Childhood Acute Myeloid Leukemia (AML) is a clinically and genetically heterogeneous malignant disease. Despite improvements in outcome over the past decades, the current survival rate still is approximately 60-70%. Cytogenetic, recurrent genetic abnormalities and early response to induction treatment are the main factors predicting clinical outcome. While the majority of children carry recurrent chromosomal translocations, 20% of patients do not show any recognizable cytogenetic alteration and are defined to have cytogenetically normal AML (CN-AML). This subset of patients is characterized by a significant heterogeneity in clinical outcome, which is influenced by factors only recently started to be identified. In this respect, genome-wide analyses have been used with the aim of defining the full array of genetic lesions in CN-AML. Recently, through whole-transcriptome massively parallel sequencing of seven cases of pediatric CN-AML, we identified a novel recurrent CBFA2T3-GLIS2 fusion, predicting poorer outcome. However, since the expression of CBFA2T3-GLIS2 fusion in mice is not sufficient for leukemogenesis, we speculated that further unknown abnormalities could contribute to both cancer transformation and response to treatment. Thus, we analyzed, by whole-transcriptome sequencing, 4 CBFA2T3-GLIS2-positive patients, as well as 4 CN-AML patients. We identified a new fusion transcript in the CBFA2T3-GLIS2-positive patients, involving Desert Hedgehog (DHH), a member of Hedgehog family, and Ras Homologue Enrich in Brain Like 1 (RHEBL1), a gene coding for a small GTPase of the Ras family. Through the screening of a validation cohort of 55 additional pediatric AML patients, we globally detected DHH-RHEBL1 fusion in 8 out of 20 (40%) CBFA2T3-GLIS2-rearranged patients. Gene expression analysis performed on RNA-seq data revealed that DHH-RHEBL1-positive patients exhibited a specific signature. These 8 patients had an 8-year overall survival worse than that of the remaining 12 CBFA2T3-GLIS2-rearranged patients not harboring DHH-RHEBL1 fusion (25% vs 55%, respectively, P=0.1). Taken together, these findings are unprecedented and indicate that the DHH-RHEBL1 fusion transcript is a novel recurrent feature in the changing landscape of CBFA2T3-GLIS2-positive childhood AML. Moreover, it could be instrumental in the identification of a subgroup of CBFA2T3-GLIS2-positive patients with a very poor outcome.