Lunit to Unveil AI Biomarker Studies at ESMO 2026, Including HER2, PD-L1 and Pancreatic Cancer Analyses

Lunit said it will unveil five research abstracts using its AI biomarker platform, Lunit Scope, at the 2026 European Society for Medical Oncology (ESMO 2026), which runs from the 23rd to the 27th. The studies cover breast cancer, lung cancer and pancreatic cancer, and examine the potential applications of AI biomarkers.

The presentations also include research related to the global HER2 analysis project, AIforHER. AIforHER is a project led by the French cancer institute ICO with participation from pharmaceutical companies and medical institutions including AstraZeneca and Daiichi Sankyo.

The goal of AIforHER is to compare and evaluate multiple AI solutions for the HER2 expression levels of breast cancer patients, and to verify whether they can be used to identify patients eligible for treatment. At this conference, Lunit plans to present the first results on HER2-ultralow breast cancer in poster form.

HER2-ultralow has recently been added to the indications for ADC treatment. As a result, the importance of accurate assessment in the HER2-ultralow area is growing.

After the first study evaluation, Lunit was selected as one of two companies to participate in the follow-up study. It then has been validating the solution in a real-world clinical setting. In this validation, it is confirming whether AI can help identify HER2-ultralow patients.

Separately, Lunit also unveiled a comparative study on PD-L1 expression assessment in patients with non-small cell lung cancer. The study was led by a project from Japan's National Cancer Center. The subjects were surgical specimens from 993 patients enrolled in the project.

Readings were conducted independently by pathologists and Lunit Scope PD-L1. The overall agreement rate was 73%. The agreement rate in the 50% or higher PD-L1 expression range was 80%, while the agreement rate in the 1% to 49% range was 95%. The agreement rate in the below-1% range was 51%.

There were 211 cases in which pathologists assessed the level as below 1%, while AI assessed it as 1% or higher. Of those 211 cases, AI classified 204 as weakly positive in the 1% to 10% range. The researchers judged that the range where assessment differences were concentrated was the 1% threshold that separates PD-L1 expression from non-expression. They said further confirmation is needed to determine how those assessment differences affect actual treatment outcomes.

In one of the studies scheduled for presentation, Lunit evaluated pathology images from 181 patients with pancreatic ductal adenocarcinoma using its analysis tool, Lunit Scope IO. After calculating the stromal ratio within the tumor tissue and the vascular endothelial cell ratio within the stroma, the researchers compared prognostic differences by KRAS G12D mutation, other mutations, and non-mutation types.

Overall, the analysis found that patient groups with higher stromal ratios had shorter progression-free survival. Median overall survival by KRAS mutation type was 43.8 months for the non-mutation group, 19.5 months for the other mutation group, and 15.1 months for the G12D mutation group.

However, an exception was identified in the G12D mutation group. In this group, the researchers found that patients with more stroma had longer progression-free survival. The researchers said that both gene mutations and the tumor-associated stromal environment need to be considered when predicting treatment efficacy in pancreatic ductal adenocarcinoma.

Based on these studies, Lunit plans to expand the use of AI-based pathology analysis to patient selection for treatment and prognosis prediction. It also plans to continue validating the technology in real-world clinical settings through cooperation with global pharmaceutical companies and research institutions.

CEO Seo Beom-seok said the studies presented this time focused on examining AI's role in areas that are difficult to judge through microscopic interpretation alone, such as HER2-ultralow. He added that Lunit will continue to work with global pharmaceutical companies and research institutions so that Lunit Scope can be used to precisely stratify patients eligible for cancer treatment.

Source: TECHWORLD · Kim Seung-gi
Original: https://www.epnc.co.kr/news/articleView.html?idxno=408488

References

• https://www.epnc.co.kr/news/articleView.html?idxno=408488