Advancing Pancreatic Cancer Treatment through Subtype-Specific Strategies Using Artificial Intelligence

Pancreatic cancer rates are on the rise, with over 2,000 new cases annually in Belgium, mainly
pancreatic ductal adenocarcinoma (PDAC). PDAC is expected to be the second most fatal cancer in
the next decade, showing a poor 5-year survival rate under 8%. It is characterized by heterogeneity
and difficult to treat, including overall absence of response to immunotherapy. Treatment options
tailored to specific subtypes, such as those focused on novel immunoregulators (Messaoudi, et al.,
2020), are urgently needed. Innovative machine learning approaches and spatial profiling
technologies, like previous work I contributed to (Michiels, et al., 2024), offer insights into tumor
biology, yet their integration into clinical practice is limited. As a surgeon-scientist trained in AI, I
want to advance subtype-specific immunomodulatory therapies for PDAC, integrating AI to diagnose
tumor heterogeneity and guide treatment decisions. To reach this, I defined the following scientific
objectives: (1) Establishing a profiling pipeline including multi-omic analysis, radiomic analysis and
AI-driven analysis of PDAC surgical resections, (2) Identification of new subtype-specific
immunomodulatory targets, and (3) Translation of this and incorporating in clinical practice. This
comprehensive approach addresses gaps in PDAC treatment, provides a more fundamental
understanding of PDAC, and sets the stage for personalized therapy and new clinical trials.