Spatiotemporal assessment of immunogenomic heterogeneity in multiple myeloma

Abstract

Spatial heterogeneity is a common phenomenon in metastatic solid tumors and an evolving concept in multiple myeloma (MM). The interplay between malignant plasma cells (PCs) and the microenvironment has not yet been analyzed in MM. For this purpose, we performed bone marrow aspirates and imaging-guided biopsies of corresponding lesions in newly diagnosed MM (NDMM) and relapsed/refractory MM (RRMM) patients. PCs were isolated and subjected to whole-exome sequencing (WES). Non-PCs were studied with next-generation flow (NGF) and T-cell receptor sequencing (TCRseq) to analyze the connection between malignant and nonmalignant cells in the bone marrow and in lesions. Although we observed a strong overlap from WES, NGF, and TCRseq in patients with intramedullary disease, WES revealed significant spatial heterogeneity in patients with extramedullary disease. NGF showed significant immunosuppression in RRMM compared with NDMM as indicated by fewer myeloid dendritic cells, unswitched memory B cells, Th9 cells, and CD8 effector memory T cells but more natural killer and regulatory T cells. Additionally, fewer T-cell receptor (TCR) sequences were detected in RRMM compared with NDMM and healthy individuals. After induction therapy, TCR repertoire richness increased to levels of healthy individuals, and NGF showed more regulatory T cells and myeloid-derived suppressor cells, regardless of depth of response. Clinical significance of imaging-guided biopsies of lesions was demonstrated by detection of monoclonal PCs in patients without measurable residual disease (MRD) in aspirates from the iliac crest as well as identification of secondary primary malignancies in MRD- patients. Furthermore, site-specific clones with different drug susceptibilities and genetically defined high-risk features were detected by our workflow.

Graphical abstract

Figure 1.

WES of the PC compartment from paired samples. (A) Number of total mutations as well as transitions and transversions from bone marrow and lesion (OL) for each individual patient. No significant differences between lesion and marrow were found for number of mutations and transitions/transversions in both NDMM and RRMM. (B) Significantly more mutations were found in patients with RRMM compared with NDMM in both locations. (C) Scatterplots of variant allele frequencies in the bone marrow (x-axis) and lesion (y-axis) in NDMM (blue) and RRMM (red). Jaccard indices were calculated to quantify overlap between paired samples. The largest numbers of unshared mutations were found in patients with EMD (indicated by black stars). BM, bone marrow.

Figure 2.

CNVs from WES. (A) Copy number plot using cnSpec from GenVisR showing gains (red) and losses (blue) for every chromosome (columns) and location per patient. Although high concordance between both locations was observed in most patients, significant differences (depicted by red boxes) were found in patients with EMD (in red). (B) Example of a newly diagnosed patient (NDMM06) with deletions of chromosomes 13, 14, 16, and 22 that were detected in the bone marrow (lower panel) and the biopsied vertebral lesion (upper panel). (C) Example of a patient with a hyperdiploid karyotype and a history of solitary plasmacytoma who showed progressive disease with a new lesion in the left humerus that extended into the extramedullary space. Although high concordance was observed between gains of chromosomes 5, 9, 15, and 19 as well as deletion of chromosome 13, differences were found for chromosome 1. Although a deletion was detected in the bone marrow (lower panel), oscillating copy numbers suggestive of chromotrypsis of chromosome 1 were assessed in the lesion (upper panel). BM, bone marrow; OL, lesion.

Figure 2.

CNVs from WES. (A) Copy number plot using cnSpec from GenVisR showing gains (red) and losses (blue) for every chromosome (columns) and location per patient. Although high concordance between both locations was observed in most patients, significant differences (depicted by red boxes) were found in patients with EMD (in red). (B) Example of a newly diagnosed patient (NDMM06) with deletions of chromosomes 13, 14, 16, and 22 that were detected in the bone marrow (lower panel) and the biopsied vertebral lesion (upper panel). (C) Example of a patient with a hyperdiploid karyotype and a history of solitary plasmacytoma who showed progressive disease with a new lesion in the left humerus that extended into the extramedullary space. Although high concordance was observed between gains of chromosomes 5, 9, 15, and 19 as well as deletion of chromosome 13, differences were found for chromosome 1. Although a deletion was detected in the bone marrow (lower panel), oscillating copy numbers suggestive of chromotrypsis of chromosome 1 were assessed in the lesion (upper panel). BM, bone marrow; OL, lesion.

Figure 3.

