2. Genetics of Blastic Plasmacytoid Dendritic Cell Neoplasms
2.1. Deletions Involving Immune Genes
Interestingly, these recurrent deletions also involve genes related to the original function of pDC, i.e., immune response, especially losses of 6q23 (
IFNGR1,
TNFAIP3), 9p21.3 (cluster of
IFNA genes), and 12p13.2-p13.1 (
CLEC2B,
CLEC4C,
CLEC4E,
TNFRSF1A) [8][15]. Thus, those deletions compromise the normal function of the cells of origin of BPDCN.
2.2. Inactivation of Genes Encoding Cell-Cycle Inhibitors and Tumor Suppressor Genes
Similarly to other hematological malignancies, deletions inactivating Tumor Suppressor Genes (TSG), such as
TP53 [9][10][11][12][16,17,18,19], responsible for genetic instability, are also found in BPDCN. The tumor suppressor gene ATM, mutated in lymphoproliferative syndromes
[13][20], may also be mutated in BPDCN
[12][19], while
RB1, involved in the regulation of the G1/S cell cycle transition, is also frequently deleted
[5][10][12][14][12,17,19,21]. Initially described in retinoblastoma, in which it modeled the principle of TSG
[15][22],
RB1 is also reported in chronic lymphocytic leukemia and Acute Lymphoblastic Leukemia (ALL)
[16][23]. Additionally, the
CDKN1B/CDKN2B/CDKN2A genes have a role in the G1/S transition, and their deletions are reported in some studies
[2][5][10][12][17][9,12,17,19,24], similar to ALL
[16][23] and lymphoma
[18][25]. This alteration in cell cycle regulation could have a crucial role in the oncogenesis of BPDCN
[10][17].
2.3. Recurrent Deletions in 5q31
Deletion in 5q are particularly recurrent in BPDCN, constituting a specific defect compared to other hematological malignancies
[3][19][10,26]. In the 5q23.3 Common Deleted Region (CDR),
HINT1 was first proposed to be a key gene
[19][26]. Indeed,
HINT1 encodes a homodimeric purine phosphoramidase, suggesting a transcriptional modulatory role. Moreover,
HINT1 deficiency would impair ATM function and thus DNA repair
[19][26]. On the other hand, the 5q31 locus would also be a key region on chromosome 5, with deletions delineating a group of unfavorable prognostic impacts
[3][10]. The glucocorticoid receptor gene
NR3C1 was found to be recurrently deleted, leading to haploinsufficiency and decreased glucocorticoid receptor activity
[3][10]. The deletions impact the polycomb complex, in particular
EZH2, with dysregulation of the
HOXA locus and plasmacytoid dendritic differentiation.
2.4. Deletions of Transcription Factors
Similarly to other hematological neoplasms, transcription factor are particularly impacted in BPDCN.
ETV6 (
TEL) is frequently mutated or deleted
[1][5][8][9][10][17][8,12,15,16,17,24], contrasting with classical defects in other leukemia where translocations are more frequent, including t(12;21)(p13;q22)
ETV6::RUNX1 in B-cell Acute Lymphoblastic Leukemia (B-ALL)
[20][27], t(5;12)(q32;p13)
ETV6::PDGFRB, t(9;12)(p24;p13)
ETV6::JAK2 or t(4;12) (q12;p13)
ETV6::PDGFRA in hypereosinophilic syndromes, and other translocations in rare cases of Acute Myeloid Leukemia (AML)
[21][28]. ETV6 invalidations are, however, known in leukemia: somatic mutations of
ETV6 remain rare in AML
[21][28], but germline mutations are also possible in the context of thrombocytopenia predisposing to AML, Myelodysplastic Syndromes (MDS), Chronic MyeloMonocytic Leukemia (CMML), B-ALL, or multiple myeloma
[22][1].
The IKAROS family (
IKZF1/2/3) is also frequently deleted
[2][9][14][23][9,16,21,29], similarly to ALL
[24][30], where it compromises lymphoid differentiation
[25][31].
ZEB2 may also be altered in BPDCN
[26][32]. This transcription factor is involved in the commitment and lineage fidelity of myeloid and lymphoid cells at various stages of hematopoiesis and is thought to play a key role in the development of various types of AML, ALL, and lymphoma
[27][33].
While translocations involving transcription factors have been widely described in ALL and AML, recurrent rearrangements in BPDCN were rare before 2017. Indeed,
KMT2A (
MLL) rearrangements had previously been described in rare cases of CD4+ CD56+ neoplams identified as BPDCN (
KMT2A::ENL and
KMT2A::MLLT1)
[28][29][34,35], but these descriptions in BDPCN have been challenged because these reported cases do not fulfill the current diagnostic criteria of BPDCN. Indeed, these cases could correspond to CD4+ CD56+ AML, because they constitute a delicate differential diagnosis for BPDCN
[30][36].
