Genomic Analyses Unveil the Pathogenesis and Inform on Therapeutic Targeting in KMT2A-PTD AML

A partial tandem duplication (PTD) in the KMT2A gene is detected in approximately 5-10% of acute myeloid leukemia (AML) and myelodysplastic syndrome (MDS). Previous studies revealed that HOX-genes are highly expressed in KMT2A-PTD AML similar to KMT2A-rearranged or NPM1-mutated AML. While the overexpression of HOX-genes is a convergent oncogenic pathway shared by these molecular subtypes of AML, the mechanism by which KMT2A-PTD induces high expression of the HOX-genes remains unclear and may be distinct from that mediated by balanced KMT2A-translocations, and NPM1 mutations.

We stipulated that analysis of common molecular features of KMT2A-PTD will help identifying mechanistic intersections pointing towards possible targeted therapies. In a total of 5420 AML cases ¹ including 793 cases from publicly available sources (Beat AML ² and TCGA ³), we identified 254 cases of KMT2A-PTD. Somatic mutational screening and karyotype analysis showed co-occurrence with DNMT3A (31%), FLT3-ITD (32%), IDH1/2 (26%), RUNX1 (28%), TET2 (18%), SRSF2, STAG2, U2AF1 mutations and deletion 5q. Strong leukemic initiators such as KMT2A-rearrangement, CBFB-MYH11, PML-RARA, RUNX1-RUNX1T1, trisomy 8 and NPM1 mutations were mutually exclusive with KMT2A-PTD and did not rely on acquisition of additional mutations linked to secondary AML ontogenesis ( e.g., ASXL1, RUNX1, SRSF2, and STAG2) different from KMT2A-PTD.

Next, we analyzed bulk mRNA-expression between mutually exclusive KMT2A-PTD, KMT2A-rearranged or NPM1-mutant high HOX-genes expressors. While several differentially expressed genes (DEG) between KMT2A-PTD and KMT2A-rearranged cases were detected, there were few DEG between KMT2A-PTD and NPM1-mutated cases. Compared to KMT2A-rearrenged cases, KMT2A-PTD had higher expression of CD34 and KIT in addition to some HOX-genes such as HOXB2 or NKX2-3. In contrast, expression of CD1D, CD14, and immune checkpoint genes including LILRB4, CD276, CD86, and SIGLEC7 was much higher in KMT2A-rearranged than KMT2A-PTD cases. These results prompted us to analyze the hematopoietic differentiation status. For this purpose, we defined three gene signatures along the HOX-targeted genes axis, termed as HOX-primitive, HOX-transient, and HOX-committed profiles. Differentiation analysis revealed almost all of KMT2A-PTD cases could be subclassified into HOX-transient stage while most of KMT2A-rearranged cases were HOX-committed. However, NPM1-mutated cases were interspersed in all the differentiation stages along the HOX-axis. We therefore assumed that there were a few DEG shared between KMT2A-PTD and NPM1-mutated cases as some NPM1-mutated AML clustered in the HOX-transient expression profile group.

Intriguingly, DEG analysis was concordant with the hierarchical clustering based on the HOX-gens axis. Indeed, CD34 and KIT were highly expressed in the HOX-transient cases, and CD1D and CD14 in the HOX-committed cases, consistent with the enrichment of KMT2A-PTD and KMT2A-rearranged AML, respectively. We can conclude that the observed signatures were unique to the HOX-stages rather than the molecular abnormalities such as KMT2A-PTD and KMT2A-rearrangements. For example, CD34 upregulation was a feature shared transversally by all HOX-transient cases, including both KMT2A-rearranged and NPM1-mutated cases. As shown in Figure, KMT2A-rearranged, KMT2A-PTD, and NPM1-mutated cases were more clearly classified in HOX-profiles than molecular abnormalities. Additionally, among the cases which molecular subtypes were not clear, those with high HOX-genes expression revealed higher MEN1 expression.

In conclusion, despite molecularly different, KMT2A-PTD, KMT2A-rearranged, and NPM1-mutated AML have similar expression patterns defined by each HOX differentiation stage. Like core binding factor AML such as RUNX1-RUNX1T1 and CBFB-MYH11, AML with higher expression of HOX-genes should be classified in the same group, HOX-AML. Given the results obtained with menin inhibitors in KMT2A-rearranged and NPM1-mutated AML, our findings open an opportunity for exploiting a therapeutic vulnerability in all HOX-AML including KMT2A-PTD AML or AML with high MEN1 expression. Since HOX-AML highly express genes according to the HOX differentiation profile, stage-specific surface proteins coded by these genes would be promising targets.