Mara John

T cell-based immunotherapies have become central to the treatment of multiple myeloma (MM), yet their efficacy depends on the functionality of endogenous T cells. How cumulative treatment exposure, particularly high-dose melphalan, together with disease-intrinsic high-risk features shapes T-cell composition and immune competence remains incompletely understood. Here, we analyzed T cell composition and function in bone marrow (BM) and peripheral blood (PB) samples from MM patients across different stages of their treatment journey using flow cytometry (BM, n=162; PB, n=1,733), single-cell RNA sequencing (n=19), and cytotoxicity assays (n=20). We reveal reduced overall T cell frequencies and CD4+/CD8+ T cell ratio, associated with lines of therapy and driven in part by depletion of naïve CD4+ T cells in gene-expression defined high risk (HR) disease. Among therapeutic agents, melphalan exerted the strongest effects on T cell populations and induced pronounced redox stress in both T cells and myeloma cell lines. This oxidative stress signature was enriched in HR patients and was reversible with N-acetyl-L-cysteine treatment. Together, these findings identify immune dysregulation as a defining feature of HR MM that extends beyond tumor-intrinsic genomic alterations and is further shaped by treatment-induced remodeling of the BM microenvironment. Given the association between higher CD4+ T cell numbers and improved CAR-T cell outcomes, our data highlight the translational importance of treatment sequencing, particularly in HR MM. One Sentence Summary: Melphalan-induced redox stress depletes CD4+ naïve T cells, particularly in patients with high-risk multiple myeloma.

Summary of clinical characteristics

Quantification of PD-1–positive cells and NK cell subsets identified by FCM

Number of cells in each cell type identified by FCM in total T cells

Representative gating strategy of NK cells

Diverse subsets of T and NK cells identified by scRNA-seq. A, UMAP of T/NK subclusters, displaying a total of 12,918 cells, grouped into 14 subclusters marked by color codes (left). UMAP of T/NK subclusters separated for each sample type, displaying cells in EMM (N = 7), EMM_BM (N = 5), and RRMM_BM (N = 6), with the colors representing the same cell type as in the global UMAP of the T/NK compartment (right). B, Dot plot highlighting the expression of canonical markers and gene signatures defining the subclusters in the T/NK compartment. The marker list was curated using Friedrich and colleagues (35), and the genes for defining the signatures are adopted from Chu and colleagues (59). C, Box plot representation of the composition of cell subtypes in the T/NK compartment, grouped by sample type and marked with different colors. D, Stacked bar plots representing patient-wise contributions to individual T/NK subclusters. The proportions sum to 100% across EMM, EMM_BM, and RRMM samples. Box plots display the median (center line), the 25th and 75th percentiles (box limits), and whiskers extending to the most extreme data points within 1.5× the IQR. Statistical comparisons were performed using the Wilcoxon rank-sum test with Benjamini–Hochberg correction for multiple testing. n.s., not significant; **, P < 0.05; ***, P < 0.01; ****, P < 0.001.

Number of cells in each cell type identified by FCM in total nucleated cells

EMM tumors display high proportions of CD16− NK cells. A, scRNA-seq data. UMAP showing clusters in the NK cell compartment, a total of 1,679 cells distinguishable by the color-coded legend (left). UMAP of NK subclusters separated for each sample type displaying cells in EMM, EMM_BM, and RRMM_BM, with the colors representing the same cell type as in the global UMAP of the NK compartment (right). UMAPs showing the expression of CD16 and CD56 in the NK cell compartment (bottom). B, Stacked bar plot showing patient-wise contribution to scRNA-seq NK cell clusters. The proportions sum to 100% across EMM, EMM_BM, and RRMM samples. Box plot indicating the proportion of NK cell subclusters in (C) scRNA-seq and (D) FCM cohorts, color-coded by sample type. E, Box plot indicating the proportion of cells in different cell-cycle phases in CD16+ and CD16− NK cell clusters across sample types. F, Dot plot showing the expression of a curated list of genes in NK subclusters across different sample types. Box plots display the median (center line), the 25th and 75th percentiles (box limits), and whiskers extending to the most extreme data points within 1.5× the IQR. Statistical comparisons were performed using the Wilcoxon rank-sum test with Benjamini–Hochberg correction for multiple testing. n.s., not significant; **, P < 0.05.