Form Follows Function: A New Paradigm of Pancreatic Cancer Progression.

Area of Science:

• Oncology
• Computational Pathology
• Cancer Biology

Background:

• Pancreatic ductal adenocarcinoma (PDAC) is known for its aggressive local invasion and vascular infiltration, yet the underlying mechanisms remain unclear.
• Understanding the link between molecular alterations and the physical behavior of PDAC is crucial for improving patient outcomes.

Purpose of the Study:

• To investigate the spatial and geometric properties of PDAC tumor structures and their correlation with invasiveness and molecular subtypes.
• To develop and validate an AI-assisted computational tool for analyzing tumor morphology and predicting clinical behavior.

Main Methods:

• Development of a large-scale, AI-assisted PDAC atlas from H&E images, annotating over 144,000 structures.
• Creation of the SHAPE computational tool for geometrizing cancer progression and analyzing morphology.
• Integration of morphological data with whole-genome, RNA sequencing, and spatial transcriptomic data from PDAC patients.

Main Results:

• Invading PDAC structures exhibit eccentric morphology and local angular coherence, closely tracking vessels and nerves.
• Eccentric structures correlate with increased copy number alterations, loss of heterozygosity on chromosome 17p, and a quasi-mesenchymal/basal-like molecular subtype.
• Upregulation of invasive programs like EMT, angiogenesis, and coagulation pathways was confirmed in eccentric structures via spatial transcriptomics.

Conclusions:

• PDAC exhibits a distinct tumor architecture characterized by specific geometric and spatial properties linked to its aggressive behavior.
• AI-driven morphological analysis offers a powerful, interpretable approach to assess PDAC aggressiveness, predict treatment response, and guide clinical management.