Review

Angiosarcoma
Robin J Young, Nicola J Brown, Malcolm W Reed, David Hughes, Penella J Woll

Angiosarcomas are rare soft-tissue sarcomas of endothelial cell origin that have a poor prognosis. They can arise anywhere in the body, most commonly presenting as cutaneous disease in elderly white men, involving the head and neck and particularly the scalp. They can be caused by therapeutic radiation or chronic lymphoedema and hence secondary breast angiosarcomas are an important subgroup. Recent work has sought to establish the molecular biology of angiosarcomas and identify specic targets for treatment. Interest is now focused on trials of vasculartargeted drugs, which are showing promise in the control of angiosarcomas. In this review we discuss angiosarcoma and its current management, with a focus on clinical trials investigating the treatment of advanced disease.

Lancet Oncol 2010; 11: 98391 Published Online May 26, 2010 DOI:10.1016/S14702045(10)70023-1 Academic Unit of Surgical Oncology, (R J Young MRCP, Prof N J Brown PhD, Prof M W Reed FRCS); Academic Unit of Clinical Oncology (Prof P J Woll FRCP), School of Medicine and Biomedical Sciences, University of Sheeld, and Department of Histopathology, Sheeld Teaching Hospitals NHS Foundation Trust, Sheeld, UK (D Hughes FRCPath) Correspondence to: Dr Robin J Young, Academic Unit of Surgical Oncology, Department of Oncology, K Floor, School of Medicine and Biomedical Sciences, Sheeld, S10 2RX, UK r.j.young@sheeld.ac.uk

Introduction
Angiosarcomas are a subtype of soft-tissue sarcoma and are aggressive, malignant endothelial-cell tumours of vascular or lymphatic origin (gure 1). Treatment is challenging in many cases and the prognosis is poor. Although the development of several vascular-targeted therapies has stimulated interest in identifying the molecular pathogenesis of angiosarcomas and clinical testing of these molecules. In this review we aim to give an overview of angiosarcoma and its current management, with a focus on clinical trials investigating the treatment of advanced disease. Much of the published data on these rare tumours are based on case series, which causes difficulty in interpret ation of results, because reporting may be selective, datasets incomplete, and treatment approaches diverse, even within the same institution. There have been no phase 3 and few phase 2 treatment trials in angio sarcomas. Furthermore, angiosarcomas are commonly included as a handful of cases within wider studies on soft-tissue sarcomas. As histological and anatomical classications are replaced by molecular classications, guidelines for diagnosis and treatment will likely be claried.

between sexes, can develop at any age, and are more common in older patients; with the cutaneous form most common in elderly white men.3,5,6 Angiosarcoma can arise in any soft-tissue structure or viscera (table 1),3,69 and cutaneous angiosarcomas typically involve the head and neck, particularly the scalp. Angio sarcomas are of endothelial cell origin and tumours arising directly from major blood vessels or the heart are rare.

Aetiology
Most angiosarcomas arise spontaneously, but there are a few reports of malignant transformation within pre-existing benign vascular lesions.10 Several well described risk factors exist (panel 1). Chronic lymphoedema of any origin is associated with the development of angiosarcoma; a phenomenon known as Stewart-Treves syndrome. Lymphoedema is one causal aetiological factor in the development of breast angiosarcomas after treatment for breast cancer.11 Lymphoedema caused by Milroys disease and chronic infections, such as lariasis,1 have likewise been linked to the development of angiosarcomas. Radiotherapy is an independent risk factor. Although the association between radiotherapy and subsequent angiosarcoma is best described for breast cancer therapy, it is not exclusive to breast lesions. A large epidemiological study of breast cancer patients using Surveillance of Epidemiology and End Results (SEER) data showed an increased risk of soft-tissue sarcomas (particularly angiosarcomas) after adjuvant radiotherapy, with a peak incidence 510 years after treatment.12 However, this link is disputed, and the heightened risk might result from concurrent lymphoedema.13,14 Interestingly, some case reports suggest that mutations
A B

Classication
Angiosarcomas are subdivided into cutaneous angio sarcoma, lymphoedema-associated angiosarcoma, radiation-induced angiosarcoma, primary-breast angio sarcoma, and soft-tissue angiosarcoma, and most reports include several angiosarcoma subtypes.1 There is some evidence that tumour behaviour might depend on site of origin,2,3 although whether differences between cutaneous, radiation induced, breast, and visceral angiosarcomas are caused by biological differences or differences in clinical presentation and treatment is unclear.

Epidemiology
About 2% of soft tissue sarcomas and 54% of cutaneous soft tissue sarcomas are angiosarcomas.4,5 The incidence of angiosarcoma has risen over the past 30 years, but whether this is a true increase is unclear. The rise could be related to greater use of radiotherapy, improved medical awareness, or a rened histopathological denition. Angio sarcomas have a similar distribution
www.thelancet.com/oncology Vol 11 October 2010

Figure 1: Angiosarcoma arising 5 years after receiving surgery and radiotherapy for breast cancer By courtesy of Sheeld Teaching Hospitals NHS Foundation Trust.

983

Review

Number (%) Head and neck Breast Extremities Trunk Liver Heart Bone Spleen Other or unknown 144 (270%) 105 (197%) 82 (153%) 51 (95%) 32 (60%) 25 (47%) 19 (36%) 14 (26%) 62 (116%)

Panel 1: Risk factors for angiosarcoma Radiation Chronic lymphoedema (Stewart-Treves syndrome) Postsurgery or radiotherapy Milroys syndrome Other types of chronic lymphoedema Exogenous toxins Vinyl chloride Thorium dioxide Arsenic Anabolic steroids Foreign bodies Familial syndromes Neurofibromatosis NF-1 Mutated BRCA1 or BRCA2 Maffucci syndrome Klippel-Trenaunay syndrome

