Clinicopathologic and Molecular Analysis of Primary Angiosarcoma of Bone – A Single Institution Experience

Abstract

Purpose:

Bone angiosarcoma (B-AS) is an exceedingly rare and aggressive vascular neoplasm, with limited therapeutic options and poor outcomes. Unlike the more prevalent clinical subsets (breast, head and neck), the pathogenesis of B-AS remains poorly defined, with no targeted therapeutic strategies available. Moreover, the often delays in diagnosis, either radiographically or pathologically, as well as the common multifocal presentation, further impacts on the feasibility of surgical management, contributing to lower survival rates. In this study we investigated the clinicopathologic and molecular characteristics of B-AS, to better define prognostic factors influencing outcomes.

Patients and Methods:

This retrospective study analyzed 22 cases of B-AS managed at a single tertiary cancer center from 1998 to 2024. Clinical and pathologic data were extracted through chart reviews and re-assessment of radiology and histology. Molecular characterization was performed in a subset using targeted next-generation sequencing (NGS).

Results:

The cohort included 14 males and 8 females (median age: 64.5 years), with tumors mostly involving femur, pelvis and spine. Twelve (55%) patients presented with disease limited to the bone, either solitary or multifocal, while 10 (45%) patients presented in addition with extraskeletal metastases at diagnosis. The skeletal distribution included 6 (27%) solitary bone lesions, with remaining 16 being multifocal (4, contiguous, 12, disseminated). A surgical procedure for the bone lesions was performed in 73% of cases, varying from intralesional curetting to limb amputation. Half of the patients received radiation and 73% chemotherapy. By molecular profiling, all tumors showed a low tumor mutational burden, with the most frequent alterations being KDR mutations and MYC amplifications. Age and chemotherapy were significantly associated with improved overall survival (OS)(p < 0.005), however, the 3-year OS was only 30%.

Conclusion:

Despite the multidisciplinary approach and orthopedic oncology expertise from a tertiary cancer center, the prognosis for B-AS remains poor. Although limited in number, the molecular profiling revealed overlapping genomic alterations with other clinical subsets of AS, having the potential of individualized patient management.

1. INTRODUCTION

Bone angiosarcoma (B-AS) is a rare malignant vascular neoplasm often characterized by multifocality and aggressive clinical course.¹ While most angiosarcomas occur in the breast and head and neck, a primary skeletal presentation remains exceptionally uncommon, accounting for less than 1% of all primary malignant bone tumors.^1,2 Due to their rare incidence, the advancements in understanding their pathogenesis and optimizing treatment strategies have been very limited over the past decades. Moreover, the frequent delay in primary diagnosis and multifocal presentation have also impacted on their surgical management and poor clinical outcome. Clinical presentation often includes localized pain, swelling, and occasional pathologic fractures, all being non-specific symptoms which may be misinterpreted as benign conditions. The clinical course of B-AS is characterized by a high rate of local recurrence and distant metastases, particularly to the lungs and other bones.^2,3 Due to its rare incidence, the molecular underpinnings of B-AS are not well defined and have lagged behind other clinical AS subsets, including MYC amplification in radiation-associated AS, KDR mutations in breast AS, and high tumor mutational burden (TMB) and u-v signature in head and neck AS.^4,5,6 Moreover, no prior studies have attempted to correlate genomic findings with clinical outcome or response to therapy. This study investigates our clinical and molecular experience with primary osseous AS, managed at a single tertiary cancer center, with available orthopedic oncology expertise, over almost a three-decade span (1998-2024). Our results highlight the clinical presentation, diagnostic complexities, molecular profiles, treatment, outcomes, and prognostic factors associated of this rare malignancy.

2. MATERIAL AND METHODS

2.1. Case Selection

The archives of the Pathology Department were searched for a diagnosis of angiosarcoma (AS) arising in the bone, diagnosed and managed at our institution over a 27-year span (1998-2024). A cohort of 22 B-AS cases was selected, in which the diagnosis was re-confirmed using modern pathologic criteria, immunohistochemistry (endothelial markers CD31, ERG)¹ and, when needed, molecular testing. In all patients a meticulous chart review was conducted to extract clinical data, therapeutic information and follow-up outcome (Table 1, STable 1). Radiologic findings and reports, H&E sections, and immunohistochemistry studies were systematically reassessed. Statistical analyses were undertaken to investigate variables influencing metastasis and overall survival (OS). Institutional Review Board approval was obtained for the purpose of this study.

