Mature plasmacytoid dendritic cells associated with acute myeloid leukemia show similar genetic mutations and expression profiles to leukemia cells.

Introduction: Mature plasmacytoid dendritic cells (pDCs) proliferation associated with myeloid neoplasms (MPDMN) are recognized as a neoplasm related to fully differentiated pDCs. Although it has been reported for many years, the genomic landscape of MPDMN is poorly understood.
Methods: We reported two patients who developed acute myeloid leukemia (French-American-British M5 subtype) coexisted with immunophenotypically mature pDCs proliferation, which fit the diagnosis of MPDMN. We sorted pDCs from myeloid blasts by flow cytometry and performed whole-exome sequencing and RNA sequencing of the two cell populations, respectively.
Results: The immunophenotypes of pDCs in both patients were positive for CD123bri, HLA-DR, CD4, CD303, CD304, and negative for CD56, CD34, CD117, and TdT. The variant allele frequency of gene mutations in myeloid blasts and pDCs were similar. The expression data showed myeloid blasts clustered tightly with hematopoietic stem cells, and pDCs from patients clustered tightly with granulocyte-monocyte progenitors/common myeloid progenitor, rather than with pDCs from the GEO platform.
Conclusion: Our study suggested that pDCs derived from the leukemic clone, evidenced by a shared mutation profile and similar transcriptional signatures between pDCs and concurrent myeloid blasts.

INTRODUCTION

Plasmacytoid dendritic cells (pDCs) are a subtype of dendritic cells and rare under normal physiological conditions.[1] Neoplasms related to pDCs are currently divided into two clinically and pathologically distinct categories: mature pDCs proliferations associated with myeloid neoplasms (MPDMN) and blastic plasmacytoid dendritic cell neoplasm (BPDCN).[2] BPDCN is a rare and aggressive neoplasm that was recognized as an entity in the WHO 2016 classification of hematopoietic neoplasms and received much attention about its genetic features and clonal evolution.[2-4] On the other hand, MPDMN, although not recognized in the WHO classification, has been reported for decades in coexistence with myeloid neoplasms, most often with chronic myelomonocytic leukemia (CMML) or, less often, with myelodysplastic syndrome (MDS) and acute myeloid leukemia (AML).[2]

The neoplastic nature of pDCs in MPDMN was first reported in a series of cases by fluorescence in situ hybridization (FISH), showing the same chromosomal abnormalities to myeloid neoplasms cells.[5,6] After that, Lucas et al demonstrated in CMML patients, that the clonal expansion of mature pDCs and monocytes shared identical somatic gene mutations in the RAS pathway.[7] In AML, the genomic landscape of pDCs from AML patients was poorly studied until two recent studies defined AML patients with pDC expansion (pDC-AML) as a distinct subset of AML, characterized by a high frequency of RUNX1 mutations.[8,9] However, there is still a paucity of in-depth research on the genomic landscape and clonal origin of MPDMN in AML.

Here, we describe two AML patients with mature pDCs proliferations diagnosed by flow cytometry (FCM) in our center. We performed detailed immunophenotypic, genetic, and transcriptional characterization of these two patients and revealed common mutation profiles and transcriptional signatures in this rare and distinctive patient cohort.

METHODS

Patients

Case 1 was a previously healthy 33-year-old male who was admitted to the hospital with fever and fatigue. Physical examination showed anemic appearance and sternal pain, without splenomegaly, lymphadenomegaly, or skin involvement. A complete blood count (CBC) revealed a white blood cell (WBC) count of 3.73 × 109/L with 13% circulating blasts, hemoglobin 90 g/L, and a platelet count of 189 × 109/L. The morphological examination of bone marrow (BM) smears revealed 76% agranular blast cells of medium and large size (Supplementary Figure S1A). Cytochemistry showed the blasts to be negative for myeloperoxidase and positive for non-specific esterase, which confirmed monocytic differentiation.

