Relapsed multiple myeloma demonstrates distinct patterns of immune microenvironment and malignant ce

Immunotherapy has shown efficacy in relapsed multiple myeloma (MM). However, these therapies may depend on a functional tumor immune microenvironment (iTME) for their efficacy. Characterizing the evolution of the iTME over the disease course is necessary to optimize the timing of immunotherapies. We performed mass cytometry, cytokine analysis, and RNA sequencing on bone marrow samples from 39 (13 newly diagnosed [NDMM], 11 relapsed pre-daratumumab exposure [RMM], and 13 triple-refractory [TRMM]) MM patients. Three distinct cellular iTME clusters were identified; cluster 1 comprised mainly of NDMM and RMM patients; and clusters 2 and 3 comprised primarily of TRMM patients. We showed that naive T cells were decreased in clusters 2 and 3, cluster 2 was characterized by increased senescent T cells, and cluster 3 by decreased early memory T cells. Plasma cells in clusters 2 and 3 upregulated E2F transcription factors and MYC proliferation pathways, and downregulated interferon, TGF-beta, interleuking-6, and TNF-αlpha signaling pathways compared to cluster 1. This study suggests that the MM iTME becomes increasingly dysfunctional with therapy whereas the MM clone may be less dependent on inflammation-mediated growth pathways and less sensitive to IFN-mediated immunosurveillance. Our findings may explain the decreased sensitivity of TRMM patients to novel immunotherapies.

Despite the advent of novel therapeutic options for treatment of multiple myeloma (MM), this disease remains incurable. While the prognosis of patients continues to improve over time1, numerous studies have shown that outcomes after relapse are poor2,3,4. Patients who are triple refractory (those progressing after treatment with a proteasome inhibitor (PI), an immunomodulatory drug (IMID), and an anti-CD38 monoclonal antibody) have particularly poor outcomes5. Tumor intrinsic and extrinsic factors drive disease refractoriness. MM is a disease comprising of multiple subclones that are genetically heterogeneous. Therapy-induced clonal selection and clonal evolution6,7,8 play a role in disease progression. In addition to changes in the plasma cell clone, alterations of the normal bone marrow (BM) immune microenvironment (iTME) can lead to tumor escape from immunosurveillance9,10,11,12. The importance of leveraging the immune system to control MM progression is highlighted by the recent use of immune-based therapies in heavily pretreated MM patients. These immune-based therapies rely on a functional iTME for their efficacy. However, the current paradigm is that immune effector cells used for cellular therapies (i.e., chimeric antigen receptor T cells [CAR-T]) are harvested after patients have failed numerous lines of immune-modulating therapies which drastically reshape the cellular composition of their immune system and may select for more aggressive clones. Therefore, while immunotherapeutic approaches have shown promising results13,14,15,16, it is not surprising that they are less effective in patients with more heavily treated triple-refractory disease14,17,18.

Understanding the transcriptomic (tumor) and “immunomic” landscape of relapsed myeloma is necessary in order to understand the drivers of disease progression and refractoriness to existing therapies. Characterizing the iTME may also inform decisions regarding the optimal timing to harvest immune effector cells for cellular therapies, and the ideal sequencing of immune-based therapies. Therefore, the aim of this study was to use a multiomics approach to characterize the malignant plasma cells transcriptome as well as the humoral iTME (cytokines), and cellular iTME in patients with newly diagnosed, relapsed, and triple-refractory MM.

The Mayo Clinic biospecimen database was searched for newly diagnosed multiple myeloma (NDMM) patients, patients refractory to a proteasome inhibitor (PI) and/or an immunomodulatory drug (IMID) but not refractory to an anti-CD38 monoclonal antibody (RMM cohort), and patients refractory to a PI, IMID, and an anti-CD38 monoclonal antibody (daratumumab, no patients treated with isatuximab were identified) (TRMM cohort). All patients except one had the following bone marrow (BM) sample types: CD138+ sorted plasma cells, CD138− sorted or unsorted mononuclear cells, and BM plasma. One patient did not have CD138− cells available and was excluded from mass cytometry analysis. At our institution, patients undergo a routine BM biopsy approximately 2–3 months post autologous stem cell transplantation in order to assess disease response. However, post-transplantation BM samples were excluded given the expected significant perturbations in the iTME at this timepoint. The electronic medical record was reviewed to obtain clinical characteristics and treatment information for included patients. All included patients had consented to have their BM samples and clinical data used for research purposes, and this study was approved by the Mayo Clinic Institutional Review Board.

