miR-182 promoter methylation is higher in elderly and unfit AML samples than in normal controls (NCs)

Our reports have demonstrated that a total of three CpG islands (CpG island 1–3) are located 6–12 kb upstream of miR-182, and the methylation frequency is higher in CpG island 3 than that in CpG island 2 and 1 in leukemic cells [15, 16]. CpG island 3, representing the actual methylation frequency, was selected for the following analyses (Figure S1A and B). DNA was extracted from BM mononuclear cells of 94 elderly and unfit AML (Table S1) at initial diagnosis and 20 NC samples to explore miR-182 promoter methylation frequency. The average methylation frequency of CpG island 3 at the miR-182 promoter was 25.9% in 94 AML samples, and this methylation frequency was significantly higher in AML than that in 20 NC samples (9.4%) by MethylTarget™ assays (P < 0.0001, Figs. 1A, S2A–B, and S3A–B). Because MethylTarget™ assays include 13 CpG sites, we subsequently analyzed the methylation frequency of individual CpG sites. 9 of 13 CpG sites had higher methylation frequency in AML than those in NC samples (Fig. 1B). The detailed 13 CpG sites for MethylTarget™ assays were indicated in Figure S4A.

Fig. 1

miR-182 promoter methylation frequency is higher in elderly or unfit AML patients than in NC samples. A MethylTarget™ assay was performed to analyze DNA methylation percentage of CpG islands 3 in 94 elderly or unfit AML patients and 20 NC samples. B The methylation frequency of individual CpG sites by MethylTarget™ assay was analyzed in 94 elderly or unfit AML patients and 20 NC samples. C and D Bisulfite-genomic sequencing was used to assess the methylation frequency in two AML patients with miR-182 promoter hypermethylation and two AML patients with miR-182 promoter hypomethylation. Four colonies were shown for each AML sample. Each row of the circle represents an individual clone. Empty and black circles represent unmethylated and methylated CpG dinucleotides, respectively (C). The statistical analysis of methylation frequency is shown (D). E BCL2 protein expression was measured in four AML samples with miR-182 promoter hypomethylation and five AML samples with hypermethylation. *P < 0.05; ***P < 0.001; ****P < 0.0001. ns: Not significant

Our results from MethylTarget™ assays demonstrated that most AML patients present relative hypermethylation at initial diagnosis, although 11.7% of AML patients exhibited hypomethylation at the miR-182 promoter (< 10%). To further validate the methylation frequency of CpG island 3, bisulfite genomic sequencing was performed in two AML samples with miR-182 promoter hypermethylation and two with hypomethylation, which were categorized based on the results of MethylTarget™ assays. Consistent with MethylTarget™ assays, methylation frequency is substantially higher in two AML samples with miR-182 promoter hypermethylation than that in two AML with miR-182 promoter hypomethylation (Fig. 1C and D). The detailed 19 CpG sites for bisulfite genomic sequencing were indicated in Figure S4B.

As reported, AML patients with high BCL2 protein expression are sensitive to VEN + HMA treatment in vitro and in vivo [9, 20, 21]. Lost expression of BCL2 protein is considered the main factor leading to resistance to VEN + HMA treatment in AML patients [22]. We then determined whether AML cells with miR-182 promoter hypermethylation were associated with high BCL2 protein levels. We measured BCL2 protein expression in five AML samples with miR-182 promoter hypermethylation and four with hypomethylation. As expected, AML samples with miR-182 promoter hypermethylation had higher BCL2 protein expression compared with those with hypomethylation (Fig. 1E).

