Abstract 1934: An artificial intelligence-based framework for BLEACH&STAIN mfIHC facilitates automated prognosis marker assessment in breast cancer

Abstract

Abstract The assessment of prognostic markers in routine clinical practice of breast cancer is currently performed using multi gene RNA panels. However, the unknown proportion of normal breast tissue in relation to malignant breast tissue can reduce the predictive value of such tests. Immunohistochemistry holds the potential for a better assessment of tumors because tumor cells can be separately analyzed. To enable automated prognosis marker detection (i.e. HER2, GATA3, progesterone receptor [PR], estrogen receptor [ER], androgen receptor [AR], TOP2A, Ki-67, TROP2, Mammaglobin), we have developed and validated a framework for automated breast cancer identification, which comprises three different artificial intelligence analysis steps and an algorithm for cell-distance analysis of 12+1 marker BLEACH&STAIN multiplex fluorescence immunohistochemistry staining, in 2004 breast cancers. The optimal distance between Myosin+ basal cells and benign panCK+ benign cells was identified as 31µm and used to exclude benign glands from the analysis combined with and deep learning-based algorithm for benign gland detection. Our deep learning-based framework discriminated normal glands from malignant glands with an area under receiver-operating characteristic curves (AUC) of 0.96 (95% confidence interval [CI], 0.92 to 0.99). The accuracy of the approach was also validated by several well-characterized biological findings, such as the identification of 13% HER2+, 73% PR+/ER+, and 14% triple negative cases in the study cohort as well as a significant higher Fraction of TOP2A in the HER2+ cases as compared to the HER2- cases (p<0.001). Furthermore, the automated assessment of GATA3, PR, ER, TOP2A-LI, Ki-67-LI, TROP2, and Mammaglobin was significantly liked to the tumor grade (p<0.001 each). Furthermore, a high expression level of HER2, GATA3, PR, and ER was associated with a prolonged overall survival (p≥0.002 each). In conclusion, a deep learning-based framework for automated breast cancer identification using BLEACH&STAIN multiplex fluorescence IHC facilitates automated prognosis marker quantification in breast cancer. Citation Format: Tim Mandelkow, Elena Bady, Magalie C. Lurati, Claudia Hube-Magg, Maximilian Lennartz, Guido Sauter, Niclas C. Blessin, Ronald Simon. An artificial intelligence-based framework for BLEACH&STAIN mfIHC facilitates automated prognosis marker assessment in breast cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 1934.