Epithelioid Haemangioma & Haemangioendothelioma of Bone

One lineage, three grades. The epithelioid vascular tumours of bone run from the benign (locally aggressive) epithelioid haemangioma, through the intermediate-grade epithelioid haemangioendothelioma (EHE), to the high-grade angiosarcoma. All three are endothelial, and they can look similar: epithelioid endothelial cells and lytic, often multifocal bone lesions. The grade therefore has to be established rather than assumed, because it determines treatment and prognosis.

Who gets them. These are rare tumours, seen across all ages. In the molecularly confirmed skeletal epithelioid haemangioma series of Tsuda 2021 (n=38), mean age at diagnosis was 38 years (range 4-75), with a male predominance of 25 to 13. Ten of the 38 presented with multiple lesions clustered in one anatomic region, the hallmark pattern across the whole spectrum.

Routinely misdiagnosed at referral. Only 6 of 38 of those skeletal lesions were recognised correctly by the referring institution, and most had been called other vascular tumours. That is why molecular confirmation has become central to the diagnosis.

EHE beyond bone. EHE occurs in bone, soft tissue, liver and lung. In the Shibayama series (n=62), the CAMTA1 subtype accounted for 59 cases, TFE3 for 2 and a variant WWTR1-ACTL6A for 1.

Endothelial neoplasms. Every tumour in the group expresses vascular markers (CD31, ERG, FLI1, CD34), which confirms endothelial differentiation.

EHE is a translocation sarcoma. Tanas 2011 showed that the WWTR1-CAMTA1 fusion arises from t(1;3)(p36;q25) and runs under the WWTR1 promoter, a promoter switch that inappropriately activates CAMTA1, a gene normally expressed only in brain. The fusion is reported in about 90% of EHEs; in the Tanas series it was present in virtually all EHEs tested and absent from all other vascular tumours, which makes it disease-defining. Nuclear CAMTA1 immunostaining, or break-apart FISH, exploits this diagnostically.

The minority variant. YAP1-TFE3 EHE occurs in younger patients, is more vasoformative, and is TFE3-positive.

The benign end is FOS-driven. Epithelioid haemangioma carries a FOS rearrangement or, in a subset, FOSB, especially in bone and in cellular or atypical cases, and the FOSB immunostain can be positive. Tsuda 2021 found FOS rearrangements in 74% and FOSB in 26% of skeletal epithelioid haemangiomas, genetically separating them from EHE and angiosarcoma.

The other FOSB tumour. Pseudomyogenic (epithelioid-sarcoma-like) haemangioendothelioma carries a SERPINE1-FOSB fusion and is FOSB-positive. It typically affects a young male with multifocal lesions in one limb, and is worth knowing as the other FOSB-driven vascular tumour.

Imaging for extent. MRI stages the local extent, and whole-body imaging screens for multifocal or multiorgan disease. EHE is FDG-avid on PET-CT; Chen 2025 reported SUVmax 9.2 and 5.1 in two mediastinal cases.

Biopsy is mandatory and must be planned. It comes before any definitive surgery. Histology grades the lesion, the immunophenotype (CD31/ERG/FLI1, nuclear CAMTA1) confirms it, and molecular studies secure the diagnosis: WWTR1-CAMTA1 or YAP1-TFE3 by FISH or fusion panel, and FOS/FOSB for the benign end.

The differential. It includes metastasis, which the regional multifocal pattern mimics, other vascular tumours and, at the malignant end, angiosarcoma.

Two related pages cover the ends of this differential: angiosarcoma, the high-grade lesion that must be excluded, and metastatic bone disease, the pattern these tumours are most often mistaken for. The simple vascular lesion at the far benign end, with its characteristic corduroy and polka-dot appearances, is haemangioma of bone, and in the spine vertebral haemangioma. Neither is part of the epithelioid group, but both enter the differential of a lytic vascular lesion.

Grade first, then extent. Treatment follows the grade and the extent of disease, is coordinated through a specialist sarcoma centre, and follows a planned biopsy.

Epithelioid haemangioma recurs; it does not spread. Local recurrence is the complication. In Tsuda 2021, 5 of 17 (29%) recurred after curettage or marginal excision and 5-year local recurrence-free survival was only 38%, yet no patient metastasised or died of disease.

EHE behaves variably. It is indolent in some patients and metastatic in others, and the factors that predict survival are set out below.

Angiosarcoma. High-grade, with a poor prognosis. Excluding it is the central diagnostic task.

Treatment complications follow the surgery performed, such as curettage defects or reconstruction after wide resection, rather than the tumour itself.

The wrong factor. "Behaviour ranges from indolent to metastatic" is the honest starting point, but it is not an answer, and the factor most people reach for turns out to be the wrong one. In a series of 62 EHEs with confirmed CAMTA1, TFE3 or WWTR1 alterations (Shibayama 2021), overall 5-year survival was 78.8%, and survival did not correlate with age, sex, tumour site, multifocality or multiorgan involvement. Multifocality remains a diagnostic clue but carried no prognostic weight here, so a patient with several lesions should not be counselled as though that alone were the bad sign.

The right factors. Two predicted survival on both univariate and multivariate analysis: tumour size over 30 mm and atypical histology, defined as at least two of high mitotic activity, high nuclear grade and coagulative necrosis, and present in 35.5% of cases. A three-tier system built from those two separated the cohort sharply.

The caveats. This is a single retrospective series of 62 patients with a median follow-up of only 34 months, drawn from all sites rather than bone alone, and the system has not been validated prospectively.