Samsung Medical Center said Tuesday that researchers have developed an artificial intelligence-based technology capable of analyzing the prognosis of gallbladder cancer patients, a breakthrough that could help advance personalized precision treatment for one of the deadliest forms of cancer.
The research team included professors Park Ju-kyung, Lee Gyu-taek and Choi Young-hoon from the hospital’s gastroenterology department, professor Kim Hong-beom from hepatobiliary-pancreatic surgery, and researcher Kim Hye-min from the Early Diagnosis Team for Refractory Cancer at the Future Medicine Research Institute.
The team said it successfully demonstrated that AI-powered spatial analysis technology can analyze the tumor microenvironment (TME) of gallbladder cancer patients and predict cancer recurrence and survival outcomes.
The gallbladder, commonly known as the bile sac, stores bile produced by the liver. Gallbladder cancer is often detected only after the disease has significantly progressed because symptoms are usually absent in the early stages, making it one of the most difficult cancers to treat, similar to pancreatic cancer.
According to Korea’s Central Cancer Registry, the five-year relative survival rate for gallbladder and other biliary tract cancers stands at 29%, second only to pancreatic cancer at 17% among major cancers with the lowest survival rates.
Gallbladder cancer is also known for its high recurrence rate even after surgery, underscoring the importance of accurately predicting each patient’s recurrence risk and survival probability when establishing treatment strategies.
Using data from 225 gallbladder cancer surgery patients and an external validation group of 41 additional patients, the research team developed an AI-based prediction model.
The model quantified and incorporated key indicators of the tumor microenvironment, including the density of tumor-infiltrating lymphocytes (TILs), the number of tertiary lymphoid structures (TLS), and fibroblast density surrounding cancer cells.
Researchers identified three major factors linked to poor prognosis: low TIL density, low TLS count, and high fibroblast density. As the number of these risk factors increased, both overall survival (OS) and disease-free survival (DFS) periods declined sharply.
Compared with the group possessing all three risk factors, patients with none of the risk factors showed recurrence and mortality risks that were 87% and 80% lower, respectively. The researchers also observed that differences in recurrence and mortality risks gradually narrowed as the number of risk factors increased.
Kim said gallbladder cancer is particularly difficult to predict among biliary tract cancers because of its extremely poor prognosis.
“This study opens the door to using AI technology for highly precise analysis of gallbladder cancer prognosis,” Kim said.
Park said the study confirmed AI’s potential as a “digital biomarker” capable of deeply analyzing the biological characteristics of cancer to predict patient outcomes.
“It is expected to significantly help provide optimized personalized treatment for individual patients following gallbladder cancer surgery,” Park said.
The findings were published in the latest issue of the International Journal of Surgery, one of the leading journals in the field of surgery.