ECP 2023 Abstracts

S58 Virchows Archiv (2023) 483 (Suppl 1):S1–S391 13 Tissue biopsies (24) and plasma (8) served for DNA/RNA extrac- tion: RecoverAll/Total/Nucleic/Acid/Isolation/kit and Thermo-Fisher, MagMAXCell-free/Total/Nucleic/Acid/Kit. DNA/RNA concentration: Invitrogen, Qubit/RNA/HS/Assay/kit, Qubit-1X-ds-DNA-HS-Assay and Qubit-4-Fluorometer. RNA reverse-transcription: Superscript-IV-Vilo- Master-Mix. Libraries: Oncomine-Focus-Assay-Chef-Ready-Library and Oncomine-Pan-Cancer-Cell-Free-Assay. Ion-GeneStudio-S5-Prime- sequencer and Ion-Reporter-software served for pathogenic-variants (PV)and variants-of-uncertain-significance (VUS) selection. Results: Age of patients:38-86years, average of 56,2, present previous clinic and imaging and Ethic Committee approval. All 24 cases had G2/ G3 mammary invasive carcinomas NST, Nottingham score 6-9; molecu- lar subtypes: 8 cases LuminalA (LA), 9-LuminalB (LB) (Ki67>25%), 2-LuminalB-like(LB-l), 3-Her2-enriched(Her2-enr) and 2-triple negative(TN). Thirteen tissue biopsies had Her2-low. NGS analysis showed PV/VUS in 14 cases. From the 6 PV, 4 are PV-PIK3CA(3-LA,1- LB,1-LB-l,1-her2-enr) and 2, AKT(1-LA,1-LB), and all in Her2-low cases. The 2 PV-PIK3CA in liquid biopsies are present also in the cor- responding tissue biopsies. The VUS- RET, SMO, MYC, appeared in tissue biopsies; a VUS-TP53 appeared in a plasma LB case, PV-negative. The PV-PIK3CA and PV-AKT do not appear together having actually common signalling pathway (PI3K/AKT/mTOR). Conclusion: These preliminary results are suggesting that NGS posi- tive PV-PIK3CA cases could be a therapy target in early breast cancer. The PV-PIK3CA appearing in the tissue and liquid biopsies of the same 2 patients, is showing that liquid biopsy is a promising tool in cancer monitoring. The association PV-PIK3CA and hormonal receptor in our cases is similar to other studies; PV-PIK3CA and PV-AKT apparently are in mutual exclusion. Her2 low cases, unexpectedly, had the major part of PV in our series. The poster was supported by the Project PN-III-P2-CERC-CO- PTE-3-2021, ONCOGUARD (Validation and implementation of the liquid biopsy as an actual method of NGS molecular profile assessment for the prognostic and therapy modulation in breast cancer), Bucuresti PS-01-007 Successful deployment of an artificial intelligence solution for primary diagnosis of breast biopsies in clinical practice M. Aslam*, A. Gunavardhan, A. Phillips, A. Savage, P. Huey, R. Nicholson, I. Krasnitsky, D. Ignatov, M. Grinwald, M. Vecsler *Betsi Cadwaladr University Health Board, NHS Wales, United Kingdom Background & objectives: We present the analysis of a clinically deployed artificial intelligence (AI) decision support solution for breast biopsies primary diagnosis utilized as first read within a digital pathol- ogy workflow in a hospital pathology laboratory setting. Methods: 7 pathologists underwent training and used the solution for prospective primary diagnosis of consecutive breast biopsies, report- ing on 355 cases (460 H&E slides), 178 cancer (50%) and 177 benign (50%). All the slides were digitized and blindly processed by the AI solution. Pathologists were presented with the AI pre-classifications on invasive, in-situ and other features, reviewed and reported the cases. Results: The AI solution demonstrated high performance when pre- classifying cases with high likelihood for cancer, with AUC = 0.99 (95% CI: 0.988, 0.999), NPV = 99% (95% CI: 0.956, 1) and PPV = 84% (95% CI: 784, 0. 888), respectively. 