AUTHORS
Tyler Freeland, B.S., Alecia Alto, BS2, Tiff Ricketts, PhD3, Gracie Jack4, Na Xie, PhD5, Sahee Park, PhD6, David Crossman, PhD7, Dong-Wook Kim, MD, PhD8, Ravi Bhatia, MD9
AFFILIATIONS
1 University of Alabama at Birmingham, Birmingham, Alabama, United States
2 University of Alabama at Birmingham, Birmingham, Alabama, United States
3 University of Alabama at Birmingham, Birmingham, Alabama, United States
4 University of Alabama at Birmingham, Birmingham, Alabama, United States
5 University of Alabama at Birmingham, Birmingham, Alabama, United States
6 Eulji University Medical Center, Uijeongbu-si, Gyeonggi-do, Korea, Rep. of South
7 University of Alabama at Birmingham, Birmingham, Alabama, United States
8 Eulji University Medical Center, Uijeongbu-si, Gyeonggi-do, Korea, Rep. of South
9 University of Alabama at Birmingham, Birmingham, Alabama, United States
ABSTRACT
Chronic myeloid leukemia (CML) patients achieving deep molecular remission have a ~50% rate of treatment‑free remission (TFR) after discontinuing tyrosine kinase inhibitor (TKI) therapy. Observations that low levels of residual leukemia cells continue to be present in most patients who remain in TFR argue against elimination of leukemia stem cells (LSC) and suggest a role for the bone marrow (BM) microenvironment (ME) in restricting leukemia regeneration. However, CML BMME characteristics during long‑term TKI therapy associated with TFR vs relapse are not well characterized. To address this gap, we analyzed BM samples collected just prior to stopping treatment to identify BMME programs associated with relapse vs. TFR.
Cryopreserved BM samples from 38 imatinib‑treated patients were studied; 14 relapsed after TKI withdrawal whereas 24 maintained TFR. Thawed cells were FACS‑sorted into CD45+CD34+ hematopoietic stem and progenitor cells (HSPC), CD45+CD14+ monocytes, CD45-CD235a-CD31+ endothelial cells, and CD45-CD235a-CD31- mesenchymal stromal cells and analyzed by bulk RNA sequencing, along with unsorted mononuclear cells (MNC), for transcriptomic changes associated with TFR vs. relapse. A leave‑one‑patient‑out (LOPO) design using Limma‑voom was used to ensure that differential expression profiles were reliable and stable, despite small numbers of replicates. Functional characterization of differentially expressed genes was performed using DAVID and gene set enrichment analysis (GSEA). BMME interactions were inferred with NicheNet.
MNC from relapsed patients showed increased expression of apoptosis resistance, heme metabolism, translation, and antimicrobial immunity genes, whereas MNC from TFR patients expressed enhanced lysosomal biogenesis genes. The HSPC population at this stage, which represents predominantly BCR-ABL-negative cells, showed elevated apoptotic resistance genes in relapse patients, but increased expression of membrane, adhesion, and Ig‑like receptors associated with extracellular interactions, and enrichment of glycolysis and hypoxia signatures, associated with HSC quiescence and self-renewal, in TFR patients. BMME monocytic and stromal cells from relapse patients were marked by strong enrichment for TNFA–NFκB and TGF-beta signatures, whereas TFR patients showed increased OXPHOS signatures. In contrast, endothelial cells from relapse patients did not show a consistent signature, whereas those from TFR patients showed increased expression of cell cycling genes, and enrichment of hypoxia, IL2, and cell cycle signatures.
Consistent with these findings, NicheNet analysis revealed heightened stromal and monocytic signaling activity associated with relapse and increased endothelial signaling with TFR. Relapse stroma cells received signals from monocyte- and HSPC‑expressed TNFSF13B and S100A12, a damage-associated molecular pattern (DAMP) molecule, associated with activation of ATF3 and CDKN1A, which regulate stress response. Relapse stroma also received IL33 signals associated with KLF6, LMNA, and NFKBIA activation, indicating an inflammatory state. Relapse monocytes received TNF, IL1B, IFNG, and IFNB1 signals from monocytes, HSPC, and stroma, associated with TNFAIP2, CD80, OLR1 and GBP2 activation, indicating inflammatory programming. Relapse HSPC received strong HMGB1 (DAMP) signals from all cell types and NOG signals from CD34 and Stroma, associated with activation of genes regulating HSC quiescence. Finally, endothelial cells from TFR patients received enhanced IGF1 signals of endothelial and stromal origin, associated with activation of cell cycling genes.
These results elucidate differential activation of BMME niche programs in patients who relapse following TKI cessation vs. those who maintain TFR. Relapse is characterized by pro-inflammatory cytokine and DAMP induced inflammatory signaling in stromal and monocytic cells. The presence of an inflammatory niche is expected to selectively enhance LSC activity and leukemic outgrowth, which requires continued TKI treatment to maintain leukemic control. In contrast, TFR is characterized by enhanced endothelial proliferation, suggesting an important role for BM microvasculature in supporting non-leukemic HSC and restricting LSC outgrowth after TKI cessation. Further resolution and functional validation will assist development of predictive biomarkers and novel therapeutic strategies to enhance TFR outcomes.
DISCLOSURES
Tyler Freeland, B.S.: No relevant conflicts of interest to declare.