AUTHORS
Prajish Iyer, PhD, Lu Yang, Ph.D.2, Sutapa Sinha, PhD3, Supriyo Bhattacharya, PhD4, Yucai Wang, MD, PhD5, Zhi-Zhang Yang, MD6, Min Shi, M.D.7, Charla Secreto8, Stephen Maxted Ansell, MD, PhD9, Hyojin Kim, PhD10, Neil E Kay, M.D.11, Wei Ding, MD, PhD12, Lili Wang, MD, PhD13
AFFILIATIONS
1 Columbia University, New York, New York, United States
2 City of Hope National Medical Center, Duarte, California, United States
3 Mayo Clinic, Rochester, Minnesota, United States
4 City of Hope National Medical Center, Duarte, California, United States
5 Mayo Clinic, Rochester, MN, United States
6 Mayo Clinic, Rochester, MN, United States
7 Mayo Clinic, Rochester, MN, United States
8 Mayo Clinic, Rochester, MN, United States
9 Mayo Clinic, Rochester, MN, United States
10 Mayo Clinic, Rochester, MN, United States
11 Mayo Clinic, Rochester, United States
12 Mayo Clinic, Rochester, MN, United States
13 Columbia University, New York, New York, United States
ABSTRACT
Despite major therapeutic advances in chronic lymphocytic leukemia (CLL) therapy, Richter transformation (RT), the aggressive progression of CLL to diffuse large B-cell lymphoma, remains incurable. Although PD-1 blockade has demonstrated clinical activity in RT, only ~30–40% of patients respond to pembrolizumab. The T cell subsets underlying response remain poorly understood. We previously reported that baseline Treg signatures could differentiate pembrolizumab responders from progressors (Iyer et al, ASH 2020) using paired samples from 3 patients. Hence, we hypothesized that pre-existing T-cell states and their differentiation trajectories determine therapeutic response to pembrolizumab in RT.
Peripheral blood T cells from 19 RT patients treated with pembrolizumab monotherapy were profiled before and after three months of therapy initiation using paired single-cell RNA sequencing and T-cell receptor (TCR) sequencing. Patients were classified as responders (R, n = 5), stable disease (SD, n = 5), or progressive disease (PD, n = 9). Cellular composition, transcriptional programs, TCR clonality, pseudo-time analysis, and spatial transcriptomics of matched lymph node biopsies were integrated to identify immune determinants of therapeutic response.
To define the T cell states associated with response to pembrolizumab, we analyzed 408,627 T cells in total and identified a pre-existing proliferative T cell state in responders. Responders exhibited a lower baseline CD4:CD8 ratio (p = 0.02), driven by expansion of cytotoxic and proliferating CD8 T cells (p = 0.008 vs PD; p = 0.001 vs SD) and CD4 Th17 cells, with depletion of naïve CD4 and Th2 subsets. The proliferating CD8 T cells showed increased expression of proliferation (MKI67, TOP2A), cytotoxic (GZMB, GNLY), and checkpoint (PDCD1, CTLA4) genes, together with enrichment of inflammatory and cell-cycle programs.
To identify T cell subsets responding to pembrolizumab, we analyzed PDCD1 expression across T cell populations and found enrichment in proliferating CD8, exhausted CD8, and CD4 Th17 cells. Comparison of PDCD1-positive versus PDCD1-negative exhausted CD8 T cells identified CCL3L1 as one of the most significantly upregulated genes in responders, suggesting a role in PD-1-responsive exhausted CD8 T cells. Because progenitor-exhausted (TPEX) cells mediate durable responses to PD-1 blockade, we next examined whether CCL3L1 marked this population. CCL3L1-positive T cells were enriched for TPEX cells expressing TCF7, SLAMF6, CXCR5, and PDCD1. Monocle 3 pseudotime analysis demonstrated sustained TPEX persistence in R from baseline through Cycle 3, whereas TPEX cells were transient and progenitor scores progressively declined in SD and PD patients. Following pembrolizumab, R retained and expanded CCL3L1+ TPEX cells, whereas PD-SDs lost these cells and accumulated terminally exhausted T cells, indicating that TPEX maintenance is associated with effective PD-1 responses. Studies defining the functional role of CCL3L1 in TPEX maintenance and anti-tumor immunity are ongoing.
Consistent with these findings, responders exhibited a less diverse, oligoclonal T-cell repertoire at baseline, with lower DE50 scores than PD or SD patients (p = 0.001 vs. PD; p = 0.02 vs. SD) and enrichment of medium-to-hyperexpanded clonotypes. Following PD-1 blockade, responders showed expansion of cycling CD8? effector clones and sharing of dominant clonotypes across CD8? effector, proliferating, and CD4? helper T-cell compartments. In contrast, non-responders maintained polyclonal, compartmentalized repertoires dominated by rare, non-cycling clones. Spatial transcriptomics validated these findings, demonstrating immune-rich niches with PD1-positive T cells, increased T-cell-B-cell proximity, and CCL3L1-associated immune interactions in responders, versus macrophage- and MDSC-dominated immunosuppressive regions in progressive disease.
Taken together, we identified a pre-existing immune ecosystem associated with response to PD-1 blockade in RT, characterized by distinct T-cell subsets, oligoclonal TCR architecture, and persistence of CCL3L1? progenitor-exhausted (TPEX) cells. Our data suggest that CCL3L1? TPEX cells sustain effective antitumor immune responses following PD-1 blockade. Peripheral immune features, including a low CD4:CD8 ratio, increased CCL3L1? TPEX scores, and reduced TCR diversity (DE50), may serve as non-invasive biomarkers for patient stratification and immunotherapy selection in RT.
DISCLOSURES
Prajish Iyer, PhD: No relevant conflicts of interest to declare.