19 % of parts have been clas- sified as suspicious by AI. The AI performance for invasive carcinoma was very high with AUC = 0.99 (95% CI: 0.988, 0.999), NPV = 99% (95% CI: 0.956, 1), and PPV = 91% (95% CI: 0.858, 0.950), respec- tively. Additional performance metrics on detection of DCIS and other features will be presented. Conclusion: We report here the successful implementation of a multi- feature AI solution that automatically imparts clinically relevant diag- nostic parameters regarding breast cancer such as subtypes/grades of invasive and in-situ cancers and other pathologic features. The solu- tion demonstrated the ability to accurately pre-classify cancer and con- tribute to workflow efficiency and that it be used as a significant aiding tool for pathologists in decision-making in routine pathology practice. PS-01-008 Impact of analytic factors of immunohistochemistry (IHC) on the incidence of human epidermal growth factor receptor 2 (HER2)-low breast cancer Y.K. Bae*, S.Y. Kwon, M.C. Kim *Department of Pathology, Yeungnam University College of Medicine, Daegu, Republic of Korea Background & objectives: HER2-low breast cancers (IHC 1+ or 2+ without gene amplification) have demonstrated efficacy for the newly developed HER2-targeting antibody-drug conjugates (ADCs). We investigated the frequency of HER2-low breast cancer and compared the IHC conditions used in different laboratories. Methods: We retrospectively reviewed primary breast cancers diagnosed at the Yeungnam University Hospital (YUH) and Keimyung University Dongsan Hospital (KUDH) between January 2022 and December 2022. The data on specimen type, HER2 status (IHC and in situ hybridization (ISH) results), and IHC conditions were collected. HER2 status was defined according to the most updated ASCO/CAP HER2 testing guidelines. Results: The pathology laboratories at YUH and KUDH performed HER2 IHC staining using the 4B5 antibody and HER2 ISH testing with the INFORM® HER2 dual ISH DNA probe (Ventana). However, there were differences in the IHC conditions, including antigen retrieval time (16 min vs 30 min), incubation time (28 min vs 32 min), tempera- ture (42˚C vs 37˚C), type of detection kit (OptiView vs UltraView), and staining platform between the two laboratories. Of the 623 cases diagnosed at YUH, 465 (74.6%) were HER2-negative, 81 (13%) were HER2-low, and 77 (12.4%) were HER2-positive. Of the 290 cases diag- nosed at KUDH, 135 (46.6%) were HER2-negative, 82 (28.3%) were HER2-low, and 73 (25.2%) were HER2-positive. Conclusion: The frequencies of HER2-positive and HER2-low breast cancers showed a difference of nearly two times between the two labo- ratories. We adjusted staining conditions using PATHWAY HER2 4 in 1 control slides to achieve the intended staining. Unlike HER2 positiv- ity, there is no gold standard or reflex test to confirm HER2-low expres- sion. Before conducting clinical trials or applying the new ADCs, each pathology laboratory should verify IHC conditions to accurately iden- tify HER2-low breast cancer. PS-01-009 HER2-low breast cancer and response to neoadjuvant chemotherapy: a population-based cohort study X. Baez Navarro*, C. van Deurzen, M.R. van Bockstal *Erasmus MC, The Netherlands Background& objectives: The clinical impact of low expression of HER2 (HER2-low) breast cancer (BC) and its relation to response to neoadjuvant chemotherapy (NAC) is not well understood. We aimed to study the asso- ciation between HER2-low BC and the pathological response on NAC. Methods: Data from the Dutch Pathology Registry was collected from BC patients treated with NAC between 2014-2022. HER2-low BC was considered as immunohistochemical (IHC) score of 1+ or 2+ with negative reflex test. We compared clinicopathologic features of HER2-0 versus HER2-low BC and assessed the correlation between HER2 status and the pathological complete response (pCR) rate after NAC, including overall survival. Results: In total 11,988 patients were retrieved. Among oestrogen receptor (ER)+ tumours (n=6,886), 67% were HER2-low, compared to 46% (n=2566) in the ER- group. Around 32% (n=207) of patients had