Aberrant single-cell phenotype and clinical implications of genotypically defined polyclonal plasma cells in myeloma

Abstract

Multiple myeloma (MM) is driven by clonal plasma cell (cPC)-intrinsic factors and changes in the tumor microenvironment (TME). To investigate whether residual polyclonal PCs (pPCs) are disrupted, single-cell (sc) RNA sequencing (scRNA-seq) and sc B-cell receptor analysis were applied in a cohort of 46 samples with PC dyscrasias and 21 healthy donors (HDs). Of 234 789 PCs, 64 432 were genotypically identified as pPCs with frequencies decreasing over different disease stages, from 23.66% in monoclonal gammopathy of undetermined significance to 3.23% in MMs (P = .00012). Both cPCs and pPCs had a comparable expression of typical lineage markers (ie, CD38 and CD138), whereas others were more variable (CD27 and ITGB7). Only cPCs overexpressed oncogenes (eg, CCND1/2 and NSD2), but CCND3 was often expressed in pPCs. B-cell maturation antigen was expressed on both pPCs and cPCs, whereas GPRC5D was mostly upregulated in cPCs with implications for on-target, off-tumor activity of targeted immunotherapies. In comparison with HDs, pPCs from patients showed upregulated autophagy and disrupted interaction with TME. Importantly, interferon-related pathways were significantly enriched in pPCs from patients vs HDs (adjusted P < .05) showing an inflamed phenotype affecting genotypically normal PCs. The function of pPCs was consequently affected and correlated with immunoparesis, driven by disrupted cellular interactions with TME. Leveraging our scRNA-seq data, we derived a "healthy PC signature" that could be applied to bulk transcriptomics from the CoMMpass data set and predicted significantly better progression-free survival and overall survival (log-rank P < .05 for both). Our findings show that genotypic sc identification of pPCs in PC dyscrasias has relevant pathogenic and clinical implications.

Graphical abstract

Figure 1.

Clonal and polyclonal cells identification in patients affected by PC dyscrasias and HD. (A) Schematic overview of cohort of samples and workflow of the study. (B) Donut charts representing the distribution of polyclonal (blue) and clonal cells (red) in PC dyscrasias and HDs. Numerical values are reported. (C) The proportions of clonal cells (red) and polyclonal cells (blue) in each sample, categorized by clinical state (top). (D) UMAP plot showing all clonal (red) and polyclonal (blue) cells, with numerical values. (E) UMAP representation of the sample space categorized by disease stage. Colors as in panel C. Numerical values and percentages are reported for each category. (F) Violin plots showing distribution of expression of genes commonly upregulated in patients with PC dyscrasias, along with annotations of the clinical classification (top). BM, bone marrow; df, degrees of freedom.

Figure 1.

Clonal and polyclonal cells identification in patients affected by PC dyscrasias and HD. (A) Schematic overview of cohort of samples and workflow of the study. (B) Donut charts representing the distribution of polyclonal (blue) and clonal cells (red) in PC dyscrasias and HDs. Numerical values are reported. (C) The proportions of clonal cells (red) and polyclonal cells (blue) in each sample, categorized by clinical state (top). (D) UMAP plot showing all clonal (red) and polyclonal (blue) cells, with numerical values. (E) UMAP representation of the sample space categorized by disease stage. Colors as in panel C. Numerical values and percentages are reported for each category. (F) Violin plots showing distribution of expression of genes commonly upregulated in patients with PC dyscrasias, along with annotations of the clinical classification (top). BM, bone marrow; df, degrees of freedom.

Figure 2.

Polyclonal cell selection and marker genes expression by clinical stages. (A) UMAP of scRNA-seq data of pPCs with colors indicating clinical clusters. Numerical values are reported for each category. (B) Dot plot displaying the top 20 marker genes that distinguish each clinical state. The x-axis lists the clinical category, whereas the y-axis lists gene names. Circle size corresponds to the number of cells in the category expressing the gene of interest, whereas the color shade correlates with the level of expression. (C) Violin plot showing expression pattern of selected PCs, adhesion/interaction, and autophagy marker genes across all the clinical stages.

Figure 3.

Transcriptional characterization of pPCs. (A) Cartoon depicting the deregulated expression profiles of pPCs, derived from 1:1 DE analyses by Wilcoxon test. Circle size corresponds to the number of cells in the category expressing the gene of interest, whereas shade correlates with the level of expression. Color code as in Figure 1C. (B-D) Gene set enrichment analysis (GSEA) plots depicting the enrichment of signal pathways based on the hallmark gene set. NES, normalized enrichment score.

Figure 4.

Transcriptomic landscape of immunoparesis. (A) Cartoon showing the pathological subset of patients (MGUS, SMM, and MM), for downstream DE analyses. Color code as shown in Figure 1C. (B) Dot plot of top 10 marker genes that distinguish patients with and without immunoparesis. The x-axis lists the clinical category, whereas the y-axis lists gene names. Circle size corresponds to the number of cells in the category expressing the gene of interest, whereas shade correlates with the level of expression. (C) DE analysis comparing the patients with or without immunoparesis within a symptomatic subset of patients, using Wilcoxon rank sum test, with adjusted P value Benjamini-Hochberg correction. No immunoparesis, blue; immunoparesis, yellow. (D) GSEA on the genes ranked by their contribution to hallmark oxidative phosphorylation, complement and IFN-γ response. NES, normalized enrichment score.

Figure 5.

Predicted interactions of PCs with microenvironment. (A) UMAP representation of PCs integrated with TME, colors indicate TME cell types. (B) UMAP plot of PCs integrated with TME colored according to immunoparesis. Color code as shown in Figure 4C. (C-D) Circos plots illustrate ligand-receptor interactions between pPCs (blue), cPCs (red), and TME, according to presence or absence of immunoparesis. Ribbon arrows indicate directionality of communication, from sender to receiver populations, whereas the arrow’s color signifies the specific sender cell type expressing the ligand. PCs were set as senders (C) and as receivers (D). DC, dendritic cells; gdT, γδ T cells; ProgB, B-cell progenitors; ProgMK, megakaryocyte progenitors; Treg, T-regulatory cells.

Figure 6.

An “hPC” signature analysis allows prediction in independent data sets. (A) HD gene signature defined by a DE analysis between HDs and MGUS, SMM, and MM. For each contrast genes significantly upregulated in HDs (Benjamini-Hochberg adjusted P < .05; log fold change >0.1) were retained. Venn diagram to show the intersection of deregulated genes within each comparison. Color code as shown Figure 1C. (B) Violin plot showing the inference of HD in the cohort of patients. (C-D) Clinical impact of HD signature in CoMMpass RNA-seq data. PFS (C) and OS (D) Kaplan-Meier curves in the cohort of patients. OS, overall survival; PFS, progression-free survival. Time is measured in years.