Case 2 was a 71-year-old male who was admitted to the hospital due to skin purpura. The CBC showed hyperleukocytosis (WBC 28.59 × 109/L), moderate anemia (hemoglobin 79 g/L), and severe thrombocytopenia (18 × 109/L). The morphological examination of BM smears showed 56% round-like shape blast cells in different sizes with pink granules (Supplementary Figure S1B). Cytochemistry results showed the blasts to be negative for myeloperoxidase and positive for non-specific esterase, which was the same as case 1.

Both patients were diagnosed as French-American-British (FAB) subtype M5 with mature pDCs proliferations. Conventional karyotyping for both patients revealed a normal karyotype. No positive results were found in leukemia-associated fusion gene screening panel (including AML1-ETO, PML-RARα, and CBFβ-MYH11, etc. Supplementary file 1) or interphase FISH for MLL rearrangements, BCR/ABL, -5/5q-, -7/7q-, and trisomy 8. Targeted next-generation sequencing (NGS) including 137 clinically relevant genes (Supplementary file 2) were conducted on both patients. The samples of two patients were collected after written informed consent was obtained. The study was approved by the local ethics committee.

Treatments

Case 1 received induction chemotherapy with conventional DA 3 + 7 (daunorubicin 60 mg/m2/day for 3 days, cytarabine 100 mg/m2/day continuous infusion for 7 days), but the patient still showed 70% of blasts in the BM on day 28, indicating no remission (NR). Subsequently, the patient was administered with an IAC regimen (idarubicin 10 mg/m2/day for 3 days, cytarabine 100 mg/m2/day continuous infusion for 7 days, and cyclophosphamide 350 mg/m2/day for day 2 and 5) as the reinduction chemotherapy. Unfortunately, complete remission (CR) was still not achieved. The patient was then started on a salvage regimen consisting of 5-azacytidine (AZA), BCL2-inhibitor venetoclax, homoharringtonine, and cytarabine (AZA 100 mg/day for 9 days, venetoclax 100 mg/day in combination with the strong CYP3A4 inhibitor posaconazole for 28 days, homoharringtonine 2 mg/m2/day for 6 days, and cytarabine 100 mg/m2/day continuous infusion for 6 days). After the third cycle of chemotherapy, the patient achieved morphological CR with negative minimal residual disease by FCM. FCM did not detect pDCs after CR was achieved. Then he subsequently underwent consolidation chemotherapies and was preparing for haploidentical BM transplantation.

Case 2 was treated with venetoclax in combination with AZA as an induction regimen. Venetoclax was given at a dose of 100 mg daily in combination with posaconazole for 28 days and AZA was given at the dose of 100 mg/day for 9 days per cycle. The patient did not reach CR after one cycle and is currently receiving the second cycle. pDCs remained detectable by FCM in this case.

The successful AML therapy led to the elimination of the pDCs component in our case 1, while pDCs persisted after the failure of induction therapy in our case 2. The successful eradication of pDCs depends on the effective treatment of AML. The characteristic of therapeutic response was consistent with those reported in the literature.[2,10]

FCM analysis and cell sorting

We used FACS Canto II (BD Biosciences, San Jose, CA) and FACS AriaIII (BD Biosciences, San Jose, CA) to analyze and sort different cell groups, respectively. The BM cells harvested from patients were stained by standard protocols with the following antibodies: anti-CD45 v500, anti-CD34 Per-CPcy5.5, anti-CD117 PE, anti-HLA-DR APC-Cy7, anti-CD123 PE-Cy7, anti-CD303 PE, anti-CD304 Per-CPcy5.5, anti-CD38 PE-Cy7, anti-CD13 APC, anti-CD16 FITC, anti-CD33 APC, anti-CD15 V450, anti-CD11b BV421, anti-CD64 FITC, anti-CD56 Percp-cy5.5, anti-CD36 FITC, anti-CD20 FITC, anti-CD71 FITC, anti-CD4 PE-Cy7, anti-CD10 PE, anti-CD38 BV421, anti-cCD79a APC, anti-MPO PE, anti-CD5 Pcy5.5, anti-cCD3 Pacific Blue, anti-CD2 PE-Cy7, anti-CD14 APC-H7, anti-CD19 Pcy5.5, anti-TDT FITC, anti-CD5 Pcy5.5, anti-CD300e PE, anti-CD15 Pacific Blue, and anti-CD33 APC. By FCM analysis, the myeloid blasts and pDCs displayed an expression profile of CD123+HLA-DR+CD34+CD117+CD303-CD304- and CD123bri+HLA-DR+CD34-CD117-CD303+CD304+, respectively.

