KMT2A

and NUP98 gene fusions are two common types of acute myeloid leukemia (AML) with poor prognosis, incorporated as independent entities in the newest WHO classification. However, their transcriptomic subclasses and optimal risk stratification remain incompletely defined. Herein, we systematically integrated and analyzed multi-omics and clinical data from 154 newly diagnosed AML patients (≥14 years) with KMT2A fusions (KMT2A-r; n=95) or NUP98 fusions (NUP98-r; n=59) treated in our center (2019–2024), companied with external validation cohorts (n=53).

KMT2A-r exhibited elevated bone marrow (BM) blasts, while NUP98-r had higher white blood cell (WBC) counts. Significantly higher positive rates of FLT3-ITD (58% vs. 3.2%, P<0.001), WT1 (32% vs. 5.3%, P<0.001), CEBPA (12% vs. 0%, P<0.001), MYC (10% vs. 1.1%, P=0.013), and IDH1 (10% vs. 0%, P=0.003) were observed in NUP98-r. Conversely, KMT2A-r AML exhibited higher mutation rates of KRAS (26% vs. 8.5%, P=0.007), KMT2A-PTD (25% vs. 0%, P<0.001), and PTPN11 (17% vs. 3.4%, P=0.012).

Unsupervised clustering of gene expression profiles identified four distinct transcriptomic subclusters in the merged cohort with KMT2A-r or NUP98-r gene fusions:

  • C1: Primitive KMT2A-r and NUP98-r (n=60, 39%),

  • C2: NUP98::NSD1(n=28, 18%),

  • C3: Mixed KMT2A-r (n=32, 21%),

  • C4: Committed KMT2A-r (n=34, 22%).

Most NUP98::NSD1 cases clustered in C2 (n=27, 96%), with only 2 cases in C1. In contrast, C1 was enriched for KMT2A::ELL (n=21, 35%), NUP98::HOXA9 (n=11, 18%), other NUP98-r (n=17, 28%), and other KMT2A-r (n=7, 12%). C3 and C4 primarily contained remaining KMT2A-r cases: 28% and 38% of KMT2A::MLLT3, 47% and 21% of KMT2A::AFDN, 16% and 24% of KMT2A::MLLT10, and 9.4% and 15% of other KMT2A-r were distributed in C3 and C4, respectively, with only one NUP98::KDM5A case in C4. Notably, C3 was associated with inferior overall survival (OS) in patients <60 years (P=0.033), while C2 predicted poor OS in patients ≥60 years (P=0.019). Standard “3+7”/Venetoclax-based induction therapy was administered in 58%/25%, 75%/14%, 69%/31%, and 68%/18% of C1–C4, respectively. Hematopoietic stem cell transplantation (HSCT) rates were similar across subclusters (46%-59%). Besides, 65%, 25%, 72%, and 71% of C1–C4 were classified as adverse risk group per ELN 2022 recommendations.

C2 had the highest WBC counts [75 vs. 24 (C1), 4 (C2), and 14 (C4)]. C4 had the highest platelet counts [84 vs. 47 (C1), 63 (C2), 30 (C3)]. FLT3-ITD (20% and 89%, P<0.001), WT1 (18% and 43%, P<0.001), and CEBPA (5% and 14%, P=0.034) mutations were predominantly observed in C1 and C2. C1 was also enriched for GATA2 (18% vs. 3.6%, 9.4%, and 0% in C2–C4, P=0.014) and STAG2 (17% vs. 0%-2.9% in C2–C4, P=0.004) mutations. C3 had higher KMT2A-PTD events [28% vs. 15% (C1), 0% (C2), and 18% (C4), P=0.014]. Recurrent spliceosome mutations including U2AF1 (n=3) and SRSF2 (n=3) were detected in C4. Subtype-specific biomarkers included PRAME (overexpressed in C3) and MYO6, etc. The pre-trained HOX-committed subtype were mostly clustered into C4, and HOX-primitive were dominantly located in C1 subclass. Besides, stemless score (LSC17) was higher in C3 subclass.

Then we developed a novel risk score incorporating age, WBC count, C3 (for <60 years), C2 (FLT3-ITD+ for ≥60 years), and mutations (SMC3 and U2AF1): [(age-44.88)/16.23×0.478 + (WBC-53.96)/80.1×0.382 + C3 (age<60 yrs)×1.26 + C2 (FLT3-ITD+ and age≥60 yrs)×1.66 + SMC3mut×1.27 + U2AF1mut×2.56],effectively stratifying patients into high- and standard-risk groups (P<0.001), validated externally. HSCT significantly improved survival in both subgroups. KMT2A-r accounted for 74% and 53% of the high-risk and low-risk KN groups. KMT2A::AFDN (23% vs. 7.6%, P=0.007) and rare KMT2A-r (16% vs. 5.4%, P=0.026) were more frequent in the high-risk group, while the low-risk group was enriched for NUP98::HOXA9-positive patients (11% vs. 1.6%, P=0.051). Transcriptomic subclasses differed significantly between risk groups (P<0.001): C1 (25% vs. 48%), C2 (15% vs. 21%), C3 (46% vs. 4.3%), and C4 (13% vs. 27%). KRAS (28% vs. 14%, P=0.036), SMC3 (6.6% vs. 0%, P=0.024), and U2AF1 (6.6% vs. 0%, P=0.024) mutations were enriched in the high-risk group.

In conclusion, our findings refine the molecular classification of KMT2A-r/NUP98-r AML, establish a clinically actionable risk stratification tool, and identify novel therapeutic targets, paving the way for precision medicine in the two high-risk genetic anomalies.

This content is only available as a PDF.
2025
Sign in via your Institution