A Novel Prognostic Model for Adult Acute Myeloid Leukemia Patients with NPM1 Mutation

Background: Approximately 25-35% of adult patients with acute myeloid leukemia (AML) harbor NPM1 mutation, characterized by aberrant cytoplasmic dislocation of NPM1 protein, which is generally associated with a favorable outcome in the absence of FLT3-ITD mutation. Due to its unique biological and clinical characteristics, the World Health Organization (WHO) classified NPM1-mutated AML as a specific leukemia subtype for myeloid neoplasms and acute leukemia in 2022. Indeed, NPM1-mutated AML encompasses a spectrum of molecular heterogeneity, with significant variability in patient outcomes. Currently, there is a lack of personalized prognostic scoring model tailored to patients with NPM1-mutated AML, especially based on molecular landscape. Therefore, a profound understanding of the co-mutational landscape profile of NPM1-mutated AML is pivotal for the development of novel prognostic scoring models, which in turn is essential for establishing precise risk stratification strategies.

Methods: The training cohort comprised 178 newly diagnosed adult AML patients with NPM1 mutations (M3 excluded), who underwent a comprehensive assessment of the molecular abnormalities through next-generation sequencing (NGS) at diagnosis, in our center diagnosed from October 2018 to November 2022. LASSO-Cox regulation was performed to select the variables screened from the co-mutation genes to construct the prognostic scoring model and nomogram. The performance of the novel nomogram was evaluated using the concordance index (C-index) and the area under the receiver operating characteristic curve (AUC). Calibration curve was employed to assess the agreement between the actual observations and nomogram predictions. All statistical analyses were performed using SPSS 23.0 (SPSS Inc., Chicago, IL) and R software version 4.2.2 (R Foundation for Statistical Computing, Vienna, Austria).

Results: A total of 178 eligible NPM1-mutated AML patients were included in the training set. 83 (46.6%) patients were male, and the median age was 59 (IQR 51-68) years. The median follow-up of the 178 patients was 26.23 (95% CI 23.31-29.16) months and the median OS had not been reached. The 3-year expected OS was 51.5% (95% CI 43.3%-59.7%). LASSO regression analysis screened variables included FLT3-ITD, DNMT3A, and IDH1/2 mutations. Cox regression model was further established based on parameters screened by Lasso regression: FLT3-ITD×0.970+DNMT3A×0.533-IDH1/2×0.772 (FLT3-ITD mutation scored 1; DNMT3A mutation scored 1; IDH1/2 mutation scored 1; 0 for other conditions). Thus 178 patients were classified into three subgroups on the basis of the risk score: low-risk (L-R, score <= 0, n = 67), intermediate-risk (I-R, 0<score <= 0.8, n = 54), and high risk (H-R, score > 0.8, n =57) subgroups. The 3-year OS for the L-R, I-R and H-R groups was 76.2% (95%CI 64.8%-87.6%), 52.7% (95%CI 37.6%-67.8%), and 25.3% (95%CI 13.3%-37.3%), respectively (P<0.001). A prognostic nomogram that integrated these screened variables from the LASSO-Cox regression model was constructed. The C-index of the training set was 0.69 (95%CI 0.632-0.748), while the AUC values at 1, 2, and 3 years were 0.707 (95%CI 0.622-0.791), 0.766 (95%CI 0.696-0.836), and 0.768 (95%CI 0.699-0.837), respectively. The results indicated favorable accuracy and discrimination. The calibration curves of the model built for predicting survival of 1, 2 and 3 years indicated a good consistency between the predicted and observed results in the training cohort.

Conclusions: We constructed a novel risk stratification model that integrated the co-mutations profile with clinical outcomes in adult AML patients with NPM1 mutations, which divided patients into three distinct subgroups with different clinical outcomes to offer a more nuanced understanding of patient survival prospects and pave the way for more precise and personalized therapeutic interventions.

No relevant conflicts of interest to declare.