Table 1. Clinicopathologic findings of the B-AS cohort.

Case	Age	Sex	Location	Tumor size	Multifocality	Mirra** classification	Extraskeletal metastasis	Follow-up	Status of   Disease
1	58	M	Pelvis, both femurs	7.7	Yes	Discontiguous	No	21	DOD
2	49	F	Humerus, scapula, skull, cervical, thoracic vertebrae	6.5	Yes	Discontiguous	No	14	AWD
3	70	M	Femur	3.5	No*	NA	No	8	AWD
4	77	M	Femur, tibia, fibula	6.9	Yes	Contiguous	Lung	11	AWD
5	47	F	Pelvis, lumbosacral, cervical spine	NA	Yes	Discontiguous	No	31	AWD
6	76	M	Femur, tibia, fibula	NA	Yes	Contiguous	Brain	3	DOD
7	75	M	Lumbar vertebrae	NA	Yes	Discontiguous	Lung, LN	1	DOD
8	51	M	Humerus, thoracic vertebra	1.6	Yes	Discontiguous	Lung, LN	121	AWD
9	74	F	Femur, pelvis, lumbosacral vertebrae	3.2	Yes	Discontiguous	Liver, spleen	2	DOD
10	72	F	Femur, pelvis	NA	Yes	Contiguous	No	17	DOD
11	80	M	Femur, pelvis, lumbar spine, appendicular skeleton	7.2	Yes	Discontiguous	No	3	AWD
12	65	M	Femur, pelvis, lumbar spine	NA	Yes	Discontiguous	No	17	DOD
13	55	F	Sacrum	7.2	No	NA	Soft tissue, LN	10	DOD
14	66	F	Pelvis	4.1	No	NA	Soft tissue	10	DOD
15	75	M	Both femur, pelvis, thoracic, lumbosacral vertebrae	NA	Yes	Discontiguous	No	17	DOD
16	33	M	Sacrum, pelvis	3	Yes	Contiguous	No	53	NED
17	44	M	Sacrum, cervical, thoracic, lumbar spine	8.1	Yes	Discontiguous	No	7	DOD
18	56	F	Radius	5.6	No	NA	Lung	16	AWD
19	64	M	Femur	3.7	No	NA	Brain	6	DOD
20	68	F	Sacrum, pelvis, both femur, foot	NA	Yes	Discontiguous	No	11	DOD
21	54	M	Thoracic, lumbosacral spine	NA	Yes	Discontiguous	No	9	AWD
22	49	M	Scapula	7	No	NA	Lung	131	NED

NA, not available; *localized; **reference#7

2.2. Clinical and Radiographic Classification

One of the common features of B-AS is the multifocal or multicentric presentation, with multiple osteolytic lesions detected radiographically in the extremities and the pelvis. Depending on the anatomic location and distribution of the skeletal involvement, the multicentric presentation has been classified as ‘solitary variant with multicentric pattern’, when involving a single bone, ‘multifocal contiguous variant’, present in multiple adjacent bones, and ‘multifocal noncontiguous variant’, defined by disseminated through multiple bones. ^7,8,9 Patients with extraskeletal spread were defined as metastatic.

2.3. Molecular Methods

A clinically validated DNA-targeted next-generation sequencing (NGS) approach using paired tumor-normal and a 505 gene-platform was applied to seven cases to identify secondary genetic alterations, employing the MSK-IMPACT assay.¹⁰ Archer FusionPlex (Archer, Boulder, CO) was performed in a single case, following the established protocol.¹¹ Cases tested for NGS were selected by the treating oncologist (requiring patient consent and blood drawn), mostly in the metastatic setting.