AML cells with miR-182 promoter hypermethylation are more sensitive to VEN treatment in vitro and in vivo

To further assess and compare the effects of VEN treatment in primary AML with miR-182 promoter hypermethylation or hypomethylation, AML with miR-182 promoter hypermethylation-1 and -2 and AML with miR-182 promoter hypomethylation-1 and -2 were treated with VEN (0.5 μM) at 24, 48, and 72 h, and viability was measured in vitro. VEN treatment significantly inhibited viability in AML cells with miR-182 promoter hypermethylation-1 and -2 (Fig. 2A and B) but did not markedly decrease viability in AML cells with miR-182 promoter hypomethylation-1 and -2 (Fig. 2C and D). Furthermore, we xenografted AML cells with miR-182 promoter hypermethylation-1 and AML with miR-182 promoter hypomethylation-1 in NSG mice and treated with or without VEN in vivo. VEN treatment substantially extended the OS in hyper-1-xenografted mice (Fig. 2E). However, VEN treatment did not affect the OS in hypo-1-xenografted mice (Fig. 2F). These results preliminarily demonstrated that AML with miR-182 promoter hypermethylation might be more sensitive to VEN treatment than AML with miR-182 promoter hypomethylation.

Fig. 2

AML cells with miR-182 promoter hypermethylation are more sensitive to VEN treatment in vitro and in vivo. A–D Cell viability was measured in two AML cells with miR-182 promoter hypermethylation and two AML cells with hypomethylation, which were treated with or without VEN (0.5 μM) for 24 and 48 h. E and F OS was measured in NSG mice xenografted with one AML cells with miR-182 promoter hypermethylation and one AML cells with hypomethylation treated with or without VEN. *P < 0.05; ***P < 0.001; ****P < 0.0001. ns: Not significant

No significant difference in baseline characteristics between AML patients with miR-182 promoter hypermethylation and hypomethylation

Ninety-four unfit newly diagnosed patients treated with AZA + VEN for at least one cycle were enrolled in this study (Table S1). The median age at diagnosis was 69 years, and 25 patients (26.6%) were older than 75 years. 42 AML patients (44.7%) were 3 or 4 according to the Eastern Cooperative Oncology Group (ECOG). For 2024 European LeukemiaNet (ELN) risk stratification [23], 44 AML patients (46.8%) were in the adverse risk group. We next analyzed patient characteristics, including main demographic and clinical and laboratory features, in AML patients with miR-182 promoter hypomethylation and hypermethylation. There were no statistically significant differences between the two groups’ baseline characteristics (Table S1).

The average time to obtain CR/CRi in AML patients with miR-182 promoter hypermethylation is shorter than that in AML patients with miR-182 promoter hypomethylation

We first assessed the CR/CRi status in AML patients with miR-182 promoter hypomethylation and hypermethylation. A total of 32 patients (68.1%) achieved CR/CRi in AML patients with miR-182 promoter hypomethylation (Fig. 3A), while the CR/CRi was 72.3% in AML patients with miR-182 promoter hypermethylation (Fig. 3A). There was no statistically significance of CR/CRi rate observed in AML patients with miR-182 promoter hypermethylation and hypomethylation by Fisher’s exact test (Fig. 3A). However, the average time to obtain CR/CRi was significantly shorter in AML patients with miR-182 promoter hypermethylation (28.9 days) than that in AML patients with hypomethylation (48.3 days) (P = 0.0003, Fig. 3B).

Fig. 3

Evaluation of AZA + VEN treatment efficiency in AML patients with miR-182 promoter hypermethylation and hypomethylation. AML patients were divided into hypermethylation (> 20.21%) and hypomethylation (< 20.21%) according to median value. A The complete remission (CR)/CR with incomplete hematologic recovery (CRi) response numbers were calculated in AML patients with miR-182 promoter hypermethylation and hypomethylation. B The average time to achieve CR/CRi was analyzed in AML patients with miR-182 promoter hypermethylation and hypomethylation. C and D The impact of miR-182 promoter hypermethylation and hypomethylation on overall survival (OS, C) and leukemia-free survival (LFS, D) among AML patients. ***P < 0.001; ns: Not significant

miR-182 promoter methylation is a prognostic biomarker for OS and LFS

We subsequently analyzed the OS and LFS in 94 AML patients. The median OS was significantly shorter in AML patients with miR-182 promoter hypomethylation (9.33 months, 95% CI 6.77–16.13) compared with those with miR-182 promoter hypermethylation (18.43 months, 95% CI 10.4–NA) (P = 0.031, Fig. 3C). In addition, the median LFS was also shorter in AML patients with miR-182 promoter hypomethylation (8.77 months, 95% CI 6.43–17.63) than in those with miR-182 promoter hypermethylation (21.27 months, 95% CI 15.77–NA) (P = 0.0087, Fig. 3D). We further assessed the number of relapsed AML patients in both groups. Fifteen AML patients with miR-182 promoter hypomethylation (31.9%) recurred, but nine AML patients with miR-182 promoter hypermethylation (19.1%) relapsed (P = 0.156).