NGF and TCRseq from paired samples in newly diagnosed and relapsed patients. (A) Scatterplots of productive TCR frequencies in bone marrow (x-axis) and lesions (y-axis) for patients with NDMM (blue) and RRMM (red). Repertoire overlaps were compared by calculating Morisita indices (MOI). In NDMM and RRMM, TCR repertoires showed significant overlap in accordance with findings from WES. (B) Comparison of the actual number of detected TCR sequences (richness estimated by chao1 indices) and clonality (Simpson clonality) showed no significant differences between both locations in NDMM and RRMM. (C) There were also no differences between both locations regarding relative abundance because distributions between lesion and marrow of small (blue), medium (red), large (green), and hyperexpanded (purple) TCR clones were comparable in NDMM and RRMM. (D) Although NGF showed no significant differences between lesion and marrow, patients with RRMM harbored fewer myeloid dendritic cells, unswitched memory B cells, Th9 cells, and CD8 effector memory T cells and more NK cells and regulatory T cells compared with NDMM. This underlines the immunosuppressive capabilities of RRMM.

Figure 3.

NGF and TCRseq from paired samples in newly diagnosed and relapsed patients. (A) Scatterplots of productive TCR frequencies in bone marrow (x-axis) and lesions (y-axis) for patients with NDMM (blue) and RRMM (red). Repertoire overlaps were compared by calculating Morisita indices (MOI). In NDMM and RRMM, TCR repertoires showed significant overlap in accordance with findings from WES. (B) Comparison of the actual number of detected TCR sequences (richness estimated by chao1 indices) and clonality (Simpson clonality) showed no significant differences between both locations in NDMM and RRMM. (C) There were also no differences between both locations regarding relative abundance because distributions between lesion and marrow of small (blue), medium (red), large (green), and hyperexpanded (purple) TCR clones were comparable in NDMM and RRMM. (D) Although NGF showed no significant differences between lesion and marrow, patients with RRMM harbored fewer myeloid dendritic cells, unswitched memory B cells, Th9 cells, and CD8 effector memory T cells and more NK cells and regulatory T cells compared with NDMM. This underlines the immunosuppressive capabilities of RRMM.

Figure 4.

Longitudinal analyses after induction therapy. (A) In 5 patients, bone marrow samples were collected after induction therapy. All patients received combination treatment with an IMiD, PI, and dexamethasone. Patient NDMM03 was changed from bortezomib, lenalidomide, dexamethasone (VRD) to daratumumab, pomalidomide, dexamethasone (Dara-PomDex) due to suboptimal response prior to intended stem cell collection. NDMM05 received daratumumab–lenalidomide, bortezomib, dexamethasone (RVD) because of cytogenetically defined high-risk disease (deletion 17p). (B) Scatterplots showing TCR frequencies and repertoire overlap between bone marrow samples at primary diagnosis (x-axis) and after therapy (y-axis). Significant differences of expanded clones were detected at primary diagnosis (blue) and after therapy (red). The patient with the least reduction of M-protein (NDMM05) showed least expanded clones. (C) Longitudinal tracking of top 10 TCR clones at primary diagnosis showed a reduction of clone size in 4 out of 5 patients. (D) No significant changes in relative abundance were observed. (E) Although no differences in TCR repertoire clonality was observed when comparing healthy individuals with NDMM, RRMM, and patients in remission after therapy, repertoire richness as estimated by chao1 indices was significantly lower in RRMM compared with NDMM and healthy individuals. Upon induction therapy, richness reached levels of healthy individuals. (F) Longitudinal immunophenotyping showed more CD3⁺ cells after induction therapy. Further analysis revealed significantly more CD3⁺HLADr⁺ cells and Tregs. Additionally, more myeloid derived suppressor cells (MDSCs) were detected, regardless of depth of response. Because Tregs and MDSCs are associated with immunosuppression in myeloma and were expanded after therapy, it raises the question of whether early changes upon induction therapy are caused by direct antimyeloma effects and not immunomodulation. KRD, carfilzomib/lenalidomide/dexamethasone.

Figure 4.

Longitudinal analyses after induction therapy. (A) In 5 patients, bone marrow samples were collected after induction therapy. All patients received combination treatment with an IMiD, PI, and dexamethasone. Patient NDMM03 was changed from bortezomib, lenalidomide, dexamethasone (VRD) to daratumumab, pomalidomide, dexamethasone (Dara-PomDex) due to suboptimal response prior to intended stem cell collection. NDMM05 received daratumumab–lenalidomide, bortezomib, dexamethasone (RVD) because of cytogenetically defined high-risk disease (deletion 17p). (B) Scatterplots showing TCR frequencies and repertoire overlap between bone marrow samples at primary diagnosis (x-axis) and after therapy (y-axis). Significant differences of expanded clones were detected at primary diagnosis (blue) and after therapy (red). The patient with the least reduction of M-protein (NDMM05) showed least expanded clones. (C) Longitudinal tracking of top 10 TCR clones at primary diagnosis showed a reduction of clone size in 4 out of 5 patients. (D) No significant changes in relative abundance were observed. (E) Although no differences in TCR repertoire clonality was observed when comparing healthy individuals with NDMM, RRMM, and patients in remission after therapy, repertoire richness as estimated by chao1 indices was significantly lower in RRMM compared with NDMM and healthy individuals. Upon induction therapy, richness reached levels of healthy individuals. (F) Longitudinal immunophenotyping showed more CD3⁺ cells after induction therapy. Further analysis revealed significantly more CD3⁺HLADr⁺ cells and Tregs. Additionally, more myeloid derived suppressor cells (MDSCs) were detected, regardless of depth of response. Because Tregs and MDSCs are associated with immunosuppression in myeloma and were expanded after therapy, it raises the question of whether early changes upon induction therapy are caused by direct antimyeloma effects and not immunomodulation. KRD, carfilzomib/lenalidomide/dexamethasone.

