Introduction
Multiple myeloma (MM) is a complex hematological malignancy marked by the uncontrolled proliferation of clonal plasma cells in the bone marrow and elevated serum monoclonal protein levels, leading to hallmark signs of end-organ damage, including hypercalcemia, renal impairment, anemia, and osteolytic lesions. High-risk cytogenetic abnormalities, such as TP53 deletion, t(4;14), t(14;16), t(14;20), and 1q21 amplifications, are associated with poor prognosis and resistance to therapy. These genetic changes drive the overexpression of oncogenes such as FGFR3, MMSET, MAF, and MAFB, and genes such as CKS1B, which increase myeloma cell proliferation, survival, and apoptosis resistance. 1 , 2 Additionally, the BM microenvironment, which is composed of cellular (stromal cells, immune cells, osteoblasts, osteoclasts) and non-cellular elements (extracellular matrix, cytokines, growth factors), interacts with malignant plasma cells, influencing their behavior, survival, and response to therapy, 3 thus playing a critical role in MM pathogenesis and progression.
Over the years, therapeutic advancements in MM have significantly increased survival rates and quality of life, with current approaches encompassing induction therapies, autologous stem cell transplantation (ASCT), maintenance strategies, and innovative emerging treatments. 4 Induction therapy is a critical first step in MM treatment, aiming to reduce tumor burden and achieve remission prior to ASCT or further therapies. The Revlimid (lenalidomide), Velcade (bortezomib), and dexamethasone (RVd) regimen is a standard induction therapy that results in high response rates and prolonged progression-free survival (PFS). 5 , 6 RVd achieves overall response rates (ORR) of 80–90% in newly diagnosed MM, with very good partial response (VGPR) rates increasing from 58% after three cycles to 87% post-consolidation. Long-term outcomes include 77% PFS at 39 months and 100% overall survival (OS) in select cohorts. 5 RVd is widely recommended for newly diagnosed patients and is also effective as salvage therapy in relapsed or refractory cases. 6 , 7 For patients unable to tolerate lenalidomide, the cyclophosphamide-bortezomib-dexamethasone (CyBorD) regimen is an effective alternative. 8 The CyBorD regimen achieves an overall response rate (ORR) of 88%, with 61% attaining VGPR or better, 39% reaching complete (CR) or near-complete responses (nCR) after four cycles, and three-year OS and PFS rates of 89% and 42.7%, respectively. 9 It induces rapid responses, reducing serum monoclonal protein by 80% within two cycles, and serves as a post-ASCT consolidation therapy to improve outcomes in patients with residual disease. 10 While RVd remains the preferred regimen due to its efficacy, CyBorD is invaluable for patients with reduced tolerability or specific clinical needs.
Both RVd and CyBorD combination therapies target myeloma cells through distinct mechanisms, including the modulation of signaling pathways and the immune microenvironment, thereby enhancing their therapeutic efficacy. 11 Lenalidomide, a key component of RVd, promotes T-cell proliferation, enhances natural killer (NK) cell activity, and shifts the immune environment toward an anti-tumor state. It reduces pro-inflammatory cytokines such as IL-6, increases anti-inflammatory cytokines such as IFN- γ and TNF- α , and suppresses immune checkpoints such as PD-1, thereby countering immune evasion and boosting T-cell responses against myeloma cells. 12 Cyclophosphamide, which is central to CyBorD, acts as an alkylating agent that induces DNA cross-linking, leading to cell cycle arrest and apoptosis in rapidly dividing myeloma cells, while its immunomodulatory effects further enhance immune responses. 13 , 14 Like both combinations, bortezomib, a proteasome inhibitor, induces apoptosis in MM cells by preventing the degradation of pro-apoptotic factors and inhibiting NF-κB activity, enhancing T-cell activation and sensitizing cells to other therapies. 15 Although dexamethasone effectively reduces inflammation, it also exerts immunosuppressive effects on T cells and NK cells, potentially increasing the risk of infection. 16
By enhancing our understanding of MM pathophysiology, we previously characterized B-cell lymphopoiesis, plasma cell (PC) differentiation, and the immune landscape of the bone marrow microenvironment across different stages of myeloma disease evolution and progression using high-dimensional mass cytometry (CyTOF). Our findings revealed distinct immunophenotypic profiles, with differential expression of key B cell signaling regulators (IRF-4, Bcl-6, c-Myc, CXCR4, MYD88, and sXBP-1), aberrant PC markers (CD319, CD269, CD200, CD117, CD56, and CD28), and stemness-associated factors (Nestin, Sox2, KLF-4, and Nanog), all of which correlated with clinical outcomes. 17 In addition, we identified stage-specific immune alterations, highlighting profound remodeling of both the innate and adaptive immune subsets throughout disease progression. 18
In this study, we conducted an in-depth investigation into the dynamic interactions between clonal PC subsets, B cell differentiation, and immune modulation in response to therapy, comparing the effects of the RVd versus CyBorD regimens. By integrating high-dimensional CyTOF profiling with clinical response data, we systematically compared responders and non-responders across both therapeutic regimens, evaluating immunophenotypic shifts, alterations in clonal dynamics, and molecular signaling perturbations associated with treatment efficacy. These findings offer crucial mechanistic insights into MM pathophysiology and therapeutic response, establishing a foundation for optimizing precision medicine approaches in MM management.
Materials and methods
Patient enrollment and sample collection
Bone marrow (BM) aspirates were collected from 105 MM patients, including untreated patients (no Tx, n = 43) and those treated with RVd- or CyBorD-treated MM ( n = 62) patients. The clinical characteristics of the patients are detailed in Supplementary Table 1. Comprehensive clinical and treatment histories were recorded at the time of sampling. BM samples were obtained during routine diagnostic procedures at Dana-Farber Cancer Institute (Boston, USA) and the Biomedical Research Center (Bratislava, Slovakia). Ethical approval was granted by the Biomedical Research Center’s local ethics committee (Protocol No. Myelom 001) and Dana-Farber Cancer Institute (Protocol No. 10-106), in accordance with the Declaration of Helsinki. Written informed consent was obtained from all participants.
Cell lines and human peripheral blood mononuclear cells
Multiple myeloma cell lines (RPMI 8226-S, RPMI-Dox40, RPMI-MR20, and MM.1S) and additional cancer cell lines (RAJI, KG-1a, Jurkat, MOLT-4, HeLa, and NTERA-2) were obtained from ATCC (Manassas, VA). The cells were cultured in RPMI 1640 (Cellgro, Mediatech, VA) supplemented with 10% heat-inactivated fetal bovine serum (FBS; Harlan, Indianapolis, IN), 100 U/mL penicillin, 100 μg/mL streptomycin, and 2 mM L-glutamine (GIBCO, Grand Island, NY) at 37 °C in 5% CO₂, except for NTERA-2, which was maintained in DMEM (Cellgro, Mediatech, VA).
Peripheral blood mononuclear cells (PBMNCs) from healthy volunteers were isolated by Ficoll-Hypaque (Pharmacia, Piscataway, NJ) density gradient centrifugation and cultured in RPMI 1640 supplemented with 20% FBS, antibiotics, and L-glutamine. Informed consent was obtained following the Declaration of Helsinki. PBMNCs were stimulated with GM-CSF (50 ng/mL, 7 d), while MM.1S cells were treated with camptothecin (6 μM, 8 h).
Antibody conjugation strategy
Antibodies, including clones, providers, and concentrations, are detailed in Supplementary Table S2. Using the MaxPAR antibody conjugation kit (Fluidigm Sciences), 100 μg of antibodies were conjugated with mass isotopes according to the manufacturer’s protocol. Lanthanide metal solutions (2.5 mM) were polymer-loaded for 40 min at 37 °C, followed by antibody buffer exchange and partial reduction of the antibody with 4 mM TCEP for 30 min. After purification, the antibodies were conjugated to lanthanide-loaded polymers for 1 h and recovered through four washing steps.
Performance assessment of metal-conjugated antibodies
Metal-conjugated antibodies were digested in 2% HCl and 2% HNO₃ (diluted in ddH₂O) and analyzed on a CyTOF 2 mass cytometer (Fluidigm). The metal content (≥100 metal atoms/antibody) was verified using blank and tuning solutions as normalization standards, and the antibody concentrations were measured at 280 nm with a NanoDrop, adjusted to 0.5 mg/mL in PBS stabilization buffer with 0.05% sodium azide, and stored at 4 °C. Antibody efficacy was validated using titration and positive controls, such as human peripheral blood mononuclear cells or cell lines (Supplementary Table S2).
Bone marrow processing
Fresh BM aspirates were collected in sodium heparinized tubes, fixed with proteomic stabilizer buffer (1.4 mL buffer/mL BM) for 10 min at RT, and stored at –80 °C. Before analysis, the samples were thawed at 4 °C, and erythrocytes were lysed with hypotonic thaw-lyse buffer (4:1 buffer-to-sample ratio) for 10 min at RT. Leukocytes were pelleted at 600 × g for 6 min, with repeated lysis as needed. The cells were washed twice with cell staining medium (CSM: PBS with 0.5% BSA and 0.02% NaN₃) and pelleted for further analysis.
