Pancreatic cancer exhibits significant heterogeneity, which poses a major challenge in selecting the best treatment for patients from the very beginning of care. Modern oncology recognizes the use of companion biomarkers to guide targeted therapy or immune checkpoint inhibitors. However, with regard to chemotherapy-which has long been the cornerstone of cancer treatment and remains crucial for most cancers-few predictive tests are available to guide the choice between monotherapy and combination chemotherapy.

Patients included in the PRODIGE 104 B - NEOPREDICT study will be those for whom the GEM transcriptomic signature is negative. This population will be treated according to the standard strategy and will be followed clinically and biologically to describe and identify the characteristics specific to this subgroup, and to compare the usual prognostic factors of this population with those of the GEM-positive population included in the parallel PRODIGE 104 A - NEOPREDICT study

Transcriptomic Signature for Patient Stratification and Prediction of Chemotherapy Sensitivity Transcriptomic signatures, derived from RNA-sequencing, capture unique patterns of gene expression that reflect the molecular phenotype of a tumor and can indicate its potential sensitivity or resistance to chemotherapeutic agents. Over the past decade, several molecular classifications of PDAC have been established (Collisson, Moffitt, Bailey), consistently distinguishing two major tumor subtypes: basal-like (poor prognosis) and classical (better prognosis). While these classifications provide strong prognostic information, they have not reliably predicted chemotherapy response in clinical practice.

Recent advances from the Dusetti/Iovanna research group have led to the development of robust transcriptomic signatures capable of predicting sensitivity to gemcitabine, irinotecan, 5-FU, oxaliplatin, and paclitaxel. These signatures-collectively referred to as Pancreas-View-were generated using preclinical models (PDX, organoids, primary cultures) and have been clinically validated, notably the gemcitabine-sensitivity signature in two retrospective cohorts and in the prospective PRODIGE 24 trial. Importantly, these signatures can be performed on FFPE-derived RNA with minimal material and rapid turnaround, making them compatible with real-world decision-making in the neoadjuvant setting.

Position of the Study Within Current Knowledge Pancreatic ductal adenocarcinoma (PDAC) remains a highly lethal disease due to late diagnosis, limited actionable biomarkers, and the modest efficacy of targeted and immunotherapeutic approaches. Its incidence is rising in Western countries, and without improved strategies, PDAC is projected to become the second leading cause of cancer-related death by 2040.

Borderline-resectable PDAC accounts for approximately 20% of cases and typically requires complex multidisciplinary management. Neoadjuvant chemotherapy (NAC)-now an international standard in BR-PDAC-offers several advantages: improved R0 resection rates, treatment of occult micrometastatic disease, and avoidance of futile surgery in rapidly progressing tumors.

Two main NAC regimens are currently used: mFOLFIRINOX and Gemcitabine + Nab-Paclitaxel. However, comparative studies in non-metastatic PDAC have not demonstrated a clear superiority of one regimen over the other, though their toxicity profiles differ substantially. As a result, selecting the optimal chemotherapy strategy remains challenging.

There is therefore a strong need for predictive biomarkers capable of guiding neoadjuvant regimen selection. The NEOPREDICT program positions itself directly within this unmet clinical need, evaluating the use of a transcriptomic gemcitabine-sensitivity signature to personalize treatment.

Inclusion Criteria:

• Borderline-resectable pancreatic ductal adenocarcinoma (BR-PDAC) as defined by the National Comprehensive Cancer Network (NCCN) v2.2025 criteria, identified on contrast-enhanced CT scan and reviewed by a local multidisciplinary pancreatic expert board including at least a medical oncologist / onco-gastroenterologist, a pancreatic surgeon, and an expert pancreatic radiologist. No central review required.
• WHO Performance Status 0-1.
• Histologically confirmed pancreatic ductal adenocarcinoma, including histological variants.
• Patient included-but not randomized-in the PRODIGE 104 A NEOPREDICT study due to a negative gemcitabine sensitivity signature (GEM-).
• Negative gemcitabine transcriptomic signature (test centrally performed in PRODIGE 104 A NEOPREDICT).
• No prior chemotherapy or radiotherapy for pancreatic cancer, and no previous definitive pancreatic cancer resection (except one cycle of mFOLFIRINOX administered while awaiting the signature result).
• Age > 18 years and < 80 years, with the possibility to include patients aged 75-80 if a standardized geriatric assessment confirms eligibility for the study chemotherapy regimen.
• Ability and willingness to comply with protocol requirements during the entire study period (treatment, scheduled visits, clinical and biological examinations, follow-up).
• Patient's non-opposition to participation in the study.
• Affiliation to the French national health insurance system.

Exclusion Criteria:

• Strictly resectable or locally advanced PDAC according to NCCN criteria.
• Distant metastases, including inter-aortocaval lymph nodes.
• Any condition contraindicating the use of irinotecan, oxaliplatin, or 5-FU.
• Complete dihydropyrimidine dehydrogenase (DPD) deficiency.
• Any uncontrolled or unstable medical condition within the past 6 months (e.g., hepatic, renal, respiratory, or cardiac insufficiency).
• Another concomitant malignancy or history of cancer within the past 3 years, except for adequately treated carcinoma in situ of the cervix or basal/squamous cell skin carcinoma.
• Pregnancy or breastfeeding.
• Patients under legal protection, guardianship, curatorship, or under judicial/administrative protection.
• Patients receiving psychiatric care or unable to provide consent.
• Inability to comply with medical follow-up for geographical, social, or psychological reasons.