Abstract
Pancreatic ductal adenocarcinoma (PDAC) remains one of the deadliest cancers, with chemotherapy as the mainstay but highly variable efficacy and toxicity. Current regimens, such as FOLFIRINOX and gemcitabine-based combinations, are selected empirically without validated biomarkers to guide choice. Several strategies have been explored to personalize therapy. Patient-derived organoids and molecular classifiers such as PurIST have improved biological understanding but have limited clinical applicability. More recently, predictive transcriptomic signatures have emerged as practical tools. GemPred identifies patients likely to benefit from adjuvant gemcitabine; GemCore, validated in both resected and metastatic tumors, is compatible with small biopsies; and Pancreas-View integrates multiple drug-specific predictors, including for all FOLFIRINOX components and gemcitabine, enhanced by AI. These approaches, retrospectively validated in large cohorts and clinical trials, consistently link predicted sensitivity with improved survival. Beyond regimen selection, signatures enable treatment de-escalation, optimize first-line choices, and identify multidrug-resistant tumors. Ongoing prospective trials will establish their feasibility, supporting transcriptomic profiling as a step toward precision chemotherapy in PDAC.