What Is the Revision-free Survival of Resurfaced Allograft-prosthesis Composites for Proximal Humerus Reconstruction in Children With Bone Tumors?

Abstract

Background

Reconstruction of the proximal humerus in children who undergo bone tumor resection is challenging because of patients’ small bone size and possible limb length discrepancy at the end of skeletal growth due to loss of the physis. There are several options for proximal humerus reconstruction in children, such as clavicula pro humero, free vascularized fibula grafting, massive bone osteoarticular allografting, endoprostheses, and allograft-prosthesis composites, but no consensus exists on the best method for reconstruction. Resurfaced allograft-prosthesis composites could be an alternative surgical option, but little is known about the results of this surgical technique.

Questions/purposes

(1) What are the complications and what is the survivorship free from reconstruction failure associated with resurfaced allograft-prosthesis composites in a small, single-center case series? (2) What Musculoskeletal Tumor Society scores do patients achieve after reconstructions with resurfaced allograft-prosthesis composites?

Methods

This study was a retrospective, single-arm case analysis in a single institution. We generally considered resurfaced allograft-prosthesis composites in children with malignant bone tumors involving the metaepiphysis of the proximal humerus in whom there was no evidence of joint contamination and in whom axillary nerve preservation was possible. Between 2003 and 2021, we treated 100 children (younger than 15 years) with bone tumors of the humerus. Thirty children (30%) with diaphyseal tumors (21 children) or distal tumors (9 children) were excluded. Among the potentially eligible children, 52 were not analyzed because they were treated with other procedures such as amputation, modular prostheses, cement spacers, free vascularized fibula grafting, and massive bone osteoarticular allografts. We included 18 children (26% of the potentially eligible children) who were treated with resurfaced allograft-prosthesis composites. There were 9 boys and 9 girls, with a median age of 10 years (range 4 to 15 years) at the time of diagnosis. A long stem (≥ 6 cm) in the resurfaced allograft-prosthesis composite was used in 9 children and a short stem (< 6 cm) was used in the remaining 9. One of the 18 children had a follow-up of less than 2 years. The median follow-up of the remaining 17 children was 4.7 years (range 2 to 19 years). The children’s medical records were reviewed for clinical and functional outcomes. We performed a competing risk analysis to calculate the reconstruction failure-free survival of resurfaced allograft-prosthesis composites. Reconstruction failure was defined as removal of the implant or allograft because of implant loosening or breakage and allograft fracture or resorption. We analyzed the children’s postoperative complications and functional outcomes at the end of the follow-up period using the Musculoskeletal Tumor Society functional scoring system.

Results

The competing risk analysis revealed that reconstruction failure was 25% (95% confidence interval 7% to 40%) at 3 years, reaching a plateau. Four of 18 children underwent surgical revision with a new reconstruction. The reasons for reconstruction revision were resorption of the allograft at the proximal part (2 children) and fracture of the allograft (2 children). Reconstruction revision was performed in 3 of 9 children who underwent reconstruction with a short stem and in 1 of 9 children who underwent reconstruction with a long stem. Several children had other complications that did not result in removal of the allograft. Allograft resorption was observed in 4 of 18 children, but no additional surgical treatment was performed. Shoulder instability or subluxation was observed in 4 of 18 children, but only 1 child underwent surgery with a reverse shoulder arthroplasty without removal of the resurfaced allograft-prosthesis composite. Limited elbow motion because of plate impingement was observed in 1 child who underwent surgical cutting of the protruding distal part of the plate. Incomplete radial nerve palsy after surgery occurred in 1 child, with spontaneous resolution after 2 months. Screw loosening occurred in 2 children who underwent surgery with removal of loose screws. Two children had a nonunion at the graft-host bone junction; 1 child underwent surgery with bone grafting and refixation of the graft-host bone junction, and the other child with both nonunion and plate breakage was treated with bone grafting and refixation of the graft-host bone junction with a new plate. Among 17 children who had a follow-up longer than 2 years, the median Musculoskeletal Tumor Society functional score at the last follow-up interval was 23 of 30 (range 20 to 26); 1 child was considered to have an excellent result (functional score ≥ 26), 15 children were considered to have a good result (functional score 21 to 25), and 1 child was considered to have a fair result (functional score ≤ 20). The Musculoskeletal Tumor Society functional score did not change after excluding 4 children who underwent replacement of resurfaced allograft-prosthesis composites (24 of 30 [range 20 to 26]). The median angle of flexion of the shoulder was 40° (range 20º to 90°), and the median angle of abduction was 30° (range 20º to 90°).

