A new study shows Lunit’s AI-powered cancer diagnostic solutions SCOPE HER2 and SCOPE ER/PR significantly improve pathologist concordance and accuracy, paving the way for improved breast cancer molecular subtype analysis for informed decision-making and precision patient care.
The study, in collaboration with Professor So-Woon Kim from Kyung Hee University College of Medicine and Professor Minsun Jung from Yonsei University College of Medicine, was recently published in Breast Cancer Research (BCR), an international peer-reviewed journal. The study aimed to evaluate the role of Lunit SCOPE in enhancing pathologists' consistency and accuracy of breast cancer molecular subtype classification, including expression levels of human epidermal growth factor receptor 2 (HER2), oestrogen receptor (ER), and progesterone receptor (PR).
Breast cancer treatment strategies and clinical outcome predictions heavily rely on the accurate determination of these receptors' expression levels. However, conventional immunohistochemistry (IHC) analysis poses challenges to classification accuracy. There can be variability in interpretation among pathologists, especially when dealing with intermediate expressions.
The study highlights Lunit SCOPE HER2 and Lunit SCOPE ER/PR, developed using a dataset of thousands of HER2, ER, and PR-stained IHC breast cancer whole-slide images. An external validation cohort of 201 breast cancer cases underwent analysis using these AI analysers.
Results from the study demonstrated a significant increase in agreement among pathologists on the status of HER2, ER, and PR, especially in cases with intermediate weakly positive expressions. Notably, AI assistance led to an increase in agreement on HER2 status from 49.3% to 74.1%, ER from 93.0% to 96.5%, and PR from 84.6% to 91.5%. This improvement was particularly pronounced in intermediate weakly positive cases with HER2 2+ and HER2 1+, showing concordance increases from 46.2% to 68.4% and 26.5% to 70.7%, respectively. Consequently, with AI assistance, the agreement among pathologists in classification of breast cancer molecular subtypes saw an increase from 58.2% to 78.6%.
The study concludes that Lunit SCOPE HER2 and Lunit SCOPE ER/PR significantly improve pathologists' concordance in classifying breast cancer molecular subtypes. These solutions hold immense potential in enhancing treatment strategies and ensuring more accurate and personalised approaches for patients.
"In this study, not only do we validate the performance of Lunit SCOPE HER2 and Lunit SCOPE ER/PR, but we also reveal the potential application of AI biomarkers in predicting treatment responses, particularly for HER2-targeted therapies. We also expect this breakthrough to open new avenues for future drug development," said Brandon Suh, CEO of Lunit. "At Lunit, we enhance healthcare providers' accuracy and concordance, and we empower them to make informed decisions that directly impact patients' lives. Our mission is to transform the landscape of personalized cancer diagnostics and therapeutics."
- Jung M, Song SG, Cho SI, et al. Augmented interpretation of HER2, ER, and PR in breast cancer by artificial intelligence analyzer: enhancing interobserver agreement through a reader study of 201 cases. Breast Cancer Res. 2024;26(1):31. Published 2024 Feb 23. doi:10.1186/s13058-024-01784-y.
