ED Sciences de la Vie et de la Santé
Characterization of extrahepatic cholangiocarcinoma and development of proteomic profiling-based diagnostic predictive models
by Arthur MARICHEZ (BoRdeaux Institute of onCology)
The defense will take place at 14h00 - Salle Vasco de Gamma Centre Médico-Chirurgical Magellan Hôpital Haut Lévêque 1 Avenue de Magellan 33600 Pessac
in front of the jury composed of
- Laurence CHICHE - Professeure des universités - praticienne hospitalière - CHU de Bordeaux - Directeur de these
- ASTRID HERRERO - Professeure des universités - praticienne hospitalière - Hôpital Saint-Eloi - CHU de Montpellier - Rapporteur
- PIETRO ADDEO - Professeur des universités - praticien hospitalier - Hôpital Hautepierrre - Hôpitaux Universitaire de Strasbourg - Rapporteur
- Aurélie BEAUFRèRE - Maîtresse de conférences - praticienne hospitalière - Hôpital Beaujon - APHP - Examinateur
- Safi DOKMAK - Professeur des universités - praticien hospitalier - Hôpital Beaujon - APHP - Examinateur
Extrahepatic cholangiocarcinoma (eCCA) is a rare malignancy with poor prognosis, for which diagnostic management remains challenging. This doctoral work aimed to improve the positive and differential diagnosis of this cancer through a proteomic profiling approach. eCCA encompasses two anatomical entities: perihilar cholangiocarcinoma (pCCA), arising at the biliary confluence from the first-order divisions to the origin of the cystic duct, and distal cholangiocarcinoma (dCCA), located in the supraampullary common bile duct. Overall 5-year survival across all stages is below 10%. Although clinical suspicion is straightforward in the setting of jaundice, general deterioration, and biliary stricture on imaging, definitive diagnosis remains elusive: histological examination of endobiliay cytological specimens (CS), the gold standard for positive diagnosis, yields a sensitivity of only approximately 50%. This diagnostic uncertainty exposes patients to repeated and potentially morbid investigations, unnecessary major surgery when the lesion is benign, and delayed initiation of palliative chemotherapy in unresectable disease. In this context, the primary objective of this work was to identify a diagnostic proteomic signature applicable to CS. Proteomic profiling of 38 CS enabled the development of a predictive model based on a stromal signature obtained through cellular deconvolution, whose performance was validated in an internal cohort of 19 patients and an external cohort of 14 patients. A prerequisite step involved the clinico-oncological and proteomic characterization of eCCA from a series of 204 patients operated at Bordeaux University Hospital between 2007 and 2023. This analysis demonstrated that overall survival and recurrence-free survival were comparable between pCCA and dCCA, with anatomical location not being an independent prognostic factor. Proteomic profiling did not distinguish between the two locations, which shared a common activation of epithelial-to-mesenchymal transition pathways, supporting the consideration of these two tumors as a single oncological entity. Beyond positive diagnosis, eCCA management raises the issue of differential diagnosis, particularly for dCCA versus pancreatic ductal adenocarcinoma (PDAC). These two anatomically adjacent cancers can be difficult to distinguish histologically when the tumor invades both the pancreatic parenchyma and the common bile duct, leading to classification as "pancreatobiliary" adenocarcinomas (PB-ADC). This diagnostic uncertainty translates into therapeutic uncertainty regarding the choice of adjuvant or palliative chemotherapy, which may impact patient survival if treatment is inappropriate. Analysis of pancreaticoduodenectomies performed for dCCA or upfront resectable PDAC at Bordeaux University Hospital between 2018 and 2023, in the era of modern chemotherapy regimens, confirmed the impact of adjuvant chemotherapy on survival in PDAC patients. This finding underscores the necessity of administering tumor-type-adapted chemotherapy, and consequently of establishing a reliable histological diagnosis on operative specimens of periampullary tumors. We therefore developed, from a cohort of 20 patients operated for dCCA or PDAC with a confirmed diagnosis, a differential proteomic signature associated with a predictive model designed to classify indeterminate PB-ADCs. Taken together, this work provides a characterization of eCCA and establishes predictive models based on proteomic profiling. These models address both positive diagnosis on CS and differential diagnosis on operative specimens, with the objective of improving diagnostic accuracy and optimizing therapeutic management in patients with eCCA.