Case reports

The three cases presented were each referred for pathological review during the process of medicolegal compensation claims for asbestos related disease. The clinical data and necroscopic information from the three patients with primary pleural epithelioid haemangioendothelioma are shown in table 1. The occupational asbestos exposure is given in table 2. Necroscopic examination showed that the pleura was the sole or major site of disease. The pathological diagnosis was independently made by at least two experienced pulmonary histopathologists after consideration of haematoxylin and eosin stained slides and a wide panel of immunohistochemical markers. In all cases ultrastructural studies were performed. In each case the parietal pleura was diffusely replaced by tumour composed of epithelioid and vacuolated cells forming small nests and cords with a minor spindle cell component. The tumour cells showed a moderate degree of cytonuclear pleomorphism and intracellular neovascular lumina containing erythrocytes were present (fig 1). Immunohistochemistry revealed that the tumour cells were positive for vimentin and with at least two of the three endothelial markers (CD 31, CD34, von Willebrand factor). In case 3 this additionally included the vascular marker thrombomodulin. The identification of intracytoplasmic Weibal-Palade organelles confirmed endothelial differentiation in cases 1 and 2 (fig 2). The diagnosis in both cases was of primary pleural epithelioid heamangioendothelioma. In case 3 the tissue was suboptimal for ultrastructural analysis. Four independent pathological opinions were sought and the overall favoured diagnosis was of primary pleural angiosarcoma (epithelioid heamangioendothelioma).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Epithelioid haemangioendothelioma with spindle and vacuolated cell forms showing neovascular lumina with erythrocytes (arrow). Stain: H & E, ×400 magnification.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Epithelioid haemangioendothelioma (case 1) containing intracytoplasmic Weibal-Palade organelle (arrow). EM, ×57 000 magnification.

Results

The results of the mineral analysis are shown in table 3. The putative latent periods of the three cases ranged from 18 to 60 years, with cumulative exposures to asbestos of 2–17 years. In cases 1 and 3 the intensity of asbestos exposure was light. In both cases ferruginous bodies were present but they were typical of asbestos only in case 2. In this case there was intermittent heavy exposure to asbestos and lung fibre burdens reflect the occupational history. Cases 1 and 3 do not show a clear cut increase in asbestos fibre levels.

The results of the immunohistochemical study of vascular marker expression in malignant mesothelioma shows that there was no immunoreactivity for CD31, CD34, or von Willebrand factor in any of the subtypes of the 92 malignant mesotheliomas tested.

Discussion

Epithelioid haemangioendothelioma is an uncommon low grade sarcoma of endothelial origin recognised in a variety of tissue sites including liver, soft tissue, lung, bone, and skin.5 It has been described as originating from the pleura6 and peritoneum.7

In the pleura the main differential diagnosis is that of diffuse malignant mesothelioma8 and clinicopathological distinction can be problematical. Firstly, a diffuse serosal growth pattern is a characteristic feature of malignant mesothelioma but this “mesotheliomatous” pattern is not specific and has been recognised in some non-mesothelial tumours, most notably carcinomas and, very rarely, sarcomas including epithelioid haemangioendothelioma. Secondly, on morphological grounds, epithelioid haemangioendothelioma may exhibit a biphasic pattern with epithelioid and spindle areas similar to biphasic malignant mesothelioma.

The accurate distinction of the two tumours has important medicolegal implications as only malignant mesothelioma is currently recognised as an asbestos related neoplasm for compensation purposes. The description of three primary pleural angiosarcomas in individuals exposed to asbestos raises two important questions. Firstly, do mesotheliomas have the capacity to undergo vascular differentiation and, secondly, if not, are pleural angiosarcomas aetiologically linked with asbestos?

Divergent differentiation of mesothelium has been described and cases of malignant mesothelioma showing leiomyoid, chondroid, osseous, and liposarcomatous patterns have been reported.9 Bona fide cases of malignant mesothelioma showing angiosarcomatous differentiation have not been described.

In order to address the issue of vascular differentiation in malignant mesothelioma an immunohistochemical analysis of 92 mesothelioma was undertaken using the three most sensitive and specific vascular markers (CD31,10 CD34,11 and von Willebrand factor12). In one recent study two of 16 malignant mesotheliomas and five of 84 mixed carcinomas showed weak non-specific cytoplasmic CD31 staining.13 In a further series14 no CD31 expression was seen in any of nine malignant mesotheliomas tested. The expression of von Willebrand factor and CD34 has not been reported in malignant mesothelioma and the results of this study support previous investigations.8The present study indicates no definite immunohistochemical evidence of vascular differentiation in any of the cohort of 92 malignant mesotheliomas studied and we conclude that vascular marker expression (and endothelial differentiation) does not occur or is a very infrequent finding in malignant mesothelioma. The use of a combination of the vascular markers CD31 and CD34 would therefore allow effective distinction of malignant mesothelioma from epithelioid angiosarcoma.

