临床 CAR-T 细胞产品的转导效率：单中心回顾性研究

Transduction efficiency (TE) is a key measure of CAR T cell manufacturing quality; however, its variability and relationship to the final product phenotype remain incompletely understood. In this retrospective study, we evaluated TE across 204 GMP-manufactured CAR T cell products, integrating manufacturing parameters, starting material composition, and immunophenotypic and transcriptomic profiles of final products. Multivariable linear regression and protocol-specific correlation analyses were performed to account for manufacturing-related confounding. TE variability was primarily associated with integrated manufacturing workflows and starting material composition. The CD4CD8-Prodigy-TransAct workflow was associated with higher TE compared with the CD4CD8-Bag-Dynabeads workflow, although contribution of individual manufacturing parameters could not be independently determined due to protocol-specific interdependencies. Within PBMC-based manufacturing, monocyte abundance was negatively associated with TE, suggesting protocol-dependent contributions of starting material composition. Higher TE was associated with a less differentiated T cell phenotype, including lower effector memory T cell (TEM) frequencies, particularly within CD4+CAR+ compartment. Transcriptomic analyses further identified associations between TE and proliferation- and oxidative phosphorylation-associated gene-expression signatures; however, these findings were exploratory and may reflect protocol-related differences. Collectively, TE variability was associated primarily with integrated manufacturing workflows, providing a real-world framework for understanding CAR T cell manufacturing performance.