Table 1: Distribution of angiosarcoma with pooled data from 534 patients3,69

in the DNA repair genes, BRCA1 and BRCA2, predispose to angio sarcomas after treatment for breast cancer.15 Familial syndromes including neurobromatosis, Maffucci syndrome, and Klippel-Trenaunay syndrome are also associated with angiosarcoma.1 Various chemicals are associated with the development of angiosarcomas, particularly within the liver. Occupational exposure to vinyl chloride is a well documented cause of hepatic angiosarcoma,16 and thorium dioxide, previously used in the 193040s in radiology, is also associated with hepatic angiosarcomas.17 Thorium dioxide has a long halflife and accumulates in the liver, spleen, and bone resulting in increased risk of hepatocellular carcinomas, angio sarcomas, and leukaemia. Other chemical carcinogens associated with angiosarcomas include arsenic, radium, and anabolic steroids.18 Reports of angiosarcoma associated with foreign bodies include accidentally retained surgical gauzes,19 vascular20 and orthopaedic prostheses,21 and gouty tophus.22 The contribution of immunosuppression to the patho genesis of angiosarcoma is uncertain. Angio sarcomas have been reported in immunosuppressed patients following renal transplant. Ahmed and Hamacher23 reported that ve (42%) of 12 patients developed angiosarcomas at the arteriovenous-stula site. The association of lymphoedema with subsequent angio sarcoma might be caused by localised immuno deciency and some epidemiological studies suggesting an association between angiosarcoma and AIDS.24 However, this hypothesis has not been conrmed, and the small number of AIDS-related angiosarcoma cases may be misdiagnosed Kaposis sarcoma. Although there is evidence for a role of human-herpes-virus 8 in the pathogenesis of Kaposis sarcoma, there is no evidence for a similar role in angiosarcoma.25

fungation, ulceration, and haemorrhage can develop.1 Tissue inltration and multifocal disease can result in contaminated resection margins despite radical surgery.7,9,26 Deeper soft tissue and visceral lesions present as an expanding mass associated with pain or discomfort (gure 2). Untreated, angiosarcoma lesions can reach sizes of 20 cm or more. Angiosarcomas principally spread haemato genously, with the lungs the most common site for metastases, and may present as pleural disease, haemorrhagic pleural effusion, or pneumothorax. Other common sites include liver, bone, soft-tissue structures, and lymph nodes.3,68,26,27

Dierential diagnosis
Angiosarcomas are included within the broad category of vascular tumours (panel 2).1 Clinical history may provide clues but a precise diagnosis requires expert histological assessment. Capillary haemangiomas typically develop during infancy and present as solitary, well-circumscribed protuberances. After radiation, atypical vascular lesions arise in areas of skin contained within treatment elds. Most of these lesions develop within 3 years of radiotherapy and, while benign, could predict an increased risk of angiosarcoma. Classic Kaposis sarcoma presents in elderly men, commonly of Mediterranean origin, with multiple cutaneous lesions involving the distal legs. AIDS-related Kaposis sarcoma initially causes small pink patches before developing into the classic blue-red plaques and nodules on the skin and mucous membranes. Gastrointestinal, splenic, and lung lesions can develop. Epithelioid haemangioendotheliomas present in middle age as solitary, painful, soft-tissue lesions, and can metastasise via the lymphatic system to the liver, lung, or bone. These tumours are less aggressive than conventional high-grade angiosarcomas, and excision with lymph-node dissection could allow long-term control. Haemangiopericytomas are
www.thelancet.com/oncology Vol 11 October 2010

Clinical presentation
Cutaneous angiosarcoma can initially resemble a bruise, or a raised purplish-red papule, is typically multifocal (gure 1) and can be mistaken for a simple benign lesion leading to delayed presentation and diagnosis. With increasing tumour size, tissue inltration, oedema, tumour
984

Review

part of a group of solitary brous tumours and generally have a good prognosis. They often present as deep softtissue masses of the legs and might be associated with hypoglycaemia caused by tumour secretion of insulin-like growth factor.

Pathology
The histological features of angiosarcoma can vary both within and between cases. Morphological differences can be subtle, and distinguishing a malignant vascular tumour from a benign proliferative or inammatory lesion with light microscopy can be difficult. Angiosarcomas are inltrative and do not have a capsule or a clear border separating normal from abnormal tissue. Abnormal, pleomorphic, malignant endothelial cells are the hallmark of angiosarcoma and can be rounded, polygonal, or fusiform and can have an epithelioid appearance. In well differentiated areas, abnormal endothelial cells form functioning vascular sinusoids continuous with normal vascular channels. These vascular sinusoids dissect between collagen bundles and are often associated with areas of monocyte inltration (gure 3). With increasingly aggressive disease, the architecture becomes more chaotic, with less clearly dened vascular spaces. Rather than a single endothelial-cell lining, the abnormal cells become multilayered and form papillarylike projections into the vascular lumen. Mitotic bodies are common, as are small clusters of erythrocytes within the cytoplasm of the abnormal endothelial cells. In poorly differentiated areas the malignant endothelial cells form continuous sheets, usually with an epithelioid morphology, and with areas of haemorrhage and necrosis, which can make differentiation from anaplastic carcinoma or melanoma difficult.1,28,29 Angiosarcomas typically express endothelial markers including von Willebrand factor, CD34, CD31, Ulex europaeus agglutinin 1, and vascular endothelial growth factor (VEGF). Immunohistochemistry is therefore important in conrming the diagnosis. Von Willebrand factor, U europaeus agglutinin 1, and CD31 are the most useful markers in poorly differentiated cases.30 However, progressive tumour dedifferentiation can lead to a loss of these markers. The absence of melanocytic markers (S100), human melanoma black-45, and melanoma anti gen can help distinguish angiosarcoma from melanoma.30 Aditionally, epithelioid angiosarcomas can express cytokeratins, leading to confusion with poorly differentiated carcinomas.

Figure 2: Hepatic angiosarcoma CT scan showing multifocal tumour in the right lobe of liver, with metastases throughout the remaining liver.

Panel 2: Classication of vascular tumours Reactive and benign vascular tumours Capillary haemangiomas Juvenille haemangioma (strawberry naevus) Cherry angioma (Campbell de Morgan spot) Pyogenic granuloma Cavernous haemangiomas Epithelioid haemangioma Vascular ectasis (naevus flammus, spider naevus) Angiomatosis Postradiation atypical vascular lesion Intermediate grade vascular tumours Kaposis sarcoma Epithelioid haemangioendothelioma Malignant vascular tumours Angiosarcoma Tumours of perviscular cells Haemangiopericytoma (solitary fibrous tumour)