2.4. Statistical analysis

All statistical analyses were conducted utilizing R software (version 4.4.1). Categorical data were evaluated using Fisher exact tests or Chi-square test, and continuous data was analyzed using the Wilcoxon test or Kruskal-Wallis test. The Cox proportional hazards regression model, log-rank test, and Kaplan-Meier estimator were employed to compare survival outcomes across groups. Survival data were visualized with the ggplot2 package.¹² Statistical significance was determined based on a two-sided hypothesis test, with a threshold of p < 0.05. Mutations were visualized and summarized using the R package ‘ComplexHeatmap’version 2.8.0.¹³

3. RESULTS

3.1. Clinical and radiographic findings

The clinical features of B-AS are summarized in Table 1. The study cohort consisted of 14 males and 8 females, with an age range of 33-80 years (median, 64.5 years). None of the cases presented in the setting of syndromic settings, such as Li-Fraumeni or POT1. Using the Mirra classification⁷, none of the cases were deemed as multicentric solitary variant (limited within one bone), while 4 cases were multifocal contiguous variant and 12 discontiguous/disseminated. Twelve (55%) patients presented with disease localized to the bone, one with a solitary bone lesion and 11 with multifocal osseous lesions (2 contiguous, 9 disseminated). The remaining 10 patients (45%) presented with extraskeletal metastatic disease at diagnosis.

Among the 6 patients presenting with solitary bone involvement, the lesions were located in the femur (n=2), and one each in the sacrum, pelvis, radius, and scapula. However, all except one patient presented with extraskeletal metastatic disease at diagnosis, including lung, soft tissue, lymph node and brain. One patient developed additional bone metastasis.

A total of 16 patients presented with multicentric skeletal lesions, 5 of them also had extraskeletal disease at diagnosis. Among the 11 patients with disease localized to the bone, the 2 contiguous variant included femur/pelvis and sacrum/pelvis bones, while the remaining 9 were diagnosed with disseminated bone lesions, including contiguous and disparate osseous lesion, involving a combination of contralateral bones, extremity and spine, etc (Figs 1, Supplem Fig 1). The remaining 5 patients had both multifocal bone and extraskeletal disease at diagnosis, with disease including a wide anatomic distribution (lower extremity, pelvis, vertebral upper extremity) as well as lung, lymph node, brain and liver.

Figure 1.

Multifocal B-AS involving three different bones in the same region around the knee joint (medial femoral condyle, fibular head, and proximal tibial metaphysis) in a 76-year-old male (case #6). A) AP and B) lateral knee radiographs demonstrate multiple aggressive appearing, purely lytic lesions (horizontal arrows) with wide zone of transition and marked cortical destruction, and minimal if any periosteal reaction. Incidental popliteal hematoma on lateral radiograph (vertical arrow). C) Sagittal T1-weighted MRI image of the right knee demonstrates multiple low T1 signal lesions in the proximal tibia (white arrows). Incidental high T1 signal mass in the popliteal fossa hematoma (yellow arrow). D) Axial T2 FS image through the dominant proximal tibial lesion demonstrates an extraosseous soft tissue component (horizontal white arrow), reactive bone marrow edema adjacent to the lesion (vertical arrow) and reactive soft tissue edema adjacent to the lesion (horizontal yellow arrow). E) Axial T1 FS and F) Axial T1 FS with contrast demonstrate the lytic lesion in the medial femoral condyle. The lesion contains a large hemorrhagic component which is bright on both T1FS pre (E) and post contrast (F) images without enhancement (yellow arrows). The smaller viable component of the tumor (white arrows in E and F) demonstrates enhancement on post-contrast image.

The greatest tumor dimension of the bone lesions from radiology imaging showed a median size of 6.1 cm (range 1.6-8.1 cm).

3.3. Radiologic findings

The radiologic review revealed lytic, destructive bone lesions with soft tissue extension, occasionally resulting in pathologic fractures. Metastatic lesions were diagnosed with CT, MRI, PET/CT, and/or bone scan. On radiographs, the tumors were usually purely lytic, with sometimes associated “honeycomb” appearance. Lesions had mostly poorly defined margins or permeative appearance, but occasional cases with well-defined margins were also noted. Among the multifocal presentation, the femur was the bone mostly involved, while the multifocal contiguous lesions were often seen on opposites sides of a single joint (most often the knee or ankle joint) (Fig. 1, Supplem Fig 1). On both radiographs and CT scans the tumors were associated with endosteal scalloping, cortical destruction and extraosseous soft tissue extension, but unlike most other primary bone sarcomas, periosteal reaction was uncommon (Fig. 1, Supplem Fig 1). The lesions avidly enhanced on contrast CT. On MRI, the lesions showed an intermediate or low signal on T1 weighted images (WI), while the T1 signal was high when hemorrhagic areas were present. Tumors showed a high signal or hyperintense on T2 WI, with occasional flow voids (due to prominent vascularity) or fluid-fluid levels. Contrast enhancement was variable but more often non-homogeneous depending on the presence of hemorrhage and necrosis. (Fig. 1, Supplem Fig 1)

