M K Heatley1. 1. Department of Histopathology, St James' University Hospital, Leeds, UK. markkheatley@hotmail.com
Abstract
AIMS: To compare the relative risk of antigen expression being detected immunohistochemically in ovarian and gastric carcinoma aggregated from studies performed for diagnostic purposes, with the relative risks of their expression in all patients in the English literature. METHODS AND RESULTS: Both types of series indicated that cytokeratin (CK) 7 expression was greater and that of CK20 and carcinoembryonic antigen less in ovarian than in gastric carcinoma (P < 0.05). Synthesis of all data available for MUC-2 suggested it was more commonly expressed in ovarian carcinoma, whereas the relative risk in papers that directly compared its expression suggested that it was more common in the gastric carcinoma (P = 0.2, NS). Aggregating all possible data suggested villin was more likely to be expressed in ovarian cancers, whereas studies in which its expression was compared directly in both tumours suggested the opposite. Although statistically significant, patient numbers were small. CONCLUSION: Provided sufficient numbers of cases are studied, analysis of studies comparing antigen expression for diagnostic purposes in tumours from two body sites is likely to be supported in the wider literature. The design of such comparative studies is informed by aggregating data from single tumour studies.
AIMS: To compare the relative risk of antigen expression being detected immunohistochemically in ovarian and gastric carcinoma aggregated from studies performed for diagnostic purposes, with the relative risks of their expression in all patients in the English literature. METHODS AND RESULTS: Both types of series indicated that cytokeratin (CK) 7 expression was greater and that of CK20 and carcinoembryonic antigen less in ovarian than in gastric carcinoma (P < 0.05). Synthesis of all data available for MUC-2 suggested it was more commonly expressed in ovarian carcinoma, whereas the relative risk in papers that directly compared its expression suggested that it was more common in the gastric carcinoma (P = 0.2, NS). Aggregating all possible data suggested villin was more likely to be expressed in ovarian cancers, whereas studies in which its expression was compared directly in both tumours suggested the opposite. Although statistically significant, patient numbers were small. CONCLUSION: Provided sufficient numbers of cases are studied, analysis of studies comparing antigen expression for diagnostic purposes in tumours from two body sites is likely to be supported in the wider literature. The design of such comparative studies is informed by aggregating data from single tumour studies.