Luminex analyses

Cytokine and chemokine protein levels in BM plasma were measured using Luminex xMAP technology. The multiplexing analysis was performed using the 65-plex immune monitoring kit (ProcartaPlex™, Thermofischer) on the Luminex 100 system. Raw data were analyzed per the manufacturer’s protocol. The following proteins were analyzed: cytokines [G-CSF, GM-CSF, IFN alpha, IFN gamma, IL-1 alpha, IL-1 beta, IL-2, IL-3, IL-4, IL-5, IL-6, IL-7, IL-8 (CXCL8), IL-9, IL-10, IL-12p70, IL-13, IL-15, IL-16, IL-17A (CTLA-8), IL-18, IL-20, IL-21, IL-22, IL-23, IL-27, IL-31, LIF, M-CSF, MIF, TNF alpha, TNF beta, TSLP]; chemokines [BLC (CXCL13), ENA-78 (CXCL5), Eotaxin (CCL11), Eotaxin-2 (CCL24), Eotaxin-3 (CCL26), Fractalkine (CX3CL1), Gro-alpha (CXCL1), IP-10 (CXCL10), I-TAC (CXCL11), MCP-1 (CCL2), MCP-2 (CCL8), MCP-3 (CCL7), MDC (CCL22), MIG (CXCL9), MIP-1 alpha (CCL3), MIP-1 beta (CCL4), MIP-3 alpha (CCL20), SDF-1 alpha (CXCL12)]; growth factors/regulators [FGF-2, HGF, MMP-1, NGF beta, SCF, VEGF-A; soluble receptors: APRIL, BAFF, CD30, CD40L (CD154), IL-2R (CD25), TNF-RII, TRAIL (CD253), TWEAK].

Mass cytometry

Our antibody panel included 36 antibodies directed against well-characterized surface markers of lymphoid cells (T cells, B cells, NK− cells). Mass cytometry antibody–metal conjugate combinations are detailed in Supplementary Table 1. Sample preparation, antibody staining, and CyTOF acquisition are detailed in the Supplemental methods. Major lymphoid cell phenotypes are defined in Supplementary Table 2.

Mass cytometry data analysis

Flow cytometry standard (FCS) files were normalized and concatenated using the Fluidigm acquisition software. FCS files were uploaded to the Astrolabe Cytometry Platform (Astrolabe Diagnostics, Inc.) where transformation, cleaning (doublets, debris), labeling, and unsupervised clustering (based on FlowSOM19) were done as previously described20. Data were transformed using arcsinh with a cofactor of 5 and the marker intensities presented in the paper are all after transformation. A minimum of 40395 CD45+ events per file were used for clustering. Since the astrolabe platform is designed to overcluster data, astrolabe-identified cell clusters were sometimes merged so that the minimum median cluster size was not smaller than the median doublet rate. As an additional method of robust cluster identification, we only considered clusters that formed discrete islands after viSNE visualization, a method that is sensitive to outliers and batch effects21. Identified clusters were then exported for downstream statistical analyses.

RNA isolation and sequencing

RNA was extracted from CD138+ plasma cells using Qiagen’s RNeasy mini kit. Total RNA concentration and quality were determined using Qubit fluorometry (Invitrogen) and the Agilent Fragment Analyzer. Using the Illumina TruSeq® RNA Exome Library Prep kiT, libraries were prepared according to the manufacturer’s instructions. The concentration and purity of cDNA libraries were checked using the Agilent TapeStation D1000. Coding regions of the transcriptome were captured by pooling four of the cDNA libraries at 200 ng each. The concentration and size distribution of the completed libraries were determined using an Agilent BioAnalyzer DNA 1000 chip and Qubit fluorometry (Invitrogen). Libraries were sequenced at up to eight samples per lane following Illumina’s standard protocol using the Illumina cBot and HiSeq 3000/4000 PE Cluster Kit. The flow cells were sequenced as 100 × 2 paired-end reads on an Illumina HiSeq 4000 using the HiSeq Control Software HD 3.4.0.38 collection software. Base-calling was performed using Illumina’s RTA version 2.7.7.

