Figure 4
From: Validation of a microRNA-based qRT-PCR test for accurate identification of tumor tissue origin
Classification example. (a) Measured levels (normalized Ct, inversely proportional to log2(abundance)) of hsa-miR-200c and hsa-miR-122 are compared for all training set samples, indicating the left and right branches of node #1 (green circles and red stars, respectively). One metastatic tumor excised from the brain (blue square) originated from a patient that had a concomitant tumor in the lung, and was, therefore, originally diagnosed as a lung cancer. However, this sample showed an uncharacteristic high expression of hsa-miR-122, a strong hepatic marker, and was consequently classified as possibly originating from the liver by the microRNA classifier. On re-examination of the metastatic brain tumor by immunohistochemistry (blinded to the results of the microRNA classifier), this tumor was indeed found to be negative for lung-specific markers: the sample was negative for immunohistochemical staining by both CK7 and TTF1, as well as CK20, CEA, CA125, s-100, thyroglobulin, chromogranin, synaptophysin, CD56, GFAP, calcitonin, and anterior pituitary hormones, whereas staining positive for CAM5.5′ and AE1/AE3. This staining pattern was compatible with hepatocellular carcinoma, prompting further staining for HEPA1 and α-fetoprotein. The tumor stained positive for both stains, consistent with a diagnosis of hepatocellular carcinoma. (b) Hematoxylin–eosin staining (upper panel) showed that the metastasis is composed of sheets of cells with abundant eosinophilic cytoplasm and round to oval nuclei. Among many immunostains used to evaluate the origin of the tumor, HEPA-1 showed strong and specific immunopositivity (lower panel).