Molecular biology
No comprehensive studies of molecular changes in angiosarcoma have been published. Reports typically explore the expression of one or two markers in a small series of tumours with limited correlation to clinical or pathological features, and are insufficient to compare subgroups such as cutaneous, visceral, or radiationinduced versus de-novo disease. Gene-expression
www.thelancet.com/oncology Vol 11 October 2010

microarray technology should help to identify unique molecular signatures for histological subtypes. This will improve the understanding of key molecular events in the pathogenesis of soft-tissue tumours. This technology may also lead to improved diagnosis, particularly in early or poorly differentiated cases, and suggest potential therapeutic targets. Because angiosarcomas are endothelial-cell tumours, there is great interest in the role of angiogenesis and the angiogenic factors associated with their pathogenesis, and how they might be used as targets for treatment. VEGF and its receptors can be overexpressed in angiosarcomas (table 2); however, there are few data on the expression of other angiogenic cytokines such as
985

Review

Figure 3: Haematoxylin-eosin stained sections of angiosarcoma Neoplastic vascular structures dissecting between collagen bundles (A; magnification 200x); varied shapes of the large malignant endothelial cells with the normal capillary vessel below (B; magnification 400x).

angiopoietins. VEGF-A is the most studied cytokine and is consistently expressed at higher concentrations in angiosarcomas than in benign vascular or normal-tissue controls.32,33 All three subtypes of VEGF receptors (VEGFR) have been detected in angiosarcomas. Loss of VEGFR-2 expression, but not VEGFR-1 or VEGFR-3, was correlated with worse prognosis in one series.31 VEGFR-2 is the main receptor mediating VEGF-A proangiogenic signalling, which suggests that in angiosarcoma VEGFR-2 expression is acting as another marker of endothelial differentiation, rather than being pathogenic. As in other tumours, abnormalities in TP53 are common in angiosarcomas.32,34 Other factors associated with aberrant growth pathways have been identied, including overexpression of Wilms tumour-1 and Galectin-3.35,36 Activation of the RAS pathway is often tumorigenic. Although immortalised murine endothelial cells form benign haemangiomas in vivo, the addition of activated HRAS produces rapidly growing, poorly differentiated angiosarcomas.37 In clinical cases, KRAS mutations have been reported in liver and cardiac angiosarcomas.3840 Downstream from RAS, experimental evidence suggests that PI3-kinase signalling is more important than the MAPK signalling cascade.37,41 However, overexpression of AKT1, a crucical component of the PI3-kinase pathway, in an immortalised murine endothelial-cell line resulted in benign vascular malformations, rather than angiosarcomas.42 Consistent with the invasive and metastatic potential of angiosarcoma, overexpression of the transcription factor ETS1 might result in increased production of metallo proteinases (particularly MMP1, MMP3, and uPA), increased extracellular proteolysis, and decreased amounts of the major basement membrane components.43,44 Furthermore, the actin-bundling motility protein, fascin, is uniformly overexpressed in angiosarcomas.33 Angio sarcomas recruit inammatory cells, particularly mast cells, that drive positive feedback loops by causing further release of cytokine growth signals, including basic broblast growth factor (bFGF) and VEGF.45 Because of the success of KITtyrosine-kinase inhibitors in treating some other sarcomas, investigators have studied KIT expression in various forms of
986

angiosarcoma. Results pooled from three separate series showed KIT positivity in 38 (50%) of 76 samples, and diffuse KIT expression in 18 (47%) of 38 samples.4648 KIT-mutation analysis has been done for a handful of angiosarcomas; however, no activating mutations in exons 11 (juxtamembrane domain) or 17 (tyrosine kinase domain) were identied. None of the molecular abnormalities reported are unique to angiosarcomas, and similar abnormalities have been identied in both carcinomas and sarcomas. Although angiosarcomas are tumours of malignant endothelial cells, increased expression of angiogenic cytokines, including VEGF and its receptors, bFGF, platelet-derived growth factor (PDGF), and the angiopoietin system, have been reported in other cancers including soft-tissue sarcomas.49

Chromosomal abnormalities
Cytogenetic analyses of angiosarcomas have been published for nine patients.50 Most of the samples were derived from metastases, poorly differentiated angio sarcomas, or radiation-induced tumours and a wide range of disparate chromosomal abnormalities were identied, none of which are specic to angiosarcoma. Common abnormalities included trisomy 5, deletions on the short arm of chromosome 7, varied abnormalities on chromosomes 8, 20, and 22, and loss of chromosome Y.

Staging
Soft-tissue sarcomas are staged using the International Union Against Cancer and American Joint Committee on Cancer (UICC/AJCC) system.1 This is based on the TNM (tumournodemetastatsis) staging system (panel 3),51 with an additional notation for histological grade. Angiosarcomas are high-grade tumours by denition, so histological grading is not used in staging.29 Although many patients with angiosarcoma (5080%) present with localised disease,79 some (2045%) have metastatic disease at presentation.79 However, reliable data are hard to obtain, because much information on angiosarcoma is derived from retrospective surveys of archived material, with all the inherent difficulties associated with reporting on incomplete datasets.

Prognostic factors
Primary soft-tissue sarcomas are associated with a 5 year survival of 5060%,52 but angiosarcomas have an overall 5 year survival of about 35%.3,6,9 Even with localised disease, the most optimistic survey suggests only 60% of patients survive for more than 5 years, with a median survival of 7 months.7 However, there are reports of some long-term survivors with metastatic disease,9 but which prognostic factors are important is unclear. Consistent with data for other soft-tissue sarcomas, patients with large tumours (>5 cm) do less well,26,27,53,54 although not in all.7 Some researchers6,7 suggest that high tumour grade is associated with a poor prognosis, others do not.26 Other factors that might
www.thelancet.com/oncology Vol 11 October 2010

Review

Study population Itakura et al31 2008 34 angiosarcoma

Number of angiogenic markers in patients 32 (94%) VEGF-A positive 4 (12%) VEGF-C positive 32 (94%) VEGFR-1 positive 22 (65%) VEGFR-2 positive 27 (79%) VEGFR-3 positive All (100%) with angiosarcoma and none with haemangioma (juvenile haemangioma) VEGF-A positive 18 (95%) with angiosarcoma, 1 (10%) with benign lesion (juvenile haemangioma)VEGF-A positive

Additional notes VEGF-A expression correlated with VEGFR-1 expression; VEGFR-2 correlated with prognosisloss of expression significantly associated with an unfavourable prognosis; VEGFR-1 and VEGFR-3 expression were not significantly associated with prognosis ..