To determine potential factors affecting OS and LFS, univariable survival analyses were conducted by Cox analyses (Table 1). The univariable analyses revealed that miR-182 promoter methylation status as a categorical variable at diagnosis had a significant impact on OS and LFS (OS: HR 0.371 95% CI 0.210–0.657, P < 0.001; LFS: HR 0.293, 95% CI 0.149–0.579, P < 0.001). In addition, the following clinical characteristics, including BM blast, ELN risk group, and TP53 mutation, were significantly associated with OS and LFS (Table 1). Additionally, variables such as achieved CR/CRi and the mutations in STAG2 and NPM1 correlated with OS but not with LFS (Table 1). Subsequently, we enrolled variables with statistical differences in univariable analyses (P < 0.05) into the multivariable analyses. As indicated in Table 2, in addition to the non-adverse ELN risk group and achieved CR/CRi, miR-182 promoter hypermethylation at diagnosis was a novel independent predictor for longer OS (HR 0.512, 95% CI 0.274–0.957, P = 0.036) and LFS (HR 0.358, 95% CI 0.175–0.733, P = 0.005).

Table 1 Univariable analysis of clinical factors for the survival in AML patientsTable 2 Multivariable analysis of clinical factors for survival in AML patients

When miR-182 promoter methylation level at diagnosis was enrolled in the univariable Cox progression analyses as a continuous variable, it still remained positively associated with OS and LFS (OS: HR 0.972, 95% CI 0.954–0.992, P = 0.005; LFS: HR 0.962, 95% CI 0.939–0.987, P = 0.003), as displayed in Table S2. In addition, when enrolled into the multivariable analyses as a continuous variable, it kept independent effects on OS (HR 0.968, 95% CI 0.947–0.990, P = 0.004) and LFS (HR 0.969, 95% CI 0.946–0.991, P = 0.007) as well (Table S3). In other words, a 3.2% reduction in mortality was significantly associated with each percent of the miR-182 promoter methylation frequency at diagnosis. Concurrently, for every percentage rise in miR-182 promoter methylation frequency at diagnosis, there was a 3.1% reduction in recurrence or mortality rate.

We then explored whether the percentage of AML patients that receive allogenic hematopoietic stem cell transplantation (allo-HSCT) affects the OS and LFS in the hypermethylated and hypomethylated groups. Three AML patients with miR-182 protomer hypomethylation (6.4%) received allo-HSCT, and two AML patients with miR-182 protomer hypermethylation (4.3%) received allo-HSCT. The percentage of AML patients receiving allo-HSCT was very low and similar in both groups. Therefore, this is not the confounding factor for survival between the two groups.

Predictive accuracy of the miR-182 promoter methylation in AML mortality risk via ROC curve

To determine the accuracy of the miR-182 promoter methylation level at diagnosis in predicting mortality risk in AML patients, we performed ROC curve analyses. The area AUC value was 0.831 for mortality at 1-year (95% CI 0.681–0.981, P < 0.001, Fig. 4A), 0.788 at 2-years (95% CI 0.649–0.926, P = 0.001, Fig. 4B), and 0.800 at 3-years (95% CI 0.667–0.933, P < 0.001, Fig. 4C). Therefore, our results demonstrated that the methylation frequency at diagnosis has a particular predictive effect on the prognosis of AZA + VEN-treated AML.

