Erythroleukemia: Classification

Abstract Acute erythroid leukemia (AEL) is a rare (2%–5%) form of acute myeloid leukemia (AML). Molecular alterations found in AEL resemble those of other AMLs. We report a classification of AELs in three major classes, with different prognosis and some specific features such as a tendency to mutual exclusion of mutations in epigenetic regulators and signaling genes.

Molecular characterization has shown that AELs from adult patients have more or less the same mutations as the other AMLs [4][5][6][7]. In 2016, we proposed a molecular classification of M6a-AELs into three main classes, NPM1-mutated, secondary-like, andTP53-mutated, and an additional fourth class for some cases that could not be included [5]. We also suggested that the erythroid phenotype could be due to specific additional mutations such as alterations in the EPOR/JAK2 pathway [8], a different cell-of-origin or both [9]. Thus, while AEL pathogenesis remains unclear, it may be linked to cellular pliancy.

F I G U R E 1
Schematic representation of the combination of the main driver mutations occurring in acute erythroid leukemia (AEL). AEL cases are classified into three major classes according to alterations in driver genes (listed at the right and in Table S2) grouped according to their function.
from the following series: (1) AELs, we had previously analyzed by targeted next generation sequencing (tNGS) and array-comparative genome hybridization (aCGH) [5] and five new cases (Table S1) (2) AELs reported in a recent study by Takeda [10] analyzed by whole exome sequencing and/or whole genome sequencing. Informed consent was obtained for all samples. To compare and pool the results on our samples (n = 55, HD) and Takeda's (n = 111, UPN), the data were analyzed for the same short list of driver genes (Table S2), that is, genes whose alterations are regularly found in AML and considered as driver genes of the disease.
Results are reported in Figure 1 and Table S3. Although few cases (not included), which did not present any mutations in the driver genes genes define the class of secondary AML [11] or, as recently designated the class of AML with myelodysplasia-related gene mutations [3,12].
Actually, class 3 AELs may be secondary too, including to treatment for a previous disease [13].
Some features of this classification can be noticed (Table S3). 5. It has been previously reported that pure AELs harbor more than one TP53 alteration [10,14,15]. This applies to nonpure AEL as well. About half of the class 3 cases of our series showed two TP53 alterations.
6. In class 2, prevalence of cases in men (45/66) could be explained by the fact that many of the driver genes (STAG2, PHF6, BCOR) are located on the X chromosome, meaning that only one loss or mutation is enough to completely inactivate the function of the protein, but the prevalence is also observed in class 3 (50/69), and this remains unexplained.
7. Interestingly, mutations in epigenetic and signaling genes (n = 66 cases) seem more or less mutually exclusive (only four doubles).
As shown here and in previous reports [5,6] AEL classification resembles that of non-AEL AML, with major classes such as TP53mutated, myelodysplasia-related and de novo-like NPM1-mutated.
Some features are however different, such as the low frequency of FLT3 alterations and more TP53 double-mutations [14,15]. [ 8,10], and GATA2 mutations may be found, but not in the majority of cases and almost exclusively in AELs with TP53 mutations (Figure 1) in which they occur in more than half of the cases [8,10]. The EPOR/JAK2 signaling pathway, when molecularly altered, is functionally active [10] and may be targeted, providing a spark of hope in the AEL therapeutic challenge. However, outside a relatively small group with alterations in signaling pathways (i.e., EPOR/JAK2) and transcriptional regulators (i.e., ERG, GATA2) involved in erythropoiesis, the reason for the particular phenotype of the majority of AELs remains obscure. kinase inhibitors).

AUTHOR CONTRIBUTIONS
NC and JA did the analyses and AG the bioinformatics. MH, SG, and NV provided samples and clinical data. DB and VGB designed the study and wrote the paper.

ACKNOWLEDGMENTS
The authors' work was supported by Inserm and Institut Paoli-Calmettes.