Adverse Impact of DNA Methylation Regulatory Gene Mutations on the Prognosis of AML Patients in the 2017 ELN Favorable Risk Group, Particularly Those Defined by NPM1 Mutation

The 2017 ELN risk stratification has been widely adopted, but some studies have suggested the outcomes are heterogenous within the ELN risk groups and may be affected by other co-existing genetic mutations. This study evaluated the impact of DNA methylation regulatory gene (TET2, IDH1/2, DNMT3A, SETBP1) mutations (DMRGM) evaluated by NGS in the outcome of AML patients in each ELN risk group. A total of 114 patients were analyzed with a median follow-up of 12 months. Overall, 30.7% (35/114) of patients had DMRGM. DMRGM status had no impact on CR rate in each ELN risk group. The OS, however, was significantly shorter in patients with DMRGM compared to those without DMRGM (median OS: 12 vs. 33 months, p = 0.0053). Multivariate analysis showed DMRGM status was an independent unfavorable factor for OS (HR: 2.704, 95% CI: 1.451–5.041, p = 0.0017). The adverse OS impact of DMRGM was only observed in the ELN favorable group (7 months vs. not reached, p = 0.0001), but not in the intermediate or adverse group. Among the favorable group with DMRGM (n = 16), DMRGM occurred predominantly in cases with mutated NPM1 (15/16, or 93.8%). Our results suggest that DMRGM adversely impact the outcomes of ELN favorable group patients, particularly those with mutated NPM1. Further studies are warranted to confirm our observations.


Introduction
Acute myeloid leukemia (AML) is a heterogeneous aggressive blood cell cancer which is the most common acute leukemia in adults [1]. Next-generation sequencing (NGS) has emerged as an important tool in the identification of mutated genes in AML [2][3][4].
Recently, large studies have identified multiple such gene mutations that significantly impact the prognostic outcomes of AML [5][6][7][8]. In 2017, the European Leukemia Net (ELN) stratified AML patients into three risk groups based on the presence or absence of specific chromosomal abnormalities and selected gene mutations [9]. As a result, the impact of Diagnostics 2021, 11, 986 2 of 11 NPM1, FLT3-ITD and CEBPA mutations was further defined based on the presence of other mutations such as RUNX1, ASXL1 and TP53. This has advanced prognosis and therapeutic options in AML. Some studies, however, have suggested that complete remission (CR) rates and outcomes are heterogenous within the 2017 ELN risk groups and may be affected by other co-existing genetic mutations [10][11][12].
One important class of mutations in AML regulates DNA methylation patterns. Genes such as DNMT3A, TET2, IDH1 and IDH2, collectively referred to here as DNA methylation regulatory gene mutations (DMRGM), have been shown to be associated with poor prognosis [5,11,[13][14][15][16][17][18]. Others, however, have not shown similar prognostic impacts of such mutations of DMRGM in AML patients [19,20]. It is well established that IDH1 and 2 are mutually exclusively in AML [21]. In addition, few studies with a large number of AML patient have demonstrated that TET2 mutation is mutually exclusive with IDH1 and IDH2 mutations [18,22]. Based on these observations, it has been suggested that the mutations of DNA methylation regulatory genes may have similar biological and prognostic effects in AML [6,15,22,23]. Of note, a prior study by Ryotokuji T et al. in 2016, combining these DMRGM as one factor, revealed the unfavorable prognostic impact of DMRGM for overall survival (OS) among AML patients [24]. These observations prompted us to explore the prognostic impact of DMRGM in each group of the 2017 ELN risk group stratification.

Patients
Adult AML patients aged 18-75 who underwent intensive induction chemotherapy at our institution from 2017 to 2020 were screened for this study. All patients had bone marrow biopsies for morphologic evaluation, flow cytometry immunophenotyping, conventional karyotyping and next generation sequencing (NGS) study for AML-associated gene mutations at diagnosis. The diagnoses and classification of AML were based on the 2016 World Health Organization (WHO) classification of AML, and risk groups were assigned using the 2017 ELN risk stratification scheme [9,25]. Patients with treatment related-AML (t-AML) and acute promyelocytic leukemia (APL) were excluded. All patients in the study received intensive chemotherapy with or without FLT3 inhibitors as appropriate and/or hematopoietic stem cell transplantation (HSCT) based on standard protocols. The study was approved by the institutional review board.

Patient Group
DMRGM were defined as mutations detected in at least one of the following genes: DNMT3A, TET2, IDH1, IDH2 or SETBP1. The patients were divided into 6 groups: groups 1 and 2 as ELN favorable group with or without DMRGM, groups 3 and 4 as ELN intermediate group with or without DMRGM, and groups 5 and 6 as ELN adverse group with or without DMRGM.