CyTOF workflow: sample preparation and data acquisition
The samples were stained with anti-human CD15-PerCP-Cy5.5 (clone HI98; BD Pharmingen) for 30 min at 4 °C. CD15⁻ cells were sorted using a FACS Aria III 5 L cell sorter (70-μm nozzle) to deplete granulocytes from the BI and BII panels (not the TIME). The cells were washed in CSM, blocked with 5 μL of Fc receptor blocking solution for 10 min at RT, and stained with specific cell surface antibody coctail in 100 μL CSM for 30 min. Aliquots of the sample were stained with either the BI or BII panel of surface antibodies (sorted or unsorted) or the TIME panel (unsorted only). After staining, the cells were washed twice in CSM and centrifuged (600 × g, 6 min, RT). For intracellular staining, BI and BII samples were permeabilized with ice-cold methanol (1 mL) and stored at −80 °C, while TIME samples were treated with FOXP3 Perm buffer (1 mL, 1 × , 30 min, RT). The cells were then washed twice in CSM and stained with intracellular antibody cocktails specific to each panel, along with 1 μL of 191/193 Iridium (Ir) DNA intercalator (Fluidigm, San Francisco, CA) in 100 μL of CSM for 1 h at RT. The cells were washed twice in CSM, centrifuged (600 × g, 6 min, RT), and subsequently washed with PBS and ddH₂O before acquisition. After washing, the cells were diluted to 0.5 × 10⁶ cells/mL in ddH₂O with 10% EQ Four Element Calibration Beads and acquired on a CyTOF 2 cytometer at 300–500 events/second.
High-dimensional data processing and SPADE analysis
The data were normalized using the Normalizer tool and EQ Four Element Calibration Beads to correct signal fluctuations, then concatenated with the Cytobank concatenation tool (Mountain View, CA). The signal intensities were arcsinh-transformed (cofactor 5), and gating strategies were applied to exclude debris, beads, and apoptotic cells. High-dimensional data were analyzed using Spanning-tree Progression Analysis of Density-normalized Events (SPADE) in Cytobank (Mountain View, CA), generating tree structures representing phenotypically similar cell clusters. Nodes in the SPADE tree were color-coded by median marker intensity (blue: low, red: high), with node size reflecting the fraction of assigned cells. SPADE clustering included 13 markers in the BI/BII panels and 33 in the TIME panel, with parameters of 1% outlier density, 10% target density, and 300/500 clusters for the BI/BII and TIME panels, respectively. Bone marrow samples from MM patients and healthy donors were analyzed together, generating unified SPADE trees for each panel. SPADE data were exported to R for further analysis, with cluster annotations based on marker expression and prior knowledge.
Interactive and advanced visualization tools
To enable interactive visualization of high-dimensional data, we developed a Shiny-based R web portal using additional packages such as shinyBS, igraph, ggplot2, Cairo, gplots, colorRamps, plotly, and DT (Supplementary Table 3) to annotate cell population clusters from SPADE analyses. Following annotation, we normalized the number of cells in each population relative to the total sample size using the formula: x = cB / cA , where x represents the normalized cell count, cB is the total number of cells in a given cluster, and cA is the total number of cells across all clusters. Median marker expression was normalized using: y = Σ ( c 1 . e 1 + c 2 . e 2 + ⋯ + cn . en ) n 1 / cB , where y is the normalized median expression of a marker, c represents the number of cells in a node, e is the marker's median expression, and cB is the total number of cells in the cluster.
For further analysis, we modified functions from the SPADEVizR package, including mergeClusters, identifyDAC, to assess marker expression across clusters. R Shiny and additional packages (e.g., shinydashboard, htmlwidgets, openxlsx, queryBuilder, rhtmlHeatmap, RColorBrewer, viridisLite, grid, and gridExtra) were used to visualize normalized SPADE results as SPADE trees, heatmaps, dot plots, and whisker/box plots. Marker regulation across multiple cell clusters was visualized using ComplexHeatmap libraries, while Venn diagrams (VennDiagram package) illustrated common marker expression changes in MM (Supplementary Table 3).
Statistical analysis
Statistical analyses were performed using GraphPad Prism. Differences between two groups were assessed with the Wilcoxon rank-sum test (Mann‒Whitney U test). Comparisons among three or more groups were performed via one-way ANOVA for mean values and Dunn's test after the Kruskal‒Wallis test for medians. The data distribution was evaluated via the Kolmogorov‒Smirnov and Shapiro‒Wilk tests, and outliers were identified by Tukey’s test. Multiple comparisons were analyzed using the Holm‒Sidak method and Tukey's test. P -values (* p < 0.05, ** p < 0.01, *** p < 0.001) are shown in the graphs and figure legends.
Results
Single-cell CyTOF profiling of immune and clonal dynamics in MM under therapies
To elucidate the effects of standard anti-MM therapies within the complex pathophysiology of MM, we conducted a comprehensive analysis of clonal PC subsets, B cell differentiation, and immune modulation in response to RVd and CyBorD treatment regimens using large-scale CyTOF profiling. Our study cohort included RVd-treated patients ( n = 47) compared to untreated MM patients ( n = 43) and CyBorD-treated patients ( n = 15) compared to untreated MM patients ( n = 43). Moreover, we stratified RVd responders ( n = 35) versus non-responders ( n = 12) and CyBorD responders ( n = 5) versus non-responders ( n = 10) to assess the immunophenotypic changes and molecular signaling alterations associated with treatment efficacy (Supplementary Table 1). Our analytical pipeline focused on profiling PC heterogeneity within and across patient groups, evaluating molecular disruptions of transcription factors and signaling regulators (CyTOF BI panel), as well as stemness-associated and aberrant PC markers (CyTOF BII panel); in addition, both panels profiled B-cell lymphopoiesis ( Figure 1A ). Furthermore, the CyTOF TIME panel was used to examine the interplay between plasma cell disease and immune status (Supplementary Table 2). Briefly, BM-isolated cells from these cohorts (either pre-sorted for BI and BII panels or unsorted for TIME panel) were stained in parallel using the three CyTOF panels, analyzed by CyTOF technology, and subjected to comprehensive bioinformatics and statistical analysis (Supplementary Table 3).
Immune modulation patterns in MM patients treated with RVd and CyBorD therapies
In our previous study, we used high-dimensional CyTOF profiling to characterize BM immune dynamics across MM progression, revealing adaptive immune suppression and increased innate immune infiltration. 18 To further investigate these immune alterations, we analyzed changes in immune cell distribution and organization in MM patients treated with RVd and CyBorD therapies. Following RVd treatment, we observed significant decreases in plasma cell (PC) clusters (PC1-3,5), plasmablasts (PB/PC2), transitional B cells (T2), unswitched memory B cells (UM1-4), and naive helper T cells (CD4Tn), while pre-BI cells and erythroblasts (Eb) increased ( Figure 1B ). Among RVd responders, PC1-4 and naive cytotoxic T cells (CD8Tn1-2) were reduced, whereas immature B cells (I2), T1,3 clusters, and B cells within the tumor-immune microenvironment expanded ( Figure 1C ). CyBorD treatment resulted in lower frequencies of PC2,7 clusters, UM2-3, and T1-2 cells, while PB/PC1 subsets were upregulated ( Figure 1D ). In CyBorD responders, only T3 and naïve B cells (N) showed a reduction Compared to non-responders ( Figure 1E ). Neither therapy significantly affected the distribution of mature CD4⁺ or CD8⁺ T cell subsets, including both central memory and effector memory cells (data not shown). These findings demonstrate distinct immune modulation patterns associated with RVd and CyBorD therapies, highlighting their differential effects on immune cell subsets, particularly plasma cells, early B cell stages, and naive T cells, in MM patients.
Distinct molecular signatures in plasmablasts and plasma cells induced by RVd and CyBorD therapies in MM patients: implications for treatment response
Comparative analysis of intra- and inter-tumoral heterogeneity in MM patients versus healthy-donor plasma cell (PC) and plasmablast (PB) clusters demonstrated a universal upregulation of MMSET (WHSC1), CD47, and Notch‐1. In two precursor PB/PC clusters (CD20–, CD27het, CD38+, CD45+, CD138–), we observed constitutive overexpression of CXCR4 (CD184), c-Myc, IRF-4, MYD88, and P -gp (CD243), together with aberrant MM markers such as CD52, CD56, and CD319. Among mature PCs (CD19het, CD20het, CD27het, CD38+, CD45lo, CD138+), we identified seven immunophenotypically and functionally distinct subsets, each defined by unique combinations of B-cell regulators and transcription factors (IRF-4, sXBP-1, Bcl-6, c-Myc), stem-cell-associated markers (Nanog, Nestin, Sox-2, RARα2), and aberrant myeloma antigens (MMSET, CD200, CD319, CD289, CD28). 17
To evaluate signaling responses associated with RVd therapy or CyBorD compared to untreated myeloma samples, significant alterations were observed in plasmablasts and plasma cell clusters. In plasmablasts, RVd therapy induced the upregulation of CD38 (PB/PC1), CD81, and CD45 (PB/PC2) while downregulating CD200, c-Myc, CD27, sXBP-1, CD362, Oct3/4, MMSET, CD289, CD38, IRF-4, Nanog, and CD319 (PB/PC2). Within plasma cell (PC) clonal clusters, RVd therapy increased CD81 (PC1-4), CD338 (PC1,2,5), FGFR-3 (PC1-3,5), Pax-5 (PC1,2,6), CD45 (PC1,4,5), IgA, CD19 (PC4,5), KLF-4 (PC6,7), CD117, CD62L, and IgM (PC5) expression. Conversely, a significant reduction in the expression of markers such as CD200, CD138 (PC1,2,4,5), MMSET (PC4-7), CD38 (PC2,6,7), c-Myc, CD269, CD319 (PC2,4), CD27 (PC2-4), CD184 (PC2,3), CD47 (PC2,5), sXBP-1 (PC2,7), CD289 (PC2,6), CD45, CD28, Nanog, IRF-4, Oct3/4, (PC2), IgM (PC3), and CD20 (PC4) was noted across several PC clusters ( Figure 2A ). In RVd responders, plasmablasts presented decreased expression of CD329 (PB/PC1) and CD319 (PB/PC2), alongside increased levels of CD81 (PB/PC1-2), IgM, IgG, CD243 (PB/PC1), and CD45 (PB/PC2). Plasma cell clusters from responders showed an increase in IgA (PC1,6), IgM (PC1,7), CD81 (PC4,7), CD52 (PC1), CD19 (PC2), CD289 (PC3), CD243 (PC4), CD138 (PC6), and IgG (PC7), along with downregulation of CD200 (PC1-3), CD56 (PC1-3), IRF-4 (PC1,2,4), Nanog (PC1,2,4,6), CD38 (PC2,4), CD269 (PC4), CD319 (PC2), MMSET, FGFR-3, Blimp-1 (PC4), CD45 (PC5), and RARa2 (PC6) compared to non-responders ( Figure 2B ).
