New Artificial Intelligence Analytic Tool



Professors of The Chinese University of Hong Kong’s (CUHK) Faculty of Medicine (CU Medicine) have developed an AI-powered analytic tool to predict the effectiveness of a complementary biomarker for immune checkpoint inhibition in non-small-cell lung cancer (NSCLC) in collaboration with South Korea’s Seoul National University College of Medicine, Sungkyunkwan University School of Medicine, Ajou University School of Medicine, and an AI start-up.


The program uses whole-slide images to do a spatial analysis of the distribution of tumor-infiltrating lymphocytes biomarkers, which can be used to predict treatment results with immune checkpoint inhibitors, which are now the first-line therapy for advanced NSCLC.


According to the findings, AI-powered analysis of tumor-infiltrating lymphocytes correlates with tumor response and progression-free survival, suggesting that it could be used to improve therapy choices in clinical practice. The Journal of Clinical Oncology just published these findings.


For advanced NSCLC with PD-L1 expression, immune checkpoint drugs are the standard treatment. Lung cancer is one of the most frequent malignancies in the world, with 1.8 million people dying from it each year. Lung cancer is the primary cause of cancer fatalities and the most frequent cancer in Hong Kong, according to the city’s cancer registry, with more than 5,000 new cases each year. NSCLC is responsible for more than 80% of all lung malignancies.


Anti-PD-1 checkpoint drugs, also known as immune checkpoint inhibitors, are a conventional first-line therapy for advanced NSCLC with PD-L1 expression. When the immune cell T-cell protein PD-1 attaches to the cancer cell ligand PD-L1, it stops the immune system from destroying cancer cells. The inhibitors function by stopping the immune system from adhering to cancer cells, allowing it to do its job and eliminate cancer cells.


According to Professor Tony Mok, the Li Shu Fan Professor of Clinical Oncology and Chairman of the Department of Clinical Oncology at CU Medicine, the outcome of anti-PD-1 checkpoint inhibitors may vary depending on the patient’s tumor microenvironment, but there is currently no standard biomarker that addresses this.


In theory, tumor-infiltrating lymphocytes are the principal activator of antitumor immunity, making them a viable biomarker for predicting treatment outcomes with immune checkpoint inhibitors, according to him. However, the existing approach of quantification is time-consuming and relies on the geographical distribution in whole-slide photographs, which limits its applicability and objectivity.


Immunological phenotypes linked to immune checkpoint inhibitor responsiveness in tumors. The combined research team’s AI-powered spatial tumor-infiltrating lymphocytes analyzer can segment and quantify several histologic components from whole-slide images, including cancer epithelium, cancer stroma, and tumor-infiltrating lymphocytes.


The three immunological phenotypes inflamed, immune-excluded, and immune-desert are then constructed via deep learning-based AI model training. In comparison to individuals with the other two phenotypes, the inflamed immune phenotype is associated with increased local immune cytolytic activity, a higher response rate, and longer progression-free survival.


This is the first study using AI-powered automated tumor-infiltrating lymphocytes analysis in advanced NSCLC, according to Professor Mok. The researchers demonstrated in this study that an AI-powered spatial tumor-infiltrating lymphocytes analyzer can predict clinical outcomes of immune checkpoint inhibitors in NSCLC patients. This could be used as a supplement to the present PD-L1 assessment, which measures tumor percentage.