Statistical methods

Hierarchical clustering of patients according to the abundance (% of CD45+ cells) of immune subsets was performed using the Ward method. The Kruskal–Wallis statistical test was used to describe differences between groups. Kaplan–Meier survival analysis was used to estimate the overall survival (defined as time between sample collection and death or last follow-up) and progression-free survival (defined as time between sample collection and death/progression or last follow-up). Pearson’s correlation was used to test correlations between continuous variables. A two-sided false discovery rate (FDR) adjusted p value of < 0.05 was considered significant. Statistical analysis was performed using JMP Pro statistical software version 14.1 (SAS Institute, Cary, NC). Differential gene expression across groups of interest was performed using the edgeR software. Gene set enrichment analysis (GSEA)22 using the hallmarks of cancer database, was performed to identify differentially expressed (FDR < 0.05) gene pathways between groups.

Patient characteristics

A total of 39 (13 NDMM, 11 RMM, and 15 TRMM) patients were included in this study. Their clinical and laboratory characteristics are shown in Supplementary Table 3. As expected, triple-refractory (TRMM cohort) patients had worse OS (median 12.4 [95% CI 6.1–15.5] months) and PFS (6.1 [95% CI 5.4–19.1] months) from sample collection than the RMM (median OS 124.1 [95% CI 25.9–131] months) and NDMM (median 110.3 [95% CI 19.3–132.4] months) cohorts. However, the NDMM cohort in this study tended to have an unusually aggressive disease course (Supplementary Fig. 1); the PFS from diagnosis was ≤12 months in 8/13 (64%) of NDMM patients, despite the fact that all patients received novel agent induction regimens, and 10/13 (77%) received triplet or quadruplet induction regimens.

Triple-refractory patients have a distinct cellular iTME

Given the small size and heterogeneity of our cohort, and to allow a less biased (by prior knowledge of patients’ relapse status) analytic approach, we agnostically grouped patients based on their cellular iTME composition and use that as our “anchoring variable” for several reasons. First, the efficacy of novel immunotherapies, which largely target plasma cells irrespective of their clonal biology, depends on a functional cellular iTME. Second, the humoral iTME (cytokines) is much less stable over time compared to the individual cellular composition, and so grouping based on cytokine expression may be less reliable. Furthermore, sample cytokine profiles are likely reflective of the cellular microenvironment, which was captured by our mass cytometry and transcriptomic data. Finally, our cohort was small and heterogeneous, as was evident by the unusually aggressive disease course of NDMM patients. For this reason, we did not use transcriptomic-based grouping or grouping based on a priori knowledge of their relapse status (NDMM, RMM, TRMM).

Cellular iTME-based hierarchical clustering recapitulated disease biology based on prior treatment exposure very accurately (Figs. 1 and 2) and identified three distinct groups: Cluster 1 comprised mainly of NDMM and RMM patients, whereas clusters 2 and 3 were comprised primarily of TRMM patients. The clinical characteristics of patients based on immune clusters are summarized in Table 1. Patients in cluster 2 were mostly (4/5 patients) triple refractory but less heavily pretreated compared to cluster 3 (2 median lines of therapy for cluster 2 versus 8 lines for cluster 3, p < 0.001). The median overall survival (mOS) from sample collection was significantly better for patients in cluster 1 (Supplementary Fig. 2). However, 4 of the 5 patients in cluster 2 had their samples collected while relapsing on daratumumab therapy. In order to ensure that recent daratumumab exposure was not the main factor altering the immune microenvironment and driving the cellular iTME clustering, we repeated the cellular iTME hierarchical clustering after excluding all seven patients that had samples collected while progressing on daratumumab-based therapy. Only two patients (both with NDMM) were clustered differently. Overall, the composition of patients within the cellular iTME clusters was similar with or without the inclusion of patients progressing on daratumumab (cohen’s kappa coefficient = 0.863, with p < 0.001), suggesting that the iTME clustering was not driven primarily by recent daratumumab exposure. These data suggest that triple-refractory patients have a distinct cellular iTME compared to non-triple-refractory patients.