Dim et al33 2007

20 angiosarcoma 5 haemangioma 19 angiosarcoma 10 benign vascular lesions

Zietz et al32 1998

Abnormal MDM2 in 68% of angiosarcoma, abnormal P53 in 53%; abnormal P53 and MDM2 expression associated with VEGF expression; no correlation between VEGF expression and tumour grade

All studies used immunohistochemistry on archived pathology slides. VEGF=vascular endothelial growth factor. VEGFR=VEGF receptor.

Table 2: Studies of angiogenic markers in angiosarcoma

predict poor outcome include old age, metastatic disease at presentation, and poor patient performance status.9,26,54 Several investigators have suggested that site of the disease affects prognosis. Angiosarcoma of the viscera (particularly liver and heart) and retroperitoneal disease are associated with poor outcome,9 as is radiationinduced disease in some series.7,55

Panel 3: TNM staging for soft-tissue sarcomas, including angiosarcoma T Tx T0 T1 T2 Primary tumour cannot be assessed No primary tumour found Tumour 5 cm diameter T1a: superficial; T1b: deep Tumour >5 cm diameter T2a: superficial; T2b: deep

Diagnostic assessment
Size of the primary lesion and presence or absence of distant metastases are the most useful determinants of treatment options. Diagnostic assessment includes biopsy, to conrm histology, and MRI to delineate the extent of the primary lesion before surgery. CT imaging of the thorax is done to exclude metastatic disease, and PET imaging may be useful to detect metastases when radical surgery is being planned.56 Although angiosarcomas can spread through the lymphatic system, the value of sentinel-lymph-node biopsy is unknown.

N Nx Cannot be assessed N0 No regional lymph-node involvement N1 Regional lymph-node involvement M Mx Cannot be assessed M0 No distant metastasis M1 Distant metastasis

Treatment
There are no randomised trials and few prospective studies, most published reports of angiosarcoma treatment are retrospective case series. Treatment has been included in management guidelines for other softtissue sarcomas, such as those published by the European Society for Medical Onocology (ESMO) and the National Comprehensive Cancer Network (NCCN). No evidencebased recommendations can be made for specic angiosarcoma subtypes and patients with this rare disease should be referred to specialist centres. Additional surgery in locally-recurrent disease to achieve a pathological complete resection may improve survival.27 Because of the high risk of local recurrence, adjuvant radio therapy, with large doses (>50 Gy), and wide treatment elds is recommended.6,26 No formal radio therapy trials have been done, but retrospective series suggest that it improves local control and overall survival,2,26 although investigators have failed to identify any benet.7,9,53,54 Radiation treatment alone is generally thought to be inadequate treatment for potentially curable disease,6 and further radiotherapy is usually avoided for radiation-induced angiosarcomas. In view of the risk of metastasis there is a rationale for using adjuvant chemotherapy. In a Japanese series of 55 angiosarcomas, multimodal treatment (surgery and chemotherapy with or without radiotherapy) was associated with a better prognosis.54 However, neither a large meta-analysis nor a recent prospective study on the use of adjuvant, anthracycline-based chemotherapy in soft-tissue sarcomas found evidence of survival benet.57,58 Although there were too few angiosarcomas to analyse
987

Local disease
Radical surgery with complete (R0) resection is the primary treatment of choice. Involved margins (R1 or R2 resection) are common because of the invasive and often multifocal nature of angiosarcomas, which confer a worse prognosis.3,7,9,26 Wide margins are therefore recommended; however, this can be difficult to achieve because of diffuse tissue inltration, tumour location (eg, cardiac angiosarcomas), and the relation of the tumour to other anatomical structures (eg, head and neck tumours).
www.thelancet.com/oncology Vol 11 October 2010

Review

Type of study

Number of patients 33 30 41 9 9 32 30

Chemotherapy

ORR

Median PFS

Median OS

Additional notes

Budd GT60 Fury MG et al3 Fury MG et al3 Fata F et al66 Nagano T et al67 Schlemmer M et al2 Penel N68

Review of RCTs Case series Case series Case series Case series Case series Phase 2

Doxorubicin-based 11 PR (ORR 33%) regimens Doxorubicin-based NR regimens Paclitaxel Paclitaxel Docetaxel Paclitaxel Weekly paclitaxel NR

NR 3754 months 40 months

NR NR NR NR NR NR 76 months

Pooled data from angiosarcoma patients enrolled in four prospective RCTs in advanced soft tissue sarcoma .. Significant difference in taxane response for disease above and below the neck (68 vs 28 months) .. .. ORR 75% (PFS 95 months) for face and scalp versus 58% (PFS 70 months) for other sites ..

4 CR and 4PR (ORR 89%) 50 months 2 CR and 4 PR (ORR 67%) 95 months 1 CR and 20 PR (ORR 63%) 2 CR/3 PR (ORR 17%) 76 months 40 months

ORR=overall response rate. PFS=progression-free survival. OS=overall survival. NR=not reported. CR=complete response. PR=partial response. RCT=randomised controlled trial. PR=partial response.

Table 3: Studies of palliative chemotherapy in locally advanced or metastatic angiosarcoma

these separately, no subgroups were identied in which adjuvant chemotherapy could be recommended. One case series reported encouraging results in 15 patients with periorbital angiosarcoma treated with neoadjuvant chemotherapy over 23 years,59 with 11 different cytotoxic and biological molecules used during this period. There is no convincing evidence to support the use of chemotherapy in the neoadjuvant or adjuvant setting after denitive surgery and radiotherapy for angiosarcoma.7,9,53

Metastatic disease
Cytotoxic chemotherapy is the primary treatment option for metastatic angiosarcoma, although the evidence base for this is limited. The main drug groups used are anthra cyclines, ifosfamide, and taxanes. Many angiosarcoma patients are elderly, so comorbidities and the risk of treatment related toxicity can limit the use of chemo therapy. Although patients with angiosarcoma have been eligible for trials in soft-tissue sarcoma, they may have been under-represented because of the difficulty in assessing geographic cutaneous disease by conventional response criteria. Published work largely comprises case series treated with various different approaches and regimens (table 3). In soft-tissue sarcomas, doxorubicin and ifosfamide as single drugs show response rates of 1636%.52 Combination chemotherapy is associated with increased toxicity but not necessarily with better outcomes.52 A large meta-analysis of anthracycline-based chemotherapy in soft-tissue sarcomas found an overall response rate of 26% and median survival of 51 weeks.60 Response rates varied with histological grade and there was some variation according to histological subtype, although angiosarcomas were not separately analysed. Similar response rates and worse survival have been reported in angiosarcoma,3,61 and the addition of cisplatin,62 vinorelbine,63 and gemcitabine64 to doxorubicin and ifosfamide is of no proven benet in angiosarcoma. Therapeutic responses to liposomal doxorubicin have been reported,3,65 showing accumulation in vascular tissue and
988

