Leukemic progenitor compartment serves as a prognostic measure of cancer stemness in patients with acute myeloid leukemia

Abstract

We systematically investigate functional and molecular measures of stemness in patients with acute myeloid leukemia (AML) using a cohort of 121 individuals. We confirm that the presence of leukemic stem cells (LSCs) detected through in vivo xenograft transplantation is associated with poor survival. However, the measurement of leukemic progenitor cells (LPCs) through in vitro colony-forming assays provides an even stronger predictor of overall and event-free survival. LPCs not only capture patient-specific mutations but also retain serial re-plating ability, demonstrating their biological relevance. Notably, LPC content represents an independent prognostic factor in multivariate analyses including clinical guidelines of risk stratification. Our findings suggest that LPCs provide a robust functional measure of AML, enabling quantitative and rapid assessment of a wide range of patients. This highlights the potential of LPCs as a valuable prognostic factor in AML management.

Keywords: acute myeloid leukemia; cancer stem cell; colony forming unit; leukemic progenitor cell; leukemic stem cell; prognostic; survival; xenotransplantation.

Graphical abstract

Figure 1

Substantial variation exists between different functional and molecular measures of leukemic stemness (A) Relationships between engraftment ability, progenitor frequencies, and LSC17 scores (NanoString assay) in a subset of n = 22 patients with AML. Vertical lines indicate the mean LPC frequency for each group. (B) Correlation between CD34 expression by flow cytometry and LSC17 scores by NanoString across a subset of n = 22 patients with AML. (C) Uniform manifold approximation and projection (UMAP) dimensionality reduction plots showing gene expression profiles of individual leukemic cells from a human patient with AML (patient #2, xenografted cells that were fluorescence-activated cell sorting (FACS) purified using human markers). Red color indicates single-cell LSC17 score >0.2 (left) or CD34 transcript expression >0 (right). The threshold of 0.2 for the LSC17 score was defined based on Figure S2A. (D) Correlation between CD34 transcript expression and LSC17 scores in single human AML cells (patient #2, xenografted cells that were FACS purified using human markers). (E) Kaplan-Meier estimates of overall survival based on CD34 expression (flow cytometry), LSC17 score (NanoString assay), LSC content (xenograft assay), or LPC content (CFU assay). n = 21 patients with AML. p < 0.05 is considered statistically significant. See also Figures S1 and S2 and Tables S1–S3.

Figure 2

Leukemic progenitor assays capture patient-specific mutations and serial re-plating ability (A) Schematic outlining LPC assays performed with cells from patient #20 (results shown in B–D). Following 14 days of incubation, colonies were imaged, quantified, and plucked for DNA extraction. (B) Representative bright-field images of abnormal myeloid colonies and erythroid colonies from patient with AML #20 (scale bar, 100 μm). (C) Colony counts from LPC assays with cells from patient #20 (counts from n = 12 wells, n = 1 patient). Data are shown as mean ± SEM. (D) The variant allele frequency (VAF) of patient-specific mutations measured in DNA extracted from bulk mononuclear cells or plucked colonies, measured by ddPCR (patient #20). E, erythroid colony. Red rectangles highlight the colony that was mutant for only one of two patient-specific mutations tested. Healthy indicates healthy umbilical cord blood cells (n = 4 colonies from n = 1 patient). Error bars represent the upper and lower limits of the Poisson fractional abundance. (E) The VAF of patient-specific mutations measured in DNA extracted from bulk mononuclear cells or plucked colonies, measured by ddPCR (n = 36 colonies from n = 6 patients). Patient IDs are indicated next to graphs. Healthy indicates healthy umbilical cord blood cells. Error bars represent the upper and lower limits of the Poisson fractional abundance. (F) Schematic showing the collection of entire 1° CFU wells to be plated into 2° CFU wells to measure the serial re-plating ability of LPCs. (G) Colony counts from serial re-plating experiments outlined in (F). Colony counts are normalized to those of 1° wells. n = 3–17 wells from each of n = 5 patients. Data are shown as mean ± SEM. p < 0.05 is considered statistically significant. See also Figure S3 and Table S4.

Figure 3

AML patient survival based on functional LSC and LPC content (A) Kaplan-Meier estimates of overall survival in LSC− and LSC+ patient subsets. Plots display all n = 111 patients tested in xenograft assays (left) or a subset of these patients treated with high-intensity chemotherapy alone (palliative- and HSCT-treated patients excluded; right). (B) Simon and Makuch estimates of overall survival in LSC− and LSC+ patient subsets, with HSCT as a time-dependent variable (only including patients who were treated with high-intensity chemotherapy, n = 75). (C) Kaplan-Meier estimates of overall survival in LPC− and LPC+ patient subsets. Plots display all n = 116 patients tested in colony-forming LPC assays (left) or a subset of these patients treated with high-intensity chemotherapy alone (palliative- and HSCT-treated patients excluded; right). (D) Simon and Makuch estimates of overall survival in LPC− and LPC+ patient subsets, with HSCT as a time-dependent variable (only including patients who were treated with high-intensity chemotherapy, n = 81). p < 0.05 is considered statistically significant. See also Figure S4 and Tables S5 and S6.

Figure 4

Functional LPC frequencies represent an independent prognostic factor in human AML (A) Forest plot showing multivariate analysis of overall survival in patients with AML treated with high-intensity chemotherapy alone (n = 60 patients). Unadjusted HRs (squares) and 95% confidence intervals (horizontal lines) are shown. (B) Kaplan-Meier estimates of event-free survival in LSC− and LSC+ patient subsets (patients treated with high-intensity chemotherapy alone; n = 53). (C) Kaplan-Meier estimates of event-free survival in LPC− and LPC+ patient subsets (patients treated with high-intensity chemotherapy alone; n = 60). p < 0.05 is considered statistically significant. See also Figure S4.