AUTHORS
Hongye Gao, MD, Yawei Zheng2, Chenfeina Feng3, Lisha Wang4, Erlie Jiang, MD5, Fang Dong, PhD6
AFFILIATIONS
1 Institute of Hematology & Blood Diseases Hospital, Chinese Academy of Medical Sciences & Peking Union Medical College, Tianjin, China, Tianjin Institutes of Health Science, Tianjin, China
2 Institute of Hematology & Blood Diseases Hospital, Chinese Academy of Medical Sciences & Peking Union Medical College, Tianjin, China, Tianjin Institutes of Health Science, Tianjin, China
3 Institute of Hematology & Blood Diseases Hospital, Chinese Academy of Medical Sciences & Peking Union Medical College, Tianjin, China, Tianjin Institutes of Health Science, Tianjin, China
4 Institute of Hematology & Blood Diseases Hospital, Chinese Academy of Medical Sciences & Peking Union Medical College, Tianjin, China, Tianjin Institutes of Health Science, Tianjin, China
5 Institute of Hematology & Blood Diseases Hospital, Chinese Academy of Medical Sciences & Peking Union Medical College, Tianjin, China, Tianjin Institutes of Health Science, Tianjin, China
6 Institute of Hematology & Blood Diseases Hospital, Chinese Academy of Medical Sciences & Peking Union Medical College, Tianjin, China, Tianjin Institutes of Health Science, Tianjin, China
ABSTRACT
BACKGROUND
Autologous stem cell transplantation (ASCT) remains an option for patients with B-cell acute lymphoblastic leukemia (B-ALL), but poor hematopoietic stem/progenitor cell (HSPC) mobilization impairs engraftment and remains a major challenge. Acute leukemia is associated with poor mobilization, yet its cellular basis is rarely characterized. We integrated clinical outcomes with single-cell RNA sequencing (scRNA-seq) of B-ALL grafts to define the burden and cellular basis of poor mobilization.
METHODS
The NICHE cohort (NCT04645199) included 228 B-ALL patients undergoing autologous stem cell collection (2004-2024); we assessed CD34+ yield and post-transplant engraftment. Patients were classified as Poor, Standard, or Good by total CD34+ yield/kg (<2, 2-5, >=5 x106, respectively). scRNA-seq of enriched CD34+ cells (119,338 cells, 22 samples; MRD-negative) included healthy bone marrow (n=4), mobilized healthy blood (n=6), B-ALL Good (n=4), Poor (n=4), Standard (n=1), and paired pre-/post-plerixafor specimens from 3 Poor donors. Donors (not cells) were the unit of biological replication for Milo, edgeR pseudobulk, pathway, and pySCENIC analyses (DEGs FDR<=0.05, |log2FC|>=0.5). Engraftment delay was also assessed by 1:1 propensity-score-matched analysis on CD34+ dose. Good and Poor CD34+ cells were xenografted at equal dose into NOG mice (n=3/group; 5.8x104 cells/mouse); peripheral-blood CD45+ chimerism was measured at weeks 4, 8, and 12.
RESULTS
Among 228 B-ALL patients, 44 (19.3%) were Poor, 86 (37.7%) Good, and 98 (43.0%) Standard mobilizers. Collection failure alone precluded ASCT in 9 (3.9%); relapse/MRD did so in 3 (1.3%). Median infused CD34+ dose was lower in Poor than Good mobilizers (1.9 vs 3.8 x106/kg), as was engraftment speed (neutrophil 14.5 vs 12 days; platelet 18 vs 13 days; both P<0.0001). In an exploratory 1:1 matched analysis of equal doses of infused CD34+ cells, Poor mobilizers showed median within-pair engraftment delays of 4.5 days for WBC (28 pairs; P<0.001) and 15.5 days for platelets (24 pairs; P=0.002). After equal-dose CD34+ xenotransplantation into NOG mice, week-4 peripheral-blood CD45+ chimerism was lower in Poor than Good (P=0.019; n=3/group).
HSPC composition did not differ between Good and Poor (PERMANOVA P=0.714; 0/3,140 Milo neighborhoods significant), pointing to functional rather than compositional remodeling. Poor HSC/MPP showed broad pathway remodeling (242 HSC, 179 MPP significant pathways; OXPHOS, MYC-target, E2F-target, mTORC1) covarying with four data-derived E2F1/YBX1 regulon axes, and lower external LT-HSC signature scores than Good HSC (mean difference -0.180; P=0.0286, FDR=0.086).
GMPs from Poor mobilizers instead showed higher OXPHOS, mTORC1, glycolysis, and NF-kB/TNF pathway enrichment. Among 10 evaluable cell types, only CLP and GMP retained Poor-associated genes replicated in both Poor-vs-Good and Poor-vs-HD contrasts, though sparse rather than broad: THSD7A+ GMP were less frequent in Poor (log2FC=-4.812, FDR=0.0042; 2.79% vs 31.98% median donor fraction, exact P=0.0286, FDR=0.057, LOSO-robust), while MS4A1+, B-primed CLP-like cells were more frequent (13.46% vs 1.88%; P=0.0286).
This THSD7A-low GMP state was externally supported: in GSE223972 (MM), plerixafor mobilization increased THSD7A+ GMP frequency versus G-CSF alone (9.44% vs 5.73%), and GSE285943 reproduced this multigene state in myeloid-enriched HSPCs.
In 3 B-ALL patients paired before/after plerixafor, pooled-HSPC pseudobulk GSEA showed attenuated TNF-NF-kB signaling and leukocyte adhesion regardless of response (FDR 1.71x10-11 to 2.41x10-7; concordant in all 3). The two responders showed concordant decreases across HSC/MPP regulon-pathway axes, with expansion of THSD7A+ GMPs, increased CD34+ counts, and contraction of MS4A1+ CLPs, whereas the non-responder showed discordant, non-restorative shifts across the same axes.
CONCLUSIONS
Poor mobilization affected ~20% of B-ALL patients undergoing ASCT with delayed engraftment. Poor HSC/MPP displayed coordinated metabolic and transcriptional remodeling and exploratory evidence of reduced primitive-state scores, consistent with post-transplant functional deterioration. Reduced THSD7A+ GMP frequency, external cross-dataset support, and expansion after plerixafor in responders nominate it as a hypothesis-generating marker for poor mobilizers and plerixafor response, warranting independent validation.
DISCLOSURES
Hongye Gao, MD: No relevant conflicts of interest to declare.