also efficacy in treating Kaposis sarcoma. In one retrospective study of 36 patients with advanced angiosarcoma, 21 (58%) received chemotherapy with various regimens, mostly doxorubicin based. 13 patients (64%) had at least a partial response, although the limited analysis suggested chemotherapy did not improve survival.7 Another retrospective study of 29 patients with metastaticbreast angiosarcoma reported responses in 14 patients (48%) with anthracycline and ifosfamide and taxane and gemcitabine combinations. No analysis was provided for progression-free or overall survival.52 Over the past decade there has been growing interest in the use of taxane chemotherapy for the treatment of angiosarcoma. Response rates in other soft-tissue sarcomas have been disappointing,66 but taxanes have antiangiogenic activity and are therefore of particular interest in the management of angiosarcomas. In a study of patients with cutaneous sarcoma, eight of nine treated with paclitaxel67 and six of nine patients treated with docetaxel responded to treatment.68 In a larger retro spective study of 32 patients with angiosarcoma, 20 patients (63%) responded to paclitaxel treatment, with a progression-free survival of 76 months.2 This study also reported more favourable responses for face and scalp tumours (overall response 75%, progressionfree survival 95 months vs 70 months) than for angio sarcomas at other sites. The only prospective phase 2 study of chemotherapy for angiosarcoma, investigated the use of weekly paclitaxel in 30 patients with locallyadvanced and metastatic disease.69 Of these patients, 19 (63%) were chemotherapy-naive, 11 (37%) had previously received anthracycline-based chemo therapy. Only ve (17%) showed a partial response; however, three of these patients proceeded to surgery and two showed a pathological-complete response. The treatment was well tolerated, although six patients developed grade 3 toxic effects, one patient developed a grade 4 toxic effect, and one patient died due to thrombocytopenia; median survival was 8 months.
www.thelancet.com/oncology Vol 11 October 2010

Review

Evaluable patient numbers Ryan CW et al73 Maki RG et al72 George et al74 Agulnik et al71 38 STS (including 9 angiosarcomas) 122 STS (including 37 angiosarcomas) 48 STS (including 2 angiosarcomas) 26 angiosarcomas

Treatment Sorafenib Sorafenib Sunitinib Bevacizumab

Overall response rate 2 PR and 14 SD (ORR 5%) overall; 1 PR and 6 SD (ORR 11%) in angiosarcoma

Median progression-free survival Median overall survival 30 months overall and 47 months in angiosarcoma 130 months overall and 135 months in angiosarcoma

1 CR, 5 PR , and 62 SD (ORR 5%) overall; 1 CR, 32 months overall and 38 months 14 9 months overall and 14 2 months 4 PR, and 21 SD (ORR 14%) in angiosarcoma in angiosarcoma in angiosarcoma 1 PR and 10 SD (ORR 2%) overall; no responses in angiosarcoma 3 PR and 13 SD (ORR 12%) NR NR NR NR

ORR=overall response rate. STS=soft-tissue sarcoma. NR=not reported. SD=stable disease. PR=partial response. CR=complete response.

Table 4: Phase 2 trials of biological therapies in soft-tissue sarcoma, including angiosarcomas

No data comparing the efficacy of anthracycline and taxane chemotherapy are available. Retrospective comparisons suggest similar response rates and survival with both,3 and consistent reports across several studies suggest that cutaneous angiosarcoma of the head and neck responds better to taxanes than does disease in other anatomical locations.2,3,67,68

Biological molecules
Although the pathogenic pathways underlying angio sarcoma are not fully understood, several groups have become interested in exploring the potential of antiangiogenic molecules in the treatment of angio sarcomas (table 4). Encouraging reports on the use of the VEGF-A monoclonal antibody, bevacizumab, include one of two patients with nasal angiosarcomas who achieved pathological-complete responses to treat ment with bevacizumab and radiotherapy.70 A phase 1/2 study of bevacizumab, in combination with docetaxel and gemcitabine chemotherapy in advanced and recurrent soft-tissue sarcomas included three angiosarcomas, with complete responses reported in two of them.71 Preliminary results of a phase 2 study of single-agent bevacizumab in 29 patients with angiosarcoma has been reported. Three partial responses and 13 patients achieving stable disease were seen in 26 evaluable patients, suggesting signicant activity.72 Use of broad-spectrum tyrosine-kinase inhibitors to target VEGFRs has been assessed in several phase 2 studies (table 4). The largest of these included 37 angio sarcomas within a larger study of 122 advanced-soft-tissue sarcomas treated with sorafenib.73 Partial responses were seen in four patients and one complete response was seen within the angiosarcoma group, giving an overall-response rate of 14% with a median progression-free survival of 38 months. By contrast, only one further partial response was seen in the remaining patients, giving an overallresponse rate of 4% and a median progression-free survival of 32 months for the overall study. Median survival was similar in the angiosarcoma and the advancedsoft-tissue sarcoma groups. Another study with sorafenib74 included nine vascular tumours among 38 patients with soft-tissue sarcoma. One patient had an unconrmed partial response and six had stable disease in the angiosarcoma group, with a median progression-free
www.thelancet.com/oncology Vol 11 October 2010

survival of 47 months and median survival of 135 months. Only one further patient with a partial response and eight with stable disease were reported among the patients with sarcoma. Their median progression-free survival was only 18 months, although overall survival was similar. Final results from a phase 2 study of sunitinib in 48 patients with metastatic-soft-tissue sarcomas included two patients with angiosarcoma.75 One patient had a partial response and ten had stable disease, but none in were reported in the angiosarcoma group. Hence, there is evidence to suggest activity of the targeted treatments, bevacizumab and sorafenib, in angiosarcoma. Further trials of targeted agents are ongoing, including a registered UK phase 2 study of the VEGFR-inhibitor axitinib (ISRCTN60791336). Biological agents are not recommended, either as mono therapy or in combination with cytotoxic chemo therapy, outside of clinical trials. There are a few case reports of responses to thalidomide in advanced or metastatic angiosarcoma,76 which too has antiangiogenic properties. Potential avenues for exploiting the endothelial origin of angiosarcoma include the use of vascular disrupting compounds, such as combretastatin and 5,6-dimethylxanthenone-4-acetic acid (DMXAA or ASA404). Interferon is an immune modulator, but also has antiangiogenic activity and in a preclinical study, interferon- inhibited transformed-murine endothelialcell lines both in vitro and in vivo through inhibition of angiogenesis.77 Additionally, there are case reports documenting a response to interferon- in combination with chemotherapy (liposomal doxorubicin) in advancedcutaneous angiosarcoma.78 Interleukin 2 has been administered as single agent and in combination with chemotherapy or radiotherapy, and a retrospective series of 20 patients treated with interleukin 2 and concurrent radiotherapy suggests that combined systemic and local arterial or direct to tumour administration might improve the progression-free survival rate.79 However insufficient evidence means interferon and interleukin-2 should not be recommended for treatment outside clinical trials.

