Stem cells origin of JMML propagating cells. (A) Cell count of pd-JAO-dc from isolated subpopulations (ie, CD34⁺, CD34^–CD33^–CD14^–CD15^–, CD34^–CD33⁺CD14⁺CD15^–, and CD34^–CD33⁺CD14^–CD15⁺) on day 15 derived from 5 patients at diagnosis of JMML; ∗∗P ≤ .01. (B) pd-JAO-dc cell count on day 15 (tBM pd-JAO-dc N = 33, CD34⁺-pd-JAO-dc N = 12), (C) time of sprouting from the 3D system (tBM pd-JAO-dc N = 13, CD34⁺-pd-JAO-dc N = 5), (D) death of the pd-JAO (tBM pd-JAO-dc N = 12, CD34⁺-pd-JAO-dc N = 3), and (E) immunophenotype analysis of pd-JAO-dc obtained from total BM and CD34⁺ enrichment maintained under hypoxia. Bars are shown as mean ± SEM. ∗∗P ≤ .01; ∗∗∗∗P ≤ .0001; ns, not significant by unpaired t test with Welch's correction. (F) Portrayal of the cellular heterogeneity landscape of pd-JAO-dc whole transcriptomic profiling. The heat map shows the cell type Enrichment Score based on the xCell software results. A deconvolution approach was applied to compare the gene expression profile data of tBM and CD34⁺pd-JAO-dc maintained under normoxia and hypoxia to gene signature of pure HSCs, progenitors, and mature myeloid derived cells. (G) Kaplan-Meier estimate survival analysis for 3D systems generated from different cell sources: CD34⁺ enriched cells and tBM of patients with JMML at diagnosis in hypoxia and normoxia, patients with acute myeloid leukemia (AML) at diagnosis, tBM, and CB-34⁺ from healthy donors. Log-rank (Mantel Cox) P values were calculated across all the subtypes.