Keratin expression in human tissues and neoplasms

Sir: Drs Chu and Weiss are to be congratulated for their comprehensive review of cytokeratin (CK) expression in human tissues and neoplasms. Their perceptive and succinct comment that such data ‘should be regarded as signposts that may, in conjunction with clinical, morphological and other immunohistochemical tumour marker data, guide one towards a correct diagnosis’ is welcome and should be widely publicized. This is because I am sure that I am not alone in having been told, by non-pathologists, that because a metastatic tumour demonstrated a positive reaction to a particular anti-cytokeratin antibody, usually an antibody to CK7 or 20, it must have originated in a particular organ system. This view may have been reinforced by accessing apparently convincing data, often unreferenced, on the internet. Although in some instances an individual cytokeratin may be almost exclusively present or absent from tumours deriving from a particular organ or system, exceptions occur even in cancers of the ovary (CK7) and colon (CK20) where almost 100% and 94% of the tumours reacted with the relevant antibody, respectively. The potential percentage of aberrant positive or negative reactions may be even greater in practice, and this is demonstrated if one applies confidence intervals to the data. I have applied confidence intervals to the proportion of cases in which a reaction to antibodies raised to CK7 and CK20 was identified, first in tumours of the female genital tract, and second in tumours originating in sites which commonly give rise to metastatic or disseminated disease, which may be included in the differential diagnoses of gynaecological malignancy (see Table 1).