Conclusion

Resurfaced allograft-prosthesis composites showed a high risk of complications, but not all complications resulted in removal of the reconstructed allograft. We used this technique mainly for very young children with small bones and for older children who underwent axillary nerve preservation. Although its success may be limited owing to a high risk of complications, a resurfaced allograft-prosthesis composite could be an alternative surgical option in order to preserve the bone stock and achieve good functional outcomes in very young children. We recommend using a long-stem resurfaced allograft-prosthesis composite, which may reduce the risk of complications.

Level of Evidence

Level IV, therapeutic study.

Introduction

The proximal humerus is the third most common location of bone sarcomas after the distal femur and proximal tibia [4, 12]. Reconstruction options after proximal humerus resection of bone sarcomas include clavicula pro humero, free vascularized fibula grafting, massive bone osteoarticular allografts, endoprostheses, and allograft-prosthesis composites (APCs) [5, 11]. The evidence supporting one type of reconstruction over the other for proximal humerus reconstruction in children with bone sarcomas is limited [4, 12], and no consensus exists on the best reconstruction method [12, 15]. Because clavicula pro humero may be associated with complications such as nonunion and limited shoulder motion, it seems to be useful for reconstruction in children with involvement of the nondominant arm [5, 8, 17]. Free vascularized fibula grafting may be associated with a longer operative time, knowledge of specific microsurgical skills, and sacrifice of autologous bone [6]. In addition, it carries the risk of complications such as fibula head resorption, graft fracture, and nonunion [6]. Articular deterioration, nonunion, resorption, and fracture are the most frequent complications associated with massive bone osteoarticular allografts [2, 12]. Another important concern when using massive bone osteoarticular grafts in children is the potential mismatch between graft size and the host bone [9]. Endoprosthesis is a reliable method of proximal humerus reconstruction after bone tumor resection in adolescents and adults. However, children younger than 9 years might have a high risk of subluxation [3, 16], infection, and aseptic loosening [10, 18].

Reconstruction with an APC combines the advantages of prosthetic and biological reconstructions [1, 13-15] by avoiding problems related to mismatch between a massive bone osteoarticular allograft and the host bone and preserving the elbow and bone stock for future revision surgery [9, 11]. To maximize the potential benefits of the APC, we used a surgical reconstruction technique consisting of a resurfacing humeral prosthesis that was cemented in a massive bone allograft; that is, a resurfaced APC (rAPC) (Fig. 1). However, whether this approach will be able to take advantage of the putative benefits noted and avoid serious potential complications has not, to our knowledge, been explored.

Fig. 1.

This figure shows a resurfacing allograft-prosthesis composite reconstruction, including (A) a bone tumor (purple) in the proximal humerus without joint contamination and (B) reconstruction of the proximal humerus with a resurfaced allograft-prosthesis composite (blue). A color image accompanies the online version of this article.

We therefore asked: (1) What are the complications and what is the survivorship free from reconstruction failure associated with rAPCs in a small, single-center case series? (2) What Musculoskeletal Tumor Society (MSTS) scores do patients achieve after reconstructions with rAPCs?

Patients and Methods

Study Design and Setting

This study was a retrospective analysis of children with tumors of the proximal humerus treated at Istituto Ortopedico Rizzoli in Bologna, Italy. The Istituto Ortopedico Rizzoli is an urban referral orthopaedic institution with large subspeciality expertise in orthopaedic oncology. Five orthopaedic surgeons participated in this study. Diagnoses were made through a pathologic analysis of surgical specimens.

