Seoul, South Korea, October 14, 2026
Lunit, a company specializing in artificial intelligence (AI) for cancer diagnostics and precision oncology, has reported new research findings showing that AI-powered pathology analysis may reveal differences in HER2 expression associated with treatment response in advanced biliary tract cancer. The findings were published in Nature Medicine and are based on exploratory biomarker analyses from the Phase 1b/2 HERBOT clinical trial. Using its Lunit SCOPE AI-powered pathology platform, the company assessed whole-slide tissue images to quantify the proportion of tumor cells expressing high levels of human epidermal growth factor receptor 2 (HER2). The analysis identified an association between the proportion of strongly HER2-positive tumor cells and response to a four-drug treatment regimen. The findings suggest that quantitative AI-based tissue analysis may provide additional biological information beyond conventional HER2 classification, potentially supporting future research into precision oncology and personalized cancer treatment. However, further validation is necessary before this approach can be established as a reliable clinical decision-making tool.
AI Analysis Identifies Differences in HER2 Expression
HER2 is a protein involved in cell growth and signaling. In some cancers, increased HER2 expression or gene amplification contributes to tumor development and can provide a target for specific cancer therapies. Conventional HER2 testing commonly classifies tumors according to immunohistochemistry (IHC) staining intensity and, where appropriate, additional molecular testing. However, tumors can contain different populations of cells with varying levels of HER2 expression. This biological variation, known as tumor heterogeneity, may help explain why patients with apparently similar HER2 classifications sometimes experience different treatment outcomes. Lunit SCOPE HER2 uses AI to analyze digitized pathology slides and classify individual tumor cells according to HER2 staining intensity. This enables researchers to quantify the proportion of cells with strong HER2 expression, designated HER2 3+, across the analyzed tissue sample. In the study, researchers examined whether this quantitative measurement was associated with treatment response among patients with HER2-positive advanced biliary tract cancer. The AI analysis found that patients whose tumors contained at least 10% HER2 3+ tumor cells achieved an objective response rate of 80%, compared with 40% among patients below that threshold. The difference was statistically significant, with a reported P value of 0.0138. These findings suggest that measuring the distribution and intensity of HER2 expression may provide additional information that conventional categorical testing does not fully capture.
HERBOT Trial Evaluates Four-Drug Cancer Treatment
The analysis used data from the multicenter Phase 1b/2 HERBOT trial, which enrolled 40 patients across 12 sites in South Korea. The study evaluated a combination of four treatments as a potential first-line approach for patients with HER2-positive advanced biliary tract cancer. The treatment regimen combined trastuzumab, a HER2-targeted therapy, with nivolumab, an immune checkpoint inhibitor, and the chemotherapy agents gemcitabine and cisplatin. The combination was investigated to determine whether targeting HER2 while also stimulating immune activity and controlling tumor-cell growth could improve treatment outcomes. Across the study population, the regimen achieved an objective response rate of 55% and median progression-free survival of 10.6 months. Objective response rate measures the proportion of patients whose tumors shrink by a predefined amount or disappear following treatment. Progression-free survival measures the time patients live without their disease worsening or progressing. The researchers also evaluated the tumor immune microenvironment using Lunit SCOPE IO, another component of the company’s AI pathology platform. This analysis classified tumors according to immune characteristics, including immune-inflamed, immune-excluded and immune-desert patterns. The findings indicated that both quantitative HER2 expression and immune-related characteristics were associated with treatment response. These observations could help researchers investigate how tumor biology influences the effectiveness of combination therapies.
Advertisement
Findings Support Further Precision Oncology Research
The study highlights the potential value of integrating AI-powered pathology, biomarker quantification and tumor microenvironment analysis into oncology research. Instead of relying exclusively on broad biomarker categories, researchers may be able to examine more detailed patterns within tumor tissue and investigate their relationship with clinical outcomes. For Lunit, the findings demonstrate how its Lunit SCOPE platform can be applied to whole-slide image analysis in clinical research. The company continues to develop AI-based tools intended to support biomarker discovery, patient stratification and precision oncology research. Nevertheless, the results should be interpreted cautiously. The analysis involved a relatively small patient population, and the reported associations do not establish that quantitative HER2 assessment can independently predict treatment benefit in other settings. Larger studies and independent validation will be needed to determine whether the approach can reliably guide treatment selection. Lunit SCOPE products are intended for research use only and are not intended for use in diagnostic procedures or clinical decisions, according to the company. The findings therefore represent a research milestone rather than a new approved diagnostic method. Overall, the Nature Medicine study illustrates how AI-based pathology analysis may help researchers understand differences in tumor biology that are not fully captured by conventional testing. Continued validation could clarify whether quantitative HER2 measurement can contribute to more individualized treatment strategies for patients with advanced biliary tract cancer and other HER2-associated malignancies.
Source: Lunit press release