Comparatively, CyBorD-treated patients demonstrated upregulation of CD38 (PB/PC1) and CD45 (PB/PC2) in plasmablasts, while reductions were observed in IgA, CD319 (PB/PC1-2), IgG, and CD243 (PB/PC1). CyBorD therapy induced plasma cell signaling responses, including increased expression of CD338 (PC1-3), Pax-5 (PC1,3-6), CD38, CD45 (PC1,4,5), Nanog (PC1,3), KLF-4 (PC1), CD329 (PC2), Nestin, CD269 (PC3), Blimp-1, sXBP-1 (PC4), FGFR-3, CD289, CD19 (PC5), and Notch-1 (PC6). Conversely, decreases were noted in CD243 (PC1,2,6), CD27 (PC1-3), CD362 (PC1,2,4), CD28 (PC2,5,6), CD184 (PC2,3), CD289, CD319 (PC2,6), MMSET (PC5,7), CD47, IgA, CD138, CD200 (PC2), CD19 (PC3), CD52, Nestin, CD269 (PC6), CD44, IRF-4, and c-Myc (PC7) across several PC clusters ( Figure 2C ). In CyBorD responders, plasmablasts showed a reduction in only CD319 (PB/PC2). In plasma cell clusters, increased expression of CD27 (PC1), CD138 (PC4), CD24 (PC5), Notch-1, CD289, and CD19 (PC6) was identified, along with reduced levels of CD184 (PC2) and Sox-2 (PC7) ( Figure 2D ).
Shared molecular alterations in plasmablast and plasma cell (PC) clusters induced by RVd and CyBorD therapies revealed overlapping signaling dynamics, characterized by upregulation of CD38 (PB/PC1), CD45 (PC1,4,5), CD338 (PC1,2), Pax-5 (PC1,6), FGFR3, and CD19 (PC5), alongside downregulation of CD27 (PC2,3), MMSET (PC5,7), CD289 (PC2,6), CD184 (PC3), CD28, CD47, and CD138 (PC2). Notably, a therapy-specific response was observed, in which treatment responders from both cohorts exhibited only CD319 downregulation in plasmablasts (PB/PC2) ( Figure 2E ). Collectively, RVd and CyBorD therapies modulate signaling in myeloma by upregulating CD81, CD38, and CD45 in plasmablasts and FGFR-3, CD338, Pax-5, and KLF-4 in plasma cells, while downregulating MMSET, CD200, c-Myc, CD319, and IRF-4. Moreover, RVd responders exhibited increased CD81 and CD52 with concomitant decreases in CD56, IRF-4, and Nanog; in contrast, CyBorD responders showed elevated Notch-1 alongside reduced CD184 and Sox-2.
Molecular alterations in B cell progenitors by RVd and CyBorD therapies in MM patients
In a previous study, we characterized the phenotypic and molecular modulations of myeloma B cell precursors/progenitors, tracing their development from hematopoietic stem cells (hsc; CD34⁺, CD38 (lo) , CD45 (lo) ) through distinct B cell maturation stages: pre-pro-B cells (CD34+ , CD38het, CD45lo), pro-B cells (CD34lo, CD38+ , CD45lo), pre-BI cells (CD34het, CD10+ , CD19lo), and pre-BII cells (CD34–, CD10lo, CD19lo, IgMlo). Across these stages, we identified the upregulation of key B-cell signaling regulators, including MYD88, MMSET, c-Myc, Notch-1, CXCR4, IRF-4, Bcl-6, and Bcl-2, as well as the ATP-binding cassette transporter CD243 and aberrant plasma cell markers (CD47, CD52, CD269, CD117, and CD25), during the transition from MGUS to smoldering MM and active MM progression. 17
In hsc, RVd-treated patients showed downregulation of MMSET and upregulation of Bcl-2, CD47, CD38, CD25, and the stemness marker CD34 compared to untreated MM patients. In pre-pro-B subset, RVd therapy led to reduced expression of MMSET and aberrant markers (CD52, CD200, CD289, CD362), while increasing CD81 levels. Similarly, in pro-B cells, RVd treatment resulted in reduced MMSET, CD52, CD289, CD362, and the stemness marker Oct3/4, along with upregulated CD81 expression. In both the pre-BI and pre-BII subsets, MMSET was downregulated, while MYD88 was upregulated. Pre-BI cells also exhibited increased CD269 and CD34, whereas pre-BII cells showed elevated CD81 and reduced IRF-4 expression under RVd therapy ( Figure 3A ). When comparing RVd responders to non-responders, responders exhibited reduced CD200 and CD45 expression in hsc, along with increased MYD88 and CD52 levels. In pre-pro-B cells, responders showed upregulated c-Myc, Bcl-2, CD25, MYD88, and CD52 expression, accompanied by reduced Nanog expression. In pro-B cells, we observed increased c-Myc, CD243, Oct3/4, and CD52 expression, while in pre-BI cells, responders showed decreased CD45 expression. In contrast, pre-BII cells in responders exhibited increased CD45, CD52, and sXBP-1 levels, along with reduced CD47 expression ( Figure 3B ).
Evaluation of early B cell progenies under CyBorD therapy revealed distinct alterations, with hsc showing reduced expression of MMSET, c-Myc, CD362, CD243, and CD52, whereas CD45, CD34, CD38, and CD25 were upregulated. In both pre-pro-B and pro-B cells, CyBorD therapy led to the downregulation of MMSET, CD200, and CD52, along with the upregulation of Pax-5 and CD34. Additional changes included reduced CD362 in pre-pro-B cells and reduced RARα2 in pro-B cells, while CD47 and CD38 were upregulated in pre-pro-B cells. Both the pre-BI and pre-BII subsets exhibited MMSET downregulation under CyBorD therapy. Furthermore, pre-BI cells showed increased CD200 and Bcl-2 expression, with reduced CD52 and CD44 expression, whereas pre-BII cells displayed elevated CD10 and reduced IRF-4 and Oct3/4 expression ( Figure 3C ). When comparing CyBorD responders to non-responders, responders exhibited downregulation of CD81 in both hsc and pre-pro-B cells, while pro-B cells showed no significant changes. CyBorD responders also exhibited reduced CD38 levels in pre-BII cells, while no significant modulation was observed in pre-BI cells ( Figure 3D ).
Shared alterations in myeloma B cell progenitors induced by RVd and CyBorD therapies included the downregulation of MMSET from hsc through pre-BII stages, and the downregulation of CD52, specifically in pre-pro-B and pro-B subsets. Moreover, hsc cells presented upregulation of CD25, CD34, and CD38, while pre-pro-B cells showed reduced expression of CD200 and CD362, and pre-BII cells demonstrated IRF-4 suppression. Notably, no common molecular alterations were identified among the responders to either therapy ( Figure 3E ). In summary, RVd and CyBorD therapies in MM patients induced distinct molecular changes, with MMSET downregulation across all B cell progenitors as a shared effect. RVd therapy upregulated MYD88, CD81, and Bcl-2, whereas CyBorD therapy increased Pax-5. Among responders, RVd responders exhibited increased MYD88, CD52, and c-Myc, while CyBorD responders showed reduced CD81 expression in early B cell progenitors.
Impact of RVd and CyBorD on B cell signaling maturation in MM patients
Further profiling of myeloma B cell maturation, from immature (I1-2; CD10lo, CD19+, IgM+, IgD–/lo, CD20–/+, CD27–) to transitional (T1-3; CD10–, CD19+, IgM+, IgDlo, CD20+, CD22+, CD27–) and naive (N; CD19+, IgM+, IgD+, CD20+, CD22+, CD27–) B cells, revealed the upregulation of key B cell signaling regulators, including MMSET, MYD88, c-Myc, CXCR4, and Notch-1. Additionally, the expression of the stem cell marker KLF-4, the ABC transporter CD243, and aberrant plasma cell markers (CD47 and CD52) showed progressive upregulation from immature to naive B cells. 17
In immature B cells, RVd-treated MM patients, compared to untreated patients, exhibited decreased expression of CD200 (I1-2), IRF-4 (I1), MMSET, CD22, CD52, and IgD (I2), alongside increased expression of Pax-5 and CD329 (I1), as well as CD81, IgM, MYD88, KLF-4, CD10, and Notch-1 (I2). Transitional B cells subjected to RVd therapy showed upregulated expression of IgM, CD22, and CD20 (T1-3); Pax-5 (T1-2); CD38 and CD47 (T1,3); KLF-4 and IRF-4 (T1); MYD88 and c-Myc (T2); and CD44 (T3), while reducing CD200 (T1-3), CD19 (T1,3), and MMSET (T3). In naive B cells, RVd therapy led to reduced expression of CD200 and MMSET, accompanied by upregulation of Pax-5, IgM, CD22, and CD81 ( Figure 4A ). In RVd therapy, responders exhibited downregulation of the stemness marker Nanog (I1-2) and upregulation of CD52 and CD20 (I1), as well as IgM and IgD (I2), compared to non-responders. Transitional B cells in RVd responders showed increased expression of CD52 (T1-3), c-Myc (T1), CD81 and CD38 (T2-3), IgM (T2), and IgD (T3). Naive B cells in RVd responders were characterized by the upregulation of sXBP-1, CD81, CD52, and c-Myc, along with the downregulation of CD45 ( Figure 4B ).
