Machine learning for Improving Risk Stratification of Hepatocellular Carcinoma (HCC) Patients Treated with Immune Checkpoint Inhibitors (ICIs)

Hepatocellular carcinoma (HCC) is the most common form of liver cancer and a major cause of cancer-related deaths around the world. Many people diagnosed with HCC at an advanced stage, facing limited treatment options and poor outcomes. Immunotherapy using immune checkpoint inhibitors (ICIs), such as atezolizumab plus bevacizumab, significantly improves patient survival and has been recommended as the standard-of-care first-line treatment for advanced HCC patients since 2020. ICIs are a type of cancer treatment that help the body’s own immune system fight cancer. Normally, immune cells have built-in “brakes” (called checkpoints) that prevent them from attacking healthy tissues. Cancer cells can take advantage of these brakes to avoid being destroyed. ICIs work by releasing these brakes, allowing immune cells to better recognize and attack cancer cells. However, not all patients respond to ICIs in the same way, and doctors currently have few tools to predict who will benefit most.

In this study, we will use a form of artificial intelligence called unsupervised machine learning to better understand differences among patients with HCC. Specifically, we will use a method called “consensus clustering analysis”, which helps group patients based on shared traits such as age, disease stage, or lab results —without us having to define those groups ahead of time. This type of analysis can reveal patterns that might not be obvious using traditional methods.

We will apply this method to data from one large clinical trial (i.e. IMbrave150 trial) that tested the new treatment for HCC using immune checkpoint inhibitors (atezolizumab plus bevacizumab). By identifying subgroups of patients with similar characteristics, we aim to see if some of these groups had better or worse outcomes with treatment, such as longer survival or fewer side effects.

Our goal is to use these insights to guide more personalized treatment in the future. If we can predict which types of patients benefit most from certain therapies, doctors may be able to choose more effective treatments from the beginning, improving care and quality of life for people living with liver cancer.