Research in the US and Sweden have determined that microscopic changes in the liver may help clinicians predict if and where pancreatic cancer metastasises.

It is known that pancreatic cancer metastases occur most frequently to the liver. As part of the current study, published in the journal Nature Medicine, researchers collected liver tissue samples from 49 patients who underwent surgery for pancreatic cancer and had no signs of metastasis at the time. Liver samples were also taken from 19 patients who did not have cancer but had undergone similar procedures for benign lesions.

After three years of follow-up, notable differences in the livers of patients with pancreatic cancer that had metastasised were identified compared to the control group who did not have cancer. The researchers found small but important differences in metabolism, gene expression, and immune cells in the liver in those who rapidly developed liver metastases compared to livers where the process took longer, or did not happen at all.

The findings allowed the researchers to develop a machine learning model to predict whether and where a pancreatic cancer would spread based on characteristics of a liver biopsy at the time of diagnosis of the initial cancer.

The researchers say the prediction model will be of use not only in pancreatic cancer but also in other cancers, such as colorectal cancer and gastric cancer, allowing oncologists to tailor cancer treatment to an individual patient’s risk of relapse.