In CyBorD-treated MM patients, compared to untreated individuals, immature B cells demonstrated increased expression of Pax-5 and CD10 (I1-2), CD45 (I1), KLF-4, Nanog, Notch-1, and CD47 (I2), while CD319 (I1) and CD22, CD200, and CD52 (I2) were downregulated. Transitional B cells under CyBorD therapy showed upregulation of Pax-5, CD20, and CD45 (T1-3), as well as KLF-4 and IgM (T1), CD22 (T2), and CD47 and CD38 (T3), with a consistent decrease in CD52 (T1-3) across all transitional clusters. In naive B cells, CyBorD therapy induced the upregulation of Pax-5, CD20, IgM, CD81, CD45, and CD22 ( Figure 4C ). Among CyBorD-treated patients, responders exhibited downregulation of CD19 and CD22 (I2) in immature cells and CD52 (I2, T1) in both immature and transitional B cells, as well as reduced CD44 (T3) in transitional B cells. In contrast, MYD88 and CD243 (T3) were upregulated in transitional B cells, while CD319 (N) was increased expression in naive B cells ( Figure 4D ).
During B cell maturation, both RVd and CyBorD therapies induced overlapping alterations in immature, transitional, and naive B cell populations, notably upregulating Pax-5 (I1, T1,2, N), CD20 (T1-3), KLF-4 (I2, T1), CD22 (T2, N ), and IgM (T1, N). Additionally, increased expression of CD10, Notch-1 (I2), and CD47 (T3) was observed, while CD22, CD52, and CD200 (I2) were downregulated. Consistent with findings in progenitor populations, no shared molecular responses were identified among responders from either treatment regimen ( Figure 4E ). Overall, RVd and CyBorD therapies modulated B-cell maturation in MM patients by inducing distinct molecular changes. RVd downregulated MMSET and CD200 across immature, transitional, and naive B cells, while upregulating Pax-5, MYD88, KLF-4, and CD81; among responders, CD52, c-Myc, and CD81 were further elevated. CyBorD decreased CD52 and similarly upregulated Pax-5 and KLF-4. In CyBorD responders, CD52 and CD44 were further reduced in immature and transitional B cells, while MYD88 and CD319 were increased in transitional and naive B cells, respectively.
Therapeutic signaling modulation of memory B cells by RVd and CyBorD regimens in MM patients
Mapping myeloma memory B cells have previously demonstrated the upregulation of key B cell signaling regulators, including MMSET, MYD88, c-Myc, CXCR4, and Notch-1, along with the stem cell marker KLF-4, the ABC transporter CD243, and the aberrant plasma cell marker CD47. These alterations were identified in unswitched memory B cell clusters (UM1-4: CD19+, IgM+, IgDhet, CD20+, CD22+, CD27+) and switched memory B cell clusters (SM1-3: IgM+, IgD–, CD27het, IgG+, IgA+). 17
In unswitched memory B cells, RVd-treated patients compared to untreated MM patients exhibited increased expression of Pax-5, MYD88, IgM (UM1-4), KLF-4, CD47 (UM2-4), IRF-4 (UM1,3), CD22 (UM1,2), CD329, c-Myc, CD44 (UM3,4), CD38 (UM1), CD24, Nanog, CD243, Notch-1, CD362 (UM3), and CD27 (UM4). Concurrently, RVd therapy led to the downregulation of MMSET (UM1), CD19 (UM1-3), CD200 (UM1,3,4), CD184 (UM3,4), CD52, and CD45 (UM3). In switched memory B cells, RVd therapy resulted in the upregulation of Pax-5 (SM1-3), CD329, FGFR3 (SM1,3), CD24 (SM1,2), MYD88, IgM (SM2,3), CD45 (SM1), KLF-4, and IgA (SM3), while downregulating CD289, CD319, CD52, and MMSET (SM1-3), CD184 and CD38 (SM1) ( Figure 5A ). Among RVd responders compared to non-responders, unswitched memory B cells showed increased expression of CD38 (UM1-4), IgM (UM2-4), CD23 (UM1,2,4), c-Myc (UM1,4), MYD88 (UM1), and IgD (UM2), with downregulation of Nanog (UM2,3) and CD19 (UM4). In switched memory B cells, RVd responders displayed upregulation of CD52 and CD28 (SM2), alongside downregulation of CD329 (SM3) ( Figure 5B ).
In unswitched memory B cells, CyBorD-treated patients compared to untreated MM patients showed upregulation of Pax-5 (UM1–4), CD45, CD20 (UM1,2,4), CD27 (UM1,2), CD24, CD44 (UM3,4), CD47, Nanog (UM2,3), CD19, IgM (UM1), CD22 (UM2), CD329, KLF-4, MYD88 (UM3), and CD81 (UM4), alongside decreased expression of CD200 (UM1,4), IgD (UM1-3), CD52, CD22, and CD19 (UM3). In switched memory B cells, CyBorD therapy resulted in increased expression of Pax-5 (SM1–3), Nanog (SM1,2), KLF-4, sXBP-1, CD27, IgM (SM1), CD329, and CD44 (SM3), along with downregulation of CD52 (SM1–3), MMSET, CD319 (SM2,3), CD28 (SM1,3), CD184 (SM1), and IgA (SM3) ( Figure 5C ). In CyBorD responders compared to non-responders, unswitched memory B cells showed upregulation of Pax-5, MMSET (UM2,3), MYD88, c-Myc, CD243 (UM2), CD319, and CD47 (UM3), alongside downregulation of IgD (UM1-3), CD52 (UM1,3), CD19 (UM3,4), and CD22 (UM4). Notably, switched memory B cells in CyBorD responders exhibited no modulation ( Figure 5D ).
In the memory B cell compartment, both treatment regimens induced a shared alteration characterized by downregulation of CD52 and upregulation of Pax5 and CD329. In addition, in unswitched memory B cells, there was increased expression of MYD88, Nanog, KLF4, CD44, CD47, IgM, CD24, and CD242, accompanied by downregulation of CD200 and CD19. In contrast, switched memory B cells exhibited decreased expression of MMSET, CD184, and CD19. No overlapping molecular signatures were observed in memory B cells from responders to either treatment regimen ( Figure 5E ). Notably, comparative profiling of memory B cell subsets in MM patients treated with RVd or CyBorD revealed that both regimens commonly downregulated CD52 while upregulating Pax5 and CD329. Unswitched memory B cells exhibited broad, regimen-specific transcriptional changes (e.g. MYD88, Nanog, KLF4), whereas switched memory B cells showed more limited and distinct alterations unique to each therapy.
Discussion
During normal B-cell development, B-cell maturation into antibody-secreting plasma cells is governed by a tightly integrated network of transcription factors and signaling adaptors. IRF4 accumulates during germinal center reactions to regulate proliferation, class switching and survival, partnering with BLIMP 1 and spliced XBP1 to drive immunoglobulin secretion while repressing B cell identity genes such as PAX5 and BCL6. 19 BLIMP 1 irrevocably commits cells to the plasma cell fate and maintains long-lived plasma cells by silencing germinal center programs. 20 IRF4 is universally upregulated in MM, often driven by translocations, that enforces an “oncogene addiction” phenotype by repressing pro-apoptotic factors such as BIM, activating anti-apoptotic programs, and engaging in a positive feedback loop with MYC, thereby correlating with aggressive disease and poor prognosis and indicating that IRF4 is a compelling therapeutic target. 21 Conversely, although PAX5 is normally silenced in plasma cells, it resurfaces in 10%–15% of CD20-positive MM through incomplete repression or post transcriptional regulation, phenocopying plasmacytoid B cell lymphomas yet paradoxically triggering apoptosis and growth arrest upon re-expression, which underscores its complex but potential therapeutic relevance. 22 Assessment of transcriptional regulators across B cell ontogeny under RVd and CyBorD treatment revealed both shared and regimen-specific modulation. Both regimens downregulated IRF4 in early progenitors and PCs but upregulated PAX5 broadly throughout B-cell maturation, indicating a shift toward B-cell identity regulation that may influence differentiation and treatment response. RVd’s unique pattern, suppressing IRF4 in immature B cells but transiently elevating it in transitional and memory subsets, along with responder-specific IRF4 downregulation in PCs, suggests that dynamic remodeling is associated with effective tumor clearance ( Figure 6 ).
