Signatures of Mesenchymal Cell Lineages and Microenvironment Factors Are Dysregulated in High Risk Myeloma

Introduction

Mesenchymal lineages including osteoblasts, adipocytes and pericytes derived from mesenchymal stem cells (MSCs) are key components of hematopoietic stem cell and myeloma (MM) niches in bone marrow (BM).

The aim of the study was to define high risk signatures of mesenchymal cells in BM of MM patients at diagnosis and post-treatment, and their biological importance.

Methods

Uniquely we have generated and extracted RNA from random BM biopsy (BM-BX) and focal lesion (FL-BX) samples that contain all cellular populations and factors of the marrow, including malignant plasma cells. By comparison of global gene expression profiling (GEP) derived from cultured MSCs, purified MM cells, and buffy coat BM samples as a source of hematopoietic cells, we identified a set of 485 probe sets encoding 350 specific MSC genes.

GEP were available from baseline BM-BX from healthy donors, patients with MGUS, SMM, and presenting cases of GEP70 low-risk (LR, n=467) and high-risk (HR, n=64) MM. Also available are paired BM-BX and FL-BX cases (n=77). To address the post-treatment role of signatures on outcome, BM-BX were taken during the first year of maintenance (n=161) and complete remission (n=132).

Associations between expression of MSC genes and progression-free survival (PFS) was assessed in Total Therapy (TT) 3-5 patients of whom 2/3 were randomly selected for training set and 1/3 used for test set.

Immunohistochemistry for EDNRA and IGFBP2 was performed in bone biopsy sections. Cell lysates and conditioned media from normal MSCs cultured alone or preconditioned with HR MM cells underwent GEP and proteomics analysis (n=4).

Results

Our analysis identified 61 and 178 MSC probe sets differentially expressed (FDR<0.05) between HR and LR patients and between random BM and FL sites, respectively. 43 probe sets were commonly changed in these two analyses highlighting the role of FL as a high risk niche. The top overexpressed MSC genes in BM-BX from HR patients were markers of pericytes and basement membrane ECM (EDNRA, COL4A1), whereas underexpressed genes were related to osteoblast (NPR3, ITGBL1, COL1A) and adipocyte (IGFBP2) activity. Expression of these genes in BM-BX gradually changed between healthy donors, MGUS/AMM, LR MM, HR MM and FL biopsies.

We developed a gene-score signature combining expression of 3 MSC genes (COL4A1, NPR3, ITGBL1), which was highly associated with outcome. A high 3-MSC gene-score signature at baseline was associated with adverse clinical and molecular parameters, ISS stage, and with shorter PFS in training (p<0.0002) and test (p<0.02) sets. GEP70 LR patients with a high MSC-gene score had shorter PFS compared to GEP70 LR patients with low MSC gene-score in the training set (p<0.008), whereas the outcome of GEP70 HR patients was not impacted by the MSC gene-score. A high 3-MSC gene-score was also detected in 26% of MGUS/SMM patients which predicted a time to therapy in SMM patients (p<0.08).

To define a favorable impact of post-treatment score we analyzed BM-BX taken during the 1^st year of maintenance, a standardized landmark time point in our Total Therapy protocol for newly diagnosed patients. We identified 3 MSC genes with highest hazard ratio (RBPMS, TGM2 and NEXN); the first two genes are involved in activated TGFβ pathway. Maintenance-based 3-MSC gene-score was associated with disease progression from maintenance therapy landmark in training (p<0.0008) and test (p<0.017) sets and could effectively differentiate GEP70 LR MM patients who are at risk for early relapse in both sets of data (p<0.0001).

We confirmed the gene expression data at the protein level using immunohistochemistry and identify EDNRA+ mesenchymal-like cells at a higher number in bones from MM vs. MGUS/SMM (p<0.0003) and between HR vs. LR (p<0.03). In contrast, IGFBP2 was highly expressed in immature adipocytes and its expression was lower in MM than SMM samples. In MSCs preconditioned with MM cells, EDNRA increased by 5 fold whereas NPR3, ITGBL1 and IGFBP2 decreased by 3 fold compared to control MSCs. Levels of the secreted factors, IGFBP2 and ITGBL1, were reduced in conditioned media from preconditioned MSCs compared to control MSCs.

Conclusion

MSC lineages and factors play a pivotal role in forming the MM niche and mediate the transition from asymptomatic to symptomatic stage and high risk.