Conclusion
Current treatment options are limited, particularly in radiation-induced angiosarcomas, with surgery as the main curative treatment option. However, wide resection margins are required, and local control is improved with
989

Review

Search strategy and selection criteria PubMed and ASCO abstracts were searched for articles in English using search terms including angiosarcoma, hemangiosarcoma, malignant hemangioendothelioma, lymphangiosarcoma, and radiotherapy, chemotherapy, drug therapy, prognosis, aetiology, pathogenesis, biology, angiogenesis, VEGF. We included papers published between January, 1989, and September, 2009, and earlier landmark studies.

adjuvant radiotherapy. Despite a substantial risk of subsequent metastatic disease there is no compelling evidence for adjuvant chemotherapy. Palliative chemo therapy is hampered by toxicities and poor response rates, and on the basis of available evidence, treatment with single-agent doxorubicin or paclitaxel is recommended. Biological therapies, in particular antiangiogenic therapies, offer hope for an angiosarcoma specic treat ment, with further prospective studies needed to clarify treatment strategies.
Contributors RJY wrote the paper. PJW supplied images and contributed to the writing of the paper. PJW, NJB, MWR, and DH provided editorial comments. DH contributed histopathology images. Conicts of interest PJW is a member of the Novartis advisory board and has received research funding from Pzer. All other authors have declared no conicts of interest. Acknowledgments RJY received funding from Yorkshire Cancer Research. References 1 Weiss SW, Goldblum JR. Enzinger & Weisss soft tissue tumors, 5th edn: Mosby; 2008. 2 Schlemmer M, Reichardt P, Verweij J, et al. Paclitaxel in patients with advanced angiosarcomas of soft tissue: a retrospective study of the EORTC soft tissue and bone sarcoma group. Eur J Cancer 2008; 44: 243336. 3 Fury MG, Antonescu CR, Van Zee KJ, Brennan MF, Maki RG. A 14-year retrospective review of angiosarcoma: clinical characteristics, prognostic factors, and treatment outcomes with surgery and chemotherapy. Cancer J 2005; 11: 24147. 4 Coindre JM, Terrier P, Guillou L, et al. Predictive value of grade for metastasis development in the main histologic types of adult soft tissue sarcomas: a study of 1240 patients from the French Federation of Cancer Centers Sarcoma Group. Cancer 2001; 91: 191426. 5 Rouhani P, Fletcher CD, Devesa SS, Toro JR. Cutaneous soft tissue sarcoma incidence patterns in the U.S.: an analysis of 12,114 cases. Cancer 2008; 113: 61627. 6 Mark RJ, Poen JC, Tran LM, Fu YS, Juillard GF. Angiosarcoma. A report of 67 patients and a review of the literature. Cancer 1996; 77: 240006. 7 Abraham JA, Hornicek FJ, Kaufman AM, et al. Treatment and outcome of 82 patients with angiosarcoma. Ann Surg Oncol 2007; 14: 195367. 8 Naka N, Ohsawa M, Tomita Y, Kanno H, Uchida A, Aozasa K. Angiosarcoma in Japan: a review of 99 cases. Cancer 1995; 75: 98996. 9 Fayette J, Martin E, Piperno-Neumann S, et al. Angiosarcomas, a heterogeneous group of sarcomas with specic behavior depending on primary site: a retrospective study of 161 cases. Ann Oncol 2007; 18: 203036. 10 Rossi S, Fletcher CD. Angiosarcoma arising in hemangioma/ vascular malformation: report of four cases and review of the literature. Am J Surg Pathol 2002 Oct; 26: 131929.

