ECP 2023 Abstracts

S64 Virchows Archiv (2023) 483 (Suppl 1):S1–S391 13 Background & objectives: Biphasic breast tumours with myoepithe- lial cell (MEC) predominance constitute heterogeneous group, rang- ing from benign adenomyopeithelioma (AME) to malignant AME (M-AME). However, definition of M-AME not clear due to rarity. We reviewed clinicopathological features of AME at our tertiary-care cancer institute. Methods: All 61 AME diagnosed between 2012-2022 were retrieved from departmental archives, of which 43 were benign (70.5%), 18 were M-AME (29.5%). Diagnosis of M-AME was based on increased mitoses (>3/10HPF), marked atypia, prominent nucleoli, necrosis, and infiltration. M-AME were further classified as M-AME in-situ (AME+DCIS), M-AME invasive (malignant epithelial and/or myeoep- ithelial component), and AME with invasive breast carcinoma (IBC). Results: Benign AME: median age 41 years (range:11-67 years), median size 2.6cm (range:0.6-12.4cm), predominantly unilateral (Left:23; right:17; bilateral:3). Core biopsy (CB) diagnosis was AME in 9, while 7 called other benign entities. M-AME: median age 48 years (range 34-90), median size 4cm (range 0.7-22cm), predomi- nantly unilateral (right:9; left:8; 1:bilateral). Among M-AME, 3 were M-AME in-situ, 6 M-AME invasive, while 9 AME with IBC. Of the latter, 6 were hormone receptor (HR)+HER2-; 2 HR-HER2+ and 1 triple negative. CB (8/18) showed M-AME (1), and IBC (4) while rest benign. All AME showed bilayered epithelial-MEC (immunohisto- chemically confirmed). M-AME showed nodal metastasis (3/18), lung metastasis (2/18) & local recurrence (3/18). Eight patients received adjuvant therapy. Conclusion: M-AME are larger and occur at an older age than benign AME. They may be misdiagnosed on CB and carry a potential for metastasis and recurrence. Complete excision with clear margins and adjuvant therapy may be indicated. Recognition of biphasic elements, confirmation with immunohistochemistry, and identification of aggres- sive histology are essential for an accurate diagnosis. PS-01-029 Assessment of intra-laboratory mitoses counting reproducibility in breast cancer using telemedicine platform I. Telezhnikova*, G. Setdikova, A. Abduraimov, S. Khomeriki, L. Zhukova *Loginov Moscow Clinical Research Center, Russia Background & objectives: Counting mitotic figures (MF) in histologic sections is an integral part of the diagnostic pathologist’s tumour evalua- tion. However, level of concordance of MF among pathologists remains low. The aim of this study was to evaluate of level concordance for MF. Methods: The slides were scanned at 40x magnification by 3DHISTECH®. Slides were uploaded to the OneCell telemedicine plat- form for pathology laboratories. For each slide, 10 fields of view 0.264 mm^2 were selected using markup tools. The evaluation was performed by 3 pathologists specializing in breast cancer. Convergence on mitosis detection and grading deviations were assessed. Results: The sample was 100 fields of view from 20 patients. Cat- egories MF were distinguished (1 – mitosis inconclusive, 2 – mitosis yes). MF categories defined as 2 in 32.8% pathologist markup 1, 35.9% pathologist markup 2, 44.3% pathologist markup 3. The mean deviation ± MF between pathologists on the convergence coefficient was 0.66 (0.55±0.78). The total absolute deviation by category on the inverted augmented slides was 0.7 for pathologist markup 1, 1 for pathologist markup 2, and 0.65 for pathologist markup 3. Grade of tumour changed in 8/20 of patients by pathologist markup 2. Overestimation of grading were a statistically significantly difference (higher) than understatement (p < 0.001). Conclusion: Level concordance for counting mitoses among patholo- gists was low, level of category concordance on the inverted aug- mented slides was also low. The authors feel it advisable to conduct training on mitoses counting using telemedicine platforms. Training of MF will improve reproducibility and accuracy. PS-01-030 Presence of crown-like structures in breast adipose tissue; dif- ferences between healthy controls, breast cancer patients and BRCA1/2 gene mutation carriers M. Zwager, I. Holt-Kedde*, H. Timmer-Bosscha, G. de Bock, C. Schröder, B. van der Vegt, M. Arjaans *Department of Plastic Surgery, University of Groningen, University Medical Center Groningen, The Netherlands Background & objectives: Crown-like structures (CLS) might influ- ence breast cancer behaviour. Therefore, we compared CLS presence between adipose tissue of breast cancer patients, BRCA1/2 gene muta- tion carriers and healthy controls, and assessed the relation of CLS numbers with clinical outcome. Methods: Breast adipose tissue sections from 259 breast cancer patients, 78 BRCA1/2 gene mutation carriers after mastectomy and 48 healthy controls after breast reduction were immunohistochemically stained for CD68. CLS presence and index (CLS/cm2) were determined and cor- related with BMI, BRCA status, tumour presence and intrinsic tumour subtype with logistic regression analyses. Correlations with tumour char- acteristics and clinical outcome were assessed. Results: CLS were more often present in breast cancer patients com- pared to BRCA carriers and healthy controls (OR:3.2; 95% CI:1.4-7.5; p=0.005). In the total study population, CLS presence was associated with the presence of breast cancer (OR:2.2; 95% CI:1.3-3.8; p=0.005) and high BMI (overweight OR: 2.1; 95% CI:1.3-3.6; p=0.004; obese OR:3.3; 95% CI:1.8-5.8; p<0.001). The same was found in the breast cancer subgroup (overweight OR:2.3; 95% CI:1.2-4.2; p=0.010; obese OR:4.2; 95% CI:2.1-8.4; p<0.001). No association with BRCA status was found. CLS presence was higher in Luminal-B-like tumours com- pared to the other subtypes (OR:2.5; 95% CI=1.3-5.2; p=0.010). In TNBC, CLS were related to lymphovascular invasion (p=0.013). No association with survival was found. Conclusion: In conclusion, CLS were more frequently present in breast adipose tissue of breast cancer patients compared to BRCA1/2 gene mutation carriers and healthy controls. Furthermore, our study provides further evidence of the association between obesity and presence of CLS. The prognostic significance and impact on clinical outcome of differ- ences in CLS numbers should be further assessed in prospective studies. PS-02 | Poster Session Digital and Computational Pathology PS-02-001 Automatic discrimination between neuroendocrine carcinomas and neuroendocrine tumours of grade 3 by deep learning analysis of H&E staining images A. Arrieta*, J. Bohoyo, M.J. Araúzo Bravo, M.R. Bella, J.C. Ferreres *Consorci Corporació Sanitària Parc Taulí, I3PT, Universidad Autónoma de Barcelona, Spain Background & objectives: Deep learning (DL) is a machine learning technology based on artificial neural networks (ANN). We have shown DL as an aid to make a selection between Neuroendocrine tumours (NETs) and Neuroendocrine carcinomas (NECs) fromNeuroendocrine neoplasms (NENs). Methods: 95 cases of NENs were selected from 2015 to 2018 from Parc Tauli Hospital database. These cases were used to develop, train and vali- date an ANN using the Keras DL learning Application Program Interface written in Python, running on top of the DL platform TensorFlow on a High Performance Cluster with 10 Graphic Processing Units