Sarcoma-like tumor of head and neck skin

A group of tumors referred to as atypical fibroxanthoma (AFX) and pleomorphic dermal sarcoma (PDS) predominantly occur in sun-damaged skin of the elderly, particularly in the head and neck region. Although this group of tumors is often regarded as of mesenchymal phenotype, the matter of histogenesis has not been entirely resolved. Evans H and Smith JL reported in 1980 that prognosis was not significantly different irrespective of whether there was a definite squamous cell carcinoma component or not, supporting a view that these are all carcinomata in nature (sarcomatoid carcinoma [SC]). There are a number of clinicopathologic studies of AFX in the literature but information on morphologically similar sarcoma-like tumors with immunohistochemical evidence of epithelial differentiation is sparse. One hundred sarcoma-like tumors (SLTs) of head and neck skin of the elderly, treated by surgical excision, were studied. Clinical information was obtained, and pathology reports and hematoxylin and eosin sections were reviewed to document size (maximum dimension), extent of invasion, mitotic count, vascular and perineural invasion, margin status, ulceration, necrosis, and the presence of actinic keratosis in adjacent/overlying skin. Immunostains examined included: pan-cytokeratins (CKs) (AE1/AE3, MNF116), high-molecular weight CKs (34βE12, CK5/6, CK14), p63, and melanocytic (S100, Melan A, HMB-45, MITF), vascular (CD31, CD34), and muscle markers (SMA, desmin, h-caldesmon) to exclude melanoma and definite sarcoma entities. The tumors were divided into AFX/PDS (G1), the SC group, which was subdivided into SLT with only p63 positivity (G2a) and SLT with CK positivity regardless of p63 status (G2b), and SLT with a minor morphologic squamous cell carcinoma component (G3). Clinicopathologic findings of each group were compared, in relation to outcomes. Age at diagnosis ranged from 51 to 96 years (median, 79 y), with M:F=11.5:1. There were 53 tumors in G1 (19AFX, 34PDS), 37 in G2 (25 in G2a, 12 in G2b), and 10 in G3. There was no statistically significant difference in clinical and pathologic parameters or survival among all 3 groups. CKs and p63 expression, size, extent of invasion, vascular invasion, perineural invasion, mitotic count, and ulcer did not affect outcome, whereas margin status and necrosis did by both univariate and multivariate analysis and by only univariate analysis, respectively. Sixty patients had multiple nonmelanomatous skin cancers. Actinic keratosis was observed in overlying/adjacent epidermis in 51 cases. Eight patients had prior radiotherapy to head skin cancers; 1 patient developed 2 separate tumors (G1 and G3) after radiotherapy. Four patients died of tumor (1 G1, 2 G2b, and 1 G3); of these, 3 cases had positive margin, and 1 had narrow margin. Our results have shown similarities of various clinicopathologic parameters between AFX/PDS and SC, raising the possibility that both entities are related, and some of the former entities may represent complete dedifferentiation (complete loss of epithelial phenotype) with a gain of mesenchymal phenotype. In addition, the difference between AFX and PDS appears to be the extent of invasiveness (stage) rather than a different histogenesis. Further investigations are needed. However, from a practical point of view, efforts should be made to excise this group of tumors with clear margins, as margin status appears to be the most important prognostic factor regardless of the presence or absence of epithelial differentiation.