Participants

Between 2003 and 2021, we treated 100 children younger than 15 years who had tumors of the humerus that were histologically diagnosed through biopsy. We considered this approach in children with malignant bone tumors in the metaepiphysis of the proximal humerus. In total, 70 of 100 children were eligible; the remaining 30 children with diaphyseal tumors (21) or distal tumors (9) were excluded. We also excluded children with joint contamination or soft tissue extension around the glenoid fossa, as well as those who did not respond to chemotherapy. Among the eligible children, 2 children, in whom limb salvage was not possible, underwent amputation. Children who had resection of the axillary nerve and those with metastases received a modular prosthesis (29). Massive bone osteoarticular allografts (6) were mainly used for children in the early 2000s. Very young children with small bones in which modular prostheses could not be used were treated with cement spacers (4 children with metastatic disease) or free vascularized fibular grafts (11 children with localized disease). Finally, very young children with small bones or older children in whom it was possible to save the axillary nerve (26% of potentially eligible children [18 of 70]) were treated with rAPCs and included in our analysis (Fig. 2). All children were treated by surgeons with experience in orthopaedic oncologic surgery.

Fig. 2.

This flow diagram represents children with bone tumors in the proximal humerus treated at our institution during the study period.

We generally selected rAPC instead of a modular prosthesis to achieve a better functional result and save bone stock (host bone) for future revision surgeries. In the early 2000s, we stopped using massive bone osteoarticular allografts because of the high risk of resorption (especially of the humeral head) and fractures that were treated with removal of the reconstructed allograft. Then, we started using rAPC because it allowed us to save the reconstruction despite the presence of resorption or fracture of the bone allograft. As an alternative to the vascularized fibula and clavicular pro humero, we considered rAPC reconstruction in very young children in whom the dimensions of the modular prosthesis did not match the bone size. In addition, rAPC was used in older children in whom preservation of the axillary nerve was possible, because we considered that functional results after rAPC would be better than those of modular prostheses.

Descriptive Data

Among the 18 children included in this study, there were 9 boys and 9 girls, with a median age of 10 years (range 4 to 15 years) at the time of diagnosis. The diagnosis was conventional osteosarcoma (9 children), Ewing sarcoma (5), low-grade osteosarcoma (1), chondrosarcoma (1), metastatic osteosarcoma in the proximal humerus (1), and metastatic adrenal neuroblastoma in the proximal humerus (1). The median tumor length was 8 cm (range 3 to 22 cm). All children underwent preoperative and postoperative chemotherapy except for two (one with low-grade osteosarcoma and one with chondrosarcoma). None of the children received radiotherapy. The median height of the children was 136 cm (range 100 to 176 cm), median weight was 34 kg (range 16 to 62 kg), and median BMI was 17 kg/m² (range 13 to 26 kg/m²) (Table 1). One patient moved to another country at 11 months and was lost to further follow-up. All other patients had at least 2 years of follow-up, and the median follow-up was 4.7 years (range 2 to 19 years).

Table 1.

Details of patients who underwent proximal humerus reconstruction with resurfaced allograft-prosthetic composites

Patient number	Sex	Histology	Age in years	Height at diagnosis in cm	Weight at diagnosis in kg	Size of tumor (maximum diameter) in cm	Stem
1	B	Osteosarcoma	13	1.75	50	11	Long
2	B	Osteosarcoma	9	1.30	34	8	Short
3	G	Osteosarcoma	9	1.34	25	9.6	Long
4	G	Chondrosarcoma	15	1.60	60	22	Long
5	B	Ewing sarcoma	8	1.33	33	19	Short
6a	B	Osteosarcoma	15	1.76	56	10	Long
7	B	Ewing sarcoma	9	1.35	39	9.5	Long
8b	B	Ewing sarcoma	4	1.00	17	6	Short
9	G	Osteosarcoma	13	1.65	38	13.5	Short
10	G	Ewing sarcoma	13	1.57	46	6	Long
11	B	Metastatic neuroblastoma	11	1.37	30	3	Short
12	G	Ewing sarcoma	10	1.22	39	4	Long
13b	G	Osteosarcoma	5	1.10	17.5	6	Short
14	B	Osteosarcoma	8	1.14	17	9.5	Short
15	G	Osteosarcoma	7	1.40	27	6	Short
16	B	Osteosarcoma	14	1.70	62	6.3	Long
17	G	Metastatic osteosarcoma	10	1.40	32	7	Short
18	G	Osteosarcoma	4	1.08	15.5	7	Long

