Comprehensive Single-Cell Immune Profiling Defines the Patient Multiple Myeloma Microenvironment Following Oncolytic Virus Therapy in a Phase Ib Trial

Abstract

Purpose: Our preclinical studies showed that the oncolytic reovirus formulation pelareorep (PELA) has significant immunomodulatory anti-myeloma activity. We conducted an investigator-initiated clinical trial to evaluate PELA in combination with dexamethasone (Dex) and bortezomib (BZ) and define the tumor immune microenvironment (TiME) in patients with multiple myeloma treated with this regimen.

Patients and methods: Patients with relapsed/refractory multiple myeloma (n = 14) were enrolled in a phase Ib clinical trial (ClinicalTrials.gov: NCT02514382) of three escalating PELA doses administered on Days 1, 2, 8, 9, 15, and 16. Patients received 40 mg Dex and 1.5 mg/m2 BZ on Days 1, 8, and 15. Cycles were repeated every 28 days. Pre- and posttreatment bone marrow specimens (IHC, n = 9; imaging mass cytometry, n = 6) and peripheral blood samples were collected for analysis (flow cytometry, n = 5; T-cell receptor clonality, n = 7; cytokine assay, n = 7).

Results: PELA/BZ/Dex was well-tolerated in all patients. Treatment-emergent toxicities were transient, and no dose-limiting toxicities occurred. Six (55%) of 11 response-evaluable patients showed decreased paraprotein. Treatment increased T and natural killer cell activation, inflammatory cytokine release, and programmed death-ligand 1 expression in bone marrow. Compared with nonresponders, responders had higher reovirus protein levels, increased cytotoxic T-cell infiltration posttreatment, cytotoxic T cells in significantly closer proximity to multiple myeloma cells, and larger populations of a novel immune-primed multiple myeloma phenotype (CD138+ IDO1+HLA-ABCHigh), indicating immunomodulation.

Conclusions: PELA/BZ/Dex is well-tolerated and associated with anti-multiple myeloma activity in a subset of responding patients, characterized by immune reprogramming and TiME changes, warranting further investigation of PELA as an immunomodulator.

Figure 1.

Study design and treatment effects of combination PELA/BZ/Dex therapy. A, Study schema showing enrollment and downstream analyses of blood/tissue samples. B, Waterfall plot showing the best percentage change in serum paraprotein among the 11 response-evaluable patients. C, Kaplan–Meier plot showing OS (red) and PFS (blue). D–G, Blood samples were collected from Patients 8–14 at baseline (pretreatment) and 24 hours after dosing on C1D1 (posttreatment). D, Cytokine levels were measured by enzyme-linked immunosorbent assay (ELISA) before (C1D1) and after (C1D2) treatment. Significant posttreatment increases were observed for C-X-C motif ligand 10 (CXCL10; C1D1: 187.000 ± 100.57 pg/mL; C1D2: 3640.00 ± 1127.90 pg/mL; P = 0.0023), CXCL11 (C1D1: 115.86 ±48.95 pg/mL; C1D2: 1636.86 ± 909.40 pg/mL; P = 0.0175), IL18 (C1D1: 227.71 ± 180.14 pg/mL; C1D2: 459.57 ± 316.19 pg/mL; P = 0.0196), C-C motif ligand 8 (CCL8; C1D1: 19.43 ± 3.64 pg/mL; C1D2: 422.71 ± 184.87 pg/mL; P = 0.0077), and tumor inhibitor of metalloproteases 1 (TIMP-1; C1D1: 76.14 ± 49.27 ng/mL; C1D2: 130.71 ± 89.11 ng/mL; P = 0.0482). *P < 0.05; ** P < 0.005. E, Flow cytometry was applied to assess the expression of CD69 on CD3⁻CD56⁺ NK cells and CD3+CD56⁻ T cells in peripheral blood monocytes (PBMC). Dot plots show representative data for CD69 expression in Patient 11. F and G, Quantification of CD69⁺ NK (CD69⁺CD3⁻CD56⁺cells/all CD3⁻CD56⁺ cells; Pre: 6.06% ± 2.08%; Post: 59.82% ± 20.42%; P = 0.0047) (F) and T cells (CD69⁺CD3⁺CD56⁻ cells/all CD3⁺CD56⁻ cells; Pre: 7.16% ± 1.30%; Post: 40.68% ± 6.26%; P = 0.0005) (G) in PBMCs. Each dot represents an individual patient. Data are presented as the mean and standard deviation. P values derived from Student t test, followed by Benjamini–Hochberg multiple comparison adjustment.

Figure 2.

