近期获批的 CAR-T 细胞与双特异性抗体疗法用于三类暴露复发/难治性多发性骨髓瘤的间接治疗比较：系统综述

Assessing the comparative efficacy of novel chimeric antigen receptor T-cell (CAR-T) and bispecific antibody (bsAb) therapies for relapsed/refractory multiple myeloma (RRMM) is crucial for informed treatment decisions. This study systematically reviewed indirect treatment comparisons (ITCs) evaluating the efficacy of approved CAR-T therapies and bsAbs in triple-class exposed RRMM.

Systematic searches identified ITCs comparing the efficacy of approved BCMA- and GPRC5D-targeted therapies: elranatamab, linvoseltamab, talquetamab, teclistamab, ciltacabtagene autoleucel (cilta-cel), and idecabtagene vicleucel (ide-cel). Screening used large language models and human researchers, with full-text review and data extraction performed by two researchers. Results were synthesized qualitatively.

Eleven matching-adjusted indirect comparisons (MAICs) were included; six compared bsAbs, four compared bsAbs and CAR-Ts, and one compared CAR-Ts. All MAICs were adjusted for differences in refractory status, cytogenetic risk, disease stage, and presence of extramedullary disease. Across all MAICs identified, elranatamab showed significant improvements versus teclistamab (ORR, DOR, PFS, OS); linvoseltamab showed significant improvements versus elranatamab (ORR, CR), talquetamab ( CR, DOR), teclistamab (DOR, PFS, OS), ide-cel ( CR, DOR, PFS), and cilta-cel (DOR); ide-cel showed significant improvements versus teclistamab (OS) and elranatamab (OS); and cilta-cel showed significant improvements versus linvoseltamab (ORR, VGPR, CR), and ide-cel (ORR, CR, DOR, PFS, OS).

Although head-to-head trials are not available, the MAICs provide hypothesis-generating estimates of the relative efficacy of approved bsAbs and CAR-Ts for triple-class exposed RRMM. Based on the available evidence, linvoseltamab and cilta-cel were frequently associated with improved efficacy across a range of comparators and endpoints; however, these findings included both statistically significant and numerically favorable results and should be interpreted cautiously in light of cross-study heterogeneity and the limitations associated with unanchored MAICs.