We appreciate the interest of Drs. Zamecnik, Michal, and Mukensnabl in our recent publication on pilomatrix carcinoma (1). In their letter, these authors challenge the specificity of shadow cells in the diagnosis of pilomatrical neoplasms. Shadow cells are cornified cells in which, as a consequence of karyolysis, nuclei have faded but vague outlines of them still can be recognized. According to the latest edition of Ackerman's book Neoplasms With Follicular Differentiation, these cells represent faulty attempts at differentiation toward hair and are found only in conditions in which some follicular matrical cells are present (2). Although shadow cells are most commonly seen in pilomatricoma and its malignant variant, the pilomatrix carcinoma, it is well known that other lesions with follicular differentiation may have shadow cells on rare occasions. These include trichoepithelioma, desmoplastic trichoepithelioma, proliferating trichilemmal tumor, and panfolliculoma (3). Additionally, the occurrence of shadow cells has already been reported in keratoacanthoma, chondroid syringoma, and in some odontogenic tumors, notably the calcifying odontogenic cyst and the complex odontoma (4–6). Recently, Dr. Zamecnik et al. have described the presence of shadow cells in extracutaneous locations, namely in adenosquamous carcinomas of the colon and uterus (7). This has led these authors to suggest that the pathologist should be aware that shadow cell differentiation may occur in visceral carcinomas and that it is not diagnostic of matrical tumors. There are unquestionable morphologic similarities between the cells they describe and folliculosebaceous matrical shadow cells, although they are not quite identical to the shadow cells seen in matrical neoplasms. In this regard, we agree with the opinion of Drs. Hitchcock and Ellington (8) in that the cells that Zamecnik et al. describe show much less regular distribution of the clear zone previously occupied by the nucleus and a more filamentous cytoplasm than matrical shadow cells. In our cases, the diagnosis of pilomatrix carcinoma was made because the tumors showed architectural features that implied malignancy coupled with cytologic characteristics of matrical differentiation. In the cases presented by Zamecnik et al. the shadow cells were present in isolation, without demonstrating the classic findings of matrical tumors. Obviously, it cannot be assumed that the superficial morphologic appearance between true shadow cells and the cells described by Zamecnik et al. is per se indicative of a tumor with matrical differentiation. Whether tumors previously interpreted as pilomatrix carcinoma could have represented cutaneous metastases from poorly differentiated visceral carcinomas with shadow cell differentiation remains, in our opinion, speculative. In such cases it could be of interest to perform immunohistochemical studies of follicle-associated keratins to identify further differences (9). Finally, the differential diagnosis between pilomatrix carcinoma and the cutaneous metastases from an undifferentiated visceral carcinoma has also important prognostic considerations. Pilomatrix carcinomas are locally aggressive tumors that have a tendency to recur, especially when they are incompletely excised. Metastasis and death have resulted from these tumors but are rare. On the other hand, the prognosis for patients with cutaneous metastasis from a visceral carcinoma is ominous, and their life expectancy from the appearance of the metastases is on average less than one year. David Hardisson, M.D. Félix Contreras, M.D. Jesús Cuevas-Santos, M.D. M. Dolores Linares, M.D.