Another piece of the AML puzzle

In this issue of Blood, Coenen et al report on a subset of pediatric acute myeloid leukemia (AML) with t(8;16)(p11;p13)/MYST3-CREBBP rearrangement demonstrating specific features with respect to cytomorphology and gene expression patterns, thus fulfilling the criteria of a distinct biological disease entity according to the World Health Organization (WHO) classification.¹ 

Within the last 2 decades the knowledge on genetic abnormalities on the cytogenetic, as well as on the molecular genetic and gene expression level in AML has increased dramatically, and recently more than 200 complete AML genomes were sequenced.²  Thorough analyses revealed a certain subset of genetic abnormalities to be disease defining (ie, causing distinct features regarding cytomorphology, patterns of additional chromosomal abnormalities and molecular mutations, gene expression profile, and clinical course). Within the WHO classification, these disease-defining genetic aberrations had been used to define unique disease entities. The number of such entities increased during recent decades and in 2008, the WHO classification system comprised 7 of them in AML. These are characterized by AML-specific fusion genes and 2 further provisional entities characterized by specific gene mutations (see figure).³ 

Here, Coenen et al¹  report on a distinct AML subgroup that clearly qualifies to become a novel genetically defined AML entity. They demonstrate for pediatric AML with t(8;16)(p11;p13)/MYST3-CREBBP rearrangements a distinct cytomorphology with 97% of cases showing a French-American British (FAB) subtype M4 or M5. This specific association has also been observed in adult AML.⁴  In this latter study, a very typical and coexisting very strong positivity of the same blasts for both myeloperoxidase and nonspecific esterase has been reported. Furthermore, in both pediatric and adult AML erythrophagocytosis is a common feature.

In AML, disease entity–defining abnormalities are considered to be mutually exclusive. They lead to a distinct morphology and gene expression profile.^5,6  Usually, they are accompanied by a distinct pattern of secondary genetic abnormalities. On the cytogenetic level, these secondary alterations mainly constitute unbalanced chromosome aberrations such as different trisomies or deletions. Moreover, accompanying molecular mutations typically affect genes encoding for proteins involved in DNA methylation, chromatin modification, signaling pathways, the spliceosome machinery or the cohesin complex (see figure).²  These mutations frequently occur concomitantly and certain patterns were evolving as our molecular knowledge constantly increased. Some mutations in candidate genes such as MLL-PTD, RUNX1 and TP53 are rarely found in combination with one of the disease-defining abnormalities, and therefore may be discussed to define entities themselves. However, their impact on cytomorphology and gene expression profile is less striking eg, in comparison with profiles driven by NPM1 or CEBPA double gene mutations.⁷ 

For pediatric AML with a t(8;16)(p11;p13)/MYST3-CREBBP rearrangement, Coenen et al¹  observed additional cytogenetic aberrations in 39% of patients, but none was recurrent. So far, data on accompanying molecular mutations in this group is limited. Gene expression analyses revealed a tight clustering of t(8;16)(p11;p13) cases. Both adult and pediatric AML cases harboring t(8;16)(p11;p13) have similarities in their gene expression profiles, underscored by a high concordance of 87%, with respect to the most differentially expressed genes found in the study by Coenen et al¹  compared with a previously reported study in adult AML.⁴  Furthermore, t(8;16)(p11;p13) cases clustered strongly together with, but separate from, MLL-rearranged AML in an unsupervised analysis. Overexpression of HOXA cluster genes was observed by microarray analysis and, together with MLL-rearranged patients, the t(8;16)(p11;p13) patients represented the only other group of pediatric AML patients that selectively activate HOXA genes without HOXB gene activation.

Interestingly, Coenen et al¹  draw attention to further similarities between t(8;16)(p11;p13) AML and MLL-rearranged AML. Both entities frequently occur in neonates and infants and both also frequently in therapy-related AML in adults, which suggests a shared sensitivity to genotoxic stress. So far, however, detailed biological data are lacking. Of note, an additional small subset of AML patients harboring a cytogenetically cryptic NUP98-NSD1 fusion occurring in 16% of children and 2.3% of adult AML with normal karyotype has recently been described.⁸  This subset is also associated with an FAB M4/M5 morphology and a characteristic HOX gene expression profile. In contrast to AML with MYST3-CREBBP fusion, and AML with MLL-fusion, HOXB cluster genes are highly expressed in addition to the high expression of HOXA cluster genes. This HOX gene signature is shared with NPM1 mutated AML and AML with DEK-NUP14 fusion, which are also both associated with AML FAB M4/M5.^1,4,7,8  On the other hand, AML harboring fusions of either RUNX1-RUNX1T1, CBFB-MYH11, or PML-RARA, or CEBPA double-mutated AML show a low or absent expression of HOXA and HOXB cluster genes.⁸ 

Disease-defining abnormalities have a strong impact on disease characteristics and are most likely the strongest pathogenetic driver event; thus, these are a promising target for novel treatment approaches. This has been shown for AML with PML-RARA fusion and translated to a dramatic increase in cure rate.⁹  However, recent studies analyzing the hierarchy of mutations revealed that disease-defining mutations are not necessarily the abnormalities occurring first during disease evolution. In their work, Krönke et al¹⁰  demonstrated that in a subset of NPM1-mutated AML the disease-defining NPM1 mutation was preceded by DNMT3A mutations. Thus, targeting disease-defining abnormalities is most promising. However, if prior premalignant mutations were present, these patients may be at higher risk for secondary or even therapy-related AML/myelodysplastic syndrome. In the study of Krönke et al,¹⁰  3 DNMT3A mutated cases that lost the NPM1 mutation at relapse showed an MLL-PTD that was not present at diagnosis. Therefore, further sequential studies and analyses with respect to mutation hierarchy are necessary to get more insights into clonal evolution in AML.

Taken together, based on the improved molecular understanding of AML, the concept of its pathogenesis has evolved in the last 2 decades from a 2-hit model, suggesting the combination of a type I (enhancing proliferation) and type II mutation (disrupting differentiation) to a revised model. This now favors an accumulation of a variable number of cytogenetic and molecular mutations that have a different impact on the disease characteristics. The number of acquired abnormalities depends on their individual leukemogenic potential. The study of Coenen et al¹  adds a valuable piece to the puzzle to obtain a clearer view on AML pathogenesis.