Drier J K, Swanson P E, Cherwitz D L, Wick M R
Arch Pathol Lab Med. 1987 May;111(5):447-52.
This study compares the immunoprofiles of 25 cutaneous and mucosal melanomas with those of 400 poorly differentiated carcinomas, using a polyclonal antiserum to S100 and murine monoclonal antibodies to keratin (KER) and epithelial membrane antigen (EMA). All malignant melanomas expressed S100 but lacked reactivity for KER and EMA. Conversely, all of 49 carcinomas (12%) that displayed S100 also exhibited positivity for both epithelial markers. The latter group of tumors included 17 carcinomas of the breast, 16 eccrine sweat gland carcinomas, five high-grade salivary glandular adenocarcinomas, four lung cancers, three pancreatic ductal carcinomas, two serous ovarian carcinomas, one clear cell adenocarcinoma of the bladder, and one uterine cervical squamous carcinoma. None of 32 epithelial neoplasms that lacked both KER and EMA (19 testicular seminomas, four hepatocellular carcinomas, eight adrenocortical carcinomas, and one Merkel's cell carcinoma) contained S100 protein. These results suggest that S100 protein is relatively nonspecific as a single immunodeterminant in the diagnostic separation of melanoma and anaplastic carcinoma. The concomitant use of stains for EMA and KER, however, obviates this problem. Finally, since it is somewhat restricted in distribution, S100 reactivity in a known carcinoma may be of some use in predicting possible primary sources for a metastasis of unknown origin.