ECP 2022 Abstract Book

Virchows Archiv (2022) 481 (Suppl 1):S1–S364 13 cytometric evaluation of full blood count for immune cell popula- tion assessment. Results: The NLR>/=5 group showed significant depression of some T -cell subgroups, activated natural killer cells and eosino- phils and significant elevation of memory B cells and neutrophils. Significant depression of the cytokines IL-2, IL-1β, IL-7, IL-13, basic FGF, IFN-γ and MIP-1α occurred in the NLR>/=5 group Conclusion: The down regulation of IFN-γ, IL-2 and IL7, involved in immunoregulation in the NLR>5 group suggests the failure of a coherent, T-cell driven, Th1-polarised response to cancer. The pan-T cell lymphocytopaenia seen in the NLR>5 group dem- onstrates an inappropriate T-cell driven immune response to tumour. The increased neutrophilia seen in the NLR≥5 group may relate to tumour microenvironmental factors that make them pro-tumourigenic. Funding: Mr Pine was supported by a grant from Leeds Cares. Dr Cummings was supported by a Wellbeing of Women project grant. Dr Orsi is supported by an NIHR clinical lectureship. Professor Quirke is supported by programme grants from Yorkshire Cancer Research (YCR) L386 and by an NIHR Senior Investigator award. OFP-01-014 An innovative artificial intelligence based application for the differentiation of invasive adenocarcinoma from pseudoinvasion in colorectal polyps L. Chen*, Z. Yang, B. Wang, D. Driman, C. Ling, Q. Zhang *Department of Laboratory Medicine and Molecular Diagnostics, Sunnybrook Health Sciences Centre, University of Toronto, Canada Background & objectives: Dysplastic glands misplaced in the submucosa (pseudoinvasion) of colorectal polyps sometimes can- not be differentiated from invasive adenocarcinoma, even by expert pathologists. We are developing an innovative application to dif- ferentiate invasive adenocarcinoma from pseudoinvasion on whole- slide images (WSI). Methods: Under low power (x2) magnification, we trained the algorithm to identify areas of interest (dysplastic glands) on WSI. Using our images and NCT-CRC-HE-100K datasets, we developed a combined non-patch and patch-based algorithm to identify 12 morphological categories in differentiating pseudoinvasion from invasive adenocarcinoma (true invasion). We used two models to aggregate the per-patch/area classification results into a final classification. Results: Slides from 130 colon polyps (70 pseudoinvasion and 60 true invasion) were digitized. Low power detection successfully identified all areas of interest. The convolutional neural network (CNN) model achieved an overall accuracy of 98% in recognizing and classifying each area/patch into categories. Based on the 9 categories currently completed, the linear model and the 3-layer CNN model show accuracy rates of 83% and 88% respectively in the final classification (true versus pseudoinvasion). Conclusion: Our AI based application mimics how pathologists work and has achieved reliable results with 9 of 12 categories used. We are confident that with the other 3 categories added into the algorithm, reliability will be increased. The algorithm will be validated using another 60 cases and diagnostic accuracy will be compared with expert GI pathologists blinded to the results. The application will be available on our website for pathologists to access worldwide (http://ai4path.ca/ ). Funding: Ontario Molecular Pathology Research Network CPTRG Pathology Image Analysis Fund OFP-01-015 MMR-deficient crypts detection by immunohistochemistry in normal colonic mucosa of patients with MMR deficient (dMMR)/MicroSatellite Instable (MSI) colorectal cancer: a helpful tool for the diagnosis of Lynch syndrome M. Svrcek*, S. Breton, I. Sourrouille, N. Basset, E. Guillerm, I. Bro- cheriou, D. Enea, P. Bourgoin, A. Duval, T. Andre, M. Muleris, F. Coulet *Sorbonne Université, AP-HP, Hôpital Saint-Antoine, Department of Pathology, France Background & objectives: The diagnosis of Lynch syndrome (LS) is challenging. We investigated whether the detection of MMR-deficient non-neoplastic intestinal crypts could be helpful in patients with inconclusive germline mutation studies of MMR genes (Lynch-like syndrome; LLS) or variants of uncertain sig- nificance (VUS). Methods: We evaluated the expression of MMR proteins in non-neo- plastic mucosa of colorectal cancers from patients with LS (n=15), including 7 with multiple CRCs (mRCCs), LLS (n=7) and VUS (n=7). Ten immunohistochemistry (IHC) slides were performed on 1 or 2 blocks of both adjacent (adj-muc) and distant (dist-muc) mucosa. A crypt ratio (number of dMMR crypts/total number of crypts) was determined. Results: dMMR crypts were identified in 12/15 patients with LS (80%). dMMR crypts tended to be more frequently observed in dist-muc [11/15 (73%) vs 8/14 (57%), p=0.157], and in mRCC [7/7 (100%) vs 5/8 (63%), p=0.244]. The ratio of dMMR crypts was significantly higher in patients with mCRC [mCRC-ratio = 0.00337 vs non-mCRC-ratio = 0.0005 (p=0.003)]. A minimum number of n=8 IHC slides analysed in adj-muc and n=6 slides in dist-muc identified all patients with dMMR crypts and 66.7% and 91.7% respectively of all patients with LS. dMMR crypts were identified in 1/7 patients with LLS (14%) and in 4/7 patients with VUS (57%). Conclusion: The detection of dMMR crypts, with a minimum number of 6 IHC slides in dist-muc, could be an integral part of the decision-making algorithm for the diagnosis of LS in patients with LLS or VUS. OFP-01-016 Tumour budding is an independent prognostic factor in stage III colon cancer patients: a post-hoc analysis of the IDEA- France phase III trial (PRODIGE-GERCOR) M. Svrcek*, D. Basile, C. Broudin, J. Emile, A. Falcoz, F. Pagès, L. Mineur, J. Bennouna, C. Louvet, P. Artru, S. Fratte, F. Ghiringhelli, T. Andre, V. Derangère, D. Vernerey, J. Taieb *Sorbonne Université, AP-HP, Hôpital Saint-Antoine, Department of Pathology, France Background & objectives: Tumour budding (Bd) is an emerging prognostic biomarker in colon cancer (CC). We explored further the significance of Bd for risk stratification by evaluating survival of stage III CC patients included in the IDEA-France phase III trial. Methods: Bd was assessed on scanned HE-stained slides and scored by central review by the Bd criteria of the 2016 Interna- tional Tumour Budding Consensus Conference (ITBCC2016) and classified as Bd1 (0-4 buds/0.785 mm2), Bd2 (5-9 buds), and Bd3 (≥10 buds) categories. Disease-free survival (DFS) and overall survival (OS) were analysed by log-rank test. Clinicopathologic features and Immunoscore® were correlated with Bd. Results: Samples of 1048 CC patients were analysed. Overall, Bd1, Bd2, and Bd3 were observed in 39%, 28%, and 33% of CC, respec- tively. Bd2 and Bd3 were associated with vascular (P = .002) and S5