Fig. 4

The area under the curve (AUC) analysis in AZA + VEN-treated AML patients. A–C AUC analysis of miR-182 promoter methylation for 1-year mortality (A), 2-year mortality (B), and 3-year mortality (C) was performed in AZA + VEN-treated AML patients

Different methylation frequencies at different disease periods in AML patients

We subsequently measured the frequency of miR-182 promoter methylation in BM cells obtained from AML patients who had undergone AZA + VEN treatment, across different time points. The average methylation frequency of miR-182 promoter was substantially lower in AML patients achieving CR/CRi than in newly diagnosed AML patients (14.1% vs 25.9%, P = 0.024, Fig. 5A). In addition, the average methylation level was substantially higher in relapsed AML patients than in AML patients achieving CR/CRi (22.6% vs 14.1%, P = 0.042, Fig. 5B).

Fig. 5

Different methylation frequencies in AML patients at diagnosis, with CR/CRi, and in relapse. A Methylation frequency at the miR-182 promoter was measured in AML patients at diagnosis and in AML patients achieving CR/CRi. B Methylation frequency at the miR-182 promoter was measured in AML patients achieving CR/CRi and in relapsed AML patients. *P < 0.05

Leukemic differentiation stage does not affect miR-182 promoter methylation status in AML cells

Although the cellular differentiation block is a fundamental characteristic of AML cells, AML cells have limited signs of differentiation [24]. Pei et al. report that AML cells in more differentiation stages, especially monocytic AML, resist VEN-based therapy [9]. Therefore, we subsequently explored whether miR-182 promoter methylation is associated with the differentiation stage. CD11b levels representing more differentiation stages were analyzed in 94 AML samples by flow cytometry. 15 of 94 (15.9%) AML samples are CD11b+ (Fig. 6A). 8 of 15 (53.3%) CD11b+ AML samples presented miR-182 promoter hypermethylation, and 7 of 15 (46.7%) CD11b+ AML samples presented hypomethylation (Fig. 6A). Therefore, AML with miR-182 promoter hypermethylation had an equal frequency of CD11b+ cells compared with AML with miR-182 promoter hypomethylation (Fig. 6A). Also, AML with miR-182 promoter hypermethylation had an equal frequency of CD11b− cells compared with AML with miR-182 promoter hypomethylation (Fig. 6A). Furthermore, the frequency of miR-182 promoter methylation was similar in AML samples with CD11b+ compared with those with CD11b− (Fig. 6B).

Fig. 6

Leukemic differentiation stage does not affect miR-182 promoter methylation status. A 94 AML samples were divided in CD11b+ or CD11b− cells. The frequencies of AML cells with miR-182 promoter hypermethylation and AML cells with hypomethylation were analyzed in CD11b+ or CD11b− AML cells. B miR-182 promoter methylation frequency was analyzed in CD11b+ or CD11b− AML cells. C CD11b and CD14 staining were performed by flow cytometer in U937 cells treated with 1 μM ATRA, 0.1 μM PMA, or DMSO (1:1000) as control (Ctrl) for 72 h. The representative plots (left) and statistical analysis of CD11b+ or CD14+ cells were shown (right). D Wright‒Giemsa staining was performed in U937 cells treated with 1 μM ATRA, 0.1 μM PMA, or Ctrl for 72 h. E Bisulfite-genomic sequencing was used to assess the methylation frequency of miR-182 promoter in U937 cells treated with 1 μM ATRA, 0.1 μM PMA, 5 μM DAC, 5 μM AZA, or Ctrl for 72 h. Each row of the circle represents an individual clone. Empty and black circles represent unmethylated and methylated CpG dinucleotides, respectively. The statistical analysis of methylation frequency is shown. ***P < 0.001; ****P < 0.0001. ns: Not significant

We next determined whether AML cells in more differentiation stages affected miR-182 promoter methylation, which was measured in U937 cells treated with all-trans retinoic acid (ATRA) and phorbol 12-myristate 13-ac-etate (PMA) as differentiation inducers. CD14/CD11b staining and Wright‒Giemsa staining demonstrated that U937 cells successfully differentiated into more mature cells (Fig. 6C and D). ATRA and PMA treatments did not affect miR-182 promoter methylation frequency in U937 cells by bisulfite genomic sequencing (Fig. 6E). In contrast, DAC and AZA treatments as positive controls significantly decreased miR-182 promoter methylation frequency [16] (Fig. 6E). These results demonstrated that differentiation could not affect miR-182 promoter methylation in AML cells.