CD1a immunopositivity in perivascular epithelioid cell neoplasms: true expression or technical artifact? A streptavidin-biotin and polymer-based detection system immunohistochemical study of perivascular epithelioid cell neoplasms and their morphologic mimics

Abstract

Perivascular epithelioid cell neoplasms comprise a family of rare neoplasms composed of morphologically distinctive perivascular epithelioid cells exhibiting a "myomelanocytic" immunophenotype. The distinction of perivascular epithelioid cell neoplasms from other tumors with melanocytic and smooth muscle differentiation can be difficult. A recent study has suggested that perivascular epithelioid cell neoplasms routinely express CD1a, a Langerhans cell-associated transmembrane glycoprotein involved in antigen presentation and that expression of this marker may be helpful in the distinction of perivascular epithelioid cell neoplasms from various mimics. We evaluated a series of perivascular epithelioid cell neoplasms and potential mimics for CD1a expression. A total of 54 cases (27 perivascular epithelioid cell neoplasms, 11 leiomyosarcomas, 10 melanomas, 6 clear cell sarcomas) were evaluated in 2 laboratories (Mayo Clinic Rochester: 31 cases, Carolinas Medical Center: 23 cases). Selected positive cases were retested at Carolinas Medical Center (11 cases) and Mayo Clinic Rochester (10 cases). Mayo Clinic Rochester methods were as follows: MTB1 clone (1:20, Novocastra, Newcastle-upon-Tyne, UK), heat-induced epitope retrieval in EDTA (pH 8.0), and Dako Advance detection system (Dako Corp, Carpinteria, CA) with background-reducing diluent. Carolinas Medical Center methods were as follows: MTB1 clone (1:30; CellMarque, Rocklin, CA), heat-induced epitope retrieval in Medium Cell Conditioner #1 (pH 8.0-9.0), and streptavidin-biotin detection system with diaminobenzidine chromogen, with and without biotin blocking. Scores were as follows: 1+, 5% to 25%; 2+, 26% to 50%; and 3+, more than 51%. Langerhans cells served as a positive internal control in all tested cases. All Mayo Clinic Rochester cases were negative. Sixteen Carolinas Medical Center perivascular epithelioid cell neoplasms (14 renal angiomyolipomas, 1 soft tissue perivascular epithelioid cell neoplasm, 1 pulmonary clear cell "sugar" tumor) showed CD1a immunopositivity (1+: 7 cases; 2+: 7 cases; 3+: 2 cases) when tested without biotin blocking, 11 of these cases were retested with biotin blocking and were negative. All non-perivascular epithelioid cell neoplasms were negative. All positive perivascular epithelioid cell neoplasms showed cytoplasmic staining only, without membranous staining. Ten Carolinas Medical Center positive perivascular epithelioid cell neoplasms were negative when retested a Mayo Clinic Rochester, using a polymer-based detection system. We conclude that perivascular epithelioid cell neoplasms do not truly express CD1a in a biologically plausible membranous pattern, but may instead show aberrant cytoplasmic immunopositivity in some laboratories. Close inspection of published photomicrographs of previously reported CD1a-positive perivascular epithelioid cell neoplasms shows an identical pattern of cytoplasmic positivity, likely reflecting abundant endogenous biotin within perivascular epithelioid cell neoplasm cells. We do not believe that there is a role for CD1a immunohistochemistry in the differential diagnosis of perivascular epithelioid cell neoplasms.