Polygonal CD34 positivity portends trichilemmal differentiation

CD34 and BerEP4 Are Helpful to Distinguish Basaloid Tricholemmoma From Basal Cell Carcinoma

Tricholemmoma, a benign follicular neoplasm with outer root sheath differentiation, typically comprises clear or pale cells, and when multiple is pathognomic of Cowden's syndrome. The tumor is probably underrecognized and in basaloid examples can be difficult to distinguish from basal cell carcinoma (BCC). We studied 55 tricholemmomas (including 15 basaloid cases) and compared immunohistochemical profile with nodular BCC from our archives. Basaloid and non-basaloid tricholemmomas had similar staining characteristics. BerEP4 was focally positive (range 10%-20%) in only 3/39 (7.7%) tricholemmomas compared with widespread positivity in BCC (90.8%, 139 of 151 cases with ≥50% tumor area stained). CD34 was expressed, usually focally (median 20%, range 10%-90%), in 52/53 (98.1%) tricholemmomas and was negative in all 21 BCCs stained. EMA staining lacked sensitivity or specificity in differentiating tricholemmoma from BCC. Five or more Merkel cells were found in 7/17 (40.1%) tricholemmomas and 1/23 (4.3%) nodular BCCs studied. In summary, immunohistochemistry is helpful in distinction between tricholemmoma, including difficult basaloid examples (BerEP4 negative or focal, CD34 positive) compared with BCC (BerEP4 widespread in most cases, CD34 negative). The presence of 5 or more Merkel cells is a relatively specific but not a particularly sensitive discriminator.

Distinguishing neurofibroma from desmoplastic melanoma: the value of the CD34 fingerprint

We have observed 'fingerprint' immunopositivity in association with perineurioma and neurofibroma. A fingerprint consists of delicate, elongated areas of positive labeling that fall between collagen bundles, thereby creating a whorled configuration that is reminiscent of a human fingerprint. At present, the differential diagnosis between early desmoplastic melanoma and neurofibroma remains challenging in a subset of cases because of overlapping histopathological and immunohistochemical features. To assess whether fingerprint CD34 reactivity could be contributory in this context, we stained 50 desmoplastic melanomas and 50 neurofibromas with CD34. Fingerprint CD34 labeling was present in greater than 30% of the proliferation in 96% (n = 48) of neurofibromas and in only 4% (n = 2) of desmoplastic melanomas. Over two-thirds of the neurofibromas exhibited a CD34 fingerprint involving more than 60% of their surface area. In the two cases of desmoplastic melanoma that showed CD34 fingerprint positivity, the staining was patchy and involved less than 60% of the tumor. In partially staining neurofibromas, areas without a CD34 fingerprint tended to occur in central lobular areas. We conclude that CD34 fingerprint immunoreactivity is useful in distinguishing neurofibroma from early desmoplastic melanoma, especially if the fingerprint involves more than 60% of a tumor's cross-sectional area.