Cells were sorted according to their surface markers by FCM. After excluding debris and doublets, an initial living cells gate was drawn on a traditional FFS/SSC gate. For case 1, based on the CD45/SSC gating strategy, the lymphocyte region (green), granulocyte region (blue), and blast region (red and black) were gated (Supplementary Figure S2A). Then among the HLA-DR+ blast cells, CD123bri+CD117-CD34- cells were sorted as pDCs (black in supplementary Figure S2A). Other HLA-DR+ cells (CD123+CD117+CD34+) were sorted as myeloid blasts (red in supplementary Figure S2A). For case 2, based on the CD45/SSC gating strategy, the lymphocyte region (green), granulocyte region (green), nucleated erythrocyte region (pink), and blast region were gated (Supplementary Figure S2B). Among the HLA-DR+ blast cells, CD123bri+CD34- cells were sorted as pDCs (black in supplementary Figure S2B). Other CD123+CD34+ cells were sorted as myeloid blasts (red in supplementary Figure S2B). HLA-DR+CD33bri+CD11b+ cells were identified and sorted as promonocytes (dark blue in supplementary Figure S2B).

DNA extraction and WES analysis

Genomic DNA was extracted from patients’ myeloid blasts, promonocytes, and pDCs sorted by FCM, respectively. Hair follicle samples were collected and used as germline controls to identify somatic mutations. DNA concentration was measured by Qubit® DNA Assay Kit in Qubit® 2.0 Flurometer (Invitrogen, USA). Sequencing libraries were generated using Agilent SureSelect Human All Exon V6 kit (Agilent Technologies, CA, USA) following the manufacturer's recommendations and index codes were added to each sample. The clustering of the index-coded samples was performed on a cBot Cluster Generation System using Hiseq PE Cluster Kit (Illumina) according to the manufacturer's instructions. After cluster generation, the DNA libraries were sequenced on Illumina platform (Illumina, San Diego, CA) and 150 bp paired-end reads were generated. Sequence alignment and variant calling were performed using the Genomon pipeline. All the downstream bioinformatics analyses were based on high-quality clean data after several steps of data processing. Valid sequencing data were mapped to the reference human genome (UCSC hg19) by Burrows–Wheeler Aligner software to get the original mapping results stored in BAM format. Samtools mpileup and bcftools were used to do the variant calling and identify SNPs and InDels.

RNA sequencing and Heatmap

Total RNA was extracted from patients’ myeloid blasts, promonocytes, and pDCs sorted by FCM, respectively. Sequencing libraries were generated using NEBNext® UltraTM RNA Library Prep Kit for Illumina® (NEB, USA) following the manufacturer's recommendations. The clustering of the index-coded samples was performed on a cBot Cluster Generation System using TruSeq PE Cluster Kit v3-cBot-HS (Illumia) according to the manufacturer's instructions. After cluster generation, the library preparations were sequenced on an Illumina platform and 150 bp paired-end reads were generated. RNA-seq data analysis was performed with a custom pipeline using Hisat2 v2.0.5 for read alignment, htseq-count for read counting, and edgeR package for differential gene expression analysis. The clustering heat map was drawn with an R-package pheatmap.