MYD88 transduces TLR signals via the NF-κB and MAPK pathways to support germinal center expansion and plasma cell longevity during normal B-cell development. 23 , 24 In MM, despite the absence of the MYD88 L265P oncoprotein observed in Waldenström’s macroglobulinemia and ABC DLBCL, intact wild type MYD88 signaling in malignant PC can promote survival, drive autocrine IL-6-mediated prosurvival cytokine production within the bone marrow niche, and confer resistance to proteasome inhibitors and immunomodulatory drugs. 25 MYD88 was consistently upregulated in unswitched memory B cells by both therapies, implicating MYD88-driven signaling in the treatment response. Notably, RVd broadly induced MYD88 and c-Myc across early progenitors and multiple B cell subsets, indicating the activation of proliferative and survival pathways that might reflect both tumor adaptation and immune engagement. CyBorD modulated MYD88 and c-Myc in restricted compartments, suggesting divergent impact on tumor microenvironment interactions and malignant cell plasticity. In normal B cell development, c-Myc promotes early B cell proliferation and transiently fuels germinal center responses under BCL6 control before yielding to IRF4 and BLIMP-1 during terminal differentiation. 26 During early B lymphopoiesis, Pax5-driven activation of CD19 and BLK cooperates with c‑Myc–mediated signaling to fuel pro‑B to pre‑B proliferation and survival, while high CXCR4 expression anchors these progenitors in the CXCL12‑rich bone marrow niche until maturation cues permit egress. 27 Emerging Notch‑1 induction upon pre‑BCR engagement then amplifies progenitor expansion via Hes1 transcriptional activation, priming cells for downstream germinal center entry. 28 Spliced XBP1 (sXBP1) drives endoplasmic reticulum biogenesis and reinforces IRF4/BLIMP-1-mediated transcription to sustain the high capacity immunoglobulin synthesis and secretory function of terminally differentiated plasma cells. 29 In MM, hypoxia driven CXCR4 overexpression in the bone marrow niche enhances tumor cell survival and angiogenesis, while ligand induced Notch 1 activation variably promotes survival or apoptotic sensitization, and MYC translocations/amplifications drive c-Myc-dependent proliferation and apoptosis evasion via Mcl-1/Bcl-xL upregulation, with t(4;14) associated FGFR3/MMSET overexpression further activating MAPK/STAT3 to block differentiation. 30 , 31 Concurrently, loss of spliced XBP1 correlates with high risk cytogenetics, advanced stage and proteasome inhibitor resistance, reversible upon sXBP1 restoration, whereas aberrant Blimp-1 reexpression in proliferating myeloma cells partners with Aiolos to repress pro apoptotic genes, highlighting both as prognostic biomarkers and therapeutic targets. 32 Both RVd and CyBorD converged on suppressing CXCR4 and FGFR3 in PC clusters, likely disrupting homing and survival signaling critical to myeloma persistence. RVd's unique induction of NOTCH1 in unswitched memory B cells and FGFR3 in switched memory B cells, along with the selective reduction of CXCR4 and sXBP1 in plasmablasts and plasma cells, suggests targeted disruption of malignant niche interactions and the unfolded protein response. In contrast, CyBorD upregulated sXBP1 and BLIMP1 in memory and plasma cell compartments, highlighting the distinct regulation of differentiation and stress response pathways. Lenalidomide induces Blimp-1 degradation via ubiquitination, derepressing targets such as CUL4A and reducing Aiolos to promote MM cell death, 32 while CRISPR Blimp-1 knockout in anti-BCMA CAR-T cells enhances memory-like traits, protein synthesis, and antitumor activity, indicating that Blimp-1 is a target to increase T-cell function and limit exhaustion. 33 In RVd responders, Blimp-1 and FGFR3 are selectively downregulated in the PC4 compartment, a change that likely diminishes plasma-cell survival/differentiation and oncogenic FGFR3 signaling. Unlike healthy plasma cells, which downregulate BCL2 upon differentiation, malignant plasma cells in the bone marrow maintain high BCL2 levels, reinforced by NF κB and STAT3 signaling, to sequester pro-apoptotic proteins and evade chemotherapy-induced apoptosis. Notably, roughly 10% of MM cases, particularly those harboring t(11;14), show BCL2 overexpression driven by promoter hypomethylation, loss of miR15/16, or clonal evolution, further supporting its role as a therapeutic target. 34 Differential BCL2 induction by each regimen in early progenitors reflects varied BCL2 dependence and reveals a vulnerability that could be exploited, for example, with BH3 mimetics or NF-κB/STAT3 pathway inhibitors, to overcome resistance in relapsed/refractory MM. Both RVd and CyBorD reprogram B-cell transcription, promoting B-cell identity (PAX5 and BCL2 upregulation in B-cell progenitors and IRF4, CXCR4, and FGFR3 downregulation in plasma cells) while inducing regimen-specific changes in pathways (MYD88, c-Myc, NOTCH1, sXBP1, BLIMP1) that likely drive differential treatment responses and reveal therapeutic vulnerabilities.
Key stemness regulators such as OCT3/4 (POU5F1), SOX2, NANOG and KLF4, together with nestin and the retinoic-acid receptor isoform RARα2, are essential for pluripotency and self-renewal, while also forming an integrated network that orchestrates early B-cell development. 35-38 During early B-cell ontogeny, low OCT3/4 cooperates with SOX2 and transient NANOG expression to maintain progenitor identity, while KLF4 and nestin peak in pro- and pre-B cells to regulate cell cycle progression, survival and cytoskeletal integrity, and RARα2-mediated retinoic-acid signaling drives germinal-center formation, affinity maturation and class switching. As cells traverse the plasmablast stage and commit to the plasma-cell lineage, expression of OCT3/4, SOX2, NANOG and nestin progressively declines, RARα2 continues to enforce lineage-specific gene programs, and elevated or experimentally induced KLF4 activity imposes cell-cycle arrest and balanced apoptosis to promote terminal plasma-cell differentiation. 35 , 36 , 38-40 In MM, aberrant reactivation of this stemness network endows CD138⁺ plasma cells with enhanced self-renewal, apoptosis resistance and clonogenic potential: OCT3/4, SOX2 and NANOG rekindle pluripotency circuits; KLF4 drives prosurvival autophagy via SQSTM1/p62; nestin stimulates PI3K/AKT- and NF-κB-dependent proliferation; and RARα2 activates the Wnt and Hedgehog pathways while upregulating ABCC3 and Bcl-2 to confer chemoresistance. 17 , 41 , 42 These changes drive tumor growth, relapse, drug resistance, and an immunosuppressive bone-marrow niche that promotes MM; therefore, we compared RVd versus CyBorD by profiling B-cell stemness. Both therapies uniformly upregulated KLF4 in immature, transitional, and memory B cells, as well as in PC clusters, suggesting the shared induction of a stemness-associated transcriptional program potentially involved in cellular plasticity and survival. The uniform elevation of KLF4 might contribute to a resilient subpopulation that persists under therapy, representing a target for strategies aimed at preventing minimal residual disease. In contrast, NANOG expression exhibited treatment- and subset-specific dynamics. While both regimens increased NANOG levels in unswitched memory B cells, CyBorD uniquely increased NANOG in additional compartments, including immature, switched memory B cells, and PCs, indicating broader activation of self-renewal pathways that may underpin adaptive resistance. Conversely, RVd selectively diminished NANOG in plasmablasts and PCs and notably suppressed NANOG across multiple B-cell stages among clinical responders. These findings suggest that downregulation of NANOG may be a hallmark of effective anti-myeloma activity with RVd, implicating NANOG as both a biomarker of response and a potential node to sensitize resistant cells. RVd suppressed OCT3/4 in pro-B, plasmablast and PC clusters, supporting the targeting of core pluripotency factors, though paradoxical upregulation in pro-B cells of responders suggests complex cell-state remodeling that may prime subsets for differentiation or immune clearance. CyBorD uniquely modulated nestin in PC clusters, suggesting that altered cytoskeletal/niche interactions that influence malignant fitness and drug sensitivity. The differential regulation of RARα2, which is decreased by CyBorD in pro-B and by RVd in responder PCs, also highlights retinoid signaling as a convergent pathway influenced by these regimens, offering a candidate axis for therapeutic modulation to disrupt tumor maintenance ( Figure 6 ). By mapping these dynamic shifts, we reveal how therapeutic reprogramming of the stemness network underlies treatment responses and resistance mechanisms, and point toward targeted strategies to disrupt malignant self-renewal and restore effective anti-tumor immunity in MM.
