Lunit Presents 7 Research Achievements at ASCO, the World's Largest Cancer Conference

Medical AI company Lunit announced on the 27th that it will present seven research results utilizing its AI biomarker platform 'Lunit Scope' at the 2024 American Society of Clinical Oncology (ASCO) conference, held in Chicago, USA, from the 31st of this month to the 4th of next month. ASCO is considered one of the world's top three cancer societies, along with the American Association for Cancer Research (AACR) and the European Society for Medical Oncology (ESMO).

Since 2019, Lunit has participated annually in ASCO to share the latest research results in the field of cancer treatment. The main research presented at this ASCO focuses on classifying breast cancer patient groups with ultra-low expression of the human epidermal growth factor receptor 2 (HER2), one of the best-known targets for anticancer therapy. Recently, it has been revealed that HER2-targeted antibody-drug conjugates (ADCs) show therapeutic effects even in HER2-low breast cancer, making it important to identify the ultra-low HER2 expression group among breast cancer patients previously classified as HER2-negative. HER2 expression levels are divided into four stages from 0 to +3, with 0 or +1 classified as negative and +3 as positive.

Using Lunit Scope, Lunit analyzed tissue slides from 401 breast cancer patients classified as HER2-negative and found that 23.6% of patients previously scored as 0 by conventional methods could be considered part of the ultra-low HER2 expression group. Additionally, among patients read as 1+ expression, HER2 expression levels could be more finely distinguished, with 51.9% of these patients showing higher HER2 expression patterns. Notably, this proportion is similar to the objective response rate (ORR) of 52.3% observed in clinical trials of Enhertu, an ADC anticancer drug, targeting HER2-low breast cancer patients. Through this, Lunit suggests that AI-based subdivision of breast cancer patients previously classified as HER2-negative into an ultra-low expression group can expand the pool of patients eligible for targeted therapy.

Lunit will also present results predicting immune checkpoint inhibitor responses in non-small cell lung cancer patients through deep learning-based chest computed tomography (CT) image analysis. An AI model trained on approximately 2,000 patients predicted treatment responders whose risk of switching to other treatments after treatment failure and mortality risk both decreased by 42%, and the median overall survival (mOS) was 16.5 months, more than twice that of the non-responder group at 7.6 months. Overall survival (OS) refers to the duration from treatment initiation to death.

In particular, when combining the newly developed CT AI model with the existing Lunit Scope IO model and the known programmed cell death protein (PD)-(L)1 biomarker, the accuracy of treatment response prediction improved further. Patients predicted as responders by all three models showed an increase in mOS up to 32 months after immune checkpoint inhibitor treatment.

Additionally, Lunit plans to present research results on ▲ prognosis prediction in hormone receptor-positive early breast cancer patients using AI ▲ phase 1a clinical trial results of an anti-CD47 drug candidate in collaboration with a domestic immuno-oncology developer ▲ prediction of immune checkpoint inhibitor responsiveness in malignant melanoma patients using Lunit Scope IO ▲ and AI-based prediction of tertiary lymphoid structures (TLS) in non-small cell lung cancer.

Seobum Seok, CEO of Lunit, stated, "Consistently presenting cancer diagnosis and treatment research results using AI technology at ASCO, the world's most prestigious cancer society, demonstrates that Lunit is leading the global medical AI market. We will continue to closely communicate with medical professionals and strive to advance AI technology reflecting clinical needs."