Lunit Predicts Immunotherapy Response in Colorectal and Renal Cell Cancer with AI Analysis

Presentation at ESMO 2025

Medical artificial intelligence (AI) company Lunit announced on October 13 that it will present three studies on predicting immunotherapy response using its AI biomarker platform, Lunit SCOPE, at the European Society for Medical Oncology (ESMO 2025) to be held in Berlin, Germany, from October 17 to 21.

The oral presentation study, conducted in collaboration with Professor Chiara Cremolini's team from the Department of Oncology at the University of Pisa, Italy, focuses on the development of a biomarker to predict the efficacy of combination therapy by adding the immunotherapy drug atezolizumab (Tecentriq) to FOLFOXIRI (FOLFIRINOX) plus bevacizumab (Avastin) in patients with proficient mismatch repair metastatic colorectal cancer (pMMR mCRC).

The research team analyzed tissue slides from 161 patients using Lunit SCOPE to quantify the density of six types of cells, including lymphocytes and tumor cells. Based on this data, the AI classified patients into Group A, with high biomarker levels, and Group B, with lower levels; approximately 70% of all patients (113 individuals) were assigned to Group A. The analysis showed that among patients who received the atezolizumab combination therapy, Group A experienced a statistically significant improvement in both progression-free survival (PFS) and overall survival (OS) compared to Group B.

Notably, there was no significant difference between Group A and Group B in patients who received chemotherapy alone. However, when atezolizumab was added to the combination therapy, only Group A showed a treatment benefit, confirming that Lunit SCOPE serves as a predictive biomarker specific to immunotherapy. To independently validate the predictive power of this biomarker, the same analysis was performed using data from 48 patients who underwent combination therapy with FOLFOXIRI, cetuximab (Erbitux), and avelumab (Bavencio). In this group, patients with higher biomarker levels demonstrated improved progression-free survival and overall survival compared to those with lower biomarker levels.

This study demonstrated that AI analysis can identify a subgroup of patients with proficient mismatch repair metastatic colorectal cancer-who typically have limited response to immunotherapy-that is more likely to benefit from such treatment. In a second study conducted with the Yonsei Cancer Center research team, the effectiveness of immunotherapy combination (nivolumab plus ipilimumab) was compared to targeted therapy alone (sunitinib) in patients with advanced clear cell renal cell carcinoma. Using Lunit SCOPE, the researchers analyzed the spatial distribution of tumor-infiltrating lymphocytes (TILs) within tumor tissues and classified patients as either 'inflamed' or 'non-inflamed' based on immune activity. Among the 125 patients who received the nivolumab (Opdivo) plus ipilimumab (Yervoy) combination, the inflamed group showed an objective response rate (ORR) of 60.5%, outperforming the non-inflamed group's 23.2%.

In contrast, among 128 patients who received sunitinib (Sutent) monotherapy, there was no observed difference in treatment efficacy based on immune phenotype. This study suggests that AI-based immune phenotype analysis could serve as a valuable predictive tool in determining first-line treatment strategies for patients with advanced renal cell carcinoma.

Seo Bumseok, CEO of Lunit, stated, "The research presented by Lunit at this year's ESMO demonstrates, with data, that AI can be linked to actual improvements in patient survival. Given that our findings have been validated across different cancer types-including colorectal, renal cell, and non-small cell lung cancer-we expect Lunit SCOPE to become a key tool for providing personalized treatment strategies in a wide range of cancers."

This content was produced with the assistance of AI translation services.