Immunohistochemical and Functional Biomarkers of Value in Female Genital Tract Lesions

To the Editor: All of us owe a debt of gratitude to Professor McCluggage for providing a comprehensive review on the use of immunohistochemistry in diagnosis in the female genital tract and, in particular, for summarizing the use of these reagents in distinguishing primary ovarian cancer from metastases from the bowel (1). This is a particular problem that we encounter often locally. There has been much debate in the past as to whether systematic review of the literature with statistical meta-analysis allows similar conclusions as narrative review (2). I decided to undertake such a systematic review and to compare the results with Prof. McCluggage's conclusions by querying the EMBASE and Medline databases to identify all those studies in which the expression of antibodies to cytokeratins 7 (CK7) and 20 (CK20), carcinoembryonic antigen, Ca125, CDX2, villin, and beta-catenin as detected immunohistochemically on histological sections was examined in cancer of the bowel and the ovary. The use of these antibodies is specifically advocated by McCluggage (1). In addition, a summary of antibody expression in human tumors prepared by a histopathologist for colleagues was accessed to identify further articles (available at www.immunoquery.com). The abstract accompanying each reference online was accessed to identify the studies in which tumors of both primary types had been investigated. The results from each of these studies were entered onto a Review Manager 4.1 statistical program; then, the relative risk was calculated. Because in many studies, the various histological types of ovarian cancer were not distinguished or the endometrioid and serous subtypes were combined, a comparison of expression in ovarian cancer of all types was made with colon cancer in addition to comparing the expression in colonic carcinoma with that in endometrioid and mucinous lesions of the ovary in the small number of studies where both were available, as in McCluggage's review. The results of the systematic review and meta-analysis are presented in Table 1. As indicated, many of the articles describing the reaction of these 8 antibodies did not distinguish between the different histological types; these data are included under the heading of Ovary (All Types) in the table. Some articles do describe the reaction of one or more of these 8 antibodies, specifically the endometrioid, mucinous, and metastatic ovarian cancers; where available, these results are presented in the table under these separate headings. The calculation of the relative risk allows the likelihood of encountering a positive reaction with a particular antibody in cancer of the bowel to be compared with the likelihood of encountering a positive reaction with the same antibody in cancer of the ovary. Thus, a relative risk of 1 suggests that a positive or negative reaction is equally likely in tumors from each organ and that it is of little discriminative value in distinguishing between them. As the relative risk gets greater or becomes less than 1, the likelihood of the reaction being encountered in tumors from one organ increases and the likelihood of the reaction occurring in tumors from the other organ diminishes; consequently, the diagnostic value of the antibody in discriminating between them improves.TABLE 1: The expression of a variety of antigens as detected in cancer of the ovary compared with the expression of these antigens in primary colonic carcinoma*In conclusion, the findings largely endorse Prof. McCluggage's conclusions and support his choice of antibodies as being of specific value in distinguishing tumors originating in the 2 body sites. They are also reassuring in that they support the use of narrative review by an expert in the field in providing a synthesis of the experimental results in the literature that can then be applied in diagnostic practice. The systematic review was limited in this situation by the small numbers of studies in which the expression of these antibodies was studied in tumors and in the failure to distinguish between the different histological types of ovarian cancer. Reassuringly, the inclusion of studies based on microarray did not seem to alter the conclusions. Mark K. Heatley, M.D. Department of Histopathology St. James' University Hospital Beckett Street, Leeds West Yorkshire LS9 7TF1