Evaluation of CD10 and procollagen 1 expression in atypical fibroxanthoma and dermatofibroma

Atypical fibroxanthoma (AFX) (dermal pleomorphic sarcoma) remains a somewhat controversial entity. Some authors have averred that AFX is a fiction, suggesting that such lesions merely represent misclassified examples of spindled squamous cell carcinoma. In addition, the immunoperoxidase confirmation of AFX has been less than straightforward and has historically been approached as a diagnosis of exclusion because of the lack of sensitivity and specificity of available "positive" reagents. Procollagen 1 (PC1) and CD10 represent recently developed immunoperoxidase reagents that have been forwarded as useful in this setting, and we sought to characterize our experience, both to confirm the utility of these antibodies and to compare them. Our investigation included 3 separate data sets. Group 1 consisted of a retrospective review of 98 consecutive cases in which PC1 was used in the evaluation of dermatopathology specimens in routine practice during a 13-month interval. Group 2 consisted of a direct comparison of 11 AFX, 11 dermatofibroma (DF), and 7 epithelioid dermatofibroma (EDF) using the CD10 reagent on cases identified by database search. Group 3 consisted of a retrospective review of 47 cases in which CD10 was used in routine practice during a 10-month interval. Group 1 included 47 AFX, 13 carcinomas, and 6 melanomas. PC1 expression was observed in 45 of 47 AFX (96%), with a strong reaction in 78% of cases. Among a comparison group of carcinomas, 13 of 13 displayed strong keratin immunopositivity and 11 of 13 (85%) lacked PC1 expression whereas 2 showed focal weak labeling. Six of six melanomas exhibited avid S100 expression and none labeled with PC1. In group 2, strong CD10 immunoreactivity was present in 11 of 11 AFX. Similarly, 11 of 11 DFs were also positive. In contrast, 6 of 7 cases of EDF lacked CD10 expression. Group 3 included 38 AFX and 9 miscellaneous spindle cell proliferations. Of the 38 AFX, 37 (97%) labeled with CD10 and in 34 (92%) the reaction was strong. PC1 immunostaining was also completed in 34 of 38 AFX from group 3 and 27 (79%) cases showed positive labeling. Our results confirm that both PC1 and CD10 can be used as positive markers of AFX. We believe that CD10 and PC1 immunostaining can be used as a useful adjunct to supplement the diagnosis of AFX, within the context of an immunoperoxidase panel. Not surprisingly, CD10 expression is also common in DF, a benign analog of AFX, with the exception of its epithelioid variant. In direct head-to-head comparison, our experience indicates that the staining of AFX with CD10 is more avid than that observed with PC1. Lastly, out data includes over 80 examples of AFX, <5% of which showed keratin labeling. Given a general lack of keratin expression, it seems unlikely that AFX merely represents poorly differentiated squamous carcinoma.