This phase 1 trial evaluated the safety and preliminary efficacy of a carcinoembryonic antigen-expressing oncolytic measles virus derivative (MV-CEA) in recurrent glioblastoma (GBM) patients. The study involved two groups: Group A received MV-CEA into the resection cavity, while Group B received MV-CEA intratumorally on day 1, followed by a second dose in the resection cavity on day 5. A total of 22 patients were enrolled, with 9 in Group A and 13 in Group B. The primary endpoint was safety and toxicity, which were well managed without dose-limiting toxicity up to the maximum feasible dose (2×10^7 TCID50). The median overall survival (OS) was 11.6 months, and one-year survival was 45.5%, showing favorable outcomes compared to contemporary controls. Secondary endpoints included assessment of viremia, MV replication, and immune response. A 22-gene interferon-stimulated gene (ISG) diagonal linear discriminant analysis (DLDA) classification algorithm was found to be inversely correlated with viral replication and tumor microenvironment remodeling, suggesting its potential for treatment personalization. The study supports the further clinical investigation of oncolytic MV derivatives and the use of ISG-based DLDA algorithms for personalized treatment strategies.This phase 1 trial evaluated the safety and preliminary efficacy of a carcinoembryonic antigen-expressing oncolytic measles virus derivative (MV-CEA) in recurrent glioblastoma (GBM) patients. The study involved two groups: Group A received MV-CEA into the resection cavity, while Group B received MV-CEA intratumorally on day 1, followed by a second dose in the resection cavity on day 5. A total of 22 patients were enrolled, with 9 in Group A and 13 in Group B. The primary endpoint was safety and toxicity, which were well managed without dose-limiting toxicity up to the maximum feasible dose (2×10^7 TCID50). The median overall survival (OS) was 11.6 months, and one-year survival was 45.5%, showing favorable outcomes compared to contemporary controls. Secondary endpoints included assessment of viremia, MV replication, and immune response. A 22-gene interferon-stimulated gene (ISG) diagonal linear discriminant analysis (DLDA) classification algorithm was found to be inversely correlated with viral replication and tumor microenvironment remodeling, suggesting its potential for treatment personalization. The study supports the further clinical investigation of oncolytic MV derivatives and the use of ISG-based DLDA algorithms for personalized treatment strategies.