allografts have ranged from one in 13 to five in 13 [1, 3, 7, 10]. Rates of pelvic allograft fracture have been reported ranging from zero of 16 to three of 14 with nonunion reported in one series as three of 16 [3, 6, 10, 13, 18].

On the other hand, bony pelvic ring reconstruction may have several benefits. Structural allograft provides an anchor for mesh and suture attachments that is not otherwise available, adding to the integrity of pelvic floor soft tissue reconstructions and theoretically lowering the rate of symptomatic visceral hernia. Hernia has been occasionally reported as a late complication of hemipelvectomy, occurring between 2 and 5 years postoperatively in one series [9], and thus most surgeons who do not reconstruct the pelvic ring still attempt pelvic floor repair after Type III resections [5, 14, 15]. This could theoretically be improved with the addition of a bony anchor. In addition, to obtain adequate oncologic margins, hip dislocation or subluxation with partial anterior acetabular wall resection may be required. Bony reconstruction of the pelvic ring may provide a way to restore bony constraint at the anterior acetabulum and again provides an anchor for capsular repair or reconstruction. Finally, although no study has evaluated pelvic pain after nonreconstruction of the pelvic ring in the posthemipelvectomy setting, it is our anecdotal experience that anatomic reconstruction of the bony pelvic ring can lead to excellent early stability and functional mobilization.

In this study, we report on a single-center series of allograft-reconstructed Type III and combined partial Type II–Type III pelvic defects. We seek to determine the postoperative functional status and complication rates including infection, symptomatic hernia, and hip instability. In addition, we report on a novel use for the concave cartilaginous surface of the proximal fibular head in reconstructing the anterior acetabular wall after combined Type III and partial Type II internal hemipelvectomy.

Patients and Methods

In this retrospective study, we reviewed all pelvic resections performed at our institution between 2002 and 2012 for local control of a malignant neoplasm of the bony pelvis (N = 167). We were interested in evaluating patients who underwent Type III internal hemipelvectomy (Fig. 1) and had allograft reconstruction of the subsequent pelvic defect at the time of resection. Patients who had acetabular resections in conjunction with their Type III internal hemipelvectomy (often known as Type II–III internal hemipelvectomy) were excluded except if the acetabular resection involved only the nonweightbearing anterior acetabular surface. During the period in question, reconstruction was the general approach used in patients undergoing these resections. Contraindications to reconstruction included infection, immunocompromised state, diabetes mellitus, active tobacco use, and surgeon preference, although we did not see these.

Fig. 1.

Type III internal hemipelvectomy refers to resection of the superior and/or inferior pubic ramus medial to the acetabulum. The area marked in blue in the photograph demonstrates a prototypical “pure” Type III internal hemipelvectomy. Our case series also include patients who had partial resections of the nonweightbearing anterior acetabulum, marked in red in the photograph.

Fourteen patients were identified between 2005 and 2013 at our hospital (Table 1) who underwent Type III resections with allograft reconstruction with six “pure” Type III resections and eight Type III resections with partial anterior acetabular resections. Surgical hip dislocation was required in all eight patients with partial anterior acetabular resections and in two patients with pure Type III resections (as a result of malignant involvement of the proximal femur in one and the need for THA at the time of resection in the other). During the period in question, three other patients were treated without reconstruction as a result of either surgeon preference or the patient choosing to not have reconstruction after a discussion of the risks and benefits.

Table 1.

Summary of patient demographics and outcomes

Case number	Age (years)	Sex	Histology	Resection	Reconstruction	Infection	Hernia	Pain	Followup (months)	Status
1	43	M	Osteosarcoma	Type III+	Hemipelvis	No	No	None	33	DWD
2	60	M	Chondrosarcoma	Type III	Humerus	No	Yes	Mild	77	NED
3	52	F	Chondrosarcoma	Type III	Femur	No	No	Mild	50*	NED
4	21	F	Enchondroma	Type III+	Femur	No	No	Moderate	51	NED
5	42	F	Chondrosarcoma	Type III+	Femur	No	No	—	4	DWD
6	62	M	Chondrosarcoma	Type III	Hemipelvis	No	No	—	19	DWD
7	43	F	Squamous cell carcinoma	Type III+	Hemipelvis	Yes	No	Severe	11	DWD
8	18	M	Osteosarcoma	Type III+	Hemipelvis	No	No	Mild	34	DWD
9	55	F	Myxofibrosarcoma	Type III+	Hemipelvis	Yes	No	Mild	19	DWD
10	41	F	Adenoid cystic carcinoma	Type III+	Hemipelvis	No	No	Moderate	18	DWD
11	61	M	Osteosarcoma	Type III	Fibula	No	No	—	7*	NED
12	12	M	Osteosarcoma	Type III	Fibula	No	No	None	23	NED
13	56	F	Chondrosacroma	Type III+	Fibula	No	No	None	16	NED
14	46	M	Chondrosarcoma	Type III+	Fibula	No	No	—	13*	NED