2.5. Recurrent MYC Rearrangements
Initially, the translocation t(6;8)(p21;q24) was sporadically reported by several studies, using conventional karyotyping. Since 2018,
MYC rearrangements (8q24) have been largely described in approximately 30% of BPDCN
[23][31][32][33][29,37,38,39], with a more frequent immunoblastoid morphology and sometimes a CD56-negative phenotype
[31][34][37,40].
These
MYC abnormalities bring BPDCN closer to high-grade B lymphoma
[2][6][31][35][9,13,37,41], but the gene partners are really different, with specific partners that could point towards pDC differentiation. Indeed, among
MYC rearranged cases of BPDCN, Sakamoto et al. confirmed the high prevalence of the t(6;8)—detected in 56% of cases involving the
RUNX2 locus at 6p21. Interestingly, Kubota et al. showed that the t(6;8) juxtaposes the promoter of
MYC to the pDCs-specific
RUNX2 super-enhancer, leading to overexpression of
MYC. In this recurrent t(6;8)(p21;q24), both
MYC and
RUNX2 are dysregulated, and cooperate together to promote survival and proliferation of the BPDCN cells. Remarkably,
RUNX2 is physiologically involved in differentiation and migration of pDCs and plays a dominant role in controlling transcription networks in BPDCN
[36][42].
Other partners of
MYC have been sporadically reported but not clearly identified (i.e., 2p12, Xq24, 3p25, 14q32). It remains to be determined if
MYC rearrangement could constitute a primary or secondary genetic event in BPDCN. In this way, the t(6;8)(p21;q24) cannot be considered as a specific genetic abnormality of BPDCN because it has been reported in follicular lymphoma
[37][43]. Lastly, a unique study showed the adverse impact of
MYC rearrangement, and this prognostic impact still needs to be confirmed by further independent studies
[31][37].
2.6. Recurrent MYB Rearrangements
In 2017, other recurrent rearrangements were described in nine of fourteen patients, including five children
[14][21]. Remarkably, all five children included in this series had a
MYB rearrangement. Of note, the previous largest report of pediatric BPDCN cases exhibited several cases with 1q and/or 6q abnormalities, or translocation t(1;6)(q21;q23)
[38][44]. These observations reveal a striking link between pediatric BPDCN and
MYB rearrangement.
MYB rearrangements create fusion transcripts between
MYB and various partner genes (
ZFAT/8q24,
PLEKHO1/1q21,
DCPS/11q24,
miR-3134/3p25)
[14][21]. The chimeric transcripts retain the MYB transactivation domain and disrupt its negative regulatory domain, which allows the maintenance of the
MYB transcriptional activity. Indeed, functional analysis of
MYB fusions revealed the activation of
MYB target genes as a result of induced
MYB activation
[14][21].
MYB is a nuclear-localized transcriptional activator in hematopoietic cells that interacts with the C/EBP complex to stimulate the transcriptional activity of
MYC,
BCL2,
c-KIT,
c-ERBB2, and other targets (
Figure 1). Its expression progressively decreases during cell differentiation, with high activity in hematopoietic stem cells and activated T-cells.
Figure 1. Interaction network between MYB, MYC, and their transcriptional targets. MYC is a strong transcriptional activator, dependent on the intracellular phosphorylation cascade signaling pathways of Mitogen-Activated Protein Kinases (MAPK) and Extracellular signal-Regulated Kinases (ERK) (p38 MAPK and MEK/ERK pathways 1 and 2). Activation of MYC induces the transcription of numerous target genes involved in proliferation, division, metabolism, and cell motility, as well as apoptosis. MYC also inhibits microRNAs (MiR150 and MiR15a) that are capable of silencing MYB expression in the basal state. Activation of MYC therefore induces activation of MYB, involved in the core binding factor (CBF) complex with CBFB, RUNX1, and CEBPA. MYB is also a transcriptional activator recognizing multiple nucleotide sequences, in a complex with CEBP. The targeted genes are involved in survival with BCL-2 and cell proliferation with c-KIT, c-ERBB2, and especially MYC. This results in an activation loop between the two transcriptional activators: BS, MYB Binding Site with MYB Recognition Element.
In contrast to
MYC,
MYB is only exceptionally rearranged in other hematological malignancies: in fact, only in rare cases of acute basophilic leukemia with
MYB::GATA1 fusion transcripts, even rarer than BPDCN
[39][45] and in T-ALL with t(6;7) and
MYB duplication
[40][46].
MYB translocations are also reported in 60–80% of adenoid cystic carcinomas, mainly with the
MYB::NFIB fusion transcript
[41][42][47,48], and in pediatric gliomas
[43][49]. For the first time, fusion transcripts appear to be recurrent and specific for BPDCN compared to other hematological malignancies.
