BACKGROUND: A significant gap in pancreatic ductal adenocarcinoma (PDAC) patient's care is the lack of molecular parameters characterizing tumors and allowing a personalized treatment. The goal of this study was to examine whole PDAC transcriptomic profiles to define a signature that would predict aggressiveness and treatment responsiveness better than done until now. METHODS AND PATIENTS: Tumors were obtained from 76 consecutive resectable (n=40) or unresectable (n=36) tumors. PDAC were transplanted in mice to produce patient-drived xenografts (PDX). PDX were classified according to their histology into five groups, from highly undifferentiated to well differentiated. This classification resulted strongly associated with tumors aggressiveness. A PDAC molecular gradient (PAMG) was constructed from PDX transcriptomes recapitulating the five histological groups along a continuous gradient. The prognostic and predictive value for PMAG was evaluated in: i/ two independent series (n=598) of resected tumors; ii/ 60 advanced tumors obtained by diagnostic EUS-guided biopsy needle flushing and iii/ on 28 biopsies from mFOLFIRINOX treated metastatic tumors. RESULTS: A unique transcriptomic signature (PAGM) was generated with significant and independent prognostic value. PAMG significantly improves the characterization of PDAC heterogeneity compared to non-overlapping classifications as validated in 4 independent series of tumors (e.g. 308 consecutive resected PDAC, HR=0.321 95% CI [0.207;0.5] and 60 locallyadvanced or metastatic PDAC, HR=0.308 95% CI [0.113;0.836]). The PAMG signature is also associated with progression under mFOLFIRINOX treatment (Pearson correlation to tumor response: -0.67, p-value < 0.001). CONCLUSION: We identified a transcriptomic signature (PAMG) that, unlike all other stratification schemas already proposed, classifies PDAC along a continuous gradient. It can be performed on formalin-fixed paraffin-embedded samples and EUS-guided biopsies showing a strong prognostic value and predicting mFOLFIRINOX responsiveness. We think that PAMG could unify all PDAC preexisting classifications inducing a shift in the actual paradigm of binary classifications towards a better characterization in a gradient.