Pancreatic ductal adenocarcinoma (PDAC) patients frequently suffer from undetected micro-metastatic disease. This clinical situation would greatly benefit from additional investigation. Therefore, we set out to identify key signalling events that drive metastatic evolution from the pancreas. We searched for a gene signature that discriminate localised PDAC from confirmed metastatic PDAC and devised a preclinical protocol using circulating cell-free DNA (cfDNA) as an early biomarker of micro-metastatic disease to validate the identification of key signalling events. An unbiased approach identified, amongst actionable markers of disease progression, the PI3K pathway and a distinctive PI3Ka activation signature as predictive of PDAC aggressiveness and prognosis. Pharmacological or tumour-restricted genetic PI3Kaselective inhibition prevented macro-metastatic evolution by hindering tumoural cell migratory behaviour independently of genetic alterations. We found that PI3Ka inhibition altered the quantity and the species composition of the produced lipid second messenger PIP 3 , with a selective decrease of C36:2 PI-3,4,5-P 3 . Tumoural PI3Ka inactivation prevented the accumulation of protumoural CD206-positive macrophages in the tumour-adjacent tissue. Tumour cell-intrinsic PI3Ka promotes pro-metastatic features that could be pharmacologically targeted to delay macrometastatic evolution.

For patients with metastatic pancreatic cancer that are not eligible for surgery, signal-targeted therapies have so far failed to significantly improve survival. These therapeutic options have been tested in phase II/III clinical trials mostly in combination with the reference treatment gemcitabine. Innovative therapies aim to annihilate oncogenic dependency, or to normalize the tumoural stroma to allow immune cells to function and/or re-vascularisation to occur. Large scale transcriptomic and genomic analysis revealed that pancreatic cancers display great heterogeneity but failed to clearly delineate specific oncogene dependency, besides oncogenic Kras. Beyond these approaches, proteomics appears to be an appropriate approach to classify signal dependency and to identify specific alterations at the targetable level. However, due to difficulties in sampling, proteomic data for this pathology are scarce. In this review, we will discuss the current state of clinical trials for targeted therapies against pancreatic cancer. We will then highlight the most recent proteomic data for pancreatic tumours and their metastasis, which could help to identify major oncogenic signalling dependencies, as well as provide future leads to explain why pancreatic tumours are intrinsically resistant to signal-targeted therapies. We will finally discuss how studies on phosphatidylinositol-3-kinase (PI3K) signalling, as the paradigmatic pro-tumoural signal downstream of oncogenic Kras in pancreatic cancer, would benefit from exploratory proteomics to increase the efficiency of targeted therapies.