Poor-prognosis cancers account for a disproportionate share of global cancer mortality despite major advances in prevention, early detection, and therapy for other malignancies. Launched in 2023, the G7 Cancer Initiative represents an unprecedented effort to address this challenge through coordinated scientific, clinical, and policy action. This report summarizes the outcomes of the G7 Cancer Conference held in Paris in June 2025, which focused on shifting poor-prognosis cancers from a paradigm of resignation to one of collective transformation. Discussions emphasized four interdependent pillars: precision medicine supported by multi-omic profiling and artificial intelligence; early detection and minimal residual disease monitoring using liquid biopsy and advanced imaging; innovative and adaptive clinical trial designs; and emerging therapeutic platforms, including nanomedicine, epigenetic therapies, and immune modulation. The meeting highlighted that scientific innovation alone is insufficient without harmonized infrastructures, equitable access, and international cooperation. Together, these elements form a new model of global cancer governance aimed at translating biological insight into meaningful survival gains for patients with historically lethal cancers.

Pancreatic ductal adenocarcinoma (PDAC) remains clinically managed largely through anatomical stage and performance status, despite major inter-patient heterogeneity and dynamic pathway adaptation during progression. In this review, we propose a hypothesis-generating, stage-aware framework in which clinical stage and established molecular subtypes are considered complementary rather than competing stratification axes. Stage-resolved transcriptomic modeling identified a directional separation across up-front resectable, borderline/locally advanced, primary metastatic and liver metastatic samples, with PC1 explaining 51.1% of the variance and separating early from metastatic states, whereas PC2 explained 30.2% and captured immune/stromal remodeling. We interpret these observations cautiously as pathway-level signals that require functional and clinical validation, not as an established biological paradigm. The model suggests a transition from MAPK-enriched, proliferative and partially immune-competent tumors toward metastatic ecosystems characterized by PI3K-AKT-mTOR engagement, MYC amplification, mitochondrial metabolism, proteostasis support and immune exclusion. We further distinguish approved standards, negative clinical experiences, early clinical signals and preclinical hypotheses. The central conclusion is that PDAC therapeutic stratification should integrate stage, molecular subtype, treatment line, tissue context and evidence level before pathway-directed interventions are clinically prioritized. Accordingly, this review should be read as a translational prioritization framework rather than as a treatment algorithm: it organizes where pathway-directed hypotheses are most biologically plausible, where they are clinically supported, and where they remain insufficiently validated for routine decision-making.

Cellular senescence has emerged as a context-dependent determinant of pancreatic ductal adenocarcinoma (PDAC) progression and therapeutic response. In early pancreatic tumorigenesis, oncogene-induced senescence acts as a robust tumor-suppressive mechanism that limits KRAS-driven malignant transformation. However, this protective barrier is progressively circumvented during disease evolution. In established PDAC, senescence acquires predominantly pro-tumorigenic functions, particularly within the stromal compartment. Senescent cancer-associated fibroblasts reshape the tumor microenvironment through the senescence-associated secretory phenotype (SASP), fostering desmoplasia, immune evasion, and resistance to therapy. In parallel, conventional chemotherapy and targeted treatments induce therapy-induced senescence in tumor cells, contributing to tumor persistence and relapse. Recent advances in senescence-associated gene signatures and SASP profiling have enabled the stratification of PDAC patients according to prognosis and immune landscape. This review critically discusses the dual roles of senescence in PDAC and evaluates emerging senescence-targeted therapeutic strategies, including senolytic and senomorphic approaches, as promising avenues to improve treatment efficacy.

Immune marker staining, collagen features, and αSMA-associated stromal analyses in KC/AKC and KPC/AKPC tumors.

Pancreatic ductal adenocarcinoma (PDAC) is one of the most lethal malignancies, with chemotherapy remaining the backbone of systemic treatment across disease stages. Although multi-agent regimens such as modified FOLFIRINOX and gemcitabine plus nab-paclitaxel-based combinations, with or without cisplatin and capecitabine (AG or PAXG), have improved outcomes in selected patients, therapeutic benefit remains highly heterogeneous and frequently limited by toxicity. This review summarizes evidence supporting shift from empirical chemotherapy selection toward a more personalized approach integrating tumour biology, molecular biomarkers, and patient-related factors, including pharmacogenetics, comorbidities, and toxicity profiles. We discuss established standards of care and highlight the strengths and limitations of available randomized data. The molecular landscape of PDAC, including homologous recombination repair deficiency, mismatch repair deficiency, and rare actionable oncogenic fusions, currently enables precision strategies in a minority of patients. The rapid development of KRAS-targeted therapies may extend molecularly guided treatment to a broader population. Beyond genomics, emerging biomarkers such as transcriptomic signatures, liquid biopsies, proteomic and metabolomic markers, and tumour microenvironment features may refine treatment selection and guide de-escalation or intensification strategies. Finally, we review biomarker-driven clinical trials and outline future directions for adaptive, biology-informed chemotherapy. Integrating molecular and clinical data into routine care may enable a more rational use of chemotherapy and improve outcomes in PDAC.

