[790] Signature Markers of Renal Epithelial Neoplasms by Hierarchical Clustering Analysis

KA Kasper, Q Tao, C Luan, JJ Wei, XJ Yang. Northwestern University Feinberg School of Medicine, Chicago, IL

Background: Distinguishing different types of renal epithelial tumors is important for clinical management; however, sometimes it is difficult to discriminate by histology alone. Many immunohistochemical markers for renal epithelial tumors have been reported for tumor subclassification. A few have been reported to be differentially expressed in different subtypes of renal cell carcinoma (RCC), but a single marker is not sensitive or specific enough to discriminate each subtype. In this study, we hypothesized that a panel of antibodies would enhance the ability of pathologists to make more accurate diagnoses of renal tumors.
Design: Tissue microarray sections of renal tumors, including 32 clear cell RCCs (ccRCC), 25 papillary RCCs (papRCC), 15 chromophobe RCCs (chRCC), & 15 oncocytomas, & 6 normal kidneys, were stained with the following panel of markers that have been previously reported to show relative specificity for certain renal tumors: glutathione S-transferase (GST- ), carbonic anhydrase 9 (CA9), alpha-methylacyl-CoA racemase (AMACR), vimentin, C-kit, & cytokeratin 7 (CK7). Hierarchical cluster analysis was performed on immunostained sections based on the staining intensity scanned from ChromoVision Automatic Cellular Imaging System. Pearson correlation and average linkage were used to generate protein and sample trees using software originally designed for analyzing gene microarray data. A dendrogram and Treeview were generated by an unsupervised cluster analysis to establish tumor signatures of selected markers. The proximity of the immunostaining levels in each tumor can be visualized.
Results: A Treeview illustrated three clusters: ccRCC, papRCC, & chRCC with oncocytomas. The majority of the ccRCCs clustered together (31/32); 12 of 25 papRCCs clustered together. The majority of the chRCCs clustered together (10/15), & the remaining ones clustered with oncocytomas. This is not unexpected as chRCCs & oncocytomas are genetically related tumors, & a good marker to differentiate them has not been fully characterized.
Conclusions: We illustrated that a panel of 6 immunohistochemical stains can be used to establish marker signatures to differentiate the most common subtypes of renal epithelial tumors. Using hierarchical clustering, we separated the distinct clusters of renal neoplasms. This method of using marker signatures provides an objective measure of a renal tumor and may assist surgical pathologists in making more accurate diagnoses. Further study is needed to better assess the clinical utility of this panel of markers.
Category: Genitourinary (including renal tumors)

Tuesday, March 10, 2009 9:30 AM

Poster Session III # 80, Tuesday Morning