Image-based subtyping of locally advanced pancreatic cancer

Pancreatic ductal adenocarcinoma (PDAC) has a very poor prognosis because the tumor is often already at an advanced stage when first diagnosed and because of resistance to conventional chemotherapy treatments. The genome, transcriptome and proteome of these tumors are highly heterogeneous. This heterogeneity is reflected in a complex tissue architecture with different molecular phenotypes: the classic phenotype with a somewhat better response rate, and the particularly aggressive quasi-mesenchymal phenotype.

There are currently no reliable biomarkers to differentiate between these subtypes in routine clinical practice. This again is partly because of the high degree of heterogeneity, which makes it harder to ensure representative sampling and draw histopathological and molecular conclusions. Medical imaging provides information on total tumor volume. Recent developments in the field of machine learning (e.g. radiomics) are showing promising results in terms of non-invasive tumor characterization and risk stratification based on medical imaging. Translational development and testing of these kinds of prognostic models require a standardized data matrix.

The NEOLAP phase II clinical trial (NCT02125136) is one of the biggest randomized prospective trials that aims to evaluate intensified neoadjuvant chemotherapy in locally advanced PDAC. The image datasets, tumor samples and clinical information collected prospectively for the NEOLAP trial offer a unique opportunity to develop and test the image-based algorithms mentioned above.

A final project summary will be provided.