Single Cell Transcriptomic Characterization of the Immune Microenvironment in Naturally Progressing Chronic Lymphocytic Leukemia (CLL)

The genetic changes occurring during natural progression and following therapy in CLL have recently been characterized. However, the role of co-existing immune cell populations during disease evolution has not been systematically evaluated. New single cell transcriptome sequencing (scRNA-seq) technologies provide a powerful approach for dissecting the composition and state of complex heterogeneous cell populations such as immune cells. We hypothesized that changes of specific immune features underlie the kinetics of progression of CLL.

To this end, we selected a discovery set of CLL samples from 6 patients previously genetically characterized by bulk whole-exome sequencing (WES). Three patients were "slow-progressors" (median time to first treatment [TTFT]: 10.7 years (yrs), range: 7.7-12.2), while the other three were "fast-progressors" (median TTFT: 1.9 yrs, range: 1.3-2.4). We processed paired samples obtained close to the time of diagnosis (T1; median yrs from diagnosis: 2.54, range 0.6-4.2) and close to first treatment (T2; median yrs to TTFT: 0.22, range 0-5.9). Non-CD19⁺CD5⁺ cells were isolated from peripheral blood mononuclear cells by flow cytometry for scRNA-seq using the 10X Genomics Chromium Controller platform. Replicates were available in 8 of 12 samples, for a total of 20 specimens. Dimensionality reduction and clustering was then performed by a novel pipeline called MUDAN (multi-sample unified discriminant analysis). We used a mixed-effects model to identify trends in cell type-specific gene expression and proportions over time across patients.

We observed changes in the composition of immune cells over time across these patients. In total, 57,069 cells were analyzed (median number of cells per sample: 4,394, range 774-8657) and the mean number of genes per cell was 906 (SD: 116.7). We identified a total of 13 immune cell types, consistent with diverse subsets of T and NK cells, myeloid cells, and hematopoietic stem cells. At T1, fast-progressors had a higher proportion of myeloid cells (dendritic cells, classical and non-classical monocytes, p<0.0001, p=0.004 and p<0.0001, respectively) as compared to slow-progressors, which showed a higher proportion of naïve and memory CD4⁺ T cells (p=0.001 and p=0.016, respectively). Over the course of disease, both fast and slow-progressors converged into a decrease in the lymphoid:myeloid ratio (L:M ratio in T1: 10.1 vs. T2: 7.3, p=0.023), due in part to a proportional increase in non-classical monocytes (p=0.004) and a decrease in CD4⁺ T cells (p=0.003).

Over time, we observed a broad upregulation of gene expression across all cell types in both fast and slow-progressors. Within the lymphoid compartment, we observed an upregulation of cell activation (CD3E, CD69, and GITR), chemotaxis (CCL3, CCL4, and XCL2), cytotoxicity (GZMK, GZMA, and NKG7) and pro-inflammatory (INFG, ISG15, and TNF) markers, especially in memory CD4⁺ T cells. In contrast, memory CD8⁺ T cells, which were marked by the inhibitory molecule KLRB1(CD161) in our cohort, showed the lowest increase in these gene signatures. TIGIT followed by LILRB2, CD160 and LAG3 were the most upregulated immune checkpoints across time, especially in memory CD4⁺ and naïve CD8⁺ T cells, and resting NK cells. LAG3 was revealed to be the only of these immune checkpoints associated with fast-progressors (p<0.05). No significant changes in the expression of PDCD1, HAVCR2 (TIM3) or TNFRSF14 (HVEM) were observed over time. Within the myeloid compartment, we observed a lack of upregulation of cell activation markers along with a marked downregulation of chemoattractants (XCL1, CXCR5 and CXCR4) and antigen presentation molecules (HLA-DOA, HLA-E, and PSME1) over time, especially in classical monocytes. Further analyses correlating CLL genetics to immune kinetics and validating these results in a larger patient cohort are ongoing.

Altogether, these findings provide fresh insights of the evolving immune microenvironment during CLL progression. Despite differences in the immune composition at the time of diagnosis, fast and slow-progressors converged into an immune activation state, mostly dominated by CD4⁺ T cells, ineffective cytotoxic cells and poor immunogenic phagocytic cells that ultimately contributed to the progression of the disease. These results provide a rationale for new immunotherapeutic strategies in previously untreated CLL patients.