Proteomics and Single-Cell Transcriptomics Reveal Differentiation Patterns of pDC-like LSCs and Poor Prognosis in AML with pDC Expansion

Background: Acute myeloid leukemia (AML) is the most common leukemia in adults with poor survival prospects, characterized by extensive heterogeneity. Recent studies have recognized that a subset of AML patients can have clonally expanded pDCs (pDC-AML). However, the heterogeneity and complex microenvironment of AML have made understanding these pDCs challenging in previous research. This study presents the first multi-omics analysis of pDC-AML, exploring its proteomic, transcriptional, and potential therapeutic strategies.

Method: Between July 2014 and October 2023, 104 patients were enrolled: 87 AML (69 non-pDC-AML, 18 pDC-AML) and 17 BPDCN. Nano-flow proteomics was performed on bone marrow (BM) samples of 6 pDC-AML, 6 non-pDC-AML, and 9 healthy donors (HDs). 10X Genomics scRNA-seq and paired scTCR and scBCR were conducted on BM samples of 8 pDC-AML, 5 non-pDC-AML, and 1 BPDCN patient. Additionally, scRNA-seq BM samples from 5 BPDCN, 5 non-pDC-AML, and 5 HDs were included from public databases (GSE185381 and GSE189431). Follow-up continued until December 25, 2023. The study adhered to the Declaration of Helsinki and was approved by the Medical Ethics Committee of Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology.

Result: The median pDC proportion in BM of pDC-AML patients was 7.8% (range 2.5-52.0%). The median OS and PFS of pDC-AML patients were shorter than non-pDC-AML patients (11.0 vs 20.0 months for OS, and 7.5 vs 11.0 months for PFS, p < 0.001). NGS identified more frequent RUNX1, NF1, and IDH2 mutations in pDC-AML patients. Flow cytometry showed absent CD56 and high CD34 and CD13 expression in pDCs of pDC-AML compared to blastic pDCs of BPDCN. Proteomic analysis revealed distinct protein patterns in pDC-AML, with upregulated ER stress pathways (GRP78, HSPA9, etc.). scRNA-seq showed increased ER stress and UPR pathways (GRP78 and XBP1.) in LSCs of pDC-AML, and upregulated IFN-α response pathways promoting pDC differentiation. Additionally, these LSCs also exhibited upregulated phenotypes related to pDCs, including cell adhesion, migration, and MHC antigen presentation. Interestingly, in pDC-AML patients, we noticed a transitional state of LSC subcluster from LSCs to pDCs, present in 7 out of 8 pDC-AML patients, but absent in all non-pDC-AML patients. Cellular differentiation states inferred using CytoTRACE indicated a low differentiated state in this cluster and survival analysis showed TCGA-LAML cohorts with a high presence of this LSC subcluster had significantly shorter OS (p < 0.01). Furthermore, in pDCs from pDC-AML patients, reduced physiological pDC functions pathways scores were observed, including type I interferon production, T cell activation, inflammatory response and antigen presentation. Additionally, scTCR-seq and scBCR-seq showed decreased clonal diversity in T and B cells, indicating a low immune response in pDC-AML patients' bone marrow.

To gain further insights into the drug resistance pathways of pDC-AML, we performed sequential scRNA-seq in pDC-AML patients, and we noticed in an MRD+ patient, after the first cycle of induction chemotherapy, abnormal pDCs were persisted, showed upregulated ERS pathway, and high presence of these residual pDCs were correlating with worse OS in TCGA-LAML cohorts (p < 0.01).

Conclusions: This study represents, at first time, the multi-omics landscape of pDC-AML patients, providing evidence that AML patients with clonally expanded pDCs form a potential subtype characterized by unique clinical characteristics, as well as distinct proteomic and transcriptomic features. Notably, it includes a specific pDC-like LSC cluster, upregulated ERS pathways and dysfunctional T and B cell state, which may be potential targets for therapeutic intervention.