* Last followup is phone call or email, not a full clinic visit with imaging; Type III+ = Type III resections that also involve the nonweightbearing anterior acetabular surface; M = male; F = female; DWD = dead with disease; NED = no evidence of disease.

Of the 14 patients treated with reconstruction, complete followup was available at a minimum of 1 year in 11 (other than those who died before the end of the first year; median, 19 months; range, 16–70 months among those surviving), one was lost to followup before a year, and two others had partial telephone or email followup. The median age of patients at the time of surgery was 45 years. Histologies included chondrosarcoma (N = 6), osteosarcoma (N = 4), atypical enchondroma (N = 1), metastatic squamous cell carcinoma (N = 1), myxofibrosarcoma (N = 1), and metastatic adenoid cystic sarcoma (N = 1; Table 1).

All patients had bony reconstruction of their pelvic defects at the time of resection. Patients received standard perioperative prophylactic antibiotic coverage. Six were reconstructed with custom-fit hemipelvis allografts and four with long bone (femur or humerus) strut allografts. In four of the patients in the latter part of our series, we used a novel method of reconstruction with a whole fibular allograft. The graft is anchored medially to the pubic symphysis using a pelvic reconstruction plate, and it provides an anchor point for soft tissue reconstruction with mesh (Fig. 2). The concave cartilaginous articular surface of the fibular head is used to approximate the anterior acetabular defect (Fig. 3). To perform this reconstruction after surgical hip dislocation and pelvic resection, a whole fibula was measured, cut, and oriented such that the fibular head concavity matched the remnant host acetabular contour. The distal end of the fibular allograft was recessed and impacted into the cancellous bone of the contralateral host symphyseal bone. A locking pelvic reconstruction “J” plate (Synthes North America, West Chester, PA, USA) was used to secure the fibula in place, spanning the anterior pelvic brim from contralateral pubis and superior ramus to intact ipsilateral ilium. The femoral head was then relocated, and remnant capsular tissue was sewn to the fibular head through drill holes or suture anchors. The postoperative radiograph demonstrates continuity of the pelvic ring achieved with the fibular allograft and the concavity of the proximal fibula in approximating the anterior acetabular defect (Fig. 4). Pelvic floor constraints were reconstructed using a heterogeneous assortment of abdominal mesh products depending on surgeon preference. In addition to the intrinsic support for the pelvic viscera offered by the defect-spanning allograft, the mesh reconstructions were sewn to the allograft fibula and remnant host bone using suture anchors for added support (Fig. 2). Overall, these pelvic floor reconstructions were performed in all but one patient.

Fig. 2.

This intraoperative photograph demonstrates the use of a fibular allograft to reconstruct the pelvic defect after Type III internal hemipelvectomy and the relationship between the bony reconstruction and soft tissue reconstruction using mesh.

Fig. 3.

The concave surface of the fibular head, shown in the inset, is used to approximate the resected anterior acetabular wall. The intraoperative photograph shows the concave surface articulating with the femoral head.

Fig. 4.

This postoperative pelvic radiograph demonstrates bony reconstruction using a fibular allograft after Type III internal hemipelvectomy.

Patients were followed routinely in the orthopaedic oncology clinic after their procedures. Postoperatively, patients were kept partial weightbearing to the affected lower extremity for at least 6 weeks and until there was radiographic evidence of graft union, and clinic notes were reviewed retrospectively to determine postoperative status. Variables assessed related to this study included ambulatory status at the time of last followup, subjective pain assessment at 12 months postoperatively, and complications including, but not limited to, symptomatic hernia, infection requiring operative intervention, symptomatic hip instability, and fracture. Local recurrence, metastatic spread, and fracture were assessed by imaging and/or tissue diagnosis. Ambulatory status, pain, hernia, and hip instability were assessed by attending orthopaedic oncologists in the clinic; no specific outcome instruments were used other than routine history and physical examination during clinical followup. At the time of last followup, seven patients (50%) had died with a median time to death after surgery of 19 months (range, 4–34 months) (Fig. 5). Two patients had local recurrence and died of their disease. Four patients had evidence of metastatic disease after surgery, all of whom died of disease. One patient had stable metastatic disease before hemipelvectomy with further metastases after surgery and died of disease.