Figure 5.

Clinical significance of imaging-guided biopsies. (A) Early, site-specific detection of relapse in a patient without MRD (MRD⁻) complete response (CR) after 5 cycles of RVD (lenalidomide, bortezomib, dexamethasone), high-dose chemotherapy, and autologous stem cell transplantation (ASCT) followed by 5 years of lenalidomide maintenance. Routine PET/CT showed a small PET⁺ lesion in the caudal right pelvis (A1). Although flow cytometry showed no signs for MRD in the iliac crest (A3), MRD positivity was confirmed by imaging-guided biopsy of the lesion (A2). Because the patient had no other signs for disease activity, active surveillance was continued. Repeated PET/CT 5 months later showed progression of the biopsied lesion. At that time point, serological relapse was also confirmed, and systemic therapy was initiated. (B) The necessity to confirm findings from PET/CT with biopsies is demonstrated by a case of secondary B-cell acute lymphoblastic leukemia (B-ALL) in a patient in MRD⁻ CR after 4 cycles of lenalidomide and dexamethasone, ASCT, and 3 years of lenalidomide maintenance. Although imaging-guided biopsy of a newly emerged lesion showed no signs for monoclonal PCs, B lymphoblasts were detected. The patient was treated with induction therapy and allogeneic transplantation, and both conditions remain in CR at the moment. (C) To investigate whether mutations from WES are accessible for therapeutic intervention, the Drug Gene Interaction database was queried. More interactions were found in PCs from lesion (OL), which emphasizes that personalized treatment approaches need to be monitored with whole-body imaging. (D) Also, prognostic factors were detected by imaging-guided biopsies (eg, TP53 mutation in a vertebral body of patient RRMM03). Lollipot plot shows all TP53 mutations in the analyzed cohort. Also, mutations associated with drug resistance (eg, PSMC3 associated with resistance to proteasome inhibitors in RRMM05) were detected by imaging-guided biopsies. BM, bone marrow.

Figure 5.

Clinical significance of imaging-guided biopsies. (A) Early, site-specific detection of relapse in a patient without MRD (MRD⁻) complete response (CR) after 5 cycles of RVD (lenalidomide, bortezomib, dexamethasone), high-dose chemotherapy, and autologous stem cell transplantation (ASCT) followed by 5 years of lenalidomide maintenance. Routine PET/CT showed a small PET⁺ lesion in the caudal right pelvis (A1). Although flow cytometry showed no signs for MRD in the iliac crest (A3), MRD positivity was confirmed by imaging-guided biopsy of the lesion (A2). Because the patient had no other signs for disease activity, active surveillance was continued. Repeated PET/CT 5 months later showed progression of the biopsied lesion. At that time point, serological relapse was also confirmed, and systemic therapy was initiated. (B) The necessity to confirm findings from PET/CT with biopsies is demonstrated by a case of secondary B-cell acute lymphoblastic leukemia (B-ALL) in a patient in MRD⁻ CR after 4 cycles of lenalidomide and dexamethasone, ASCT, and 3 years of lenalidomide maintenance. Although imaging-guided biopsy of a newly emerged lesion showed no signs for monoclonal PCs, B lymphoblasts were detected. The patient was treated with induction therapy and allogeneic transplantation, and both conditions remain in CR at the moment. (C) To investigate whether mutations from WES are accessible for therapeutic intervention, the Drug Gene Interaction database was queried. More interactions were found in PCs from lesion (OL), which emphasizes that personalized treatment approaches need to be monitored with whole-body imaging. (D) Also, prognostic factors were detected by imaging-guided biopsies (eg, TP53 mutation in a vertebral body of patient RRMM03). Lollipot plot shows all TP53 mutations in the analyzed cohort. Also, mutations associated with drug resistance (eg, PSMC3 associated with resistance to proteasome inhibitors in RRMM05) were detected by imaging-guided biopsies. BM, bone marrow.