The Immunoscore in Colon Cancer (CC): Comparison of the digital workflow to conventional histology for the quantification of CD3 and CD8 positive and tumor-infiltrating lymphocytes C. Geppert1, J. Kolwelter1, S. Nolte1, J. Galon3, F. Pages4, A. Lugli2, T.T. Rau2 , A. Hartmann1 1Institute of Pathology, University Hospital, Erlangen, Germany, 2Institute of Pathology, University Bern, Bern, Switzerland 3Laboratory of Integrative Cancer Immunology, Cordeliers Research Center, Paris, France 4Service d'Immunologie Biologique, Hôpital Européen Georges Pompidou, Paris, France Hole-slide-imaging, scanned in 20x; Mirax Midi, 3D-Histech. Detection of tissue and segmentation of tissue regarding heterogenity CD3 /CD8 Segmentation Detection CD3/CD8 Aims: The Immunoscore project is an ongoing international study including 23 centers worldwide validating the prognostic impact of CD3 and CD8 in colon cancer. As a participating center our institute accomplished all the steps included in the workflow. The aim of this study is to compare a complete digital pathology approach to classical histological methods. Slide annotation with tumor-area & normal tissue; definition of the invasive margin by software (standard of thickness 600µm). Methods: The Immunoscore is defined by the density of CD3 positive T-lymphocytes and CD8 positive cytotoxic T-lymphocytes in the tumor center (TC) and the invasive margin (IM) of colon cancer using whole slides imaging of tumor slides. The digital workflow includes the following steps: 1) Slides are scanned and digitally segmented into small areas by the Definiens Architect software; 2) the annotation must be performed by a pathologist; 3) the invasive margin is calculated and the overall stained lymphocytes including the hot spots (best of 5 tiles counted) are quantified automatically by software; 4) validation of the selected hot spots and cell detection by a pathologist; 5) calculation of optimal cut-offs and Immunoscore-analysis. IM CD3 IM CD8 Performed immunohistochemistry of CD3 & CD8 e.g. in hot–spots: in invasive margin (IM) and in the tumor-center (TC) with detection and counting of positive immune-cells (red); tumorcells were not counted (20x). tumor negative CD3/CD8 positive Results: The average number of the digitally quantified CD3 and CD8 positive lymphocytes in the included 447 colon cancer-cases were for CD3 (in 5 tiles): 4688 cells for TC and 4137 cells for IM and for CD8: 2013 cells for TC and 1677 cells for IM, respectively. The average of lymphocyte density for CD3 was 2147 cells/cm2 with a range of 91 cells to 33890 cells (TC) and 2237 cells/cm2 with a range of 31 cells to 18188 cells (IM) and for CD8 (in 5 tiles) was 2080 cells/cm2 (TC) with a range of 108 cells to 71212 cells (TC) and 911 cells/cm2 with a range of 193 cells to 14882 cells (IM). The area that was analyzed depends on each tumor slide of the individual case. The average of the analyzed area was for CD3/CD8 2,17 cm2 (TC) and 1,84 cm2 (IM). TC CD8 TC CD3 Validation of selected hot-spots and cell count: Brown-intensity of positive cells for CD3/CD8 show normal separation part of cells in tumor (TC + IM) Intensity of stained cells in tumor (CT +IM) Conclusions: The digital workflow proposed by the Immunoscore task force is a reliable basis for an accurate quantification of CD3 and CD8 positive lymphocytes in colon cancer compared to a semi-quantitative scoring system used in conventional histology like 10HPF or estimation of the percentage of cells. Therefore, a possible perspective is the implementation of the Immunoscore workflow in the daily surgical pathology practice. These analyses can provide additional aspects like heterogeneity of the immune response. The prognostic power of the Immunoscore is based on this accurate quantification approach (prognostic data sets with significance in KM- curves – data not shown – multi-center data under publication). Hole-slide imaging and analysis of CD3/CD8 can provide a heat-map for validating the heterogenity of the immune reaction done by Definiens architekt64 ©. www.fau.de/einrichtungen/fakultaeten/med