Type: Oral
Session: 722. Allogeneic Transplantation: Acute and Chronic GVHD and Immune Reconstitution: Monitoring, Understanding and Optimizing GVHD Interventions
Hematology Disease Topics & Pathways:
Genomics, Computational biology, Biological Processes, Technology and Procedures, Machine learning
We found that severe (Grade 3-4) GvHD patients exhibit higher cumulative frequencies, diversity, and number of unique alloreactive clones compared to mild (Grade 1-2) GVHD patients (p<0.05). Our temporal analysis showed that severe GvHD patients have a significantly higher number of expanded alloreactive clones with long-term persistence (after day 68; p=0.021). To further explore the dynamics of T cell clonotypes, we adapted a machine learning algorithm for the decomposition of temporal patterns (Nazaret et al. 2023) which resolves the timing of clonal expansion and their differences between patients. We found distinct temporal patterns associated with alloreactive clones from different grades of GvHD (p<6.6e-3). While patients with no (2 patterns) GvHD showed alloreactive clones with decreasing frequency over time, patients with mild (1 pattern) and severe (3 patterns) GvHD were more likely to exhibit transient or persistent patterns.
In gut biopsies from post-transplant patients, we applied our recently developed computational tools (Nazaret et al. 2023; He et al. 2024) to analyze T cell trajectories and their spatial context. The tissue composition of T cell states in post-transplant biopsies was significantly altered compared to normal donors. We found infiltrating activated cytotoxic T cells (upregulation in GZMA, GZMB, GZMH, GZMK, IFNG) and proliferating CD8+ T cells to be significantly enriched in severe GvHD patients compared to mild or no GvHD patients (p<4.9e-3). Interestingly, TCR clonotype profiling revealed that these infiltrating activated cytotoxic and proliferating CD8+ T cells have lower clonal diversity (higher Gini index) in severe GvHD patients (p<4.6e-2). Surprisingly, we found enrichment of regulatory T cells in severe GvHD patients (p<7.1e-3), although they exhibit phenotypes of cytotoxicity, AIM2 inflammasome activation, and IL-6 signaling (PRF1, GZMA; GBP2, GBP5; IL6ST, STAT3, JAK1) compared to Tregs from no/mild GvHD patients or normal donors (p<3.0e-4).
Trajectory analysis of single-cell transcriptomic data highlighted two disease-associated cell state transitions, with severe GvHD patients enriched in more effector and proliferating states, whereas this phenotypic shift is not observed in normal donors and is less prominent in mild GvHD. Lastly, we applied our Starfysh (He et al. 2024) algorithm to integrate spatial transcriptomics data from patients with varying grades of GvHD and identified grade-specific hubs as regions with distinct cell type compositions. Our spatio-temporal modeling of T cells in GvHD patient specimens identified T cell phenotypes and spatial niche types enriched in patients exhibiting severe GvHD.
In summary, by integrating MLRs, high-throughput TCR sequencing, single-cell transcriptomics, and spatial transcriptomics, we have characterized the temporal and spatial dynamics of alloreactive T cells in human GvHD. This comprehensive approach enables the identification of biomarkers associated with GvHD severity and highlights the potential for targeted therapies to improve patient outcomes in allo-HCT.
Disclosures: Reshef: Takeda: Research Funding; J&J: Research Funding; Synthekine: Research Funding; Incyte: Consultancy, Research Funding; Atara Biotherapeutics: Research Funding; Sanofi: Research Funding; TCR2: Research Funding; Genentech: Research Funding; Bayer: Consultancy; CareDx: Research Funding; Cabaletta: Research Funding; BMS: Research Funding; Immatics: Research Funding; Sana Biotechnology: Consultancy; TScan: Consultancy, Research Funding; Orca Bio: Consultancy; Quell Biotherapeutics: Consultancy; Autolus: Consultancy; Abbvie: Research Funding; Precision Biosciences: Research Funding; Gilead Sciences: Consultancy, Research Funding; Allogene: Consultancy.