Triple-refractory relapses are associated with shrinking of the CD4 T-cell pool and increased terminally differentiated T cells

We then considered lymphoid cell subset differences across the three clusters. We noted no differences in B cell subset frequencies between the three immune clusters. There was also no significant difference in the overall NK-cell infiltration between the clusters. However, one CD39+/NKG2D− NK-cell subset (NK cell-5), which is thought to have immunosuppressive properties23,24 and unable to recognize MM cells25, was increased in cluster 3.

When considering T-cell differences, the overall T-cell infiltration was significantly lower in cluster 3 patients compared to clusters 1 and 2 (23% vs 40% vs 51%, respectively, p < 0.001). The CD4 to CD8 ratio was significantly higher in cluster 1 compared to clusters 2 and 3 (2.0 vs 0.61 vs 1.15, respectively, p < 0.001). CD4+ T cells were highest in cluster 1 compared to 2 and 3 (median 26.97% of CD45+ cells versus 16.3% and 7.9%, respectively, p < 0.0001). However, CD8+ T cells were highest in cluster 2 (median 18.4%, 36.3%, and 9.5% of CD45+ cells for clusters 1, 2, and 3, respectively, p < 0.0001).

The differential abundance of specific immune subsets is shown in Fig. 3. Cluster 1 was characterized by an increased abundance of the T1, T2, and T12 populations which had phenotypes consistent with naïve CD4 and CD8 T cells, as shown in Fig. 4. The CD8/CD161+ T17 subset was also highest in cluster 1. Other CD4/CD161+ T-cell populations (T6, T7) were lowest in cluster 3 but preserved in cluster 2. CD161+ T cells are thought to represent T-cell subsets with similar transcriptional and phenotypic features, high proliferative capacity, and the ability to secrete high levels of IFN gamma and TNF alpha26,27. Cluster 2 was characterized by a higher abundance of several immunosenescent, terminally differentiated (CD27/CD28−, CD57/KLRG1+/−) populations (T10, T16, T18, T19, T20, TCRgd-3) as well as the T14 population characterized by tissue homing (CCR5+) and early differentiation markers (CD27/28+) but also exhaustion markers (PD-1, TIGIT). Cluster 3 was characterized by a decrease of the following subsets: the T8 and T15 populations, with a central memory (CCR7/CD127+) phenotype; the CCR5+ T9 effector memory population, a subset necessary for maximal T-cell-mediated antitumor responses28; the CD103+ T13 population, thought to consist of tumor-reactive, tissue-resident memory T cells29,30; and the TCRgd-1 subset expressing markers of early differentiation (CD27, CD28), tissue homing (CCR5), and tumor cytotoxicity (CD226)31.

These data suggest that triple refractoriness is associated with shrinking of the T-cell pool overall, especially CD4+ and naïve subsets. Consequently, within some triple-refractory patients, early memory T-cell subsets decrease and in others senescent, terminally differentiated T-cell subsets accumulate.

Triple-refractory relapses are associated with proliferative plasma cell clones that downregulate immune effector molecule signaling pathways

Since patients in clusters 2 and 3 above, enriched in TRMM cases, had a distinct iTME compared to cluster 1 (comprised primarily of NDMM and RMM patients), we performed GSEA of the malignant plasma cells and compared cluster 1 to the combined cluster 2 and 3 groups. In patients from clusters 2/3, E2F targets, G2M checkpoint genes, and Myc targets were increased (Fig. 5), suggesting that malignant cells from clusters 2/3 upregulated transcriptional networks related to DNA replication, cell division, and MYC signaling. This group also showed downregulation of various inflammatory pathways, most notably tumor necrosis factor alpha (TNFa) and interferon (IFN) signaling. Of note, we found no cytokines differentially expressed between the three immune clusters identified above or when considering clusters 2 and 3 together. For this reason, we do not present the cytokine data that can be made available upon request. These data suggest that more heavily pretreated patients have highly proliferative malignant clones that downregulate pathways required for tumor immunosurveillance.