HMGA2^high AML share an immature transcriptional signature. (A) Dot plot representation of HMGA2 and CD34 expression assessed by total RNA-sequencing in the Leucegene cohort of 452 primary AML specimens. Log-transformed scale is used to better visualize AMLs with low expression of these markers. Extreme AML groups of HMGA2 expression are depicted, and dotted line defines HMGA2 positivity threshold defined in our previous prognostic paper.¹⁶  (B) Gene Set Enrichment Analysis plot comparing HMGA2^high (RPKM >2; n = 39) vs HMGA2 null (RPKM = 0; n = 83) transcriptomic signatures in primary AML specimens. Results obtained for EPPERT_HSC_R gene set that includes upregulated genes in HSC-enriched populations. (C) MDS constructed using expression data (100 genes presenting the largest standard deviations between samples, excluding HMGA2) obtained from inv(16), CK, MLL-rearranged (MLLr), t(15;17), t(8;21), NPM1 mutated (N), and other subgroups, presenting either high (>2 RPKM) or null HMGA2 expression. The subgroup denominated as “others” is composed of samples presenting none of the alterations defining the plotted groups. (D) Same analysis as in panel C with the exclusion of CK AML specimens. (E) Single-cell transcriptomic overview obtained from the integration of 5 primary AML specimen data sets from Petti et al.²⁹  Only leukemic cells, defined by the presence of at least 1 somatic mutation, are displayed. Black arrow: non-cycling HMGA2+ cells, red arrow: cycling HMGA2+ cells. Cell type annotations were adopted as published: HSC, megakaryocyte-erythroid progenitors (MEP), common myeloid progenitor (CMP), natural killer (NK), and natural killer T (NKT) (left). Uniform Manifold Approximation and Projection (UMAP) plot of HMGA2, stem cell, and proliferative genes (right). FDR, false discovery rate; NES, normalized enrichment score.