3.2. Pathologic findings

Matching the radiographic findings, the gross resection specimens showed hemorrhagic and destructive bone lesions involving one or multiple adjacent bones (Fig. 2, Supplem Fig 2). The majority of cases showed a solid growth pattern and displayed epithelioid morphology. Tumors also displayed variable vasoformative features, typically set in confluent pools of blood. The vascular channels were composed of capillary or small sized blood vessels, often with inter-anastomozing channels, lined by highly atypical endothelial cells (Fig 3). The lesional cells showed eosinophilic cytoplasm and enlarged nuclei with diffuse hyperchromasia and marked degree of nuclear pleomorphism (Fig 3). An increased mitotic activity and coagulative type necrosis were seen in all cases (Fig 3). In the initial diagnostic biopsies, not uncommonly, scant lesional tissue was present, with either extensive hemorrhage or well-formed vascular channels (Supplem Fig 3).

Figure 2. Gross specimen of a left hip disarticulation (case#6).

Multiple foci of angiosarcoma were noted including femoral diaphysis (12.0 cm), tibia (4.0 cm) and distal thigh/gluteal area. In addition, a large hematoma was noted in the popliteal area (as described above).

Figure 3. Pathologic spectrum of angiosarcoma of bone.

(A) The tumor composed of large epithelioid cells with ovoid vesicular nuclei and prominent nuclei. (case#13, 200x, H&E). (B) Pleomorphic epithelioid cells with abundant amphophilic to basophilic cytoplasm, multinucleated nucleus arranged in solid pattern (case#15, 200x, H&E). (C) Focal area demonstrates vasoformative pattern (case#16, 200x, H&E). (D) coagulative necrosis (case#12, 100x, H&E)

3.3. Surgical procedures and chemoradiation management

Thirteen patients (59%) were diagnosed by an image-guided core biopsy, 6 cases (27%) with open or incisional biopsy, and 3 cases (14%) by intra-operative curetting.

All patients except for one (case #3) presented with either multicentric bone lesions and/or with extraskeletal metastatic disease at diagnosis. The only patient presenting with localized bone disease (case #3), was a 70 year-old male with a right distal femur lesion, who was treated with curetting and retrograde nailing, followed by adjuvant radiation. Among the 11 patients presenting with multicentric bone lesions but no extraskeletal metastases, a surgical procedure was applied in 7 patients (32%): 4 being treated with an en-bloc resection (hip, 3, humerus,1), 2 had fixation and currenting of femur, and 1 was managed with a cervical stabilization due to instability.

Of the total 10 patients presenting with concurrent bone lesions and extraskeletal metastases, 5 had solitary and 5 multicentric skeletal lesions. Among the 5 patients with solitary bone disease, 3 patients were managed surgically, 2 with an en-bloc resection of the radius and scapula, respectively, and one patient with marginal resection of the mass and plate fixation for a femoral lesion. All patients with multifocal bone lesions were managed surgically: 3 with en-bloc resection (above-the-knee amputation, hip disarticulation, and femur resection), while 2 patients received curetting with fixation for the humerus and kyphoplasty for the lumbar vertebra.

Among the 10 patients presenting with disseminated disease outside the bone, 3 underwent surgical procedures for extraskeletal metastases, one each for soft tissue, lung, and brain (metastasectomy).

Among the 6 patients presenting with solitary bone tumors, 3 received radiation (RT), one as adjuvant for femur the lesion, and two palliative (sacrum and pelvis). Of the 16 patients with multifocal osseous lesions, 9 had palliative RT for one or more bone lesions. Two additional patients had RT for brain metastases.

Chemotherapy was administered in 16 (73%) patients, all for metastatic lesions. There were various chemotherapy regimens applied first line, including doxorubicin (6 cases, one case in combination with oralatumab), adriamycin and ifosfamide (4 cases), gemcitabine and docetaxel (3 cases, one with bevacizumab), docetaxel (2 cases), and one each with paclitaxel, sorafenib and pembrolizumab. A subset of patients (n=13, 59%) had more than 2^nd lines of therapies, including ifosfamide and etoposide, ipilimumab and nivolumab, nivolumab and larotrectinib.

