Background: Acute erythroid leukemia (AEL) is a rare subtype of AML characterized by erythroid predominant proliferation and classified into two subtypes with pure erythroid (PEL) and myeloid/erythroid (MEL) phenotypes. Although several reports described gene mutations in AEL, genotype phenotype correlations have not fully been elucidated with little knowledge about feasible molecular targets for therapy. Methods: To understand the mechanism of the erythroid dominant phenotype of AEL and identify potential therapeutic targets for AEL, we analyzed a total of 121 adult AEL cases with the median age of 60 (23-87), using whole genome/exome sequencing of 35 cases, followed by targeted-capture sequencing of 387 genes together with 1,279 SNP loci for copy number measurements in all cases. Among these, 21 were also analyzed by RNA sequencing. Genetic profiles of these AEL cases were compared to those of 409 cases with non-erythroid AML (non-AEL) including 195 cases from The Cancer Genome Atlas. Six patient-derived xenografts (PDX) were established from AEL with JAK2 and/or EPOR focal gain/amplification/mutation. PDX cells were inoculated into immune-deficient mice and tested for their response to JAK1/2 inhibitor. Results: According to unique genetic alterations, AEL was classified into 4 genomic groups (A-D). Characterized by TP53 mutations and complex karyotype, Group A was the most common subtype (48/121; 40%) and showed very poor prognosis. Remarkably, almost all the PEL cases (12/13; 92%) were categorized into Group A. Conspicuously, 75% of PEL cases with TP53 mutation had focal gain/amplifications/mutations of JAK2 (5/12; 42%), EPOR (7/12; 58%), and ERG/ETS2 (1/12; 8%) loci on chromosomes 9p, 19q, and 21q, respectively, while 34% of MEL cases with TP53 mutation had focal gain/amplifications/mutations of JAK2 (2/29; 7%), EPOR (7/29;24%), and ERG/ETS2 (7/29;24%) loci, frequently in combination. Group B was characterized by frequent NPM1 mutations, in contrast to the frequent co-mutation of FLT3 in the corresponding subgroup of NPM1-mutated cases in non-AEL, whereas NPM1-mutated patents in this group lacked FLT3 mutations but had frequent PTPN11 mutations (8/16; 50%), which were much less common in non-AEL (15/101; 15%). All cases in Group C (n=22, 18%), another prevalent form of AEL, had STAG2 mutations and classified in MEL. Prominently, 68% (17/25) of STAG2-mutated AEL cases had KMT2A-PTD, which was rarely found in non-AEL cases. The remaining cases were categorized into Group D, which was enriched for mutations in ASXL1, BCOR, PHF6, RUNX1 and TET2. We also identified recurrent loss-of-function USP9X mutations in this group, which were previously reported in ALL with an upregulated JAK-STAT pathway. In RNA sequencing analysis, AEL cases exhibited gene expression profiles implicated in an upregulated STAT5 signaling pathway, which was seen not only in those cases with JAK2 or EPOR focal gain/amplification/mutation, but also in AEL without these amplifications, suggesting that aberrantly upregulated STAT5 activation might represent a common molecular signature of AEL. Survival analysis revealed that TP53 mutation is a poor prognostic factor in AEL and non-AEL and no statistically significant difference between AEL and non-AEL with TP53 mutation. Intriguingly, 19p gains/amplifications were associated with a significantly poor prognostic prognosis in TP53-mutated AEL cases. Based on this finding, we evaluated the effect of a JAK inhibitor, ruxolitinib, on 6 PDX models established from AEL having TP53 mutations and JAK2 and EPOR mutation/amplification. Of interest , ruxolitinib significantly suppressed cell growth and prolonged overall survival in mice engrafted with 4 PDX models with STAT5 downregulation, although the other 2 models were resistant to JAK2 inhibition with persistent STAT5 activation. Conclusion: AEL is a heterogeneous group of AML, of which PEL is characterized by frequent amplifications/mutations in JAK2 and/or EPOR. Frequent involvement of EPOR/JAK/STAT pathway is a common feature of AEL, in which a therapeutic role of JAK inhibition was suggested.