Immunohistochemistry in surgical pathology. The case of the undifferentiated malignant neoplasm

An ideal immunohistochemical screening panel would be one in which each antibody is 100% sensitive and specific for the target cell type (e.g., markers for epithelial neoplasms, lymphomas, sarcomas, etc.). Anyone who has practiced immunochemistry is well aware that this situation does not exist. High sensitivity is hindered by the loss of key antigens through formalin fixation and routine tissue processing. Although sensitivity can be improved by dealing with fresh-frozen, lyophilized, or plastic-embedded specimens, these procedures are often perceived as inconvenient by pathologists. Specificity is a more insidious problem. With the advent of monoclonal antibody technology, many individuals equated the monospecificity (e.g., marking one antigen a determinant) with tissue specificity. This, of course, is not the case as determinants recognized by one monoclonal antibody may be expressed on cells of different lineage. High sensitivity and high specificity are important for different reasons. By definition, an undifferentiated neoplasm lacks morphologic features to unequivocally substantiate sarcoma, lymphoma, carcinoma, or melanoma. Thus, antibodies with low sensitivity that fail to mark a significant percent of cases will provide inconclusive or erroneous information. The failure of an antibody to stain a particular tissue could be a true-negative (valuable information) or a false-negative (misleading information) result. Obviously, when antibodies have a sufficiently low sensitivity, their use is a liability rather than an advantage. Specificity is obviously important. When an undifferentiated neoplasm is found to be "positive" for a particular marker, there is a tendency to immediately categorize the neoplasm. In this setting, when histologic features of cellular differentiation are totally lacking, an extreme degree of trust is being placed on the immunohistochemical technique. From our earlier discussion it is apparent that perfect sensitivity and specificity do not exist among most immunohistochemical reagents. Accordingly, the safest approach is to use a panel of antibodies that will disclose anomalous immunohistochemical reactions (e.g., a neoplasm positive for both keratin and LCA). Specimens from such cases should be carefully evaluated with additional monoclonal antibodies and scrutinized by light microscopy. Furthermore, while immunohistochemistry provides for rapid and cost-effective diagnosis, electron microscopy may still contribute valuable information. Despite our best intentions and desires, it is also clear that a small percentage of "undifferentiated" neoplasms will remain undifferentiated. Quality control and quality assurance are two final, but important, issues to address. An extraordinary large number of variables in tissue selection, fixation, and processing can skew results.(ABSTRACT TRUNCATED AT 400 WORDS)