Clinical follow-up and outcome

Follow-up data was available in all patients with duration ranging from 1-131 months (median, 11). At last follow-up, 12 (55%) patients were dead of disease (DOD), 8 (36%) patients being alive with disease (AWD), and 2 (9%) showed no evidence of disease (NED). Among the 12 patients who died of disease, all presented with either multifocal bone lesions (n=9) and/or extraskeletal metastasis (n=6) at diagnosis. Within this group, 8 patients were managed surgically for the osseous lesions with an en-bloc resection (n=6), and one each with fixation and currenting or kyphoplasty. For the extraskeletal metastases, a metastasectomy was performed in 2 patients (soft tissue and brain).

Among the 8 patients AWD at last follow-up, only one patient (case #3) presented with a localized solitary femoral bone lesion which was managed with curetting and fixation. One additional patient had a solitary bone lesion but presented with concurrent extraskeletal metastases and was treated with an en-bloc resection for the primary radius lesion. The remaining 6 patients presented with multicentric bone lesions, with 3 of them showing synchronous extraskeletal metastatic disease. All except one patient were managed surgically for the bone lesions, including en-bloc resection (n=2), and one each with above-knee amputation, fixation and curreting (n=2), and one cervical stabilization. The only non-surgical candidate (case #21) presented with disseminated bone lesions (involving thoracic and lumbosacral spine) and was treated with chemotherapy alone.

Two patients remained NED. The first patient (case #16) was as 33-year-old male with multifocal contiguous lesions in the sacrum and pelvis, and inguinal lymph node metastasis. After starting weekly paclitaxel chemotherapy, he received radiation therapy for the pelvic metastatic lesions (2700cGy/3fr), followed by excision of the inguinal lymph node metastasis. He then continued chemotherapy with the same regimen over four years and follow-up imaging showed no evidence of disease. The second patient (case #22), a 49-year-old male presented with a solitary radiation-associated AS in the scapula (prior history of radiation for Hodgkin’s lymphoma) and a lung metastasis. He received three-cycle of neoadjuvant chemotherapy (adriamycin, ifosfamide, and mesna) with an excellent radiologic response. He then underwent radical resection of the right scapula and glenoid with negative margins, showing profound treatment response in the form of extensive hyalinization and necrosis (95% of tumor). He then received three-cycles of adjuvant chemotherapy (ifosfamide and liposomal doxorubicin). Subsequently, he received Muramyl tripeptide phosphatidylethanolamine (MTP-PE) as an immune activating agent. The patient remains free of disease after 131 months follow-up.

3.4. Molecular findings

3.4.1. DNA-Targeted NGS (MSK-IMPACT)

MSK IMPACT assay was performed in 7 cases with adequate material (Fig. 4). All cases exhibited a low tumor mutational burden (TMB), ranging from 0.8-4.4 mutations/Mb. Only one case (case #12) demonstrated a high fraction of altered genome (FGA) of 0.2, while the others < 0.1. Two cases demonstrated KDR missense mutation (case #11, #19 – g.55984867G>C (L88V), g.55968612A>G (I684T). One case (case # 16) showed MYC amplification (8q24.21), which was co-amplified with AGO2 (8q24.3). One case had a TP53 in-frame deletion with a concurrent POT1 missense alteration. Two harbored LATS2 (13q12.11) alterations, one being missense mutation (case #16) and the other a deep deletion (case #12).

Figure 4. Oncoprint summarizing the main genomic alterations identified by MSK-IMPACT.

3.5. Statistical analysis

The analysis evaluated various factors influencing overall survival (OS), such as age, gender, stage at primary diagnosis, multifocality, size, location, diagnostic procedure, definitive procedure, chemotherapy, and radiation therapy, see Supplem Table 1. OS rates were 60% and 30% at 1 and 3 years, respectively. The patient’s age and chemotherapy showed a significant association with overall survival (p < 0.005) (Fig 5). There was no association between OS with gender, stage at primary diagnosis, multifocality, size, location, diagnostic procedure, definitive procedure, and radiation therapy. Similarly, there was no correlation between the presence of extraskeletal metastasis with other factors. Moreover, there was no difference in survival between cases with solitary versus multicentric bone lesions (contiguous or discontiguous) (p = 0.89), nor between localized bone disease (solitary or multifocal) and those with bone plus extraskeletal metastases (p = 0.33) (SFig. 4).

