Clinical and Molecular Characteristics of NPM1MT De Novo AML (NPM1MT dnAML) Differ from NPM1MT therapy-associated AML (NPM1MT tAML)

Background: NPM1-mutated AML accounts for 30% of all adult AML cases and frequently carries a favorable prognostic impact when enriched by a normal karyotype and the absence of FLT3-ITD mutations. Although the clinical and molecular characteristics of NPM1-mutated de novo AML (NPM1^MT dnAML) are well known, the impact of NPM1^MT within the context of therapy-associated AML (NPM1^MT tAML) remains controversial. Recent work by Othman et al. suggests NPM1^MT tAML may represent a biologically de novo leukemia, displaying similar clinical, genomic, and transcriptional profile to NPM1^MT dnAML and should be treated as a single disease entity. We sought to externally validate this by analyzing the clinical outcomes, mutational landscape, and molecular features of a large cohort of patients with NPM1-mutated AML.

Methods: We conducted a large retrospective analysis of adult AML patients with NPM1^MT by screening public databases, cBioPortal (Cerami et al., 2012) and AACR GENIE (v 13.1), and expanded with a published metanalytic cohort of various sub-studies from Cleveland Clinic Foundation from 2012-2021 (Awada et al., Blood 2021 and Kewan et al., Nature Communications, 2023). Baseline clinical and molecular characteristics were noted. We analyzed the molecular data in the context of overall survival (OS), clinical parameters, coexisting mutations, karyotype, AML subtypes, ELN-favorability, and gene expression data.

Results: We screened a total of 15,742 adult AML patients, and 4,135 eligible patients were separated into 5 cohorts: 2,156 patients (52%) with NPM1^MT dnAML, 1233 patients (30%) with NPM1^WT sAML, 289 patients (7%) with NPM1^MT sAML, 397 patients (10%) with NPM1^WT tAML, and 60 patients (1.5%) with NPM1^MT tAML. NPM1^MT dnAML and NPM1^MT sAML patients had a median age of 60 and 72 yrs. [18-91 vs 27-91 yrs., p =<0.0001 resp.]. NPM1^MT tAML also had a median age of 72 yrs. [range 24-84]. Females were more represented in NPM1^MT dnAML compared to both sAML cohorts [51% vs 45% and 39%; p =<0.0001, <0.0001] and both tAML cohorts. Abnormal karyotype enriched NPM1^WT tAML more than NPM1^MT dnAML [72% vs 27%, p =<0.0001], whereas normal karyotype was more prominent in NPM1^MT dnAML compared to NPM1^WT sAML, NPM1^MT sAML, and NPM1^WT tAML cohorts [72% vs 55%, 58% and 59%; p =<0.001, <0.001 and 0.0243 resp.]. FLT3-ITD^NEG status dominated all 5 cohorts, NPM1^MT dnAML and NPM1^WT t-AML had similar FLT3-ITD^NEG mutants [83% vs 86%, p =0.52], and NPM1^MT sAML had the highest FLT3-ITD^POS hits (32%). NPM1^WT sAML cohort was profoundly hyperproliferative in bone marrow blast response (33.5%) whereas NPM1^MT dnAML and NPM1^MT tAML had similar responses (73% and 73%, p =0.3). All five cohorts were pancytopenic, the degree of pancytopenia was comparable in NPM1^MT dnAML vs NPM1^MT t-AML as shown below in Fig 1-A.

NPM1^MT dnAML (Fig.1-B) had the best survival vs NPM1^MT tAML [24.74 vs 15.9 mo., p =<0.0016]. NPM1^MT s-AML had the worst survival across all the cohorts, although comparable to NPM1^WT tAML [8.73 and 9.76 mo., p =<0.0001]. When we exclude FLT-ITD, NPM1^MT dnAML has the highest median OS, and median OS for NPM1^MT t-AML rises as well [25 vs 23 mo., p =0.0196], the remaining three cohorts had comparable OS (10.97 vs 10.63 vs 11.05 mo.). Next, we narrowed down to only ELN-favorable risk groups, median OS was only numerically higher for NPM1^MT dnAML vs NPM1^MT tAML [25 vs 23 mo., p-value =0.1348]. In terms of co-mutational profile, NPM1^MT dnAML and NPM1^MT tAML share many similarities though not with associated mutations: NPM1^MT tAML is enriched by TP53 (10%), and SF3B1 (5%), and NPM1^MT dnAML by RAD21 (5%) and CEBPA (5%)

To understand gene-expression differences, we analyzed publicly available RNA-seq data for dnNPM1^MT AML vs NPM1^MT tAML and discovered that NPM1^MT dnAML had higher expression of cMYC as well as FOXO1.

Conclusion: Although dnNPM1^MT AML and NPM1^MT tAML appear to have distinct molecular and clinical profiles, a narrow focus on ELN-favorable patients demonstrates that both entities have similar clinicogenomic outcomes. When evaluated in a large cohort of AML patients with added granularity from sAML cohorts, we notice several differences in survival and molecular profile that converge as more unfavorable factors, such as abnormal cytogenetics and FLT3-ITD^POS status are removed. Ongoing work at Karmanos Cancer Center is focused on evaluating differences in treatment response and gene-expression analysis based on RNA-seq.