Evidence of Targetable Immune Dysfunction in the Bone Marrow of Patients with Intermediate/High-Risk Myelodysplastic Syndrome Refractory to Hypomethylating Agents

Background Growing evidence indicates that a dysfunctional adaptive immune response contributes to the pathogenesis of myelodysplastic syndrome (MDS). Intermediate/high (I/H)-risk MDS is associated with immunosuppression, and median overall survival in hypomethylating agent (HMA)-refractory patients (pts) remains poor at < 6 months. Thus, new immune checkpoint-targeting therapies may be important for restoring effective anti-tumor T-cell mechanisms in these pts. Here we aim to evaluate the relationship between immunomodulatory and regulatory targets in lymphocytes and CD34+ blasts, cytogenetics and prior treatment regimens to inform future novel immunotherapy combinations.

Methods Bone marrow aspirates from HMA-refractory, I/H-risk pts (n = 18) and normal donors (NDs; n = 20) were assessed by flow cytometry (FC), immunohistochemistry and RNA sequencing (RNAseq). An independent cohort of treatment-naive, I/H-risk pts (n = 9) was assessed for regulatory T-cell (Treg) counts by FC. T-cell subsets (Treg, T-naive, T-effector memory, T-central memory, T-late effector memory) were identified by FC based on established markers (CD3, CD4, CD8, CD197, CD45RO, CD127, CD25, FOXP3). Activation/proliferation status of T-cell subsets was measured by HLA-DR and Ki-67. MDS blasts were identified by CD45, CD34 and CD117. Immune checkpoint markers programmed death-1 (PD-1) and programmed death-ligands 1 and 2 (PD-L1/2) were measured on T cells and CD34+ blasts. T-cell immunoglobulin and ITIM domain (TIGIT) was also measured on CD4 and CD8 T cells. T cells (CD4, CD8) and PD-L1 were identified in decalcified bone marrow biopsies by immunohistochemistry. Whole transcriptome gene expression analysis was performed using the Illumina TruSeq RNA Access Library Prep Kit on bulk bone marrow aspirates.

Results T cells were detected in bone marrow aspirates from HMA-refractory, I/H-risk pts at levels comparable to those in NDs. Treg counts (cells/µL) were increased in HMA-refractory vs treatment-naive pts (P= 0.017), with the highest Treg count associated with complex cytogenetic karyotypes (≥ 3 abnormalities). We observed increased T-cell activation and proliferation in these pts compared with NDs (MDS vs ND medians [% positive cells]; HLA-DR−/Ki-67+: 1.23 vs 0.75; HLA-DR+/Ki-67+: 1.71 vs 0.33; HLA-DR+/Ki-67−: 14.14 vs 4.80). No trends were seen in CD4-naive and memory subsets, whereas CD8-naive cells were lower and late effector memory cells were higher than in NDs (P= 0.035 each). We observed higher expression of immune checkpoint and regulatory molecules (PD-L1, PD-1, TIGIT) in MDS except for PD-L2, which was minimally detected (< 5%). Frequency (%) of PD-L1 on CD34+ blasts was higher in pts with MDS than in NDs (P= 0.02) and was associated with complex karyotype (P= 0.026) and Treg frequency (P= 0.021). PD-1+ T cells were higher in pts with MDS than in NDs (CD4, P= 0.004; CD8, P= 0.012), with a trend toward increased PD-1 expression associated with number of prior therapies. TIGIT expression on CD4 T cells correlated with PD-1 (P= 0.008) and Treg frequency (P= 0.022). Lymphocyte-activation gene 3 expression by RNAseq was also higher in MDS vs ND marrow (P < 0.001). Some signals, such as PD-L1 and Tregs, correlated by FC and RNAseq.

Conclusions Bone marrow in pts with I/H-risk MDS demonstrate evidence of prolonged T-cell activation, an increase in suppressive immune cell populations (i.e., Tregs) and expression of immune checkpoint molecules (PD-L1, TIGIT) compared with normal bone marrow. Patients with complex karyotype had subtle differences, potentially relevant for identifying subsets responsive to checkpoint inhibitors. These data provide new insight into the immune landscape of HMA-refractory I/H-risk MDS.