ECP 2022 Abstract Book

Virchows Archiv (2022) 481 (Suppl 1):S1–S364 13 Background & objectives: Heterogeneity of metastatic pattern of breast cancer hampers treatment decisions. Large scale studies into underlying biological and patient specific factors are necessary to progress for future treatment of metastatic diseases. Methods: Pathological records of 4831 patients diagnosed with breast cancer who underwent autopsy between 1974 and 2010 were retrieved from Dutch nationwide pathology databank (PALGA). Natural language processing (NLP) methods were applied to extract data from autopsy reports. Named entity recognition was based on SNOMED codes that are utilized in PALGA. Results: The accuracy of data retrieval with NLP was above 0.9 and recall was 0.94 for majority of cases. Our model outperformed manual extraction of value of interest. We identified 2622 (54.2%) of patients with metastatic disease. Invasive ductal carcinoma, and mucinous carcinoma more frequently metastasized to lung and liver, whereas for invasive lobular carcinoma and mixed type this was to bone and liver respectively. There was no statistically significant association between lateralization and metastatic pat- terns, except for kidney (right= 5.0%, left= 8.9%, bilateral= 13.3%, p= 0.007). In a subgroup of patients, we found that (ER+/HER2+) patients, were more likely to metastasize to liver and bone, com- pared to (ER-/HER2+) patients. Conclusion: This is the first large-scale study that demonstrates artificial intelligence methods are efficient for extracting informa- tion from Dutch pathology reports. We show differences in fre- quencies and combinations of metastatic sites between histological subtypes. The patterns and frequencies identified in this autopsy study may reflect the underlying biology of metastatic breast can- cer and potentially influence the future follow-ups and patient-tai- lored treatment strategies depending on other clinical correlations. OFP-10-003 Automated prognosis marker assessment in 2,004 breast cancers using an artificial intelligence-based framework for BLEACH&STAIN mfIHC T. Mandelkow, E. Bady, J.H. Müller, N.F. Debatin, M.C.J. Lurati, C. Hube-Magg, N.C. Blessin, G. Sauter* *University Medical Center Hamburg, Germany Background & objectives: Prognostic markers in routine clinical practice of breast cancer are currently assessed using multi-gene panels. However, the fluctuating tumour purity can reduce the predictive value of such tests. Immunohistochemistry holds the potential for a better risk assessment. Methods: To enable automated prognosis marker detection (i.e. HER2, GATA3, progesterone- [PR], oestrogen- [ER], and andro- gen receptor [AR], TOP2A, Ki-67, TROP2), we have developed and validated a framework for automated breast cancer identifica- tion, which comprises three different artificial intelligence analysis steps and an algorithm for cell-distance analysis of 11+1 marker BLEACH&STAIN multiplex fluorescence immunohistochemistry (mfIHC) staining in 2’004 breast cancers. Results: The optimal distance between Myosin+ basal cells and benign panCK+ cells was identified as 25 μm and used to exclude benign glands from the analysis combined with several deep learn- ing-based algorithms. Our framework discriminated normal glands from malignant glands with an AUC of 0.96. The accuracy of the approach was also validated by well-characterized biological find- ings, such as the identification of 13% HER2+, 73% PR+/ER+, and 14 triple negative cases. Furthermore, the automated assessment of GATA3, PR, ER, TOP2A-LI, Ki-67-LI and TROP2 was signifi- cantly liked to the tumour grade (p<0.001each). Furthermore, a high expression level of HER2, GATA3, PR, and ER was associated with a prolonged overall survival (p≥0.002 each). Conclusion: A deep learning-based framework for automated breast cancer identification using BLEACH&STAIN multiplex fluorescence IHC facilitates automated prognosis marker quanti- fication in breast cancer. OFP-10-004 Upgrade rate and predictive factors for benign breast intra- ductal papilloma on core biopsy in Vancouver, Canada T. Salisbury*, S. Koonmee, L. Ali, O. Ondic, R. Bhan, K. Pivo- varcikova, A. Gurung, R. Alaghehbandan *Department of Pathology, Faculty of Medicine, University of Brit- ish Columbia, Royal Columbian Hospital, Vancouver, BC, Canada Background & objectives: The management of benign intraductal papillomas (IDPs) diagnosed on core biopsy is controversial. We aim to determine the upgrade rate of IDPs diagnosed on core biopsy in subsequent surgical excision specimens and to identify associated clinical, pathologic, and radiologic factors. Methods: This is a retrospective population-based study of all breast papillary lesions diagnosed on core biopsy from 2017-2019 in Fraser Health Authority in Greater Vancouver, Canada. Patient demograph- ics, histopathologic, and radiologic findings were analysed. Upgrade was defined as atypical ductal hyperplasia (ADH), ductal carcinoma in situ (DCIS), encapsulated papillary carcinoma (EPC), solid papil- lary carcinoma (SPC), and invasive carcinoma on surgical excision. Results: A total of 129 patients with benign IDPs diagnosed on core biopsy were included. The overall upgrade rate to atypia or malignancy was 9.3% (12/129) on final excision. This included 7 with ADH, 7 with DCIS, EPC or SPC, and 1 with invasive carci- noma. Predictors of upgrade included older age (55.6 vs 66.1 years, p < 0.0001) and larger lesion size (11.1 vs 15.1 mm, p = 0.001). Older age (≥ 55 years) (OR [95%CI] 5.3 [1.04-27.08]) was an independent predictor of upgrade. In our study, location (central vs peripheral) and BI-RADS radiologic category were not associated with predicting upgrade. Conclusion: Our findings support surgical excision of IDPs diag- nosed on core biopsy in women aged 55 years or older with large lesions, while a conservative approach (close clinical follow-up) may be warranted for younger women with smaller lesions. OFP-10-005 Evaluation of dual colour-dual in-situ hybridization (D-DISH) for HER2/neu testing in breast cancer S. Desai*, A. Rathi, P. Shah, A. Sahay, A. Patil, T. Shet *Tata Memorial Centre, Homi Bhabha National Institute, India Background & objectives: To standardize and validate the HER2 Dual ISH DNA Probe Cocktail (D-DISH) assay by VENTANA for HER2/neu testing in breast cancer using FISH as the gold standard and to assess the interobserver variability in interpreting D-DISH. Methods: HER2/neu IHC, FISH, and D-DISH assay by HER2 Dual ISH DNA Probe Cocktail Assay (Ventana Medical Systems, Inc., Tucson) were performed on 120 breast carcinoma cases. Sensitiv- ity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy were calculated against D-DISH results for four pathologists (FISH-gold standard). The absolute agreement between FISH and D-DISH ratios, average signals was assessed. Results: A concordance of 98.3% was observed between FISH and D-DISH assays. D-DISH showed a sensitivity of 95.92%, speci- ficity of 100%, PPV of 100%, and NPV of 97.26%, with respect to FISH. Cohen’s kappa statistic was 0.96 demonstrating perfect agreement. Intraclass correlation coefficient (ICC) values of 0.97 (HER2), 0.92 (CEP17), and 0.97 (HER2/CEP17) were observed. Interobserver variability showed an almost perfect agreement S40