11 Stewart FW, Treves N. Lymphangiosarcoma in postmastectomy lymphedema; a report of six cases in elephantiasis chirurgica. Cancer 1948; 1: 6481. 12 Huang J, Mackillop WJ. Increased risk of soft tissue sarcoma after radiotherapy in women with breast carcinoma. Cancer 2001; 92: 17280. 13 Virtanen A, Pukkala E, Auvinen A. Angiosarcoma after radiotherapy: a cohort study of 332,163 Finnish cancer patients. Br J Cancer 2007; 97: 11517. 14 Karlsson P, Holmberg E, Samuelsson A, Johansson KA, Wallgren A. Soft tissue sarcoma after treatment for breast cancera Swedish population-based study. Eur J Cancer 1998; 34: 206875. 15 West JG, Weitzel JN, Tao ML, Carpenter M, West JE, Fanning C. BRCA mutations and the risk of angiosarcoma after breast cancer treatment. Clin Breast Cancer 2008; 8: 53337. 16 Bosetti C, La Vecchia C, Lipworth L, McLaughlin JK. Occupational exposure to vinyl chloride and cancer risk: a review of the epidemiologic literature. Eur J Cancer Prev 2003; 12: 42730. 17 Ron E. Cancer risks from medical radiation. Health Phys 2003; 85: 4759. 18 Locker GY, Doroshow JH, Zwelling LA, Chabner BA. The clinical features of hepatic angiosarcoma: a report of four cases and a review of the English literature. Medicine (Baltimore) 1979; 58: 4864. 19 Ben-Izhak O, Kerner H, Brenner B, Lichtig C. Angiosarcoma of the colon developing in a capsule of a foreign body. Report of a case with associated hemorrhagic diathesis. Am J Clin Pathol 1992; 97: 41620. 20 Weiss WM, Riles TS, Gouge TH, Mizrachi HH. Angiosarcoma at the site of a Dacron vascular prosthesis: a case report and literature review. J Vasc Surg 1991; 14: 8791. 21 McDonald DJ, Enneking WF, Sundaram M. Metal-associated angiosarcoma of bone: report of two cases and review of the literature. Clin Orthop Relat Res 2002; 369: 20614. 22 Folpe AL, Johnston CA, Weiss SW. Cutaneous angiosarcoma arising in a gouty tophus: report of a unique case and a review of foreign material-associated angiosarcomas. Am J Dermatopathol 2000; 22: 41821. 23 Ahmed I, Hamacher KL. Angiosarcoma in a chronically immunosuppressed renal transplant recipient: report of a case and review of the literature. Am J Dermatopathol 2002; 24: 33035. 24 Goedert JJ, Cote TR, Virgo P, et al. Spectrum of AIDS-associated malignant disorders. Lancet 1998; 351: 183339. 25 Schmid H, Zietz C. Human herpesvirus 8 and angiosarcoma: analysis of 40 cases and review of the literature. Pathology 2005; 37: 28487. 26 Pawlik TM, Paulino AF, McGinn CJ, et al. Cutaneous angiosarcoma of the scalp: a multidisciplinary approach. Cancer 2003; 98: 171626. 27 Lahat G, Dhuka AR, Lahat S, et al. Outcome of locally recurrent and metastatic angiosarcoma. Ann Surg Oncol 2009; 16: 250259. 28 Calonje E, Fletcher CDM. Vascular Tumors. In: Fletcher CDM, ed. Diagnostic Histopathology of Tumors. 3rd edn.: Churchill Livingstone; 2007: 4176. 29 Weiss SW, Lasota J, Miettinen MM. Angiosarcoma of soft tissue. In: Fletcher CDM, Unni KK, Mertens F, eds. WHO classication tumours of soft tissue and bone. IARC Press; 2002: 17577. 30 Ohsawa M, Naka N, Tomita Y, Kawamori D, Kanno H, Aozasa K. Use of immunohistochemical procedures in diagnosing angiosarcoma. Evaluation of 98 cases. Cancer 1995; 75: 286774. 31 Itakura E, Yamamoto H, Oda Y, Tsuneyoshi M. Detection and characterization of vascular endothelial growth factors and their receptors in a series of angiosarcomas. J Surg Oncol 2008; 97: 7481. 32 Zietz C, Rossle M, Haas C, et al. MDM-2 oncoprotein overexpression, p53 gene mutation, and VEGF up-regulation in angiosarcomas. Am J Pathol 1998; 153: 142533. 33 Dim D, Ravi V, Tan J, Hicks D, Wong M. The actin-bundling motility protein fascin and vascular endothelial growth factor (VEGF) are universally over-expressed in human angiosarcoma. Proc Am Soc Clin Oncol 2007; 25: 10068. 34 Naka N, Tomita Y, Nakanishi H, et al. Mutations of p53 tumorsuppressor gene in angiosarcoma. Int J Cancer 1997; 71: 95255. 35 Ueda T, Oji Y, Naka N, et al. Overexpression of the Wilms tumor gene WT1 in human bone and soft-tissue sarcomas. Cancer Sci 2003; 94: 27176.

990

www.thelancet.com/oncology Vol 11 October 2010

Review

36 Johnson KD, Glinskii OV, Mossine VV, et al. Galectin-3 as a potential therapeutic target in tumors arising from malignant endothelia. Neoplasia 2007; 9: 66270. 37 Arbiser JL, Moses MA, Fernandez CA, et al. Oncogenic H-ras stimulates tumor angiogenesis by two distinct pathways. Proc Natl Acad Sci USA 1997; 94: 86166. 38 Przygodzki RM, Finkelstein SD, Keohavong P, et al. Sporadic and Thorotrast-induced angiosarcomas of the liver manifest frequent and multiple point mutations in K-ras-2. Lab Invest 1997; 76: 15359. 39 Weihrauch M, Bader M, Lehnert G, et al. Mutation analysis of K-ras-2 in liver angiosarcoma and adjacent nonneoplastic liver tissue from patients occupationally exposed to vinyl chloride. Environ Mol Mutagen 2002; 40: 3640. 40 Garcia JM, Gonzalez R, Silva JM, et al. Mutational status of K-ras and TP53 genes in primary sarcomas of the heart. Br J Cancer 2000; 82: 118385. 41 LaMontagne KR Jr, Moses MA, Wiederschain D, et al. Inhibition of MAP kinase kinase causes morphological reversion and dissociation between soft agar growth and in vivo tumorigenesis in angiosarcoma cells. Am J Pathol 2000; 157: 193745. 42 Perry B, Banyard J, McLaughlin ER, et al. AKT1 overexpression in endothelial cells leads to the development of cutaneous vascular malformations in vivo. Arch Dermatol 2007; 143: 50406. 43 Naito S, Shimizu K, Nakashima M, et al. Overexpression of Ets-1 transcription factor in angiosarcoma of the skin. Pathol Res Pract 2000; 196: 10309. 44 Dictor M, Bendsoe N, Runke S, Witte M. Major basement membrane components in Kaposis sarcoma, angiosarcoma and benign vascular neogenesis. J Cutan Pathol 1995; 22: 43541. 45 Yamamoto T, Umeda T, Nishioka K. Immunohistological distribution of stem cell factor and kit receptor in angiosarcoma. Acta Derm Venereol 2000; 80: 44345. 46 Miettinen M, Sarlomo-Rikala M, Lasota J. KIT expression in angiosarcomas and fetal endothelial cells: lack of mutations of exon 11 and exon 17 of C-kit. Mod Pathol 2000; 13: 53641. 47 Hornick JL, Fletcher CD. Immunohistochemical staining for KIT (CD117) in soft tissue sarcomas is very limited in distribution. Am J Clin Pathol 2002; 117: 18893. 48 Komdeur R, Hoekstra HJ, Molenaar WM, et al. Clinicopathologic assessment of postradiation sarcomas: KIT as a potential treatment target. Clin Cancer Res 2003; 9: 292632. 49 DuBois S, Demetri G. Markers of angiogenesis and clinical features in patients with sarcoma. Cancer 2007; 109: 81319. 50 Wong KF, So CC, Wong N, Siu LL, Kwong YL, Chan JK. Sinonasal angiosarcoma with marrow involvement at presentation mimicking malignant lymphoma: cytogenetic analysis using multiple techniques. Cancer Genet Cytogenet 2001; 129: 6468. 51 Sobin LH, Wittekind CH. TNM classication of malignant tumours, 6th edn (UICC). New York: Wiley-Liss, 2002. 52 Mocellin S, Rossi CR, Brandes A, Nitti D. Adult soft tissue sarcomas: conventional therapies and molecularly targeted approaches. Cancer Treat Rev 2006; 32: 927. 53 Sher T, Hennessy BT, Valero V, et al. Primary angiosarcomas of the breast. Cancer 2007; 110: 17378. 54 Naka N, Ohsawa M, Tomita Y, et al. Prognostic factors in angiosarcoma: a multivariate analysis of 55 cases. J Surg Oncol 1996; 61: 17076. 55 Vorburger SA, Xing Y, Hunt KK, et al. Angiosarcoma of the breast. Cancer 2005; 104: 268288. 56 Maeda T, Tateishi U, Hasegawa T, Ojima H, Arai Y, Sugimura K. Primary hepatic angiosarcoma on coregistered FDG PET and CT images. AJR Am J Roentgenol 2007; 188: 161517. 57 Woll PJ, van Glabbeke M, Hohenberger P, et al. Adjuvant chemotherapy (CT) with doxorubicin and ifosfamide in resected soft tissue sarcoma (STS): Interim analysis of a randomised phase III trial. Proc Am Soc Clin Oncol 2007; 25: abstr 10008. 58 Adjuvant chemotherapy for localised resectable soft-tissue sarcoma of adults: meta-analysis of individual data. Sarcoma Meta-analysis Collaboration. Lancet 1997; 350: 164754.

