Evaluation of Efficacy and Safety of Chimeric Antigen Receptor-Natural Killer (CAR-NK) Cells in Breast Cancer: A Systematic Review and Meta-Analysis

Background: For almost two decades now, chimeric antigen receptor-natural killer cells (CAR-NK) have been investigated in pre-clinical breast cancer models, yet clinical evidence on efficacy remains scarce. This meta-analysis provides pooled evidence of pre-clinical CAR-NK effectiveness and safety in breast cancer and an overview of current clinical trials to support clinical translation. Methods: Following PRISMA guidelines and a registered protocol (PROSPERO, CRD420251131530), PubMed, Web of Science, and Scopus were searched up to 30 June 2025 for pre-clinical CAR-NK studies in breast cancer. Clinical studies were retrieved from clinicaltrials.gov and the International Clinical Trial Registry Platform (ICTRP) up to 1 March 2026. Pre-clinical studies without in vivo data or non-human CAR-NK cells were excluded. Primary outcomes were tumor burden (ratio of means, ROMs) and survival (median survival ratio, MSR). Data were analyzed in JASP™ and risk of bias (RoB) was assessed using SYRCLE’s tool. Results: Fourteen pre-clinical studies (38 CAR-NK treatment groups targeting EGFR, HER2, tissue factor, CD70, mesothelin, or folate receptor, with peripheral blood as the primary NK source, and a 5–10 million cell dose) and 11 early-phase clinical studies (targeting HER2, TROP2, PD-L1, MUC1, or NKG2D ligands under ongoing investigation) were included. In pooled pre-clinical analysis, CAR-NK significantly reduced tumor burden against untreated and unmodified/mock controls (ROM 0.311 [0.22–0.44] and 0.42 [0.33–0.53], p < 0.001, respectively). Survival was also prolonged significantly (MSR 1.47 [1.15–1.87], p = 0.010 vs. untreated; 1.30 [1.09–1.60], p = 0.007 vs. unmodified/mock NK cells). Subgroup analyses indicated improved efficacy with peripheral blood source and 5–10 M dosing. No treatment-related toxicities were reported. CAR-NK persistence was generally higher than unmodified/mock NK cells. Discussion and Conclusions: Significant heterogeneity was observed in ROM analysis which the multi-level meta-analysis configured as intra-study interventional variability. There was moderate RoB in pre-clinical studies. Published results from clinical trials remain limited, highlighting early stages of investigation. Overall, CAR-NK therapy demonstrated consistent pre-clinical efficacy and safety, supporting further translational and clinical evaluation in breast cancer.