Figure 2

Image analysis workflow for immunohistochemical staining quantification. (A-D) Prostate cancer tissue microarrays were stained by immunohistochemistry. The method of image analysis performed independently for each marker was dependent on the staining pattern: default malignant epithelial area (A), malignant epithelial nuclei (B), tumor-associated stromal area (C), or tumor-associated vasculature (D). (E-G) Genie Histology Pattern Recognition software (Aperio) subclassified tumor areas into malignant epithelium (dark blue), stroma (yellow), and glass (light blue). (H) Markers with heterogeneous positivity were evaluated by Color Deconvolution (Aperio) to quantify staining within malignant epithelial areas. (I) Markers with predominantly nuclear localization (CCND1, MKI67, SIAH2, and TP53) were evaluated by the Nuclear algorithm (Aperio) to quantify staining within nuclei of malignant epithelium. (J) Markers with significant stromal positivity (EI24 and HA) were evaluated by Color Deconvolution (Aperio) to quantify staining within tumor-associated stroma. (K) The microvascular marker (CD34) was evaluated by the Microvessel algorithm (Aperio) to quantify additional metrics including average vessel area, average vessel perimeter, average lumen area, average vascular area, and microvessel density. Scale bars represent 50 μm.