Dear Editor,

DNA methyltransferase 3A (DNMT3A) is wildly recognized as a tumor suppressor gene. Its deficiency leads to expanded hematopoietic stem cells (HSCs) pool, blocked HSCs differentiation, genomic instability, and a risk of malignant transformation in clonal hematopoiesis [1]. DNMT3A mutation (DNMT3A^Mut) is prevalent in adults, particularly in monocytic, and cytogenetically normal cases, affecting 25% of acute myeloid leukemia (AML) patients [1]. Although the distribution pattern of DNMT3A^Muts has been well characterized in AML, its prognostic significance remains controversial.

To better understand the reasons behind DNMT3A^Mut prognostic heterogeneity, we conducted a retrospective study, as detailed in two previous studies [2, 3]. Our findings show that DNMT3A^Muts with stable expressed mutants are associated with poor prognosis in AML patients without receiving hematopoietic stem cell transplantation (HSCT). In this study, we enrolled 485 adult de novo AML patients, of whom 98 (20.2%) were found to have DNMT3A^Muts. Our results indicate the distribution pattern of DNMT3A^Muts and clinical characteristics of AML patients with these mutations were similar to previous reports. In our cohort, patients with DNMT3A^Muts showed a relatively shorter overall survival (OS), relapse-free survival (RFS) and disease-free survival (DFS) (Figure 1A).

We analyzed how different therapeutic strategies, with or without HSCT, affected the prognosis of various subgroups of DNMT3A^Mut patients. In the HSCT group, DNMT3A^Mut patients exhibited comparable OS, RFS, and DFS with DNMT3A wild-type (DNMT3A^WT) patients (Figure 1A). Conversely, DNMT3A^Mut patients showed poorer OS, RFS, and DFS in the non-HSCT group, and the disparity between DNMT3A^Mut and DNMT3A^WT patients was more pronounced in the non-HSCT group compared to the overall cohort (Figure 1A). Therefore, HSCT overcame the poor prognosis of DNMT3A^Mut, indicating DNMT3A^Mut patients without receiving HSCT were primarily responsible for the unfavorable outcomes.

To further explore key factors contributing to poor prognosis of DNMT3A^Mut patients, we focused on DNMT3A^Mut types in the non-HSCT group. Near 50% of DNMT3A^Mut variants in AML are heterozygous DNMT3A^R882, the hotspot mutation [1]. As reported, DNMT3A^R882 played a dominant-negative role against DNMT3A^WT via formatting dimers, leading to genome-wide hypomethylation [1]. However, DNMT3A^non-R882 has been infrequently studied. Recently, Yung-Hsin Huang et al. systematically studied protein stability of 253 disease-associated DNMT3A^Mut variants and their methyltransferase activity, and provided a comprehensive view on the implications of various DNMT3A^Mut variants [4]. In our cohort, we identified 112 DNMT3A^Mut variants. We first categorized DNMT3A^Mut patients into two groups: DNMT3A^R882 and DNMT3A^non-R882. Except for OS, both of DNMT3A^R882 and DNMT3A^non-R882 patients a had worse RFS and DFS compared to DNMT3A^WT patients, indicating that DNMT3A^R882 and DNMT3A^non-R882 did not effectively distinguish between prognostic outcomes (Figure 1A). Accurately, nearly half of DNMT3A^non-R882 variants were missense mutations, displaying similar mutant stability and diminished methyltransferase activity with DNMT3A^R882, so we further divided DNMT3A^Mut patients into stable and instable DNMT3A^Mut groups (Supplementary Materials and Methods). Notably, stable but not instable DNMT3A^Mut patients exhibited worse OS, RFS and DFS compared to DNMT3A^WT patients (Figure 1A). Additionally, we observed similar results when comparing stable DNMT3A^Mut to nonstable DNMT3A^Mut patients (nonstable DNMT3A^Mut included both instable DNMT3A^Mut and DNMT3A^WT) (Figure 1A). Thus, stable DNMT3A^Mut was responsible for poor prognosis of DNMT3A^Mut patients in the non-HSCT group.

As stable DNMT3A^Mut serves as a strong prognostic factor for AML patients, we were interested in exploring whether differences existed in distribution patterns and clinical characteristics between stable and instable DNMT3A^Mut. In our cohort, we identified stable DNMT3A^Mut in 51 patients and instable DNMT3A^Mut in 25 patients, while 22 DNMT3A^Mut cases were undefined. Apart from a relatively higher frequency of NPM1 mutation in stable DNMT3A^Mut patients, the distribution pattern of stable DNMT3A^Mut was similar to that of instable DNMT3A^Mut. Furthermore, compared to 412 nonstable DNMT3A^Mut patients, 51 stable DNMT3A^Mut patients presented a distribution pattern similar with DNMT3A^Mut compared to DNMT3A^WT patients. Definitely, stable DNMT3A^Mut did not exhibit a specific distribution pattern. We also compared the baseline clinical characteristics of stable DNMT3A^Mut with those of instable DNMT3A^Mut or nonstable DNMT3A^Mut patients. However, stable DNMT3A^Mut patients did not obviously exhibit distinct clinical features. Although there were no significant statistical differences, the stable DNMT3A^Mut group showed a relatively lower complete remission rate and a higher relapse rate compared to both instable DNMT3A^Mut and nonstable DNMT3A^Mut groups, regardless of whether all patients or non-HSCT patients were considered, which possibly contributed to poor prognosis of stable DNMT3A^Mut patients.

