Composition and Fitness of T and NK Cells in Extramedullary Myeloma Tumor Microenvironment

Background:

Extramedullary disease (EMD) is an aggressive manifestation of multiple myeloma (MM), when clonal plasma cells (PCs) become independent on the bone marrow (BM) microenvironment and invade distant tissues and organs. The incidence of EMD is increasing and is associated with drug resistance and poor prognosis. The efficacy of modern immunotherapy is highly dependent on the patient's own immune system. However, there is limited knowledge about the composition of the EMD microenvironment and the fitness of its effector immune cells.

Aims:

To investigate the composition and fitness of T and NK cells in extramedullary tumor and corresponding BM microenvironment

Methods:

Biopsy of EMD soft tissue tumor was performed in relapsed/refractory MM patients and samples were processed immediately after surgery (Jelinek, Leukemia, 2024). Single-cell RNAseq was performed using Chromium GEM Single Cell 3' reagent kit v3.1 (10x Genomics) and cell suspensions from 7 EMD tumors, 5 BM from time of EMD relapse (EMD_BM, 4 paired with EMD tumors), and 5 unrelated RRMM_BM without EMD. The computational pipeline combined scanpy python framework and R packages SoupX for ambient RNA filtering and scDblFinder for doublet removal. Cytotoxicity and dysfunctional score of T cells were based on Li et al., 2019, Cell. For flow cytometry (FC) assessment, four 8-color panels were used consisting of CD3, CD4, CD8, CD16, CD27, CD45, CD45RA, CD56, CD57 as backbones (and other activation or inhibitory molecules) for dissection of T and NK cell subsets in EMD (N=6) and EMD_BM (N=5).

Results:

Single-cell RNA-seq analysis yielded a median of 3,489, 1,409, and 4,611 cells for EMD, EMD_BM, and RRMM_BM samples, respectively. The median proportion of PCs was 92% for EMD and 36% for RRMM_BM samples. Interestingly, the median number of PCs in EMD_BM samples was only 0.2%. The median proportion of cells in the T/NK cluster was 7% for EMD, 18% for EMD_BM, and 20% for RRMM_BM.

Semi-automated cell type annotation using Celltypist, the Leiden clustering algorithm, and literature review resulted in the identification of 7 distinct T cell and 2 NK cell clusters. We observed a substantially lower number of CD4+ T cells in EMD compared to unrelated RRMM_BM (median 2% vs 31% of T cells; p = 0.02) and EMD_BM (12%; p = 0.05). Importantly, this finding was validated by FC (median 10.7% vs 31.6% for EMD vs EMD_BM; p = 0.05). The remaining six T cell clusters were all CD8+ with markedly lower estimated cytotoxicity (e.g. FAM65B, IL7R, PLEK, etc.) and dysfunction (e.g. LAG3, TIM3, PD1, TIGIT, etc.) scores in EMD compared to EMD_BM or RRMM_BM. This trend was attributed to different composition of T cells between the groups. EMD had the highest proportion of cells in clusters with low cytotoxicity and high dysfunction, including CD8_proliferating_MKI67 (median 14% vs 2% and 0.2%; p = 0.03 and 0.02 for EMD_BM and RRMM_BM), CD8_exhausted-like_Tox (3% vs 0% and 0%; p = 0.09 and 0.13), and in the cluster specific almost exclusively for EMD that we annotated as CD8_MTRNR2L12 (19% vs 2% and 1%; p = 0.05 and 0.05). Additionally, using FC we found an increased percentage of T cells positive for exhaustion marker PD-1 (39.4% vs 15.9%; p = 0.03) in EMD compared to EMD_BM and higher proportion of central memory CD8+ T cells (61.9% vs 31.9% of CD8+ T cells).

Analysis of NK cell compartment revealed significantly higher proportion of CD16- NK cells (median 70.3% vs. 7.5% and 6.8%; p=0.007 and 0.003) compared to CD16+ cytotoxic NK cells (29.7% vs. 92.5% and 93.2%), a well-known phenomenon in solid oncology (Rebuffet et al. 2024, Nature Immunology). This finding was further confirmed by FC (37.9% vs 7.6% of total NK cells; p = 0.016). Importantly, CD16- NK cells from EMD exhibited significantly higher expression of inhibitory receptor NKG2A compared to EMD_BM (p < 0.001).

Conclusion:

In this study, we revealed that majority of T cells in EMD tumor microenvironment are CD8+ T cells with impaired cytotoxicity and increased exhaustion compared to BM CD8+ T cells. For the first time, we demonstrated that NK cells infiltrating EMD tumors are represented dominantly by CD16- NK cells (in contrast to BM) resembling the situation in solid cancers. Moreover, EMD CD16- NK cells have significantly higher expression of NKG2A checkpoint compared to BM CD16- NK cells suggesting potential therapeutic opportunity with NKG2A inhibitors in EMD MM patients.