Author(s): Carvalho JC
AIMS: To develop an immunohistochemical strategy for distinguishing renal oncocytoma (RO) from the eosinophilic variant of chromophobe (ChRCC), and papillary (PRCC) and clear cell (CRCC) renal cell carcinoma containing eosinophilic cytoplasm in core biopsy specimens. METHODS AND RESULTS: Cluster analysis was performed on immunohistochemical data from 21 RO, 16 ChRCC, 16 CRCC and 20 PRCC patients. A panel of CK7, C-kit, S100A1 and vimentin clustered into four groups. Cluster A (94% ChRCC) expressed C-kit and CK7 and lacked S100A1 and vimentin. Cluster B (95% RO) expressed C-kit, S100A1, focal CK7 (single or small clusters of cells) and lacked vimentin. Cluster C comprised a mixture of PRCC and CRCC with no expression of C-kit or CK7 and variable S100A1 and vimentin. PRCC with strong expression of CK7 clustered into group D. A panel of S100A1 (positive) and focal CK7 expression distinguished RO from ChRCC with 91% sensitivity and 93% specificity. A panel of vimentin (negative) and C-kit (positive) distinguished RO from CRCC with 83% sensitivity and 86% specificity and RO from PRCC with 79% sensitivity and 88% specificity. CONCLUSIONS: Hierarchical cluster analysis is an effective approach to analyse high-volume immunohistochemical data to generate an optimal panel in the differential diagnosis of oncocytoma from its mimics.