Single cell analysis of chronic myeloid leukemia bone marrow from time of TKI discontinuation reveals stem cell alterations associated with recurrence versus treatment free remission

Abstract Treatment free remission (TFR) is the ultimate goal for chronic myeloid leukemia (CML) patients, but only ~20% of all CML patients are able to achieve it. The fundamental reasons for CML recurrence are that leukemic stem cells (LSC) can survive tyrosine kinase inhibitor (TKI) treatment and outcompete non-leukemic hematopoietic stem cells (HSC) if treatment is stopped. Previous studies suggest that LSC persist in patients maintaining TFR, since BCR-ABL continues to be detectable using sensitive assays (Ross D et. al. Blood 2013;122(4)). However, it is unknown why LSC are repressed while HSC reconstitute the blood system in TFR, but LSC expand and outcompete HSC in recurrence. We hypothesize that HSC regulation may play an active role in TFR, and studying HSC may reveal new avenues of treatment to improve TFR incidence. We studied bone marrow (BM) samples obtained from CML patients who achieved deep molecular remission after long-term TKI treatment, prior to stopping treatment in a TFR attempt. Our study included samples from 19 patients aged 24-71 (mean 52.5), 12 females and 7 males, who had received an average of 7 years of Imatinib (range: 2.6-13.5 years). 7 patients maintained TFR (TFR patients), whereas 12 patients experienced recurrent disease (recurrent patients). Cryopreserved BM mononuclear cells (MNC) samples were thawed, underwent dead cell removal and a portion were enriched for CD34+ cells. Equal numbers of MNC and CD34+ cells were analyzed by CITE-Seq. After hashtag and antibody derived tag labeling, cells were processed using the 10X Genomics platform and sequenced on the NovaSeq platform. The data were processed with the Cell Ranger pipeline, assessed for quality, and cells were projected onto hematopoietic and immune atlases for cell type annotation based on Zhang et. al. Nat Immunol. 2024;25:703 Applying a gene expression profile for BCR-ABL+ CD34+ cells (Giustacchini et. al. Nat Med. Jun 2017;23:692)to the HSC cluster to differentiate BCR-ABL+ from BCR-ABL- HSC, we determined that TFR patients had a median of 4.4% (range 0-16.7%) BCR-ABL+ HSC, while recurrent patients had 1.5% (0-6.2%) BCR-ABL+ HSC. In contrast, CML patients at diagnosis (n=5) had a median of 82.8% (range 63-97.5%) BCR-ABL+ HSC, while healthy controls (n=4) did not have any BCR-ABL+ HSC detected, supporting the validity of this approach. We compared HSC between TFR and recurrence patients for differentially expressed genes (DEG) and pathway enrichment using EdgeR and Limma mixed effects analysis, correcting for patient-specific effects. We observed increased expression of genes related to transcriptional regulation (AGO1, DDX42, STAT5A), signal transduction (ACVR1B, GAB1, ILKAP), and cell cycle (CTC1, ERAL1) in recurrent patient HSC, while TFR patient HSC showed increased IL16, a leukocyte migration marker, and the ribosomal biogenesis marker C1orf109. Enrichment analysis using DAVID revealed TNF-induced inflammation to be upregulated in HSC from recurrent patients. GSEA analysis of HSC also showed IFNa, IFNy, and TNF gene signatures to be increased in recurrent patients. Comparison of leading edge genes for TNF signaling pathways in HSC against other cells using the CellChat database indicated an increase in IL18R1 and TNFRSF1B interactions in TFR patients, while TNFRSF1A and FAS interactions were increased in recurrence. NicheNet analysis corroborated these results, with TNF showing high prior interaction potential with TNFRSF1A in recurrence. Finally, NicheNet analysis also revealed that recurrent patient HSC tended to have higher signaling through PF4, JAM2/3, and TGF-B, known to be associated with regulation of HSC state and localization. Our data indicate that HSC in BM from patients in deep remission are predominantly BCR-ABL-, but that small numbers of BCR-ABL+ HSC may be detected both in patients that subsequently recur as well as those that maintain TFR. HSC from patients that recur after TKI withdrawal show evidence of altered TNF and TGF-B signaling compared to HSC from patients that maintain TFR, indicating an association of HSC inflammatory signaling with recurrence. Additional studies are warranted to investigate the mechanisms underlying persistent HSC inflammation after restoration of normal hematopoiesis and to determine whether inflammatory signaling can reduce the competitive fitness of HSC against LSC, enhancing susceptibility to leukemia recurrence when treatment is withdrawn.