Patient number	Reconstruction failure	Other complication	Surgery for the complication (time from first surgery to subsequent surgery in months)	Follow-up in months and disease status	MSTS score at final follow-upc	Bone stock (% of the total length of humerus)
1	No	Nonunion	Addition of an autograft because of nonunion (22)	52, CDF	26	50% (16/32 cm)
2	Fracture of allograft at the proximal part	No	Renewal of allograft and reinternal fixation (17)d	216, CDF	23	39% (11/28 cm)
3	No	Loosening of screw	Removal of screw (72)	73, CDF	21	52% (13/25 cm)
		Mild subluxation	No
4	No	Soft tissue recurrence	Re-resection of tumor (25)	55, CDF	24	20% (6/30 cm)
		Mild resorption of allograft	No
5	No	No	No	132, CDF	26	14% (4/28 cm)
6a	No	No	No	11, followed in another country	20	12% (4/34 cm)
7	No	Mild resorption of allograft	No	40, CDF	25	23% (7/30 cm)
8b	No	Mild resorption of allograft	No	56, CDF	20	39% (7/18 cm)
9	No	Loosening of screw	Removal of screw and adjustment of plate (62)	142, CDF	25	45% (15/33 cm)
		Mild subluxation	No
10	No	Severe subluxation of shoulder	Conversion to reverse shoulder arthroplasty (32)	33, CDF	23	48% (16/33 cm)
11	No	No	No	45, DOD	25	44% (12/27 cm)
12	No	Mild subluxation	No	236, CDF	22	60% (18/30 cm)
13b		Plate breakage at the junction of allograft and host bone because of nonunion	Reinternal fixation without removal of prosthesis (6)	63, CDF	24	41% (9/22 cm)
	Resorption of allograft at the proximal part		Renewal of rACP (19)
14	No	Impingement of plate at elbow	Adjustment of plate	24, CDF	24	15% (3/20 cm)
		Elbow limitation or limb length discrepancy (> 5 cm)/radial nerve palsy	No
15	Fracture of allograft at the distal part	No	Insertion of intercalary prosthesis (26)	230, CDF	23	60% (15/25 cm)
16	Severe instability of shoulder		Conversion to reverse shoulder arthroplasty (30)	79, CDF	23	53% (18/34 cm)
		Resorption of allograft at proximal part	No
17	No	No	No	39, DOD	21	50% (15/30 cm)
18	No	Mild resorption of allograft	No	24, CDF	23	56% (10/18 cm)

^aThe patient was excluded from the analysis of functional outcomes owing to a short follow-up period.

^bPatients received a fibula allograft.

^cMSTS scores at the final follow-up interval were evaluated at various timepoints.

^dThe patient underwent a second revision with renewal of rACP at 176 months after the first surgery because of fracture of the allograft at the proximal part, and underwent a third revision with a modular prosthesis at 192 months after the first surgery owing to resorption of the allograft.

MSTS = Musculoskeletal Tumor Society functional scoring system; B = boy; CDF = completely disease-free; G = girl; DOD = death of disease; rAPC = resurfaced allograft-prosthetic composite.

Description of Treatment

Preoperatively, all children were evaluated with plain radiography, MRI of the humerus, and chest CT to assess the extent of the tumor and presence of distant metastases. Resection was performed through a standard deltopectoral approach to achieve wide margins. Moreover, the length of the resected specimen was carefully measured for later reconstruction. Preoperatively, images of the hosts and donors were compared in order to best fit the fresh-frozen allografts. When the humerus was so small that no appropriate donor could be found, a fibular allograft was considered. Massive bone allografts were resected at the anatomic neck, and the medullary canal was reamed to hold the resurfacing prosthesis. The stem of the resurfacing prosthesis was cemented into the massive bone allograft, in both fibula and humerus allografts. The massive bone allograft was then fixed to the host bone with a titanium or steel plate (4.5 or 3.5 mm, depending on the size of the bone) with locking and nonlocking screws on the lateral side of the humerus. A medial plate was added to reinforce the initial fixation in 5 children who underwent a long resection. Different resurfacing implants were used over time: The Bigliani/Flatow prosthesis (Zimmer Biomet) was used in 8 children, and the GLOBAL CAP humeral head prosthesis (DePuy Synthes) and SMR Resurfacing prosthesis (Lima Corp) were used in 1 child each. A 3D-printed custom-made resurfacing prosthesis was used in 8 children (Fig. 3).

