A big challenge when treating and researching pancreatic cancer is its heterogeneity: the degree of aggressiveness of the tumours varies, as does the response to treatments. Several classification models have been previously proposed, but each one identified different subtypes.
A new study , directed by Núria Malats at the Spanish National Cancer Research Centre (CNIO), has compiled the information from all of these models and uses algorithms to integrate it into a new consensus classifier. In addition, the team have developed a website and an application to facilitate its use: the RNA of a tumour sample is sequenced, the data is uploaded to the website or the application, and the tumour subtype is obtained.
From the best treatment to new risk factors
The new classifier establishes two subtypes based on the information from the tumour cells, and another two based on that of the stroma, the set of healthy cells surrounding tumour cells that also influence the effectiveness of treatments. The study is published in Genome Medicine, and its first author is Pablo Villoslada , a postdoctoral researcher with Malats’ group.
The new model may indicate whether the tumour is a subtype that responds best to one of the most common treatments for this cancer. Hence its importance for clinical practice.
The classifier can also be used to study risk factors for pancreatic cancer, indicates Malats. “We know that smoking increases the risk, albeit slightly, but consistently. Perhaps what is happening is that we have studied it for pancreatic cancer in general, and it turns out that there are subtypes that increase the risk, and others that do not. Now that we are able to separate these subtypes, we can check this.”
Both the website and the app are already up and running. They are freely accessible and include the previous classifiers, to make comparisons or have that data.
Genome Medicine
Experimental study
Human tissue samples
Development of a consensus molecular classifier for pancreatic ductal adenocarcinoma
14-Nov-2025
The authors declare no competing interests.