Considering the aetiological role of asbestos, in the largest published series of 14 malignant vascular tumours involving the serous membranes8 only one case of epithelioid angiosarcoma had a prior history of asbestos exposure. In order to ascribe mesothelioma to exposure to asbestos certain conditions must be met including (1) accurate diagnosis, (2) reliable evidence of exposure to amphibole asbestos, (3) an appropriate latent period, and (4) an absence of confounding aetiological factors such as irradiation or erionite mineral fibre exposure.

The diagnosis of these three cases has been discussed. The evidence of asbestos exposure can come from various sources including occupational history, light microscopic identification of asbestos bodies in lung tissue sections, and mineral fibre analysis of tissue digests. The reliability of occupational history is strongly dependent on “recall” which varies between occupations and is subject to bias whereby individuals can underestimate or overestimate their exposure to asbestos. Electron microscopic mineral fibre analysis allows for the accurate determination of asbestos fibre burden and fibre type at death. For case 2 all parameters indicated an increased crocidolite asbestos fibre count consistent with an occupational exposure. However, for cases 1 and 3 the mineralogical results were equivocal. Case 1 worked most of his life as an underground coal miner but for a period of about two years (prior to mining) worked as a labourer demolishing sheds said to contain asbestos roofing materials. Light microscopy showed occasional ferruginous bodies but no “classical” asbestos bodies. Mineral fibre analysis showed a low level of tremolite, a contaminant amphibole of chrysotile (white) asbestos ore. It is possible that the major asbestos exposure in this case was to chrysotile. However, it is well known that the biopersistence of chrysotile is low and, with clearance over a 32 year period, the fibre may have been undetectable by mineral analysis at death. A further interesting finding in this case was that 90% of the non-fibrous particulate matter was mica. The mica consisted of very large plates and was not typical of that usually seen in association with coal mining. Mica has been associated with ferruginous body formation (which are sometimes mistaken for “classical” asbestos bodies) and pleural plaques,15 and an awareness of this is important to prevent a false attribution to asbestos, particularly in the presence of a diffuse pleural tumour. There is no known association of mica with malignant mesothelioma. In summary, only the occupational history indicated exposure to asbestos. In case 3 there was a positive occupational history of exposure to asbestos as a shipyard welder but the chrysotile and amosite asbestos fibre levels detected were not clearly elevated above those identified in the background control population. Many ferruginous bodies and iron coated platy material were present, as frequently occurs in welders.16

An appropriate latent period (from initial exposure to asbestos to death of the patient) is an important factor in the ascription of an asbestos induced neoplasm. In a study of 1690 cases of malignant mesotheliomas the latent period was in excess of 15 years in 99% of cases.17 In comparison, in a series of occupational related (hepatic) angiosarcomas following exposure to vinyl chloride monomer18 the latent period was 8–32 years with a mean latency in excess of 20 years. In this series of pleural epithelioid haemangioendothelioma the tumour latency period was similar to that seen in asbestos related malignant mesotheliomas and ranged from 18 to 60 years. It is of course recognised that, whilst the pathogenic mechanisms of vinyl chloride and asbestos in neoplastic transformation may be considerably different, the studies do highlight the fact that long latent periods exist between initial exposure to the carcinogen and death of the patient in the two tumour sites (liver and pleura).

With respect to confounding factors, hepatic angiosarcomas have been aetiologically associated with radiotherapy, anabolic steroids, vinyl chloride monomer, and thorotrast. None of the patients in this series was known to have any of these risk factors. Because of their rarity, no risk factors for pleural angiosarcomas have been clearly defined.

In summary, three cases of primary epithelioid haemangioendothelioma of the pleura in patients who claimed prior exposure to asbestos are presented. The results of the immunohistochemical study do not support the view that endothelial differentiation occurs in malignant mesotheliomas. We advocate the introduction of endothelial markers such as CD31 and CD34 into the routine immunohistochemical panel used in the identification of malignant mesothelioma so that effective distinction from epithelioid haemangioendothelioma can be made. With regard to the aetiology (that is, asbestos), it is recognised that the three cases were identified during pursuit of medicolegal compensation claims for asbestos related disease and therefore demonstrate referral bias. However, further multicentre case-control studies with mineral fibre analysis should be prompted by the findings of this series to determine whether there is a genuine and consistent association between epithelioid haemangioendothelioma and exposure to asbestos and, if so, to determine whether there are similar dose responses and latency periods as in malignant mesothelioma.