Figure 5.

Kaplan-Meier curve showing improved overall survival in patients treated with chemotherapy.

4. DISCUSSION

This study investigates the clinicopathologic features and treatment outcomes in a cohort of B-AS managed at a single tertiary cancer center with orthopedic oncology expertise. Most previous studies have documented the clinicopathologic characteristics of B-AS, in the settings of multi-institutional studies.^2,3 Pain and soft tissue swelling represent the most common symptoms at presentation, however, due to their non-specific nature have limited impact in establishing the diagnosis.¹⁴ An important aspect of B-AS pathogenesis is the predilection for multifocal osseous involvement, often seen even in the absence of other extraskeletal metastases.^2,3 This phenomenon is similar to the multicentric presentation of other vascular bone lesions, at the benign or low grade end of the spectrum, including epithelioid hemangioma, pseudomyogenic hemangioendothelioma, etc.^15,16,17 This unique feature, known as angiotropic spread, occurs via metaphyseal veins, with either antegrade or retrograde pattern, without evidence of systemic metastases. The importance of distinguishing multicentric bone involvement versus extraskeletal spread stems from the old literature documenting a more indolent course with a favorable prognosis in multifocal B-AS compared to solitary bone lesions, that are more often associated with distant metastases to the lungs.⁹ This debate in our opinion remains unresolved, as many of the so-called ‘low grade angiosarcomas’ in the old studies may not represent true angiosarcomas, as the molecular hallmarks of epithelioid hemangioma and epithelioid hemangioendothelioma were not yet defined. Radiographically, the characteristic pattern described is that of clustering of skeletal lesions within a single limb or anatomic region, such as the shoulder or hip. In the current study, 16/22 (73%) patients presented with multifocal osseous lesions. This incidence is significantly higher (p = 0.01) compared to the 40% multifocality detected in a large study of malignant vascular bone tumors (n=63).¹⁷ However, this latter cohort included both hemangioendothelioma and B-AS, and 22/63 cases were based only on radiograph images.¹⁷ Our results show that 55% of patients presented with disease localized to bone, while remaining had in addition extraskeletal metastases. Localized, solitary disease at diagnosis is extremely rare in this setting, with only one patient presenting with a single bone lesion (femur) in this series, and all the remaining patients with a solitary bone lesion were diagnosed with concurrent extraosseous metastatic disease. Across the multifocal skeletal involvement, none were limited to a single bone, instead being either contiguous or disseminated to multiple bones. Despite this high prevalence of multifocal disease at presentation, a surgical procedure was applied in a majority of patients (73%), and half received radiation. None of the prior studies stratified the management or survival based on the solitary, contiguous, and discontiguous skeletal presentation, and grouped the multifocal bone involvement as part of the distant metastases together with non-osseous spread.^2,3 Thus, in one study 35/80 (44%) patients were deemed as metastatic at presentation, including 8 (23%) patients with multifocal bone lesions.² The 5-years OS of this metastatic cohort was 8% (95% CI 0–20) and the 1-year OS was 22% (95% CI 8–37).² In another study, 24/60 (40%) patients presented with metastatic disease, which included half with multivisceral metastases, and 8 (33%) with only multifocal bone involvement, with none being alive at 5-years.³ These series corroborated with ours reveal a highly dismal outcome of multifocal B-AS.

The advanced disease at presentation also relates to the non-specific clinical symptoms and radiographic findings. In the presence of multifocal lesions, either isolated within one bone or multiple sites, the radiologic differential diagnosis typically includes the more common entities such metastatic disease and multiple myeloma. B-AS may be suspected when the multifocal lytic lesions are confined to one region.^18,14,17,19 Moreover, lesions lacking cortical involvement may be radiographically occult, such lesions being better evaluated on CT or MRI.¹⁷

As the pathologic and molecular features of B-AS have evolved over the years, results from older clinical series are often compromised by the inadvertent inclusion of benign or low-grade vascular tumors.^15,20 Diagnostic challenges are not infrequent, either due to limited biopsy material, extensive hemorrhage, and necrosis, with scant lesional cells, or due to morphologic overlap with other epithelioid vascular neoplasms. The genetic alterations of these lesions have been only recently defined, with more than half of epithelioid hemangioma of bone harboring FOS or FOSB gene rearrangements²¹, while the hallmark of epithelioid hemangioendothelioma is either the WWTR1::CAMTA1 or YAP1::TFE3 fusion.^22-24 Microscopically, B-AS are composed of overtly malignant, often epithelioid, endothelial cells either arranged in solid sheets or showing variable degrees of vascular channel formation. Accordingly, the cases in this series showed high-grade cytologic atypia, brisk mitotic activity and necrosis.