Fig. 3.

A 4-year-old boy had Ewing sarcoma of the right proximal humerus. (A) A preoperative radiograph shows an osteolytic lesion with a periosteal reaction. (B) T2-weighted MRI with fat suppression shows an abnormal high-intensity signal with soft tissue invasion. (C) A radiograph shows sclerotic change in the lesion after preoperative chemotherapy. (D) A postoperative radiograph shows reconstruction with a resurfaced allograft-prosthesis composite using a fibula as a massive bone allograft and a custom-made resurfacing prosthesis (Adler Ortho) fixed with a medial bone allograft and lateral plate (Proximal Lateral Humeral Locking Plates, Zimmer). Radiographs show allograft consolidation at (E) 6 months and (F) 6 years after surgery.

The median total operation time was 4 hours (range 2 to 8 hours). The median blood transfusion was 1 unit (range 0 to 4 units). The median length of proximal humeral resection was 14 cm (range 8 to 28 cm), and the median allograft length used for reconstruction was 14 cm (range 9 to 30 cm). After resection of bone tumors, the median length of the remaining humerus (bone stock) was 45% (range 12% to 60%) of the total length of the humerus. We used humerus allografts in 16 children and fibula allografts in 2. The median size of spherical implants in the epiphysis was 40 mm (range 30 to 52 mm). A long stem (≥ 6 cm) was used in 9 children and a short stem (< 6 cm) was used in the remaining 9. We decided to use a short stem in two fibula bone allografts because of the small size of the graft, while for humerus bone allografts, the choice of a short or long stem was based on the surgeon’s preference. All resurfacing prosthesis stems were cemented into the graft. The surgical margins were considered wide in all children.

Aftercare

Postoperatively, the upper limb was immobilized in a thoracobrachial abduction brace for 4 to 6 weeks. After removal of the brace, children were allowed to use the upper limb for basic daily activities, but engagement in sport activities remained restricted until 8 months after surgery.

Description of Follow-up Routine

Surveillance for local and systemic recurrence consisted of clinical and imaging assessments (radiography of the humerus and CT of the lung) every 3 months for the first 2 years, followed by every 6 months up to 5 years and annually thereafter.

Primary and Secondary Study Outcomes

Our primary study goal was to analyze reconstruction failure-free survival at 3 years after surgery, which was calculated using a competing risk analysis. Reconstruction failure was defined as removal of an implant or allograft because of implant loosening or breakage and allograft fracture or resorption. Procedures such as screw or plate removal and revitalization or addition of autografts for nonunion were not considered reconstruction failure. Our secondary study goals were to assess other complications that were not considered reconstruction failure and to evaluate patients’ postoperative function at the final follow-up, based on MSTS score.

Radiographic and CT images were reviewed and analyzed by three study authors (CE, HA, and AA). Union of the allograft-host bone junction was considered present when the junction line was no longer visible or the junction was bridged with periosteal bone on three of the four cortices using AP and lateral radiographic views or CT images. Although we cannot define the exact time of bone union, we reviewed radiographs every 3 months postoperatively until bone union was confirmed.

Functional outcomes were evaluated using the MSTS functional scoring system, which assesses pain, function, emotional acceptance, hand positioning, and manual dexterity [7]. Functional outcomes were defined as follows: excellent (≥ 26 points), good (21 to 25 points), and fair (≤ 20 points). Functional outcomes and shoulder ROM were evaluated at the final follow-up interval [7]. Shoulder ROM was objectively evaluated with a goniometer. We excluded 1 of 18 children from the functional analysis; this patient was lost to follow-up 11 months after surgery.

Ethical Approval

Ethical approval for this study was obtained from the ethical review board of Comitato Etico di Area Vasta Emilia Centro (approval number: 0011245).

