CELL INTELLIGENCE · 肿瘤细胞治疗研究

Recurrent pancreatic ductal adenocarcinoma (PDAC) includes circulating tumor DNA (ctDNA)-defined molecular residual disease (MRD), isolated or oligometastatic relapse, disseminated measurable recurrence, and heavily pretreated refractory disease. These states differ in tumor burden, treatment urgency, tissue accessibility, and immune competence and should not be treated as one population. Immune checkpoint blockade has minimal activity in unselected PDAC, and randomized trials have not shown consistent benefit from adding checkpoint inhibitors to chemotherapy. However, rare microsatellite instability-high/mismatch repair-deficient (MSI-H/dMMR) tumors and selected antigen- or human leukocyte antigen (HLA)-defined subgroups support precision immune intervention. This Mini Review evaluates biomarkers for immunotherapy and adoptive cell therapy, including ctDNA kinetics, tumor mutational burden, KRAS and neoantigen profiles, antigen presentation and target-antigen density, effector immune state, myeloid and fibroblast-mediated suppression, and spatial immune exclusion. It compares checkpoint blockade, personalized and KRAS-directed vaccines, chimeric antigen receptor T-cell (CAR-T), chimeric antigen receptor natural killer-cell (CAR-NK), and T-cell receptor (TCR)-engineered therapies by recurrence setting, evidence maturity, feasibility, and failure mechanisms. Checkpoint blockade is currently most defensible for MSI-H/dMMR disease; vaccines and cell therapies remain investigational and are best tested in biomarker-selected trials, especially in low-burden or molecular recurrence. We propose a five-step workflow to classify recurrence, identify actionable biomarkers, match evidence-appropriate interventions, assess feasibility, and monitor response and immune escape. This framework supports trial design and disciplined clinical translation rather than empiric immunotherapy in unselected recurrent PDAC.