Pancreatic ductal adenocarcinoma shows early dissemination, stromal remodeling, and therapy resistance, but how tumor programs co-evolve with immune and stromal changes across stages remains unclear. We built a stage-resolved transcriptomic atlas using bulk RNA sequencing of FFPE samples from 443 untreated tumors spanning resectable, locally advanced, primary metastatic, and liver metastatic disease. Integrative modeling identified ten transcriptional programs with monotonic dynamics during progression. Early stages were enriched for epithelial differentiation and immune-stimulatory signals. Advanced stages showed increased cytoskeletal remodeling, vesicular trafficking, oxidative stress responses, and mitochondrial metabolism. Pathway analysis revealed enhanced PI3K-AKT-mTOR signaling and MYC target engagement in metastatic disease. Immune deconvolution showed loss of CD8+ T cells, dendritic cells, and M1 macrophages. Compact gene signatures stratified stage and survival and generalized to TCGA and ICGC cohorts. Functional assays confirmed greater invasiveness, altered redox balance, and mitochondrial dependence in advanced disease, suggesting stage-associated metabolic vulnerabilities.

Clinical characteristics and mutation profiles of the patient-derived organoids used in this study.

Analyses of TGFβ ligand/receptor expression, reporter activity, inferred signaling interactions, and PSC differentiation in ATM-deficient tumor models.

PURPOSEIn 2023, the G7 Cancer Initiative was launched by Australia, Canada, France, Germany, Japan, the United Kingdom, and the United States with the aim of enhancing global cancer control, and with poor-prognosis cancers as a priority. To facilitate effective collaboration among G7 Cancer, we aimed to address the lack of standardized definitions and coordinated initiatives across countries.METHODSWe examined how the G7 Cancer Initiative countries defined poor-prognosis cancers, objectively classified them, quantified their burden, and assessed national response strategies. A review of national cancer plans was conducted together with an expert email survey to evaluate definitions and classifications. Poor-prognosis cancers were identified based on 5-year net survival (NS) below 30% and a mortality-to-incidence (M/I) ratio over 0.75 using CONCORD-3 and Global Cancer Observatory 2022 data.RESULTSPancreatic cancer was consistently categorized as a poor-prognosis cancer, while some countries also included liver, esophageal, stomach, and some brain cancers. For lung cancer, classification varied depending on the definition used. These six cancers accounted for a major share of cancer deaths, with lung (18%-23%) and pancreatic (6%-10%) cancers contributing the most. National strategies differed, with Australia, France, and Japan implementing specific policies for poor-prognosis cancers, while others addressed them indirectly or not at all.CONCLUSIONTo enhance cancer outcomes for poor-prognosis cancers, the G7 Cancer Initiative should coordinate efforts through joint programs focused on early detection, treatment, and policy alignment. Standardized definitions and collaborative action are essential to strengthening the global poor-prognosis cancer response.

PSC-induced chemotherapy resistance and in vitro and ex vivo assays assessing apoptosis, invasion, wound closure, and drug response after perturbation of tumor–CAF signaling.

Nucleic acid therapeutics offer unique opportunities to develop personalized precision therapies by responding specifically and effectively at the gene level to evolving pathogens. However, many nucleic acid drugs employ a single-target strategy, which is effective in monogenic disorders, yet can be ineffective in treating diseases characterized by complex and interacting molecular events. Nucleic acid combinations capable of simultaneously modulating multiple pathways and harnessing concerted effects constitute a promising approach to combatting complex, dynamically evolving and heterogeneous diseases. Here, we report a combination strategy utilizing small activating RNA to activate tumor suppressor genes alongside small interfering RNA to silence oncogenes for therapeutic potential and clinical relevance in cancer treatment using patient-derived tumor models of pancreatic cancer. Transcriptomic and proteomic profiling of individual patient tumors enabled rational design of patient-specific combinations, which exhibited superior anticancer efficacy through coordinated regulation of multiple genes. Importantly, diverse combinations were tailored to individual patient tumor models to elicit therapeutic responses, while certain specific combinations demonstrated therapeutic efficacy across different tumor models that shared similar genetic features, highlighting the flexibility and pan-treatment potential. Our findings not only confirm the efficacy of the combination approach for cancer treatment, but also underscore its potential applicability to other complex malignancies.