To determine whether stable DNMT3A^Mut was an independent risk factor for poor prognosis in non-HSCT patients, we divided patients into stable DNMT3A^Mut group and nonstable DNMT3A^Mut group, and displayed univariate analysis for OS, RFS, or DFS, respectively. Notably, stable DNMT3A^Mut exhibited an adverse effect on OS, RFS and DFS. Following this, we conducted multivariate analyses, confirming that stable DNMT3A^Mut was an independent adverse risk factor for OS, RFS or DFS in the non-HSCT AML patients (Table S1).

Given its strong prognostic role, we sought to validate the clinical significances of stable DNMT3A^Mut in external cohorts. Herein, we analyzed data from The Cancer Genome Atlas TCGA (TCGA) database, and found that stable DNMT3A^Mut consistently indicates poor prognosis in the non-HSCT patients (Figure 1B).

Although most evidences supported that DNMT3A^Mut was a biomarker for poor prognosis in AML, the role of DNMT3A^Mut in AML has not been widely accepted. In our cohort, DNMT3A^Mut patients exhibited a shorter OS, RFS, and DFS in the entire cohort, consistent with results from most other reported cohorts. Most stable DNMT3A^Mut variants are distributed across the methyltransferase domain, which is known to act as a dominant negative, resulting in decreased DNA methylation activity and a poor prognosis. HSCT eliminates leukemia cells, improves the hematopoietic function of the bone marrow, and enhances patient prognosis. A study has shown that HSCT improves outcomes of DNMT3A^Mut AML patients [5], and our findings also indicate that HSCT reduces the prognostic gap between DNMT3A^Mut and DNMT3A^WT patients. In contrast, the prognostic disparity existed between DNMT3A^Mut and DNMT3A^WT patients in the non-HSCT group. DNMT3A^R882, the hotpot mutation in DNMT3A^Mut variants, was recognized as the primary factor contributing to the poor prognosis associated with DNMT3A^Mut [1]. Nearly 50% of DNMT3A^Mut variants are attributed to DNMT3A^R882, but significant heterogeneity also exits in DNMT3A^non-R882 [1]. A recent study indicated that certain missense DNMT3A^non-R882 variants also shared similar protein stability and methyltransferase activity features with DNMT3A^R882, thus, classifying DNMT3A^Mut into DNMT3A^R882 and DNMT3A^non-R882 was not accurate [4]. Herein, we classified DNMT3A^Mut as stable or instable DNMT3A^Mut according to mutant stability, and found that stable DNMT3A^Mut, rather than instable DNMT3A^Mut, was an independent predictor for relatively shorter OS, RFS and DFS. Furthermore, stable DNMT3A^Mut primarily worsened the prognosis for AML genetic subtypes, indicating stable DNMT3A^Mut was a dominant contributor for poor prognosis in the non-HSCT patients. In the future, we aim to demonstrate the prognostic value of stable DNMT3A^Mut through prospective multicenter clinical studies.

Collectively, stable DNMT3A^Mut, but not instable DNMT3A^Mut, should be regarded as a strong predictor of poor prognosis in AML patients without receiving HSCT, thereby providing insights into the mechanism underlying AML pathogenesis.

Author Contributions

Xiang Zhang designed this study. Xiang Zhang and Xiong Ni collected clinical data and updated follow-up. Xiong Ni and Jin Jie guided clinical managements for patients. Xiang Zhang, Lixia Liu, and Jiayue Qin displayed data analysis. Xiang Zhang and Jiayue Qin wrote the manuscript. Jie Jin provided critical comments on this study. Xiong Ni and Jie Jin revised the manuscript. All authors have read and approved the final manuscript.

Ethics Statements

This study was approved by the ethical review committees of the First Affiliated Hospital of Zhejiang University School of Medicine (IIT20220659A) and Changhai Hospital (B2022-035). All procedures in studies involving human participants were performed in accordance with the ethical standards of the institutional research committee and with the 1964 Helsinki Declaration and its later amendments. Written informed consent was obtained from these patients.

Conflicts of Interest

The authors declare no conflicts of interest.