Oesogastric and pancreatic adenocarcinomas are poor-prognosis cancers. Incidence of pancreatic cancer drastically increases to such an extent that it will become the second cause of cancer's mortality by 2030. A major challenge is to optimize the therapies for localized setting, when oxaliplatin-based chemotherapy is the standard, before and after surgical excision. Because in 50% of cases oxaliplatin triggers a grade 2-3 sinusoidal obstruction syndrome (SOS) which increases post-operative morbidity, decreases histological response to chemotherapy, increases tumor recurrence, and aggravates the risk of chemotherapy-induced peripheral neuropathy (CIPN). There is an urgent need to better understand the biological processes involved in SOS, in order to prevent and treat it without stopping or reducing oxaliplatin administration. The biological link between oxaliplatin and SOS has not been described, but recent murine experiments argue for HMGB1 to be the mediator released after exposure to oxaliplatin and inducing SOS, and thereafter CIPN. To date, no biomarker is established between murine and patient analyses, and the release of HMGB1 after oxaliplatin treatment and its effect on hepatic parenchyma is not described in patients. Investigators hypothesized is that HMGB1 would also been increased in patients after oxaliplatin treatment, and correlated to the development of SOS and CIPN. If confirmed, personalized treatment will be possible to target this pathway. Therefore, investigators propose to dynamically explore this hypothesis in localized oesogastric and pancreatic cancer patients who will be routinely managed by an initial laparoscopy and post-oxaliplatin surgical excision.