MYB may play a key role in the leukemic transformation process, similar to
MYB::GATA1 rearrangements in acute basophilic leukemia. However, the frequency of
MYB rearrangements is very uncertain and possibly higher in young patients, as suggested by the original study
[14][21].
Interestingly, an analog of
MYB,
MYBL1 would also be rearranged in a very similar way, with an identical functional impact
[31][37]. Finally,
MYB,
MYBL1, and
MYC defects would be mutually exclusive, delineating distinct groups of patients
[31][37].
2.7. Mutation Landscape
2.7.1. A Myeloid-like Profile
In addition to these cytogenetic defects, the mutation landscape of BPDCN has been particularly studied. However, given the rarity of BPDCN, only small cohorts have been studied so far (less than 30 cases), and usually by targeted high-throughput sequencing. Nevertheless, 22 Whole Exome Sequencing (WES) have been performed
[7][9][44][14,16,50]. On a first stratum, the mutation landscape is quite similar to myeloid neoplasms
[45][46][47][48][49][50][51,52,53,54,55,56], with a high prevalence of mutations involving epigenetics (
TET2,
ASXL1) and splicing (
ZRSR2,
SRSF2,
U2AF1)
[9][11][51][52][53][16,18,57,58,59]. These mutations rather suggest an early process before leukemic transformation, as is well described in AML, disturbing the DNA methylation balance, modifying chromatin access and the splicing processes
[54][60]. In multistage leukemogenesis models, epigenetics and splicing mutations would be present from the pre-leukemic stages
[55][61], and their frequency increases with age. Of note, mutations of
TET2 are found in 40 to 60% of cases
[9][11][53][16,18,59]. Interestingly, loss-of-function of ZRSR2 impairs pDC activation and apoptosis after inflammatory stimuli with intron retention, promoting pDC expansion. Of note, being located on the X chromosome, this enrichment of ZRSR2 in BPDCN fits well with its predominance in males
[52][58]. Although mutations of
NPM1 were initially described in BPDCN
[9][16], this has not been confirmed since, and this is not consistent with the nature of these mutations defining a mutually exclusive subtype of AML
[22][1]. In contrast, sub-clonal mutations of signaling pathways can be found in 5 to 20% of BPDCN, especially
FLT3,
KIT,
KRAS, and
NRAS mutations
[9][12][14][26][44][45][53][56][57][58][16,19,21,32,50,51,59,62,63,64]. This profile is close to that of CMML and is consistent with a common clonal origin of BPDCN and CMML cells demonstrated in a few patients suffering from the two neoplasms
[7][45][14,51]. The leukemic model would include shared epigenetic mutations, with secondary emergence of a BPDCN clone and another clone leading to CMML
[48][59][54,65] or AML
[60][61][62][63][66,67,68,69].
2.7.2. Some Lymphoid-like Features
Associated with these “myeloid-like”, key deleted transcription factors or tumor suppressor genes
IKZF1,
ETV6,
RB1,
ATM, and
TP53 can also be mutated in some cases (5–10%), also resulting in an invalidation
[2][4][6][7][11][64][65][9,11,13,14,18,70,71]. Notably, biallelic invalidations of
ETV6 argue for a primordial early event, possibly overexpressing the BPDCN oncogene
TCL1A [4][8][66][11,15,72].
IKZF1 loss-of-function, either by deletion or mutation, would lead to the increased cell interactions in BPDCN. BPDCN also exhibit
KMT2D and
SYNE1 mutations or losses, previously reported in follicular lymphoma
[6][64][13,70]. Overall, the most characteristic feature of BPDCN would be that combination of myeloid-like and lymphoid-like abnormalities (
Figure 2).
Figure 2. Genomics and transcriptional landscape of BPDCN. The genomic landscape of BPDCN include a combination of myeloid-like and lymphoid-like mutations and cytogenetic defects within a complex landscape, with frequent complex karyotypes. The transcriptional program of BPDCN is made of a diversity of original factors: RUNX2, MYB, IFN pathway, neural processes, cholesterol metabolism, corticoresistance factor, and original oncogenic factors.
2.8. Prognostic Factors
A major challenge to determine the mutation landscape in BPDCN is to establish a molecular prognostic stratification, as in AML with the
European Leukemia Network recommendation
[67][73]. Nevertheless, this point remains a tricky issue because of the rarity and diagnostic challenges of BPDCN. Some adverse prognostic factors have still been proposed, particularly mutations involving
ETV6,
TP53, or
NRAS/KRAS [9][16], as well as biallelic deletions of 9p21.3
[1][2][8,9], deletions of
NR3C1 [3][10], abnormal karyotype or numerous abnormalities detected
[68][69][74,75], and rearrangements of
MYC [31][70][37,76]. At this time, it is unfortunately impossible to integrate all of these factors into a score stratifying patients, and most of these prognostic abnormalities need to be confirmed.