Statistical Analyses
The median follow-up for survival was calculated according to the Kaplan-Meier method [26]. Patient baseline characteristics between DNA methylation regulatory genes mutated or unmutated groups were performed by using the Mann-Whitney test for continuous variables and by using a chi-square test for categorical variables. The Kaplan-Meier method was used to estimate the distribution of OS in each group by the 2017 ELN risk stratification [27]. The log-rank test was used for univariate analysis to compare the OS difference between groups. Post hoc pairwise comparisons for the log-rank test with Sidak adjustment were performed to prevent type I error. The Cox proportional hazard regression model was used for multivariable analysis [28]. CR rates were compared by chi-square test or Fischer's exact test based on sample sizes. Two-tailed statistical significance at a level of 5% was used for statistical analysis. All analyses were performed using SAS 9.4 (SAS Institute, Cary, NC, USA).

Patients' Characteristics
A total of 114 patients were studied (Table 1). There were 69 males (60.5%) and 45 females (39.5%) with a median age of 61.5 years (range, . The median follow-up was 12 months. Patients were nearly equally distributed among favorable, intermediate, and adverse per 2017 ELN. DMRGM were present in 30.7% (35/114) of patients. Table 1 describes the clinical and laboratory features as well as AML classification of patients by DMRGM status. There were no significant differences in age, gender, white blood cell (WBC) counts, bone marrow (BM) blasts, peripheral blood (PB) blasts, or HSCT rate between DMRGM positive versus negative groups.

Survival Analysis
Among the entire cohort of patients, the OS of patients with DMRGM was significantly worse than that of patients without DMRGM (medial OS: 12 months vs. 33 months, p = 0.0053, Figure 1A). Multivariate analysis showed that DMRGM status was an independent unfavorable prognostic factor for OS (hazard ratio (HR): 2.704, 95% confidence interval (CI): 1.451-5.041, p = 0.0017, Table 3a).  We subsequently examined the impact of DMRGM on survival within each 2017 ELN risk group. For the ELN favorable group (n = 37), the median OS of patients with DMRGM was significantly shorter than those without DMRGM (7 months vs. not reached, p = 0.0001, Figure 1B). Of note, among the ELN favorable group, patients with mutated NPM1 I (n = 21) DMRGM showed similar impact on median OS (DMRGM vs. no DMRGM: 7 months vs. not reached, p = 0.016). In multivariate analysis, the DMRGM status remained an independent unfavorable prognostic factor for OS (HR: 6.882, 95% CI: 1.24-38.184, p = 0.0274, Table 3b). In contrast, DMRGM showed no significant impact on OS in either the ELN intermediate group or ELN adverse group ( Figure 1C,D). Of importance, the median OS of the ELN favorable group with DMRGM was significantly shorter than that of the ELN intermediate group (7 months vs. not reached, p = 0.0078, Figure 2) and also appeared shorter than that of the ELN adverse group, although not statistically significant (7 vs. 12 months, p = 0.9937, Figure 2).  We subsequently examined the impact of DMRGM on survival within each 2017 ELN risk group. For the ELN favorable group (n = 37), the median OS of patients with DMRGM was significantly shorter than those without DMRGM (7 months vs. not reached, p = 0.0001, Figure 1B). Of note, among the ELN favorable group, patients with mutated NPM1 I (n = 21) DMRGM showed similar impact on median OS (DMRGM vs. no DMRGM: 7 months vs. not reached, p = 0.016). In multivariate analysis, the DMRGM status remained an independent unfavorable prognostic factor for OS (HR: 6.882, 95% CI: 1.24-38.184, p = 0.0274, Table 3b). In contrast, DMRGM showed no significant impact on OS in either the ELN intermediate group or ELN adverse group ( Figure 1C,D). Of importance, the median OS of the ELN favorable group with DMRGM was significantly shorter than that of the ELN intermediate group (7 months vs. not reached, p = 0.0078, Figure 2) and also appeared shorter than that of the ELN adverse group, although not statistically significant (7 vs. 12 months, p = 0.9937, Figure 2).

Genetic Association of DMRGM with Other Mutations in the 2017 ELN Favorable Risk Group
Among the ELN favorable group with DMRGM (n = 16), DMRGM occurred predominantly in cases with mutated NPM1 (15/16, or 93.8%, Figure 3). The remaining one case concurred with CBFB-MYH1 fusion ( Figure 3). The DMRGM in mutated NPM1 cases appeared independent of FLT3-ITD mutations (eight patients had FLT3-ITD low , and seven patients were without FLT3-ITD).

