Utility of immunohistochemistry in distinguishing ovarian sertoli-stromal cell tumors from carcinosarcomas

Abstract

Poorly differentiated Sertoli-stromal cell tumors and carcinosarcomas of the ovary both show biphasic epithelial and stromal patterns and may both show heterologous stromal elements, presenting a difficult differential diagnosis. We studied the immunohistochemical profile of Sertoli cell differentiation in human testes and applied these findings to the ovarian tumors. Eleven Sertoli-stromal cell tumors, six carcinosarcomas of the ovary, and 11 testes (six fetal, one infant, and four adult) were studied using antibodies to cytokeratin AE1:AE3 (AE1:3), cytokeratin CAM 5.2 (CAM), epithelial membrane antigen (EMA), vimentin, desmin, muscle-specific actin (MSA), S-100 protein (S-100), CA 19-9, CA 125, carcinoembryonic antigen monoclonal (CEA-M), carcinoembryonic antigen polyclonal (CEA-P), and placental alkaline phosphatase (PLAP). In the fetal testes, immature gonadal stroma and sex cord areas stained with vimentin (six of six cases), AE1:3 (five of six cases), and CAM (six of six cases). Sertoli cells in immature gonadal stroma areas, sex cords, and seminiferous tubules of normal fetal, infant, or adult testes never showed immunoreactivity for EMA, S-100, CA 19-9, CA 125, CEA-M, CEA-P, or PLAP. All Sertoli-stromal cell tumors stained with AE1:3 and CAM in areas of Sertoli cell differentiation (11 of 11 cases) but did not stain with EMA, PLAP, CEA-P, CEA-M, CA 19-9, CA 125, or S-100 (none of 11 cases). Carcinosarcomas expressed AE1:3 and CAM in all epithelial areas (six of six cases) and most stromal areas (five of six cases). Carcinomatous areas of carcinosarcoma also showed immunoreactivity for EMA (six of six cases), CA 125 (two of six cases), PLAP (two of six cases), CEA-P (two of six cases), and CEA-M (one of six cases), while stromal areas of carcinosarcoma expressed EMA (four of six cases) and S-100 (four of six cases). Heterologous stromal elements were present in three of 11 Sertoli-stromal cell tumors (two showed skeletal muscle and one showed both skeletal muscle and cartilage differentiation) and in four of six carcinosarcomas (one skeletal muscle, one cartilage, and two cartilage and skeletal muscle). All skeletal muscle heterologous elements expressed desmin, vimentin, and MSA. The heterologous cartilage in carcinosarcoma stained with S-100 (three of three), while the one case of heterologous cartilage in Sertolistromal cell tumor did not. These results suggest that ovarian Sertoli-stromal cell tumor can be distinguished from carcinosarcoma by the absence of staining for EMA, PLAP, CEA, CA 125, or CA 19-9 in epithelial areas of Sertoli-stromal cell tumor. Unlike carcinosarcoma, stromal areas of Sertoli-stromal cell tumor did not express S-100, even the one case containing heterologous cartilage. While we noted a lower disease-free survival rate in carcinosarcomas compared with Sertoli-stromal tumors, we studied too few poorly differentiated Sertoli-stromal tumors to evaluate the clinical utility of distinguishing this subgroup from carcinosarcomas.