Immunohistochemical phenotype of ovarian granulosa cell tumors: absence of epithelial membrane antigen has diagnostic value

• Published in: Human pathology Vol. 25; no. 1; p. 60
• Main Authors: Costa, M J, DeRose, P B, Roth, L M, Brescia, R J, Zaloudek, C J, Cohen, C
• Format: Journal Article
• Language: English
• Published: United States 01-01-1994
• Subjects: Adolescent; Adult; Aged; Aged, 80 and over; Antigens, Neoplasm - metabolism; Child; Diagnosis, Differential; Female; Granulosa Cell Tumor - metabolism; Granulosa Cell Tumor - pathology; Humans; Immunohistochemistry; Membrane Glycoproteins - metabolism; Middle Aged; Mucin-1; Ovarian Neoplasms - metabolism; Ovarian Neoplasms - pathology; Phenotype
• ISSN: 0046-8177
• DOI: 10.1016/0046-8177(94)90172-4

Granulosa cell tumors (GCTs) represent 1.5% to 3% of primary and 6% to 10% of malignant ovarian neoplasms, and present little diagnostic difficulty in the typical case; however, other primary or metastatic tumors may mimic their various histologic patterns. For this reason, immunohistochemistry can be used to supplement routine histology to help determine a final tissue diagnosis. Previous reports on the utility of antibodies to intermediate filaments vary, as some investigators found keratin to be uniformly negative in GCTs while others reported immunoreactivity for keratin in 20% to 68% of cases. To determine the immunophenotype of granulosa cell tumors and to discover which antibodies are useful in differentiating GCTs from histologic look-alikes, we studied 52 GCTs, including 24 typical cases, 23 cases in which the diffuse pattern predominated, and five juvenile cases, with a panel of commercially available antibodies using an automated immunohistochemistry system. Immunoreactivity for granulosa cells in GCTs was as follows: 17 cases (32.7%) reacted with cytokeratin AE1/AE3, six cases (11.5%) reacted with cytokeratin MAK-6, three cases (5.8%) reacted with cytokeratin CAM 5.2, no case (0%) reacted with epithelial membrane antigen, 52 cases (100%) reacted with vimentin, no case (0%) reacted with desmin, 48 cases (92.3%) reacted with smooth muscle actin, and 26 cases (50%) reacted with S-100 protein. No attempt was made to quantify staining of background thecoma-like or fibroma-like elements in GCTs. Immunoreactivity was independent of the histologic subtype of GCT. Cytokeratin immunoreactivity showed a globoid pattern of staining and was consistent with the expression of 52.5-kD and 45-kD cytokeratins (8 and 18 of Moll's classification). For this reason, the presence of cytokeratin immunoreactivity by itself cannot be used to differentiate a primary or metastatic carcinoma from a GCT. The presence of smooth muscle actin and the absence of epithelial membrane antigen immunoreactivity are additional features that are characteristic of a GCT. S-100 protein immunoreactivity is a finding limited exclusively to GCTs among sex cord stromal tumors, and its presence may have some role in differentiating between Sertoli-stromal cell tumors and GCTs. Since epithelial membrane antigen immunoreactivity is present in many of the histologic look-alikes of GCTs, such as metastatic or primary carcinoma, the absence of staining in GCT has diagnostic value.