Page 44 - 2021_12-Haematologica-web

L. Zalmai et al.
mutated cases involved mainly males (M/F=8/3), with a median age of 70 years (range, 55-87 years).
Discussion
The World Health Organization 2017 classification clear- ly recognizes BPDCN as a form of pDC neoplasm, while MPDCP is still insufficiently described and probably under- diagnosed8. MPDCP mainly concerns CMML and more rarely MDS or AML. Our study focused on pDC-AML, using a phenotypic and mutation characterization on sorted population of 15 cases. Our gateway for the inclusion of cases is the pDC-like morphology of some blastic cells detected by cytologists in French hospitals. This point pre- cludes us from determining the prevalence of pDC-AML among AML, due to a recruitment bias. We highlight in this study an excess of pDC in a context of AML in order to dif- ferentiate them from BPDCN and to precisely dtermine their molecular profile, not well-described so far. The clini- cal presentation of our cohort was close to that of BPDCN, with elderly patients and a clear male predominance, but skin lesions were less frequent (25% of cases vs. 90% in BPDCN), linked to either pDC (two cases) or myeloblast (two cases) infiltration. The excess of pDC in BM aspiration of pDC-AML was well over the pDC infiltrate in healthy donor BM and PB, considered to be below 1% of total nucleated cells24,26,27 and confirmed by our data. Moreover, pDC are frequently gathered on BM smears, compatible with aggregates and in the only case of BM biopsy, clus- tered nodules of pDC were present, as reported by pathol- ogists with islands of pDC in MPDCP.13,21,24 In the two cases with prior history of CMML, pDC were not in high excess prior transformation on BM smears. However, in the absence of available sample, flow cytometry assays were not assessed at the CMML stage.
Blasts exhibited an immature phenotype with expres- sion of CD34, CD117 and TdT, which distinguishes these cases from BPDCN. They also frequently express myeloid markers, and the expression of CD123 is definitely weak- er than pDC from the same samples or pDC blasts from BPDCN. Of note, strong pDC or cDC lineage markers (CD303, CD304, cTCL1, ILT7 and CD1c, CD141, CD11c) were never expressed on blasts. Focusing on the 11 M0- AML cases, blasts are immature with markers of commit- ment to the myeloid lineage (CD13, CD33 without MPO) plus expression of lymphoid markers in some cases; this could be a sign of an original progenitor origin or onco- genic dysregulation in pDC-AML compared to other AML.
Only four cases of pDC-AML showed higher infiltration of pDC than blasts: three at diagnosis (N2, N8 and N34) and the fourth at day 81 post-induction (N35). In these 15 cases, the phenotype of pDC is different from pDC in healthy donors and in BPDCN. CD56, is not expressed in our 15 cases, similarly to physiological mature pDC, only CD56+ when stimulated by Flt3L during maturation.28,29 Conversely CD56 is almost systematically expressed in BPDCN5,30,31 except for rare cases.7,32,33 Moreover, the pDC in pDC-AML always expressed CD303 (100%) while 20 to 30% of BPDCN are CD303- 6,31 and BPDCN express a lower intensity of CD303 than pDC of pDC- AML (data not shown); cTCL-1 expression was substantial- ly lower on pDC of pDC-AML compared to pDC of BPDCN and even pDC of healthy donors (Figure 3B).
Finally, CD34 is sometimes expressed on pDC (33% of cases), whereas CD34 is almost never expressed on BPDCN blasts. This phenotype is closer to normal pDC in some respects (CD56–, CD303+, cTCL1+low). pDC lineage maturation is divided in three stages with progressive acquisition of CD303 and CD304 plus downregulation of CD34 and CD117 and gradual loss of CD13, CD33 or CD22 expression34. In our series, the expression of CD34 (six cases) and CD13, CD33 or CD22 (three cases) makes these pDC closer to the intermediate stage of pDC matu- ration (CD34+, CD303+, CD304+/-, CD123+). Moreover, aberrant markers of myeloid or lymphoid lineage (CD13, CD33, CD11b, CD22, CD7, CD5) on both blasts and pDC in the same patient support the idea of a common origin of these populations, with maturation of CD34+ blasts towards the pDC lineage by downregulation of CD34 and upregulation of CD123, CD303 and CD304. Reinforcing this hypothesis, we clearly show a maturation continuum between blasts and pDC or monocytes (Figure 1D) with CD34 downregulation and upregulation of CD45, pDC markers (CD123, CD304, CD303) on the pDC cells or CD14, CD64 on monocytes, as recently high- lighted.14,15 Another study also described similar cases of leukemia associating immature CD34+ myeloid blasts and CD34+/- CD56– pDC, but considered these cases as an immature subgroup of BPDCN.35 Altogether, a variety of teams described similar cases under different denomina- tions: MPDCP with myeloid neoplasm (MN), pDC prolif- erations associated with MN, AML/MN with PDC differ- entiation, AML/MDS-pDC, leukemia of ambiguous line- age, or immature group 1 of BPDCN.14–17,22,23,35,36 There is a need to refine this poorly defined MPDCP, where the “mature” denomination should be omitted because it only refers to the absence of the blastic morphology of BPDCN.8
Our data and others highlight that pDC are present at all stages of maturation in pDC-AML, even CD34+, because they keep their potential for maturation14,15 as already described in pDC-CMML.24 In addition, pDC-AML appear to be quite similar to pPDC-CMML, as clonal monocytes are also frequently detected in our cohort (14 of 15 AML). In this regard, all cases of MPDCP described so far fall within our definition of pDC- AML/CMML/MDS. Interestingly, the expression of CD123 on both blasts and pDC opens up the potential to use new therapies targeting CD123, such as tagraxofusp, IMGN632 or CD123 CAR-T cells.37-40 Crucially, these pDC-AML may respond differently from other AML or BPDCN.
Very few molecular data have been obtained to assert the clonal link between pDC and leukemic cells in CMML and AML.21-23 In our series, using sorted cell populations, we show that the same mutations are shared by blasts, pDC, monocytes and even cDC, thus confirming the neo- plastic origin of pDC and their shared clonal origin with blasts, monocytes and cDC (for one tested case). Of note, no unique mutation was identified in pDC but not in the blast compartment, or vice versa, albeit with the limitation that only 70 genes were tested. Thus, no specific mutation arising during the lineage commitment was highlighted. Blasts seem to be broadly blocked in an immature undif- ferentiated stage in pDC-AML compared to pDC-CMML, and maintain their ability to mature towards pDC, mono- cytes or even cDC (Figure 5A and B). The immature CD34+ blasts could be a proliferation, at the very least, of
3064
haematologica | 2021; 106(12)