Multidimensional Analysis of Relapsed Bone Marrow Reveals Cold Immune Microenvironment Post-TMLI in Acute Leukemia Transplant Patients

Introduction: Patients with refractory leukemia face poor outcomes, highlighting a significant unmet clinical need. To address this challenge, we developed total marrow and lymphoid irradiation (TMLI) as a new conditioning regimen for hematopoietic cell transplantation (HCT), which enables dose escalation to the bone marrow (BM) and lymph nodes to improve survival. We recently reported a phase 2 trial of TMLI (2000 cGy in 10 fractions) with cyclophosphamide and etoposide as a conditioning regimen for refractory acute leukemia patients (n=74). The median age was 40 years (range: 16.5-59.3 years), and the median blast percentage was 20% (range: 0-95%). Patients received HCT from either sibling donors, matched unrelated donors, or mismatched unrelated donors. The trial showed promising results: Two-year progression-free survival and overall survival estimates, respectively, were 34% (95%CI: 22-46%) and 48% (35-61%) in AML, and 22% (7-43%) and 39% (18-60%) in ALL. The objective of our correlative study was to conduct a comprehensive analysis of the BM microenvironment in patients, with a particular focus on identifying immune cell populations and the functional states that are associated with either relapse or response.

Methods: The clinical correlative study received approval from our Institutional Review Board (IRB). Patient BM cells were collected at different times, and mononuclear cells (MNCs) were isolated and cryopreserved. Between 60-150 days (when most relapses occur), we collected BM samples from 14 patients who had relapsed and from 14 patients who did not relapse and eventually experienced long-term remission. The MNCs were thawed and stained with antibodies, using standard procedures, against validated markers for immune cells, hematopoietic stem cells, endothelial cells, and mesenchymal stem cells. Data were acquired by Aurora spectral flow cytometry. Transcriptional profiling of patient BM (12 samples) associated with complete remission MRD-negative; n=5), low disease (MRD-positive --15%; n=3), high (≥25%; n=4) disease burden was assessed by Nanostring. Further, transcriptional profiling at the single cell level was carried out on one remission and one relapse sample using 10X genomics, and data were analyzed using Seurat. Multiplex immunofluorescence was performed on patient iliac core biopsy samples for spatial immune cell profiling.

Results: BM flow cytometric analysis showed reduced mean CD3⁺ CD8⁺ T cell frequencies in relapsed sample (3.4%) compared to CR (9.2%). NanoString transcriptomic profiling of BM MNCs from patients who relapsed after TMLI-HCT showed a significantly higher ratio of CD163+ve M2 macrophages to cytotoxic CD8⁺ T cells (p=0.03), suggesting an immune suppressive microenvironment, the ratio progressively increasing from responder to low disease to high blast. Because bulk RNAseq cannot identify a specific cell type, we turned to scRNAseq to identify transcriptional changes at the single-cell level. The single-cell RNAseq analysis of BM cells (CD45⁺) showed a 2-fold reduction in the frequency of CD3⁺/CD8⁺ T cells in relapsed BM samples compared to levels from patients in remission. Further, the differentially expressed genes from T cell clusters showed 1.5-2-fold increased T cell exhaustion marker genes (PD1 and LAG3) in relapse over remission. Multiplex immunofluorescence of samples from relapsing patients revealed phenotypic and spatial diversity among immune cells, with blast and CD8+ T cells adjacent to macrophages, a configuration that is known to protect AML cells and suppress immune response.

Conclusion: In relapsed patients, the immune and transcriptional profiling of patient BM post-HCT revealed an immunosuppressive, pro-tumorigenic microenvironment. The multidimensional analysis highlighted a "cold" immune microenvironment in the BM at relapse, characterized by T cell exhaustion, reduced cytotoxic activity, and increased immunosuppression, likely allowing leukemic cells to evade immune surveillance, and contributing to disease progression. Future studies should evaluate T cell functions, macrophage interactions, and pathways associated with relapse using larger sample sizes. These insights will aid in developing regimens that combine TMLI with immunotherapy to enhance immune modulated response and will also facilitate identifying predictive biomarkers of relapse for early intervention.