Reovirus detection in multiple myeloma cells following PELA/BZ/Dex combination therapy and inducement of host immune response. A, Representative pre- and posttreatment bone marrow biopsy images stained for PELA RNA in pre- (left) and posttreatment (right) biopsy samples. Scatter plot (bottom) shows the quantification of percentage cells with positive staining [Pre-Nonresponder (NR): 0.00% ± 0.00%; Pre-Responder (R): 0.00% ± 0.00%; Post-NR: 53.93% ± 30.02%; Post-R: 49.55% ± 39.16%]. Black boxes indicate the area shown in insets. B–D, Representative posttreatment biopsy samples showing co-expression of PELA RNA with various protein markers. B, The Nuance system was used to convert reoviral RNA to blue (left) and CD138 to red (middle), with co-expression shown in yellow (right). C and D, The Nuance system was used to convert reoviral RNA to green (left). Junctional adhesion molecule A (JAM-A) (C) or PD-L1 (D) was converted to red (center), and co-expression images are shown in yellow (left). White boxes indicate the area shown in insets. Scale bar 60 μm for B and D, 100 μm for C. E, Representative image showing detection of reovirus protein in posttreatment biopsy samples. Scatter plot (bottom) shows the quantification of percentage cells with positive staining (Pre-NR: 0.00% ± 0.00%; Pre-R: 0.00% ± 0.00%; Post-NR: 1.10% ± 1.12%; Post-R: 11.85% ± 10.33%; Post-NR vs. Post-R, P = 0.042, unpaired t test). F–H, Representative images of pre- (left) and posttreatment (right) bone marrow biopsies stained for PD-L1 (Pre-NR: 0.44% ± 0.55%; Pre-R: 0.75% ± 1.50%; Post-NR: 7.34% ± 8.12%; Post-R: 5.80% ± 4.06%; All Pre vs. All Post, P = 0.0078, Wilcoxon signed-rank test) (F), CD8 (Pre-NR: 12.96% ± 3.53%; Pre-R: 9.05% ± 2.97%; Post-NR: 19.50% ± 7.53%; Post-R: 23.53% ± 6.95%; Pre-R vs. Post-R, P = 0.0112, paired t test) (G), and activated caspase-3 (Pre-NR: 0.06% ± 0.48%; Pre-R: 1.13% ± 0.79%; Post-NR: 2.08% ± 2.31%; Post-R: 3.88% ± 2.08%; Pre-R vs. Post-R, P = 0.024, paired t test) (H). Scale bar 500 μm. Numbers indicate the percentage of positive cells in each image. Black boxes indicate the area shown in insets. Scatter plots (bottom) show the quantification of percentage cells with positive staining. All scatter plots are presented as the mean and standard deviation. NR, n = 4; R, n = 4.

Figure 3.

Visualization of the custom IMC bone marrow panel. A, A custom IMC panel was designed consisting of antibodies against 31 markers to interrogate the multiple myeloma TiME in the bone marrow. B, Multiplexed image from Patient 9 posttreatment showing tumor–immune cell interactions: monocyte/macrophage marker (CD14⁺), NK cell receptor (NKG2D+) and its ligand ULBP-2/5/6, multiple myeloma marker (CD138⁺), T cell markers [CD4⁺, CD8a⁺, and forkhead box P3 (FOXP3)⁺], as well as the NK cytotoxicity receptor, NKp46⁺, and the T and NK-cell cytotoxicity marker, Granzyme. C, Patient 7 posttreatment sample showing the interaction between the inhibitory NK cell receptor NKG2A and its ligand, HLA-E. D, Patient 9 posttreatment sample showing HLA-ABC⁺ cells expressing cytotoxic T cells (CD8a⁺). E, Patient 9 posttreatment sample showing HLA-DR and CD4 co-expression. F, Patient 10 posttreatment sample showing dendritic cell cross-priming: CD4⁺ (T helper cells), CD8a⁺ (cytotoxic T cells), and reovirus⁺CD163⁺ cells (M2 macrophages). G, Image from Patient 10 posttreatment showing apoptotic multiple myeloma cells [CD138⁺ and cleaved caspase-3 (CC3⁺)]. Histone H3⁺ is used to highlight nuclei. H, Image from Patient 1 posttreatment showing the immune checkpoint interaction between PD-L1⁺ multiple myeloma cells (CD138⁺) and PD-1⁺ T cells (CD3⁺). I, Image from Patient 1 posttreatment showing the inhibitory receptor, lymphocyte-activation gene 3 (LAG3), on NK cells (CD16⁺). J, Patient 1 posttreatment sample showing IDO⁺ positive monocyte/macrophages (CD68⁺). K, Patient 9 posttreatment sample showing T cell immunoglobulin mucin-3 (TIM3), an immune checkpoint receptor, on T cells (CD3⁺). L, Patient 10 posttreatment sample showing Ki-67 (proliferation marker) in T cells (CD3⁺) and multiple myeloma cells (CD138⁺). M, Patient 1 posttreatment sample showing the plasma cell markers CD138⁺ and MUM1/IRF4⁺. N, Patient 1 posttreatment sample showing B cells (CD20⁺), megakaryocytes and endothelial cells (CD31⁺), leukocyte adhesion marker (CD11b⁺), monocyte/macrophages (CD68⁺), granulocytes (CD15⁺), and histone H3⁺.