Genetic Association of DMRGM with Other Mutations in the 2017 ELN Favorable Risk Group
Among the ELN favorable group with DMRGM (n = 16), DMRGM occurred predominantly in cases with mutated NPM1 (15/16, or 93.8%, Figure 3). The remaining one case concurred with CBFB-MYH1 fusion (Figure 3). The DMRGM in mutated NPM1 cases appeared independent of FLT3-ITD mutations (eight patients had FLT3-ITD low , and seven patients were without FLT3-ITD).

Discussion
The results of this study indicate that DMRGM commonly occur in AML and they adversely impact the prognosis of AML patients based on the 2017 ELN risk classification. The impact of these mutations was mainly seen in the favorable risk group, particularly those with NPM1 mutation, while they had no impact on the ELN intermediate or adverse risk category.
Although the prognostic impact of DMRGM as a group of genes on AML patients has not been extensively studied, multiple studies on individual genes involving DNA methylation regulations correlate with our findings. Several studies suggest that DNMT3A adversely impacts OS effect on AML patients [6,14,15]. In a meta-analysis of 4500 AML patients by Shivarov et al., DNMT3A mutations showed a significantly worse prognosis, particularly in patients with cytogenetically normal AML (CN-AML) [29]. IDH1/2 mutations have also been shown to adversely impact OS in AML patients [16,17]. In a recent meta-analysis, IDH1/2 mutations showed no OS affect in the whole AML population; however, IDH1 mutation conferred a worse OS in patients with CN-AML (OS: HR, 1.21; 95% CI, 1.01-1.46) [30]. Additionally, similar to our findings, others have shown that IDH mutations tended to occur with NPM1 but not FLT3-ITD mutations [5,17]. Furthermore, some studies reported that the negative impact of IDH 1 and 2 mutations was only seen in mutated NPM1 with wild-type FLT3, but not in other AML patients [17,31]. With respect to TET2 mutations, a meta-analysis by Liu et al. in 2552 AML patients showed that it was associated as an adverse prognostic indicator only in CN-AML (OS: HR, 1.43, 95% CI: 1.16-1.75, p = 0.001) [32]. Tian et al. evaluated the impact of TET2 in 373 adults with CN-AML patients. They reported that TET2 mutation was an unfavorable prognostic factor leading to shorter median OS as compared to wild-type TET2, particularly in FLT3-ITD negative, NPM1 positive patients (9.5 vs. 32.2 months, p = 0.013) [33].

