Although majority of pediatric acute myeloid leukemia (pAML) undergo complete remission (CR) after chemotherapy, more than 30% of which eventually relapse, leading to a dismal outcome. Chemo-resistant cells at CR, measured as minimal residual disease (MRD), are thought to be origin for relapse. Although more sensitive detection of MRD has been achieved through genetic mutations, the pathological and clinical significance for most of the detected mutations remain unknown, largely due to a lack of understanding for the biological context where the mutation resides. The idea of retention of leukemia stem cells (LSCs) as a mechanism of chemo-resistance has not been demonstrated post-therapy in patients. Here we applied single cell RNA sequencing (scRNA-seq) on 14 pAML who met the criteria for clinical CR (n=11) and partial remission (PR, n=3) to determine the cellular heterogeneity and cancerous feature of the residual cells that can survive chemotherapy at remission. These patients carried common genetic mutations such as AML1-ETO, CBFB-MYH11 and FLT3-ITD, which represents genetically diverse WHO subtypes.

To maximize the power to detect tumor cells that can survive chemotherapy, we used 54 unsorted total bone marrow (BM) and/or peripheral blood (PB) nucleated cells collected at diagnosis and Day 26 (D26) of the first cycle of chemotherapy for 10X genomics' scRNA-seq. Cells from each pAML were compiled into one UMAP together with normal reference using unsupervised clustering to distinguish tumor clusters from normal. Clusters were defined to be tumor if patient's diagnosis contributed at least 80% of the cells, and were confirmed by the presence of somatic mutations and/or known AML mRNA expression signature associated with chromosome translocations. By projecting onto the closest normal hematopoietic cells based on transcription features, the tumor cells were classified as one of the 12 cell types (HSC/MPP, LMPP, GMP, MEP, E/B/M, CLP, classical/nonclassical Monocyte, cDC, pDC, and pre/inmature Neutrophil like cells). Consistently with findings in adult AML, cell populations were heterogeneous at diagnosis with 5-9 distinct clusters. Interestingly, majority (9/11) of the patients had 1-6 tumor clusters detected (mean 52 cells per cluster) at D26 post-chemotherapy. These D26 residual tumor cells possessed mutations originally detected by genomic sequencing and/or known AML signatures, and consistently clustered with tumor cells from diagnosis. These residual tumor cells accounted for average 1.4% of total BM cells at D26, while the morphological examination and flow cytometry analysis of MRD showed average 4.0% and 0.71% of tumor cells.

To further evaluate the chemo-resistant potential of identified residual tumor cells, we focused on three distinct features known to be associated with chemo-resistance in mouse models, including LSC activity, active oxidative phosphorylation (OXPHOS) or leukemic-regenerating cell (LRC) state. Among the total 18 residual tumor clusters detected at D26, 33.3% (6/18) exhibited expression signatures associated with at least one chemo-resistant features. The remaining clusters consisted more differentiated progenitor/monocyte-like cells. Specifically, three patients (1 PR and 2 CR) had the HSC/MPP or LMPP-like clusters possessing strong LSC and OXPHOS-associated signatures. The remaining one patient had cDC-like cluster expressing reported LRC signature. Importantly, all these four patients had either unfavorable cytogenetics or persistence of driver mutations detected by PCR.

Taken together, these data showed that pediatric AMLs represented heterogeneous populations at both diagnosis and remission. Among the residual tumor clusters that survived chemotherapy, a small fraction (6/18) were HSC/MPP-like, LMPP-like and cDC-like cells with known chemo-resistant expression features. These findings provide the first in vivo characterization of cellular heterogeneity in chemo-treated pAML with complete remission. Further studies are needed to determine the molecular characteristics of these residual cells that may convey chemo-resistance and to determine whether the presence of these cells are associated with increased risk of relapse.

Disclosures

No relevant conflicts of interest to declare.

Author notes

*

Asterisk with author names denotes non-ASH members.

Sign in via your Institution