Figure 1
From: Validation of a microRNA-based qRT-PCR test for accurate identification of tumor tissue origin
(a–d) Development of a diagnostic test for determining tissue of origin: data from two array platforms (a, b) was used to select a subset of 104 candidate microRNAs that were further investigated by qRT–PCR in a training set (c); a standardized test was developed, measuring expression of 48 microRNAs, and was validated on an independent set of 204 blinded FFPE tumor samples (d). Graphs in each box show separation of epithelial from non-epithelial samples (node #5 in the binary decision tree; Figure 2) using hsa-miR-200c and hsa-miR-148b in different datasets, obtained using different platforms and sample sets. Data from the preliminary study26 (a) and from the custom arrays (b) is shown in normalized fluorescence units on a logarithmic scale. Data from the qRT–PCR training set (c) and test validation (d) is shown in inverted normalized Ct, proportional to log2(abundance). The gray area (with higher levels of hsa-miR-200c) marks the region classified as epithelial (left branch) at this node. The classification rule determined for the qRT–PCR training set data (c) was used as is for the validation dataset (d).
