T cell fitness and tumor specificity predict immunotherapy response in multiple myeloma Free

T cells are increasingly recognized as central effectors in the therapeutic response across a wide spectrum of treatments for multiple myeloma, including bispecific T cell engagers (BTCEs), CAR-T cell therapies, and immunomodulatory agents used in standard-of-care regimens. Despite distinct mechanisms of action, these therapies converge on the requirement for a competent and responsive T cell compartment. As such, dissecting the immunological synapse between T cells and myeloma cells, the key interface of recognition and cytotoxic engagement, holds promise for identifying actionable biomarkers of response and resistance, and may ultimately guide the rational sequencing and personalization of immunotherapies.

To define the functional and antigenic landscape of the anti-myeloma T cell response, we analyzed bone marrow samples from n=15 patients with newly diagnosed MM (NDMM) using paired single-cell RNA and TCR sequencing. This approach enabled high-resolution profiling of the T cell compartment coupled with extensive funtional testing and led to the identification of over 100 TCRs with putative tumor reactivity. Myeloma-reactive clonotypes were rare, accounting for a median of 1.29% of total marrow T cells, but exhibited distinct transcriptional states, including upregulation of effector molecules, chemokine receptors and transcription factors associated with cytotoxicity and tissue residence, while lacking markers of terminal exhaustion.

To directly infer the antigen targets of these TCRs, we integrated HLA class I immunopeptidomics performed on primary CD138⁺ myeloma cells with TCR deorphanization strategies. To functionally confirm TCR:antigen pairing, we synthesized and cloned selected myeloma-reactive TCR sequences and expressed them in patient-derived CD8⁺ T cells. Notably, several validated TCRs recognized non-mutated myeloma-associated antigens, supporting the idea that self-antigen-directed immunity contributes to disease control in early MM.

Building on these data, we developed a scRNA-sequencing based classifier, capable of recognizing and quantifying the transcriptional signature of tumor-reactive T cells in single-cell datasets, called TfiT. This classifier successfully distinguished tumor-reactive from bystander and virus-specific T cells and outperformed existing models derived from solid tumors in predicting reactivity within MM samples (AUC=0.895, prospectively validated across n=9 NDMM patients).

Importantly, the baseline abundance of TfiT-positive T cells prior to therapy initiation significantly correlated with clinical outcomes: Among n=19 NDMM patients treated with standard-of-care quadruplet therapy (Daratumumab-VRd), higher frequencies of TfiT-classified T cells predicted achievement of a complete response (p=0.0213), suggesting that the phenotypic and functional state of the T cell compartment before treatment is a critical determinant of therapeutic efficacy.

To evaluate whether T cell fitness dynamics also track immunotherapy response at later disease stages, we applied the TfiT model to a cohort of n = 16 patients with relapsed/refractory MM treated with BCMA-targeting bispecific T cell engagers (BTCEs). TfiT-classified T cells were again enriched in responders at baseline and expanded during therapy, indicating that the transcriptional hallmarks of tumor-reactive T cells not only predict response but dynamically track therapeutic engagement over time. In contrast, non-responders failed to exhibit such expansion, suggesting impaired T cell priming or immune escape.

Collectively, our findings support a model in which transcriptionally defined, tumor-reactive T cells are key determinants of immunotherapy response in multiple myeloma. Their frequency, phenotype, and activation state reflect the endogenous immune competence of the host and can serve as biologically grounded biomarkers to monitor both spontaneous and therapy-induced anti-tumor immunity. Beyond defining the transcriptional correlates of tumor reactivity, we present a scalable framework to map TCR antigen specificity for the identification of personalized, myeloma-specific TCRs. Lastly, the TfiT signature offers a predictive and longitudinal biomarker across multiple lines of MM immunotherapy, with potential to guide treatment selection, monitor response, and inform sequencing strategies in a rapidly evolving therapeutic landscape.