Chromophobe renal cell carcinoma: a comparative study of histological, immunohistochemical and ultrastructural features using high throughput tissue microarray

Aims: In some cases distinction between chromophobe renal cell carcinoma (CRCC), oncocytoma and clear cell (conventional) renal cell carcinoma (eosinophilic variant) using routine light microscopy remains problematic. The present study investigates the level of agreement in the diagnosis of CRCC, as well as the histological features most frequently used for this diagnosis by two pathologists with a special interest in renal neoplasia. The sensitivity and specificity of immunohistochemical markers in cases with overlapping histological features in the diagnosis of CRCC were also studied. Electron microscopy was performed, as a diagnostic gold standard, on all of the cases. Methods and results: Thirty-two renal tumours with predominantly eosinophilic cytoplasm were reviewed in a blinded fashion by two pathologists. The diagnosis and morphological features used to render each diagnosis were tabulated. Validation of the utility of keratin 7 and 20, epithelial membrane antigen (EMA), vimentin, CD10, parvalbumin, RCC antigen, antimitochondrial antibody and Hale's colloidal iron was performed by the construction of a tissue microarray (TMA) master block. Based on histological criteria alone, overall agreement on the diagnosis of these tumours was reached in 69% of the cases, while there was total disagreement in 12%. In 59% of the cases, total agreement was reached in classifying the case as a CRCC based on histology alone. Kappa statistics for interobserver variability were calculated as only slight agreement (kappa = 0.3). The histological features most frequently associated with a diagnosis of CRCC were accentuated cell borders (87%) and a combination of hyperchromatic wrinkled nuclei (79%) and perinuclear halos (74%). The most sensitive and specific marker for CRCC was parvalbumin (sensitivity 0.91; specificity 1.0). The immunohistochemical profile of EMA+/ vimentin- was useful but had low specificity (sensitivity 0.75; specificity 0.4). CD10 had the highest sensitivity (1.0) but worst specificity (0.25) for CRCC. Keratin 7 had high sensitivity (0.83) but fairly low specificity (0.37) for CRCC. Hale's colloidal iron and the RCC antigen marker were not contributory. Finally, the antimitochondrial antibody was found to be fairly sensitive (0.83) for excluding CRCC. Conclusions: A small but significant proportion of renal tumours with cells having eosinophilic cytoplasm cannot be classified, even by experienced pathologists, based on histology alone. In these cases it is imperative to use markers with known sensitivity and specificity for the diagnosis of CRCC.