Fig. 5.

The Kaplan-Meier survival plot demonstrates the proportion of patients surviving during the course of followup in this series.

Results

In terms of overall postoperative functional status, all patients remained ambulatory after surgery. At latest followup, five patients were able to ambulate without assistive devices, and the remaining nine used a walker, cane, or crutches. We also assessed pain at 12 months after surgery. Pain was rated as “none” in five patients, “mild” in two patients, “moderate” in two patients, and “severe” in one patient.

Six patients had major complications requiring operative interventions. Two patients had infection; one had a symptomatic hernia; two had hip instability; and one had hardware failure with nonunion. A single patient underwent left Type III internal hemipelvectomy and had a symptomatic left inguinal incisional hernia necessitating repair 15 months later. At the time of his internal hemipelvectomy, this patient underwent reconstruction of the abdominal wall with Kugel-type Bard mesh patch repair, and the rectus abdominis muscle was anchored to the allograft. The patient developed postoperative bleeding requiring reexploration on the first postoperative day during which it was discovered that the mesh reconstruction had failed and was replaced. Fifteen months later, there was obvious visceral herniation in the left inguinal region at the site of the internal hemipelvectomy incision, and a defect adjacent to the existing mesh was discovered and repaired with a Marlex plug (Bard PerFix Lot# HUTA1345; C.R. Bard Inc, Murray Hill, NJ, USA) and Parietex mesh (Lot #PIL00545; Covidien AG, Mansfield, MA, USA). Two patients had an infection treated with irrigation and débridement. One patient experienced hip instability, and this patient was treated with further surgery. She initially had undergone surgical hip dislocation during her pelvic resection with reconstruction of the joint using the hemipelvis allograft host capsule. She subsequently had femoral head collapse with superior and lateral dislocation within 3 months of surgery. Pathology demonstrated reactive changes and marrow fibrosis, consistent with avascular necrosis that was likely a consequence of surgical hip dislocation rather than failure of the allograft to maintain anatomic positioning and hip stability. THA was performed approximately 5 months after pelvic resection, and she was subsequently able to walk with an assistive device although with some subjective feeling of hip instability without any actual episodes of dislocation. One patient had a THA as part of the bony reconstruction during the initial pelvic resection, but the hardware dislocated on the first postoperative day; the patient presented with a neurologic deficit, and there was radiographic evidence of dislocation. Intraoperatively, the polyethylene liner was found to be dislocated anteriorly, putting stretch on the sciatic nerve. The patient underwent revision of the femoral and acetabular components. This patient’s initial surgery was complicated by extensive radiation, large tumor mass, and obesity. One patient had pain in his left hip 4 months postoperatively and had radiographic evidence of hardware failure. He was initially managed postoperatively with restriction of strenuous activity, and his pain improved significantly. He did continue to have some discomfort, however, and underwent plate revision with iliac crest bone graft 13 months postoperatively. At 6 months status postrevision, he reported no pain and ambulates unassisted.

Discussion

Bony reconstruction after pelvic resections medial to the acetabulum, known as Type III internal hemipelvectomy, is traditionally not undertaken given the preserved continuity of the weightbearing axis and complications associated with allograft reconstruction. Over the last several years, however, we have reconstructed these defects to provide an anchor for soft tissue reconstruction to prevent visceral hernia formation and to provide a bony constraint to the anterior acetabulum to prevent hip instability and maximize postoperative ambulatory function. Because the standard of care is to not reconstruct, a review of the functional status and complications associated with allograft reconstruction of these defects was undertaken in this study. All patients in our series remained ambulatory postoperatively and pain at 1 year was subjectively assessed to be “none” or “minimal” in seven of the 11 patients for whom complete followup was available. We observed five complications requiring operative intervention, and they included infection, symptomatic hernia, and hip instability. Our series demonstrates that patients who have undergone allograft reconstruction of Type III defects have good postoperative function, although traditional problems such as persistent pain, hip instability, and infection were still seen.

