Proteomic Profiling as a Diagnostic Test for Extrahepatic Cholangiocarcinoma on Cytological Samples : Validation of a Proof-of-concept.

Cholangiocarcinoma is a cancer with an increasing incidence and a poor prognosis (< 5% of all stages at 5 years). The extra-hepatic form is the most common and is rarely resectable at diagnosis (30% of cases). This is due to the late onset of symptoms, which often indicate that the disease is too advanced. It can also be explained by the diagnostic difficulties encountered with this cancer. In fact, the reference diagnostic technique is histological testing on endo-biliary cytological samples taken by biopsy and/or brushing during catheterisation of the bile ducts or by interventional radiology. The problem is that the sensitivity of this test is of the order of 50%, with an additional diagnostic uncertainty of the order of 20 to 30%, due to cellular atypia that are not sufficiently clear to confirm a cancerous pathology, although it is not possible to exclude it. This leads to diagnostic error and delay, repeated invasive examinations and sometimes major surgery for stenoses that are ultimately benign when the surgical decision has been taken without a reliable histological result. Additional diagnostic techniques are needed to diagnose a suspected extrahepatic cholangiocarcinoma stenosis, particularly at the molecular level using proteomics and, above all, proteomic profiling. Proteomic profiling enables a patient's diagnosis to be oriented towards a profile (benign or malignant) by comparing the proteins identified in formalin-fixed, paraffin-embedded tissue (cytological sample) with a set of proteins identified within a reference diagnostic signature obtained from previously analysed benign and malignant samples. Within Inserm Unit 1312 - Team 3, a proof of concept for proteomic profiling has been developed as a diagnostic tool in the face of suspected extrahepatic cholangiocarcinoma stenosis, with the development of a diagnostic signature. The aim of this project is to validate this proof of concept on a large retrospective monocentric cohort.