Statistical Analysis

Values are presented as means or medians, as appropriate, based on a Shapiro-Wilk analysis. The statistical analysis was performed with SPSS version 26 (IBM Corp) and R (R Foundation for Statistical Computing). Cumulative incidence functions were estimated as the probability of dropout because of the presence of competing events (death of the children or failure of rAPC reconstruction).

Results

Survivorship and Complications

A competing risk analysis was performed in all 18 children, showing that reconstruction failure-free survival was 25% (95% confidence interval 7% to 40%) at 3 years, reaching a plateau (Fig. 4). The median time to revision was 28 months (range 19 to 30 months). We did not consider plate breakage as reconstruction failure. Surgical revision was performed in 3 of 9 children who underwent reconstruction with short-stem rAPC and in 1 of 9 children who underwent reconstruction with a long-stem rAPC. The reasons for revision were resorption of the allograft at the proximal part (2 of 18 children), fracture of the allograft at the proximal part (1 of 18 children), and fracture of the allograft at the distal part (1 of 18 children). At the time of the first revision surgery, 3 patients underwent conversion from rAPC to modular prostheses. One patient who had an allograft fracture at the distal part received revision surgery with an intercalary modular prosthesis. One patient who had allograft resorption underwent revision surgery with reverse modular shoulder arthroplasty. One patient underwent multiple revision surgeries with a new rAPC twice and final revision surgery with a modular prosthesis (owing to allograft fracture in the first and second revision surgeries and allograft resorption in the third revision surgery). One patient underwent revision surgery with a new rAPC because of allograft resorption.

Fig. 4.

This graph shows a competing risk analysis for survival of reconstructions with a resurfacing allograft-prosthesis composite free from removal of the graft or implant. Reconstruction failure-free survival was 25% (95% CI 7% to 40%) at 3 years and reached a plateau.

Complications that were not treated by removal of the reconstructed allograft included allograft resorption, instability or subluxation, plate impingement, incomplete radial nerve palsy, and screw loosening (Table 2). Allograft resorption occurred in 4 of 18 children. Instability or subluxation of the shoulder occurred in 4 of 18 children, one of whom underwent reverse shoulder arthroplasty without removal of the rAPC. We observed limited elbow motion because of plate impingement in 1 child who underwent surgery with removal of the protruding distal part of the plate. Incomplete radial nerve palsy occurred in 1 child, with spontaneous resolution after 2 months. Screw loosening in the allograft without evidence of nonunion was observed in 2 of 18 children, both of whom underwent removal of loose screws. These children did not undergo removal of the allograft because screw loosening was not related to nonunion. Two of 18 children had nonunion at the junction of the allograft and host bone; 1 child underwent surgery with bone grafting and refixation of the graft-host junction, and the other child had nonunion and plate breakage and was treated with bone grafting and refixation with a new plate at the graft-host junction. Local recurrence in soft tissue occurred in 1 of 18 children 25 months after surgery. This child underwent additional surgery with excision of the local recurrence.

Table 2.

Complications after proximal humerus reconstruction with resurfaced allograft-prosthetic composites in children with bone tumors

Complication		Incidence	Removal of the implant or allograft	Additional surgery without removal of the implant or allograft
Soft-tissue failure	Subluxation of shoulder joint	4		1
Graft-host nonunion	Nonunion	2		2
Structural failures	Allograft resorption	6	2
	Loosening of screw	2		2
	Fracture	2	2
	Plate breakage	1		1
Infection	Infection	0
Tumor progression	Soft tissue recurrence	1		1
Pediatric failure	Limb length discrepancy (> 5 cm)	1
Other	Limitation of elbow	1
	Impingement of plate	1		1
	Incomplete radial nerve palsy	1

Functional Outcomes

In 17 of 18 children who had follow-up longer than 2 years, the median MSTS functional score at the last follow-up was 23 of 30 (range 20 to 26). One child had an excellent functional result (MSTS functional score ≥ 26), 15 children had good functional results (MSTS functional score 21 to 25), and 1 child had a fair functional result (MSTS functional score ≤ 20). The MSTS functional score did not change, after excluding 4 children who underwent replacement of rAPCs (24 of 30 [range 20 to 26]). The median angle of flexion of the shoulder was 40° (range 20º to 90°), and the median angle of abduction was 30° (range 20º to 90°).

