□ The research team led by Senior Researchers Yoonhee Lee from the Division of Biomedical Technology and Gyogwon Koo from the Division of Intelligent Robot at DGIST (under President Kunwoo Lee) has developed a technology that distinguishes lung cancer gene mutations solely by measuring the "stiffness" of exosomes—tiny particles released from cancer cells in the bloodstream—using atomic force microscopy (AFM). Their study enables rapid and precise analysis of individual exosomes and is expected to advance into a new liquid biopsy-based diagnostic technique for lung cancer.
□ Non-small cell lung cancer (NSCLC) is the most common type of lung cancer, accounting for more than 85% of all cases. However, because it rarely shows noticeable symptoms in the early stages, it is often diagnosed at an advanced stage, making treatment difficult. NSCLC continues to have a high mortality rate, and developing new diagnostic technologies that enable early detection and treatment remains a significant challenge in the medical field. In particular, conventional tissue biopsies place a considerable burden on patients and have limitations in repeated testing. As such, non-invasive liquid biopsy technology utilizing blood-derived information has recently attracted attention.
□ The research team led by Senior Researchers Yoonhee Lee and Gyogwon Koo at DGIST isolated exosomes from NSCLC cell lines with distinct genetic mutations (A549: KRAS mutation, PC9: EGFR mutation, PC9/GR: EGFR-resistant mutation). Using AFM, the team measured nano-scale physical properties of individual exosomes at high resolution, including surface stiffness and height-to-radius ratios.
□ They found that the exosomes derived from A549 cells exhibited significantly higher stiffness, which indicated that alterations in the cell membrane lipids caused by KRAS mutations were also reflected in the exosomes. In contrast, exosomes derived from PC9 and PC9/GR cells showed similar properties, suggesting a correlation with their shared genetic background. These findings demonstrate that the physical properties of exosomes vary depending on the genetic mutations of the cancer cells from which they originate.
□ To precisely classify these nanomechanical characteristics of exosomes, the research team employed AI technology. Height and stiffness data of exosomes obtained through AFM were visualized and used to train a deep learning-based convolutional neural network (DenseNet-121) model to classify their cell lines of origin. Exosomes derived from A549 cells were distinguished with a remarkably high accuracy of 96%, and the overall average AUC reached 0.92. This demonstrates the potential of a next-generation liquid biopsy platform capable of high-precision classification based solely on the physical properties of exosomes, without the need for fluorescent labeling.
□ Senior Researchers Yoonhee Lee and Gyogwon Koo stated, "This study presents a new diagnostic potential to distinguish lung cancer with specific genetic mutations using only a small amount of exosome samples. We plan to actively pursue the practical application of this technology by integrating a high-speed AFM platform in clinical sample validation."
□ This study was conducted with Senior Researcher Yoonhee Lee from the Division of Biomedical Technology and Gyogwon Koo from the Division of Intelligent Robot at DGIST serving as corresponding authors, and postdoctoral researcher Soohyun Park from the Division of Biomedical Technology and Dr. Youngkyu Kim from Columbia University as co-first authors. The research findings were published online in the renowned chemistry journal Analytical Chemistry on July 8, 2025.