Figure 4.

PhenoGraph, tSNE, and cluster visualization. A, Unsupervised clustering was performed using the PhenoGraph algorithm. Clusters were annotated on the basis of the interpretation of each cluster's unique expression profile. B, ImaCytE visualization software highlights each cluster in a unique color on the original segmentation mask. C, tSNE dimensionality reduction algorithm shows that each cluster is unique and separate. The central legend is shared between panels B and C. D, Comparison of the frequencies of each cluster observed for each patient between pre- and posttreatment samples. Patients are grouped into nonresponders (Patients 9, 12, and 14) and responders (Patients 1, 7, and 10). The proportion of cytotoxic T cells is significantly higher in the responder group compared with the nonresponder group for both pretreatment [median (interquartile range): R, 8.96% (1.48%); NR, 5.73% (1.26%); P = 0.02, unpaired t test] and posttreatment samples [R, 9.73% (4.27%); NR, 2.98% (2.86%); P = 0.02, unpaired t test].

Figure 5.

Nearest neighbor analysis. A, Representative MCD viewer images depicting the proportions of cytotoxic T cells (green) in the most commonly identified posttreatment neighborhoods in bone marrow samples from nonresponders (top, 11,940 cells) and responders (bottom, 25,900 cells). CD138⁺ multiple myeloma cells are shown in red. White boxes indicate the area shown in insets. Scale bar, 100 μm. B, ImaCytE spatial analysis highlighting the most abundant neighborhoods identified for both nonresponders (top) and responders (bottom). MM, multiple myeloma; CTC, cytotoxic T cell; MP, macrophage; Gran, granulocyte. C, Nearest neighbor violin plots comparing distances from various immune cell types to all myeloma cells in nonresponders (NR, green) and responders (R, red) in pre- (top, NR: 18,679 cells; R: 14,114 cells) and posttreatment (bottom, NR: 11,940 cells; R: 25,900) samples. Smaller distances indicate closer phenotypes. Black labels on each violin plot represent the median distance. Median (interquartile range) values are as follows. (pretreatment, top) Cytotoxic T cell: NR, 34.29 (33.70); R, 24.83 (23.08); P = 0.46. Macrophage (CD68⁺): NR, 32.87 (54.70); R, 29.72 (35.03); P = 0.974. NK (NKG2A⁺): NR, 42.93 (52.50); R, 32.54 (32.69); P = 0.483. T helper: NR, 52.00 (53.64); R, 55.48 (54.32); P = 0.94. (posttreatment, bottom) Cytotoxic T cell: NR, 64.57 (59.09); R, 21.33 (20.30); **P = 0.0093. Macrophage (CD68⁺): NR, 74.63 (75.62); R, 25.76 (33.83); P = 0.46. NK (NKG2A⁺): NR, 37.49 (74.93); R, 34.53 (46.66); P = 0.94. T helper: NR, 46.38 (56.71); R, 35.99 (45.51), P = 0.69. D, ImaCytE spatial analysis highlighting neighborhoods of low immune activity around a single non-IP myeloma cell (top, 5,514 cells analyzed) and increased immune activity around a single IP myeloma cell (bottom, 2,771 cells analyzed). CTC, cytotoxic T cell; MM (IP), IP multiple myeloma. MP, macrophage. E, Nearest neighbor distance violin plots comparing distances from various immune cell types to IP (green, 2,771 cells) and non-IP (red, 5,514 cells) myeloma cells in posttreatment samples. Smaller distances between MM cells and immune cells indicate closer proximity. Numeric labels on each violin plot represent the median distance. Median (interquartile range) values are as follows. Cytotoxic T cell: MM, 47.33 (57.86); MM (IP), 20.48 (19.91); *P = 0.028. Macrophage (CD68⁺): MM, 34.33 (62.62); MM (IP), 31.40 (41.07); P = 0.28. NK (NKG2A⁺): MM, 55.06 (62.66); MM (IP), 24.45 (31.35); P = 0.95. T helper: MM, 54.54 (57.62); MM (IP), 27.56 (30.10), P = 0.31. All comparisons were performed using Wilcoxon signed-rank tests (unpaired for comparisons between response type, paired for comparisons between treatment time points), followed by Benjamini–Hochberg multiple comparison adjustment.