Among malignant plasma cell-specific targets, CD319 (SLAMF7) is ubiquitously expressed on normal and malignant plasma cells, conferring superior diagnostic sensitivity and prognostic stratification. 43 CD56, expressed in 65%–80% of MM cases but is absent in normal plasma cells, is linked to osteolysis and extramedullary spread. Its absence correlates with poor outcomes by increasing MMP-9-mediated invasion, whereas its overexpression enhances proteasome inhibitor-induced apoptosis. 44 In 30% of MM patients, aberrant CD117 expression activates PI3K/AKT and MAPK signaling to promote malignant plasma cell survival; CD117⁺ cells may reflect favorable marrow profiles, but CD117⁺CD56⁻ phenotypes are associated with aggressive, drug-resistant disease. 45 MMSET, a histone methyltransferase overexpressed in t(4;14) MM (15% of cases), catalyzes H3K36 demethylation to activate oncogenes (MAF, c-MYC), represses tumor suppressors via Sin3a/HDAC1/LSD1 interactions, and upregulates cell-cycle and anti-apoptotic genes, correlating with poor prognosis. 46 Finally, CD44 variants (v6, v9) are upregulated in aggressive and extramedullary MM, mediating hyaluronan-driven proteasome inhibitor resistance through γ -secretase-mediated cleavage of CD44’s intracellular domain, and elevated soluble CD44 levels are associated with poor prognosis. 47
Both RVd and CyBorD therapies induced broadly overlapping immunophenotypic changes across the B cell continuum, reflecting shared mechanisms that affected myeloma biology at multiple differentiation stages. Early downregulation of MMSET, CD200, and CD362 in B progenitors likely impaired epigenetic and microenvironmental support essential for malignant propagation. Concurrent suppression of CD52 in early and immature B cells may reduce tumor-cell survival and niche retention, while upregulation of CD47 in transitional and memory B cells may increase “don’t eat me” signaling, contributing to immune evasion. Both regimens consistently increased CD81 expression in naive B cells and PC clusters, indicating that the modulation of adhesion and signaling pathways is relevant to differentiation and therapeutic responsiveness. The upregulation of CD329 (Siglec-9) and CD44 in memory B cell compartments highlighted the importance of immunomodulatory and adhesion molecules potentially critical for myeloma cell migration and immune interactions. The downregulation of MMSET and CD319 in memory B cells and PCs suggested the impairment of epigenetic regulation and tumor-promoting signaling. Importantly, both treatments also decreased CD28, CD47, and CD200 expression in plasma cells-markers associated with tumor growth and immune evasion, indicating a therapeutic impact on malignant survival and microenvironmental crosstalk. The elevation of CD338 (ABCG2) in PCs may have reflected selection of drug-resistant subclones, highlighting a potential mechanism of treatment escape ( Figure 6 ).
Among additional plasma cell-associated markers, CD200, normally restricted to early B‐cell precursors and memory B cells, is aberrantly overexpressed approximately in 70% of MM cases, where it facilitates immune evasion, correlates with advanced disease, reduced overall survival, and the adverse rs1131199 GG genotype. 48 In parallel, CD28, ubiquitously expressed on plasma cells, is upregulated in high-risk MM, where CD28-CD86 autocrine signaling activates PI3K/Akt and NF-κB pathways, effectively substituting for IGF-1/IL-6 signaling to promote cell survival, and engages bone marrow dendritic cells to induce IL-6 and IDO production, further sustaining tumor persistence. 49 By contrast, CD81, a tetraspanin sustaining CD19 trafficking in the BCR coreceptor complex, persists on maturing plasma cells; in MM, high CD81 marks poorly differentiated CD19⁺CD81⁺ clones with distinct mutation profiles that drive chemoresistance, relapse, and rapid progression from smoldering to active disease. 50 Several cell surface markers implicated in MM pathogenesis include BCMA (CD269), which is essential for plasmablast and plasma cell survival and is uniformly overexpressed on MM cells despite γ -secretase-mediated shedding that generates soluble BCMA. Membrane-bound BCMA activates NF-κB to sustain proliferation and immune suppression and serves as a key target in MM immunotherapy. 51 Moreover, syndecan‐2 (CD362), a heparan sulfate proteoglycan, is upregulated in hsc and early B‐lineage cells across MGUS to relapsed MM, where it activates NF-κB, PI3K/Akt, and Notch-1 to drive proliferation, survival, and drug resistance, while its sulfated glycosaminoglycans promote niche adhesion. 17 Siglec-9 (CD329), enriched on switched memory B cells in MM, delivers inhibitory signals via ITIMs to recruit SHP-1/2 and dampen anti‐tumor immunity, correlating with disease progression. 17 , 52 CD47, “don’t eat me” receptor, is upregulated from MGUS to MM, especially in cases with 17p deletions and high β2-microglobulin, to bind SIRPα on macrophages, inhibit phagocytosis, and associate with proteasome inhibitor resistance and poor survival. 17 , 53 Finally, CD52, which normally attenuates BCR signaling and enforces tolerance in naïve and memory B cells, is largely lost on CD45⁻CD38⁺⁺ MM plasma cells; its residual expression in a minor CD45⁺ subset aligns with c-MAF/FGFR3 translocations but has not translated into clinical benefit from alemtuzumab. 54
Despite these parallels, RVd and CyBorD impart distinct molecular footprints. RVd uniquely upregulated CD47 and CD81 in early B-cell progenitors and transitional B cells, along with CD269 (BCMA) in pre-BI cells, while downregulating CD289 and CD362 in early B-cell progenitors, suggesting enhanced targeting of immune-evasion and pro-survival pathways that may increase malignant-cell persistence and resistance to clearance, which could help explain its clinical efficacy in disrupting malignant niches. The responders to RVd specifically showed decreased CD200 and CD47 in early B cell progenitors, together with increased CD52 and CD243 ( P -glycoprotein) expression, potentially reflecting adaptive changes in immune regulation and drug efflux that affect treatment sensitivity. During the transitional stages of B-cell maturation, RVd uniquely elevates CD81, CD329, and CD44, and downregulated CD200, changes that may reduce immune evasion, alter adhesion and homing to supportive niches, and thereby modulate microenvironmental interactions. CyBorD imparted unique modulation by decreasing CD362, CD243, and CD52 in HSCs and early progenitors, as well as CD200 and CD44 in early B-cell progenitors, while increasing CD47 and CD200 in B-cell progenitors and transitional B cells, indicating differential impacts on immune checkpoints, cell‒cell adhesion, and progenitor survival. Conversely, CyBorD specifically decreased CD319 and CD52 in immature and transitional B cells, which may alter signaling important for activation and niche retention. In the memory B cell compartment, RVd induced CD243 and CD362 in unswitched memory B cells while decreasing MMSET, and lowered CD289 in switched memory B cells, changes that may influence antigen responsiveness, epigenetic regulation, and migratory behavior. In contrast, CyBorD uniquely increased CD81 and CD44 in memory B cell populations, suggesting that enhanced adhesion and signaling could affect trafficking and cell survival. In mature plasma cells and plasmablast clusters, RVd induced CD81 and CD117 expression while suppressing BCMA (CD269), CD289, CD200, MMSET, and CD319, indicating broad suppression of survival and adhesion molecules and potential disruption of plasma-cell niches and pro-tumor signaling. The observed downregulation of CD329 also suggested a diminished immune-evasion capacity. Conversely, CyBorD increased CD329 in plasma cells but reduced CD243, CD44, CD362, and CD52, which may reflect distinct remodeling of adhesion, drug-resistance, and immune-escape pathways. Overall, both regimens remodel B cell phenotypes to impair tumor-supportive epigenetic and microenvironmental programs while leaving distinct molecular footprints in adhesion, immune-checkpoint, and drug-resistance pathways that may drive differential response and resistance ( Figure 6 ).
Specific immunophenotypic signatures may inform first-line therapy selection between RVd and CyBorD. Both regimens reinforced B-cell identity, suppressed tumor-supporting markers (MMSET, CD200, CD362, CD52), and increased KLF4, indicating shared anti-myeloma mechanisms and potential targets to prevent relapse. RVd demonstrated broader MYD88/c-Myc activation, reduced NANOG expression, and distinctive CD47/CD81 modulation, consistent with enhanced clearance of MYD88/c-Myc–driven disease. In contrast, CyBorD activated BLIMP1 and sXBP1, implicated retinoid pathways, and remodeled adhesion and immune-evasion markers, suggesting preferential efficacy in plasma cell–dominant or stress-adapted myeloma phenotypes. Following RVd or CyBorD failure, key therapeutic vulnerabilities included persistent immune-evasion signals (CD47, CD200), drug-efflux activity (CD338/ABCG2), and altered adhesion and epigenetic regulators (CD81, CD44, MMSET). Targeting these pathways, together with BCL2, MYD88, c-Myc, FGFR3, and CXCR4 signaling, may help overcome resistance, while modulation of BCMA and CD329 supported antigen-directed strategies, and inhibition of stemness and niche factors (NANOG, OCT3/4, nestin, RARα2) could prevent relapse. Collectively, these findings provide a framework for integrating immunophenotypic profiling into therapeutic decision-making, enabling more precise selection and sequencing of myeloma treatments to improve clinical outcomes.
Supplementary Material
Supplementary material
Supplementary information Materials and Methods.
Supplementary material
Supplementary Table S1.
Supplementary material
Supplementary Table S2.
Supplementary material
Supplementary Table legends
Supplementary material
Supplementary Table S3
Acknowledgments
ZV performed the experiments and analyzed the data, DC performed the experiments and analyzed the data, and contributed to the writing of the manuscript. GB, MH and LK performed the bioinformatics analyses. MEM and KS performed the experiments. LD, ML, EK and DMD contributed with clinical specimens and collected the clinical data. KCA edited the manuscript. JJ conceived and designed the study, performed experiments and analyzed data, collected clinical data and wrote the manuscript.
Supplemental material
Supplemental data for this article can be accessed at https://doi.org/10.1080/15384047.2025.2603100 .
Disclosure statement
The authors declare the following competing interests: EK declares honoraria (Amgen, Janssen, Genesis Pharma, Pfizer, Takeda, GSK) and research support (Amgen, Janssen, Pfizer); KCA declares advisory role (Pfizer, Amgen, Astrazeneca, Janssen, Precision Biosciences), board membership (C4 Therapeutics, Raqia, NextRNA, Window, Mana, Starton) and ownership interests (C4 Therapeutics, Oncopep, NextRNA, Raqia). Other authors declare no competing financial interests.
Data availability statement
All data that support the findings of this study are presented in the paper or available upon reasonable request.
Ethical approvals
Ethical approval was granted by the Biomedical Research Center’s local ethics committee (Protocol No. Myelom 001) and Dana-Farber Cancer Institute (Protocol No. 10-106), in accordance with the Declaration of Helsinki. Written informed consent was obtained from all participants.
