Pancreatic cancer, which has a poor prognosis, is becoming increasingly common in industrialized countries. Chemotherapy is the main treatment available for metastatic disease. Response to chemotherapy is currently assessed using imaging to monitor changes in the size of lesions undergoing treatment. Radiomics is a new method of analyzing quantitative data extracted from imaging that can be used to develop prediction algorithms and evaluate response to treatment. There is little robust radiomics data available for evaluating response to first-line treatments for pancreatic adenocarcinoma. Our main hypothesis is that quantifying tumor-related architectural changes could enable early identification of patients who will not respond to chemotherapy.

Pancreatic adenocarcinoma is a cancer with a poor prognosis, and its incidence is increasing, particularly in industrialized countries. Systemic chemotherapy, the main treatment available, has improved survival in patients with metastatic pancreatic adenocarcinoma. Following the results of the study by Conroy et al., the combination of 5FU + oxaliplatin + irinotecan (FOLFIRINOX) is one of the standard first-line chemotherapy regimens for metastatic pancreatic adenocarcinoma.

The population of patients with metastatic pancreatic adenocarcinoma is heterogeneous, with different survival profiles. This is part of a multifactorial picture that is still poorly understood (molecular, terrain, other factors, etc.). Response to chemotherapy is currently assessed using the RECIST 1.1 method, which takes into account changes in the size of lesions during treatment. However, this method is sometimes limited, as it does not always take into account the underlying pathophysiology.

Radiomics consists of analyzing quantitative data extracted from imaging, which allows prediction algorithms to be developed. Several studies have suggested that tumor architecture could be a biomarker of response and survival in patients with esophageal, lung, or colorectal cancer.

A study of 41 patients with advanced pancreatic cancer treated with gemcitabine-based chemotherapy evaluated the impact of radiological architectural assessment on treatment response. The results showed that tumor size, asymmetry, and standard deviation of density at baseline were associated with progression-free survival, and that tumor size and standard deviation of density at baseline were associated with overall survival. These results remain to be confirmed in a larger population, hence the value of conducting our work with data from our database.

Our main hypothesis is that quantifying tumor-related architectural changes could enable early identification of patients who will not respond to treatment. The overall objective would be to tailor the therapeutic strategy to each individual patient.

Inclusion Criteria:

• Adult patients (≥18 years)
• Diagnosis of acute alcoholic hepatitis (ICD10 C25) in a pathology report, followed by manual verification in clinical documents

Exclusion Criteria:

• Patients younger than 18 years at diagnosis

Patients diagnosed with pancreatic cancer across all AP-HP hospitals. Data will be collected retrospectively from the AP-HP Clinical Data Warehouse (EDS).