RESULTS

The immunophenotypes of leukemic cells and pDCs

By FCM analysis, case 1 had a major myeloblast population (60%, CD34bri+/CD117+/HLA-DR+/CD38+/CD13+/partially CD7+/partially CD36+/partially TdT+/CD33dim+/CD123dim+) as well as a small subset of pDC population (15%) (Fig. 1A). Case 2 had 9.6% myeloid blasts (CD34bri+/CD38bri+/CD13+/CD33+/CD117+/HLA-DR+/CD123+/partially cCD79a+/partially TdT) and 20.4% promonocytes (CD33bri+/CD64bri+/CD13+/CD11b+/CD123+/HLA-DR+/CD38+) as well as 8.61% pDCs. The immunophenotypes of pDCs in both patients was positive for CD123bri, HLA-DR, CD4, CD303, CD304, and negative for CD56, CD34, CD117, and TdT (Fig. 1B), which conform to the diagnosis of MPDMN.[2,10,11]

Figure 1

Immunophenotypic features of two cases. (A) Flow cytometry histograms of case 1 showed the predominant myeloid blast population (red) with CD34bri+CD117+HLA-DR+CD38+CD13+CD33dim+CD123dim+ immunophenotype and PDCs subpopulation (black) with CD123bri+HLA-DR+CD303+CD304+CD4+CD56-CD34-TdT- immunophenotype. (B) Flow cytometry histograms of case 2 showed 9.6% myeloid blasts (red) with CD34bri+CD38bri+CD117+HLA-DR+CD13+CD33+CD123+ immunophenotype, 20.4% promonocytes (blue) with CD34bri+CD64bri+CD13+CD11b+HLA-DR+CD38+CD123+ immunophenotype and PDCs (purple) with CD123bri+HLA-DR+CD303+CD304+CD4+CD56-CD34-TdT- immunophenotype.

The mutation profiles of myeloid blasts and pDCs

NGS detected BCOR p.C254Vfs∗12 mutation with variant allele frequency (VAF) of 84.8%, RUNX1 p.D137 E138insG mutation with VAF of 25.3%, and RUNX1 p.W106C mutation with VAF of 50.0% in case 1. For case 2, DNMT3A p.R882H mutation with VAF of 46.40%, NRAS p.G13R mutation with VAF of 2.9%, RUNX1 pS265Efs∗290 mutation with VAF of 53.6% were observed by NGS. Capillary electrophoresis discovered both patients carried an FLT3-ITD mutation with an allele ratio of 0.13 and 0.33, respectively.

In the WES analysis for case 1, we detected 25 somatic mutations (21 SNVs, 3 deletions, and 1 insertion) in the sorted myeloid blasts, 21 somatic mutations (17 SNVs, 3 deletions, and 1 insertion) in the sorted pDCs and 32 somatic mutations (28 SNVs, 3 deletions, and 1 insertion) in the unsorted BM. For case 2, we detected 23 somatic mutations (22 SNVs and 1 insertion) in the sorted myeloid blasts, 22 somatic mutations (21 SNVs and 1 insertion) in the sorted pDCs, 23 somatic mutation (22 SNVs and 1 insertion) in the sorted promonocytes, and 26 somatic mutations (25 SNVs and 1 insertion) in the unsorted BM. Among them, shared gene mutations were observed in myeloid blasts, pDCs, and promonocytes at the same nucleotide in both cases (Supplementary Table). For case 1, identical mutations of RUNX1, BCOR, BCORL1, and other genes were detected in both sorted pDCs and myeloid blasts. For case 2, identical mutations of FLT3, EP300, DNMT3A, RUNX1, and other genes were detected in sorted pDCs, myeloid blasts, and promonocytes. The detailed list of gene mutations is shown in Supplementary Table . Irrespective of the silent mutations, the similarities of VAF of gene mutations between the myeloid blasts (including the promonocytes from case 2) and pDCs indicated that they shared the same clonal origin (Fig. 2A–C), which suggested that pDCs originated from and were a part of the malignant clone.

Figure 2

Scatter plot of VAFs and heatmaps of expression profiles from different cell populations. (A) Myeloid blasts and pDCs from case 1 showed similar VAF of gene mutations. (B) Myeloid blasts and pDCs from case 2 showed similar VAF of gene mutations. (C) Promonocytes and pDCs from case 2 showed similar VAF of gene mutations. (D-E) The expression levels of genes are indicated by the color bar: red indicates higher levels and blue lower levels. Myeloid blasts from case 1 and case 2 clustered tightly with HSC. pDCs from case 1 and case 2, promonocytes from case 2 clustered tightly with GMP/CMP. CLP = common lymphoid progenitor, CMP = common myeloid progenitor, GMP = granulocyte-monocyte progenitor, HSC = hematopoietic stem cell, MEP = megakaryocyte-erythrocyte progenitors, MPP = multipotent progenitor, NK = natural killer, PBMC = peripheral blood mononuclear cell.