References
1. Zanwar S , Rajkumar SV . Current risk stratification and staging of multiple myeloma and related clonal plasma cell disorders . Leukemia . 2025 ; 39 : 2610 – 2617 . doi: 10.1038/s41375-025-02654-y . 40702148 PMC12589131
2. Cardona-Benavides IJ , de Ramon C , Gutierrez NC . Genetic abnormalities in multiple myeloma: prognostic and therapeutic implications . Cells . 2021 ; 10 : 336 . doi: 10.3390/cells10020336 . 33562668 PMC7914805
3. Fotiou D , Katodritou E . From biology to clinical practice: the bone marrow microenvironment in multiple myeloma . J Clin Med . 2025 ; 14 : 327 . doi: 10.3390/jcm14020327 . 39860333 PMC11765558
4. Zavaleta-Monestel E , Quesada-Villasenor R , Barrantes-Lopez M , Arguedas-Chacon S , Campos-Hernandez J , Rojas-Chinchilla C , Quesada-Villaseñor R , Barrantes-López M , Arguedas-Chacón S , Campos-Hernández J , et al. Advancements in the treatment of multiple myeloma . Cureus . 2024 ; 16 : e74970 . doi: 10.7759/cureus.74970 . 39744254 PMC11691229
5. Joseph NS , Kaufman JL , Dhodapkar MV , Hofmeister CC , Almaula DK , Heffner LT , Gupta VA , Boise LH , Lonial S , Nooka AK . Long-term follow-up results of lenalidomide, bortezomib, and dexamethasone induction therapy and risk-adapted maintenance approach in newly diagnosed multiple myeloma . J Clin Oncol . 2020 ; 38 : 1928 – 1937 . doi: 10.1200/JCO.19.02515 . 32298201 PMC7587409
6. McCaughan GJ , Gandolfi S , Moore JJ , Richardson PG . Lenalidomide, bortezomib and dexamethasone induction therapy for the treatment of newly diagnosed multiple myeloma: a practical review . Br J Haematol . 2022 ; 199 : 190 – 204 . doi: 10.1111/bjh.18295 . 35796524 PMC9796722
7. Richardson PG , Durie BG , Rosinol L , Mateos MV , Dispenzieri A , Moreau P , Rosiñol L , Kumar S , Raje N , Munshi N , et al. Clinical perspectives on the optimal use of lenalidomide plus bortezomib and dexamethasone for the treatment of newly diagnosed multiple myeloma . Haematologica . 2023 ; 108 : 2894 – 2912 . doi: 10.3324/haematol.2022.282624 . 37608773 PMC10620581
8. Areethamsirikul N , Masih-Khan E , Chu CM , Jimenez-Zepeda V , Reece DE , Trudel S , Kukreti V , Tiedemann R , Chen C . CyBorD induction therapy in clinical practice . Bone Marrow Transplant . 2015 ; 50 : 375 – 379 . doi: 10.1038/bmt.2014.288 . 25599165
9. Reeder CB , Reece DE , Kukreti V , Chen C , Trudel S , Hentz J , Noble B , Pirooz NA , Spong JE , Piza JG , et al. Cyclophosphamide, bortezomib and dexamethasone induction for newly diagnosed multiple myeloma: high response rates in a phase II clinical trial . Leukemia . 2009 ; 23 : 1337 – 1341 . doi: 10.1038/leu.2009.26 . 19225538 PMC2711213
10. Figueiredo A , Atkins H , Mallick R , Kekre N , Kew A , McCurdy A . Cyclophosphamide-bortezomib-dexamethasone compared with bortezomib-dexamethasone in transplantation-eligible patients with newly diagnosed multiple myeloma . Curr Oncol . 2020 ; 27 : e81 – e85 . doi: 10.3747/co.27.5385 . 32489256 PMC7253752
11. Lu K , Wang W , Liu Y , Xie C , Liu J , Xing L . Advancements in microenvironment-based therapies: transforming the landscape of multiple myeloma treatment . Front Oncol . 2024 ; 14 : 1413494 . doi: 10.3389/fonc.2024.1413494 . 39087026 PMC11288838
12. Guo H , Yang J , Wang H , Liu X , Liu Y , Zhou K . Reshaping the tumor microenvironment: the versatility of immunomodulatory drugs in B-cell neoplasms . Front Immunol . 2022 ; 13 : 1017990 . doi: 10.3389/fimmu.2022.1017990 . 36311747 PMC9596992
13. Dabbish E , Scoditti S , Shehata MNI , Ritacco I , Ibrahim MAA , Shoeib T , Sicilia E . Insights on cyclophosphamide metabolism and anticancer mechanism of action: a computational study . J Comput Chem . 2024 ; 45 : 663 – 670 . doi: 10.1002/jcc.27280 . 38088485
14. Swan D , Gurney M , Krawczyk J , Ryan AE , O'Dwyer M . Beyond DNA damage: exploring the immunomodulatory effects of cyclophosphamide in multiple myeloma . HemaSphere . 2020 ; 4 : e350 . doi: 10.1097/HS9.0000000000000350 . 32309787 PMC7162079
15. Pellom STJr , Dudimah DF , Thounaojam MC , Uzhachenko RV , Singhal A , Richmond A , Shanker A . Bortezomib augments lymphocyte stimulatory cytokine signaling in the tumor microenvironment to sustain CD8+T cell antitumor function . Oncotarget . 2017 ; 8 : 8604 – 8621 . doi: 10.18632/oncotarget.14365 . 28052005 PMC5352426
16. Chimote AA , Alshwimi AO , Chirra M , Gawali VS , Powers-Fletcher MV , Hudock KM , Conforti L . Immune and ionic mechanisms mediating the effect of dexamethasone in severe COVID-19 . Front Immunol . 2023 ; 14 : 1143350 . doi: 10.3389/fimmu.2023.1143350 . 37033961 PMC10080085
17. Jakubikova J , Cholujova D , Beke G , Hideshima T , Klucar L , Leiba M , Jamroziak K , Richardson PG , Kastritis E , Dorfman DM , et al. Heterogeneity of B cell lymphopoiesis in patients with premalignant and active myeloma . JCI Insight . 2023 ; 8 ( 3 ): e159924 . doi: 10.1172/jci.insight.159924 . 36752202 PMC9977435
18. Cholujova D , Beke G , Klucar L , Drgona L , Valuskova Z , Leiba M , Kastritis E , Dorfman DM , Anderson KC , Jakubikova J . CyTOF profiling of bone marrow immune dynamics across myeloma stages . Oncoimmunology . 2025 ; 14 : 2542333 . doi: 10.1080/2162402X.2025.2542333 . 40820826 PMC12363513
19. Willis SN , Good-Jacobson KL , Curtis J , Light A , Tellier J , Shi W , Smyth GK , Tarlinton DM , Belz GT , Corcoran LM , et al. Transcription factor IRF4 regulates germinal center cell formation through a B cell-intrinsic mechanism . J Immunol . 2014 ; 192 : 3200 – 3206 . doi: 10.4049/jimmunol.1303216 . 24591370
20. Shapiro-Shelef M , Lin KI , Savitsky D , Liao J , Calame K . Blimp-1 is required for maintenance of long-lived plasma cells in the bone marrow . J Exp Med . 2005 ; 202 : 1471 – 1476 . doi: 10.1084/jem.20051611 . 16314438 PMC2213334
21. Hagman J . Critical functions of IRF4 in B and T Lymphocytes . J Immunol . 2017 ; 199 : 3715 – 3716 . doi: 10.4049/jimmunol.1701385 . 29158346
22. Lin P , Mahdavy M , Zhan F , Zhang HZ , Katz RL , Shaughnessy JD . Expression of PAX5 in CD20-positive multiple myeloma assessed by immunohistochemistry and oligonucleotide microarray . Mod Pathol . 2004 ; 17 : 1217 – 1222 . doi: 10.1038/modpathol.3800169 . 15195108
23. Kang SM , Yoo DG , Kim MC , Song JM , Park MK , O E , Quan F , Akira S , Compans RW . MyD88 plays an essential role in inducing B cells capable of differentiating into antibody-secreting cells after vaccination . J Virol . 2011 ; 85 : 11391 – 11400 . doi: 10.1128/JVI.00080-11 . 21865389 PMC3194946
24. Tian M , Hua Z , Hong S , Zhang Z , Liu C , Lin L , Chen J , Zhou X , DeFranco AL , Hou B . B Cell-intrinsic MyD88 signaling promotes initial cell proliferation and differentiation to enhance the germinal center response to a virus-like particle . J Immunol . 2018 ; 200 : 937 – 948 . doi: 10.4049/jimmunol.1701067 . 29282308
25. de Groen RAL , Schrader AMR , Kersten MJ , Pals ST , Vermaat JSP . MYD88 in the driver's seat of B-cell lymphomagenesis: from molecular mechanisms to clinical implications . Haematologica . 2019 ; 104 : 2337 – 2348 . doi: 10.3324/haematol.2019.227272 . 31699794 PMC6959184
26. de Barrios O , Meler A , Parra M . MYC's fine line between B cell development and malignancy . Cells . 2020 ; 9 : 523 . doi: 10.3390/cells9020523 . 32102485 PMC7072781
27. Weber TS . Cell cycle-associated CXCR4 expression in germinal center B cells and its implications on affinity maturation . Front Immunol . 2018 ; 9 : 1313 . doi: 10.3389/fimmu.2018.01313 . 29951060 PMC6008520
