A Suboptimal Response in CML Is Associated with Expression of the Immune Checkpoint VISTA

Introduction: Tyrosine kinase inhibitors (TKIs) have revolutionized the treatment of chronic myeloid leukemia (CML) leading to molecular remission in most patients. The next challenges are decreasing resistance and improving the odds of treatment-free remission. Understanding determinants of disease persistence and suboptimal responses is critical for improving these probabilities. The role of the immune system in the control or eradication of CML has been demonstrated in multiple settings including successful graft-vs-leukemia effect and the requirement of active T-cells and natural killer (NK) cells for sustained remission prior to therapy discontinuation. Herein, we aim to demonstrate a comparison of immune determinants of response in CML in those who achieve a deep molecular response (MR4.5) compared to suboptimal responders.

Methods: We prospectively collected longitudinal blood and bone marrow samples from patients receiving treatment for CML and characterized them using cytometry by time of flight (CyTOF). 20 patients that achieved MR4.5 at any time during their treatment were compared to a cohort of 32 patients with suboptimal response to therapy according to the European LeukemiaNet (ELN) response milestones (Table 1). We custom-designed this CyTOF panel (36-plex) to include markers of leukemia/stem progenitor cells, lineage defining, intracellular signaling, immune exhaustion and activation markers. Samples were barcoded, pooled and processed simultaneously to overcome technical variations and batch effects. CD45^hi cells (leukocytes) were grouped using PhenoGraph clustering, based on expression of these markers in the earliest samples available, collected at various timepoints during treatment.

Results: Analysis of differential expression patterns revealed 25 distinct clusters. Patients with suboptimal response to therapy had a higher proportion of CD45^hiCD3+ cells (T-cells) compared to those with an optimal response (median of 59% vs 53%, P=0.04). Notably, this included a higher proportion of exhausted CD8+ cytotoxic T-cells in patients with suboptimal response characterized by high Eomes and low Tbet expression (median of 8.0% vs 1.9%, P=0.04). In contrast, CD56+ NK-cells were more prevalent in the optimal response group (median of 5.5% vs 3.2%, P=0.04), particularly an interferon gamma releasing (CD56^hiCD16^loTbet+) subpopulation (median of 4.1% vs 2.5%, P=0.04). While CD68+ cells (macrophages) were detected at similar rates in both the optimal and suboptimal response groups (median of 16.5% vs 15.5%, P=0.07), the phagocytic subpopulation (CD16-CD14+CD68+HLA-DR^lo) was more prevalent in those who achieved an optimal response (median of 15.1% vs 13.2%, P=0.02). Patients with suboptimal response had a higher proportion of CD68+HLA-DR^loVISTA^hi cells (median of 5.6% vs 1.8%, P=0.004), referring to macrophages expressing the immune regulatory ligand V-domain immunoglobulin suppressor of T cell activation (VISTA). Like PD-L1 and other B7 family ligands, VISTA attenuates T-cells response to antigens and VISTA overexpression has been associated with resistance to immune checkpoint inhibitors in solid tumors. Furthermore, cells characterized by CD66b+CXCR4+VISTA^lo were increased in patients with optimal response (median of 20.6% vs 0.6%, P=0.01), indicating a higher proportion of mature granulocytes with low expression of VISTA in the optimal response group (Figure 1).

Conclusion: We found that higher expression of the immune checkpoint VISTA, in addition to markers of T-cells exhaustion is associated with suboptimal response to therapy in CML. These findings highlight the importance of the immune microenvironment in response to targeted therapies such as TKIs.