59 DeMartelaere SL, Roberts D, Burgess MA, et al. Neoadjuvant chemotherapy-specic and overall treatment outcomes in patients with cutaneous angiosarcoma of the face with periorbital involvement. Head Neck 2008; 30: 63946. 60 Van Glabbeke M, van Oosterom AT, Oosterhuis JW, et al. Prognostic factors for the outcome of chemotherapy in advanced soft tissue sarcoma: an analysis of 2,185 patients treated with anthracyclinecontaining rst-line regimensa European Organization for Research and Treatment of Cancer Soft Tissue and Bone Sarcoma Group Study. J Clin Oncol 1999; 17: 15057. 61 Budd GT. Management of angiosarcoma. Curr Oncol Rep 2002; 4: 51519. 62 Asmane I, Litique V, Heymann S, et al. Adriamycin, cisplatin, ifosfamide and paclitaxel combination as front-line chemotherapy for locally advanced and metastatic angiosarcoma. Analysis of three case reports and review of the literature. Anticancer Res 2008; 28: 304145. 63 Anderson SE, Keohan ML, DAdamo DR, Maki RG. A retrospective analysis of vinorelbine chemotherapy for patients with previously treated soft-tissue sarcomas. Sarcoma 2006; 2006: 15947. 64 Gautam U, Hurley J, Silva OE, Benedetto PW, Robles C. Gemcitabine: an active chemotherapeutic agent for angiosarcoma. Proc Am Soc Clin Oncol 2002; 21: abstr 2931. 65 Skubitz KM, Haddad PA. Paclitaxel and pegylated-liposomal doxorubicin are both active in angiosarcoma. Cancer 2005; 104: 36166. 66 Verweij J, Lee SM, Ruka W, et al. Randomized phase II study of docetaxel versus doxorubicin in rst- and second-line chemotherapy for locally advanced or metastatic soft tissue sarcomas in adults: a study of the european organization for research and treatment of cancer soft tissue and bone sarcoma group. J Clin Oncol 2000; 18: 208186. 67 Fata F, OReilly E, Ilson D, et al. Paclitaxel in the treatment of patients with angiosarcoma of the scalp or face. Cancer 1999; 86: 203437. 68 Nagano T, Yamada Y, Ikeda T, Kanki H, Kamo T, Nishigori C. Docetaxel: a therapeutic option in the treatment of cutaneous angiosarcoma: report of 9 patients. Cancer 2007; 110: 64851. 69 Penel N, Bui BN, Bay JO, et al. Phase II trial of weekly paclitaxel for unresectable angiosarcoma: the ANGIOTAX Study. J Clin Oncol 2008; 26: 526974. 70 Koontz BF, Miles EF, Rubio MA, et al. Preoperative radiotherapy and bevacizumab for angiosarcoma of the head and neck: two case studies. Head Neck 2008; 30: 26266. 71 Verschraegen CF, Quinn R, Rabinowitz I, Arias-Pulido H, Muller C. Phase I/II study of docetaxel (D), gemcitabine (G), and bevacizumab (B) in patients (pts) with advanced or recurrent soft tissue sarcoma (STS). Proc Am Soc Clin Oncol 2008; 26: abstr 10534. 72 Agulnik M, Okuno SH, Von Mehren M, et al. An open-label multicenter phase II study of bevacizumab for the treatment of angiosarcoma. Proc Am Soc Clin Oncol 2009; 27: abstr 10522. 73 Maki RG, DAdamo DR, Keohan ML, et al. Phase II study of sorafenib in patients with metastatic or recurrent sarcomas. J Clin Oncol 2009; 27: 313340. 74 Ryan CW, von Mehren M, et al. Phase II intergroup study of sorafenib (S) in advanced soft tissue sarcomas (STS): SWOG 0505. Proc Am Soc Clin Oncol 2008; 26: 10532. 75 George S, Merriam P, Maki RG, et al. Multicenter phase II trial of sunitinib in the treatment of nongastrointestinal stromal tumor sarcomas. J Clin Oncol 2009; 27: 315460. 76 Raina V, Sengar M, Shukla NK, et al. Complete response from thalidomide in angiosarcoma after treatment of breast cancer. J Clin Oncol 2007; 25: 90001. 77 Taylor KL, Oates RK, Grane R, Leaman DW, Borden EC, Lindner DJ. IFN-alpha1,8 inhibits tumor-induced angiogenesis in murine angiosarcomas. J Interferon Cytokine Res 2006; 26: 35361. 78 Burns BT, Blakey SA, Harris WB. Complete response of metastatic angiosarcoma to liposomal doxorubicin and interferon-2a. Proc Am Soc Clin Oncol 2002; 21: 2939 (abstr). 79 Ohguri T, Imada H, Nomoto S, et al. Angiosarcoma of the scalp treated with curative radiotherapy plus recombinant interleukin-2 immunotherapy. Int J Radiat Oncol Biol Phys 2005; 61: 144653.

www.thelancet.com/oncology Vol 11 October 2010

991