28. Kang JA , Kim WS , Park SG . Notch1 is an important mediator for enhancing of B-cell activation and antibody secretion by Notch ligand . Immunology . 2014 ; 143 : 550 – 559 . doi: 10.1111/imm.12333 . 24913005 PMC4253503
29. Hu CC , Dougan SK , McGehee AM , Love JC , Ploegh HL . XBP-1 regulates signal transduction, transcription factors and bone marrow colonization in B cells . EMBO J . 2009 ; 28 : 1624 – 1636 . doi: 10.1038/emboj.2009.117 . 19407814 PMC2684024
30. Das SK , Lewis BA , Levens D . MYC: a complex problem . Trends Cell Biol . 2023 ; 33 : 235 – 246 . doi: 10.1016/j.tcb.2022.07.006 . 35963793 PMC9911561
31. Giorgiutti S , Rottura J , Korganow AS , Gies V . CXCR4: from B-cell development to B cell-mediated diseases . Life Sci Alliance . 2024 ; 7 : e202302465 . doi: 10.26508/lsa.202302465 . 38519141 PMC10961644
32. Hung KH , Su ST , Chen CY , Hsu PH , Huang SY , Wu WJ , Lin F , Tsai M . Aiolos collaborates with Blimp-1 to regulate the survival of multiple myeloma cells . Cell Death Differ . 2016 ; 23 : 1175 – 1184 . doi: 10.1038/cdd.2015.167 . 26823144 PMC4946885
33. Battram AM , Mane-Pujol J , Moreno DF , Oliver-Caldes A , Carpio J , Cardus O , Mañé-Pujol J , Oliver-Caldés A , Rodríguez-Lobato LG , Urbano-Ispizua Á , et al. Genetic disruption of Blimp-1 drastically augments the antitumor efficacy of BCMA-targeting CAR T cells . Blood Adv . 2025 ; 9 : 627 – 641 . doi: 10.1182/bloodadvances.2024013209 . 39642314 PMC11847098
34. Cao Q , Wu X , Zhang Q , Gong J , Chen Y , You Y , Shen J , Qiang Y . Mechanisms of action of the BCL-2 inhibitor venetoclax in multiple myeloma: a literature review . Front Pharmacol . 2023 ; 14 : 1291920 . doi: 10.3389/fphar.2023.1291920 . 38026941 PMC10657905
35. Bernal A , Arranz L . Nestin-expressing progenitor cells: function, identity and therapeutic implications . Cell Mol Life Sci . 2018 ; 75 : 2177 – 2195 . doi: 10.1007/s00018-018-2794-z . 29541793 PMC5948302
36. Ghaleb AM , Yang VW . Kruppel-like factor 4 (KLF4): what we currently know . Gene . 2017 ; 611 : 27 – 37 . doi: 10.1016/j.gene.2017.02.025 . 28237823 PMC5391259
37. Zhang J , Gao Y , Yu M , Wu H , Ai Z , Wu Y , Liu H , Du J , Guo Z , Gaetano C . Retinoic acid induces embryonic stem cell differentiation by altering both encoding RNA and microRNA expression . PLoS One . 2015 ; 10 : e0132566 . doi: 10.1371/journal.pone.0132566 . 26162091 PMC4498831
38. Kallas A , Pook M , Trei A , Maimets T . SOX2 is regulated differently from NANOG and OCT4 in human embryonic stem cells during early differentiation initiated with sodium butyrate . Stem Cells Int . 2014 ; 2014 : 298163 . doi: 10.1155/2014/298163 . 24707296 PMC3951062
39. Marks E , Ortiz C , Pantazi E , Bailey CS , Lord GM , Waldschmidt TJ , Noelle RJ , Elgueta R . Retinoic acid signaling in B cells is required for the generation of an effective T-Independent immune response . Front Immunol . 2016 ; 7 : 643 . doi: 10.3389/fimmu.2016.00643 . 28066447 PMC5179524
40. Schoenhals M , Jourdan M , Seckinger A , Pantesco V , Hose D , Kassambara A , Moreaux J , Klein B . Forced KLF4 expression increases the generation of mature plasma cells and uncovers a network linked with plasma cell stage . Cell Cycle . 2016 ; 15 : 1919 – 1928 . doi: 10.1080/15384101.2016.1191709 . 27230497 PMC4968905
41. Guo W , Wang H , Chen P , Shen X , Zhang B , Liu J , Peng H , Xiao X . Identification and characterization of multiple myeloma stem cell-like cells . Cancers (Basel) . 2021 ; 13 : 3523 . doi: 10.3390/cancers13143523 . 34298738 PMC8306148
42. Riz I , Hawley TS , Hawley RG . KLF4-SQSTM1/p62-associated prosurvival autophagy contributes to carfilzomib resistance in multiple myeloma models . Oncotarget . 2015 ; 6 : 14814 – 14831 . doi: 10.18632/oncotarget.4530 . 26109433 PMC4558117
43. Soh KT , Tario JDJr , Hahn T , Hillengass J , McCarthy PL , Wallace PK . CD319 (SLAMF7) an alternative marker for detecting plasma cells in the presence of daratumumab or elotuzumab . Cytometry B Clin Cytom . 2021 ; 100 : 497 – 508 . doi: 10.1002/cyto.b.21961 . 33017079 PMC8940539
44. Li L , Li X , Shang A , Zhao Y , Jin L , Zhao M , Shen W . Prognostic significance of CD56 antigen in newly diagnosed multiple myeloma: a real-world retrospective study . Medicine (Baltimore) . 2022 ; 101 : e30988 . doi: 10.1097/MD.0000000000030988 . 36221376 PMC9542762
45. Zheng D , Zhu M , Li Q , Wan W , Chen Y , Jing H . Dual negativity of CD56 and CD117 links to unfavorable cytogenetic abnormalities and predicts poor prognosis in multiple myeloma . J Clin Med . 2022 ; 11 : 6524 . doi: 10.3390/jcm11216524 . 36362752 PMC9653854
46. Popovic R , Martinez-Garcia E , Giannopoulou EG , Zhang Q , Zhang Q , Ezponda T , Shah MY , Zheng Y , Will CM , Small EC , et al. Histone methyltransferase MMSET/NSD2 alters EZH2 binding and reprograms the myeloma epigenome through global and focal changes in H3K36 and H3K27 methylation . PLoS Genet . 2014 ; 10 : e1004566 . doi: 10.1371/journal.pgen.1004566 . 25188243 PMC4154646
47. Kikuchi J , Kodama N , Takeshita M , Ikeda S , Kobayashi T , Kuroda Y , Uchiyama M , Osada N , Bogen B , Yasui H , et al. EMD originates from hyaluronan-induced homophilic interactions of CD44 variant-expressing MM cells under shear stress . Blood Adv . 2023 ; 7 : 508 – 524 . doi: 10.1182/bloodadvances.2022007291 . 35930695 PMC9979770
48. Alapat D , Coviello-Malle J , Owens R , Qu P , Barlogie B , Shaughnessy JD , Lorsbach RB . Diagnostic usefulness and prognostic impact of CD200 expression in lymphoid malignancies and plasma cell myeloma . Am J Clin Pathol . 2012 ; 137 : 93 – 100 . doi: 10.1309/AJCP59UORCYZEVQO . 22180482 PMC3632327
49. Nair JR , Carlson LM , Koorella C , Rozanski CH , Byrne GE , Bergsagel PL , Shaughnessy JP , Boise LH , Chanan-Khan A , Lee KP . CD28 expressed on malignant plasma cells induces a prosurvival and immunosuppressive microenvironment . J Immunol . 2011 ; 187 : 1243 – 1253 . doi: 10.4049/jimmunol.1100016 . 21715687 PMC3152989
50. Paiva B , Puig N , Cedena MT , de Jong BG , Ruiz Y , Rapado I , Martinez-Lopez J , Cordon L , Alignani D , Delgado JA , et al. Differentiation stage of myeloma plasma cells: biological and clinical significance . Leukemia . 2017 ; 31 : 382 – 392 . doi: 10.1038/leu.2016.211 . 27479184 PMC5439510
51. Shah N , Chari A , Scott E , Mezzi K , Usmani SZ . B-cell maturation antigen (BCMA) in multiple myeloma: rationale for targeting and current therapeutic approaches . Leukemia . 2020 ; 34 : 985 – 1005 . doi: 10.1038/s41375-020-0734-z . 32055000 PMC7214244
52. Lustig M , Chan C , Jansen JHM , Brautigam M , Kolling MA , Gehlert CL , Bräutigam M , Kölling MA , Baumann N , Mester S , et al. Disruption of the sialic acid/Siglec-9 axis improves antibody-mediated neutrophil cytotoxicity towards tumor cells . Front Immunol . 2023 ; 14 : 1178817 . doi: 10.3389/fimmu.2023.1178817 . 37346044 PMC10279866
53. Sun J , Muz B , Alhallak K , Markovic M , Gurley S , Wang Z , Guenthner N , Wasden K , Fiala M , King J , et al. Targeting CD47 as a novel immunotherapy for multiple myeloma . Cancers (Basel) . 2020 ; 12 : 305 . doi: 10.3390/cancers12020305 . 32012878 PMC7072283
54. Salama ME , Du S , Efimova O , Heikal NM , Wendlandt E , Toydemir RM , South S , Perkins SL , Hussong JW , Zhan F . Neoplastic plasma cell aberrant antigen expression patterns and their association with genetic abnormalities . Leuk Lymphoma . 2015 ; 56 : 426 – 433 . doi: 10.3109/10428194.2014.931951 . 24913506 PMC4408538