Prior molecular analyses in both bone and soft tissue AS have shown common amplifications of 2q and 17q, suggesting a potential pathogenic overlap irrespective of their location.⁵ In one of the largest angiosarcoma genomic study to date, encompassing 39 cases, tested either by whole genome sequencing or by angiogenesis gene screen, there were 8 B-AS included.²⁵ Half of the B-AS showed no detectable alterations, while the other half showed PTPRB loss of function mutations (n=3) and MYC gene amplifications as recurrent gene alterations.²⁵ In keeping with these observations, our results show shared genomic alterations with radiation-associated AS (MYC amplification) and primary breast AS (KDR mutations).⁴

In summary, our findings underscore the highly aggressive behavior of B-AS and the therapeutic challenges faced in the setting of advanced stage and multifocal bone disease at diagnosis. Clinical presentation with solitary or multifocal lesions limited to one bone is extremely rare, highlighting the early disseminated stage of this disease. Despite a multidisciplinary approach, including surgery and chemoradiation modalities, applied in a tertiary cancer center, the survival remained dismal. Large, multi-institutional genomic studies are needed to further expand our understanding of the genomic landscape of this disease in order to guide targeted therapeutic strategies.

Supplementary Material

Suppl. Fig. S1 & S2

Supplementary Figure 1. 73-year-old female with loco-regional multifocal B-AS involving right sacrum, bilateral pelvic bones and throughout the right femur, including both sides of the right sacroiliac joint and both sides of the right hip joint. A), B), C), D) Images from PET/CT scan. CT images demonstrate multiple lytic lesions: A) in the bilateral posterior iliac bones (horizontal arrows) and the right sacrum (vertical arrows), including right ilium cortical destruction without associated periosteal reaction. B) in the anterior and posterior columns of the acetabulum (arrows). Fused PET/CT images C) and D) corresponding to CT images from A) and B) demonstrate marked FDG uptake in the lytic bone lesions, with standardized uptake values (SUV) between 17 and 19, consistent with hypermetabolic tumor. E) Coronal T2 weighted FS image of the inferior pelvis and proximal right femur demonstrate a few high signal lesions in the acetabulum (yellow arrow) and the proximal femur (white arrows) consistent with multifocal angiosarcoma. F) Additional lesions are located in the mid and distal femoral diaphysis on coronal T2 weighted FS image (arrows). The two inferior lesions are in the medullary cavity, while the superior-most lesion (short white arrow) involves the cortex only. Reactive soft tissue edema (yellow arrow) is present adjacent to the larger lesions.

Supplementary Figure 2. Gross specimen of a right femoral resection for an 8 cm lesion within the proximal femur, associated with pathologic fracture (case #10). An additional lesion (4.0 cm) was noted in the diaphysis. Additional curating of the acetabulum was also positive for epithelioid angiosarcoma.

Suppl. Fig. S3

Supplementary Figure 3. Pathologic features and immunohistochemistry of biopsy specimen. (A) The biopsy specimen showed well-formed, anastomosing vascular spaces lined by hyperchromatic atypical endothelial cells. (case#14, 200x, H&E). (B) Hyperchromatic epithelioid cells arranged in solid pattern (case#15, 200x, H&E). (C) Immunostaining showed positive results for ERG (case#15, 200x) and (D) CD31 (case#15, 200x).

Suppl Fig. S4

Supplementary Figure 4. Kaplan-Meier curve showing no difference in survival between cases with solitary versus multicentric bone lesions (contiguous or discontiguous) (p = 0.89).

Suppl. Table S1

Supplementary Table 1. Treatment modalities applied for the B-AS cohort.

Data Availability Statement:

The data that support the findings of this study are available from the corresponding author upon reasonable request.