Our study has several limitations. This is a retrospective case series of a small number of patients with a relatively short followup, which is a common design flaw when attempting to examine rare diseases with high mortality rates. In addition, we were not able to provide an internal control group with which to compare our results, because our center does not commonly leave Type III resections unreconstructed. The comparison between unreconstructed and reconstructed Type III defects is a difficult one to make not only because of small numbers, but also because of heterogeneous patient populations, disease processes, extents of resection, need for surgical hip dislocation, and other confounding variables. Three patients at our institution, who were not included in this series, underwent Type III internal hemipelvectomy and were not reconstructed. This raises the concern for selection bias with regard to who is and is not offered reconstruction. At our institution, the standard of care has been to reconstruct these defects, which differs from what is most widely reported in the literature but, as a result, makes this bias less of a concern in terms of preselecting a population of patients who may have better outcomes with allograft reconstruction. Several patients in our series have since died, which is inherent with the disease condition being studied but may raise concern for transfer bias. Complete followup of at least 12 months was, however, available in 11 of 14 patients in our series, including patients who died. Lastly, because of the retrospective nature of the study, we were not able to use objective measures such as the Musculoskeletal Tumor Society score, EQ-5D, SF-36, and other patient-centered metrics.

Our results demonstrate a positive functional result in patients who undergo allograft reconstruction after Type III internal hemipelvectomy. All patients were able to ambulate after surgery, and five of 14 were able to do so without an assistive device. We also report positive results in terms of pain with five of 10 patients for whom subjective assessment of pain at 1 year postoperatively was available reporting pain as none or mild. We are not aware of other studies that have focused specifically on functional outcomes after Type III internal hemipelvectomy with which to compare our results. Others have shown, however, good functional results in patients who had reconstruction after any form of internal hemipelvectomy [2]. In the largest case series of internal hemipelvectomy patients to date, Chao et al. [2] compared functional outcomes between patients who were and were not reconstructed in 111 patients and found low rates of late recipient site complications (dehiscence, hernia, allograft infection, osteomyelitis, wound infection, and seroma) in reconstructed patients. They speculated that significant differences between patients who are and are not reconstructed may be seen in long-term functional outcomes as patients begin more strenuous physical activity, which they may be even more likely to do after a Type III resection compared with more debilitating resections. Longer duration of followup and a comparison of outcomes with nonreconstructed patients would help to establish whether any differences truly exist.

We report complications in our series that are comparable to those previously described in the allograft literature. In reviewing these complications, it is important to acknowledge that there are several risks in allograft reconstruction that are widely reported and serve as reasons why the standard of care is to not reconstruct. These complications must be weighed against the potential benefits that may be reported in this series. For example, we report an operative infection rate of two of 14 patients, which is within the range of what has been previously reported for other series of allograft reconstructions of the pelvis [1, 3, 7, 10, 11]. One patient developed a symptomatic inguinal incisional hernia that was discovered to be the result of a defect in the abdominal wall adjacent to the mesh that was placed at the time of his pelvic resection. Mesh reconstruction of the abdominal wall without bony reconstruction to prevent visceral herniation has been well described. Although one series found no hernias and good functional outcome with this technique in eight patients with mean followup of 9.5 years [14], it is our experience that the addition of allograft provides a more secure bony anchor for both the mesh and the abdominal and hip musculature and thus may offer improved long-term protection against visceral herniation. This is especially relevant because visceral hernias have been reported to occur late, several years after internal hemipelvectomy [9].

There is some evidence that anterior hip instability can occur after anterior acetabular resections. Trousdale [16] reported on a patient with excessive anterior acetabular version who experienced nine anterior hip dislocations before being successfully treated with periacetabular osteotomy to retrovert the acetabulum. His discussion of the variation in femoral head coverage with different degrees of acetabular anteversion is one factor that may play a role in the varying degrees of hip stability in patients who undergo Type III internal hemipelvectomy and the subsequent advantage of anterior coverage with bony reconstruction. With regard to the novel use of the fibular head cartilaginous surface, we are unaware of any previous descriptions of this technique. The concave articular surface of the fibular head provides a good anatomic match for the anterior acetabular surface that is occasionally resected with Type III internal hemipelvectomies and may contribute to better hip stability and potentially better hip function relative to nonreconstruction or other forms of allograft. However, this will need to be tested by future, larger series.

In summary, we report the results of a series of patients who underwent allograft reconstruction after Type III internal hemipelvectomy. We found positive functional results in terms of pain and ambulation but concomitant operative complications including hernia, infection, and hip instability. The use of a fibular allograft in particular is a novel method for approximating an anterior acetabular defect and may warrant future study. However, the most important unanswered question remains whether, on balance, any benefits that may accrue to these patients as the result of reconstruction are offset by a relatively high likelihood of undergoing secondary or revision surgery.