Pancreatic ductal adenocarcinoma (PDAC) has one of the lowest cancer survival rates. Recent studies using RAS inhibitors have opened the door to more efficacious therapies, although their beneficial effect is still limited mainly due to the rapid appearance of tumor resistance. Here, we demonstrate that genetic ablation of three independent nodes involved in downstream (RAF1), upstream (EGFR), and orthogonal (STAT3) KRAS signaling pathways leads to complete and permanent regression of orthotopic PDACs induced by KRAS/TP53 mutations. Likewise, a combination of selective inhibitors of KRAS (RMC-6236/daraxonrasib), EGFR family (afatinib), and STAT3 (SD36) induced the complete regression of orthotopic PDAC tumors with no evidence of tumor resistance for over 200 d posttreatment. This combination therapy also led to significant regression of genetically engineered mouse tumors as well as patient-derived tumor xenografts (PDX) in the absence of tumor relapses. Of importance, this combination therapy was well tolerated. In sum, these results should guide the development of new clinical trials that may benefit PDAC patients.

Single-cell RNA sequencing is a powerful approach for characterising cellular heterogeneity and elucidating transcriptional programs that drive tumour plasticity, therapeutic resistance, and disease progression. In this study, we present a single-cell transcriptomic dataset comprising 41 patient-derived cell cultures (PDCs) established from pancreatic ductal adenocarcinoma (PDAC). The dataset was generated using Evercode™ technology, which is based on Split Pool Ligation-based Transcriptome sequencing (SPLiT-seq). This scalable method enables high-throughput profiling across multiple samples without droplet-based microfluidics, allowing efficient capture of inter-sample heterogeneity. This functionally annotated collection of experimentally derived models reveals transcriptional heterogeneity both within and across PDCs, enabling in-depth exploration of PDAC cell diversity, to support the development of computational tools for single-cell analysis, and to guide functional studies on tumour cell subpopulations. This resource constitutes a valuable reference for computational and experimental studies aiming to decipher PDAC heterogeneity and identify therapeutic vulnerabilities.

KRAS, a key member of the RAS proto-oncogene family, encodes a small GTPase involved in regulating cell proliferation, differentiation, and survival through signaling cascades such as MAPK/ERK, PI3K/AKT/mTOR, RalGEF/RalA/B, JAK/STAT3, and NF-κB. Although KRAS mutations, especially at codons 12, 13, or 61, lead to its activation and contribute to uncontrolled growth, these changes alone are not enough to fully transform a normal cell. KRAS activation requires cooperation with other genetic or epigenetic events, such as inactivation of p53 or p16INK4a, tumor suppressor genes to overcome critical cellular barriers, for example, senescence and apoptosis. This need for cooperation reflects the complexity of the oncogenic process, requiring simultaneous deregulation of multiple signaling pathways for malignant transformation. Indeed, in experimental models, mutant KRAS expression in normal cells often induces oncogene-induced senescence rather than unlimited proliferation. These findings highlight that KRAS functions more as an initiator of tumorigenesis than as an autonomous driver. Thus, in depth understanding of the genetic context in which KRAS operates is essential for the development of effective and personalized therapeutic strategies.

Pancreatic ductal adenocarcinoma (PDA) transcriptomic profiling has identified prognostic subtypes, yet patient-specific first-line therapies remain elusive. Here, we stratified PDA tumors by mRNA translation rates, a frequently dysregulated step in gene expression, using translatome profiling of 27 patient-derived xenografts (PDXs). Unsupervised analysis revealed a distinct tumor subset with low global protein synthesis but sustained translation of Integrated Stress Response (ISR) mRNAs, including ATF4. These ISR-activated cancer cells exhibited broad chemoresistance and apoptosis resistance, yet were auxotrophic for serine due to loss of PHGDH and CBS expression, impairing serine and cysteine biosynthesis. This vulnerability correlated with improved overall survival in patients with low expression of both enzymes. Notably, cancer-associated fibroblasts (CAFs) reprogrammed by ISR-activated cells, shifting from myCAF to iCAF phenotype with reduced collagen synthesis and glycine-to-serine conversion, produced serine and sustained tumor growth in amino acid-depleted environments. Our findings demonstrate the power of translatome profiling to reveal stable, drug-resistant PDA cell states and identify a targetable CAF-tumor metabolic symbiosis, opening new avenues for therapeutic intervention in this highly lethal malignancy.