Discussion
The results of this study indicate that DMRGM commonly occur in AML and they adversely impact the prognosis of AML patients based on the 2017 ELN risk classification. The impact of these mutations was mainly seen in the favorable risk group, particularly those with NPM1 mutation, while they had no impact on the ELN intermediate or adverse risk category.
Although the prognostic impact of DMRGM as a group of genes on AML patients has not been extensively studied, multiple studies on individual genes involving DNA methylation regulations correlate with our findings. Several studies suggest that DNMT3A adversely impacts OS effect on AML patients [6,14,15]. In a meta-analysis of 4500 AML patients by Shivarov et al., DNMT3A mutations showed a significantly worse prognosis, particularly in patients with cytogenetically normal AML (CN-AML) [29]. IDH1/2 mutations have also been shown to adversely impact OS in AML patients [16,17]. In a recent meta-analysis, IDH1/2 mutations showed no OS affect in the whole AML population; however, IDH1 mutation conferred a worse OS in patients with CN-AML (OS: HR, 1.21; 95% CI, 1.01-1.46) [30]. Additionally, similar to our findings, others have shown that IDH mutations tended to occur with NPM1 but not FLT3-ITD mutations [5,17]. Furthermore, some studies reported that the negative impact of IDH 1 and 2 mutations was only seen in mutated NPM1 with wild-type FLT3, but not in other AML patients [17,31]. With respect to TET2 mutations, a meta-analysis by Liu et al. in 2552 AML patients showed that it was associated as an adverse prognostic indicator only in CN-AML (OS: HR, 1.43, 95% CI: 1.16-1.75, p = 0.001) [32]. Tian et al. evaluated the impact of TET2 in 373 adults with CN-AML patients. They reported that TET2 mutation was an unfavorable prognostic factor leading to shorter median OS as compared to wild-type TET2, particularly in FLT3-ITD negative, NPM1 positive patients (9.5 vs. 32.2 months, p = 0.013) [33]. Though it is well established that mutated NPM1 has favorable outcomes in AML, our results and the above studies suggest that DMRGM may adversely impact the OS of patients with mutated NPM1. Although many of the above studies considered the effects of cytogenetic findings and NPM1 mutation, none of these studies specifically looked into the impact of these gene mutations among different ELN risk groups.
Our approach of considering DMRGM as a group is similar to the study by Ryotokuji et al., in which the investigators integrated four genes (IDH1/2, TET2 and DNMT3A) involving DNA methylation regulation as a group and analyzed its prognostic impact on AML. In their study, cases with DMRGM had a significantly poorer OS than those without DMRGM in all cases of AML, in agreement with our findings [24]. However, this study did not evaluate the prognostic effect of DMRGM by ELN risk stratification. Our study further found that the adverse impact of DMRGM on AML patients was only in patients in the ELN favorable risk group. However, the mechanisms leading to the adverse impact remain unclear and require future studies. Of note, the adverse impact appears not associated with the complete response (CR) rate. In the entire cohort, the CR rate of patients with DMRGM was not different from the patients without DMRGM. Similarly, DMRGM status did not impact the CR rate within each ELN risk group.
In contrast to our observation, some studies have shown that mutations of DNA methylation regulatory genes may not impact the prognosis of AML patients. For example, Gaidzik et al. reported no significant survival impact of DNMT3A mutation in a cohort of 1770 adult AML patients [19]. Shen et al. stated that DNMT3A, but not IDH1/2, mutations adversely impacted OS in a cohort of 605 AML patients [20]. Moreover, Mason et al. reported that NPM1-mutated AML patients with TET2, IDH1/2 mutations had better OS than NPM1-mutated patients without such mutations [34]. The mechanisms leading to these discrepancies remain uncertain. Further studies with larger sample sizes are needed to fully illustrate the prognostic impact of DMRGM in AML.
Grouping DMRGM as a functional group of genes to study the impact on the survival of AML patients can be controversial. For example, SETBP1 mutations in AML are relatively uncommon compared to TET2, IDH1/2 and DNMT3A, and the prognostic impact of SETBP1 mutation has not been well documented. However, it is well established that IDH1, IDH2 and TET2 are usually mutually exclusive in AML [5,18,22,35]. In addition, mutations of these three genes have similar epigenetic impacts leading to global DNA hypermethylation [22,[36][37][38]. DNMT3A has also been reported to play an important role in DNA methylation, and several studies have demonstrated a correlation between DNMT3A mutation and DNA methylation, although the function and biological consequences of DNMT3A mutations have yet to be fully elucidated [38][39][40]. Additionally, particular mutations of DNMT3A may have different effects on methylation regulation. Some studies have suggested that SETBP1 is involved in methylation [41,42]. In the current study, only two patients had SETBP1 mutations, both in the ELN adverse risk group, and DMRGM did not impact survival outcome in this group. Future studies, particularly for SETBP1-mutated patients, are needed to validate the approach of considering DMRGM as a functional group.
Among the ELN favorable risk group, the cases with mutated NPM1 show strong concurrence with DMRGM in our study. This finding agrees with previous studies which have reported that DNMA3A, IDH1/2 and/or TET2 frequently co-occurred with NPM1 mutation in AML patients [4,5,43]. The study by Papaemmanuil et al. revealed that 73% (319 of 436) of NPM1-mutated AML patients are positive for mutations in at least one of the following genes: DNMT3A, IDH1, IDH2, and TET2 [5]. Our findings confirm that DMRGM frequently concur with NPM1 mutation and suggest these mutations may adversely impact OS.
Although our study suffers from limitations of small sample size, particularly only 37 patients in the ELN favorable group with 16 of them carrying DMRGM, and retrospective analysis, it can have important clinical implications if confirmed by others. The current recommendation is to advise against allogeneic stem cell transplantation in patients with favorable risk disease in CR1. Thus, ELN favorable risk patients, particularly those defined by mutated NPM1, with DMRGM may not receive adequate treatment. Our findings suggest the importance of including DNA methylation regulatory genes in the targeted Diagnostics 2021, 11, 986 9 of 11 NGS panel along with the genes used to define current ELN risk groups to further evaluate the clinical significance of DMRGM.

Conclusions
In conclusion, our results indicate that DMRGM are common in AML patients and adversely impact the prognosis of AML patients, particularly those in the 2017 ELN favorable risk group defined by mutated NPM1. The latter patients had OS similar to that of patients in the ELN adverse risk group. Further studies with a large sample size are warranted to confirm our observations.