Plasmacytoid dendritic cell expansion defines a distinct subset of RUNX1-mutated acute myeloid leukemia

Abstract

Plasmacytoid dendritic cells (pDCs) are the principal natural type I interferon-producing dendritic cells. Neoplastic expansion of pDCs and pDC precursors leads to blastic plasmacytoid dendritic cell neoplasm (BPDCN), and clonal expansion of mature pDCs has been described in chronic myelomonocytic leukemia. The role of pDC expansion in acute myeloid leukemia (AML) is poorly studied. Here, we characterize patients with AML with pDC expansion (pDC-AML), which we observe in ∼5% of AML cases. pDC-AMLs often possess cross-lineage antigen expression and have adverse risk stratification with poor outcome. RUNX1 mutations are the most common somatic alterations in pDC-AML (>70%) and are much more common than in AML without pDC expansion and BPDCN. We demonstrate that pDCs are clonally related to, as well as originate from, leukemic blasts in pDC-AML. We further demonstrate that leukemic blasts from RUNX1-mutated AML upregulate a pDC transcriptional program, poising the cells toward pDC differentiation and expansion. Finally, tagraxofusp, a targeted therapy directed to CD123, reduces leukemic burden and eliminates pDCs in a patient-derived xenograft model. In conclusion, pDC-AML is characterized by a high frequency of RUNX1 mutations and increased expression of a pDC transcriptional program. CD123 targeting represents a potential treatment approach for pDC-AML.

Graphical abstract

Figure 1.

Increased pDCs in a subset of AML. (A-B) Flow cytometric identification of pDCs (aqua population represents pDCs). (C) pDC proportion (percentage of white blood cells) in BM aspirates from patients with AML and normal controls (median ± IQR). AML, AML without pDC expansion; Controls, normal subjects; pDC-AML, AML with pDC expansion. (D) CD123 levels on CD34⁺ blasts and pDCs (median ± IQR). (E) Hematoxylin and eosin stain of BM biopsy from a representative patient with pDC-AML (inset: anti-CD34 immunostain). (F) Anti-CD123 immunostain of the patient from panel E. (G) Hematoxylin and eosin stain of BM biopsy from another representative patient with pDC-AML (inset: anti-CD34 immunostain). (H) Anti-CD123 immunostain of the patient in panel G. (I-J) Wright-Giemsa stain of flow-sorted leukemic blasts (I) and pDCs (J) from pDC-AML. **P < .01, ***P < .001. Original magnification, ×200 (E-H) and ×1000 (I-J). MFI, mean fluorescence intensity.

Figure 2.

RUNX1 mutations in pDC-AML. (A) Oncoplot of mutations in pDC-AML. (B) Lollipop graph of RUNX1 mutations in pDC-AML. (C) VAF of the major mutations compared with blast percentages. AML, NOS, AML not otherwise specified; AML-MRC, AML with myelodysplasia-related changes; t-AML: therapy-related AML.

Figure 3.

Leukemic blasts from pDC-AML and AML with RUNX1 mutations have greater differentiation propensity to pDCs in vitro. (A-C) The sorted leukemic blasts were cultured in vitro for 2 weeks and subjected to immunophenotyping by flow cytometry. (D) pDC proportions in the culture after 2 weeks were compared. Data are mean ± SD. (E) Time course experiments showed that a high pDC proportion persisted from the leukemic blasts of pDC-AML. Data are mean ± SEM. (F) Wright-Giemsa staining of sorted pDCs (were CD303⁺ and had bright CD123 and HLA-DR expression and were negative for CD11b, CD14, CD19, CD3, and CD56) from the in vitro culture. AML RUNX1, AML with RUNX1 mutations but no pDC expansion; AML no RUNX1, AML without RUNX1 mutations or pDC expansion. *P < .05, ***P < .001. Original magnification, ×1000 (F).

Figure 4.

Leukemic blasts differentiate into pDCs in vivo. (A-B) BM cells were harvested from the primary NSG mice receiving pDC-AML leukemic cells, as described in supplemental Figure 4 and supplemental Table 6. The unsorted BM cells, purified leukemic blasts, and purified pDCs were injected IV into secondary NSG mice. Engraftment of human CD45⁺ cells was evaluated from peripheral blood 4 weeks (A) and 8 weeks (B) after transplant. (C) pDC-AML phenotype in the secondary NSG mice receiving purified leukemic blasts. Data are mean ± SD. (D-E) Representative flow plots showing hCD45⁺ cells in the secondary NSG mice receiving purified blasts (D) or pDCs (E). (F) Representative flow plot showing pDC-AML phenotype in hCD45⁺ cells from panel D. ***P < .001.

Figure 5.

Leukemic blasts from pDC-AML upregulate a pDC transcriptional program. (A) Principal component analysis of gene expression in normal marrow CD34⁺ cells, normal pDCs, pDCs from pDC-AML, blasts from pDC-AML, blasts from AML without RUNX1 mutations or pDC expansion, and blasts from AML with RUNX1 mutations but no pDC expansion. (B) pDC transcriptional program, as evaluated by normalized single sample GSEA scores for each group. Scores are calculated based on expression levels of pDC genes. (C) Upset plot displaying overlap of upregulated genes among blasts from pDC-AML, blasts from AML without RUNX1 mutations or pDC expansion, and blasts from AML with RUNX1 mutations but no pDC expansion, pDCs from pDC-AML, and normal pDCs. All groups are compared with normal marrow CD34⁺ cells. (D) Heat map showing the 30 upregulated genes shared among blasts from pDC-AML, blasts from AML with RUNX1 mutations but no pDC expansion, and normal pDCs but not with blasts from AML without RUNX1 mutations or pDC expansion. (E) Gene ontology analysis showing upregulated pathways in blasts from pDC-AML compared with normal marrow CD34⁺ cells. (F) Expression levels of a subset of IFN-related genes upregulated in pDC-AML. These genes are also among the subset of genes depicted in panel D. **P < .01.

Figure 6.

Tagraxofusp treatment results in significant reduction of leukemic blasts and pDCs in vivo. (A-C) BM cells harvested from primary NSG mice were injected IV into secondary NSG mice. The mice were treated by intraperitoneal injection for 1 cycle with PBS, tagraxofusp 0.1 mg/kg per day, or 0.2 mg/kg per day. (A) One week after the last dose, hCD45⁺ cells were examined in peripheral blood. (B) Leukemic blasts and pDC proportions are shown in hCD45⁺ compartments. (C) CD34⁺ leukemic blast proportions in total white blood cells (WBC) were also evaluated. Data are mean ± SD. (D-F) Representative flow plots show leukemic blasts and pDCs after treatment with PBS (D), tagraxofusp 0.1 mg/kg per day (E), or tagraxofusp 0.2 mg/kg per day (F). (G) BM cells were harvested from mice treated with PBS or tagraxofusp 0.1 mg/kg per day. CD34⁺ leukemic blasts and pDCs were enumerated. **P < .01, ***P < .001.