Transcriptional profiling of myeloid blasts and pDCs

Finally, we compared the gene expression profiles of myeloid blasts and pDCs in both patients to normal hematopoietic cells, including hematopoietic stem cells (HSC), granulocyte-monocyte progenitors (GMP), common myeloid progenitor (CMP) and dendritic cells, etc (Fig. 2D, E). The expression data of other hematopoietic cells come from the GEO platform (https://www.geoplatform.gov/). Gene expression profiling revealed 547 upregulated genes and 713 downregulated genes between myeloid blasts and pDCs from case 1, 1891 upregulated genes and 1866 downregulated genes between myeloid blasts and pDCs from case 2, respectively (Fig. 3). Myeloid blasts clustered tightly with HSC, pDCs from patients clustered tightly with GMP/CMP, rather than with pDCs from the GEO platform, suggesting that pDCs from AML patients presented a similar immunophenotype to normal pDCs, but were essentially myeloid cells based on gene expression profiles.

Figure 3

Volcano plots displaying differential expressed genes between myeloid blasts and pDCs from case 1 (A) and case 2 (B), respectively. The y-axis corresponds to the mean expression value of log 10 (P value), and the x-axis displays the log2 fold change value. The red dots represent the up-regulated expressed transcripts (P < .05); the green dots represent the down-regulated expressed transcripts (P < .05).

DISCUSSION

MPDMN, albeit a rare and elusive entity[2] has been observed and reported in the setting of myeloid disorders for many years.[5,6,12,13]. Due to the immunophenotypic similarity to normal dendritic cells, the pDCs used to be considered non-malignant or reactive.[2,10] With the development of detection technology, researchers gradually realized that pDCs originated from malignant clone.[7-9] Two recent excellent studies[8,9] defined a subset of RUNX1 mutated AML with pDCs proliferations (pDC-AML). Neither study diagnosed pDC-AML as MPDMN due to CD34 expression on pDCs, which indicated an immature status of pDCs. Mutation analysis showed shared genetic abnormalities between sorted pDCs and myeloid blasts.

In our study, we reported two patients who developed AML (FAB M5 subtype) with coexisting immunophenotypically mature pDCs proliferation, which fit the diagnosis of MPDMN. We sorted pDCs from myeloid blasts by FCM and performed WES and RNA-seq of the two cell populations. We not only proved that pDCs and concurrent myeloid blasts share the same mutation but also showed that these two groups of cells exhibited similar gene expression profiles. pDCs were confirmed to be derived from the leukemic clone based on both genomics and transcriptomics. Our data, along with those from other groups, demonstrated that pDCs at different stages of maturation from AML patients exhibited the same nature of origin. Interestingly, consistent with recent studies which reported a high incidence of RUNX1 mutation[8,9] and FLT3-ITD mutaion[14] in pDC-AML patients, our two patients also carried both RUNX1 and FLT3-ITD mutations..Therefore, we speculate that FLT3-ITD mutation may also be a characteristic mutation in these patients, especially in patients with FAB M5 subtype. Moreover, neither patient was sensitive to regular chemotherapy in our study. Tagraxofusp, an FDA-approved CD123-targeting agent has shown therapeutic potential in a patient-derived xenograft model.[9] Whether MPDMN patients will benefit from the CD123-targeting treatment remains unknown and warrants further study.

To conclude, we demonstrated molecular evidence that mature pDCs phenotypic cells in AML patients were myeloid cells and a part of the leukemic clone, by analyzing genetic mutations and transcriptional expression profiles of pDCs and myeloid blasts. The disadvantage of this study is the lack of research on pDCs function in vitro and the limited number of cases. We can only look forward to further research in patients with a similar diagnosis in the future.