Discussion

Surgical options for reconstructing the proximal humerus after bone tumor resection in children include biological or prosthetic implants [9]. Various reconstruction techniques have been reported, including clavicula pro humero, free vascularized fibula grafting, endoprostheses, massive bone osteoarticular allografts, and allograft-prosthetic composites [5, 9, 11, 15]. The main challenge of these reconstructions in children is their small bones; therefore, using a massive bone allograft or endoprosthesis may result in a size mismatch with host bone after resection [9]. In addition, reconstruction of the proximal humerus is associated with bone loss [17]. The rationale of using rAPC was to prevent the loss of bone stock, achieve acceptable articular congruency, and preserve as much bone as possible for future revision surgery. Our results suggested that reconstructing the proximal humerus with rAPC in children in whom the axillary nerve and deltoid can be preserved may be considered as an alternative surgical option to endoprostheses or massive osteoarticular allografts. Clavicula pro humero and free vascularized fibula grafting are two other possible surgical techniques for very young children. However, clavicula pro humero may sometimes result in poor functional outcomes, and free vascularized fibula grafting is a complex procedure that usually involves collaboration with plastic surgeons and requires advanced surgical skills [5, 9]. Therefore, in very young children with no other available alternative for reconstruction, rAPC could be a surgical option. Despite the high risk of complications, most children in our series retained their allografts, achieving good functional outcomes at the last follow-up interval.

Limitations

The primary limitation of our study is the small number of patients and the absence of a direct comparison with other surgical techniques because of the rarity of malignant tumors in the proximal humerus of young children. Nevertheless, we could assess functional results and complications in a limited number of children who underwent rAPC. There are no other series reporting the results of this surgical technique in children with tumors of the proximal humerus. Second, there might have been selection bias regarding children undergoing rAPC for tumors of the proximal humerus. We selected children in whom preservation of the axillary nerve was possible for reconstruction with rAPC, while potentially excluding children with large tumors who would be treated with larger amounts of soft tissue resection to achieve a margin. Third, there were no infections, which could be attributed to the generally lower infection risk among children (especially very young children) than among adults. Additionally, the selection of rAPC was based on the possibility of preserving the axillary nerve, enabling sufficient soft tissue coverage during reconstruction and thereby contributing to a reduced risk of infection. We were unable to determine the exact time of union of the allograft-host bone junction because it would have been necessary to evaluate patients very frequently (weekly) and at the same intervals all the time. We were only able to retrospectively evaluate the children’s radiographs every 3 months, and three of the authors agreed regarding union of the allograft-host bone junction. Finally, the median follow-up duration was too short to determine how this reconstruction might work when the patient reaches skeletal maturity and how durable it will be in adulthood. In addition, the MSTS score was gathered retrospectively from the medical records of children at their last follow-up visit. Late complications related to abnormal shoulder articulation may become more apparent with longer follow-up.

Survivorship and Complications

Complications were frequent in this series; in fact, most patients treated with this approach had a complication, revision operation, or both. The revision surgery rate was nearly 25% at 3 years. Given the high rate of revision and the frequent complications mentioned, we conclude that this approach is an option, but it should be reserved for patients for whom there are no reasonable alternatives. APCs are an attempt to address the limitation of endoprostheses and massive osteoarticular allografts [12]. An APC may preserve bone stock, and a resurfacing prosthesis avoids the need for an exact match between the size of the host bone and bone graft [9]. We reported the results of using rAPC in children who underwent resection of the proximal humerus for bone tumors, showing that rAPC, similar to other surgical techniques, was associated with a high risk of complications. On the other hand, rAPC could preserve bone stock in very young children who may need surgical treatments in the future. An endoprosthesis to reconstruct the proximal humerus in children is one of the most commonly used surgical techniques [12]. However, despite innovations in materials and designs, the proportion of patients who undergo revision or removal of prostheses remains high. Infection seems to be the most common cause of failure, followed by aseptic loosening and shoulder instability [9]. In a retrospective study analyzing 25 children who had an endoprosthetic replacement of the proximal humerus after resection of bone sarcomas, 9 of 25 children had proximal subluxation of the endoprosthesis and 3 children had deep infections [9]. In another study analyzing 35 children with sarcomas of the proximal humerus who underwent resection and reconstruction with extendable endoprostheses [16], the overall reconstruction survival at 10 years was 74% (26 of 35 patients). The most common cause of failure was aseptic loosening (4 of 35 patients) [16]. Subluxation at the proximal humerus occurred in 54% of children (19 patients) and it was the most common complication, seen more commonly in children younger than 9 years (86% of patients). The authors suggested that biological reconstruction may be a better option for these very young children [16]. Furthermore, endoprosthetic replacement after proximal humerus resection may be difficult in children because of the small medullary canal of the humerus and possible cortical resorption at the bone-implant interface [9].