Pancreatic cancer remains one of the deadliest malignancies, with a 5-year survival below 13% and limited response to standard therapies. Conventional pancreatic ductal adenocarcinoma (PDAC), which accounts for over 90% of cases, dominates the landscape. Yet rare subsets of histological and molecular variants - such as medullary, colloid, intraductal papillary mucinous neoplasm-associated, and acinar cell carcinomas - have distinct biology and actionable alterations. These tumors are more frequently KRAS wild-type and are enriched for homologous recombination deficiency, DNA mismatch repair defects, microsatellite instability-high status, and, notably, recurrent gene fusions, unveiling promising opportunities for precision-guided therapies in a disease that has long been resistant to standard approaches. These molecular features identify subgroups more likely to benefit from platinum-based chemotherapy, PARP inhibition, immune checkpoint blockade, or targeted agent approaches with limited activity in unselected, conventional PDAC. Emerging stratifications, including MAPK pathway status and age at diagnosis, further refine the identification of actionable subgroups. Although limited, clinical evidence, including basket trials and real-world observations, demonstrates that biomarker-guided interventions can yield meaningful and durable responses, with improved progression-free and overall survival. Incorporating a structured diagnostic algorithm that recognizes rare subtypes and integrating comprehensive genomic profiling into routine practice represents a paradigm shift from non-stratified to mechanism-driven therapy in pancreatic cancer. This narrative review summarizes the clinicopathological characteristics, molecular landscapes, and therapeutic opportunities of rare PDAC subtypes, highlighting how precision medicine can reshape prognosis and treatment in a historically hard-to-treat disease.

BACKGROUND:A resemblance of pancreatic tumours to their native tissue architecture remains largely unexplored, while it may reveal novel insights into healthy and diseased tissue. OBJECTIVE:This study aims at generating a spatially resolved map of human pancreatic duct cell populations in the native tissue and in tumours, that is, pancreatic ductal adenocarcinoma (PDAC) and adenosquamous carcinoma of the pancreas (ASCP). DESIGN:New datasets were acquired with several spatial transcriptomics platforms integrated with public single-cell RNAseq datasets and validated by multiplex immunofluorescence. Cell lines and primary human cell cultures were genetically manipulated. RESULTS:Groups of Keratin-5+ cells in larger ducts have a gene signature reminiscent of stem cells and (supra)basal cells from other tissues. At single cell resolution, this group comprises ∆Np63+ basal cells (BAS) and ∆Np63- supra-basally residing luminal-B cells (LUM-B). The latter express previously unreported MUC4 and MUC16 and are distinct from other luminal cells in the ducts. In cancer, BAS and LUM-B signatures associate with basal-like (BL) PDAC and correlate with lower survival. However, PDAC exhibits a random spatial pattern and fragmented native expression programmes while ASCP preserves the identity of LUM-B and BAS in a spatially unmixed pattern. ∆Np63 drives cell plasticity to BAS, conserved from the native tissue to cancer. CONCLUSION:Spatially distinct duct cell populations are revealed, and the extent of preservation of the native cell identities in pancreatic cancer underpins distinct tumour identities. This warrants separate consideration in research and therapy.

Abstract Signal-driven relocation of GalNAc-transferases (GALNTs) from the Golgi to the ER, termed GALA, promotes tumour growth, but its effects on glycosylation are unclear. Unlike N-glycosylation, which is co-translational, O-glycosylation initiates post-translationally in the Golgi. Here we show that GALA subverts this arrangement in pancreatic ductal adenocarcinomas (PDAC) and in murine pancreatic tumours, where it stimulates growth. Quantitative glycoproteomics on a cellular model reveals a substantial expansion of the O-glycoproteome, consisting in thousands of sites across hundreds of proteins, ER-resident proteins, cell-surface receptors, secreted factors, and extracellular matrix components. Profiling of murine tumours and patient-derived xenografts confirms widespread activation and cross-species conservation of glycosylation patterns. Structural analysis reveals that GALA-specific residues have solvent accessibility as low as N-glycosylation sites, and below Golgi O-glycosylation or phosphorylation sites, indicating that ER O-glycosylation occurs co-translationally. By inverting the normal temporal sequence of folding and glycosylation, cancer cells generate alternative glycoforms that foster tumour growth.