Our study was designed to assess the oncological outcome of proximal humerus osteosarcoma treated by limb salvage surgery and chemotherapy. Furthermore, we compared the functional outcomes after limb salvage using two main methods of reconstruction: biological reconstruction using vascularized grafts with shoulder fusion (Fig. 3) and nonbiological reconstruction using a PMMA spacer or endoprosthesis with a mobile shoulder. Finally, we assessed the prognostic factors that affect both the oncological and functional outcomes. We assessed oncological outcome in terms of 5-year survivorship, local recurrence and chest metastasis, while functional outcome was assessed using the MSTS score.

A Preoperative investigations. Left: X-ray of the Rt shoulder showed an extensively sclerotic lesion with a periosteal reaction, as indicated by a Codman triangle and sunburst appearance. Center: Axial-cut MRI of the shoulder with T1 gadolinium tumor enhancement. Right: Coronal-cut MRI demonstrating a high signal intensity on T2-weighted imaging. B Resected specimen. C Immediate postoperative X-ray (left) and imaging performed at 12 weeks of follow-up (right), with fusion at the humerus–implant junction readily visible. D At 6 months of follow-up, the patient showed satisfactory elbow flexion ROM (range of motion), varying between 10 and 120 degrees, while shoulder abduction was limited to 40 degrees

The 5-year survivorship in our cases was 65%, which is similar to that reported by Wittig et al. [17], who managed 23 patients with a proximal humerus resection for stage IIA and IIB together with an endoprosthesis. Yao et al. [18] estimated the 5-year survival rate at 71%. In 2009, Gupta et al. in 2009 reported on 23 cases, almost all of whom were treated by limb sparing, and noted that survivorship markedly decreased with time from 77% at 5 years of follow-up to 57% at 10 years [9].

Our study showed that local recurrence occurred in 18% of the patients and was not statistically correlated with any of the studied prognostic factors, especially resection length, deltoid resection, and pathological fractures on presentation. Our study obtained the same result as Gupta et al. [9] in terms of local recurrence, which occurred in 13% in their cases and was highly linked with positive margins.

Chest metastasis occurred in 41% of patients, which is higher than in Yao et al.’s study, where distant metastasis occurred in 30% [18].

In comparison with patients with osteosarcoma at other sites who were also managed in our institution, proximal humerus osteosarcoma patients showed higher incidence rates of local recurrence and chest metastasis along with lower 5-year patient and limb survivorships compared to distal femur, proximal tibia and proximal femur osteosarcoma patients (Table 5). This was also found by many other studies [12, 19].

Therefore, we believe that osteosarcoma of the humerus has a slightly worse oncological outcome compared to osteosarcoma around the knee and a better outcome than spinal and pelvic osteosarcomas [7, 11]. This conclusion was also drawn in other studies, such as a South Korean study by Cho et al., who similarly identified a proximal humerus osteosarcoma as having a poorer survivorship compared to osteosarcomas at other anatomical extremities [6]. However, a review of 345 osteosarcoma cases was performed in 1975 by Campanacci et al., who observed that tumors affecting the proximal half of the femur and humerus had a poorer outcome [2].

The mean MSTS score of our patients was 25 (83%); the lowest score was 22 (73%), while the highest was 29 (96%). The functional outcome in our study was comparable to that reported by Wittig et al., who used an endoprosthesis for reconstruction and achieved an MSTS score of 80–90% [17]. Vitiello et al. also found an excellent functional outcome in a patient with proximal humerus chondrosarcoma managed by a wide resection and 3D-printed custom-made prosthesis [16].

In our study, both nonbiological reconstruction and biological reconstruction yielded the same mean MSTS score of 25; however, this comparison was not statistically significant (P = 1). This is because, whatever the reconstruction modality used, it ultimately acts as a hanger for the upper limb to preserve elbow and hand function. This is mainly due to the resection of the rotator cuff muscles (as well as the deltoid muscle in some cases). In cases with shoulder fusion, some range of motion of the shoulder is still present due to scapulothoracic movement. This overall lack of superiority of any reconstruction method over the other has been reported by several previous studies [13, 15, 18].

The functional outcome in our study was not affected by whether the resection length was less or more than 15 cm (proximal or distal to the deltoid insertion) (P = 0.444). However, in another study done to assess the functional outcome and shoulder instability of the reconstruction of proximal humerus metastases, El Motassime et al. found that patients with a resection length of > 10 cm had worse outcomes than those who had a resection length of 10 cm. They chose 10 cm as this was the minimum resection done considering the size of the smallest module of prosthesis used. However, in their study, the rotator cuff muscles and the deltoid were preserved in some cases, as they were unaffected by the tumor (metastasis); this was not the case in our study, which assessed only primary aggressive osteosarcoma [8].

The poorer functional outcome noticed in our younger patient group was also found in a study by Yao et al., who encountered constraints such as a small intramedullary canal, compliance with immobilization and poor soft-tissue coverage due to insufficient remaining adjacent tissues in this patient group [18].

Strong points in our study include a relatively large sample of exclusively proximal humerus osteosarcoma cases in a single center, with surgery and follow-up done by the same team. To the best of our knowledge, this is the highest cumulative number of cases of proximal humerus osteosarcoma currently studied. Another strong point of our study is the use of biological and nonbiological reconstruction modalities in almost all of our 34 patients, who showed similar functional outcomes to patients who had received an endoprosthesis elsewhere. The cost effectiveness without compromising the functional outcome should be considered as well in the management of proximal humerus osteosarcoma.

Our study encountered limitations that should be taken into consideration when interpreting the results, such as the combination of pediatric and adult populations, given that these groups require different reconstruction modalities, meaning different functional outcomes and different oncological outcomes. Also, being a retrospective study, it potentially allows selection bias; however, this can be justified by the rarity of the tumor at this site as well as the long follow-up required to evaluate the oncological outcome. Future systematic reviews and a meta-analysis analyzing the outcomes of proximal humeral osteosarcoma in particular are required to provide more solid data.