A study reported that free vascularized fibula grafting or clavicular pro humero could be an option in very young children in whom endoprosthetic replacement seems to be inappropriate [16]. However, the risk of complications of these biological reconstructions may be very high. The risk of aseptic loosening, implant breakage, and avascular necrosis ranges from 18% to 30% [5-7]. In a retrospective study analyzing 11 children who underwent reconstruction of the proximal humerus with free vascularized fibula grafting, 7 children had a fracture and most of them needed reconstruction revision [9]. A recent study reported the results of 11 children who underwent free vascularized fibula grafting for primary bone tumors of the proximal humerus, showing that 7 patients had fractures, 4 had transient nerve palsies, and 2 experienced avascular necrosis of the graft [14]. In addition, free vascularized fibula grafting could be complicated by the risk of fibula head resorption [6] and donor site morbidity [5, 9].

Functional Outcomes

Our results showed that patients who had rAPC to reconstruct the proximal humerus had reasonable functional results. Most patients had MSTS scores that would be considered “good,” which means they still had functional limitations, pain, or both, but in reconstructions that are this complex, this is unsurprising. One small study analyzed the MSTS functional score in children who underwent proximal humerus reconstruction after bone tumor resection; the mean MSTS functional score was 20.9 in prosthetic reconstructions, 23.3 in massive osteoarticular allograft reconstructions, 22.9 in free vascularized fibula grafting reconstructions, and 23.6 in APC reconstructions [9]. Limited active shoulder function after proximal humerus resection of bone tumors has been reported, and similar functional results have been found with the use of endoprostheses, massive bone osteoarticular allografts, or APC reconstructions [9]. The use of rAPC may reduce some of the problems associated with a massive bone graft or prosthesis alone [9]. A massive bone osteoarticular allograft provides an anatomic fit with bone stock restoration and tendon insertions, enhancing glenohumeral stability. The prosthesis reduces the risk of subchondral fracture and resorption of the bone allograft [9].

Conclusion

Similar to the other surgical techniques, reconstruction with rAPC after proximal humeral resection for bone tumors may have many serious complications; in fact, most children treated with this approach had a complication, a revision, or both. The revision rate was nearly 30% at 3 years; considering the high rate of revision along with the frequent complications noted above, we conclude that this approach is an option, but it should be reserved for patients in whom there are no reasonable alternatives. This surgical technique can provide attachment sites for the capsule and tendons and can preserve bone stock. Resurfacing prostheses provide mechanical support against subchondral collapse, saving bone stock for possible future surgical treatments. We believe this is best used in patients in whom the axillary nerve and deltoid can be preserved. We cannot show this approach is superior to free vascularized fibula grafting, endoprostheses, or massive osteoarticular allografts alone until this approach is directly compared with these alternatives, but we suggest this as a possible alternative treatment for select patients. Although its success is limited by a high risk of complications, rAPC seems to be an option to preserve the bone stock in very young children with proximal humerus tumors. In our patients, if revision was performed after rAPC reconstruction, replacement with a new APC reconstruction or modular prosthesis could be performed. Based on our results, a long stem in an rAPC reconstruction may reduce the risk of complications. Multicenter prospective studies with a larger population, longer follow-up, and comparisons with other reconstruction options will be necessary to confirm the results of our study.