CELL INTELLIGENCE · 肿瘤细胞治疗研究

Chimeric antigen receptor (CAR) T-cell therapy has achieved remarkable success in hematologic malignancies but continues to face significant barriers in solid tumors such as ovarian cancer (OC), primarily due to antigen heterogeneity and on-target, off-tumor toxicity. Traditional discovery methods that rely on bulk tissue analyses are inadequate to overcome these challenges. This review proposes a paradigm shift toward a comprehensive, multi-omics, single-cell pipeline for the grading and validation of safe and effective CAR targets in OC. We establish a three-tiered homogeneity classification system (Grades A through C) that directly informs engineering strategies, from monotherapy for Grade A targets to combinatorial targeting for Grade B targets and deprioritization for Grade C targets. Using CLDN18.2 as a validation case, we demonstrate how this pipeline classifies targets according to homogeneity levels and prescribes specific therapeutic solutions. We detail a sequential framework that integrates bulk omics atlases, single-cell and spatial transcriptomics, glycoproteomics, and immune phenotyping to deconstruct both tumor complexity and the immunosuppressive microenvironment. These data feed into a computational prioritization funnel that scores candidate antigens based on tumor specificity, homogeneity, tumor microenvironment (TME) fitness, and essentiality. We then outline a tiered experimental validation pathway, from patient-derived organoids (PDOs) to in vivo safety models. The practical utility of this atlas-driven approach is demonstrated through a case study on claudin 18.2 (CLDN18.2), in which integrated multi-omics data diagnosis inform rational engineering strategies such as combinatorial targeting and affinity tuning to mitigate limitations. Finally, we discuss reporting standards, practical feasibility, and future directions, emphasizing the transformative potential of artificial intelligence and collaborative atlases. This holistic pipeline aims to bridge the critical gap between computational discovery and clinically viable, next-generation CAR-T therapies for ovarian cancer.